EIT rs tiles 0.5: Nari inter for Heal at Da 1 CERT Y p Ton IR See inside of back cover for catalog card. Public Health Service Publication No. 1000-Series 2-No. 1 For sale by the Superintendent of Documents, U. S. Government Printing Office Washington 25, D. C. - Price 30 cents NATIONAL CENTER| Series 2 For HEALTH STATISTICS | Number 1 VITALand HEALTH STATISTICS DATA EVALUATION AND METHODS RESEARCH Comparison of Two Vision-Testing Devices A study to compare visual acuity as measured by the Sight-Screener and the Sloan Letter Chart. Washington, D.C. June 1963 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Anthony J. Celebrezze Luther L. Terry Secretary Surgeon General CAT. FOR PUBLIC HEALTH NATIONAL CENTER FOR HEALTH STATISTICS Forrest E, Linder, Ph.D., Director Theodore D, Woolsey, Assistant Director O. K. Sagen, Ph.D., Assistant Director U.S. NATIONAL HEALTH SURVEY Theodore D, Woolsey, Chief Alice M, Waterhouse, M.D,, Medical Advisor James E, Kelly, D.D.S., Dental Advisor Walt R. Simmons, Statistical Advisor Arthur J, McDowell, Chief, Health Examination Survey Philip S. Lawrence, Sc.D., Chief, Health Interview Survey Robert T, Little, Chief, Computer Applications Public Health Service Publication No. 1000-Series 2-No. 1 PREFACE HEAL This is one of the special methodological studies carried out during the 5 cycle of the Health Examination Survey prograra' of the U.S. National Health Survey to calibrate certain of the tests and measurements used in the standard- ized special health examination of the Survey. Adequate data were not available on the re- liability of some of the tests used or on the com- parability of test results with those obtained by other frequently used methods which were not practical for this Survey. Such information was needed to evaluate examination findings—to de- termine the actual extent of variation existing within the population tested apart from the vari- ation due to the measurement devices used. The measurement of visual acuity was one of the areas in which such a calibration study was needed. Since the examinations had to be conducted uniformly by a number of different examiners in a limited amount of time and in a space too small for testing with the usual wall charts, it was necessary to use a portable meas- uring device. The instrument selected was the Sight-Screener. However, the vision test in the Survey was intended to provide a measurement of sight across the entire scale of visual acuity, rather than just a visual screening. Information was lacking or inadequate on the comparability of measurements obtainable from the different Sight-Screener instruments used and on the com- parability of the Sight-Screener test results with those obtained from a standard wall chart. For these reasons, the U.S. National Health Survey contracted with the Pennsylvania State College of Optometry to conduct a calibration study on the Sight-Screenmer instruments. Dr. Vernon I. Ryan, Assistant Professor of Optom- etry, directed the project which was carried out in the Eye Clinics of the College and provided consultation in the preparation of this report. Arrangements were made with the Wilmer Ophthalmological Institute of the Johns Hopkins University for the use of an improved Snellen- type chart developed by Dr. Louise L. Sloan at the Institute. The design used in the study was developed by Donald Loveland, who was assigned to act as liaison between the Pennsylvania State College of Optometry and the Health Examination Survey Branch during the data collection phases of the study. 056 CONTENTS Page Preface --=-cmmmmm meme eee i Introduction =====--eeee cmc m cece meee ———————— Relevant Research Findings----====cmocmmomommmeoae 1 Study Design---====-ccm mmm 2 Description of Tests and Controls-----=---occmcmmomcmcaoooo 2 Sight-Screener----- cocoon 2 Sloan Charts-==--=eecccm mcm cece meee 3 The Study Group----===--cemomm mmm eee 4 Findings = ======= comme mma 5 Instrument Differences----====ceomcemommccmmcceceeeeean 6 Sight-Screener and Sloan Test Differences-----=---cac--- 6 Summary and ConclUuSIiONS == == === m= mmm mm mmm mm mm sm mm em 14 DDCEAIIET TAD C8 tiveness om is io $2 i ES 17 Appendix I, Target-Specifications and Record Forms ------- 31 Appendix II. Some Technical Notes -=====-c-=ccccmmmmoaa___ 33 SYMBOLS Data not available====cemmcm coca Category not applicable--==camemecmaaaaaa- Quantity zZero===-==m=mmmoe ec mmeeeeeeeee Quantity more than O but less than 0,05---- Figure does not meet standards of reliability or precision------ecececaao-- COMPARISON OF TWO VISION-TESTING DEVICES INTRODUCTION This report presents the findings from a research study of visual acuity testing conducted by the Pennsylvania State College of Optometry under contract with the U.S. National Health Survey. The purpose of the study was: 1. to determine the comparability between the test results that might be expected from the two American Optical Company Sight-Screener instruments used in the first cycle of the Health Examination Survey, and 2. to obtain information on the compara- bility of results from the Sight-Screener instruments and an improved Snellen- type test on a population whose visual acuity was no better than that which might be expected in the general adult population reached through the first cycle of the Health Examination Survey. Relevant Research Findings Sulzman, Cook, and Bartlett’ investigated the reliability of visual acuity measures obtained from several screening devices, using as sub- This report was prepared by Jean Roberts of the U.S. National Health Survey staff. jects 128 naval personnel and others from the New London Submarine Base whose visual acuity tended to be normal or near normal. In their study, the test-retest reliability of the Sight- Screener was found to be slightly greater than that for commercial Snellen-type charts but slightly less than that for an improved Snellen- type chart developed at the Base. The reliability of measures of acuity for distance vision ex- ceeded those for near vision. The relationship between results from the Sight-Screener and from clinical tests was as- sessed by Fonda, Green, and Heagan? among 41 aviation medical examiner students from Randolph Field. In their study, the determinations of visual acuity using the Sight-Screener did not vary more from the determinations utilizing clinical tests than the clinical tests varied among themselves. The comparability of visual acuity test re- sults is dependent upon many factors. The effect of illumination of the test target on visual acuity has been demonstrated in investigations of Lythgoe, Hecht * and many others. These studies show that, within a middle range of luminances, visual acuity is directly proportional to the log- arithm of the luminance when the contrast be- tween test object and background remains con- stant, Low luminance exaggerates the effect on acuity of uncorrected errors of refraction. High luminance minimizes the effect of errors of refraction, Cobb and Moss’ and Ludvigh® showed that acuity increases with increasing contrast between target and background. In her review of the research done with measurements of visual acuity, Sloan’ concluded that, on the basis of available research, the best test situation exists if the contrast is at least 84 percentand when the background brightness is maintained constant within the limits of about 12 to 18 millilamberts (11 to 17 foot-candles). The selection of the ""end-point' or criterion for scoring the tests will also affect the meas- urements of visual acuity. A number of studies, including that of Lythgoe? and others, demon- strate that a more accurate measure of visual acuity is obtained at the normal illumination level if the standard of at least 7 out of 10 correct answers is adopted as the criterion of being able to see (read) the test object. STUDY DESIGN Visual acuity for near and distance vision was determined- by the Sight-Screener and by Sloan Charts (improved Snellen-type charts de- veloped at the Wilmer Institute) for each person in the study group during the 3-month period from June 19 to September 18, 1961. The tests were administered without glasses, and then the appropriate parts were repeated if the examinee wore glasses and had them with him. The order of administration of the tests was randomized so that it would be possible to assess the effect that eye fatigue and other factors might have had on the two visual acuity test series. The Sight-Screener was used first on even-numbered days and the Sloan test first on odd-numbered days. Numbers were assigned examinees in the order in which they were ad- mitted to the project. The distance tests were administered first for those with even numbers, the near tests first for those with odd numbers. For any particular subject, the near-far order was the same for both Sight-Screener and Sloan testing. Right eye, left eye, and binocular acuity were always measured in that order. Subjects wearing glasses were tested first without glasses and then with the glasses. Two Sight-Screener instruments from the National Health Survey were used in the study. Instrument "A" was used at the start and through the first complete week of the project. Instru- ment "B" was used during the second and third weeks, and the instruments alternated biweekly thereafter. Because of the large number of vol- unteers available in the early stages of the proj- ect, three-fourths of the group were tested on Instrument A and one-fourth on Instrument B. Different examiners administered the Sight- Screener and the Sloan tests for a given subject. In all, 21 examiners were used during the study. With the exception of the project director, who also did some of the testing, none were assigned for more than a 2-week period. In this way, it was possible to minimize measurement variance attributable to any one examiner. Preprinted record cards, containing the test letters at each acuity level for the Sight-Screener and Sloan tests, were used for recording the test results (see Appendix I). These contained sep- arate sections for near and distance vision as well as for tests with and without glasses. A maximum of 24 tests would have been given a subject who wore glasses for near and distance vision. DESCRIPTION OF TESTS AND CONTROLS Sight.Screener This instrument uses the stereoscopic prin- ciple to achieve the optical equivalent for dis- tance in testing visual acuity. Near vision is tested without the interposition of lenses. Monocular visual acuity is measured under conditions of binocular seeing. Both eyes view the illuminated slide with vectographic lettering but only the eye that is being tested can see the letters. This is achieved by means of polarized light and polarizing screens near the lenses of the eyepiece. In addition, monocular acuity is tested in such a way that the subject is unaware of which eye is being checked. With the sup- pression test (see the last line of figure 1) it is possible to determine before starting the acuity testing whether the vision is substantially poorer in one eye than the other, Use of the monocular occluder over the better eye for such persons prevents any possibility of overrating the acuity in the poorer eye in these cases. However, oc- cluders were not used in this study so that it would be possible to obtain information on the Zz E CE c¢noz EBNC Tr N F ET coze ze NE Newz E F TC bpzeEN BNE cnzn Figure 1. Sight-Screener target. circumstances under which this overrating may be expected. Identical targets are employed for the opti- cal equivalent of distance and for near vision; but the lines for testing the right eye, left eye, and binocular vision differ (see the top three lines in figure 1). With the tests of near acuity, the target is 14 inches from the eye and about 20 degrees be- low the primary position. The far target is at the optical equivalent of 20 feet simulated by means of lenses. The headrest and mechanical positioning of the eyepiece make it possible to maintain the target distances consistently. Since the Sight-Screener is essentially a screening instrument, it does not provide for the measurement of visual acuity at as many levels as are usually represented on a good wall chart or near-test card. The acuity scale is coarse for the poorer levels, ranging from 20/200 down to 20/50, since there are only four steps and few letters; but it has five steps within the range for better acuities from 20/50 to 20/10 (see Appendix I). Only one letter is provided for testing at the 20/200 and 20/100 levels, two letters at 20/70 688263 O - 63 - 2 and four letters at each of the other levels. The design of the letters follows the Snellen principle, without the serifs—the height or width of the letter being five times the width of the lines in the letter. To 'pass' or be able to read a particular level no errors are allowed in groups withone or two letters and only one error is permitted in groups of four letters. The visual acuity level reached corresponds to that for the group of letters farthest to the right which the examineeis able to read with no more than the allowable number of errors. Sloan Charts The improved Snellen-type near and distance charts used in this study were those developed by Sloan®at the Wilmer Ophthalmological Institute of Johns Hopkins University. The charts utilize 10 capital letters—Z N H R V K DC O S—designed in accordance with the Snellen principle, except that serifs are omitted. Experimental evidence indicates thatthese letters are about as nearly equal in legibility as can be obtained when simple capital letters are used. Moreover, the average difficulty offered by these letters has been shown to be equal to the difficulty in visual resolution offered by Landolt rings having breaks at horizontal and vertical loca- tions. Hence, these letters meet the recommen- dations made by the Committee of Optics and Visual Physiology of the American Medical As- sociation in 1916° and again in 1930.1 Six of the 10 letters are the same as those on the Sight- Screener targets, while 4 differ. Specifications for the size of letters on the Sloan Charts, as well as the Sight-Screener tar- gets, are shown in Appendix I. On the Sloan distance charts there are 13 gradations in letter size for the range from 20/13 to 20/200 (fig. 2). The Sloan near chart provides for measure- ment at 14 levels from 16/12.8 to 16/256 (cor- responding to a range of 20/16 to 20/320 in the notation for distance testing). To '"pass' or be considered able to read at a particular level no more than 3 errors were allowed if the line contained 9 or 10 letters and no more than 2 errors if there were 5, 6, or 8 letters. For lines with 1, 2, or 4 letters, the same criterion was used for scoring as in the VISUAL ACUITY CHART FOR 20 FEET VISUAL ACUITY CHART FOR 20 FEET suoan ums oan ums g 1 1 i NZ TITOoOQUO 33 < N 1< xX << XA A < on on I A oO oN Z Tow °c0 OU N nA I < o I! KDVRH NZ COS - Pf VRNHZ DC SKO {i HNORGC zZsvDK * i NRoZH ovVSDK = Figure 2. Sloan distance chart. Sight-Screener tests. The visual acuity level reached corresponds with the line farthest from the top of the chart which the examinee is able to read with no more than the allowable number of errors. Target illumination was maintained within a range of 12 to 18 foot-candles on both near and distance charts throughout the study. The far target was at a distance of 20 feet, that for near vision at 16 inches. While the near target distance for the Sloan test differed from that in the Sight-Screener, appropriately scaled near targets were used for each so that the test results would be comparable. The device which supported the near Sloan target at the standard distance contained a chin rest and occluders which could be moved so as to cover the eye not being tested (fig. 3). The metal backing on which the near target rested was curved so that the en- tire card would be roughly 16 inches from the eye in normal position. With these charts, the same target was used for testing the right eye, left eye, and binocular vision. Targets for near and distance vision differ, 4 Figure 3. Sloan near target mounted. THE STUDY GROUP The study group consisted of 502 English- speaking, literate adult volunteers ranging in age from 17 through 79 years with a wide range of visual acuity. They were selected from the patients of the Eye Clinics at the Pennsylvania State College of Optometry, friends and relatives accompanying the patients, and the staff of the Eye Clinics, during the 3-month period from July 19 to September 18, 1961. Clinic patients were given the test batteries in the study before under - going their regular examinations in the clinic. Only those persons who had no obvious handicap, such as an inconveniencing infirmity, lack of in- telligence, or language barrier, were admitted to the study. The presence of an ocular pathology without discomfort did not bar acceptance. While no attempt was made to select a ran- dom sample of the adult population under 80 years, the group did include a substantial number of persons throughout the entire age span and over the range of visual acuity of concern in the first cycle of the Health Examination Survey. The age-sex composition of the group is shown in table A. Sixty percent were of the white race (table 1). Proportions of men and women aged 45-64 years and of nonwhite women aged 25-44 years were slightly larger than might have been expected. Table A. Number of persons and percent distribution of study group, by age and sex Age Total Male Female All ageS-===mrmmmmmmmmeme mee ———————— 502 238 264 Percent distribution Total-===mremree ccc ——————————————— 100.0 47.5 32.3 17-24 year§--===-==-em-eecececcc cc —————— 11.0 6.8 4.2 25-44 yearS-======--mmemmeeee meee ———— 34.9 13.4 21.5 45-64 yearS-=--mmmmmemmmmeeme meme —————————— 41.3 20.9 20.4 65-79 years-=mrrmmmmmcm mmm ————————— 12.8 6.4 6.4 Fifty-six percent of the group (282 persons) FINDINGS were tested with and without glasses. Nearly three-fourths of these persons (205) needed a correction made in their lenses. The remaining 220 persons were tested without glasses only. Roughly one-third of these persons were in need of glasses for either near or distance vision or both. Among persons under 45 years of age, 60 percent were found to have uncorrected binocular distance visual acuity of 20/20 or better on both tests, the proportion dropping to less than 10 percent for those 65 years of age and older (see table B and tables 2-5). Comparison is made here between test re- sults from the two Sight-Screener instruments for groups of subjects tested under identical con- ditions and between Sight-Screener and Sloan scores for the same individuals on comparable tests. For making these comparisons, visual acuity was expressed in terms of the size of the visual angle in minutes subtended by the optotypes—the width of the lines in the smallest letters read correctly in accordance with the study criterion. Averages were then converted to decimal or to Table B. Proportion of the study group with normal binocular vision without glasses, by age Acuity test 17-24 years 25-44 years 45-64 years 65-79 years Sight-Screener Distance=========mmeemmeu= 60 Near=========cmemcmcceeanoa 72 Sloan Distance===========mm=ee--= 62 Near========cmeccmccecaaen 69 Proportion with 20/20 vision or better 65 37 2 49 1 2 64 37 9 58 3 2 Snellen notation in the text tables presentedhere. The decimal values shown are the reciprocals of the visual angle measurements. The more com- monly used Snellen fractions give in the numer- ator the distance of the test target from the ex- aminee. The denominator is the distance at which the particular line (or block of letters) should be read correctly by a person with normal vision. Scores of 2.0, 1.0, and 0.5 minutes (visual angle size), for example, would be equivalent to scores of 0.5, 1.0, and 2.0, respectively, in decimal notation and to 20/40, 20/20, and 20/10, respec- tively, in Snellen notation. Persons whose visual acuity was less than 20/200 on a particular Sloan or Sight-Screener test of near vision were excluded from both comparable test parts, since this was the lower limit measurable with the Sight-Screener targets. Instrument Differences Determination of the reliability of the Sight- Screener instruments, as measured by the com- parability of scores attained on two of theinstru- ments used in the Health Examination Survey, was one major concern of this study. As indicated in the Study Design, three- fourths of the group were tested with Instrument A and one-fourth with Instrument B. Eachofthese was further subdivided into four groups according to which tests were given first—the near or far and the Sight-Screener or Sloan. (See tables 10- 13.) Fatigue does not appear to have affected the test results appreciably for those tested on either instrument. In general, scores attained on the Sight-Screener were no better when that battery was given first than when it followed the Sloan series. Similarly, subjects did no better on the first near or far test than on the second com- parable test regardless of whether the near or far battery was given first. An indication of the comparability of scores attained on the two instruments is shown in table C and figure 4. Visual acuity scores for persons tested on Instrument A did not differ significantly from scores attained by those tested with Instru- ment B on any of the 12tests—monocular (right and left eye) and binocular, near and distance vision with or without glasses. The differences between average scores shown here is no greater 6 than would be expected through chance alone in samples of this size. It may be seen in figure 4 that the distribution of scores obtained on the two instruments is similar. Sight-Screener and Sloan Test Differences The second major purpose of this study was to obtain information on the comparability be- tween Sight-Screener and Sloan test results for a population whose visual acuity was no better on the average than that which might be expected in the general adult population to be reached through the first cycle of the Health Examination Survey. To do this it is necessary to take into ac- count the fact that measurements of visual acuity are affected by a number of external factors not all of which are directly related to the compara- bility of the Sloan and Sight-Screener tests. Re- call of identical lines on the testtarget, the effect of practice, fatigue resulting from taking two tests in sequence without rest, and differences in the test targets are some of those which need to be considered. Recall and practice. —As mentioned pre- viously, the near and far targets for the Sight- Screener are identical, although the lines for testing the right eye, left eye, and binocular vi- sion differ. If recall of the target letters substantially affected measurements on these instruments, acuity scores would be better on distance tests for those given the near tests first and on the near tests for those given the distance onesfirst, provided this occurred in the absence of similar results on the Sloan. This assumes that the sub- ject recognizes the letters he was able to recall from the first test more quickly the second time he sees them and, hence, would have time to resolve letters further along on the target within the normal test time limits. Comparison is made here only on tests given without glasses since these were the series given first in each battery. Average scores for the subgroups in the study, as shown in table D, indicate the possibility of recall or some other factor affecting results on the near tests given without glasses—for both monocular and binocular vision. The near un- corrected acuity on the Sight-Screener, but not on the Sloan tests, is significantly better on the Table C. Average scores on Sight-Screener and Sloan Tests for groups in which Sight- Screener Instruments A and B were used! Sight-Screener A Group Sight-Screener B Group Test? Sight-Screener Sloan Sight-Screener Sloan WITHOUT GLASSES Distance vision Monocular---------ccccccccc mem —— .38 % Jhb 42 Binocular--=====--ermmmccc cc —————— 46 +52 +33 .54 Near vision Monocular-----=-==---cccceemmeaan- +27 +32 +25 .28 Binocular-----===-===-ccececenn—- +33 +33 +27 +29 WITH GLASSES Distance vision Monocular----====-cmemecmmemm enn .67 .66 .73 .78 Binccular----==-=-ccecmmmm meee +31 +35 .89 +99 Near vision Monocular--------emeeece cece ———— .60 .58 .61 +39 Binocular----==---mmmeccc ccc —————— «42 .70 .65 +653 1Scores given in decimal notation. 2Monocular tests are those in which the right eye and the left eye were examined separately. The results were combined for this and subsequent tables unless otherwise indicated. average for those given the far tests first than it is for those given the near tests first. While a similar pattern may be seen with the Sloan scores, the differences are not statistically significant. Average uncorrected distance acuity scores when the near tests are given first do not differ from those when the far tests are given first any more than would be expected through chance alone. Consequently, the Sight-Screener tests do not show any consistent evidence that recall has affected the results here. The possibility of recall also exists on the Sloan tests, since the lines for the right eye, left eye, and binocular vision are identical. If learn- ing the target letters substantially affected the scores, there would be a consistent increase from the right eye to the left eye to the binocular scores on the Sloan tests in the absence of a similar pattern or to a greater extent than on the Sight-Screener scores for the same persons. Table E shows a systematic improvement in average scores on the successive types of test. However, the differences between the cor- responding Sight-Screener and Sloan scores are insignificant; hence, recall does not appear to have appreciably affected the Sloan test results. Instead, the pattern here may indicate an im- proved score resulting from practice in both tests. 50 Distance © 30 ) S c d eS & 10 + fr — |NStrument A fi weeensene Instrument B J 1 | | 1 d 0 0.2 0.4 0.6 0.8 1.0 70 5 Distance Q ® ki o ° Cc © ~~ a ; | 0 0.2 0.4 0.6 0.8 1.0 lL Visual acuity TESTS WITHOUT GLASSES TESTS WITH GLASSES 50 Near 30 7] 10 *) 7] Tar” | | | 1 1 0 0.2 0.4 0.6 0.8 1.0 60 0 0.2 0.4 0.6 0.8 1.0 Visual acuity Figure 4. Percentage distribution of binocular visual acuity, in decimal notation, on Sight-Screener Instruments A and B. Fatigue.— The administration in sequence of two complete visual acuity tests to the same per- son without rest might be expected to produce fatigue, although available experimental evidence would indicate that the series used in this study are probably not of sufficient length to do so. If this factor did in fact affect the scores, the study group could be expected to perform better on the test battery given first. Average scores for distance monocular vi- sion, uncorrected, on the Sloan tests exceeded those on the Sight-Screener regardless of which battery was given first (table F). On the binocular distance tests without glasses, the group given the Sight-Screener test first did better on the Sloan. Only on the distance binocular tests with glasses did the group given the Sloan tests first do better on their first than on their second comparable test. 8 Half of each of these two groups—those given the Sloan first and those given the Sight- Screener first—started with the near tests and half with the far tests. In the foregoing compari- son, the effect of fatigue may have been masked when those starting with the near tests were combined with those who started with the distance tests. If these persons are now separated and com- parison made between Sight-Screener and Sloan scores for the appropriate subgroups, the sub- jects tended to do no better, if as well, in the tests which they took first than they did on the comparable tests parts administered later (table G). Even scores on the binocular distance tests with glasses do not differ more than would be expected through chance alone in samples of this size. Table D. Average scores attained onSight-Screener and Sloan Tests for those given near tests and those given far tests first Average score on Test Sight-Screener Sloan Near first Far first Near first Far first WITHOUT GLASSES Distance vision Monocular ===mm= mm —m———————————————— +39 40 43 42 Binocular---==-==meeemm cece c————— .49 47 .54 .51 Near vision Monocular---====rm;e eee ——————————— a +29 «29 .30 Binocular mere mmm mmm mm rm mm oe om om .28 .35 1 «35 WITH GLASSES Distance vision Monocular--====mmmm__————— om om om “12 .65 74 .64 BLITOCULAIL = ww mimimn.omimmism m sm i tind .86 81 .98 +79 Near vision MONOCULAL =m mmm mmm mom om mm mt et mm mn em +37 +62 +35 .61 Binocular---==m=mmm_e_———————————— .70 .69 +71 .66 Scores given in decimal notation. Consequently, fatigue does not appear tohave affected the test results substantially. These find- ings are consistent with those of Rabideau'! who found that fatigue did not affect testresultsin se- ries of tests consisting of eight different targets each presented 20 times in succession without rest. Test target differences.—The effective il- lumination and target-background contrast were within generally acceptable limits for both tests. However, two essential differences do exist be- tween the targets which may account for the slightly better scores on the Sloan than on the Sight-Screener. At each level, the Sloan charts provide more letters for practice—10 letters each for more than half of the levelsas compared with a maximum of 4 letters for the Sight-Screen- er. Also the Sloan letters are more nearly com- parable in difficulty than are those on the Sight- Screener targets. Test comparison. — Assuming that the Sight- Screener and the Sloan tests are measuring the same aspects of visual acuity in a similar man- ner, scores attained on the two tests by the same individual should differ only by chance, other factors being equal. It has been shown that for persons in this study, recall of the test target letters and fatigue have not affected test results appreciably. Yet, scores attained on the Sloan battery tended to be slightly better than those on the Sight-Screener. Table E. Average scores attained on Sight-Screener and Sloan Tests for the right eye, left eye, and binocular vision! Average score without Average score with glasses glasses Test Distance Distance toasts Near tests tests Near tests Sight-Screener Right eye---==-cccccmmmmcmaacaaaa CA + 26 .64 «55 Left eye--=-=-ceccmcmcmccccceeeem .40 .27 .73 +83 Binocular---------c-cccmcmcaao 48 +31 .83 .70 Sloan Right eye---=-=-cccccmccmcnccnaaan 40 «258 .66 .56 Left eye-----=-cmcmmcm mmm yan .30 v 13 .60 Binocular=-==----coccccmomccnaaaaoo +53 +33 .88 .68 Scores given in decimal notation. Table F. Average scores on Sight-Screener and Sloan tests for those given Sight-Screener or Sloan first! Sight-Screener tests Sloan tests first— first—average scores average scores Test Sight- Sight- Screener Sloan Screener Sloan WITHOUT GLASSES Distance vision Monocular--=--=----cc-oomomooooooo .38 42 41 Jab Binocular----==----cemommmmcme no 47 +33 49 .53 Near vision Monocular-----==---cccecmcmaooaaono 27 .30 .26 «29 Binocular=-------c---eommmcmmmaoooo .31 34 .32 +33 WITH GLASSES Distance vision Monocular=====--cecem ccm ancanoo .67 .68 .70 71 Binocular-------=---ccmocccmcoaa .83 .84 .83 «93 Near vision Monocular-==-=cccccm emcee eee «39 .60 +30 .56 Binocular-----cccccmmcccamac nana 72 .73 .68 .64 Scores given in decimal notation. Table G. Average scores on Sight-Screener and Sloan tests for selected groups, accord- ing to which tests were given first! Average scores with Distance and Sight- Distance and Sloan Test Screener tests first tests first Sight- Sight- Screener $1530 Screener Sloan DISTANCE VISION Without glasses MOTIOCULLAT ww om mm mm mm me me .38 41 42 43 BLAIOCUI LAL mmm mmm mw mm mm me mr mm ime vA .50 «51 +52 With glasses MOTIOCU LAL =m mmm smi stm sm 0 mm ie mio ime 48 +63 .62 .67 .68 BinoCULAT =e mm mw mw mm mm om mm mm em me .84 «dd 77 .87 Near and Sight- Near and Sloan Screener tests first tests first NEAR VISION Without glasses MOTIO CU LAI mimi immo to mm mt tee .24 +29 +33 37 BiriOCULAI = mm om oom oom im om smi 0 a 2D 33 +28 +30 With glasses Monocular----------=-=------------ .54 .56 .60 .54 BINOcul al = =m mim mio i cmt sn em 0 si ne .68 w?3 «73 +89 Scores given in decimal notation. Only on uncorrected distance tests of monocular and binocular vision, however, were the mean differences statistically significant (table H), As may be seen in figure 5 and tables 6-9, the dis- tribution of scores on Sight-Screener tests is similar to that for the corresponding Sloan tests. A test of the comparability of the entire dis- tribution on each Sloan test with its counterpart on the Sight-Screener series required combining the scores on each into six groups—20/20 or 688263 O - 63 - 3 better, 20/30, 20/40, 20/50, 20/100, and 20/200. The percentage of examinees reaching the var- ious levels for binocular vision are shown in figure 5. Only for uncorrected monocular dis- tance and binocular near vision do the distribu- tions differ more than would have been expected through chance alone. The former comparison gives a chi-square value of 20 which is signifi- cant at the 1 percent level; while the near bin- ocular test shows greater dissimilarity between Table H. Visual acuity from Sight-Screener and Sloan Tests with and without glasses! rest EESTI | erie core | (Sight Screner WITHOUT GLASSES Distance vision Monocular=======ceccm cee eee .40 42 +0.80 Binocular---===-cceccm orm .48 “33 +0.82 Near DEO EUL GI = mse sos sm hse -.27 +29 +0.82 Binocular======= cme eee Me | +33 +0.83 WITH GLASSES Distance vision Monocular=======cc cme eee .68 .69 +0.75 Binocular---====meomee emcee .83 .88 +0.70 Near Monocular=---==cecme meme meee .59 +58 +0.67 Binocular--==-===cememe meee +70 .68 +0.71 Scores given in decimal notation. scores with a chi-square value of 25, It is to be expected that this coarser grouping will tend to mask some differences between tests that might be observable if acuity were measurable at as many and the same levels on the Sight-Screener. A high degree of association was found be- tween Sight-Screener and Sloan scores in this study, better on tests without than with glasses. This is true despite the differences noted between the test targets and the wide range of visual acuities among the study group. As indicated in table H the correlation between scores for un- corrected visual acuity ranged from +0.80 for distance monocular tests to H).83 for binocular near tests. The extent of agreement between test scores for one group on near binocular vision without glasses (the 323 persons tested on Sight- Screener A) is shown in figure 6. Scores on the tests with glasses were not as highly correlated, presumably because of the substantial number of these persons in need of refractive changes in their lenses. The lack of agreement between scores on tests without glasses was primarily of two types. On the one hand, there were those persons whose acuity for one eye was substantially better than for the other. These persons tended to rate better on Sight-Screener tests for the eye with the poor- er acuity than they did on the Sloan test. Since occluders were not used for Sight-Screener tests, the eye not under test often could see a faint ghost image of the target in use and hence read further than the eye under test would normally have been able to read. The second type of problem was evident for other examinees both on monocular and binocular 50 Distance w o T Percentage 80 Distance 60 1 Sloan ere Sight- wennene e.. Screener Percentage H o 20 0 0.2 0.4 0.6 eX} 1.0 Visual acuity TESTS WITHOUT GLASSES 50 Near 70 Near 60 — 20 0 0.2 04 0.6 0.8 1.0 Visual acuity Figure 5. Percentage distribution of binocular visual acuity, in decimal notation, on Sight-Screener and Sloan tests. tests—the examinee who tested as high as 20/100 level on the Sloan test but was unable to read any of the Sight-Screener target. For these persons, it may be that other visual problems such as astigmatism make it more difficult for them to read the vectographic lettering in the Sight- Screener, with the limited number of chances allowed at each level, than it is to read the Sloan charts. Despite those cases in which scores on the two tests differed by two or more acuity levels, a correlation of +0.82 was found on binocular distance tests without glasses and +0.83 for bin- ocular near test (uncorrected), as indicated pre- viously. Comparison of these results for uncor- rected binocular acuity with the test-retest reli- abilities found by Sulzmanet al, ! showed a slight- ly lower degree of association for distance tests in this study but a higher correlation than for the near tests on both the Sight-Screener and the New London (an improved Snellen-type) tests, as indicated below: Tests Test-retest binocular, reliability uncorrected (Sulzman et al.) New London: Far +0.88 Near +0.75 Snellen: Far +0.80 Sight-Screener Far +0.84 Near +0.77 If the test conditions in the New London study were comparable to those in the present investi- 3 Uncorrected neor binoculor vision 080 067} v 050 EE Li 040 $S-SI correlation r=+0.81 Visual ocuity on Sloan 0.20 0.10 TIT TTT 1 1 J i 067 1.00 133 2.00 (20/200) (20/10) Visual acuity on Sight-Screener (A) Figure 6. Scatter diagram of visual acuity on Sight-Screener (A) and Sloan tests. gation, one might conclude that the scores on the Sight-Screener compare as well with the Sloan as they do with repeated tests on the Sight-Screen- er or the standard Snellen Charts. This was true in spite of the fact that the New London group had visual acuities substantially better than those in the present study—roughly 20/20 as compared with 20/40 on the average in distance tests. A further comparison of Sight-Screener and Sloan test results over the entire range of vision from 20/15 to 20/200 is shown in table J for one of the groups in this study. Here it was assumed that persons reaching a particular level or thresh- old of visual acuity in fact have acuities spaced over the interval between that level and the next higher measurable level, and that the distribution of Sight-Screener scores will be similar to that for the Sloan within the interval. On this basis it was possible to estimate roughly the number and proportion of persons expected to have reached the intermediate thresholds which are not meas- urable on the particular test. It may be seen that, as with the grouped scores, the differences be- tween the estimated distributions are negligible. Hence, if the above assumptions are valid, the scores obtained on the Sight-Screener in this study were, in general, comparable to those ob- tained on the Sloan tests throughout the entire range of vision from 20/15 to 20/200. SUMMARY AND CONCLUSIONS Comparison was made of visual acuity as determined by the Sight-Screener instruments and by Sloan charts (an improved Snellen-type) for a group of 502 English-speaking, literate men and women, aged 17 through 79 years, with visual acuity correctable to 20/200 or better. Testing was done without glasses for the en- tire group and then the appropriate parts were repeated if the examinee wore glasses and had them with him, The two Sight-Screener instruments in the study were those used in the first cycle of the Health Examination Survey. Part of the study group was tested on one instrument, the remain- der on the other. Results obtained on the two in- struments were compared. Comparison was also made of results obtain- ed on Sight-Screener and Sloan tests for the same persons and at the various measurable acuity levels. The effect on these test scores of recall or practice, fatigue, and target differences were considered. Visual acuity scores attained by the study group show: 1. There was no difference in scores on the two Sight-Screener instruments that could indi- cate essential differences between the two de- vices. In all 12 tests, the distribution of scores is similar and mean differences are no greater than would be expected through chance alone. 2. Neither fatigue nor recall of target letters appear to have affected scores on either test battery. 3. Target differences do appear to have af- fected test scores to some extent, Scores on the Sloan tended to be slightly better in general than those on the Sight-Screener. However, only on uncorrected monocular and binocular distance tests did the average scores differ more than would be expected through chance. 4. Scores at the various measurable levels on the two tests appear to be essentially com- parable if it is assumed that persons reaching a particular acuity level in fact have acuities dis- tributed over the next higher interval in accord- ance with the distribution of Sloan scores within that interval. It is then possible to estimate roughly the proportion that could be expected to Table J. binocular distance vision Actual and estimated distributions of visual acuity scores for uncorrected Actual distribution Estimated distribution of scores of scores Visual acuity (Snellen notation) Ser comer Sloan Sg Sloan Total number of persons----- 465 465 465 465 Percent distribution Total--=====ccccmcmmcm mean 100.0 100.0 100.0 100.0 20/10-======mcmmmmmm mmm m mm mm ene L.5 --- --- 20/13-==-=-mcmmmmm meme 17.4 15:9 17.4 20/15-====--mmmmmm mmm meee 17.6 3.2 3.9 20/16-==-=mmm mmm meee . 11.8 9.0 7:9 20/20-=======mmmmmmm meme meee eo 26.5 15.5 L745 15.5 20/25-=====mmmm meme 13.1 11.7 13.1 20/30-======---mmmmmmmm meme 18.5 7.7 6.8 7:7 20/40-====mmmmmm meme 8.8 10.8 8.8 10.8 20/50-=====mmmmmm mmm mmm mmm 6.2 5.4 6.2 5.4 20/60-====--ccmmmm meme eee 3.9 2.9 3.9 20/70-====-cmmmmmm meme meee 4.7 1.8 2.4 20/80-==-====-mmmmmmmmm— meee 4.9 5.4 2,5 20/100--======-c-mmmm meme m mmm 12.3 3.2 6.9 3.2 20/125-====--mcmmmm mmm mmm 2.4 X.5 2.4 20/160-========-mmmc mmm mem mmm me 2,2 1.4 2.2 20/200-======-=cmcmmmmmmmmm mem 3:9 1:7 1.0 1.7 reach intermediate levels. pected to reach.) 5. A correlation of +0.80 or better between (It is not possible, however, to predict from this an intermediate score that a particular individual could be ex- on each of the tests given without glasses. This is as high or nearly as high as the test-retest reliability found for the Sight-Screener and for the standard Snellen charts in the New London Sight-Screener and Sloan scores was obtained Submarine Base study. REFERENCES i H. Sulzman, E. B. Cook, and N. C. Bartlett, “Visual Acuity Measurements With Three Commercial Screening Devices,” Progress Report No. 2 on Bureau of Medicine and Surgery Research Project No. X-493, February 7, 1946, as revised by E. B. Cook. Medical Research Department, U.S. Naval Submarine Base, New London, Conn., April 22, 1948. 2G. E. Fonda, E. L. Green, and F. V. Heagan, Jr., “Comparison of Results of Sight-Screener and Clinical. Tests,” Project No. 480, Report No. 1. 27th AAF Base Unit, AAF School of Aviation Med- icine, Randolph Field, Texas, September 4, 1946. 3R. J. Lythgoe, “The Measurement of Visual Acuity,” Medical Research Council, Special Report Series No. 173. London, His Majesty’s Stationery Office, 1932. 4s, Hecht, “Relationship Between Visual Acuity and [llumina- tion,” Journal of Physiology, 11:25, January 1928. Sp. W. Cobb and F. K. Moss, “Relation Between Extent and Contrast in Liminal Stimulus for Vision,” Journal of Experimental Psychology, 10:350, August 1927. by, Ludvigh, “Effect of Reduced Contrast on Visual Acuity as Measured With Snellen Test Letters,” Archives of Ophthalmology, 25:469, March 1941. %. L. Sloan, “Measurement of Visual Acuity,” Archives of Oph- thalmology, 45:704-725, June 1951. 81. L. Sloan, "New Test Charts for the Measurement of Visual Acuity at Far and Near Distances,” American Journal of Ophthal - mology, 48(6): 807-813, December 1959. 9k. Jackson, M. M. Black, A. E. Ewing, W. B. Lancaster, and R. Fagin, "Committee of Standardizing Test Cards for Visual Acuity,” Transactions of the Section of Ophthalmology of the American Med- ical Association, 1916, pp. 383-388. 10g Jackson, et al., “Report of American Committee on Optics and Visual Physiology: Report on Tests and Records of Visual Acuity,” Transactions of the Section of Ophthalmology of the American Medical Association, 1930, pp. 358-363. G. F. Rabideau, *Differences in Visual Acuity Measurement Ob- tained With Different Types of Targets,” Psychological Monographs 69(10), No. 395, 1955. Table 1. 10. 31. 12: 13. DETAILED TABLES STUDY POPULATION CHARACTERISTICS Percent distribution of selected characteristics of study group—race and source of referral, by age--======= === ccm eee VISUAL ACUITY BY AGE Sight-Screener Tests without glasses showing visual acuity, by age-=====--=----= Sight-Screener Tests without glasses according to percent distribution of visual acuity, by @ge---=== =m ee ee eee m Sloan Tests without glasses showing visual acuity, by age-===-====c-cccaccaaaaa. Sloan Tests without glasses according to percent distribution of visual acuity, COMPARISON OF TEST RESULTS Visual acuity thresholds attained on Sight-Screener and Sloan Tests for distance vision without glasses======= === mmm eee eel Visual acuity thresholds attained on Sight-Screener and Sloan Tests for near vi- sion without glasses===== == mm mmm ee ee eee Visual acuity thresholds attained on Sight-Screener and Sloan Tests for distance vision with glasses======m mmm mmm oo eee Visual acuity thresholds attained on Sight-Screener and Sloan Tests for near vi- sion with glasses======= mmm eee eee Distance visual acuity without glasses—mean, standard deviation, and correlation of Sight-Screener and Sloan scores for the eight study subgroups----------=ccaoo Near visual acuity without glasses—mean, standard deviation, and correlation of Sight-Screener and Sloan scores for the eight study subgroups-=-------cecocaaeaa- Distance visual acuity with glasses—mean, standard deviation, and correlation of Sight-Screener and Sloan scores for the eight study subgroups---=-------==-c-o Near visual acuity with glasses—mean, standard deviation, and correlation of Sight-Screener and Sloan scores for the eight study subgroups=-==-==---=--cccoeaoo Page 18 Ie 20 21 22 23 24 25 26 27 28 29 30 Table 1. Percent distribution of selected characteristics of study group—race and source of re- ferral, by age All 17-24 25-44 45-64 65-79 Chavacteristic ages years years years years Number -----=--- mmm memmememmemmme meme ——e—————— 502 35 175 208 64 Percent---=-----====--m-m--eecemmmm—ee—oeo—en 100.0 110 34,9 41.3 12.8 Percent distribution Race White----===-==cmeercccc ccc cec cece emcee mmm ———— 59.6 7.0 18.4 25.8 8.4 Nonwhite--=====mecemrme ccc c emcee cme — mm ———— 40.4 4.0 16.5 15.5 4.4 Source Clinic patientee wom mmm mmm mmmmmimm mmm mm ——— 26.1 1.6 5.8 14.5 4.2 Friend or relative-===--rrrrmeeceeecce eee ee —————— 66.5 7.6 26.5 24.4 8.0 Other------===eeemcmcccmc ccc ccc cece —e mmm mm mmm mm 7.4 1.8 2.6 2.4 0.6 Table 2. Sight-Screener Tests without glasses showing visual acuity, by age Test and acuity! All 17-24 25-44 45-64 65-79 ages years years years years DISTANCE VISION Monocular Total======-=ccecccccm ccm emcee mmm mmm mam 918 91 334 382 111 20/20 or better-------cecceccceecenme— aaa (1.04) 286 45 152 83 6 20/30===mmmmmmmm mmm mc meee meee cee ————— 0.7) 234 18 89 107 20 20/40 =mmmmmmmm Mmmm mmm cece meee mmm (0.5) 83 10 22 41 10 20/50 =mmmmm mmm meee meee mma (0.4) 67 4 19 29 15 20/100" ==mmmmmmmm meme cee c meme mmm ———— (0.2) 169 10 36 86 37 20/200 ===m=mmmmm meee meme meee mm mm mmm —————— (0.1) 79 4 16 36 23 Binocular Total=====cmmeeemem ccc cccc cc cccccc meee me mm 469 47 168 196 58 20/20 or better-=--==-=----c-eccecmccem————= (1.04) 214 29 107 73 5 20/30 ===mmmmmemeec emcee meee e mmm m— mmo mone (0.7) 88 8 26 39 15 20/40 ===mmmmmmmm meee meee mmm emma —— eo (0.5) 42 2 14 24 2 20/50 ====mmmmmmmm meee meee meme mm— mmm (0.4) 29 - 4 17 8 20/100 =====mmmmmmc meme e emcee cmm———— aa (0.2) 78 6 15 34 23 20/200 ====mmmmmee emcee emcee mmm meme mm (0.1) 18 2 2 5 5 NEAR VISION Monocular Total===m=mmmeccccem ccc ccc ce mmm mm ——— 861 94 335 342 90 14/14 or better------=-=--ee-cmccecaccooooo- (1.04) 192 52 133 4 3 IY A EE ttt EEE EL EEE EEE (0.7) 131 18 90 20 3 14/28==mmmmmm mmm eee m mmm mmm (0.5) 60 5 29 24 2 14/35=mmmm mmm mcm c cece cece mm meen mmm (0.4) 56 6 16 24 10 14/70==mmmmmmmm cece ccc ccc c mec c meme em ene mmm ——— (0.2) 246 5 46 157 38 14/140===mmmmmmce ec ccmc mcm m mmm mmm en (0.1) 176 8 21 113 34 Binocular Total======mmm-eccccccccccecccmeeec oem —————— 457 48 170 187 S32 14/14 or better-=-==----ccccccccmemceaaoaaa= (1.04) 137 33 98 5 1 14/2] ===mmmmmmmme meee ee eee memmemem—mm—— aon (0.7) 50 5 27 14 4 14/28 ==mmmmmmm meme cee m mmm mmm (0.5) 40 3 13 21 3 14/35===mmmemm meee cece cme enema (0.4) 28 1 6 15 6 14/70 ==cmmmemmmmmc cme ccm c emcee meme 0.2) 150 4 18 103 25 14/140 ===mmmmmmmme meee cme meme (0.1) 52 2 8 29 13 visual acuity in Snellen notation with decimal equivalent shown in parentheses. Table 3. Sight-Screener Tests without glasses according to percent distribution of visual acuity, by age 1 All 17-24 25-44 45-64 65-79 Test and acuity ages years years years years DISTANCE VISION Percent distribution Monocular Total=-emme meme cee cece cece 100.0 100.0 100.0 100.0 100.0 20/20 or better---=---ce-m oom eeeeeeeeee 31.2 49.5 45.5 21.7 5.4 20/30m === mmm mm mmm ee eee a 25.5 19.8 26.6 28.0 18.0 20/40====cmm mmm mmm mm .0 11.0 6.6 10.7 9.0 20/50= === mmm mem mee eee 2.3 4.4 5.7 7.6 13.5 20/100======= mmm meen ——— 18.4 11.0 10.8 22.5 33.3 20/200=======m mmm eee 8.6 4.4 4.8 9.4 20.7 Binocular Total=mmme meme eee emcee eam 100.0 100.0 100.0 100.0 100.0 20/20 Or better=====-=--c-- omen 45.6 61.7 63.7 37.2 8.6 20/30====== mmm mmm meme emma 18.8 17.0 15.5 19.9 25.9 20/40===== mmm mmm mee eee eee 9.0 4.3 8.3 12.2 3.4 20/50= === mmm mm mm meee een 6.2 - 2.4 8.7 13.8 20/100====== mmm mmm eee 16.6 12.8 8.9 17:3 39.7 20/200====== mm mmm me eee 3.8 4.3 1.2 4.6 8.6 NEAR VISION Monocular Total-====-ecem mm mcc cc ceee EE 100.0 100.0 100.0 100.0 100.0 14/14 or better-----=-=-ccmccmm ccm 22.3 55.+3 39.7 1.2 3.3 J EE te 15.2 19.1 26.9 5.8 3.3 J EE a Ee 7.0 543 8.7 7.0 2.2 14/35 mmm mmm mm mmm mm mee ee eee 6.5 6.4 4.8 2.0 11.1 14/70= =m mmm mm mm mee ee ee ee em 28.6 5+3 13.7 45.9 42.2 14/140==== mmm meme eee 20.4 8.5 6.3 33.0 37.8 Binocular Total=mmmmm meme me eee eee 100.0 100.0 100.0 100.0 100.0 14/14 or better======-=---ocmo meee 30.0 68.7 57.6 2.7 1.9 14/2) mmm mmm mm mm ee eee meee 10.9 10.4 15.9 1.5 1.7 14/28 mmm mmm meme ee eee eee 8.8 6.2 7.6 11.2 5:8 14/35 mm mmm mm mn eee ee eee 6,1 2.1 3.5 8.0 11.5 14/70= === mmm mmm mm ee eee 32.8 8.3 10.6 55.1 48.1 14/140== mmm mmm mm me eee eee een 11.4 4.2 4.7 15.5 25.0 Visual acuity in Snellen notation. 20 Table 4. Sloan Tests without glasses showing visual acuity, by age 1 All 17-24 25-44 45-64 65-79 Test and aeuity ages years years years years DISTANCE VISION Monocular Total-=-====-ccemecm ccc cme memo 911 86 33 381 113 20/20 or better--=-=----mememeeee— ee ——————————————— 310 49 165 87 9 20/30====mmcemmem meee eee eee meee mmm ee ———— 184 12 56 104 12 20/40====mmmmmmmem meme m meee ecm e mcm come ————— 86 6 22 47 11 20/50====mmmmmmcmmm meme m ee mm ee mee —————————————— 91 7 27 38 I 20/100~===m=m=memmeme meee meee meee meme meme ———————— 138 7 37 62 32 20/200-====m=mmemmmemmm eee m em meee e mmm em —————— 102 5 24 43 30 Binocular Total-==m=mmrecc ccc ccc; ce; — ee ——————————— 464 45 168 194 57 20/20 or better==-=-==rmrmemeeece— ee —————————————— 208 27 109 71 1 20/30==mmmmmmmme meme meee mee m meme mm —————— 97 8 18 53 18 20/40==mmmmmm mmm mmm mm —————————————— 50 5 14 21 10 20/50-====mmmme meme ee meee meee eee ——————————————— 25 - 8 10 7 20/100-====mmmmmem meme mm ——————————————— 55 4 12 24 15 20/200 = =m mmm mom mm mm mmm mn 29 3 7 15 6 NEAR VISION Monocular Total=mrmrer rrr errr — cee ——————————————— 839 90 331 331 87 16/16 OF DRL EEE =m mmm mmm mm mm wr om mom rm mw moan oe mor av ot se om creo 169 52 114 1 2 16/20 mmm mmm mm mm nm 138 15 97 26 - 16/32m mmm mmm mm mm om mm nn 80 9 38 28 5 16/40= mmm mmm mmm om on mm nm mm nn cn 44 2 14 22 6 16/80====memcmmnc cmc —————————————————— 223 5 46 133 39 16/160 cm mm mmm om mm om mm om 185 7 22 121 35 Binocular TOCA Ln mmm mmm om mm om om om om mm mm te 452 46 170 180 56 16/16 or better=-=-=----e-emcececceccccccece—c————— 119 33 83 2 1 16/24 mmmmemr men m nmr — rm — nm ———————————————— 73 5 43 25 - 116/32 0m er wm mm nt 52 3 14 27 16/40===mmmec mmr mmm mmm ——————————————— 21 1 8 11 16/80 mmm mmm mom mm mm mm 104 2 11 66 25 16/160~==mmmmmmmmm cece ——————————————————— 83 2 11 49 21 lyisual acuity in Snellen notation. 21 Table 5. Sloan Tests without glasses according to percent distribution of visual acuity, by age i All 17-24 25-44 45-64 65-79 Test and aculcy ages years years years years DISTANCE VISION Percent distribution Monocular TOLER Lom mmm wm mm mm mm mm a a a 100.0 100.0 100.0 100.0 100.0 20/20 or better======----memmmmee eee eee 34.0 57.0 49.8 22.8 8.0 20/30=====mmmmmmmee eee eee eee eee ————— 20.2 14.0 16.9 27.3 10.6 20/40=mmmmmmmmmm meme eee meee ————— 9.4 7.0 6.6 12.3 9.7 20/50===mmmmmmme meme eee meee ———— 10.0 8.1 8.2 10.0 16.8 20/100======m=mmm—m meee eee meme mm ———— 15.1 8.1 11.2 16.3 28.3 20/200 ====mm-mmmme meee memes meee em ———— 11.2 5.8 7.3 11.3 26.5 Binocular TOE Lm om om ome ow mom mm om am mm a mm 100.0 100.0 100.0 100.0 100.0 20/20 or better====-=--m--eceeee eee eeee—————n 44.8 60.0 64.9 36.6 1.8 20/30===m==mmmmmm mmm e meee ———— 20.9 17.8 10.7 27.3 31.6 20/40== =m mmm mmm eee eee eee 10.8 5 6 8.3 10.8 17.5 20/50===m==mmm mmm meme meee meee ———— 5.4 * 4.8 5.2 12.3 20/100 =mmmmmm me ——————————————————— 11.9 8.9 7.1 12.4 26.3 20/200 =m=mmmmmmmmmmm cee eee eee meme eee ———————— 6.2 2:2 4.2 7.7 10.5 NEAR VISION Monocular Total =mmmrm mmm ————————————————— 100.0 100.0 100.0 100.0 100.0 16/16 or better========-=--ceemee eee eeeem—eeeen 20.1 57.8 34.4 0.3 2,3 16/24 mmmmmm mmm meee em 16.4 16.7 29.3 7.9 0.0 16/32===mmmmm mmm mm meme eee meee meme 9.5 10.0 11.5 8.5 5.7 16/40= == mmm mmm mmm meee eee mmm 5.2 2,2 4.2 6.6 6.9 16/80===mm= mmm mmm mmm meee eee emma 26.6 5.6 13.9 40.2 44.8 16/160= == === mmm mm ee me eee 22.1 7.8 6.6 36.6 40.2 Binocular Total---==mmemeemc ccc; ccc cece ce ——— 100.0 100.0 100.0 100.0 100.0 16/16 or better======-==e--eeomee emcee meee 26.3 71.7 48.8 1.1 1.8 16/24mmmmmm mmm me eee ee eee eee eee 16.2 10.9 25.3 13.9 0.0 YY EE 11.5 6.5 8.2 15.0 14.3 16/40= mmm mmm mmm me ee ee eee eee 4.6 2.2 4.7 6.1 1.8 16/80 === mm mmm mm mem meee 23.0 4.3 6.5 36.7 44.6 16/160===mmmmmmm mmm ee mee eee 18.4 4.3 6.5 27.2 37.5 1yisual acuity in Snellen notation. 22 Table 6. Visual acuity thresholds attained on Sight-Screener and Sloan Tests for distance vision without glasses Sight-Screener Tests Test part and threshold’ Sloan Test Total Instrument A | Instrument B Monocular tests Total number tested------=--===c----- 892 892 642 250 Percent distribution Total----=====cmeemmmmmcmm—mmmeeenn 100.0 100.0 100.0 100.0 20/20 or better--==-=-=====ce------- (1.04) 34.8 32.1 31.2 34.4 20/30=====---=c-eecmemmeemem——————no (0.7) 20.6 26.2 26.5 25.6 20/40====mcmmmecmm meme mmm oe (0.5) 9.6 9.3 9.5 8.8 20/50=====-mmcc-ecmemme meme mmm (0.4) 10.2 7.4 7+3 7.6 20/100-=======-c=cmccmmemmmm mmm momo (0.2) 15.0 18.5 17.8 20.4 20/200----====m-mem-eemmmm—me—————oo (0.1) 9.8 6.5 7:8 3.2 Binocular tests Total number tested----===c-c------- 460 460 332 128 Percent distribution Total-======memeeece meee ce mee —————- 100.0 100.0 100.0 100.0 20/20 or better----===-=-----m-m=--a- (1.0+) 45.2 46.6 46.7 46.1 20/30=========st commen mmm mmo (0.7) 21.1 18.9 17.8 21.9 20/40======--mmmmmmmm mmm mm mmmmmm mo (0.5) 10.9 9.1 9.6 z.8 20/50 === === mmm ————————— (0.4) 5.4 6.3 6.3 6.3 20/100-=======-ce-mmmem meme mmo o- (0.2) 11.7 16.1 15.4 18.0 20/200-==========--mmmmmmm—mm——momo (0.1) 5.7 3.0 4,2 oi lyisual acuity score in Snellen notation with decimal equivalent shown in parentheses. 23% Table 7. Visual acuity thresholds attained on Sight-Screener and Sloan Tests for near vision without glasses Sight-Screener Tests Test part and threshold’ Sloan Test Total Instrument A | Instrument B Monocular tests Total number tested=======-=ccceca--a 811 811 581 230 Percent distribution Total=s-emmeecc cece ccc ccm ccc ee 100.0 100.0 100.0 100.0 1.0 or better-=-==--=;m-ecccecccee——————- 20.8 23.7 23.8 23.5 EE att 17.0 16.2 15.7 17.4 0 5mm mm mmm me eee 9.9 7.4 8.4 4.8 0 4mmmm mm mmm meee meme m 5.4 6.8 7.2 5.7 JO 27.2 29.4 28.9 30.9 JO EE 19.7 16.5 16.0 17.8 Binocular tests Total number tested==-=======c--e---- 442 442 313 129 Percent distribution Total=mm=m=mec ccm cc ccc 100.0 100.0 100.0 100.0 1.0 or better-=====--e-eececceceec———————— 26.9 31.0 31.6 29.5 0.7mm mmm eee eee 16.5 11.3 11.5 10.9 tat 11.8 9.0 10.2 6.2 0ubmmmm mmm mmm meee meee een 4.8 6.3 7.0 4.7 J tt 23.3 33.1 31.9 35.7 Oulmmmmm mmm eee eee emcee eee 16.7 9.3 2.7 13.2 Lyisual acuity score shown just in the decimal equivalent of the Snellen fraction since the Snellen ratios differ for the two test series. 24 Table 8. Visual acuity thresholds attained on Sight-Screener and Sloan Tests for distance vision with glasses Test part and threshold! Sight-Screener Tests Sloan Test Total Instrument A | Instrument B Monocular tests Total number tested-==========ccee--- 485 485 342 143 Percent distribution Total-=====emeerece ccc meee c cme ema 100.0 100.0 100.0 100.0 20/20 or better========cm-eeccccccecece——— 56.7 51.6 50.3 54.5 20/30======mmmemeeecmemeeeeeee meee 24,7 29.3 28.1 32.2 20/40=====m=mccmme cece ecm ce cme — ema a 8.0 8:2 9.1 6.3 20/50=======--emme meee —— mmm 4,3 4.7 5¢3 3.5 20/100=======-=-meeemmee cme ————————— 3.7 5.6 6.7 2.8 20/200======mmmmcme meee eee mem em —————————— 2.5 0.6 0.6 0.7 Binocular tests Total number tested========e-ece-e-- 245 245 173 72 Percent distribution Total======e-ececcc ccm m ccc ccc ceca 100.0 100.0 100.0 100.0 20/20 or better-===-==--mecccccmcccnnaa——— 723.5 67.0 66.5 68.1 20/30=======mmemmeeeeeeeecee meme ————— aa 15.9 21.2 19.7 25.0 20/40==m==mmmmmmm meme mmm —————— “ PW 6.5 8.1 2.8 20/50======mmmmmem meme meme mm mee ——— ea 1.6 2.0 1.7 2.8 20/100-======mmemmememm eee mmm ———————— 2.0 3.3 4.0 1.4 20/200========cmemmmemmm——a——— wo 1.2 - - 2 visual acuity score in Snellen notation. 25 Table 9. Visual acuity thresholds attained with glasses on Sight-Screener and Sloan Tests for near vision Sight-Screener Tests Test part and threshold! Sloan Test Total Instrument A | Instrument B Monocular tests Total number tested------------ 535 535 382 153 Percent distribution Total-====meeem ccc ceeee ee 100.0 100.0 100.0 100.0 1.0 or better--------cecmomcmeo-- 32.5 37.5 37.7 36.6 0.7-mmmmmim meme eee 37.8 36.4 35.6 38.6 0.5-==mcmmcmcccemceeccemmeeme———— ma 12.5 10.8 10.7 11.1 SS 5.8 7.3 7.9 5.9 4 J 9.3 Zod 7.1 7.2 4 J ppp 2.1 0.9 1.0 0.7 Binocular tests Total number tested------------ 272 272 195 77 Percent distribution Total-=-=-=-cecemmmcccm meee 100.0 100.0 100.0 100.0 1.0 or better=-==----ecemeoccmanaonoon 47.1 53.6 55.4 49.4 UC) mmm 0 FE Th 36.4 32.4 29.7 39.0 ET TE 7.0 5.9 6.7 3.9 OY Lm mo sw 1.8 2,9 3.6 1.3 0. 2enccnannmmnn mmm mmm ——————— 7.4 4.8 4.1 6.5 Ou lemme meme eee eee 0.4 0.4 0.5 a Lvisual acuity score shown just in the decimal equivalent of the Snellen fraction since the Snellen ratios differ for the two test series. 26 Table 10. Distance visual acuity without glasses—mean, standard deviation, and correlation of Sight- Screener and Sloan scores for the eight study subgroups Naber Sight-Screener Sloan Test and subgroup tested in Correlation subgroup Standard Standard Meza deviation Mean deviation Monocular vision Far 1st, SS lst, SS-Inst. A----- 170 2.56 2.438 2.28 2.129 +0.87 Far lst, SS 1st, SS-Inst. B----- 57 2.95 2.218 2.88 2.227 0.79 Near 1st, SS lst, SS-Inst. A---- 179 2.79 2.829 2.47 2.334 0.84 Near 1st, SS 1st, SS-Inst. B---- 53 1.93 1555 2.07 2.15] 0.74 Far 1st, S1 lst, SS-Inst. A----- 145 2.47 2.170 2.32 2.320 0.77 Far lst, S1 lst, SS-Inst. B----- 69 2.21 2.047 2.30 2.116 0.80 Near 1st, S1 lst, SS-Inst. A---- 148 2.65 2.565 2.26 2.161 0.80 Near lst, S1 1st, SS-Inst. B---- 71 2.11 1.646 2.28 2.114 0.79 Binocular vision Far 1st, SS 1st, SS-Inst. A----- 87 2:23 2.197 2.00 2.062 0.89 Far 1st, SS lst, SS-Inst. B----- 28 2.35 1.687 1.91 1.327 0.87 Near 1st, SS lst, SS-Inst. A---- 921 2.05 2.092 1.85 2.315 0.70 Near 1st, SS lst, SS-Inst. B-=--- 27 1.76 1.275 1.67 1.950 0.57 Far 1st, S1 lst, SS-Inst. A----- 76 2.03 1.860 1.86 1.447 0.77 Far lst, S1 1st, SS-Inst. B----~- 37 1.85 1.390 2.09 2.255 0.64 Near lst, S1 lst, SS-Inst. A---- 78 2.33 2.262 1.96 1.810 0.85 Near 1st, S1 1st, SS-Inst. B---- 36 1.61 1.085 1.66 1.325 0.73 Scores given in minutes of visual angle subtended by target optotypes. The mean score of 2.56 minutes would be equivalent to 20/51.2 in the Snellen notation or 0.39 in the decimal notation. SS —Sight-Screener; Sl-Sloan. 27 Table 11. Near visual acuity without glasses—mean, standard deviation, and correlation of Sight-Screener and Sloan scores for the eight study subgroups! With &l Sight-Screener Sloan Test and subgroup tested in Correlation subgroup Standard Standard Moen deviation Mean deviation Monocular vision Far lst, SS 1st, SS-Inst. A----- 166 3.48 3.070 3.57 3.348 +0.88 Far lst, SS lst, SS-Inst. B----- 56 4.13 3.602 4.51 3.921 0.82 Near lst, SS lst, SS-Inst. A---- 162 4.27 3.409 3.78 3.234 0.82 Near 1st, SS lst, SS-Inst. B---- 51 4.55 3.591 4.00 3.413 0.77 Far lst, S1 lst, SS-Inst. A----- 145 3.94 3.240 3.84 3.264 0.81 Far 1st, S1 lst, SS-Inst. B----- 72 4.21 4.205 4.37 3.762 0.84 Near 1st, S1 lst, SS-Inst. A---- 137 4.24 3.447 4.01 3.400 0.75 Near 1st, S1 1st, SS-Inst. B---- 70 4,52 3.580 4.38 3.685 0.79 Binocular vision Far 1st, SS lst, SS-Inst. A----- 85 3.06 2.812 3.17 3.152 0.79 Far lst, SS 1st, SS-Inst. B----- 29 3.67 3.087 3.63 3.210 0.89 Near lst, SS lst, SS-Inst. A---- 86 3.30 2.532 2.84 2.160 0.81 Near 1st, SS lst, SS-Inst. B---- 28 4,23 3.275 4.07 3.767 0.66 Far lst, S1 lst, SS-Inst. A----- 77 2.74 2.217 3.21 3.232 0.69 Far 1st, S1 lst, SS-Inst. B----- 36 3.15 2.327 3.07 2.592 0.78 Near 1st, S1 lst, SS-Inst. A---- 75 3.70 3.245 3.55 3.267 0.81 Near 1st, S1 1st, SS-Inst. B---- 38 3.94 3.325 3.33 3.160 0.89 Igcores given in minutes of visual angle subtended by target optotypes. The mean Sight-Screener score of 3.48 minutes would be eauivalent to 14/48.7 in the Snellen notation or 0.29 in the decimal notation. The mean Sloan score of 3.57 minutes would be equivalent to 16/57.1 in the Snellen notation or 0.28 in the decimal notation. 28 Table 12. and Sloan scores for the eight study subgroups Distance visual acuity with glasses-—mean, standard deviation,and correlation of Sight-Screener Sight-Screener Sloan Number Test and subgroup tested in Correlation subgroup Standard Standard Mean deviation Mean deviation Monocular vision Far lst, SS lst, SS-Inst. A----- 77 1.64 1.393 1.72 2.086 +0.72 Far 1st, SS 1st, SS-Inst. B=-=--- 37 1.5) 1.361 1.46 1.525 0.89 Near lst, SS lst, SS-Inst. A---- 100 1.47 0.872 1.39 1.244 0.67 Near 1st, SS lst, SS-Inst. B---- 26 1.12 0.343 1.09 0.355 0.55 Far 1st, S1 1st, SS-Inst. A----- 67 1.51 0.809 1.56 1.286 0.78 Far 1st, S1 lst, SS-Inst. B----- 41 1.49 0.879 1.35 0.683 0.73 Near 1st, S1 1st, SS-Inst. A---- 98 1.44 1.087 1.45 1.398 0.90 Near 1st, S1 lst, SS-Inst. B---- 39 1.26 0.408 1.18 0.495 0.62 Binocular vision Far 1st, SS 1st, SS-Inst. A----- 40 1.23 0.652 1:52 1.982 0.54 Far lst, SS lst, SS-Inst. B----- 18 1.08 0.517 1.05 0.775 0.80 Near 1st, SS lst, SS-Inst. A---- 50 1.28 0.765 1.07 0.665 0.80 Near 1st, SS 1st, SS-Inst. B---- 13 0.98 0.262 0.92 0.210 0.74 Far lst, S1 lst, SS-Inst. A----- 34 1.31 0.805 1.16 0.585 0.83 Far 1st, S1 lst, SS-Inst. B==---- 21 1.28 0.702 1.13 0.512 0.70 Near 1st, S1 1st, SS-Inst. A---- 49 1.13 0.460 1.06 0.567 0.70 Near 1st, S1 1st, SS-Inst. B---- 20 1.10 0.317 0.90 0.217 0.42 IScores in minutes of visual angle subtended by target optotypes. 29 Table 13. Near visual acuity with glasses-—mean, standard deviation, and correlation and Sloan scores for the eight study subgroups! of Sight-Screener Sight-Screener Sloan Number Test and subgroup tested in Correlation subgroup Standard Standard Mean deviation Meg deviation Monocular vision Far lst, SS lst, SS-Inst. A----- 88 1.52 0.798 1.61 1.207 +0.54 Far lst, SS lst, SS-Inst. B----- 37 1.51 0.868 1.47 0.808 0.87 Near lst, SS lst, SS-Inst. A---- 110 1.91 1.455 1.82 1.479 0.77 Near lst, SS lst, SS-Inst. B---- 28 1.66 0.848 1.60 1.041 0.42 Far 1st, S1 lst, SS-Inst. A----- 83 1.69 1.230 1.69 1.056 0.58 Far lst, S1 1st, SS-Inst. B----- 43 1.74 1.454 1.81 1.225 0.60 Near 1st, S1 lst, SS-Inst. A---- 102 1.70 1.135 1.86 1.501 0.69 Near 1st, S1 lst, SS-Inst. B---- 45 1.60 1.107 1.79 1.110 0.87 Binocular vision Far lst, SS lst, SS-Inst. A----- 46 1.30 0.557 1.41 0.887 0.70 Far lst, SS lst, SS-Inst. B----- 18 1.36 0.675 1.24 0.762 0.81 Near 1st, SS lst, SS-Inst. A---- 55 1.54 1.045 1.37 0.800 0.68 Near lst, SS lst, SS-Inst. B---- 14 1.22 0.302 1.36 0.872 0.54 Far 1st, S1 lst, SS-Inst. A----- 43 1.41 0.927 1.49 1.410 0.72 Far 1st, S1 lst, SS-Inst. B----- 23 1.89 2.022 2,03 2.560 0.95 Near lst, S1 lst, SS-Inst. A---- 51 1.33 1.300 1.45 0.970 0.58 Near 1st, S1 lst, SS-Inst. B---- 23 1.49 0.935 1.43 0.742 0.87 IScores in minutes of visual angle subtended by target optotypes. 30 APPENDIX | TARGET-SPECIFICATIONS AND RECORD FORMS Specifications for the sizes of optotypes (width of the lines in the letters) and the number of letters of each size on the Sloan and the Sight-Screener targets used in the study are shown in Appendix I. The Sloan charts contain optotypes ranging in size from 10.00 to 0.65 minutes for distance and from 16.00 to 0.80 minutes for near vision. These form a series in which the steps are approximately equal on a log- arithmic scale with a gradation of 0.1 log unit. This means that the size of the letters of each successive line is approximately 26 percent larger than that of the following line. Slight deviations from an exact geo- metric progression are used to maintain relatively simple numbers for specifications in visual angle and Snellen notation. Sight-Screener targets contain optotypes ranging from 10.00 to 0.50 minutes for both near and distance vision. The first column in table I gives the visual angles in minutes subtended by the width of the lines in the letters of both distance and near test targets. The second column gives the decimal equivalent of the Snellen notation for letters in the targets of both tests. PH3-3767 SIGHT-SCREENER CALIBRATION STUDY 6-61 Sight -Screener data Examinee number Instrument WITHOUT GLASSES RED series (even) 9 8 7 6 5 I 3 2 1 200 100 70 50 40 30 20 15 10 R (2) E F Te DZEN BNEC EHZN FZEC TFEO ZNHT L(3)N F ET CDZE ZBNE NEHZ ECNF ETOF NTZH B (4) 2 E CE CNDZzZ EBNC ZHEC NEDF OEFT THNZ BLACK series (odd) R (2) E F TC DZEN BNEC EHZN FZEC TFEO ZNHT L (8) N F ET CDZE ZBNE NEHZ ECNF ETOF NT zH B (4) 2 E CE CNDZ EBNC ZHEC NEDF OEFT THNZ Tested by Figure 1. Record card for Sight-Screener scoring. The next four columns give the Snellen notation for the letters used for distance and for near vision on the Sight-Screener and on the Sloan tests. The number of letters at each level on the targets is shown in the last three columns of the table. Figures 1 and 2 contain samples of the record cards used for recording the test findings in the study. Two record cards per subject were used for the Sight- Screener tests—one for those without and one for those with glasses. Four cards per subject were used for the Sloan tests—one each for distance vision, uncorrected; near vision, uncorrected; distance vision, corrected; and near vision, corrected. As the test was adminis- tered, the examiner drew an oblique line through the letters that were misnamed. If a line (a block of letters on the Sight-Screener target) could not be attempted, he drew a horizontal line in the record card through the letters at that and any subsequent levels. Explanation of the scoring is contained in the section on "Descrip- tion of Tests and Controls." PHS-3785 SIGHT-SCREENER CALIBRATION STUDY. Sloan data. 6/61 Exam No. WITH/WITHOUT GLASSES FAR (even) (R) (13) K (12) DV(11) HC ZS (10)0RNKHS (9)DVRCKH (B)0ZNSHVCD(T)NRKCSZHVDO(6)SDKHNOCVRZ (5)HOCZRDSVN(4)KDVRHNZCOS (3)VRNHZDCSKO (2) HNORCZSVDK (L)NRCZHOVSDK (L (13)K (12)pV (11) HC ZS (10)ORNKHS (9DVRCKH (B0OZNSHVCD(TINRKCSZHVDO(6)SDKHNOCVRZ (5)H0CZRDSVN(MKDVRHNZCOS (3)VRNHZDCSKO (2JHINORCZSVDK(L)NRCZHOVSDK (B) (13)k (2) DV(11)HC ZS (W0JORNKHS (9)DVRCKH (BOZNSHVCD(T)INRKCSZHVDO(6)SDKHNOCVRZ (5)HOCZRDSVN(MKDVRHNZCOS(3)VRNHZDCSKO (2)JHINORCZSVDK(L)ONRCZHOVSDK 200 Tested by. Figure 2. Record card for scoring on Sloan distance test. 31 Table I. Specification of the sizes of letters and number of letters, Sight-Screener and Sloan Targets for testing distance and near visionl Snellen ratios for letter sizes used Number of letters at Visual angle Decimal cach level Ir oiooras equivalent Distance charts Near charts Sloan subtended at of Snellen Sight standard test ratio (Re- S BS distance (20 ft., | ©iProcal of | Sight- | Bight= | sn RN ctanee io ND visual Screener Screener San. istance | p; stance | Near 14 in., 16 in.) angle) (20 ft.) (20 fc.) (14 in) (16 in.) | and near .0625 FR. cee . ww 16/256 ¥ wn 2 wn 5 .0800 ve vee — 16/200 o “vie 6 .1000 20/200 20/200 14/140 16/160 1 ) 8 +1250 wa 20/160 iw 16/128 Sina 2 10 .1600 wi 20/125 aie 16/100 — 4 10 .2000 20/100 20/100 14/70 16/80 1 6 10 .2500 “ee 20/80 _., 16/64 “em 6 10 we BOF 20/70 “5 14/49 "i 2 ae +3333 wen 20/60 wie 16/48 ce 8 10 .4000 20/50 20/50 14/35 16/40 4 10 10 .5000 20/40 20/40 14/28 16/32 4 10 10 .6667 20/30 20/30 14/21 16/24 4 9 10 .8000 ces 20/25 “ew 16/20 "am 10 10 1.0000 20/20 20/20 14/14 16/16 4 10 10 1.2500 —_ 20/16 —_ 16/12.8 wi 10 10 1.3333 20/15 Ve 14/10.5 ew 4 Hay “ee 1.+35385 "ew 20/13 Hd wy 10 “sin 2.0000 20/10 ho 14/7 4 w . LAdapted from L. L. Sloan, “New Test Charts for the Measurement of Visual Acuity and Far and Near Distances,” American Journal of Oph- thalmology 48(6): 809, December, 1959. This is the size of the visual angle of resolution in minutes of arc subtended by the width of the lines in the test letters used at each threshold level. 32 APPENDIX II SOME TECHNICAL NOTES In this study visual acuity was measured at arbi- trarily selected points, determined by the size of the letters in the targets, along the continuum of possible letter sizes from 0.50 to 16.00 minutes of visual angle which the component lines of those letters would sub- tend when viewed from the standard distance by the normal eye. The points or levels at which measure- ments were taken differed for the two tests at all but six points, as indicated in Appendix I. It was assumed in the analysis of the study data that persons reaching a particular acuity level actually had acuities uniformly distributed over the interval between that level and the next higher level (the level with the next smaller size letters) measurable on the particular test. All analysis was done in terms of minutes of visual angle. Findings in the text tables have been converted into the reciprocal of the visual angle size (of the letters), called the ''decimal' nota- tion, or into the Snellen notation, since the latter two notations are so frequently used to express visual acuity. The Snellen fraction contains in the numerator the standard distance between the subject and the test target and in the denominator the distance at which the smallest letters discriminated (read) by the subject would be read by the '"mormal' eye. The decimal nota- tion is the decimal equivalent of the Snellen fraction. In testing the significance of the difference between mean scores in the Sight-Screener or the Sloan tests for different groups of examinees, the determination has been one of whether the two samples of examinees may be regarded as independent samples drawn from the same normal population, i.e., testing thehypothesis that the true difference between the mean scores is zero. If the above hypothesis is true, the variable n n, (n, +n 2) X~-3 / (n,- 1) 5 +(n,-1) 5,’ has the t-distribution with n +n, freedom. Here the mean and variance of the first sam- ple of size n_ are denoted by 1 2 n, +n, - 2 degrees of 1 n n _ 1 1 1 1 X=— YY x. and s?2 = i T(x. - %)?2 yp & X n,- Z while § and s,” are the corresponding characteristics of the second sample of size n,. When assessing the significance of the mean dif- ference between scores on the Sight-Screener and Sloan tests for the same group of persons, account has been taken of the correlation between scores onthe two tests, since these are in fact two highly correlated measures of the same characteristic for each person. In this case, the variable =(X=-V + - 2 u=(X-y) 8 8 Tey 2s Sy n has the t-distribution with n-1 degrees of freedom. In testing the hypothesis that two samples—the Sloan or the Sight-Screener scores for two different groups of examinees, or the Sloan and Sight-Screener scores for the same persons—are independent samples drawn from the same population, with respect to the visual acuity characteristic, the X i of homoge- neity was used. Here the hypothesis being tested is that in the two independent samples being compared there are r constants P, veees PL (for the r acuity levels) with Y p; = 1 such that the probability of a result be- longing to the i acuity level is equal to p; in both samples. In this case 2 1 % v¥ 1 1 X" =nn3 — n, n 5Lth\ RN o 2 2 2 (ny +n) : % n, © n ~ x, + n, +n n,n, I HY 2 with r-1 degrees of freedom. Here xX; and y; are the number of persons in the two series reaching the i h acuity level. The S-percent level was used for determining sig- nificance for all statistical tests unless otherwise in- dicated in the text, 33 U. S. GOVERNMENT PRINTING OFFICE : 1963 O - 688263 See inside of back cover for catalog card. Public Health Service Publication No. 1000-Series 2-No. 2 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington 25, D.C. - Price 45 cents NATIONAL CENTER| Series 2 For HEALTH STATISTICS | Number 2 VITALand HEALTH STATISTICS DATA EVALUATION AND METHODS RESEARCH Measurement of Personal Health Expenditures Development and testing of a brief questionnaire on family medical and dental expenditures for use in the Health Interview Survey. Washington, D.C. June 1963 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Anthony J. Celebrezze Luther L. Terry Secretary Surgeon General NATIONAL CENTER FOR HEALTH STATISTICS Forrest E, Linder, Ph.D., Director Theodore D, Woolsey, Assistant Director O, K. Sagen, Ph.D., Assistant Director U. S. NATIONAL HEALTH SURVEY Theodore D, Woolsey, Chief Alice M, Waterhouse, M.D,, Medical Advisor James E, Kelly, D.D.S., Dental Advisor Walt R. Simmons, Statistical Advisor Arthur J, McDowell, Chief, Health Examination Survey Philip S. Lawrence, Sc.D., Chief, Health Interview Survey Robert T, Little, Chief, Computer Applications CO-OPERATION OF THE NATIONAL OPINION RESEARCH CENTER AND THE UNIVERSITY OF CHICAGO Under legislation establishing the National Health Survey, the Public Health Service is authorized touse, insofar as possible, the services or facilities of other Federal, State, or private agencies. The methodological study in this report was performed under a contractual arrangement with the National Opinion Re- search Center, The University of Chicago. Public Health Service Publication No. 1000-Series 2-No. 2 The Health Interview Survey of the National Health Survey (NHS) is designed to collect infor- mation on illness, accidental injuries, a variety of measures of disability, and hospitalization experi- ence as well as the demographic and social char- acteristics of members of the families that come into the sample. The collection of data on personal health expenditures in a general purpose health interview such as this presents many problems. One difficulty is that a detailed interview on family health expenditures could make the entire interview too long and detract from the reliabil- ity of many items since the interview sometimes takes up to an hour and ranges over a variety of topics—both the regular topics mentioned above and special topics that are added from time to time. Furthermore, the kind of data needed on health expenditures includes as an essential part statistics on the frequency distribution of annual expenditures for families and persons. This means that the questions must cover expenditures for the entire year before the date of interview. Re- call of minor expenses for a period as long as a year is subject to much error. (Repeated visits to the same family would seem the obvious solu- tion, but the Health Interview Survey plan at the present time calls for interviewing each family only once.) Despite these problems, it was felt that the Health Interview Survey could make a contribu- tion to statistics on this topic if theobjective was limited to estimates of the total bill for all health expenditures for each person. Only major sub- items such as hospital care would be shown sep- arately, and these only if there was evidence in- dicating their validity. Emphasis in the analysis would be placed on families and individuals with large annual expenditures—the right-hand tail of the frequency distribution. For this and for break- downs of the population by age, sex, and resi- dence, the Health Interview Survey sample would have an advantage due to its relatively large size—roughly 10,000 different households each quarter. Consequently, the National Health Survey made a contract with the National Opinion Re- search Center (NORC) of the University of Chicago to develop a brief set of questions which could be used to classify families and individuals into broad classes according to the amount of their personal health expenditures. Working with the Bureau of the Census, which carries out the Health Interview Survey on behalf of the National Health Survey, NORC was then to test the ques- tionnaire in the field, NORC had previously con- ducted two detailed cost studies in collaboration with the Health Information Foundation,! and the results from these studies suggested that the methods used could serve as a good basis for further developmental work. The basic plan of the study was to use a cri- terion source of information with which to com- pare the yield of the abbreviated questions on ex- penditures. This was accomplished by a designin which the original health interview, which in- cluded the proposed new questions, was carried out in a sample of households, and a subsequent intensive interview was conducted in the same lodin W. Anderson, Patricia Collette, and Jacob J. Feldman, “Family Expenditure Patterns for Personal Health Services)* Health Information Foundation. Research Series 14, 1960. households. This latter interview was essentially like that previously used by NORC and it provided a detailed study of expenditures against which the shorter interview would be matched. An additional feature of the study was an in- vestigation of the relative effectiveness of secur- ing data by means of a direct interview and a self-enumeration questionnaire. The questionnaire that was tested incor- porated the proposals of NORC with certain modi- fications suggested by the staffs of the Bureau of the Census and the National Health Survey. The interview was carried out by Census interviewers under the supervision of that agency. The second or criterion interview was administered by NORC staff, and the analysis of results including a de- tailed report on the study was also the responsi- bility of NORC. For those methodological studies which the National Health Survey initiates but does not directly conduct, staff members are assigned to provide liaison with the research organization and to convey the viewpoint of the National Health Survey. For this study Mr. Elijah L.. White pro- vided the liaison, and he also edited the con- tractor's report for the present publication. Preface-==-=cemmm mmm ee ee eee I. Study Design===== === mmm me eee eee 1, INEroduction ===== === mmm ee eee eee Review of Previous Research on Medical Care COStS====mmmeeeaaaan Specifications for the Study-=======cc mmm Interviewing DesSign---==--==- comm mmm meee Brief Description of INtervViews==== === mmc m mmo eee Findings-=-=====cm mmm mmm ee eee Mean Values of Personal Health Expenses--==--ecoemmmomccmmccaaoo Net Frequency Distributions of Personal Health Expenses----====ac-- Total Health EXpenses=-=====c comme mm meee Hospital Overnight EEXpenses--==--coemmmmmm mmm Hospital —Not Overnight IXpenses--==-==-==cccmommmmmmmmmmeo DOCLOr EXPENSES === === == comme mmm eee eee Medicine EXpenses-=======-cmm momo momo Dental EXpenses==-==--=c comm mmo meee eee Other Health EXpenses-—==-==cc momo m mmm meee Gross Differences in Frequency Distributions of Personal Health EXpenses------mo oom m mmo eee Reporting of Personal Health Expenses in Relation to Personal Characteristics =—======mmm mmm mmo ee eee Analysis of Major Discrepancies-===-=-=eemmmmmm mca Overnight Hospitalizations=======cc common eee Not-Overnight Hospitalization----=====cccmmmmmm mmm DOCLOr EXPENSES ======= mmm moe Medicine EXpensSes=======-ce oom m cee eee Dental EXpenses-==--=—— comme eee All Other Health EXpenSes=======cm comme eee eee II. Conclusions and Recommendations=======m ccm m moomoo Appendix I. Questionnaires Used in Study-======mecmmmmmmmmcmeeeeeee oo A. Supplemental Questions Used in the National Health Survey--- NHS-4(a) Direct Interview === ooo e mom momo NHS-4(b) Self-Enumeration Interview=----=-ecococmmmmacaaoo B. Criterion Questionnaire Used by the National Opinion Research Center-====ceeemccccccccaaa meee em Main Criterion Questionnaire---===-ceecmmmmmcccccccceeeee Hospital Supplement===--== ccm mmm eee Major Condition Supplement---=====m momo meee Appendix II. Detailed Tables Showing Gross Differences in Reported Health EXpenses---=-ccoc mmm mmo aoc oo UG W Wr — fot | ft ft EE 11 11 17 17 17 27 27 31 31 31 31 31 35 36 36 37 38 36 40 46 51 53 SYMBOLS Data not available===m==mmmccmm cee eeeee Category not applicable--==-=-ececeeaaaao Quantity ZerO====m=mcmmmemmc cme —m—————— Quantity more than O but less than 0,05---- Figure does not meet standards of reliability or precision---------ccacaeao PERSONAL HEALTH EXPENDITURES The following research report was prepared by the National Opinion Research Center of the University of Chicago under contract with the National Health Survey Division, National Center for Health Statistics. Paul N. Borsky, Senior Study Director, and Jacob J. Feldman, Di- rector of Research, directed the project for NGRC and were responsible for the analysis and report presented here. Galen Glockel supervised the field work; J. Robert Banacki was in charge of data processing; and Harold Levy prepared the extensive tabulations. |. STUDY DESIGN INTRODUCTION In contemplating the regular collection of reports on personal medical outlays, the Public Health Service decided that it would like to se- cure the necessary information by adding a num- ber of supplementary questions to its continuing National Health Survey (NHS) questionnaire. The problem arose as to how many and what kinds of questions were required to achieve a reasonable level of validity. Presumably, the more detailed and precise the probing, the greater the accuracy of the expenditure estimates. For administrative reasons, however, and to avoid an overly lengthy and costly interview, it was necessary to try to minimize the number of different kinds of ques- tions and still maintain an acceptable level of overall validity. The object of this study was to experiment with different approaches in order to test the feasibility of alternative methods. REVIEW OF PREVIOUS RESEARCH ON MEDICAL CARE COSTS The first phase in planning this study was to review prior, relevant research. The National Opinion Research Center (NORC) had conducted two national studies of medical care costs, and in collaboration with Columbia University had conducted a special area study of medical costs, Other universities have conducted special stud- ies, and their experiences and reports were also evaluated. Special tabulations were prepared from the 1958 national study of NORC. As part of the 1958 sampling procedure, eightoverall summary ques- tions that made up the screening questionnaire were asked prior to the very detailed and lengthy personal interview. It was possible, thereby, to compare reports of total family medical care costs as revealed by the short-screening ques- tionnaire with the totals consolidated from the de- tailed interview. Such a comparison might sug- gest the efficacy of using a relatively small num- ber of questions in securing medical costs data. When only three class intervals were used, over 84 percent of all family reports of total med- ical expenses were the same for summary as for detailed interviews (table 1). About 10 percent of the summary reports of total family expenses were understatements, about 6 percent were over - statements. Even when as many as seven class intervals were used, about two-thirds of all summary and detailed interview reports were in agreement. The summary form resulted in understatement for about 23 percent of the families and inoverstate- ment for about one-half that percent (table 2). A detailed evaluation of the cases in which major discrepancies existed between the sum- mary and criterion forms revealed the following major problems in the summary form: Table 1. Comparison of summary and detailed interview reports of total family medical expenses, by three class intervals: NORC, 1958 Intervals of expenditure Intervals of expenditure Total Under $200 $200-499 $500+ Number | Percent || Number | Percent | Number | Percent | Number Percent Summary interview Detailed interview TOCalmmmmmmm nm -12,207 100.0 1,214 55.0 622 28.2 371 16.8 Under $200=-===--- mmmmmm———— | 1,289 58.3 1,138 51.4 143 6.5 8 0.4 S200%49 Dw mm mimirmim mmm sm mim om et ot er we v0 562 25.5 61 2.8 432 19.6 69 3.1 $500+-====mm- 0 356 16.2 15 0.8 47 2.1 294 13.3 Total-===-m==m=ee————- 2,207 100.0 Same class interval----- --=- 11,864 84.4 Summary less=-====-==--- —— 220 10.0 Summary greater------------ 123 5.6 } Table 2. Comparison of summary and detailed interview reports of total family medical expenses, by seven class intervals: NORC, 1958 Intervals of expenditure Intervals of expenditure Total Under $100 $100-199 $200-299 $300-399 $400-499 $500-999 $1,000+ Num- | Per- Num- { Per- | Num- | Per- | Num- | Per- | Num- | Per- | Num- | Per- | Num- | Per- | Num- | Per- ber cent ber cent | ber cent | ber cent | ber cent | ber cent | ber cent | ber cent Summary interview Detailed toggrviey Total=rmmmmmmm————— 2,207 | 100.0 750 | 34.0 | 464] 21.0 300 | 13.6 1971! 8.9 1251 5.7 268 | 12.1 103 | 4.7 Under $100-=--==-==-=-=--=- 861 33.0 683] 30.9 144 6.5 15 0.7 12] 0.6 2; o0.1 51 0.2 - - $100-199---=----=--cmunu- 428 19.4 45 2.1 266 | 12.1 82| 3.7 25 1:1 21 0.3 3| 0.1 - - $200-299--- 283 12.8 8| 0.4 37 1.7 157 7.1 54 2.4 13| 0.6 12] 0.5 2] 0.1 $300-399~ mmm mmmn——————— 187 8.5 6| 0.3 7] 0.3 28 1.3 84 3+8 42 1.9 20| 0.9 - - $400-499--mmmmmmmmmm————— 92 4.2 1 2( 0.1 6| 0.3 12| 0.6 36 1.6 35 1.6 - $.50099G mm www mm mm om mm mmm me 257 11.6 5 . 57 0:2 10| 0.4 9 0.4 24 1.1 171 7.8 33 $1,000+-mmmmmmmm———————— 29 4.5 2 31 0. 2 0.1 - 1{ 0,1 22 1.0 68 3.1 Total--=mmm=mmme——- 2,207 | 100.0 Same class interval------ 1,465 66.4 Summary less--=========-= 506 | 22.8 Summary greater 236 10.8 1. Inclusion of charges incurred prior to period covered by survey . Inclusion of free medical care . Omission of second or third hospitalization . Omission of pregnancy hospitalizations . Omission of anesthetist and surgeon fees . Omission of medical insurance payments . Omission of nonprescription drugs . Omission of routine medical examinations . Omission of nondoctor treatments and tests 10. Duplication of reported charges Using the experiences from the above analy- sis and information secured from other studies, a short interview form was developed and pre- tested in several versions. After consultation with staffs of the Bureau of the Census and the National Health Survey, final forms were de- veloped for the study. Ww No NO 00 NON UT SPECIFICATIONS FOR THE STUDY After detailed discussions, the following ad- ministrative specifications were established for the survey: 1. The period covered would be 1 year prior to the short interview 2. The short interview should require about 10-15 minutes on the average 3. The basic unit for recording medical charges would be the individual member of the household 4. Every member regularly residing in the household at the time of the short inter- view would be enumerated 5. Any adult member of an immediate family may respond for other members of the family 6. In cases of subfamilies or nonfamily mem- bers residing in a household, separate in- terviews would be secured on the detailed interviews 7. Information on total personal health ex- penses (not payments) incurred during the past year would be secured from totaling major components of expense 8. Information on medical care charges would be grouped into four or five class intervals 9. Persons covered by complete prepayment medical plans such as the Kaiser Health Plan in California and the Health Insur- ance Plan of Greater New York would be excluded from the study 10. To simulate eventual field conditions when the personal health expenditure questions would be incorporated in the National Health Survey, the regular health survey questions would be asked of all respond- ents participating in this special meth- odological study at the time of initial in- terview 11. To reduce costs and facilitate field work, a subsample from the urban part of the NHS sample would be selected in such a way that it overlapped NORC's national probability sample of areas INTERVIEWING DESIGN Since the Bureau of the Census was expected eventually to administer the medical care ex- penditure supplement, it was decided tohavereg- ular health interviewers conduct the current health interview survey and ask the short form questions. This would provide a realistic field test of the short questionnaire. NORC interview- ers would then conduct the longer detailed inter- view which would be used as the criterion of validity when comparing the responses with the short interview. So that no special efforts would be exerted which might bias the comparisons, the initial interviewers would not be informed of the followup NORC interview. Since self-enumeration has been found useful in other work of the National Health Survey, it was decided to split experimentally the sample of short interviews with half being directly interviewed and half being given a self-enumeration form. The actual field procedures were as follows: 1. A total of 442 households was selected from the regular NHS sample of primary sampling areas which overlapped with NORC areas. These health interviews were conducted during October 1960. It is important to note thatthe sampleofhouse- holds used in this study does not repre- sent a national sample, and the population included (mostly urban) is small. There- fore, the data on expenditures should not be used as estimates of national levels of expenditure. 2. Half of the sample households were se- lected at random and given the self-enu- meration form of the shortinterview—the form for this interview is included in Ap- pendix I. The form included a letter tothe respondent which briefly described the purposes of the personal health expendi- ture study and asked that the questionnaire be completed and mailed within 5 days to the Bureau of the Census. (A self-ad- dressed envelope requiring no postage was furnished.) Prior to filling out the form, the respondent was urged to consult other members of the family and tocheckavail- able records in order to make the report as accurate as possible, At the end of the regular health survey interview, the re- spondent was asked for his telephone num - ber and told that a callback might be necessary if the information was incom- plete. . The other half of thehouseholds was asked the same questions on personal health ex- penditures as part of the regular health interview survey. These respondents were not asked to check records or to consult with others in the family, and very few actually did. About a week following the health interview, each directly enumerated respondent was sent a letter stressingthe importance of more detailed information and asking him to check records and to consult other family members on a sug- gested list of medical cost items prior to a second interview, . About 3-4 weeks following the initial inter- view, NORC interviewers got in touch with the respondents who originally provided the data for both halves of the sample and arranged for a followup criterion inter- view, The initial contact was with the original respondent, but if told another member of the household was more knowledgeable, the NORC interview was arranged with the better informed person. In many instances, several members of the household contributed information to the NORC interview. If the self-enumeration form was not re- turned within a week or if the form was incomplete and failed to pass a review in the regional field office, a followup letter, telephone call, or personal contact was made. Of the 442 assigned families, 402 were subse- quently interviewed by NORC. In four cases, how- ever, interviewer reports strongly questioned the validity of the NORC interviews. The respondents were reported as hostile and uncooperative, and a careful review of the answers revealed such discrepancies and conflicts in answers thatit was decided to discard these cases from the analysis. Of all the cases assigned, 91 percent were interviewed by NORC and 90 percent were found to be usable and complete. Only 4 percent were refused or terminated before completion and for the remaining 5 percent, followup contacts were cancelled because of the time and cost involved. Approximately 10 percent of the families that belonged to the self-enumeration subsample and with which NORC had successfully conducted an interview had not, even after followup, returned their forms by mail. Instead, short-form data pertaining to their health expenses had been col- lected by phone or personal interview. However, because recourse to such nonresponse proced- ures would remain necessary in the event that self-enumeration was adopted by the National Health Survey for the collection of health expense information, in all the tabulations presented in this report the individuals in these anomalous families were treated as if their expenses had actually been reported initially through self-enu- meration, In a number of subsequent tables (3-10 and Appendix II), a category labeled "NHS self-enu- meration—failed edit'' appears. This category is composed of : (1) individuals from families whose mail-returned, self-enumeration forms originally failing edit because of missing or incomplete in- formation were passed after followup; (2) families for which the editorial followup was unsuccessful; (3) families that failed to return the self-enumera- tion form but were subsequently interviewed by phone or personal visit; and (4) certain other cases shown in the table below. Thus, the ''failed edit" rubric is somewhat of a misnomer. About two-fifths of the families in the category were not even subject to the normal editing procedure be- cause the information was collected from them by direct interview rather than self-enumeration. The category in question might well be viewed as ""'self-enumeration—problem cases," as can be seen from the detailed breakdown given below of response data for households originally inter- viewed by NHS and assigned for followup by NORC. Number Percent Total assigned---=====m=eemcemececncneeen= 442 100 Total not completed--==-==--mcecmceenennx 44 10 Refusals and breakoffs--- 16 4 Too costly to followup- 24 5 NORC data questionable- 4 1 Total completed-=====mmemcmmmecc cece een 398 90 NHS direct interviewS=--====-ecceccana- 193 43 NHS self-enumeration Returned self-enumeration formby mail Passed edit without followup-==-=---- 146 33 Passed edit after followup==-=------- 29 6 Failed edit—no successful followup 7 2 Did not return self-enumeration form by mail Information obtained by phone------ 16 4 Information obtained by personal interview------mmmmemce nnn nn ——— 4 1 Other Miscellaneous extraordinary cir- CUMSLANCeS====mmmmmmm ee —————————— 3 1 In all, detailed reports were obtained by NORC for 1,203 persons, of which short self- enumeration forms were initially secured from 617 and direct interviews were held with 586. BRIEF DESCRIPTION OF INTERVIEWS Six questions were asked in both of the short- form questionnaires with the actual language modified to fit self-enumeration and direct inter - viewing. A general definition of health expenditures was given on the self-enumeration form with emphasis on exclusion of free care. The exact dates for the year covered in the report were specified at the top of the form. The questions dealt with overnight hospitalizations (in this re- port overnight hospitalization refers to stays of 1 or more nights); not overnight hospitalizations (the great majority of these stays represents out- patient care); doctor charges; costs of medicine, prescriptions, and tonics; dental costs; and all other health expenses. Both short forms are in- cluded in Appendix I. The NORC criterion interview was different from that used in most attitude or opinion studies. In most ordinary interviews directed toward un- derstanding personal attitudes and opinions, care is taken to insure absolute privacy for individual interviews; sequences of questions are strictly maintained; questions are standardized; and con- flicts in answers are never challenged. Since the goal in this study was to obtain as complete and accurate a record of costs as possible, the pres- ence of and consultation with other family mem- bers was encouraged; conflicts in answers were questioned and eliminated; changes in order and language of questions were permitted; and prob- ing was directed to clarifying any ambiguities and inaccuracies in response. The NORC criterion interview started with questions about hospitalizations. If a person had been hospitalized during the past year, a special detailed supplement was filled out on hospital ex- penses for each hospitalization. Costs of room and board and other hospital charges were sep- arated from doctor charges before, during, and after hospitalization for a given condition. Ex- penses for medicine, tests, X-rays, and special equipment were some of the other items separate- ly probed. Then, a series of questions was asked about major illnesses, chronic conditions, and expensive illnesses. and a special supplement was completed for each reported major condition. The detail of this supplement paralleled the hospital forms and separated free care from costs levied on the respondent. Expenses paid by the respond- ent's health insurance or billed to the respondent or his family were covered by the study. Free care was defined as that paid for by workmen's compensation, nonprofit organizations, charitable or welfare groups, government welfare, and mil- itary or veteran agencies. Minor illnesses involving home and office visits were third among the items probed, fol- lowed by an enumeration of costs of medicine. A special question dealing with costs of eyeglasses was followed by detailed inquiries on costs of special appliances and equipment. Dental costs were covered by a series of six questions on types of possible treatment. A final question in- cluded charges of nondoctor personnel, such as chiropodists, chiropractors, or nurses. ll. FINDINGS MEAN VALUES OF PERSONAL HEALTH EXPENSES Overall, the total average personal health expense reported to NORC was $102.14 per per- son. The comparable total reported on the short interview forms was $105.13—only 3 percent more than the amount recorded on the detailed interview, The six separate components of per- sonal health expenses, however, were not as ac- curately reported on the short questionnaires. Hospital expense not involving overnight stays, as will be discussed later, was a frequently mis- understood category and was generally overstated in the short interview reports. Doctor and dental expenses were also generally overreported inthe short forms, offsetting somewhat under statements of overnight hospital stays, medicine, and all other health expenses. In evaluating percent dif- ferences, care must be taken to consider the av- erage amounts of each category of expense. For example, since the average expenditure for out- patient care reporied to NORC was under $2.00, an average difference of a small amount would be computed as a sizeable percent difference. Direct interviewing by NHS produced more accurate overall reporting of medical and dental expenditures. Offsetting differences in average component costs resulted in a chance reporting of identical average total costs on the short NHS in- terviews and NORC detailed interviews. The pat- tern of overstatements and understatements on component expenses, however, persisted with doctor, dental, and hospital-not-overnight ex- penses being overreported on NHS interviews and other health expenses being underreported. In general, NHS self-enumeration reports showed almost 6 percent higher average total health expenditures. A closer inspection of the data, however, reveals that self-enumeration re- ports which passed field edit for completeness of reporting were almost 4 percent less than NORC reports, while those which failed field edit were greatly overstated by 23 percent on the average. Of the 617 persons included on self-enumeration reports, 437 passed field edit and 180 did not. Table 3 presents these findings on average ex- penditures. NET FREQUENCY DISTRIBUTIONS OF PERSONAL HEALTH EXPENSES Total Health Expenses The offsetting of individual over statements by other understatements produced a remarkable similarity of net frequency distributions in per- sonal health expenses reported on both short forms used by NHS and the detailed interviews of NORC. On the average, NHS cost intervals were less than 1 percent different from NORC reports (table 4). Direct enumeration by NHS again reflected greater accuracy, but the average differences reported by self-enumeration were only 1 percent greater than direct interviewing. As expected, the self-enumeration reports that passed edit were almost as accurate as direct interviewing with the fail edit reports accounting for most of the NHS discrepancies. Hospital Overnight Expenses With only 9 percent of all persons reporting hospitalizations involving overnight stays, it was not surprising that the average difference in net frequency distributions for these hospitalization expenses was less than half of 1 percent. NHS direct interviews and NHS self-enumeration forms which passed edit were about the same inaverage net differences. The fail edit forms as expected were the least accurate (table 5). Hospital —Not-Overnight Expenses Only about 3 percent of all persons reported care at a hospital not requiring overnight stays. This item was often confused with doctor expenses and will be discussed with other discrepancies more fully in another section. On the average, net frequency distributions differed by approximately 2 percent with the direct interview distribution coming somewhat closer to the criterion than did the self-enumeration. These data are presented in table ©. Table 3. Reported mean values of personal health expanses reported to NHS and NORC, by type of 1 expenditure: 60 Type of expenditure Source and type of Hospital interview 1 All Total Doctor Medicine | Dental other : Not Overnight overnight Total NORC-=====mcmmmmmmc cme e ecm e $102.14 $22.53 $1.69 $28.89 $23.85 | $16.95 $8.23 NHS ===-=-mcmmmmmmmmcm meme emo 105,13 20.46 4.74 33.41 20.66 18.30 Z+36 Difference---=-====c-=ccauc-- +$2.99 -$2.07 +$3.05 | +%4.52 -$3.19 | +$1.35| -$0.67 Percent difference--------- 2.9 9.2 180.5 15.6 13.4 8.0 8.1 NHS direct interview NORC-=========mmmmmcmcmeae $100.49 $21.28 $0.95 30.95 $23.04 | $14.67 $9.60 NHS direct-----=-=--ecee--- 100.49 19.10 2,91 32.36 20.58 17.65 7.30 Difference---------=-=---- - -$2.18 +$1.96 [ +$1.41 -$2.46 | +$2.98 | -$1.70 Percent difference------- - 10.2 206.3 4.6 10.7 20.3 17.7 NHS self-enumeration NORC-=======mommmmmmmmmemo $103.71 $23.73 $2.39 | $26.94 $24.62 | $19.11 $6.92 NHS self--=-=---occencaaaao- 109.53 21:75 6.47 34.41 20.74 18,93 7.23 Difference-===-=====cu-u- +§5.82 -$1.98 +$4.08 | +$7.47 -$3.88 | -$0.18 +$0.31 Percent difference------- 5.6 8.3 170..7 27 «7 15.8 1.0 4.5 NHS self-enumer- ation—passed edit NORC-====m==-mmmmmm mmm e $95.37 $17.72 $2.28 | $24.21 $21.50 | $21.91 $7.75 NHS-==mm mmm mmm m meee em 92.00 13.70 4.33 27.50 18.98 21..47 6.02 Difference-=---========= -$3.37 -$4.02 +$2.05 | +$3.29 -$2.52 | -$0.44 | -$1.73 Percent difference----- 3+3 22.7 90.0 13.6 11.7 2,0 22.3 NHS self-enumer- ation—failed edit NORC=====mm mmm meme meee em $123.94 $38.31 $2.66 $33.57 832.19 | $12.31 $4.91 NHS- === meme mm cme meee 152.09 41.29 11.67 51.18 25.02 12.76 10.17 Difference------------- +$28.15 +$2.98 +$9.01 | +$17.61 -$7.17 | +80.45| +$5.26 Percent difference----- 22.7 7+8 338.7 52:5 22.3 3.7 107.1 Table 4. Net frequency distributions of total personal health expenses reported to NHS and NORC, by intervals of expenditure: 1960 NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents TOtaLmmmm mmm mmm mmm mmo moo oe eee 100.0 100.0 '0.9 Under $25mmmmmmmme mmm me rn mm mm mn mn nm mm nm nn 41.1 41.8 +0.7 $25-49-nmmmmmmenm mmr m mmm mmm mem mmm ——————————————— 18.0 16.2 -1.8 $50 m0 mm mmm mmm mm mm mm me 16.0 15.6 -0.4 $100-299==mmmmmmmmr erm m mmm mem mmm em em ———————————————— 16.5 17.3 +0.8 5 B00 oe mm 2 8.4 9.1 +0.7 Direct interview, 586 respondents TOtal=mmmmmmmm mmm mmm momen 100.0 100.0 11.0 Under $25-====m=rmrmecmcccc ccc eee eee eee ——————————————— 41.1 40.0 -1.1 §25=4T mmm mmm mm mm mm mn mm nn nm 18.3 17.0 -1.3 $5.50 m G0 rm mw am mm mm mm 2 2 2 2 2 2 16.0 17.6 +1.6 S100=299 mm mmm mm wm om om mm mm mm mm 2 0 te 16.2 16.9 +0.7 $300+mmmmm mmm mmm ———————————————————————————— 8.4 8.5 +0.1 Self-enumeration, 617 respondents TOtal=====m moomoo meee meme meee ee | 100.0 100.0 13.8 Under $25-=-==mm-mecceccecccc ec cc cece meee mmm ———— 41.0 43.4 +2.4 177 15.4 -2.3 16.1 13.8 -2.3 16.8 17.7 40.9 8.4 9,7 +1.3 TOE Lm mmm os wm mm mm a 1 0 100.0 100.0 11.2 Under $25-=-=mmmrmmmeccccc ccc cece eee eee — ee —————————— 39.6 40.7 +1.1 16.9 17:8 +0.9 17.8 14.7 -3.1 18.1 19.0 +0.9 7.6 7.8 +0.2 Total-==-==reemrccccc ccc cece ccc ccc cee ee ——————————— 100.0 100.0 14,0 Under $25-=mmmmemeccme ence —————————————————— 44.4 50.0 +5.6 19.4 9.5 -9.9 11.7 11.7 = 13.9 14.4 +0.5 10.6 14.4 +3.8 ] Total average difference. Table 5. intervals of expenditure: 1960 Net frequency distributions of hospital overnight expenses reported to NHS and NORC, by NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 100.0 100.0 10.4 91.0 90.4 -0.6 2.0 2.6 +0.6 2.7 3.2 +0.5 2.0 1:5 -0.5 2.3 2.3 - 100.0 100.0 10.6 91.6 90.9 -0.7 $1-99 1.5 2.1 +0.6 $100-199 2.7 3.4 +0.7 $200-299 2.3 1.6 -0.7 $300+ 1.9 2.0 +0.) 100.0 100.0 10.4 No expense 90.4 90.0 -0.4 $1-99 2.4 3.1 +0.7 $100-199 2.6 2.8 +0.2 $200-299 1.8 1.4 -0.4 $300+ 2.8 2.7 -0.1 100.0 100.0 10.6 No expense 9L.l 91.7 +0.6 $1-99 2.5 3.0 +0.5 $100-199 2.7 3.0 +0.3 $200-299 1.6 0.9 0.7 $300+ 2.1 1.4 -0.7 Self-enumeration—failed edit, 180 respondents TOtalem===m==-==mmm-————————————— mem m mm mmm ——————— 100.0 100.0 11.4 NO eXpenSe-=====-mmmemcmc ccc c ce — cc eee — eee — eee ——————————— 88.9 85.5 - 30h $1-99=m-mmmmeem emcee meee meme meme meee seme em ———————————— 2.2 34 +12 $100-199--=-==--emmmmcmm meee mmm eee meme meme emcee em ———— 2.2 2.9 +0.7 $200-299-====----mmeeem mmm m meee eee cme m meee mmm——e ee — 2.2 Zl -0.1 $300+-====-mmmmmee meee mem mm meme mmm e eee ee m————————— 4.5 6.1 +1.6 i : Total average difference. Table 6. Net frequency distributions of hospital——not overnight NORC, by intervals of expenditure: 1960 expenses reported to NHS and NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 100.0 100.0 11.9 %.7 91.9 -4.8 1.4 4.0 42.6 0.7 1.8 Jul 0.6 1.2 +0.6 0.6 1.1 +0.5 100.0 100.0 11.6 96.3 92.8 -3.53 2.4 4.6 +2.2 0.8 1.6 +0.8 0.5 - -0.5 - 1.0 +1.0 100.0 100.0 12.5 97.3 91.0 -6.3 § Lm 2mm mm mm me ee ee eee 0.5 3.4 +2.9 $25=40 mmm mmm mmm emma meme 0.5 2:1 +1.6 §50=99 = mm mmm mm ee ee ee ee eee 0.6 2.4 +1.8 § LOO == mm mm mm mm mm ee meme 1.1 ; - Self-enumeration—passed edit, 437 respondents TOtalmm mmm mmm mm mm ee ee meee 100.0 100.0 12.5 NO expense~========mmmmecc cece ——————————_—_————— 97.5 91.5 -6.0 § Lm 2m mmm mm ee ee eee 0.7 3.9 +3.2 § 25m mmm mm mm me ee eee - 1.8 +1.8 $50 m9 mmm mmm ee ee eee 0.7 1.9 +1+2 § 100+ mmm mmm mm eee mem 1.1 0.9 -0 .2 Self-enumeration—failed edit, 180 respondents Tota Lemmon mmm mm mm me ee ee meee 100.0. 100.0 12.8 NO expense--=-====mmm emo ——————————————————— 96.6 89.5 -7.1 § Lm 2mm mm mm ee ee eee - 2.2 +2.,2 $2549 mmm mm mm mm me eee 1.7 2.8 41.1 $50=99 mmm mmm me ee meme 0.6 3.9 +33 § L100 = mmm mm mm mm ee eee eee 1.1 1.6 +0.5 } Total average difference. 10 Doctor Expenses Over half of all persons reported some doctor expenses during the past year. On the average, total net frequency distributions varied only about 1 percent. It was interesting to note thatthe short interviews understated the small expenditures of less than $25 and tended to overstate slightly the larger categories of expenses. Direct NHS inter- views produced the greatest accuracy with theav- erage net frequency difference amounting to less than one-half of 1 percent, Data from interviews which failed edit again showed the greatest vari- ability as shown in table 7. Medicine Expenses Net frequency distributions reported on NHS interviews for medicine expenses were only a little over 1 percent different, on the average, from NORC interviews. Direct and self-enumera- tion interviews which passed edit were about the same, but those which failed edit were consistently less accurate. '"Zero'' expenditures were com- bined with those "under $25" because of an arti- fact in NORC reporting. In many cases, a family total for nonprescription medicine had to be ar- bitrarily divided evenly among family members by NORC. On NHS self-enumeration interviews this was not so; thus many more ''zero' expendi- ture discrepancies resulted (table 8). Dental Expenses Over one-third of all persons reported some dental expenses with the average net frequency interval difference between NHS and NORC inter- views amounting to only 1 percent. Only minor differences are noted in table 9 for the different types of NHS interviews. Other Health Expenses Other health expenses included suchitemsas eyeglasses, hearing aids, crutches, braces, and other medical appliances as well as costs for nurses, chiropractors, and other practitioners without a medical degree. On the average, net frequency distributions reported to NHS and NORC were within less than 2 percent of one another. Direct interviewing was more accurate, but the differences were small as can be seenin table 10. GROSS DIFFERENCES IN FREQUENCY DISTRIBUTIONS OF PERSONAL HEALTH EXPENSES Almost two-thirds of all persons reported total medical expenditures in the same class in- terval in both the short NHS interview and the more detailed NORC interview. Differences, as shown in table 11, were not significant between NHS direct interviews and self-enumeration, The precise magnitude of thediscrepancy be- tween the NHS and NORC interviews for a given category of expenses was quite arbitrary. This magnitude seemed to be a function of the number and size of the class intervals employed and the shape of the expenditures distribution. Since these factors varied, comparisons as to the relative accuracy of the reporting with regard to different types of expenses would not be fruitful. The level of agreement was artificially high for relatively infrequent types of expenses, like those for hospital care. Hence, the percents of coincident classification exclusive of the 'no ex- pense'' cell of the diagonal are presented in table 11. It is interesting to note that in table 12only 9 percent of all persons reported to NORC and NHS total expenses which were two or more class in- tervals different. About one-fourth of all persons reported only one class interval difference in total health expenses. Likewise, all categories except doctor expenses showed less than 9 percent of all persons reporting two or more class interval dif- ferences. Clearly, most of the discrepancies re- ported to NHS on the short interview were inac- curacies involving only one class interval dif- ference. Details on gross differences in reported health expenses for all six categories of expenses are shown in Appendix II. REPORTING OF PERSONAL HEALTH EXPENSES IN RELATION TO PERSONAL CHARACTERISTICS In the discussion on netand gross differences in reported frequency distributions of personal Table 7. Net frequency distributions of doctor expenses reported expenditure: 1960 to NHS and NORC, by intervals of NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 100.0 100.0 pr 44.3 44.5 +0.2 31.3 28.1 “3.2 9.3 10.7 +1.4 6.8 3 40.5 5.6 5.7 40,1 2.7 3.7 #41..0 100.0 100.0 '0.5 No expense-----=-===memeececcccc emcee ccc eee —————————— 40.8 40.0 -0.8 SLs oe ws is 32.4 32.1 -0.3 A 10.5 10.2 -0.3 50-99 =m mmm mm mmm em ee ee eee 7.0 7.9 +0.9 T0010 wom mm sm rs mm wp 6.6 6.6 : RE 2:7 3.2 +0.5 Total 100.0 100.0 2.0 No expense 47.8 48.8 +1.0 $1-24mmmmmmme em 30.3 24.3 -6.0 $25-49mmmmaauan 7:9 11.2 43.3 6.7 6.8 +0. 4.7 4.7 - 2.6 4,2 +1..6 100.0 100.0 11.7 46.9 49,2 +2.3 29.3 24.3 -5.0 10.1 12:1 +2.0 746 Z43 -0.1 4.1 4.6 +0.5 2.0 2.3 40.3 100.0 100.0 3.9 NO expense=====-=m-m-m oom eee eee em 50.0 47.7 -2.3 $1=2bmmmmmmm mm mm mmm mmm mmm mmm mmm meme mmm mmm mmm —————— 32.8 24.5 -8.3 825-49 mmm mm ee ee eee em 2.8 8.9 +6.1 TL 44 5.0 +0.6 S100 = 109m mmm mm mm 6.1 5.0 -1.1 $2004 == =m mmm mmm ee eee 3.9 8.9 +5.0 Iotal average difference Table 8. Net frequency distributions of medicine expenses reported to NHS of expenditure: 1960 and NORC, by intervals NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 1 Total---===-===mececrccre cece ce — eee eee —— eee ——————— 100.0 100.0 1.4 73.8 76.5 +2.7 14.4 11.8 -2.6 7 +6 7.5 -0.1 42 Heol - Totale--=-=====mmmemmmmmmemmmme mmm meemmeme meee oeon 100.0 100.0 y.3 Under $25---==--mccmmememmmec meee ccmcece ee meeeeeesmmmee ene 73.7 75.4 +144 $25=49=mmmmmm meme meme em emeeeee—oecmesooom = 15.0 12.8 2,2 G50 m0 = = = mm rr rm 6.5 7.4 +0.,9 SLO =m mmr rr rm mr 4.8 4.4 -0. Self-enumeration, 617 respondents Totalem=mmmmmmmmmmmmmmmmmmmmm mmm meee mmeemeooooe 100.0 100.0 '2.0 Under $25======--mmeocmecccc cece eee mme meee seeem—e——— 13.9 77.4 +343 $25-49----mmmmmmm mmm meme eee memes m meee ee mess ee————— 13.8 10.9 -2.9 $5009 = = mmm mn tm mm rr mn mm 8.8 7.6 at $100+===-==mmmmeem meee m mm meee eee meee mmm meses —————————— 3.5 4,1 +0.6 Self-enumeration—passed edit, 437 respondents TT or rt mm ee se 100.0 100.0 11.2 UNAEY $25 mmm som mom mem mm mm mm rm ar rm a 15.3 77.1 +1.8 $25-49=- mmc cnm mmm reer mmm ———— ee ———————————————— 13.3 11.5 -1.8 §50=99====-mmmme meme meme meee eee emmeeoomseme———— 8.9 8.2 -0.7 $100+---====memmmm mmm meee —————————— meme es ———————————— 2.5 3.2 +0,7 Self-enumeration-—-failed edit, 180 respondents TOtale======m=m=memmmem—m—eeeceeee—e—em———————————— 100.0 100.0 13.2 Under $25---==-=--c--emmmcmcmeee emcee ccm mememee—mmeem——o— 70.6 78.3 +7.7 $25-49-=-m=mmmmemmme memes meee eee eee eee em eem ee eeee——————— 15.0 945 «5,5 §50=99-====mmmmeem meme meme meee mmm eee eee emmeseses—————— 8.3 6.1 “2,2 $100+---====mmemrmememee cess meee meee meee seme esseeee—————— 6.1 6.1 - 1 Total average difference. Table 9. Net frequency distributions of dental expenses reported expenditure: 1960 to NHS and NORC, by intervalsof NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 100,0 100,0 4.3 6443 64.3 - 20.5 17.8 -2.7 6.3 8.7 +2.4 4.2 3.7 -0.5 4.7 5.5 +0.8 100.0 100.0 11.7 64.0 65.0 #+1:0 21.9 13.3 -3.6 Ged 7.8 1.5 4,4 348 -0.6 3.4 5.1 +1,7 100,0 100,0 1.3 64.5 63.7 -0.8 19.3 17.3 -2.0 63 9,6 +343 4.1 3.6 -0.5 5.8 5.8 - 100.0 100.0 }.7 63.4 62.9 -0.5 Gm 2m mm mmm me ee ee mmm een 17.8 15.6 -2.2 Rl EE 7.4 11.2 +3.5 $50=99 mmm mmm meme eee 5.0 3.4 -1.6 $100 =m mmm mm mmm me ee eee 6.4 6.9 +0..5 Self-enumeration failed edit, 180 respondents 100.0 100.0 '1.6 No expense 67.2 65.5 -1.7 $1-24 22.8 21.7 -1.1 $25-49 3.9 5.6 +107 $5099 mm mmm mm mm mm ee meee 17 3.9 +242 $100+=mm=mmmmmmm mmm mmm mmm mmm mm meme mme eee eee ———— 4b 3.3 -1.1 1 Total average difference. Table 10. Net frequency of other personal health expenses reported to NHS and NORC, by intervals of expenditure: 1960 NHS intervals of expenditure NORC NHS Difference All persons, 1,203 respondents 100,0 100,0 1.6 78.0 81.6 +3.6 9.5 6.3 -3,2 8.5 7.8 -0.7 3.1 3.3 +0.2 0.9 1.0 40.1 100.0 100.0 41.2 78.8 81.7 +2.9 8.7 6.5 =2,2 7.7 7.5 -0.2 3.4 2.1 -0.3 1.4 1.2 -0.2 100,0 100.0 19.0 77.3 81.5 +4.,2 10.2 6.2 -4,0 9.2 8.1 -1.1 2.8 3.4 +0.6 0.5 0.8 +0.3 100.0 100.0 19.5 75.9 82.2 +6.3 10.3 5.9 -4.4 9.6 8.7 -0.9 3.7 2.7 -1.0 0.5 0.5 - 100.0 100.0 12.2 80.6 80.0 -0.6 10.0 6.7 -3.3 8.3 6.7 -1.6 0.6 5+0 +44 0.5 1.6 +1.1 1 Total average difference. Table 11. Comparison of personal health expenses reported to NORC and NHS according to type of NHS interview and level of agreement, by type of expenditure: 1960 Type of expenditure Type of NHS interview and Hospital level of agreement Viodd Total ov Not Doctor : Dental | Other er- cine : over- night night Total Percent distribution Same class interval----==cccccccnmanaan- 63.2 95,3 92.1 62.6 75.4 81.1 84.2 NHS greater-=-=-=----e-ceccccecceceeenax 19.9 2.8 6.2 20.7 11.3 10.7 6.0 NHS leSS======m=m-mmcccc ccc ccc cece ————— 16.9 1.9 1.7 16.7 13.3 8.2 9.8 NHS direct interview Same class interval-------=-ccccceceeea- 63.8 295.9 92.9 62.8 15.5 80.6 85.3 NHS greater---=-====memeccceceecceceane- 20.1 2.4 5.2 20.2 11.6 10.9 4.7 NHS leSS=====mmmmmcm ccc ccm meme eee 16,1 1.7 1.9 17.0 12.9 8.5 10.0 NHS self-enumeration Same class interval--=-===-ceeccmcaaanaax 62.6 94.6 21.6 62.58 75:1 81.9 83.1 NHS greater---==----mmeccccccecce cee ———— 19.7 3.2 7:1 21.0 11.3 10.4 7-1 NHS leSS=====mmmmeceec ccc ccc cece eee 17.7 2.2 1.3 16.2 13.6 2:7 9.8 Adjustment for no expense all persons No expenses-NORC-NHS====-=====m-cmmamaanx Y 89.6 90.9 34.0 1 58.9 73.8 Same class interval with expenses------- 1 5.7 1.2 28.6 1 22.2 10.4 Different class interval with expenses-- 1 4.7 7.9 37.4 L 18.9 15.8 IRecause of the combination of the “no expense’ and “%1-24” class intervals, a meaningful adjustment is not possible for the “rats” ”” anc ‘medicine’ categories. Table 12. Percent of persons according to the number of class interval differences and level of agreement in personal health expenses reported to NHS and NORC, by type of expenditure: 1960 Type of expenditure Class interval differences Hospital and level of agreement Madi Total OvEE~ Not Doctor | J o Dental | Other night | over: Total Percent One class interval difference from NORC Total=-mrmrmeemer re ————————————— did 3.4 4,5 27.3 19.5 14.5 8.8 NHS greater----------c--eccemecmmaanannx 15.5 2,2 3.3 15.4 8.5 8.0 247 NHS less-======scmeommccmm emcee emem mmm 22,2 1,2 1,2 12.5 10.0 6.5 6.1 Two or more class interval differences from NORC Total====-=e-mme ccc cc cca cc cee — en m 9. 1.3 ub 9.5 6.1 4,4 2.0 NHS greater--=--=--meemeececcceccceece———— 4.4 0.6 2.9 i 2.8 2.7 ro NHS leSS-=====mmmmeeemeccc ccc eee —— a 4.7 0.7 0.5 4.2 3:3 1.7 3.7 health expenditures, the offsetting of overstate- ments by understatements of expenses was noted. While this tendency to balance errors in report- ing on the short interviews produced desirable overall accuracy, it is important in evaluating the expenses of different population groups to becer- tain that there are no systematic biases inreport- ing related characteristics to specific personal characteristics such as sex, age, or education. An analysis of these items which will be pre- sented in this section, revealed no serious biases related to these factors. The class intervals em- ployed for each of the expense categories in this analysis were the same as those which appear in the corresponding sections shown in Appendix II. Sex The degree of consistency between the class intervals of the personal health expenditures re- ported to NHS and those reported to NORC was generally about the same for males and females. For the NHS direct and self-enumeration inter- views combined, a difference in the level ofclass interval coincidence of as much as 5 percent oc- curred only in the case of expenses for doctor care (table 13). This difference was to some extent an artifact of the lower proportion of females who reported zero expenditures for doctor care. Age As can be seen from table 14, there was, in general, little systematic variation between indi- viduals of different ages in thedegreeofclass in- terval coincidence of the expenses reported for them. A higher level of utilization of the particu- lar type of service by one age group accounted for most exceptions. For instance, there was less agreement between the dental expenses reported to NORC and those reported to NHS for the 15-34 age group than for the other age groups. Thus in the 15-34 year age group, a lower proportion of persons reported zero dental expenses, and there was less probability of coincidence for the ''zero- zero'' category of reporting in the NHS and NORC interviews. One result for which there was no ready ex- planation was the exceptionally low coincidence level of the doctor expenses reported for theold- est age group in the self-enumeration sample. Since a disproportionate number of the discrepan- cies appeared to have come from the "problem families''—those who originally failed edit and those who failed to return the self-enumeration form and therefore had to be interviewed by tele- phone—the finding provided no basis for the im- provement of the self-enumeration form itself. It was not clear whether the respondents were bas- ically unreliable or whether questioning by the NHS supervisor of the original self-enumeration response led to an inflation of the expense figures (or greater accuracy than the NORC criterionin- terview figure). In any event, the discrepancy even for this deviant group did not appear to be so large as to present an insurmountable prob- lem. Because of the high level of interest in the health expenses for those persons 65 yearsofage or older, separate tabulations were made for the 55-64 year group andthe 65 and over group. These two groups did not appear to differ appreciably in their coincidence level. They have been combined in table 14 because of the extremely small num- bers of cases in each of these age-classintervals (112 persons, 55-64 years of age; 103 persons, 65 years and over). Education of Family Head Table 15 shows the comparisons between the personal health expenses reported to NHS and those reported to NORC, each individual being classified in terms of the educational attainment of the head of the family of which he was a mem- ber. The pattern of response coincidence was quite erratic—so much so that it defied a ready interpretation. The differences in coincidence level between subgroups were considerably larger when the classification was on the basis of the educational attainment of the head of the family than when it was on the basis of the individual's sex and age. This may have been due to the fact that the sam- pling variance of the differences between sub- groups was probably considerably greater when classified by a characteristic of the family head (or the family as a whole) than whenclassified by a characteristic of the individual. In the former case all the individuals in a given family, and therefore all the individuals for whom a givenin- formant reported, appeared in the same sub- 17 Table 13. Comparison of NORC and NHS personal health expenses according to type of expenditure and level of agreement,by sex: 1960 Type of expenditure and level of agreement Male Female TOTAL EXPENDITURE Total Number of respondents’ =======mo mmm om ooo eee SANE CLABES JANCGUNB ommimmiion on toto abd 00 HH 0 8 dE SE NHS | GA CQL wren oncom om msm om mom om m0 to 0 0 2000 20 000 0,2 8 0 0 A B00 oe NHS 1 QB www om woes om sm oom mm sm 1 0 om a 0 0 0 0 0 0 0 0 ot ott £0 0 NHS direct interview Number of respondents!-=--=--mccccmmom eee eee Same class InCelvalmmmmmsmm mimes wm i re a NES SGA EQN wr wn momo iim am om hw 0 0 0 a to NHS LE wim onion ss mm con om mn oo mt on om 1 0 0 0,0 eh fm NHS self-enumeration Number of respondents l-=--=coocom mmm eee eee Same ClaS8 LNCer VAL mre mmm me mo mmm mmm om mm on om om om om 0 om ss 0 0 0 em em me NHS Brea Cen =m mmm mmm mmm vm on ono on om mm on mm mm tm mm 2 tt mt NEIS. 1158 mr mow om mms sme cst mmm mom 0 0 0 00 0 00, 00, 0 0 HOSPITAL OVERNIGHT Total Same class interval---=-=--=ermmecccccccce cee m er — mm — em —————————————— INES G0 CQ mm miomssomi smn 1 om sm 1 1 0 2 0 1 0 A 100 0 a IHS! | G85 mwrvmmmrimwsm mmm msm msm tm om kt 0 a, 5 te mt NHS direct interview Same Class Interval mrmmmmr mmm mmm mmm mmm on 0 20 om om 0 mg mt NHS greater==-==smmrecmr ccc ccc ccc cc ——————————————————————————— IHS 1.0185 mew mem mms mom om os om m0 0 0 000 00 000 900 00 013010 30 0 0 et 0 NHS self-enumeration Same class Intervaleermmrrmmmmremem meee ea. NHS. Sl @ BE QL mmm ees mm own comm mem ow 20 00 0 0 0 0 00 00000 0 0 0 0 09.000 NHS 1 @SS mm mmm mm mom cmon om om mo om om om mm mm me HOSPITAL—NOT OVERNIGHT Total Same CLASSE INICEILVAL mmm minim inom mmm om om nm on nm mo on a mt NHS greater =rimmmmniehi=men nar A waa aR eS nA Ae EER eS Ra Si NHS LeSSr mmm mmo moo on om on om oo on om ot on am mm mm mm mm 2 0 0 0 0 NHS direct interview Same class interval=====sewannsmmnnm- 0 8 8 a 8 0 NHS. G0 Cm ommins mien mcm mk su oF ht oS 0 0 0 ho od Sh A tb NHS leSS-—==r=rmmrrmmc rr cc ccc ccc ccc; ce ———————————————————————— NHS self-enumeration Same Class INCeLVAL= mmm mmm mm mmm mmo mm oom mio om wom om mo ow i om 0 20 2 0 nm NHS greater----====-e-eececccccc ccc ccc ccc; ccc cece mee mmm mm ——————————— NHS leSS====-mmmcmc cmc ccc cmc ce ccc cmc ccc m cmc — -———— mmm me ————— Percent distribution 573 272 301 62.0 19.7 18.3 62.5 19.1 18.4 61.5 20.2 18.3 « oo » oe oo \O 00 W Oo Ww « oe wws oo « eo oOo ou « o © wws 630 312 318 64.7 21.2 14.1 63.5 19.2 17.3 WwW «+ © avowunm OOH « + NEN wie ® >. Ao HON «oe NWO © Ipercentages for each type of expenditure are based on the same number of respondents throughout this table. Table 13. Comparison of NORC and NHS personal health expenses according to type of expenditure and level of agreement, by sex: 1960—Con. Type of expenditure and level of agreement Male Female DOCTOR EXPENSES Percent distribution Total Same class interval---=-=eecemmcecccce esse eee eee ——————— mmm ———————— 65.1 60.5 NHS greater------emm-emececc cece cc; cece cece eee meme messes e—s—————————— 17.1 24.0 NHS leSS-=====mmm-memeeccc cece ccc c cede eee meme meme ee ee m——————— 17.8 15.5 NHS direct interview Same class interval==-===--memememececccccce cee cee eee emee—e— em mmm 64.7 61.2 NHS greater-=-==memmmeemmeecc ccc eee ceed eee m meee e mmm eemeeee——————— 15.1 24.7 NHS leSS=====mcmccmmmmm cme eee m emcee emcee mmmmommmo-——--oo 20.2 14.1 NHS self-enumeration Same class interval------===-sememeceeme cme mm mmm ememceceeme— oo 63.5 59.7 NHS greater--=-===m--mem emcee meee meee m mmm mmm mmm ——— 18.9 23.3 NHS LQ 08 weer em sam tm, 1 0 0 15.6 17.0 MEDICINE EXPENSES Total Same class interval----=-=---emeeccce ceed meee emcee eee meme eee ————— 75.6 74.9 NHS Srealer =m mm mm mmm mm om om on om om mm mn mm tc mn tn 0 ew 10.5 12,7 NHS leSS=======mm-mem mmm eee meee mmm em meemm em emmmmmm meee 13.9 12.4 NHS direct interview Same class interval--==-=-----ecmmmem cece e emma 76.9 74.7 NHS greater---===-m--m-m;ceccc ccc — cee eee eee eee eee e—e—e————————— 9.9 13.1 NHS leSS-=====mmm-memmec cece ccc ccc eee eee me meee mmm ———- 13,2 22, NHS self-enumeration Same class interval-=--=--m---emmemcmmccm cece cmcmcc cc ccemcoe meme 75.1 25:2 NHS greater-===-remmemrece ccc cece er; cece ee eee ee ———————————————————— 10.0 12.3 NHS leSS=====--c-cmcmcc mmm m meee meme eee eee cc—cco-—-oo 14.9 2.5 DENTAL EXPENSES Total Same class interval-=w---emeecemmeccecccc eee ee eee mm eee —————————————— 80.6 81.6 NHS greater-=====e---emome eee ccc emcee ccm me ecm —eemem memo m— memo 10.7 10.8 NHS leSS-=m===mmmmee cece ccc ce —; e-em meme mmm mee mm m= ————— 8.7 7+6 NHS direct interview Same class interval---====---mmececmcccmc ccm meee cmc ce ccc — 77.6 83.3 NHS greater-=---===seermeeecc;cc cs; ec; eee; e eee —— ee ————————————————— 12.5 9.3 NHS leSS-======mc comm meme eee mmm meme meme mmm mmm mmo 9.9 7.4 NHS self-enumeration Same class interval------=-cecmemmcc cme mmm mmm mcm mmm meme me 83.4 80.2 NHS greater--=====c---mo mec meee mmemcmcmeooo—-—e-— 9.0 11.9 NHS 1@E mmm mmm om om on on om onion 50 00 4m mm et 0 0 0 7.6 7.9 Table 13. Comparison of NORC and NHS personal health expenses and level of agreement,by sex: 1960—Con. according to type of expenditure Type of expenditure and level of agreement Male Female Same class interval NHS greater NHS leSSm mmm mm mm mmm mmm ee ee eee ee ee eee ee mmm meee Same class interval NHS greater NHS less Same class interval NHS greater NHS less OTHER HEALTH EXPENSES Percent distribution 20 Table 14. Comparison of NORC and NHS personal health expenses according and level of agreement, by age: 1960 to type of expenditure Type of expenditure and level of agreement Under 15 15-34 35-54 5 5+ TOTAL EXPENDITURE Total Number of respondentsl-===mm mmc oe Same class interval--==--ccccco mmm NHS greater--=--==-cccccc mcm em NHS lesS=====mmmemmoc mcm ccc m een NHS direct interview Number of respondents! ====== moomoo Same class interval------=-c-m mmo NHS greater-----==-==-- mmc cmc ccc ccc NHS LOSS crm car ot mre mm 0 hf i 00 NHS self-enumeration Number of respondents! ---=ce oom m momo oo Same class interval--=-=---=-m ooo eee NHS greater----=---m-comm eco cece eee ee em NHS leSS======== === momo eee meee HOSPITAL OVERNIGHT Total SAME CLABES LNCELVAL mmm mmm mm omni mt mm tm mm a NHS (SORE amor mein cos vw ci 3900100 00 490 0 0 0 i 00 0 NHS legge mmmmmmmaimsimmmaiinion mins omisinieie sees meee ssw onesies NHS direct interview Same Class INCELVAL = rm mm mmm mmm mm mmm mm mm mmm mmm mo ot ot mi mm co om NHS Beata r= mim mms oti mim mmm mim mi sion) 0 snot od i om tin 0 0 0 00 tn 10 0 NUS 1 @88 wm mim minions ion sm iim m/s 0m 0 0 8 0 0m 0 60 2 0 0 8 NHS self-enumeration Same Class INCEYVAL =m mmm immo mm mmm mmm mmm um mmm mmm ob im mm om om mon mm NHS Br @a lar mimimm mmm minors sm on isi om mt mh 0 6 mo 0 te mo NHS legg~==merremms=rrinssewmns@idnannsnicaannesaanesamnses san HOSPITAL—NOT OVERNIGHT Total Same Class INLET Val == mm mmm mmm mmm mim mm im mom om om on on 0 on on 0 0 0 0 2 0 0 NES GLOBE mirmivmion win m0 m0 mm i 0 a 0 0 0 0 0 NUS LEGG mm mmmiomion'on mmm on om anim mm. mm. 0 1 0 0 000A 0 A A 0 0 A 0 90 100 NHS direct interview Same class interval-----==--emmececccccec ccc eee — ee ————— NHS grealar=mmrmmmm mmm mm mmm mim mm mmm om om om om mm 0 0 0 tm 0m 50 0 mm we NHS leSS=====-mmrmmecccc cece ccc; cece; cece, ————————————— NHS self-enumeration Same class interval====--m-ccmeccmce cme NHS S08 EG mmm own 0 mm mm om ow a 0 mm so NHS leSS~===-mmmmcmrme ccc ccc ccc ccc; ccc ccc ccm, ———————————— Percent distribution 385 63.9 19:7 16.4 194 65.5 19.6 14.9 191 62.3 19.9 17.8 «ee “oe U1 oo avon ANN oo «eo aA W uN Ww eo ANN 268 63.1 18.6 18.3 128 66.4 20.3 13.3 140 60.0 17.1 22.9 =o “oe « on ano ONO HOO = \O 00 OW 00 335 62.7 21.8 15.5 162 61.1 22.8 16.1 1723 64.2 20.8 15.0 HUN “eo. PEE oo 0 Ow wn © wor NOR «eo ~NEN 215 115 62.3 19.1 18.6 Ipercentages for each type of expenditure are based on the same number of respondents throughout this table. 2) Table 14, Comparison of NORC and NHS personal health expenses according to type of expenditure and level of agreement,by age: 1960—Con. ‘ Under Type of expenditure and level of agreement 15 15-34 | 35-54 55+ DOCTOR EXPENSES ee Percent distribution Total Same class interval-----=--=-memccccccceccc cence ——————— 63.1 62.3 64.2 59.5 NHS greater===== mmo ooo on eee eee eee eee eee m meen me 21.8 22.0 18.2 21.9 NHS leSS==m==mmmmmce ccc ccc ecm em mee —————— 15,1 15.7 17.6 18.6 NHS direct interview Same class interval----=-----eeeceececcccccccccc cece ————- 61.9 61.7 63.0 66.0 NHS greater----=---=--cceccccmccceccccccccc ccc — 24.7 22.7 16.7 14.0 NHS leSS=== moomoo mmc cme e emcee mm meme men 13.4 15.6 20.3 20.0 NHS self-enumeration Same class interval----=-=---ecccccccccccccccc ccna 64.4 63.6 65.3 33, NHS greater=====--c--eecemc cence ccc ccc ce cc mce meee mee 18.8 20.0 19.7 28.7 NHS leSS=======--memecccccec ccc ccc cc cc cmc ccc m ccm e cece 16.8 16.4 15.0 17.4 MEDICINE EXPENSES Total Same class interval----=--==ee-mcccccmccce ccc cmc cee ———— 76.1 71.2 77.6 68.4 NHS greater---=-==--cccmccemc cece ccc ccc cecee cece meee ————— 10.6 9.0 12.2 14.4 NHS leSS=====-mmemmccm cece ccm c cece cece ne ne 13.3 13.8 10.2 17.2 NHS direct interview Same class interval----=-=-=-e-ececccccccccccc ecm ———— 74.2 83.6 17+2 66.0 NHS greater-----------eeecccmccccccccec cece ccc cee m————— 12.4 8.6 11.1 15.0 NHS leSS===-==-cmmmm eee mm mmm ————— 13.4 7.8 11.7 19.0 NHS self-enumeration Same class interval------=--meeececcccccme cece ————— 78.0 71.4 78.0 70.4 NHS greater-=--=-m-meec cm ee cece eee ec meme em ————— 8.9 9:43 13.3 13.9 NHS leSS====mmmmmem meee ccc cee m meee eee ———— 13.1 19.3 8.7 15.7 DENTAL EXPENSES Total Same class interval-=====-eeeeoecmmm eee eee eee 84.9 71:3 84.5 81.4 NHS greater----==mmmmmecec;ccccc; cc; ; ccc ccc; —— eee ee —————————— 7.0 16.4 9.8 11.6 NHS leSS===mmmmemmrercccc ccc ccc; ;; cc; — eee sem ————————————— 8.1 12.3 5.7 7.0 NHS direct interview SAME CLASS INTELVAL mmr mms own somo mm mm sm tt am tw 86.6 71.1 81.5 79.0 NHS greater---eremrmemeesre cece ——————————————————————————— Tv 16.4 9.2 13.0 NHS leSS-mmrmmmrmmmc ccc ccc; —— ccc ——— eee —————————————————— 547 12.5 9.3 8.0 NHS self-enumeration Same CLASS INUCETVAL mmm wm mmm mmm mmm im me ow or me ow mm mn mm 0 om. om 83.2 71.4 87.3 83.5 NHS Greater = mmm mmm mmm mmm ms mom mm mm mt om 2m 2 0 0 40 0 a 6.3 16.4 10.4 10.4 NHS leSS--====mm-memccccccc emcee cece ccc c cece. ————— 10.5 12,2 2.3 6.1 22 Table 14. Comparison of NORC and NHS personal health expenses according to type of expenditure and level of agreement,by age: 1960—Con. Type of expenditure and level of agreement Unger 15-34 | 35-54 55+ OTHER HEALTH EXPENSES Percent distribution Total Same class interval==-=---===-=--eme-mmoeocco--o-ooemmemmomo—oooo= 90.4 85.1 78.5 79.5 NHS Ga @ 1 = wim emer ase mes om wm mm mm, me, 8 5 et ee 3.1 5.6 8.7 7+9 NHS leSS===mm=mmmmmmcmc cece emm cme me ooo o-ooee—me—oo——ooooo-- 6.5 9.3 12.8 12.6 NHS direct interview Same class interval 91.8 83.6 79.0 84.0 NHS greater--=-==-==-==--ee----ccec===--- 2.0 6.2 8.0 3.0 NHS leSS===m=mmmmmmm-mm mo mee meee mmmm— ooo memos -o-—--——--o- 6.2 10.2 13.0 13.0 Same class interval----=------c-mmememmommccocoooo-oo-o-oo—oo- 89.0 86.4 78.1 75.7 NHS greater---------=---mmmmommoo ceo ceco—-——mo—ooo—ooooo 4.2 5.0 9.2 12.2 NHS leSS==m=mmmmmc—mmmm meee meme ccc ee—ome--osoo—ee-———o-o- 6.8 8.6 12.7 12,1 23 Table 15. Comparison of NORC and NHS reports of personal health expenses according to type of expenditure and level of agreement, by education of family head: 1960 Type of expenditure and level of agreement Elementary High school College TOTAL EXPENDITURE Total Number of respondents! --==m-mmoomo mmo Same class interval--=----=--cmmccm cece eeeee een NHS greater---=----=--c-cccccccc ccc ccc ccc —————— NHS lesSs=====mm mmm mmo cece ce eee eee eee eee ee ee NHS direct interview Number of respondents! ====m moomoo mmol Same class interval---ece cea mm cee NHS greater-------==-==-meccemceese cece cece cece cea ———— NHS lesg--==--=--==memememeeece ee m— cee cee c eee ee. ————— NHS self-enumeration Number of respondents! ====mmmmommoo ome Same class interval------=--c-memeec cee eceecceneeeea eo NHS greater--------=-----cccccccmcc ccc ccccccme cea NHS lesS§=======mmmmm emcee cee cece eee meme eee HOSPITAL OVERNIGHT Total Same class interval--=-=--=--cccmcce eee emeeeeeeee oo NHS greater====-=-crceecccccccamnmnnsnem rer nn nnn meee -—— NHS leSS=====m-mmmm mmo cece meee e cee meee eem emma n NHS direct interview Same class interval--=----cccccmc cme eeeee em NHS greater----=-=---ce-ecc ccc ccc cc mm em em eee mmm J SE EE NHS self-enumeration Same class intervale-=---e-ecccmcecccaceeeeemmcaccaaaaa- NHS greater-=-===----m-c cece cece meee cee cee memeeem NHS leSS====m mmm meme eee eee ee eer ee HOSPITAL—NOT OVERNIGHT Total Same class interval--=---==me-cccmo cece een NHS Breater «=< =m mmm mimi mmm mmm mom mom mm on mo on mm NHS lesS=--======--crocm mcm ccc cece meme meena NHS direct interview Same class interval---=------cecmccc cece eee NHS greater====--======mecccccacccnem mmr ener nen nee -- NHS les§--=--=---=m-mmcceccccccc ccc ccc ccc cece ———————— NHS self-enumeration Same class interval------=m--cecmccc meee NHS Sraator === === mm mmm ummm mim mw mimo om mn 0 0 0 0 NHS lesSS=======mm-mm mmc ecm meme mm eee e en Percent distribution 426 206 220 516 62.6 20.3 17.1 252 264 57. 19. ww Hn ono + 00 NWS wHw UN Huw ano ER Huw HW 229 110 119 61.6 2nd aN O sw SN o oun ws sos © wo Ipercentages for each type of expenditure are based on the same number of respondents throughout this table. Number of respondents in this table excludes those where education of head was unknown. 24 Table 15. Comparison of NORC and NHS reports of personal health expenses according to type of expenditure and level of agreement, by education of family head: 1960—Con. Type of expenditure and level of agreement Elementary | High school College DOCTOR EXPENSES Percent distribution Total Same CLOSE LTIEET VA Lymm ermimiuiioss wm ohooh eh ew 66.7 64.0 52.0 NHS (STC 0 CE wm wm wm sm own mm moat mom to 3 90 at 0 17.6 21.3 24.4 NHS Lamm coe cow cw comm om mm mmm mm mm om mmm to 0 15.7 14.5 23.6 NHS direct interview SANE CLASSE LATENT mmr oo ms 6 i a 0 5 67.0 67.5 46.4 INES (BTC AEET wm mmm mm tm stm 1 ew 8 0 0 0 00 16.3 19.0 30.0 NHS leSS========m-mmmememcemceeeceecemcememecemm————————— 16.5 13.5 23.6 NHS self-enumeration Same class interval 66.4 60.6 59.7 NHS greater---------- 18.6 23.9 19.3 NHS less-=wmrmmewremmemee renee mmr ————— 15.0 15.5 21.0 MEDICINE EXPENSES Total Same CLASSE IVEY VA Lummi mivsimimsimim mis mi mie nfo 6m A we 0 0 76.3 73.0 77.7 NHS S002 Le www mom wn mm mt 0 11.7 12.8 2.3 IHS Lo wwe mmm om mmm mm mm om mt mm 0 0 0 12.0 14.2 14.0 NHS direct interview Same class inCerval-—====~==scc asc A 70.4 79.8 77.3 NHS STC AEE ww rm mm mmm om mm sm 2 a 0 0 0 0 0 0 0 14.6 11.5 6.4 NHS less-rmrrm=rrmmmeemmm meee rrr rr 15.0 8.7 16.3 NHS self-enumeration SAME CLABE AIICOIVE Lv mmm mm me mrimy moomoo mmo ed 0 81.8 66.7 78.2 NHS greater 0 14.0 10.1 NHS losgwem mm mmm mmm mmm rm mm crm cr cr ro 9.1 19.3 11:7 otal Same Class INErval= mmm smimmmmm ims im mmiominise mim 81.9 82.6 74.2 NHS: Se OEY iow m im mmm i 10 0 ee i a A 10.8 Gel 15.3 WHS: L080 ws comms am mo mm cm er 0 7.3 8.3 10,5 NHS direct interview Same class Interval===rarmme sims me i ————————. 78.6 86.5 68.2 NHS: SX QL mies: imtosticn fish fe i 0 0 e000 11.2 6.3 21.8 NHS, Lome mm om mim on ons mm mn mm em mt 10.2 7.2 10.0 NHS self-enumeration Same CLASS LICEE YA Lrmmimmm mim sh m mimo eo ait 85.0 78.8 79.8 NHS SCTE mrmomwmcs mom ssimmim sm t te 0 t 0 0 10.5 Al.7 9.3 IHS! 0/8 im mmm mmm cm own mm mcm 0, 0 4.5 9.5 10.9 25 Table 15. Comparison of NORC and NHS reports of personal health expenses according to type of expenditure and level of agreement, by education of family head: 1960—Con. Type of expenditure and level of agreement Elementary | High school College OTHER HEALTH EXPENSES Percent distribution Total Same class Interval imme sive wis ues swmin mw mi ————— 84.3 84.3 83.4 NHS Sraalar == mm mmm mom mm mm mm mw om mm om mn 0 2 2 0 2 2 4.9 6.0 1.9 NHS LS mm mm mom mm mmm oo om mm a mt 2 2 0 00 0 0 2 0 0 0 0 10.8 9.7 8.7 NHS direct interview Same class interval-========r=memmmmmm—— em ————————————————— 84.9 85.7 84.5 NHS greater-----------=--------meom momen mcmmce eee 4.9 4.0 5.5 NHS les§---===-emmrcmeecceccccccce ccc; ccc eee ———— 10.2 10.3 10.0 NHS self-enumeration Same class interval--==mmmmm =e e—————— an vn nm on oe 0 om 0 on 83.6 83.0 82.4 NHS Graber = mm mmm mmm om mw om om om mw am mm a 2 m0 2 om ta 3.0 8.0 10.1 NHS Lo Gm mm mim mmm mim sm ow mm 0 a a 0 11.4 9.0 7.5 26 group, while in the latter case they were likely to appear in several subgroups. This hypothesis, of course, suggests a substantial positive intraclass correlation of reporting discrepancies within fam- ilies, a not too unlikely possibility. Among the unexplainable results is the ex- treme unreliability of the NHS direct interview reports of doctor and dentist expenses for indi- viduals in families headed by someone who has completed at least 1 year of college. While this was in part an artifact of the more widespread use of physician and dental services by persons in families headed by individuals with a college education than by persons in families headed by someone with less education, this factor could hardly account for the entire discrepancy. If it did, one would expect a similar pattern of devia- tion in the self-enumeration cases, however, this did not occur. As unsatisfying as such anexplana- tion is, it would appear most prudent to view the poor performance of the college families on the NHS direct interview as a sampling anomaly rather than as a substantive finding which neces- sitates special modificationof the directinterview form. ANALYSIS OF MAJOR DISCREPANCIES For the purposes of detailed analysis, it was desirable to isolate a subgroup of individuals for which the two sets of expense reports were par- ticularly inconsistent. It was felt that an exami- nation of such cases would provide leads as to how the short interview forms might be improved. A "major discrepancy'' in any particular cat- egory of expense was defined as a difference of two or more class intervals between the NHS and NORC interviews. In addition, differences of $100 or more within the highest (open-ended) class in- terval were also treated as major discrepancies. Those individuals for whom there was atleastone major discrepancy were drawn into the special analysis. Many individuals had several major discrepancies—one with respect tothe grand total and one with respect to each of several different expense components, There were other individuals, of course, for whom there were major discrep- ancies with respect to two or more expense com- ponents but no discrepancy with respect to total expenses due to the compensating character of the errors. Table 16 presents a distribution of the 348 cases which involved at least one category with a major discrepancy. The 170 cases in table 17 are those whose total expenditures reported to NHS were higher than those reported to NORC by two intervals or $100 more in the open-ended inter- val. Similarly, the 161 cases distributed in table 18 reported lower expenses to NHS than to NORC. Also shown in table 18 are 17 cases with total ex- penses reported in the same interval in both in- terviews but with a major discrepancy in at least two categories that counterbalanced each other in the total expense category. Considering only the discrepancy cases re- porting greater expenditures to NHS (table 17), the categories that contained the greater amount of overstatements were doctor expenses (67 per- cent), medicine costs (44 percent), and other health expenses (21 percent). The patterns were the same for direct interviewing and for self- enumeration. Of the discrepancy cases reporting less ex- penditure to NHS (table 18), the leading categories were medicine costs (59 percent), doctor ex- penses (46 percent), and other health expenses (30 percent). As inthe overstatement group above, no large differences were observed between the patterns of discrepancy in direct interview and self-enumeration reports. A detailed examination of each category of expenditure provides some insight into the reason for these discrepancies. In general, however, the overstatements reflected exaggerated estimates given in response to the abbreviated question- naire, while the under statements were attributable to omissions of items which were elicited in the more detailed questions. Each of the 348 discrep- ant cases was individually examined to determine the reasons for NHS-NORC differences and each component of health expense is analyzed sep- arately below. Overnight Hospitalizations Of the overstatements for overnight hospital expenses in the short interviews, exactly one-half of the major discrepant cases were apparently due to general exaggeration of charges. Thesein- 27 Table 16. Distribution of all major discrepancy cases according to type of expenditure and level of agreement, by type of interview: 1960 Type of interview Type of expenditure and level of agreement Total NHS direct as Self Number | Percent Number | Percent | Number | Percent Total expenditure Total ====--mmmm ccc eee 348 100 155 100 193 100 NHS greater-------=---commmeocmmme meme m 170 49 78 50 92 48 NHS leSS==mmmmmm mm mmm ee eee ee eee em 161 46 71 46 90 46 Same class interval--=-==-=-ccocmooommonoooo 17 5 6 4 11 6 Hospital overnight NHS greater---=-=--ecccmcme cee eee 32 9 12 8 20 11 NHS leS§==mmm mm mmm mmm eee eee 28 8 12 8 16 8 Same class interval--==-===-=cecccommoaoaooo 21 6 7 4 14 7 No expense----=-=-===mccccmmcocen cece ——— 267 77 124 80 143 74 Hospital—not overnight NHS greater--------=-mecmo meme meeemee 43 12 13 8 30 16 NHS les§========mm mmm meee eee 10 3 3 2 7 3 Same class interval--=-====-ceccemmmmooooooo 8 2 4 3 4 2 NO expense=======mm oom ome meee meee eee 287 83 135 87 152 79 Doctor expenses NHS greater-----=----cc-cccmcccmmceeeeeee oo 135 39 57 37 78 41 NHS leSs==mmmmmmm mmo meee eee emma 90 26 49 32 41 21 Same class interval--=====-cccemmmeoom aaa 65 19 24 15 41 21 No expense=--==----mccmmm cece eee en 58 16 25 16 33 17 Medicine expenses NHS greater-----------ceocmcomcmemcmeeeeeme en 94 27 44 28 50 26 NHS lesS==mmmm mmm mmm eee eee eee 126 36 56 36 70 36 Same class interval--=-------mcmmcmmmccaaaao 105 30 46 30 59 31 No expense------==--mmcccccc ccc cece cn ——— 23 7 9 6 14 7 Dental expenses NHS greater---------c-ecmcco mcm cceeeeee eee 56 16 26 17 30 16 NL 34 10 12 8 22 11 Same class interval--=--===-ececemocmmmannaaaoo 89 26 39 25 50 26 No expense------=-cmecccccmc occ ee—eee een 169 48 78 50 91 47 Other health expenses NHS greater-=-=====--cm coe mm emcee 46 13 18 11 28 14 NHS lesS===mmmmm mom m eee e emcee cece 67 19 37 24 30 16 Same class interval--==-=====ecoccmmmmceoaaooan 35 10 12 8 23 12 No expense---===--cmomcmmc mcm eee 200 58 88 57 112 58 Ipercentages for each type of expenditure are based on the same number of respondents throughout this table. 28 Table 17. Distribution of major discrepancy cases in which NHS was greater than NORC in total expenses according to type of expenditure and level of agreement, by type of interview: 1960 Type of interview Type of expenditure and level of agreement Total NHS direct Sisal Number | Percent || Number | Percent | Number | Percent Total ==========mmmmmmre meme ———— 170 100 78 100 92 100 Hospital overnight NHS greater------=--===mememecmenneee eee ———— 29 17 11 14 18 20 NHS legg~-=~=======—=memememmm mmm nnn 5 3 3 4 2 2 Same class interval---------meececccecceen—.— 8 5 2 3 6 6 No expense-----===-mmmccececcccce cece ————— 128 75 62 79 66 72 Hospital—not overnight NHS greater-—=--=-=-===<==cmmmmmmmnam nnn 32 19 10 13 22 24 NHS leS§=====mmrmmemccccc cece cee meen 5 3 2 3 3 3 Same class interval-----==--eecceccecmenanann 3 2 - - 3 3 No expense---=========meummmmemenme ene ————— 130 76 66 84 64 70 Doctor expenses NHS greater---------mmeceececcccccccc cece ———— 113 67 52 67 61 66 NHS leSg~=-=-=====-m-mecemncc mcm, ———————— 10 6 5 6 5 6 Same class interval---=--=--cecccceacccnneaan 26 15 11 14 15 16 NO expense--====-mmmcecceccce cece eee ——————- 21 12 10 13 11 12 Medicine expenses NHS greater--------=--ceeceeccceccccc eee ——— 75 44 36 46 39 42 NHS leSg-=mmmmmmmmmmmm— 24 14 10 13 14 15 Same class interval---- 55 33 24 31 31 34 NO QE DEINE mim oi mim i ito eo mm hh 1 00 16 9 8 10 8 9 Dental expenses NHS greater-=-====-=---mc-ceecceeecemee cee me 42 25 21 27 21 23 NHS legs-=====mmm-mmmmec ccs e nen ———— 3 2 - - 3 3 Same class interval------=---ccccecccccncnaan 48 28 25 32 23 25 NO @XpPENS@== === == =m mmm mmm mmm mmm mm mom mom om me on we 77 45 32 41 45 49 Other health expenses NHS greater======sc==—=remsmmmsmmenmeannnnmnnn 33 21 14 18 21 23 NHS leS§-=====mmmmemeececc cece cece. ——— 17 10 8 10 9 10 Same class interval----==----emcecemccecee——— 14 8 6 8 8 9 No expense--=====m==mmmemeem cece cece cee ———— 104 61 50 64 54 58 percentages for each type of expenditure are based on the same number of respondents throughout this table. 29 Table 18. Distribution of major discrepanc cases in which NHS was less than or the same as NORC in total ex- penses according to type of expenditure and level of agreement, by type of interview: 1960 NHS interview less in total expense NHS and NORC interview same in total expensel Type of expenditure Self- and level of agreement Total Direct enumeration Self- Total Direct SrugEration Number | Percent Number | Percent | Number | Percent POLEL mmm sia mins 161 100 71 100 90 100 17 6 11 Hospital overnight NHS greater-=--=rmmemmemeae—— 2 1 1 1 L 1 1 - 1 NHS lesS§----==-==occomcnnnn 21 13 9 13 12 13 2 - 2 Same class interval-------- 12 8 5 7 7 8 1 - 1 No expense---==mmmmemmm-e——— 126 78 56 7° 70 78 A3 6 7 Hospital—not overnight NHS greater--------------=-- 10 6 3 4 7 8 1 - 1 NHS less-===mrmmmmmemen———— 4 3 1 1 3 3 1 - 1 Same class interval-------- 5 3 4 6 1 1 - - - No expensSe~-=====m===mm=mm=m 142 88 63 89 75 88 15 6 9 Doctor expenses NHS greater=--====rmomemmem—- 13 8 3 4 10 11 9 2 7 NHS less-======-==-cmemnna- 74 46 42 59 32 36 6 2 4 Same class interval-------- 37 23 11 16 26 29 2 2 - No expense--=========c-==== 37 23 15 21 22 24 - - - Medicine expenses NHS greater---------------= 13 8 6 8 7 8 6 2 4 NHS leS§-====-=c=cececeena= 94 59 43 60 51 57 8 3 5 Same class interval-------- 47 29 21 30 26 29 3 1 2 No expense---=======c=ece-- 7 4 1 2 6 6 - - - Dental expenses NHS greater---=-------=---- 10 6 4 6 6 6 4 1 3 NHS les§=======c=-eeceamean-= 27 17 11 15 16 18 4 1 3 Same class interval--=----- 39 24 14 20 25 28 2 - 2 No expense--=-=--=-==-e-e=e=== 85 53 42 59 43 48 7 4 3 Other health expenses NHS greater----------==---- 6 4 1 2 5 6 5 3 2 NHS lesS==-==-===mceceemunnn 48 30 29 41 19 21 2 - 2 Same class interval-- - 19 12 6 8 13 14 2 - 2 NO expense-====-=====-c-==-= 88 54 35 49 53 59 8 3 5 Io percentages shown because totals are too small. “Percentages for each type of expenditure are based on the same number of respondents throughout this table. 30 dividuals for whom charges were exaggerated are compensated for by an almost equal number for whom charges were underestimated (16 and 13 individuals, respectively). The two most frequent specific causes for overstatement were wrong data and duplication of charges. NORC screened for the reported date of last hospitalization and was able to find five cases where thehospitaliza- tion occurred prior to the survey year. In four cases, costs for hospital room and board and miscellaneous hospital charges were included more than once, resulting in the inflation of the total estimate of costs. A study of the 28 understatements of hospital expense showed that 29 percent were found to have forgotten about a single hospitalization, 18 percent forgot the second or third hospitalization, and 7 percent included the overnight costs in the hospitalization not-overnight category (table 19). Not-Overnight Hospitalization This category was cviginally separated from overnight hospitalizations to avoid being over- looked if not specifically mentioned. Actual field experience, however, indicated that the respond- ents were often confused about the meaning of this item. As shown intable 20,28 percent of the over- statements erroneously included regular doctor visits, 7 percent included overnight hospitaliza- tion, 2 percent mistakenly included nondoctor care, and 21 percent included the wrong date for the expense. The costof not-overnighthospitalizations was included under other categories in 70 percent of the understatements—30 percent doctor expense, 20 percent medicine, and 20 percent overnight hospitalizations. Of course, the numbers of cases were small, and this must be considered in eval- uating the percentages. Doctor Expenses Over 70 percent of all overstatements of doctor expenses could not be attributed to any- thing more than general overstatement. Of the remaining specific reasons, free care and wrong date were the most numerous. The major reasons for understatements of doctor expenses were the omissions of serious illnesses. In 26 percent of the understatements a major condition was forgotten; in 23 percent a chronic illness. Forgetting to report visits to hospitals and office visits each accounted for 17 percent of the understatements. Omission of minor illnesses and charges for anesthetists and sur- geons each accounted for about 10 percent. These data are presented in table 21, Medicine Expenses Virtually all of the overstatement in medi- cine expense appeared to be general exaggeration. Free care and errors in allocating family costs were other reasons for overstatement. Understatements, however, were due to more specific reasons. Almost one-third were due to omission of a major condition which also re- quired special medication. About one-fifth of all understatements were due to forgetting general medicines and tonics or failing to allocate gen- eral medicines to all family members. Other rea- sons are presented in table 22, Dental Experses As in the case of doctor and medicine ex- penses, overstatements of dental expenses were largely general exaggerations of cost, accounting for 91 percent of the differences. Understatements can be attributed to omis- sions of general examinations in 65 percent of the cases, to forgetting about fillings in 62 per- cent, and to omission of treatments, extraction, and bridgework in lesser numbers of cases. These details are listed in table 23. All Other Health Expenses The biggest single reason for both overstat- ing and understating this category of expense was the mistake made in the cost of eyeglasses (15 and 72 percent, respectively). Omission of non- doctor care amounted to an additional 15 percent of the understatements and the omission of other medical appliances accounted for the remaining underreporting. These details are presented in table 24, 31 Table 19. Reasons for major discrepancies in overnight hospitalizations: 1960 Discrepancy Number | Percent Total OVErSTALOMBIVE Sw mmsmmimm mime iu sia iw oii om sw 00 18 00 8 ot 5 8 8 i i 0 32 100 WEONE QaL@m mmm mmm mmm om om om mm mm mm mm or 0 me 0 0 om 5 16 Duplication of charges —=mermw=ressn cnn sn mmmm wm mm om wm om wom om ow - 12 Included OULPILLIENL Caren = =mmmmm =m mw mmm me mmm mmm mmo oo orm oo oro 0 on om mm 0 0 2 6 Included surgeon fee-==r-rmr=srmmrm==mccecscm memes — rn. ————————————————————— 2 6 Error in eS pordayl=imessins =n dies dm Ws wn om Hm mo on am 1 3 Included doctor VigiLg-=~=s=~eesdummunemioes 1 3 Included. HedioingBme mm mmm mmm wm me 1 3 General error in estimate 16 50 Total understatement§========m== === cme c esse ——————————————————————— 28 100 Omitted only hospitalization-=========remeemer erence nem erm e—————.————— 8 29 Omitted second or more hospitalizations-----=-====emeemceenee mc cn nen ene ———— 5 18 Reported as not-overnight stay-=-=======s=meceeememceenee eee — nese. —————— 2 7 General error in estimate--========me=cemcmccce semen ——————————————— 13 46 Table 20. Reasons for major discrepancies in not-overnight hospitalizations: 1960 Discrepancy Number | Percent Total overstatements' 0 8 00 43 100 Included doctor office visits 12 28 Wrong date--==-==-meemmemccec cece mem meee eee memes ————————————— 9 21 Included overnight stay----===--=-cccccccecccccccccce ccc cece cece 3 1 Included free care----=--==-=memececcccccccmccceeceeee meee ee eee mm ————— 3 7 Wrong respondent-=--=-===mmmececceaccccece cece cece e seme esse eee —————————— 2 5 Included nondoctor care-==-==-=m=m--mmccccc cece eee eee ee eeeeeeeee—e———— 1 2 General error in estimate-=-==-----coccoceccc cece 13 30 Total understatements’ A 10 100 Included in doctors=--===-=-ceecccccc ccc meee meee mem meme eee 3 30 Included in medicine-========cecoccm ccc eee 2 20 Included in overnight stay--==--===----ccmeccccccccc cece m eee m———— 2 20 Omitted only hospitalization 1 10 Omitted lab test§-=---==--ceccecccccc ccc ccc ccccccc ccm ——— 1 10 Error in other treatments--====-=m--ccecccc ccc ccc e eee meee—————— 1 10 I7otal refers to number of cases although more than one reason may be given for the discrepancy. Therefore, individual numbers and percent- ages may add to more than their totals. 32 Table 21. Reasons for major discrepancies in doctor expenses: 1960 Discrepancy Number Percent Total overstatementsSl-=---c ccm om m oo mmm een Included free care-=--=-==-ceccccccccccccccccccccccccccccccccccccccccccccccc———— Wrong date====-cccecmccc cmc mecmccceccccceccceecce———- Wrong respondent=-=-----cccccmccccccmcccccccceeceeeeeeeceeeccecceeceee———————— Included medicine====-mccccc ccc ccc cccccmmccccccccccccccccccccaaa Included outpatient clinice===mremccccm ccc cccmccmmc cece Included current pregnanCy=--=---=--ceeeccccccccccccccccccecccmcceeceec———————— Duplication of viSitS====cmeccmc occa Included nondoCtOor=======ecccccmcccccccccccaccccccccccccccccaccccc meme ———————— Error in home viSitS=eececcccccc cmc c ccc ccc ccc cmccccccccccccccccccccceeeam Error in office visSitS-==eeecemc ccc cccccccccccccccccccccece eee General error in estimate---==-c-ccccccccccccccccccccccccccccccccccccccc cca Total understatementsl—m==ee oom ooo eee ee Omitted major condition-=-===ecccccc cmc mee mae Omitted chronic illneSS=====meccmeccccccccccccccaaccccccccccccccccccccccccce——a Omitted visits in hospitale=-===ccecmccm mcm eee Omitted office ViSitS==m=memcc comme ccc ccdcccceeeae Omitted minor illnessSeS==-====ceecccccccccccccccccmcccccccccccccccccccccccco—aa Omitted anesthetiStS====ececcccccmcaccmcccccaaccccccccccccccccccccccccccmc————— Omitted SUrgeon-======-ceccccccc ccc cccccccccccccceeeceeceeceeme————————— Omitted tests and treatments pre- and post-hospital care------=--ccecccccccaaa- Care included in hospital outpatient-=--=-=--=-cemcm momma Omitted insurance payment====-=== eee came eee Omitted clinic visits in hospital caseS========cccemcccccccccccccmcccccccceaao General error in estimate--===ecc ec cec cmc ccc cccccccmcccccccccccccccccccma 135 100 — AHF MFHRFRFWLWANMPOL —- FREER NWWWOH ~ 100 Total refers to number of cases although more than one reason may be given for the discrepancy. Therefore, individual numbers and percent- ages may add to more than their totals. Table 22. Reasons for major discrepancies in medicine expenses: 1960 Discrepancy Number | Percent Total overstatementsl-=====m comma eee 94 100 J YR Fb of EE ae 2 2 Included family total=====-ceemc moomoo coe ecm ccm eee cece mee mee eee 1 1 Error in respondent==== === eco o momo coe eee meee eee emen i i Duplication with doctor=====m=c common 1 1 Duplication with hospital--=====ccmemc ammo coca 1 1 General error in estimate=======eoe meen occ o ccc ecceccccemmcemee eee 90 96 Total understatementsl=m===m emo 119 100 Omitted major condition medicinee=======- comm como eeeee 36 30 Omitted miscellaneous medicines=======ccccc moomoo ccc 26 22 Family total not allocated======== eee c momo c ccc ccccccccmeec een 2 18 Omitted medicine for hospitalized condition=-===-eeccccccmmm mcd 19 16 Omitted doctor suggested medicine---=-===c-ccc comm momo eee 15 13 Omitted prescriptions==--=====c comm momma mn 13 il Wrong respondents==== === momo ome eee eee ee eee eee eee emmmmm meee 3 3 Wrong date--=--===== cmc mmmm mmm mmm mmm 1 1 Included in doctor charges--=====-==cc common a 2 Total refers to number of cases although more than one reason may be given for the discrepancy. Therefore, individual numbers and percent- ages may add to more than their totals. 33 Table 23. Reasons for major discrepancies in dental expenses: 1960 Discrepancy Number | Percent Total overstatements! === mmm moomoo oo oem 56 100 VWEONE TEOPOTHARTIE S = ww me ve ve me ve es mo 0 2 0 020 0 0 0 0 0 0 0 0 to tm tt 3 5 Free care---------ceccmecccc cmc m meee eee m meee meee meme meee—e——————oa— 2 4 Wrong dateS-=====mmeme-eeee seem m meee m mmm m meee meee eee eee ee ee ——————— 1 2 Included in doctorS=----=-==-c-cccmcmc emcee eee mm meme meme cccccmoeo-o 1 2 General error in estimate----------cccemcccmcccmmm meme mm emer mmcmmemmmemeo 51 91 Total understatementsl==mmm moomoo moomoo 34 100 Omitted examination----=-=-=ceecccccmec cece cece cece ecm emcee mmm m meme 22 65 Omitted fillingS-===---cmccmccmmm cece ccc cc ccm 21 62 Omitted other treatmentS----=-=---ccemmccccccc cece emcee c ccc ccc me mmo 14 41 Omitted extractionS---------ccccmccccmc mcm ce cmc c ccc c ccm mm mmm mm eo —— 12 35 Omitted bridgework-=-=-====-emceecccccccc ccc ccc ccc cc ccm mmm c ccc c cae eee 10 29 Wrong date==-=-=--mcecceccceme meme m eee meee mmememm meee ccceee————- 1 3 Wrong respondent---=-—==----e meee ene m mmm meme meme mmm mmm mmm mee 2 6 ITotal refers to number of cases although more than one reason may be given for the discrepancy. Therefore, individual numbers and percent- ages may add to more than their totals. Table 24. Reasons for major discrepancies in other health expenses: 1960 Discrepancy Number Percent Total overstatementsl-mmmmm momo moc o cee ee em Error in glasses------=c-—cocmmcmmc meee cme mmme meme mmm ——e———— Free care included-===---mcccmcmmc mcr creme cccmmcccmccmcc cmc cm cmc m mem ——— Wrong dates======—= moomoo momo eee meme mea Wrong respondent ====== === comma me ee ee ee cmeeee emcee emceeee—————— General error in estimate-=-===ccmmmmmc coco mmmcm mmm nme Total understatements!====-m moomoo oo meee Omitted glasSesS=========- cco mmm emccceceemeemee— Omitted nondoctor care--=-=====ececccccceccccccccccceccmemmeee eee m———————— Omitted elastic hoSe========cceccmmmcecccccccccccccccccccmcccccc ccc ccm cee ——— Wrong respondent===-—===-me mmo o come eee ccmeeemeemeececemeeemee———— Omitted wheel chajr-----=ccccemcmmmmcccccccceccccccccccccmc ccc ccc ccc ccc mmm ———— Omitted artificial leg----===----=c=ccccmcem ccc ccccceccmeemmeen—e— ea Omitted special COrSE@t===m====m mmm cocoon cm meee emma Omitted arch SUPPOTtS====m= =m mmm mo oe eee eee mee Wrong date-====m mmm mm mmm eee ee ee eee mmm meme mmm e mem 100 ANN F& 100 72 15 4 3 3 1 1 3 1 Total refers to number of cases although more than one reason may be given for the discrepancy. Therefore, individual numbers and percent- ages may add to more than their totals. 34 ll. CONCLUSIONS AND RECOMMENDATIONS As compared with a detailed criterion inter- view, the short questionnaire provided reasonably accurate information on total personal health ex- penses and on major components of the expenses. The quality of short-form reporting did not appear to vary from the detailed interview to such a degree as to indicate any serious bias in the es- timates of expenditures by sex, age, or educa- tional attainment of the head of the household. Short direct interviews were slightly more accurate than self-enumerated reports. However, depending on the degreeof accuracy desired, self- enumeration may be used as a reasonable sub- stitute if direct interviewing time is limited. In making overall estimates of health ex- penses, there was a general tendency for re- spondents to inflate estimates. Specific detailed probing reduced this general overstatement, but also added some costs which were overlooked. Although there were substantial discrepan- cies in detailed comparisons, the overreporting tended to offset the underreporting so that the marginal frequency distributions and mean esti- mates of health expenses were not substantially different in NORC and NHS interviews. Study of cases in which there were major discrepancies between expenses reported on the short NHS questionnaire and the detailed NORC interview indicate that NHS estimates could be improved by the measures listed below. Consolidate not-overnight hospital outpatient and emergency care in doctor expenses. Stress actual dates of period covered by survey. Screen ior actual date of last hospitalization to avoid errors in dates. Probe for second and third hospitalizations during the survey year. Use reminder probes on major conditions and chronic illnesses in connection with doctor and medicine expenses. Use other specific reminder probes on hos- pital, office, anesthetist, and surgeon charges in aggregating doctor expenses. Allocate general medicine, tonics, and vita- min costs among all family members. Use reminder probes on dental cost question to assist recall of types of treatment. Use a separate question on costs of eye- glasses, including any charges for exami- nations. Use appropriate reminder probes for medi- cal appliances. The extent to which suggested additional probes are appropriate depends, again, onthede- gree of refinement desired relative to the prob- lems of maintaining a short form questionnaire that is administratively practical. Although the short form tested in this study yielded reasonably accurate overall estimates, any of the above measures that could be incorporated would fur- ther refine the data. 35 APPENDIX | QUESTIONNAIRES USED IN STUDY A. Supplemental Questions Used in the National Health Survey Two methods were used to collect the data on personal health expenditures in the initial inter- view of the National Health Survey in order to in- vestigate the relative merits of data collection. In half of the households the questions were asked as a part of the direct interview while a self- enumeration questionnaire was leftin the remain- ing half of the households. NHS-4(a), reproduced below, was the ques- tionnaire used in direct interview. NHS-4(b) is a reproduction of the letter and form left in each of the self-enumeration households. A separate page of questions was left for each person in the latter households in a booklet form. 36 B. Criterion Questionnaire Used by the National Opinion Research Center This questionnaire was composed of three parts: the main criterion questionnaire, a hospital supplement, and a major condition supplement. Supplements were to be completed for each sepa- rate hospitalization episode and each major condi- tion. The latter was defined as any condition which met one of the following criteria: a current pregnancy which did not involve hospitalization; surgery, bone setting, or other operations which did not involve hospitalization; conditions which involved expenditure of $50 or more which did not involve hospi- talization; and chronic illness determined on original house- hold interview survey. DIRECT INTERVIEW QUESTIONNAIRE FORM NHS-4 (a) (9-17-60) COSTS FOR MEDICAL AND DENTAL CARE FOR PAST 12 MONTHS Ask question 1 only for persons who were in a hospital (nursing home, sanitarium) overnight or longer during the past 12 months. Ask questions 2 - 6 for EVERYONE. (Check one box): (1) [] In hospital (——— times) 3 Not in hospital 1. (a) How much did your hospital bills come to for the (one time, two times, etc.) you were in the hospital this past year? In case you don't know the exact amount of the bills, give the best estimate you can. [] No bills (Free care) $ (b) Does this amount cover ALL of the hospital charges--for example, in addition to the cost of the room, does it include charges for the operating (or delivery) room, anesthesia, X-rays, tests, special treatments, etc? If not included originally in 1(a), correct amount in 1(a) From bills or records: [7] All charges included [] Some missed - Now Corrected Not from bills or records: [] All charges included [] Some missed - Now Corrected (c) Was any part of the hospital bill paid for by insurance, whether paid directly to the hospital or paid to you or your family? If ““Yes '’ to 1(c), ask: (d) Was the part paid by insurance included in the amount (in 1(a) ) you gave me? If not included originally in 1(a), correct amount in 1(a) [J No part paid by insurance [] Part paid by insurance and included in 1(a) [] Part paid by insurance, not originally in 1(a) - Now Corrected (e) Besides these hospital bills you have already told me about, how much did the bills for 2 Ne any special nurses at the hospital come to? special $ nurses 2. (a) During the past 12 months did you go to a hospital for any minor operations, emergency Y N treatment, outpatient clinic services, X-rays, tests or any thing like that for which you (Ives J% did not stay ovemight? If “Yes '’ to 2(a), ask: (b) How much altogether did these kinds of hospital bills come to? $ (c) Was any part of these hospital bills paid for by any insurance, whether paid directly to the hospital, the doctor or to you or your family If “Yes” to 2(c), ask: (d) Was the part paid by insurance included in the amount (in 2(b) ) you gave me? If not included originally in 2(b), correct amount in 2(b) [7] No part paid by insurance [7] Part paid by insurance and included in 2(b) [] Part paid by insurance, not originally in 2(b) - Now Corrected 3. (a) How much did all of your doctors’ and osteopaths’ bills come to for the past 12 months? This amount should include all doctors’ bills for home and office visits as well as for clinics and hospitals. [CINo doctors’ bills $ (b) Does this amount cover all doctors’ bills --for example, operations or treatments, check- ups or examinations, X-rays, tests, etc? If not included, correct amount in 3(a) From bills or records: All charges included [C7] Some missed - Now Corrected Not from bills or records: [7] All charges included [77] Some missed - Now Corrected (c) Was any part of these doctors’ bills paid for by insurance, whether paid directly to the doctor or to you or your family? If “'Yes’’ to 3(c), ask: (d) Was the part paid by insurance included in the amount (in 3(a) ) you gave me? If not included, correct amount in 3(a) [7] No part paid by insurance [7] Part paid by insurance and included in 3(a) [] Part paid by insurance, not originally in 3(a)- Now Corrected 4. During the past 12 months, about how much did you spend (for yourself) for prescriptions, medicines, tonics, vitamins, pills and things like that? [7] No ex- penses for medicines, etc, 5. (a) How much did all of your dentists’ bills come to for the past 12 months? | No dental bills $ (b) Does this amount cover all dental expenses--for example, all fillings, extractions, cleanings, X-rays, bridgework, dental plates, straightening of teeth, etc.? If not included, correct amount in S(a) From bills or records: [] All charges included [] Some missed - Now Corrected Not from bills or records: [7] All charges included [] Some missed - Now Corrected 6. We are interested in OTHER medical expenses you may have had during the past 12 months but we don't want to include any insurance premiums you may have paid. About how much did all OTHER medical expenses come to for you during the past 12 months-- for example, things like eye glasses, hearing aids, braces, chiropractors’ fees, home nursing care and the like, not counting those you have already told me about? [] No other medical expenses 37 SELF-ENUMERATION QUESTIONNAIRE FORM NHS-4(b) Form Approved (9-16-60) Budget Bureau No. 68-6013 U. S. DEPARTMENT OF COMMERCE BUREAU OF THE CENSUS WASHINGTON 25, D. C. Dear Friend: The Bureau of the Census is conducting a special survey on the cost of medical care as collecting agent for the U.S. Public Health Service. This study, when combined with other infor- mation, will serve to answer important questions about health and medical care costs in our Nation. The Census interviewer who called at your household was asked to leave this form in order that all of the family members could take part in answering these questions, and that bills, receipts, and other records can be consulted. Please mail the completed form within five days. A self-addressed envelope which requires no postage has been provided for your convenience. Your cooperation in answering these questions will be a definite public service. The in- formation will be given confidential treatment by the Bureau of the Census and the U.S. Public Health Service. Nothing will be published except statistical summaries. Thank you. Sincerely yours, (oto (Ruins Robert W. Burgess Director Bureau of the Census CONFIDENTIAL - This information is collected for the U.S. Public Health Service under authority of Public Law 652 of the 84th Congress (70 Stat 489; 42 U.S.C. 305). All information which would permit identification of the individual will be held strictly confidential, will be used only by persons engaged in and for the purposes of the survey, and will not be disclosed or released to others for any other purposes (22 FR 1687). USCOMM-DC 11431 P-60 l COSTS FOR MEDICAL AND DENTAL CARE DURING THE PAST 12 MONTHS From: To: Pl #ASE ANSWER THE QUESTIONS BELOW FOR: Name of person GENERAL INSTRUCTIONS . IF YOU DO NOT HAVE EXACT AMOUNTS FROM BILLS OR RECORDS, GIVE THE BEST ESTIMATE YOU CAN. 2. COUNT: All bills paid (or to be paid) by the person himself, his family or friends and any part paid by insurance whether paid directly to the hospital or doctor, or paid to the person or his family. If you do not know the exact amount paid by the insurance, estimate it. 3. DO NOT Any amounts paid (or to be paid) by: COUNT: Workmen’s compensation Non-profit organizations such as the ‘Polio Foundation” Charitable or Welfare Organizations Military Services or Veterans Administration Federal, State, City, or County Government 1 (a) Was this person in a hospital, (nursing home, sanitarium) overnight or longer during the past 12 months? Yes [CJ] No (Go to Question 2) (b) How much did all of the voor bills come io for the past 12 months? Count all bills for all hospital stays during the past 12 months. Be sure to include all hospital charges and not just the cost of the room and board, for example, $ charges for the operating or delivery room, anesthesia, X-rays, tests, special treatments, etc. (c¢) IN ADDITION TO the amount given above, how much were the bills for the surgeons, anesthetists, or other doctors’ visits at the hospital? [_] No doctors’ bills for $ hospital services. (d) Besides these hospital bills, how much did the bills for any special nurses at the hospital come to? [J] No other bills for $ special nurses. 2 (a) During the past 12 months did this person go to a HOSPITAL for any operations, emergency treatment, out-patient clinics, X-rays, tests or any things like that for which he did not stay overnight? Yes [J No (Go to Question 3) (b) How much altogether did these kinds of hospital bills come to for this person? $ (c) IN ADDITION TO the amount given above, how much were the bills for the surgeons, anesthetists, or other doctors’ visits at the Yon for this kind of care? [] No doctors’ bills for this kind of care $ ANSWER QUESTIONS 3, 4, 5 and 6 FOR THIS PERSON WHETHER OR NOT HE (OR SHE) HAS BEEN IN THE HOSPITAL DURING THE PAST 12 MONTHS. 3. IN ADDITION TO any amounts that you may have listed in Questions 1 and 2, how much did all of the doctors’ or osteopaths’ bills for this person come to for the past 12 months for home, office, or clinic visits? Be sure to include all doctors’ bills for all treatments, check-ups, X-rays, shots, tests, and the like. [J No doctors’ bills $ 4. About how much was spent for this person for prescriptions, medicine, tonics, vitamins, pills, and things like that during the past 12 months? [CJ No costs for medicine, etc. $ 5. How much did all of the dentists’ bills for this person come to for the past 12 months? Be sure to include all dental costs for fillings, extractions, cleanings, X-rays, bridgework, dental plates, straightening of teeth, etc. $ [1 No dentists’ bills 6. About how much did all other medical expenses for this person come to for the past 12months? Include the costs of all such items as eye glasses, hearing aids, braces, chiropractors’ fees, home nursing care, and the like, not included above. $ [C7] No other medical expenses PSU No. Segment No. Serial No. Column No. FOR OFFICE USE ONLY NORC 429-2 Budget Bureau No: 686013 10/60 MAIN CRITERION QUESTIONNAIRE FAMILY NO. RESPONDENT'S NAME: PHONE NO.: ADDRESS: SUGGESTED TIME TO CALL: RECORD OF CALLS Date Time Results 1st 2nd 3rd 4th Notes: This Main Questionnaire has: Hospital Supplements (number) Major Condition Supplements Time interview began: Time interview ended: For this family, year ago means to Date of interview: Signature of Interviewer: Names of Family Members: Relation to Head: Age: Sex: 40 INTRODUCTION. Hello. I'm and I'm working on the health survey for the Public Health Ser- vice. We talked to you a short time ago about your health and medical care and arranged to see you again today. ASK QUESTION 1 FOR EACH MEMBER SEPARATELY, BEGINNING WITH FIRST MEMBER. 1. Let's start by talking about hospital, sanatorium, convalescent or nursing home care. When was the last time (NAME OF MEMBER) was a patient in a hospital, sanatorium, convalescent or nursing home--even if (he) (she) didn't stay overnight? (What year and month was it?) Never o iv o oo 28 5 » a2 3 9 «+ » vo a « 8 = Mye than year 8g0 .« « « « & « + ss wi 4 + » » 2 Year ago Or 1ess + « + + o so ¢ + 4 4 44 4 3 FOR EACH MEMBER REPORTING HOSPITALIZATION, ETC., YEAR AGO OR LESS, 2. ASK ONLY IF THERE IS A MARRIED WOMAN UNDER 45 IN THE HOUSEHOLD: A. And can you tell me if (MARRIED WOMAN UNDER 45) has been pregnant (at any other time) during the past 12 months--I mean from to 7 *B. IF "YES": Is that a current pregnancy, was the baby born already, or did it end in a miscarriage? **¥C., IF BABY ALREADY BORN, OR IF MISCARRIAGE: Did (WOMAN) go to hospital in connection with this (birth) (miscarriage)? FILL IN A HOSPITAL SUPPLEMENT BEFORE YOU ASK Q. 2. Y88 « vo 2 vo ¢ 2 os 0 0 5 5s ov 2 +» + J ml¥ NO « soos ov wid o oo wis vwwnw 2 Current » =» + » + ww mw ss » v wi sv w 10 =» LF Baby DOIN + « ¢ ¢ o 5 5 4 ¢ oo 5 + + +» 4» 2¥% Miscarriage « « «+ « «+ + + 4 4 4 4 4 44. . 3Z¥% Yes « « + so vo oo 0 2 2 2 a 0 + 0 0 + oll - WE BO ¢« wanna s mwmds vawwowoenws ZF #IF CURRENT PREGNANCY OR BIRTH OR MISCARRIAGE WAS NOT HOSPITALIZED, ENTER CONDITION ON MAJOR CONDITION SUP- PLEMENT AND FILL OUT AFTER ASKING QUESTION 5. #4 IF WOMAN HOSPITALIZED, FILL OUT HOSPITALIZATION SUPPLEMENT BEFORE ASKING QUESTION 3. 41 3. ASK FOR EACH MEMBER SEPARATELY, BEGINNING WITH FIRST MEMBER. A. *B. In the past twelve months, I mean since has (NAME OF MEMBER) had any surgery--like a boil lanced or tonsils removed--or any broken bones or fractures set, outside of a hospital? IF "YES": What was the condition? 4. And *B. now about any (other) expensive illness (in addition to w In these past twelve months--that is, since have you (has PERSON) had any (other) illness, ry cident or condition for which the charges were as much as $50 altogether for doctor's care, medicine, treatments and tests, and so on? Please combine what you had to pay and what any insurance paid in figuring the $50 cost. IF "YES": What (was) (were) the condition(s)? hat you've already told me): Yes + « iv + + vow ov www ows wow wv 1O= LE HAND WRITE CARD TO RESPONDENT AND READ Q. S, INSERTING EACH CONDI TION LISTED IN COLUMN FOR THAT PERSON. THEN REVIEW ANSWERS AND MAKE CORRECTIONS BY PUTTING LINE THROUGH CONDITION WHICH SHOULD BE LEFT OUT AND BY WRITING IN CONDITIONS WHICH SHOULD BE INC LUDED. 5. During the first interview, you mentioned that (PERSON) had (CONDITIONS LISTED) (none of the following conditions) during the past 12 months--that is, since . Now, I'd like you to check this list again to make sure we haven't made any mistakes or left something out. All correct first interview? Yes + + 4 + + +t ver ts eee rere oe 1Bal WO 4 wv ww sv p00 8 3 05mm oo 0% sO STOP! BEFORE ASKING Q. 6, PREPARE A MAJOR CONDITION SUPPLEMENT FOR EACH DIFFERENT CONDITION MENTIONED IN Q's. 2 = S FOR EACH MEMBER. STARTING WITH FIRST MEMBER, ASK Q's. 6 - 8 BEFORE PROCEEDING TO THE NEXT MEMBER. 6. Up to now we've been talking about illnesses or conditions that may have involved quite a bit of expense during the past twelve months. A. Aside from what we've talked about, (have you) (has PERSON) seen a doctor or osteopath at your: home for any (other) illness or condition, even minor ones during the past year--that is, since ? *IF "YES" TO A, ASK B - D: B. What were the illnesses or conditions requiring these home visits? How many times altogether (have you) (did PERSON) see a doctor at your home? And how much did all these doctor's calls to (you) (PERSON) at home cost you, including both what you had to pay and what insurance paid? Now I'd like to find out about any other minor medical expenses. 16- 1* EB 2» wes 2 9 C2 8 $ SEE « » NO uo ooo oo smd ns 0 sm os www © times 7. A. Aside from what you've already told me about, (have you) (has PERSON) seen a doctor or osteopath at his office or at a clinic for any (other) illness or condition, even minor ones, during the past year-- since 2 *IF "YES" TO A, ASK B ~ D: What were the illnesses or conditions? How many times altogether (did you) (did PERSON) see the doctor at his office or clinic for these conditions? And how much, altogether, did these office or clinic visits cost you, what you paid and what insurance paid? Yes + vo ov ws + + ss wis so vw ww +» » L7= 1% times including $ 8. HAND YELLOW CARD TO RESPONDENT AND ASK A - C. CIRCLE "YES" OR "NO" CODE FOR EACH ITEM IN PART A, BEFORE ASKING PARTS B AND C. Now just to be sure we haven't left anything out,-- A. In addition to the visits we've already counted in other questions, did (you) (PERSON) see a doctor, osteopath, nurse, or doctor's assistant during the past es Yo year for any of the following conditions which you may have forgotten to tell me about? (1) A routine check-up OF eXOIINELIONT « os & o ® @ + 3 sw 0 4's » ® iv + + w « « * 1x 2 (2) Any shots, inoculations, vaccinations, X-rays or tests? + « + + + + + + «4 1% 2 (3) Any chest or throat conditions like a cold, flu, virus, sinusitis, tonsilitis, bronchitis or pneumonia? + « « « « + ¢ + tt 4 4 4 4 eet see eee eee 1x 2 (4) Measles, whooping cough, or Chicken POX? + + + = + + o + = = + = + o + o + os » 1x 2 (5) Any injuries like cuts, sprains, burns, bruises, broken arms or legs? . . . . 1* 2 (6) Digestive flu or other stomach LIOUDLE? + + + « 4 oo « + + + + o os + + + + » 1* 2 (7) Any ear, eye or skin Infections? + + + + + + + + o = a + + + ss 4 44... 1* 2 (8) Any serious headaches or backaches? + + + + + 4 4 4 4 4 os + so ss oo a» 1% 2 (9) Any kidney or bladder infections? + « + « o + o oo = + 4 + va 4 eee. a 1* 2 *IF "YES" TO ANY ITEM, ASK B AND C AFTER ASKING ALL NINE ITEMS FOR THAT PERSON. B. Now, not counting any visits you've already mentioned in other questions, about how many times did (you) (PERSON) see the doctor for these conditions? . « + « + times C. And how much did he charge you for these additional visits? . « « « « ¢ « + « + $ Xx **DON'T FORGET TO GO BACK TO QUESTION 6, AND ASK Q's. 6 - 8 FOR THE NEXT PERSON. 9. ASK PART A - D FOR EACH MEMBER SEPARATELY, BEGINNING WITH FIRST MEMBER. Now a few questions about the cost of medicines, drugs and vitamins and things like that. A. In the past 12 months, since , in addition to the cost of medicines and drugs you may already have told me about for (PERSON), how much were you charged for any $ prescription medicines for any of the other illnesses or conditions (PERSON) had? B. What about other medicines, drugs or vitamins that some doctor may have told (PERSON) $ to use last year--how much were you charged altogether for them? C. And about how much did you spend for (PERSON) on other drugs and medicines that were $ not suggested by a doctor--like aspirins, vitamins, tonics, laxatives, mouth washes, cold pills, nose drops, cough medicines and things like that? USE THIS BOX FOR COSTS OF MEDICINES WHICH CANNOT BE DIVIDED AMONG INDIVIDUALS D. Then the total cost of medicines and drugs you've told me about for (PERSON) comes to $ . Does that sound about right? (IF NOT, MAKE THE NECESSARY CORRECTIONS. ) $ 10. A. ASK FOR EACH PERSON: Have you (has PERSON) had glasses made or replaced, or had Yes + « oo +» 1B 1% your) (his her) eyes examined in the past 12 months (other than that we've N 2 talked about)? DB +s HEE *IF "YES," ASK B AND C: B. How much were you charged for the glasses? $ Total B + C C. And (aside from the charges that you've already told me about), how much was the $ $ (additional) charge, if any, for the eye examinations and treatments? Included in bill for glasses [] Included in dr bills reported earlier [] (ENTER AMOUNT OR CHECK ONE OF TWO BOXES, AS APPROPRIATE.) 43 11. ASK EVERYBODY: A. And during the 12 months since did you have to buy or rent the follow- ing for anyone in the family? TASK FOR EACH ITEM. IF ANYONE HAD ITEM, CIRCLE "YES" OR "NO" CODE FOR EACH PERSON. IF NO ONE HAD ITEM, CHECK "NO ONE" BOX BE- LOW.) (Who?) (For whom?) Yes No (1) Ahearing aid? + « « + + o 4 + 4 4 + 0 0a NO ONE [J] 1x 2 (B) Crobelies? + « io 5s & » #5 + ¢ # & ® % % # NO ONE [7] 1% 2 (BY AGTUBST vw vim + ¢ £8 ME + 3 MEE ks ow NO ONE [J 1* 2 (4) Braces? . « + « « NO ONE [] 1* 2 (5) Avheel chair? « + « + « NO ONE [J 1x 2 (6) Anything else we haven't already included? (SPECIFY) No ONE [J 1* 2 *B. IF "YES" TO ANY ITEM ABOVE: About how much did (this) (these items) for (PERSON) cost during the past year--including what you paid and what insurance may have covered? (Not counting the glasses.) $ 12. Now a few questions about some other kinds of health expenses and we'll be through. A. Has (PERSON) .iad any dental care during the past 12 months, that is since ? Yes v 8 v 29 1% No . . . 2 *IF "YES," ASK B - H: Yes No B. Did (PERSON) have an examination of teeth or cleaning . . « « « « « « « 1 2 C. Did (he) (she) have any fillings or in-lay work? . « « «+ + « « « « « + « 1 2 D. Did (he) (she) have any teeth pulled or other mouth surgery? . . « « « « « + « + 1 2 E. Did (he) (she) have any teeth straightening work? . . + « « « « « « « « « « « 1 2 F. Did (he) (she) have any bridgework or replacement of teeth? . . . « . +. « + «. . . 1 2 G. Did (he) (she) have any other dental work, like gum treatments, X-rays and things like that? « « « o o + + oo oo so os os 2 os os 8 oo + 8 + + 2 2 8 8 + oe eee oH 2 H. About how much altogether did all this cost you during the past 12 months-- includ- ing what you paid and what any insurance may have covered? . . . +. + + + « « «+ $ 13. Besides what you've already told me about, (have you) (has PERSON) seen any of the following during the past year? Yes No A. A chiropodist or foot doctor? + « « « « & ¢ 4 4 4 4 4 te eee ee ee 20. 1* 2 B. A physiotherapist? + « « « « « «to 0 «0 0 a 4 4 bee eee eee eee 21. 1% 2 Co AChiTopractor? « + + » « + + oo + 2 3 3 # #2 3 5 4. 8.8 % & 8 4 4 8 + © & ® 22- 1% 2 De AvVisiting nurse? « « « « + + o 2 5 + + 4 8 5 + + 4 ¢ 8 8 vs 5» 2 vs « xv vo» 23- 1% 2 E. Or some other medical person, other than those you've already told me about? . .| 24- 1% 2 *F. IF "YES" TO ANY ITEM: About how much did it cost you for these visits? $ 14. Now would you check to see if I have this all right? It cost (NAME) A. For hospital care (Q. 13 of Hospital Supplement) + + + « + «+ + « + « + o $ B. For operations or bone settings outside a hospital and for other major conditions (G: 8, MaJor CORILLION] « ¢ wo #5 + 5 8'4 % + » Ww o's 2 Lo T* os 20 0% + 2 $ Co For doctor calla at nome (Q: BD) « + v5 v « s 2 F & % + ® ®# 5 8 6 © ® ¢ + $ D. For doctor office calls (Qe 7=D) + + + « v + 4 +o 4 ov 4 os 0 so 0 os 0 0 os $ E. For other doctor charges (Qe 8=C) « « « « « v «+ ov + +t +o so ot 0 0 so ov ou $ F. For other medicines and drugs not included already (Q. 9-D) + + « « + + + + + $ G. For glasses (Qe LO=BHC) + + « ¢ + « « + o « + oo o os + + os + so so + 2 vv 2 2 ss $ He For other special items (Qu 11uB) « « + sv + + x 0 0 + + + so 24 3 sa + $ I. For dental expenses (Q. 12-H) + « + + « « « « « + = = + = 2 2 + 2 + so os 0 ¢ + + + $ J. For other medical services (Q. 13-F) « ov © #» # # ow ® ¢ wx wx ® 5 v w= x + + $ 44 X. That makes a total of §______. Does that sound right? (IF NO, MAKE NECESSARY CORRECTIONS.) INTERVIEWER SUMMARY SHEET (To be filled out immediately after you have left respondent's home) Is there any information in this interview about which you feel dubious? Yes [] * No [J XIF "YES," ANSWER A AND B: A. Which information do you question? B. Why are you doubtful about it? 2. In general, how accurate would you say the cost data are? Inaccurate Fairly accurate Completely accurate 0 0 0- 3. Had this family completed a self-administered questionnaire? Yes [] * No (J X¥IF "YES": What was respondent's reaction to your asking for a personal interview after (he) (she) had already provided data on the self-administered form? (PLEASE DESCRIBE IN DETAIL. IF ANY PROBLEMS OR RESENTMENT AROSE, EXPLAIN WHAT MADE YOU THINK THERE WAS RESENTMENT, WHY RESPONDENT WAS ANNOYED AND HOW YOU HANDLED PROBLEM. ) 4. ANSWER IF RESPONDENT DID NOT FILL OUT SELF-ADMINISTERED QUESTIONNAIRE: Those respondents who were not asked to return self-administered questionnaires received a letter from Public Health Service. The letter said that a second interviewer was coming and asked them to check their records before the second interview. A. When you called for an appointment interview, did this respondent remember receiving this letter from Public Health Service? Seemed to remember getting letter [J Could not remember getting letter [J B. Was there any indication that respondent had checked records and prepared information in advance of interview? Yes, Seems to have checked records [J No, Probably did not check records [J] 5. Was there any indication that respondent had discussed any of the medical cost items with other household members BEFORE YOU INTERVIEWED HIM? Yes, had obviously discussed[] No indication of discussion[] 6. A. Which person was your respondent? That is, who gave most of the information? B. Did anyone else help provide information for the interview? Yes [J * No OJ *IF "YES," ANSWER (1) AND (2): (1) Who helped? (2) On which parts of the interview did someone else help? 7. Was there any evidence that you could have gotten more accurate information in any area 1f you had interviewed some other member of the family? Yes [J * No OJ *IF "YES," ANSWER A AND B: A. To get what information? B. Who should have been interviewed and why? 8. Did respondent refer to or check bills or records during any part of interview? Yes (J * No (J X¥IF "YES": Which bills or records did respondents check to give you information? (PLEASE DESCRIBE IN DETAIL.) 9. Please attach an extra page (including family number) to give us any additional information which you think would help us to understand the interview or to get a clearer picture of this family's medical experiences and expenses. 10. Date Interview Completed: 11. Interviewer's Signature: 45 NORC 429-3 10/60 HOSPITAL SUPPLEMENT Family No.: Respondent's Name: Name of Hospitalized Member: Hospitalized Member No.: 1. What made it necessary for (PATIENT) to go to the hospital? (ENTER KIND OF ILLNESS, ACCIDENT OR CONDITION. ) 2. A. Was (PATIENT) in the hospital for (this condition) (these conditions) on Just one occasion since last or was (he) (she) there for this condition on more than one occasion? One occasion « «+ . + 30-1 Two occasions «+ « « «+ + 2 Three or more occasions 3% *USE ANOTHER HOSPITAL SUPPLEMENT TO RECORD SEPARATE DATA ON 3rd, 4th, ETC. HOSPITALIZATIONS. B. About how many days did (PATIENT) stay in the hospital (the first time) (the second time), and what kind of room did he have--private, semi-private or ward? (ENTER, IN THE APPROPRIATE SPACES BELOW, THE NUMBER OF DAYS PATIENT STAYED IN EACH TYPE OF ROOM. IF "EMERGENCY ROOM ONLY" OR IF "OTHER OUT-PATIENT CARE ONLY," CHECK APPROPRIATE BOX.) First Hospitalization Second Hospitalization Kind of Room Enter No. Days Kind of Room Enter No. Days Private Private Semi -private Semi-private Ward Ward Emergency room only « « o + « + oo so + + ++ oo Emergency room ONLY « « « « oo « oo oo + 0 4 O Other out-patient only « « « + + + + + + « » « « [1 * Other out-patient only « « « « « + » + «+ + « « « [ * *IF PATIENT WAS JJOT ASSIGNED A BED, BUT RECEIVED ONLY NON-EMERGENCY CARE AT THE OUT-PATIENT DEPARTMENT OR CLINIC, SKIP TO Q. 14 AND ENTER CONDITION LISTED IN Q. 1 ON MAJOR CONDITION SUPPLEMENT FOR LATER QUESTIONING. 3. ASK QUESTION 3 SEPARATELY FOR EACH HOSPITAL STAY (Q. 2-A). RECORD DATA ON FIRST AND SECOND HOSPITALIZATION BE- LOW AND USE ANOTHER SUPPLEMENT FOR ANSWERS ON THE THIRD, FOURTH, ETC. HOSPITALIZATIONS. A. Now I'd like to ask you a few questions about the hospital expenses not counting any separate doctor bills or surgeon's fees--(On the first, second hospitalization) Were any of the hospital expenses paid directly to the hospital by any of the following groups--by Blue Cross, some other hospital plan or insurance com- pany, an employer of someone in the family, Workmen's Compensation, a non-profit or government agency? (CIRCLE "YES" OR "NO" CODE FOR EACH ITEM IN PART A.) bol Hospitalization Re Hospitalization Yes No Apount Yor No fount (OL) Blue CIO88 + + » s ws # rs sv 0 +s sw ww +s « » Sld% 2 3 3* 4 (2) Another hospitel plan or insurance company . . . 32-1% 2 § >* 4 (3) An employer + + « « « os a os 4 a + woe woo. 331% 2 $ 3* 4 $ (4) Workmen's Compensation « + + « « « oo oo « + «341% 2 § * 4 $ (5) Non-profit agency (SPECIFY) 351% 2 § 3* 4 (6) Government agency (SPECIFY) 36-1% 2 $ 3* 4 § (7) (ENTER TOTAL FOR EACH HOSPITALIZATION) + + + « + « « « « + § $ ¥B. FOR EACH "YES" IN A ABOVE: And how much altogether did (GROUP OR AGENCY) pay the hospital? (ENTER IN APPRO- PRIATE SPACE UNDER B ABOVE.) 46 4. (In addition to this) Still not counting any doctor's or surgeon's fees, how much did you have to pay the hos- pital--including anything you may have collected directly from any insurance group con the (first) (second) hos- pitalization? First « « « «+ $ Second . . . $ 5. Then the total hospital bill for the (first) (second) stay amounted to First Second (Q. 3-B + Q. 4) $ $ Is that right? YE8 « +» wm ¢« 2 wo ww #2 & Xk IF NO, MAKE NECESSARY CORRECTIONS IF TWO HOSPITAL STAYS, BE SURE TO INCLUDE BOTH ON THIS SUPPLEMENT. IF MORE THAN TWO HOSPITAL STAYS, DON'T FORGET TO FILL OUT ANOTHER HOSPITAL SUPPLEMENT! ! 6. A. Did (PATIENT) have any kind of operation or have any broken or dislocated bones set while in the hospital (that time) (either time)? Yes « oo vw v5 + 2» ww » 3s 1% No + «ov 0 0 0 0 0 0 ou 2 _*IF "VES," ASK B AND C: B. Was there more than one kind of operation (bone setting) or only one? More than one « « « « « . 38-1 Only one + « « « « o o 2 C. What kind of operation(s) or bone setting(s) was (it) (the first one) (the second one, etc.)? First Second, 47 7. ASK ONLY IF THERE WAS SURGERY OR A DELIVERY ("YES" TO Q. 6-A): Now I'd like to ask you a few questions about any doctor bills you may have had in connection with the (first) (second) operation. A. Who was the doctor who did the (first) (second) operation (set the broken bones)? NAME: 1st NAME: 2nd B. Was there a doctor's bill in addition to the hospital bill for the (first) (second) operation? First Second Yes «. . . 39. 1%* 40- 1%* NO «+ # » 2% 2% *IF "NO," ASK C. THEN SKIP TO Q. 8. C. Why not? 41- IF "YES," ASK D - G: D. Was any of the surgeon's bill paid directly to the doctor by any of the following groups: Blue Shield, another surgical or medical plan or insurance company, an employer of someone in the family, Workmen's Compensation or a non-profit or government agency? lst eration 2nd Operation D. E. D. E. Yes No Amount Yes No Amount (1) Blue BRIE « + wv 4% + + ABE 4% ww we a 42-14 2 § a 5 $ (2) Other surgical, medical plan or insurance . . . 43-1# 2 § 4 5s $ (5) Bployer ws + s+ 5 5 50 + 8 Fw bs 6 EE 4 4.1% 2 $ aft 5 $ (4) Workmen's Compensation + « « « « + + & oo 4 45.1% 2 $ 4 5 $ (5) Non-profit agency (SPECIFY) 46-1+ 2 $ 4+ 5 $ (6) Government agency (SPECIFY) 47-1# 2 § a 5s $ #E. FOR EACH "YES" IN D ABOVE: And how much did they pay? (ENTER IN APPROPRIATE SPACE UNDER E ABOVE.) F. ASK EVERYBODY WHO SAID "YES" TO Q. 7-B: (In addition to this) How much did you have to pay the surgeon (the first time) (the second time) including anything you may have collected directly from any insurance group? First $ Second $ G. ‘Then the total charge for surgery (the first time) (the second time) amounted to (AMOUNT IN E + F)-- First $___ ~~ Second §$ Is that right? YES « a vv vv eee eee ad IF NO, MAKE NECESSARY CORRECTIONS 8. ASK EVERYONE: Altogether, about how many doctors and osteopaths has (PATIENT) seen in the past year about this condition--both before, during and after (he) (she) was in the hospital? NUMBER OF DOCTORS AND OSTEOPATHS: IF NONE, SKIP TO Q. 11 BELOW. A. How many times did doctors and osteopaths visit (PATIENT) while (he) (she) was actually in the hospital, approximately? HOSPITAL VISITS: B. Since last year at this time, how many times did a doctor or osteopath come to the home to see (PATIENT) about this condition? HOME VISITS: C. During the last 12 months, how many times did (PATIENT) see a doctor or osteopath about this condition at the doctor's office? DOCTOR'S OFFICE VISITS: D. (Besides this) And how many times did (PATIENT) go to a doctor's office for shots, X-rays, or some other kind of test or treatment for this condition from a nurse, technician or another assistant when (he) (she) DIDN'T SEE THE DOCTOR AT ALL? OTHER DOCTOR VISITS: E. And how many times did (PATIENT) go to an out-patient clinic or emergency room at a hospital, or to an in- dustrial clinic, or some other clinic for examination, tests, or care in connection with this condition 2? eis past year! CLINIC VISITS: 9. A. Now could you tell me if any of these doctor visits were paid for directly to the doctor by any of the fol- lowing groups--Blue Shield, another medical plan or insurance company, an employer of someone in the family, Workmen's Compensation, a non-profit or government agency? (CIRCLE "YES" OR "NO" CODE FOR EACH.) A. Yes CG) Blue Seid « vv +» v #05 + v2 Ew rs BHT ys Paw vm wees pw ALE (2) Another medical plan or an insurance COMPANY « s o + a + « + = + « oo o « oo 49-1% (3) Buployer « + s + 5s w# 58 2 $4 % 4 ww ws svewwiwrvewes ews BOIE (4) Workmen's Compensation + + + + « « « « « & ¢ « ¢ + ¢ ¢ os + os 5s 55 ss ss Slal¥ (5) Non-profit agency (SPECIFY) 52.1% (6) Government agency (SPECIFY) 53.1% B. No Amount 2 $ 2 $ 2 $ 2 $ 2 $ 2 $ . $ *B. FOR FACH "YES" IN A ABOVE: And how much did (EACH) pay the doctor? (ENTER IN APPROPRIATE SPACE IN B ABOVE) 10. In addition to what you've already told me about, how much did you have to pay the doctor for: ASK A - E AS THEY APPLY: A. IF ANY HOSPITAL VISITS: His hospital visits? B. IF ANY HOME VISITS: His home visits? C. IF ANY OFFICE CARE BY DOCTORS OR NURSES: The office visits, including all charges for ex- aminations, shots, tests, X-rays, and things like that which either the doctor, his nurse, or assistant did? D. IF ANY CLINIC VISITS: And how much did you pay the clinic for the clinic visits? E. So that the total amount you had to pay for these visits was (ADD A - D), including any- thing you got directly from any insurance . . « « « « « « « «0 tt 0 0 0 e000 0. Yes . + «+ + [STN IF NO, MAKE NECESSARY CORRECTIONS 11. A. Did (PATIENT) have any other expenses in the hospital you haven't already told me about that were not covered on the hospital bill, but were billed separately--things like (READ THROUGH LIST, CIRCLING "YES" OR "NO" CODE FOR EACH ITEM.) A. B. Yes No Amount (1) An anesthetist's fee that was NOT charged on hospital bill? . . . + « « + « + . 541% 2 $ (2) A pathologist's SEPARATE charge for laboratory tests? . « « « « + + + + « « + . 551% 2 $ (3) A radiologist's SEPARATE charge for X-ray tests or treatments? . . . . . . . . 561% 2 $ _ (4) Special hospltal NUrSINE? « + + « + + « + + + + + + + 2 2 ss + ss 0 +s 6 0+ o 5TA* 2 $ (B) ORygen® + i vw 5% ¢ so 3 9 4 8 B 4% + + ba Ei +s um wt vue we + vw » SBE 2 $ (6) An ambulance? « o + + « + + o « + + + + 4 a 4 4 at a ee ae eae eae es. 59.1% 2 $ (7) Anything else? (SPECIFY) 60-1% 2 $ ENTER TOTAL § *B. FOR EACH "YES" IN A ABOVE: And how much was (ITEM) including anything insurance paid as well as what you paid? (ENTER ABOVE) 49 12. ASK EVERYBODY: A. (a) (2) (3) (4) (5) (7) (8) *B, During the last twelve months since , did (PATIENT) have any other expenses for this condition-- that is, not counting what you've already told me about? Expenses for things like: (READ THROUGH LIST, CIRCLING "YES" OR "NO" CODE FOR EACH ITEM.) A. Yes Medicines the doctors or hospital prescribed for this condition? . . . . . . . . . 611% Other non-prescribed medicines for this condition? « + « « + « « ¢ ¢« + + « « « « & 62.1% X-28y HEELET + wv 4 « vw 8 2 Hy GW HE Le HE Ys rE EE WEE ws ow OTE Other special tests like blood tests, electrocardiograms, urine analyses, and soon? 64.1% Special treatments like X-ray treatments, hear or diathermy treatments, massages, GOA BO ONT + o 5,» + » # & 5's 2 3 » 5 8 + +» 5% #4 # s +a 4 8s sown ss ass BOLE Home nursing care (for which you were charged)? + « « + « + o « « « = « o + « « « 66-21% Any medical equipment or appliances like braces, crutches, wheel chair, a vaporizer oranything like that? « « vw w 5 « si & 5 & & vi Hv 0 & 8 § ® wis 8 oo w 8 s & « & BT1¥* Anything else? (SPECIFY) 68-1% B. No Amount 2 $ 2 $ 2 $ 2 $ 2 $ $ $ 2 $ ENTER TOTAL $ FOR EACH "YES" IN A ABOVE: And what was the cost of (ITEM)--including anything that insurance covered, as well as what you paid? (ENTER ABOVE.) 13. ASK EVERYBODY: let's check over these expenses for this condition again to make sure I have them all correct. (ENTER Now AMOUNTS BELOW AND INCLUDE EXTRA SUPPLEMENTS IF THREE OR MORE The £06aL DILL WAS + « « « « + + os s + = + + + os o + + + + 2 s so v 2 vs + 8 so + +» $ The $050) DL1L fOr BUTBELY WBE + « wv o ¢ # o 5 3 » 5 ss 8» 5 & & ¢ 8 si 8 5 38 % » $ The total for doctor visits paid directly Dy insurance was « « « « « » « « « « « «+ $ The total doctor visite YOU DPAIA WEIE « « o « # ¢ + & o sv 3 3 4 6 + & & 2 #8 + = $ The total for other expenses in the hospital you had tO Pay Was + « « « « « « + « » $ And the total for all other expenses you just mentioned was + « « + « « o « « + + $ Toot moles a EraBA YOUBL OF + + o 3 o + 6 5 0% 8 s. 8 #8 % 6 bh 6% 5 ¢ WB @® 5» $ Does that sound right? Yes « + « ¢« v=» HOSPITALIZATIONS FOR THIS CONDITION.) . wn 7-G . 10-E 11 ceppee © cee eee. 1¥ IF NO, MAKE NECESSARY CORRECTIONS IMPORTANT! ENTER GROSS TOTAL FROM Q. 13 IN APPROPRIATE LINE ON Q. 14 OF THE MAIN CRITERION QUESTIONNAIRE BEFORE GOING ON TO ANOTHER SUPPLEMENT OR BACK TO MATIN QUESTIONNAIRE. Q. 2 OR Q. 3. SUPPLEMENT. DON'T FORGET TO ASK QUESTION 14 ON THIS 14. ASK EVERYBODY: Now was (PATIENT) in the hospital for any other illness, accident, or condition during these past 12 months? 50 IF YES: Fill out another Hospital Supplement for this person. IF NO: Return to Q. 2 or to Q. 3 in the Main Questionnaire, depending on where you left off. Family No.: NORC 429-4 MAJOR CONDITION SUPPLEMENT 10/60 Respondent's Name: Patient's Name: Patient's Condition: No.: 1. What sort of care or treatment did (PATIENT) get in connection with (CONDITION)? (For example, a boil lanced, shots, heat or physical therapy treatments, a bone set, etc.) 2. Altogether, how many doctors and osteopaths has (PATIENT) seen in the past year for this condition--that is, since ? * *IF NONE, SKIP TO Q. 4-B. 3. Since last year, how many times did a doctor or osteopath come to the home to see (PATIENT) about this condition? * *IF ANY VISITS, ASK Q. 3-A. *IF NO VISITS, SKIP TO Q. 4. A. How much were your total charges for these home visits--including anything that in- surance may have paid you or your doctor? $ 4. A. During the last year, since , how many times did (PATIENT) see a doctor or osteopath about this condition at his office, at a clinic, or a hospital out-patient department? B. (Besides this) And how many times did (PATIENT) go to a doctor's office when (he) (she) didn't see the doc- tor at all but got shots, X-rays, or some other kind of test or treatment for this condition from a nurse, technician or another assistant? C. IF ANY OFFICE OR CLINIC CARE: How much did the doctor(s) charge for all the care (PATIENT) got at the office clinic)? Let's include all bills for examinations, shots, tests, X-rays and so on given by the doctor or his assistants, and also include anything insurance may have paid you or your doctor. 5. ASK EVERYBODY: A. Did (PATIENT) have any other expenses for this condition in the past year--for A. Be things like: (READ EACH ITEM ON FOLLOWING LIST AND CIRCLE "YES" OR "NO" CODE Any Amount of FOR EACH. IF "YES" TO ANY ITEM- ASK B AND C.) expense? expense Yes No (1) Medicines the doctor(s) prescribed for this condition? . « « « « + « + «+ 1* 2 $ (2) Other non-prescribed medicines? + + « + + vo oo + + + ov 4 4 40a a 1* 2 $ (3) Reavy Lesko? = + + +. 5 6 % § $C AR § + BH BLK SL EWE SO MEE SE 1x 2 $ (4) Other special tests like blood tests, electrocardiograms, urine analyses BNA BO OUT ~ ve 2 ow wie ¢ Sim Gd E83 RATES BER AL LEER EEE I» 2 $ (5) Special treatments like X-ray treatments, massages, heat or diathermy treatments or any other kind of treatment? . . + + « + « «vo 0 0 1* 2 $ (6) Home nursing care (for which you were charged?) « « « + « « + « « o + + « & 1* 2 $ (7) Any medical equipment or appliances like braces, crutches, wheel chair, a vaporizer or anything like that? + + + « 4 «ov 4 4 +o ov vo 0 0 os 0 os 1% 2 $ (8) Any other medical expense for this condition not already covered? (SPECIFY) 1* 2 $ *B., FOR EACH ITEM FOR WHICH ANY EXPENSES INCURRED: What were (PATIENT'S) total expenses for (ITEM)--including anything that insurance may have paid? (ENTER AMOUNT ON APPROPRIATE LINE UNDER B ABOVE. ) C. IF EXPENSES INCURRED FOR ANY OF ABOVE ITEMS, ENTER TOTAL HERE AND CHECK: That makes a total of $ for other expenses. Right? $ 6. That makes: A. For home VISItS « « o « oo oo ot 0 + + 4 o + to oo 4 vt ee ee eee $ Q. 3-A Bi Tor office or cline vislhs « » +" w « » » ws #5 & v ww x 4 3 www vw $ Q. 4-C OC. For ollor SRpEnSEs « « « ono + + CH 4 ¢ 3 A HH 8 + 2 HE 8 Fy EWES $ Qe B5 - 9 ed Le cd | | | IL TE | — @ 0 I 2 3 0 1 2 3 0 2 3 o 1 2 3 0 I 2 3 8 3 o © 8 © 150 CASE #30 CASE #3 CASE #43 CASE #44 CASE #48 2 3 Time After Challenge (in hours) Figure 1. Response to the standard glucose tolerance test, 10 men, Milan, Phase II. The findings trom this phase of the study are indicated in detailed table IV and figure 1. While they are consistent with the limited data already in the literature, it must be admitted that it was startling to see them. It was clear that with the resources available, stable estimates of the amount of difference associated with different procedures could not be obtained, even if the specific group under study were a true population sample, but that it might be possible to establish the existence and direction of such differences. It was decided to continue with the original pro- gram of study. Phase III The final study group consisted of 24 indi- viduals. The following eight procedures were under study: A. Overnight fast, challenge of 100 grams B. Overnight fast, challenge of 50 grams C. One hour after meal, challenge of 100 grams D. One hour after meal, challenge of 50 grams E. Two hours after meal, challenge of 100 grams F. Two hours after meal, challenge of 50 grams G. Three hours after meal, challenge of 100 grams H. Three hours after meal, challenge of 50 grams In procedure A blood specimens were taken before challenge, and at %, 1, 1%, 2, and 3 hours after challenge. In the other procedures, blood specimens were taken just before challenge and 1 hour after challenge. Urine specimens were taken about 90 minutes after challenge. Again, the in- stitutional fare was not altered. As in Phase I, all food eaten on the day of the test was recorded and its carbohydrate content estimated. The volunteers were ranked according to their Phase I blood glucose level 1 hour after challenge. All persons with blood glucose levels at or above 110 mg.% 1 hour after challenge were selected for Phase III (fig.2). In the end, 15 participants came from this group and 9 came from the group with 1-hour blood glucose levels less than 110 mg.% in Phase I. The 24 participants were placed in 8 groups. Each group consisted of one person chosen ‘at random from the high end of the scale, one from the middle and one from the low. The group was then assigned at random to one of the eight procedures. This initial assignment determined the order in which these three persons moved through the succession of procedures. The final assignments are shown in detailed table II. Since each procedure was duplicated, each person was to be challenged 16 times during Phase III. With only minor exceptions, challenges were given at weekly intervals. During all of Phase III only 1 episode of vom- iting was noted within the 24 hours after glucose administration. Subject number 46 had this ex- perience 6 hours after a 50-gram dose. However, he had had bouts of epigastric discomfort and vomiting for years. Indeed, his fifteenth pro- cedure was deferred because such an episode had begun on the preceding day. There were several lapses in the execution of this design. In spite of an effort to solicit only men who would be expected to remain at Milan for the entire period of the study, 2 of the original 24 participants were transferred and 1 was paroled before completing the full series of Phase III, The two transferred men were replaced at random from the remaining volunteers. Several men were 4 Blood Glucose Level (mg. %) Cases O 50 100 150 200 250 30 03 43 ANS Participant, Phase IIT 29 [C1 Non-participant, Phase IIL 05 e Discharged before 24 completion of study 3) ee Veins unsatisfactory for venipuncture 34 e's Chronically ill 40 37 06 58 02 25 39 22 1 10 41 08 04 a7 07 32 38 42 27 23 19 ol 50 09 33 36 26 59 a8 a6 49 35 28 45 a4 21 20 50 100 150 200 250 Blood Glucose Level (mg. %) } 0 Figure 2. Blood glucose levels 1 hour after challenge, 45 men, Milan, Phase |. given the wrong dosage one week. The correct dosage was administered to these men after com- pletion of the remaining scheduled tests. Almost all specimens were obtained precisely as intended, but occasionally the time between meal and chal- lenge was different from that planned. Whenever a significant lapse was noted (see detailed table II), an effort was made to supply the correct procedure at a later time during Phase IV. Phase III started January 24 and ended May 8, 1962, Individual test findings are recorded in detailed table III. Phase IV The final 6 weeks afforded time and labora- tory support to perform only three fasting tests on each of the remaining 22 participants in Phase ITI. It was decided to use the following three procedures, each after 3 days of at least 250- 300 grams of carbohydrate daily: 1. A standard 100-gram oral glucose tol- erance test (SGTT). 2. A cortisone glucose tolerance test (CGTT). A uniform procedure was adopted whereby a dose of 62)% mgs. of cortisone was administered 8% to 9 hours and again 2% to 3 hours before a morning fasting challenge of 100 grams of glucose. In contrast, the original procedure of Fajans and Conn® would have called for 50 mg. doses instead of 62% mg. doses of cortisone for individuals under 160 pounds in weight (subjects 21, 29 and 48) and the glucose dose would have been 1.75 grams per kilogram of "ideal body weight". 3. The prednisone glycosuria test (PGT) as described by Joplin, Fraser, and Kee- ley? followed directly by another 100- gram glucose tolerance procedure. The mean SGTT values were to be compared with means from Phase II and Phase III (procedure A) to check the hypothesis that there should be no difference; i.e., that added carbohydate had been unnecessary. The CGTT and PGT were included to explore their potential applicability and useful- ness for population studies. The CGTT would also provide another means for clinical classification of carbohydrate tolerance. It appeared infeasible to perform an intravenous glucose tolerance test. Laboratory Methods Blood specimens were shipped on water ice from Milan at the end of each day of tests to the Diabetes Field Research Unit in Brighton, Mas- sachusetts, for determination of glucose concen- tration by the Somogyi-Nelson Method.!® A re- view of technical variability encountered during the study (Appendix I) supports a conclusion that the work of this laboratory was reliable and consistent from week to week and that shipment did not significantly alter the results. Urine samples were tested by the field staff at Milan with a glucose oxidase impregnated tape (''Tes- Tape" produced by Eli Lilly Company, Indianap- olis).!! Quantitative urinalyses during Phase IV were done in Brighton by the Froesch and Renold method, 12 BLOOD GLUCOSE LEVEL 1 HOUR AFTER CHALLENGE The primary purpose of the study was to see how blood glucose levels 1 hour after challenge were influenced by differences in the amount and time of glucose challenge. One method of evaluating this is the comparison of mean 1- hour blood glucose levels in response to each procedure, averaged for all 24 persons in the main study (Phase III). The mean value for each procedure is shown in table 1 and figure 3. With table 2, which gives the standard deviations of response, the means reveal several of the major findings. As expected, the response to a 100-gram oral glucose load was greater than to 50 grams. The difference between the mean of all 100-gram and of all 50-gram procedures combined was 9.4 mg.%. If the mean levels for individuals at different times of challenge are considered, there are al- together 96 comparisons of a 50-gram with a 100- gram challenge. The level was greater after a 100- gram challenge in 65 of these 96 comparisons. In other words, in the majority of instances the 100-gram challenge leads to higher 1-hour blood glucose levels than the 50-gram. Futhermore, the mean 1-hour level was higher with a challenge of 100 grams than with a challenge of 50 grams, whether the glucose load was given to a fasting individual or was given Effect of Challenge Dose 150 [* V// 50-gram challenge gas 100-gram challenge # o E 100 ® > o 4 o ” o 0 2 © © o 2 @ 50 c o @ = 0, | hour 2 hours 3 hours Overnight Time Between Meal and Challenge Effect of Time Between Meal and Challenge oo oA 7 _ i | ga | hr. 2hrs. 3hrs. Over- I he 2hrs. 3hrs. Over- night ht nig Figure 3. Mean blood glucose level 1 hour after challenge, by various procedures, 24 men, Milan, Phase Ill. 1, 2, or 3 hours after a meal. However, the difference was not statistically significant for the fasting test. For both the fasting tests and the tests given 1 hour after a meal, only 13 of the 24 persons had a higher response to the 100- gram challenge than the 50 gram. When the challenge was given 2 or 3 hours after a meal, the 1-hour level was, respectively, 11.0 mg.% and 16.5 mg.% higher with a 100-gram challenge than with 50; and was higher inthe first instance for 18 of the 24 persons and in the second for 21 of the 24 persons. These observations refer only to levels 1 hour after challenge. Except where a 100-gram challenge was administered after an overnight fast, this study undertook to measure blood Table 1. Mean level of blood glucose before challenge and 1 hour after challenge: 24 men, Milan, Phase III Mean level (mg.%) Time from meal to challenge Before challenge 1 hour after challenge of Total 100 grams 50 grams 100 grams 50 grams 1 hour-======memecca=--- 89.5 87.8 91.2 106.8 100.9 2 hours==============-- 84.2 83.4 85.0 113.8 102.8 3 hours-=--====m=ce==-- 77.0 76.8 77.3 118.6 102.1 Overnight----=========- 78.4 77 .4 79.3 109.4 205.2 glucose levels only before challenge and 1 hour after challenge. It is conceivable therefore that the peak response was as high to the 50-gram challenge as to the 100-gram challenge, but that the peak came at a different time. It will be shown, however, thaturine glucose concentrations tended to be higher after a 100-gram challenge than after 50, which would seem to argue for generally higher levels of blood glucose after 100 grams. The mean blood glucose level 1 hour after a 50-gram challenge appeared to be the same whether the challenge was given to a fasting individual or 1, 2, or 3 hours after a meal. On the other hand, the mean blood glucose level 1 hour after a 100-gram challenge was af- fected by the time the challenge was given. If it was given 2 hours after a meal the level was higher than if the challenge was given 1 hour after a meal. The response level was still higher if the challenge was given 3 hours after a meal. While this "trend" was statistically significant in terms of mean levels for the group, it was not Table 2. compelling for individuals. In fact, it was noted only in six individuals. In 18 cases, however, the response to a 100-gram challenge given 3 hours after a meal was greater than the response to the same challenge given 1 hour after a meal, so that we are justified in considering this effect of time after meal as generally true. It does not follow, however, that the level after a 100-gram challenge is higher following an overnight fast than it is when challenge is ad- ministered 3 hours after a meal. In fact, the data suggest that the level is lower. This difference, however, is not statistically significant and is found in only 14 of the 24 persons tested. Figure 4 illustrates specific test results for three individuals selected from the low, middle, and high portions of the response scale. Each of the eight procedures studied was per- formed twice on each subject in successive weeks. For the eight procedures taken as a group, the differences in variability of level 1 hour after challenge are not statistically significant. How- ever, this conclusion does not allow for the fact Variation of blood glucose levels before challenge andl hour after challenge: 24 men, Milan, Phase III Time from meal to challenge Variation (mg.%) Before 1 hour after challenge of challenge 100 grams 50 grams Average of absolute differences 12.0 14.8 20.2 12.6 12.3 15.3 11.3 23.1 17.3 6.4 20.5 14.2 Standard deviation of response 10.5 13.4 17.3 11.2 11.4 14.0 10.8 19.9 17.2 549 18.7 12.1 NOTE: If d is the absolute difference between replicates of a given measure and there are n pairs of replicate measures, the average of the absolute differences is 5 d and the standard deviation of response is > 4? n 2n Blood Glucose Level (mg.%) 150 — 100 50 — | 51 -—— 2d Challenge After Overnight Fast 100-gram 100-gram 50-gram 50-gram challenge challenge challenge challenge ) | | _J 0 | 2 3 Challenge After Overnight Fast 150 100 200 150 — Challenge After Overnight Fast CASE #43 Challenge | Hour After Meal 150 ~ 50 CASE #26 Challenge | Hour After Meal 150 |” 100 |— 0 | 0 | CASE #45 Challenge | Hour Atter Meal 200 150 Challenge 2 Hours After Meal 1507] 100 — 50 Challenge 2 Hours After Meal 150 [~ — . Wo SP 100 2” Challenge 2 Hours After Meal 200 [ 150 Challenge 3 Hours After Meal 150 [~ 100 [— 50 |= Challenge 3 Hours Atter Meal 150 [ 100 (— 50 [— Challenge 3 Hours After Meal 200 150 100 100 |— 100 100 Be 50 |— 50 [~ 50|— 0 | | | J 0 J Gd Ohman 0 | 2 3 0 | 0 | 0 | Time After Challenge (in hours) Figure 4. Blood glucose response to specified glucose challenge, Subjects 43, 26, and 45, Milan, Phase Il]. that each variance includes a relatively fixed technical variability and that the statistical test is close to the level of significance. In any event, it is well recognized that technical error tends to make it more difficult to demonstrate differences that are actually present. In this case, the most reasonable conclusion from the data is that (appearances to the contrary) the variability in level 1 hour after challenge is not the same for all eight procedures. Similarly, there is no indication in the data that variability in response is any greater for those persons with high levels of blood glucose than those with low levels. To take a specific example, when a 100-gram challenge is given a fasting individual, the blood glucose level at 1 hour andits variation have a rank correlation of -.05. It cannot even be demonstrated by these data that there is a statistically significant difference in variability of response among different persons. Since both these conclusions are implausible on a priori grounds, they may be modified to this statement: In this specific study group, any differences that did exist between persons in their variability of response at 1 hour after challenge could not be demonstrated with the procedures, the laboratory methods, and the ex- tent of replication that was used. To be conservative then, all statistical tests were performed on the assumption that the varia- bility of level differed from person to person and from procedure to procedure. Table 3. Mean level of blood glucose THE MEAL Persons presenting themselves for examination in the usual survey may arrive after a breakfast, lunch, or dinner of variable carbohydrate con- tent, or in a nearly fasting state, It was decided early in planning this study that it would be impossible to evaluate all these factors with the resources available, and that attention would be focused on the effect of giving a challenge at varying times after a meal. However, a de- scription of each participant's previous meal was obtained each time he presented himself during the main study; and the test assignments were arranged so that where the test was given after a meal, half of the study group would always come in after breakfast and the other half after lunch. The "breakfast" and 'lunch'' groups were quite similar in glucose response. Their mean blood glucose levels in Phase I, when everyone received a 100-gram challenge after breakfast, were 81.0 and 75.5 mg.% before challenge, and 124.4 and 113.4 mg.% after challenge. Similarly, their morning fasting levels during the main study were nearly identical, 79.8 and 77.5 mg.%; and when a challenge of 100 grams was given after an overnight fast, the blood glucose levels rose to the same level in both groups, 110.3 and 108.5 mg.%. The only discordant note is the difference in response to a 50-gram challenge after an overnight fast, the "breakfast group before challenge and 1 hour after challenge, ac- cording to meal preceding challenge: 24 men, Milan, Phase III Mean level (mg.%) 1 hour after challenge of Time from meal to challenge Before challenge 100 grams 50 grams After After After After After After breakfast | lunch | breakfast | lunch | breakfast | lunch 1 hour----==-=mmemm meme meme eee B6.5)] 922.5 100.4 | 113.3 97.2 | 104.5 2 hours------==-=-cecemmenee———— 81.6 | 86.8 105.4 | 122.2 96.4 | 109.2 3 hours----=-==-=-emcecce cca 73.4 | 80.6 110.0 | 127.1 96.9 107.2 NOTE: 12 men received challenges after breakfast, 12 after lunch. rising to 101.8 1 hour after challenge and the "lunch" group to 108.6. However, the weight of the evidence favors the conclusion that under the same circumstances the two groups had essentially the same blood glucose levels. Hence where mean levels after a meal differ, it seems reasonable to attribute most of the differences to the content of the meal or to the time of day. Of course, comparisons between tests after break- fast and after lunch will also reflect any di- urnal rhythm or differences in recent physical work. Table 3 gives average levels of response for these two groups under the various test pro- cedures. Blood glucose levels both before chal- lenge and 1 hour after challenge were higher after lunch than after breakfast. When a challenge was given after lunch, the rise was greater than when the same challenge was given after break- fast. The level 1 hour after challenge, when the challenge (either 50 or 100 grams) was given after breakfast, was lower than when the chal- lenge was given to fasting individuals, although the differences were trivial and not statistically significant, Similarly, with a 50-gram challenge given after lunch, the levels 1 hour after chal- lenge were indistinguishable from those obtained from a 50-gram challenge given after an overnight fast. A 100-gram challenge given after lunch, however, yielded 1-hour levels distinctly higher than did the same challenge given after an over- night fast. All of these differences may, of course, reflect differences between the persons assigned to the two groups (although this is unlikely), but the study was not designed to sort out this kind of factor with great precision. It is worth noting, however, that the carbohydrate intake at breakfast tended to be higher than at lunch, although the kind of carbohydrate eaten at these meals is not the same and may conceivably have different effects on the glucose tolerance test. The range and mean carbohydrate intake during the main study are given for each person in the study group in detailed table IV. URINE GLUCOSE A semiquantitative glucose oxidase tape method specific for glucose was used to test urine specimens collected 1% hours after each of 10 the 16 glucose loading tests performed on each subject. Data are given in detailed table III and summarized in table 4. No negative urine was obtained when the 1-hour blood glucose level was over 160 mg.%, whereas no urine specimen showed even a trace of glucose when the l-hour blood glucose level was below 60 mg.%. Nine persons had positive urine with some frequency (at least 7 times out of 16). Their urine glucose findings may be roughly quantified by using the test scale (1, 2, 3, 4), assigning a value of % for a glucose trace, and zero for a negative urine. A person's response to the repli- cates of one procedure may be combined and com- pared with the parallel statistic for another pro- cedure. If this approach is used to compare all 100-gram tests for these nine persons with all their 50-gram tests, the average score for the 100- gram tests is 0.847 more (the difference having a standard deviation, Sp/ym in the notation of Appendix II, of 0.276). In short, the 100-gram challenge elicited a significantly higher concen- tration of glucose in the urine of these nine sub- jects than the 50-gram challenge. This statement also applies to the 24 persons taken as a whole. THE STANDARD GLUCOSE TOLERANCE TEST The study yielded a large amount of data relating to the standard glucose tolerance test. In the Phase II pretest, 10 men were given 100- gram challenges twice after an overnight fast, and successive blood specimens were taken. Four of these men were not participants in sub- sequent tests, but the other 6 and another 18 of the original volunteers did participate in the main study, where the same sort of test was administered in replicate. After Phase III was completed, 22 of these 24 men were also given a single standard glucose tolerance test preceded by a 3-day period of high carbohydrate intake. Thus, there were 28 men with at least one pair of standard glucose tolerance tests and there were 6 men with 5 standard glucose tolerance tests. These various data are presented in de- tailed table IV and summarized in table 5 and figure 5. They indicate that while the fasting blood glucose level of an individual is most stable, each of the levels between % and 3 hours after challenge has a standard deviation between 12 and 18 mg.%. This variability, of course, compli- cates the evaluation of the glucose tolerance test when the test results fall relatively close to whatever critical values are used for diagnosis. To evaluate changes in clinical status onthe basis of single standard glucose tolerance tests is Table 4. Urine glucose scores: especially hazardous in light of the high vari- ability of the individual tests, It is worth noting that the variability of the ensemble of measurements taken in a standard glucose tolerance test is actually greater than appears from table 5. This may be seen by the following: Add all the blood glucose values for a single standard glucose tolerance test for each of the 28 pairs of 3-hour tests performed during 24 men, Milan, Phase III Total urine glucose scores Case number All tests 100 gram tests |50 gram tests 0l-==-mmm mmm mmm eee 4 0 4 O4mremmmmm mmm c creme ——————— 1 2 0 05=-—mmmm meme eee em e 0 0 0 (OF EE 1/2 1/2 0 10-==mcmmmmmmmcm ee e ee mmm ee eem 0 0 0 1lecmmmmmm mmr meee 1/2 1/2 0 20 == mmm meme emma 9 51/2 31/2 2] mmm mmm meme mmm emma 38 20 18 A EEE EEE EEE EE EEE PEE EEE 1/2 1/2 0 26=m=mmmmmmmeme cece 0 "0 0 28mm mmm mmm mmm meme emma 46 27 19 AR et EE EE EEE 11/2 0 11/2 Fm hm SH SR A 1/2 1/2 36mm meme e nee 31 1/2 16 1/2 15 femme meme 4 f3mmmm mmm emma 11/2 1 1/2 ffmmmmmm mmm mmm mmm mmm ————— 20 13 7 4S mmm mmm meme eee 29 1/2 15 142 14 LR Ee tattle 9 6 1/2 21/2 4] mmm mmm meme eee eee eee 0 0 0 48mm mmm meee meme eee 71/2 5 1/72 2 49 mmm mmm eee e em 1 1/2 0 11/2 50mm mmm mmm mmm meee meme eee 0 0 0 DR EE EE EEE EE EE EE EE EEE 1 1 0 NOTE: Urine determinations are made 90 minutes after challenge. Negative urine is given a score of 0; trace, }, readings of 1,2,3, or 4 plus are scored 1,2,3, or 4. There were 8 tests with a challenge of 50 grams, 8 with a challenge of 100 grams, 16 tests altogether. Table 5. Variation of blood glucose levels on standard glucose tolerance tests ac- cording to time after challenge: 28 men, Milan, Phases II and III Variation (mg.%) Time after challenge Average of Standard absolute deviation differences | of response 0 hour--------- bob 3.8 1/2 hour------- 16.8 15.2 1 hour--------- 20.0 18.0 1 1/2 hours---- 14.1 12.5 2 hours-===---- 11.9 13.5 3 hours---=----- 16.1 14.7 NOTE: If d is the absolute difference between replicates ofa given measure and there are n pairs of replicate measures, the av- erage of the absolute differences is 1/n ¥ d and the standard de- viation of response is s 42 2n Tests in Phases II and III were given without a special prepar- atorv diet. Phases II and III. The standard deviation between replicate sums is 42.0 mg.%. If it is assumed that the variation at one time after challenge is independent of the variation at any other time, the figure computed from table 5 would be 32.8 mg.%; the difference is statistically significant. As already noted, the glucose tolerance tests done in Phases II and III were undertaken with- out any special preparatory diet. Such diets were developed to correct any possible caloric or carbohydrate deprivation, either of which tends to reduce tolerance to a standard challenge! As a special check on this factor, a series of standard glucose tolerance tests were performed during Phase IV on all 22 remaining persons who had participated in the main study. Some of the group were given these tests during 1 week of Phase IV; the remainder were given the tests the following week. The test with preparatory diet was done only once on each person, It will be seen from table 6 that the levels for tests given without the 3-day preparatory diets were, if anything, lower than the comparable results with the preparatory diet, although the differences 12 were not statistically significant. Hence, it can be argued that the normal prison fare consti- tuted preparation enough. A record was made of the carbohydrate con- tent of the last meal for each subject in each of his nonfasting tests, The range and mean carbohydrate content of each subject's meals in Phase III are given in detailed table V, It will be noted that the meals were generally more than adequate in carbohydrates, suggesting that the subjects were actually receiving a diet resembling the customary glucose tolerance preparatory diet. During Phase III, only two subjects gained or lost more than 4 pounds. Selected discordant 1- hour blood sugar values from duplicate tests re- vealed that about as many are associated with differences in the carbohydrate content of the respective meals in the same direction as with differences in the opposite direction. Thus, vari- ation in the recent carbohydrate intake does not seem to be a suitable explanation for dis- cordant blood glucose values. In Phase IV, 2 weeks after the standard glucose tolerance test, a cortisone glucose toler- ance test was performed and evaluated in accord with the method of Fajans and Conn.® This test and the Phase III and Phase IV 3-hour glu- cose tolerance tests are summarized clinically Table 6. Mean level of blood glucose on standard glucose tolerance tests with and without 3-day preparatory diet: 24 men, Milan, Phase III and IV Mean level (mg.%) Time after challenge Without With preparation | preparation (Phase III) | (Phase IV) 0 hour---=------ 77.6 78.7 1/2 hour===--=-- 111.8 119.0 1 hour---+«-=e-- 108.3 107.3 1 1/2 hours-=--- 97.1 102.2. 2 hours======== 91.5 102.1 3 hours-======-= 68.5 76.5 Blood Glucose Level (mg. %) 200 150 100 No Preparatory Diet em Phase TI-First test CASE #20 — . — Phase II-Second test = = —= Phase TII-First test 200 wenerrneen Phase III-Second test 3-Day Preparatory Diet = = — Phase IV 150 |- 100 50 50 0 | 1 1 1 ds Q 1 | 1 iS 0 | 2 3 0 | 2 3 CASE #21 CASE #44 200 [~ 200 150 150 100 CASE #43 100 50 50 0 1 1 | I ———ds 0. l | L l 0 | 2 3 0 | 2 3 CASE #28 CASE #48 250 [ 250" 200 150 100 200 — 150 100 50 50 . ~. \ 0 L 1 l L ro 0, 0 | 2 3 0 Time After Challenge (in hours) Figure 5. Response to the standard glucose tolerance test, § men, Milan, Phases II, Ill, IV. for each subject according to established criteria in detailed table VI. Considering as "abnormal any fasting blood glucose levels above 100 mg.% during Phase III and clinically definite or suspect abnormalities of either of the two procedures in Phase IV, the following four examinees are seen to have manifested definite or suspect evidence of decreased carbohydrate tolerance at least twice: 01, 21, 26, 28. RANKING INDIVIDUALS Ultimately, any glucose tolerance testis eval- uated by a decision that the blood glucose level is either high or low. Hence, if one glucose toler- ance test ranges a set of persons from low to high in the same order as another test, itmay be con- sidered as equivalent to that test. Ifa rank corre- lation of 1.00 is found between two tests, this means that the individuals are ranked in exactly the same order by both tests. If the rank corre- lation is 0.00, there is no similarity at all in the order. Where only 24 persons are being evaluated, a rank correlation of 0.34 is indistinguishable from no correlation. For presentpurposes, nega- tive correlations are equivalent to none. Table 7 exhibits the mean blood glucose con- centration 1 hour after 50-gram and after 100- gram challenge for each person in Phase III and the corresponding ranks. The rank correlation between the average of all 100-gram procedures and the average of all 50-gram procedures is 0.93. Inspection of table 7 and of figure 6 con- firms that the 100-gram and 50-gram procedures do, indeed, rank individuals with remarkable con- sistency. This does not answer the question of how well a single casual 1-hour test compares with the deliberate test experience. For this purpose the Phase I data may be used. As these were the initial tests performed on each subject and were done at various times after a meal, they are quite comparable to tests performed in epidemio- logic surveys. The rank correlation of the aver- age of all 100-gram procedures in Phase III with the single 100-gram testin PhaseIwas 0.63. A similar inquiry may be made of the re- lation of blood glucose levels before and after challenge. There is, of course, a drop in blood 14 175 T T 150 wl Mean for all 100-Gram Tests (mg. %) ~ wn I | 100 — » -1 Blood Glucose Level | Hour After Challenge 75 | | : 75 100 125 150 Blood Glucose Level | Hour After Challenge Mean for all 50-Gram Tests (mg. %) Figure 6. Correlation of mean blood glucose level after the 50- gram challenge with mean blood glucose level after the 100-gram challenge, 24 men, Milan, Phase Ill. glucose levels (before challenge) from 1 to 2 to 3 hours after a meal. Nonetheless, it is con- ceivable that if due allowance is made for shifts in the scale, a casual blood specimen obtained without any deliberate preparation or delay could measure glucose tolerance quite well. Certainly such a test procedure would have distinct advan- tages for survey work. In terms of this study, the question can be phrased: how does the blood glucose level before challenge relate to the blood glucose level after challenge, and how is this relationship affected by time since last meal? The rank correlations between levels before challenge and 1 hour after challenge are given in table 8. Needless to say, this study does not allow the complexities of response to carbohydrate challenges in close succession to be evaluated (for that is what a meal followed by a glucose drink amounts to), but it can be said that when the challenge is given within 1 or 2 hours after a meal the corre- lation of ranks before and after challenge is quite striking. Table 7. Mean level of blood glucose l hour after challenge and ranking of individuals: 24 men, Milan, Phase III Mean level (mg.%) Ranking Case number 0 100 50 100-gram 50-gram -gram -gram All tests tests teats All tests teshs oe Ome mmm 118.3 122.2 114.4 18 18 20 OF ww mm mmm mmm mows 87.7 92.3 83.1 6 7 3 05--=mmmmmmmm—————— 86.7 88.8 84.6 3 3 6 O08wemmmmmm mm ————— 106.8 117.8 95.7 13 15 9 A —— 87.6 90.8 84.4 5 6 5 TL vw i em 86.9 90.3 83.5 4 5 4 mmm wt im mim 109.4 115.1 103.7 15 14 15 0 ei so lo 139.2 142.4 133.9 22 23 23 25= = mmm mn mmm ———— 100.7 106.9 94.5 11 11 8 D5: on ba ome 114.9 121.1 108.8 17 17 18 DB wir sh so im 122.1 137.0 107.2 19 20 17 2 ini mio mice mmm 96.9 94.9 98.8 9 9 10 3 jm mr mc es mc 114.0 120.9 107.1 16 16 16 him mr 144.1 162.7 125.5 24 24 21 dim meri tii mts 105.3 108.1 102.4 12 12 14 43m remem wm erence 81.6 84.4 78.7 1 1 1 Gly mmm im 107.4 113.0 101.9 14 13 13 45mm mmm mmm mmm 143.3 141.4 145.1 23 22 24 bf =mammmam mmm 131.0 134.3 127.7 21 19 22 BT mmm iimmnis min in 87.8 85.2 90.4 7 2 7 48ewmmnm nme meme 84.4 89.6 79.3 2 4 2 40mm mm nm mmm mm mn 96.8 93.0 100.5 8 8 12 50=mmmmmmm———————— 99.8 100.3 99.2 10 10 11 50mm 126.3 139.4 113.2 20 21 19 It 1s instructive to compare the rank corre- lations just discussed with the rank correlation of replicated standard glucose tolerance tests. When persons who had fasted overnight were given the 100-gram challenge during Phase III, and this procedure was repeated 1 week later, the rank correlation between their levels 1 hour after challenge was 0.68. This is not surprising, given the high variability of response, but it does raise the question whether this generally accepted standard procedure has much inherent advantage over any of the other seven procedures under investigation. The urine tests donot lend themselves equally well to rank correlation techniques, since the majority of persons in the study seldom if ever "spilled" glucose into their urine even after an 100-gram challenge and were consequently tied in rank. However, using the scoring system previously described, it is evident by comparison of tables 4 and 7 that the nine persons who tended to "spill" after challenge usually ranked high in blood glucose level after challenge. Figure 7 shows the relation of average blood glucose level 1 hour after challenge to the composite urine glu- cose score for each of the 24 persons in Phase III. Table 8. cose levels after challenge: Rank correlation of blood glu- before challenge and 1 hour 24 men, Milan, Phase ITI Glucose challenge Time from meal to challenge 100 grams [50 grams 1 hour---=-====e=e-- .81 74 2 hours-=======-=--- .68 .54 3 hours========---- .40 sal Overnight=-========= .26 Jab 73 T T T Mean Blood Glucose Level (mg. %)- All Tests 100 §- . : g 75 = ol 1 1 TI 0 10 20 30 40 50 Urine Glucose Score - All Tests Figure 7. Correlation of mean blood glucose level 1 hour after challenge with urine glucose score 90 minutes after challenge, 24 men, Milan, Phase lll. INCIDENTAL OBSERVATIONS There were a few occasions in which in- dividual idiosyncrasy seemed to be the explanation for discordant duplicate tests or a peculiarity in the glucose tolerance curve. A discordant pair of values for an individual may arise from labeling errors, failure of examinee to follow instructions, laboratory mistakes or some other defect in techniques, or may only reflect the inherent variability of an individual. That there is no safe way for deciding which factor prevails in a given case is illustrated by data for examinee 28. His four glucose tolerance tests in Phases II and III (shown in figure 5) were highly variable 2 hours after the challenge. On the first test, the level was 100mg.%; 1 week later itwas 30.5 mg.%. This difference of 69.5 mg.% should be compared with the average difference of 14.6 mg.% for all 10 men in Phase II, including examinee 28. However, 3 months later duplicate standard glu- cose tolerance tests were done on the same man with almost identical results. The first week showed 101.5 mg.% at 2 hours; the second week it was 40.0 mg.%. What on the first pair of tests seems to betoken a technical error, on the second pair of tests may more logically be attrib- uted to examinee idiosyncrasy. Clearly, in a small series of cases one or two very peculiar individuals can produce a dis- torted picture of the general population. This does not appear to have happened in this study, for even the idiosyncrasy just mentioned has only a minor effect on the mean values. Therefore, even though clinicians have a natural interest in the unusual case, the kind of study undertaken here can shed little light on cases of this type. DISCUSSION One must be most circumspect in generalizing from a study such as this. For one thing, the group was limited in size. For another, this was an unusual group, living under unusual circum- stances. The age range was limited; the group included only men. Any number of artifacts might have intruded on the study. Some peculiarity of the prison diet might have affected the results. Usually the subjects spent the hour after glucose loading in comparative idleness during which they often smoked cigarettes. This is quite different from the examination routine of either the Te- cumseh Study or the Health Examination Survey, in both of which participants are occupied during the period between challenge and venipuncture and have little opportunity for smoking. There was an epidemic of Type B influenza during the eighth and ninth weeks of Phase III. Each subject was routinely asked at each session whether he had a cold or fever or other infection, and it was only during these weeks that any excess number of respiratory infections was noted in the study group. However, the subjects report- ing these symptoms revealed no consistent alter- ation in response to glucose load in comparison with asymptomatic periods. One disturbing feature of the study group is the low mean blood glucose level after challenge. The response to a 100-gram challenge is dis- tinctly less than noted in the Tecumseh Study, while the response to a S0-gram challenge is less than that found in the Health Examination Survey. On the other hand, their response levels were comparable to those found by Wilkerson and his associates!?in another prisoner group. There does not seem to be an obvious explanation for these findings. These various qualifications are not entered to deprecate any findings of this study. Inthe last analysis, no study can stand by itself. It must be integrated with the findings of other related studies and must be repeated by other investi- gators on other study groups before its meaning becomes clear and certain. While there have no doubt been numerous informal observations made of the factors in- vestigated in this study, there are relatively few solid data in the literature. Maclean!’ an early worker with the glucose tolerance test, observed that ''. . . after a certain dose is reached, about 25 grams, further increase in the amount of sugar does not increase the actual height of the resulting hyperglycaemia." Of course, laboratory tech- niques then in use measured something more than blood glucose, so that his findings are not neces- sarily in contradiction to this study. In any event one clear finding in this study is that a 100-gram challenge yields a somewhat higher blood glucose level than a 50-gram challenge. Irving and Wang!® in a study which essen- tially yielded replicate standard glucose tolerance tests on a series of 12 persons, found, as in this study, large variability in the results. The vari- ability in level of their subjects, while somewhat greater than that for subjects in this study at fasting and at ’, 1, 1%, and 2 hours after challenge, appears to be of about the same magnitude. What differences do exist may be accounted for by two facts: (1) the measurement of blood glu- cose concentration in their study was done on capillary blood and could be expected to have a greater measurement variability than determi- nations made in this study; and (2) the prior preparation was deliberately varied from one test to the other. A study of replicate standard glucose tolerance tests was made by Freeman, Looney, and Hoskins!” on 35 men, 30 of whom were schizo- phrenic. Blood glucose was determined by the Folin-Wu method. The average difference between replicate specimens taken fasting and %, 1, 2, and 3 hours after challenge was 9.0, 25.8, 29.9, 20.3, and 15.2 mg.%. These are all greater than the comparable figures for our stuay group. If it is assumed that the fasting glucose level is highly stable, the greater variability reported in fasting tests in their study suggests a greater technical variability in the measuring technique than obtained in this study. Unger '® studied the variability of standard glucose tolerance tests using a group of food handlers with a casual postprandial blood glucose level on screening of less than 130 mg.%. ""With- in-person" standard deviations may be computed from his published data. For 7 men under age 40 and for 10 men over 40 the standard deviations of the 1-hour blood glucose level were 23.9 and 29.7 mg.%, respectively. The corresponding fig- ures for 15 women under 40 and 17 women over 40 were 32.8 and 25.9 mg.%. The figure for the inmate volunteers in this study was 18.0 mg.%. Again, part of the difference may be technical, since the standard deviation of fasting levels was also higher than in this group. For men over 40 it was 7.6 mg.% as compared with 3.8 mg.% for this study group. At 2 hours, however, the stand- ard deviation for men over 40 in Unger's group was 6.7 mg.% as contrasted with 11.7 mg.% for this group. Summing up the various comparisons, it ap- pears that the variability of response to chal- lenge found in this study, high though it was, probably represents a conservative estimate of this factor. One interesting finding in this study is that the response to a 50-gram challenge appears to be quite insensitive to the interval since prior meal. Whether the 50-gram challenge is given fasting or 1, 2, or 3 hours after a meal the blood glucose level 1 hour after challenge appears to be the same. Data from the Health Examination Survey suggest that the effect of time from last meal to a S0-gram challenge is not as trivial as appears from this study, and there are some anomalies in the results of the Milan Study itself which suggest special caution be used in interpreting the findings with respect to the 50-gram challenge, Still, the effect of time after meal seems definitely greater with a 100- gram challenge than with 50. Finally, it must be said quite explicitly that this study cannot be used to decide whether any specific tolerance test is best for determining the presence or absence of diabetes. What the 17 study does strongly suggest is that any of the procedures under investigation will tend to rank persons with respect to glucose tolerance in about the same order from low to high, that a casual glucose tolerance procedure yields re- sults quite similar to a standardized procedure, but that any procedure will yield variant results when repeated on the same individual. SUMMARY AND CONCLUSIONS Using 24 male prisoner volunteers 40-52 years of age as subjects, l-hour oral glucose tolerance tests performed under eight different arrangements were compared. Challenges were given with both 50 and 100 grams of glucose, and were given after an overnight fast and 1, 2, and 3 hours after breakfast or lunch. Each procedure was performed twice. The 16 tests for each subject were performed at weekly intervals. Fasting 100-gram tests were extended to 3 hours. Subsequently, 22 of the subjects were given three clinical tests after added dietary carbohydrate: the standard glucose tolerance test, the cortisone glucose tolerance test, and the pred- nisone glycosuria test. The following findings were noted: A challenge of 100 grams of glucose yielded slightly but consistently higher mean blood glucose levels 1 hour after challenge, and significantly higher concentrations of urine glucose, than did a 50-gram challenge. Despite this fact, individuals with high levels 1 hour after a 100-gram challenge also had rela- tively high levels after a 50-gram challenge and mean response to the two loads appeared to rank individuals in almost the same order. The four subjects classified clinically as exhibiting some evidence of deficient carbohydrate tolerance were ranked high by both the 50-gram and 100- gram tests. In contrast with the more uniform 1-hour levels of the group given 100 grams of glucose at various intervals after breakfast, response levels of the apparently similar group tested after lunch with 100-gram challenges increased with time after meal. On the other hand, the 50-gram test revealed no significant correlation of re- sponse level with interval after meal in either group. These findings should be treated with some reserve, An individual's blood glucose level after over- night fast was highly stable but his level under other circumstances was variable. In particular, the variability of response to challenge after an overnight fast was of the same magnitude as variability of response when the challenge was administered after a meal. Administration of a glucose challenge when- ever a person comes infor examination, no matter when or what he last ate, appears to be an en- tirely reasonable method of testing for carbo- hydrate tolerance. REFERENCES ly, H. Moyer and C. R. Womack, ‘Glucose Tolerance: A Com- parison of Four Types of Diagnostic Tests,”” American Journal of Medical Sciences, Vol. 219, No. 2, 1950, p. 161. LR W. Conn and S. S. Fajans, ‘The Prediabetic State,’ Amer- ican Journal of Medicine, Vol. 31, 1961, p. 839. 3y. S. National Health Survey, ‘‘Plan and Initial Program of the Health Examination Survey,’’ Health Statistics, Series A, No. 4. Wash- ington, U. S. Government Printing Office, 1962. 47, A. Napier, “Field Methods and Response Rates in the Tecumseh Community Health Study,’’ American Journal of Public Health, Vol. 52, 1962, p. 208. 3]. M. Leichsenring and E. D. Wilson, ‘Food Composition Table for Short Method of Dietary Analysis’’ (2d revision), Journal of the American Dietetic Association, Vol. 27, 1951, p. 386. 6A. de P. Bowes and C. F. Church, Food Values of Portions, Commonly Used, 8th edition. Philadelphia, Lippincott, 1956. 7B. K. Watt and A. L. Merrill, Composition of Foods--Raw, Processed, Prepared, U.S. Department of Agriculture Handbook No. 8. Washington, U. S. Government Printing Office, 1950. 8s. 5. Fajans and J. W. Conn, ‘‘The Early Recognition of Dia- betes Mellitus,”’ Annals of the New York Academy of Sciences, Vol 82, 1959, p. 208. 9r. G. Joplin, R. Fraser, and K. J. Keeley, ‘‘Prednisone-Glyco- suria Test for Prediabetes,’’ Lancet, Vol. II for 1961, p. 67, 1961. 10y, Somogyi, ‘Determination of Blood Sugar,’’ Journal of Bio- logical Chemistry, Vol. 160, 1945, p. 69. 114, B. O'Sullivan, N. Kantor, and H. L. C. Wilkerson, ‘‘Com- parative Value of Tests for Urinary Glucose,’’ Diabetes, Vol. 11, 1962, p. 53. 12 R. Froesch and A. E. Renold, “Specific Enzymatic Deter- mination of Glucose in Blood and Urine Using Glucose Oxidase, ’’ Diabetes, Vol. 5, 1956, p. 1. 137.-w. Conn, “Interpretation of Glucose Tolerance Test,” American Journal of Medical Sciences, Vol. 199, 1940, p. 555. 14y. L. C. Wilkerson, H. Hyman, M. Kaufman, A. C. McCuiston, and J. O’S. Francis, ‘‘Diagnostic Evaluation of Oral Glucose Tol- erance Tests in Non-Diabetic Subjects After Various Levels of Carbohydrate Intake,’ New England Journal of Medicine, Vol. 262, No. 21, 1960, p. 1047. 15Hugh Maclean, Modern Methods in the Diagnosis of Treatment of Glycosuria and Diabetes, Sth edition. London, Constable and Company, Ltd., 1932. 16% up. Irving and I. Wang, ‘The Effect of the Previous Diet on Glucose Tolerance Tests,” Glascow Medical | aurnal, Vol. 35, No. 11, 1954, p. 275. 17y, Freeman, J. M. Looney, and R. G. Hoskins, ‘‘Spontaneous Variability of Oral Glucose Tolerance,” Journal of Clinical En- docrinology and Metabolism, Vol. 2, 1942, p. 431. 18g, H. Unger, ‘The Standard Two-Hour Oral Glucose Toler- ance Test in the Diagnosis of Diabetes Mellitus in Subjects With- out Fasting Hyperglycemia,’’ Annals of Internal Medicine, Vol. 47, 1957, p. 1138. Table Le II. III. Iv. V. VI. VII. 20 DETAILED TABLES Blood glucose and urine glucose findings: 45 men, Milan, Phase I-----=-==-=---- Assignment of subjects to experimental procedures: 24 men, Milan, Phase III---- Blood glucose and urine glucose findings according to procedure used: 24 men, Milan, Phase IIl-=-===--e-ececcc cece cece ccc eee cece eee ecm ee em ee meme —————— Standard glucose tolerance tests: 28 men, Milan, Phases II, III, and IV---==---- Carbohydrate intake within 4 hours preceding glucose challenge: 24 men, Milan, JE I ED Clinical classificationof study participants by specified tests: 24 men, Milan, Phases III and IV,======c--cecccmccccc ccc cece ccc ccc cece cece me meme me ———————— Heights, weights, and changes in weight: 24 men, Milan, Phase IIIl--=--=-====--- Page 2) 22 23 26 28 29 30 Table I. Blood glucose and urine glucose findings: 45 men, Milan, Phase I First week Second week Case number | Blood glucose levels ae Case number | Blood glucose levels Js Before 1 noe ¥ Before 1 poup th ee challenge challenge challenge challenge challenge challenge 01% em mmmmeem 74.0 116.0 | Negative 3l---==-mmm- 74.0 80.0 | Negative 2 me mmm mmm 107.5 84.5 | Negative 32---cmmme- 103.0 100.5 | Negative DB mmm mms 76.5 64.0 Trace BF in me wn 87.0 120.0 | Negative B4% «mem mtr 67.5 97.0 | Negative 3h4mmmmmmnnn- 77.0 82.0 | Negative 05%=mmcmmmmmn 95.0 78.0 | Negative 359mm 79.5 138.0 1+ Bomar 85.5 83.0 2+ KL 65.0 121.5 2+ OF mms mmm 134.0 100.0 2+ 37-=mmmmmmam 82.5 83.0 | Negative 1]: —— 86.5 96.5 | Negative 38-mmmmmmmm 69.5 102.0 Trace 09C-mmmmmmmam 81.5 120.0 Lt 39mm 70.0 87.0 | Negative 10? emma 87.5 90.0 Negative 40==mmm mmm 69.0 82.0 | Negative 1) Re —— 78.0 88.5 | Negative flame 92.0 95.5 | Negative 1 LJ —— 66.0 114.0 | Negative 42% amma 69.5 102.0 | Negative 20 Preemie 87.5 226.0 2+ 43% emma 87.0 75.0 | Negative 21% meen 68.0 158.0 2+ bh enn mennn 66.0 155.0 3+ Gm i 91.5 87.5 | Negative 45% mma 80.0 144.5 2+ 23% emma 69.0 113.0 | Negative [[46"--------- 75.5 132.0 | Negative Daou er we 54.5 79.0 | Negative 478 —mcmaeeee 82.5 99.5 | Negative DG emmmmmmnie ns 96.0 86.0 | Negative |[[48%-=------- 100.5 127.0 | Negative 7 70.0 122.0 | Negative [49%--------- 69.0 133.0 | Negative Frm nm 82.0 106.0 2+ LL 88.5 118.0 | Negative 1; ST — 77.0 144.0 3+ 58mm mmmmmmn- 69.0 83.5 | Negative ZG mmr em 77.0 75.0 | Negative 1c R——— 72.0 122.0 | Negative 30mm mmm 78.0 56.0 | Negative Selected for Phase III. bSelected for Phase III but discharged and replaced by case number 29. “Veins unsuitable for venipuncture. dSelected for Phase III but discharged and replaced by case number 08. 21 Table II. Assignment of subjects to experimental procedures: 24 men, Milan, Phase III Week Case number 1, 2 3, 4 5, 6 7: 8 9, 10 11, 12 13, 14 15, 16 Morning group 28, 10, 50=-=m== mmm——- D H F G B A Cc E 21, 48, 1ll-==c=cmem————— BE D A Cc E F G B 49, 33, Obmmmmmemmmmeeeee A F H E ¢ D B é 20, 01, 43==-m-mmmmeeam- F A D B G H E Cc Afternoon group 46, 59, 42=mmmmmmmmaaaan B E G F D Cc A H 45, 26, 05==mm=mmmmmaan- E B c A H G F D 4b, 36, 4] =mmmmmmmmmmman c G E H A B D F 08, 29, 23-emmecemeennn- G C B D F E H A Key to Procedures A = 100 grams challenge after overnight fast E = 100 grams challenge 2 hours after meal B = 50 grams challenge after overnight fast F = 50 grams challenge 2 hours after meal C = 100 grams challenge 1 hour after meal G = 100 grams challenge 3 hours after meal D = 50 grams challenge 1 hour after meal H = 50 grams challenge 3 hours after meal Lapses Case number 01, 04, 10, 11, 20, 28, 33, 42, 43, 48, 49, 50—no lapses Case number 05,21,26,44,45, 47 were given challenges of 100 grams in week 13, when they should have been Suen challenges of 50 grams. These results were discarded and the correct procedures were one on week 17 Case number 46 was ill week 15. This procedure was completed week 17. Case number 59. The values obtained during week 15 were considered highly improbable for this person and discarded. The procedure was completed on week 18. Case number 36. The values obtained on week 15 were considered highly improbable for this person and dis- carded. On week 13 this person was given 100 grams of glucose instead of 50. These losses could not be made up later. Case number 23. The value before challenge obtained on week 16 was considered highly improbable for this person and discarded. On week 13 this person was given 100 grams of glucose instead of 50. This procedure was completed on week 17. Case number 29. Replaced case number 19, the original assignee, after week 1. Week 1 procedure was com- pleted (by accident) on week 13. Week 13 procedure was completed week 17. Case number 08. Replaced case number 35, the original assignee, after week 5. It proved impossible to 22 make up all the lost ground. Week 1 procedure was completed week 13. Week 3 procedure was completed week 17. Week 13 procedure was completed week 15. Week 15 procedure was completed week 19. Duplicates of procedures B and C (assigned to weeks 5 and 4,respectively) were never completed. * NOTE: Letters shown in body of table refer to procedure. Table III. Blood glucose and urine glucose findings according to procedure used: 24 men, Milan, Phase III A Cc D E F G Case number 1 2 1 2 1 2 3 2 1 2 1 2 1 2 | 2 Blood glucose level (mg. %) before challenge Olessmnmeme 77.5] 70.5] 99.0| 103.0 | 102.0 | 113.0 | 89.0 112.5| 80.0 | 81.5| 73.5| 68.0] 69.0 74.0 | 79.0 75.0 04mm mmmmmmm 69.5| 66.5] 77.0| 67.0| 62.0| 60.5| 79.0] 58.0| 60.0 | 82.0 75.0| 60.5] 50.5| 63.0 86.0 63.5 05mmmmmmmm= 74.5] 73.5] 88.5| 82.5| 55.0 56.5| 69.0| 57.0| 58.0| 75.0 | 78.0 | 74.5| 91.0 | 80.5 76.5 67.5 08wmmmmmm= 79.0 | 87.5] 74.5 ---] 65.0 ---| 93.0| 90.0| 89.0| 88.0| 95.0 97.5 99.5| 83.0 | 75.5 71.0 10=-mmmm——- 24.0| 70.0] 75.0) 65.5| 64.5| 67.0] 78.0| 48.5| 65.0 | 63.0| 86.0 67.0| 76.5| 66.0 78.0 77.5 1l-m-mmm——- 71.0| 8o0.0| 63.0] 76.5| 70.5| 64.5| 72.5| 72.5| 91.0 | 88.5| 82.0| 74.5| 97.0 78.0 89.0 90.5 20==mmmm——- 83.5| 81.5| 78.5| 66.0 100.5|103.5| 72.0] 101.0 95.0 | 73.0 | 82.0] 111.0] 65.5| 83.0 76.0 63.5 2l---mm—--- 109.0 | 102.5 | 104.0 | 114.5 | 127.5 | 104.0 | 156.5 | 150.5 | 135.0 | 126.5 | 111.5 | 141.0 | 108.0 111.5 | 104.0 | 131.0 23wemmmm——— 76.0 ---| 80.5| 72.5|100.5| 96.0|101.0| 78.0 | 82.0] 71.5| 97.0| 97.0| 91.5| 62.5) 89.0 92.0 26--mmmm——— 75.0 | 72.0 |100.5| 68.0] 110.0 |120.0| 92.0] 87.5|129.5| 91.0 | 97.5|101.5| 74.5|120.5| 67.0 110.0 28-mmmmm mn 78.5| 79.0] 71.0] 80.0] 132.0 135.0] 108.5] 104.0 | 72.0 | 110.0 | 85.0 | 72.5| 75.5| 67.0 | 56.5 55.0 29mm mmmmm—— 65.5| 68.5| 75.0] 75.0| 84.0| 68.0| 90.5| 71.0| 71.5| 59.0 | 64.0| 70.0 | 67.0 | 61.0 | 63.5 71.0 33mm 76.5] 71.0] 70.0] 78.0] 96.5] 99.0| 93.0 101.0 | 64.0 | 90.5 111.0 | 87.0 | 69.0 | 84.5] 72.0 81.0 36-mmmmm——— 66.0 68.5] 66.0 69.0 (102.0 | 90.0 | 111.5 ---| 79.0] 83.5| 77.0 ---| 72.5] 84.0] 97.5 71.5 42mm mmm 83.0| 86.0 91.5| 85.5| 76.0| 66.5| 79.0| 72.0| 98.0 | 87.5| 86.5| 92.5| 89.5| 89.5] 84.0 109.0 43mmmmmm 78.0| 81.5] 77.0| 79.5] 72.0| 55.0 56.0| 62.5 49.5| 75.0 72.0 57.5| 35.5| 57.5] 46.0 41.0 4h mmm mmm 74.5 74.0] 76.0) 79.0] 63.0| 65.0] 95.5| 91.0| 78.0 93.5| 77.0| 76.0| 76.5| 86.5 | 78.0 83.0 45mm mmm mmm 81.5| 79.5| 86.0| 85.0 166.5 |148.0| 163.0] 156.5 | 89.5 |104.5| 95.0 | 100.0 | 79.0 | 71.5} 72.0 78.5 46-=mmmmm- 73.0| 83.5] 71.5| 84.0| 97.5| 78.5|113.0| 129.0 94.5| 99.5| 82.0| 97.0| 84.0] 70.0 | 67.0 73.5 47mmmmmm——— 75.0| 68.0] 70.5| 77.5] 70.5| 80.0| 76.0| 87.0 | 88.5| 80.0 102.0 79.0 74.5| 78.5] 81.0 82.0 48mm mm———- 75.0! 74.0] 69.0 72.0| 55.5] 66.5| 68.0| 89.0| 79.0 | 63.0 | 82.5| 70.0 | 57.5| 58.0 | 69.5 76.0 49m mmm mmm 74.0] 78.5| 73.5] 70.5| 70.5| 74.0| 55.0] 79.5| 65.5| 69.0 58.5| 64.5| 63.5] 86.0 | 62.0 60.5 50~ m= mmm=—- 86.5! 99.0 |104.0| 92.0| 98.5] 87.5| 65.5| 99.5| 69.5 | 82.0 |100.5| 93.0| 68.0 | 71.0 | 81.0 76.5 59mm mm mmm 77.5| 72.0] 70.0| 80.5] 93.5|116.0| 87.0] 104.5| 88.0 | 93.0 |101.0| 82.0 | 82.0 | 83.5| 74.0 85.0 23 Table III. Blood glucose and urine glucose findings according to procedure used: 24 men, Milan, Phase III—Con. A B Cc D E F G H Case number Blood glucose level (mg. %) 1 hour after challenge 0l----=muun 147.0 | 126.5 | 136.0 | 100.0 | 108.0 | 111.5| 98.5 | 148.5] 134.5 104.0 | 89.0] 102.0 | 149.5] 96.5] 131.5] 110.0 04==meomenn 93.0 [ 110.0 | 66.5| 87.5] 83.5| 91.0 94.0 76.0 94.0 82.0 | 84.0 82.5| 90.0 | 94.5] 101.5 73.0 05-=mmmmunn 72.0 | 86.0 | 93.5| 72.5| 78.0 | 86.5| 77.0| 82.5| 69.0 63.5 1104.5] 74.5 |134.5| 121.0 | 86.5 86.0 08-==mmmeen 115.5 ( 121.5 | 109.0 -== |. 140.0 === | 98.0 76.0 94.0 115.0 | 92.5 96.5 102.5] 114.0 | 105.0 79.5 10---=mmna- 80.0 | 81.0 | 85.5| 80.0| 99.0| 77.5| 83.5| 94.5| 77.5 91.5] 76.5| 94.5|118.0| 101.5 | 82.5 78.5 ll-mmmeeee 102.5 1101.0 | 66.0 | 92.0| 83.5| 89.0 | 98.5| 89.0| 85.0] 82.0 97.0 79.0 | 101.0 | 78.5] 78.0 68.5 20-======ne 86.5 | 84.0 | 111.0 | 101.0 | 108.5 | 123.5 | 83.5] 107.5] 132.0 147.5 | 83.0| 89.0 | 91.0 148.0 127.5] 127.0 2l-mmmmmmm 181.0 | 151.0 | 140.5 | 143.0 | 133.0 | 95.0 | 149.5 | 136.5| 145.0 150.0 | 124.5 | 148.0 | 156.5 | 127.5 | 131.5 | 114.0 23mm mma 141.5 | 109.5 | 107.5 | 114.5 | 107.0 | 108.5 | 86.5| 69.0] 87.5 87.0 78.5 | 111.07 91.5) 122.5] 93.5 95.3 26===mm=mn= 140.5 | 134.0 | 114.5 | 100.0 | 100.0 | 121.0 | 86.0 | 112.0] 133.0 121.0 1129.0 129.0 | 89.0 | 130.0 | 88.0 | 112.0 28-mmmmmm mm 136.5 1195.0 | 143.5 | 133.0 | 117.0 | 137.5 | 107.5 | 83.5] 108.5 | 137.5 100.0 | 129.5 | 113.0 | 151.0 | 86.0 74.5 29--mmmmman 75.0 | 83.0 109.5 | 114.0 | 107.0 | 112.0 | 102.5 | 64.0] 96.5 104.0 | 98.0 | 100.5 | 84.5| 97.5| 58.5| 143.5 128.0 | 86.5| 94.0 | 95.0 126.0 | 143.5] 100.0 | 119.0] 107.0 142.0 | 99.5] 115.5 | 124.5 110.0 | 125.0 | 108.5 137.5 | 112.0 | 81.5| 99.0 | 137.0 | 124.5] 110.5 we=| 197.5] 195.5 | 150.0 we= (214,0({ 183.5 | 154.0] 148.5 LY 90.0 | 105.5 | 102.5| 96.0 | 92.5| 85.5] 96.5| 99.5 117.0 | 125.0 | 125.5 | 106.5 | 112.5 | 136.5 | 84.5 | 108.5 43-mccmmaen 78.0 | 110.0 | 78.0 | 81.0 81.0 | 64.5| 79.0| 82.0| 84.0 86.0 | 84.0 | 81.0 | 86.5| 85.5| 67.0 71.5 bhmmmmmee ee 120.5 | 101.0 | 108.0 | 120.5 | 83.0 | 133.0 | 130.5 | 104.0 101.0 | 109.0 | 93.0 | 92.0 | 147.0 109.5] 69.0 98.0 45mm mmmmmen 103.0 | 113.0 | 137.0 | 130.0 | 124.0 | 138.5 | 146.5 | 156.0 140.0 165.5 | 142.5 | 156.0 | 169.5 | 178.0 | 146.0 147.0 46 =mmmm am 126.5 126.5 | 125.5 | 159.5 | 104.0 | 127.0 | 128.0 | 118.0 159.5] 165.0 | 109.0 | 137.0 | 131.0 | 135.0 | 131.0 | 113.5 47=mmmmmmem 88.5| 51.5 84.5 94.5| 80.5| 82.0 112.0 116.0] 99.0 95.0 70.0 | 70.5 | 118.5| 66.5| 94.5 81.0 48-mmmmeeen 108.5 119.5] 91.0 | 71.5| 76.0 | 78.0| 94.5| 79.0| 88.5 62.0 | 60.0 75.0 | 104.0 80.0 78.5 85.0 41.5 (103.5 | 112.0 | 127.0 | 95.5| 83.5] 65.5] 111.5 108.0 [ 110.0 | 63.5 | 113.5 | 106.5 95.5 | 103.0 108.0 85.0 [111.0 | 118.0 | 90.0 | 117.0 | 87.0 | 65.5| 87.0 75.0 | 95.0 (113.0 | 130.0 | 113.0 | 117.0 | 110.0 80.0 39---mmmenn 132.5 | 118.0 | 118.0 | 105.0 | 150.0 | 157.0 | 90.0 | 137.5 148.0 | 146.5 | 108.0 | 97.5 |137.5| 125.5] 111.5 | 138.0 24 Table III. Blood glucose and urine glucose findings according to procedure used: 24 men, Milan, Phase III—Con. A B Cc D E F G H Case number 1 2 1 2 1 2 1 2 i 2 1 2 1 2 1 2 Urine glucose 1% hours after challenge Ol----====- N N N N N N N 2+ N N N 2+ N N N N Olymmmmmmmmm N N N N N N N N 1+ N N N N N N N 05--===m==m N N N N N N N N N N N N N N N N 08-==mmmmmm Tr N N -- N --- N N N N N N N N N N 10-==mmm=mm N N N N N N N N N N N N N N N 1l1---mmmmmm N N N N N Tr N N N N N N N N N N 20-======== N Tr N N N N N Tr 2+ N 24 Tr 1+ Ir 1+ 2l-=mmmmm- 2+ 2+ 3+ 1+ 3+ 1+ 3+ 2+ 3+ 3+ 4+ 2+ 2+ 2+ N 2+ 23=mmmmm mm N N N N N N N N TE N N N N N N 26-==m=mm=- N N _N N N N N N N N N N N N N 28----===== 4t 4 3% 3+ 4+ 2+ 1+ 3+ 4+ 2+ 1+ 3+ 3+ 3+ 2+ 2+ 29-=mmmmm—- N N N N N N TE N N N N 1+ N N N N 33---mmmmm- N N N N N Tr N N N N ; N N N N N N 36----====- 3+ Tr N N N 3+ 2+ - 3+ 2+ 3+ -- 2+ 3+ 2+ 3+ 42mm mmm N Tr N 2+ Tr N Ir N 1+ 2+ 1+ Tr N N N N J N 1+ Ix N N N N N N N N N N N N N bhymmmmmmmm 1+ 1+ 1+ N N 3+ 1+ 1+ 2+ 1+ 2+ 1+ 3+ 2+ 1+ N 45-=mm mmm N Tr 1+ 1+ 2+ 2+ 2+ 2+ 3+ 3+ 2+ 1+ 2+ 3 2+ 3+ 46mm mmm N 1+ Tr N TE Tr 1+ N Tr 2+ N 1+ 1+ 1+ N N f]mmmmmmm mm N N N N N N N N N N N N N N N N 48mm mm mmm Tr N N N 2+ Tr N 1+ N N + N 2+ TE N N 49mmmmmm mm N N N N N N N 1+ N N N Tr N N N N 50===mm==m= N N N N N N N N N N N N N N N N 59~mmmmmm—— N N N N N N N N N N N N N 1+ N N NOTES: Letters refer to procedure used (see key shown on table II). Numbers 1 or 2 rerer to the time (first or second) that the procedure was administered. N - negative; tr- trace 25 Table IV. Standard glucose tolerance tests: 28 men, Milan, Phases II, III, and IV Blood glucose levels (mg. %) by time after challenge Urine glucose Case number and test series 0 , 1 - . 5 1% hours 3 hour's after after hour hour hour hours hours hours challenge | challenge 22:5 132.0 147.0 107.5 113.5 66.0 | Negative | Negative 70.5 144.0 126.5 111.0 103.5 61.0 | Negative om 75.5 118.5 126.0 129.0 91.0 | 104.0 | Negative | Negative 79.0 89.5 74.5 91.0 77.0 76.5 1+ -—— 75.0 75.5 68.5 66.0 71.0 75.5 | Negative | Negative 69.5 106.5 93.0 725.0 93.0 70.0 | Negative —— 66.5 102.5 110.0 66.0 92.0 75.0 | Negative | Negative 73.5 107.5 73.5 56.5 65.5 63.5 I+ Trace 74.5 81.0 72.0 80.5 65.0 76.0 | Negative | Negative 73.5 84.5 86.0 61.5 81.0 73.5 | Negative | Negative 81.0 86.0 78.5 72.0 75.0 77.0 | Negative | Negative 87.5 114.0 121.5 99.5 127.0 102.53 Trace Trace 79.0 107.0 115.5 119.0 118.5 98.0 | Negative Negative 78.5 163.0 140.0 115.0 103.0 74.0 2+ — 80.0 123.5 149.0 145.0 114.5 96.5 2 1+ 74.0 81.0 80.0 74.0 81.0 48.0 | Negative | Negative 70.0 79:0 81.0 88.5 85.5 54.0 | Negative | Negative 73.0 98.0 17.5 76.0 98.0 82.0 | Negative | Negative 21.0 89.0 102.5 94.0 84.0 89.0 | Negative | Negative 80.0 94.0 101.0 -——— 95.0 79.0 -— Negative 93.5 113.5 104.0 90.0 86.0 83.0 | Negative | Negative 86.0 124.5 125.5 122:5 88.5 74.0 Trace —— 84.0 121.0 99.0 72.0 55.0 77.0 Trace 0 83.5 112.5 86.5 100.5 88.0 68.5 | Negative -—— 81.5 117.5 84.0 95.0 68.5 69.5 Trace -— 79.5 171.5 153.0 150.0 119.5 55.5 1“ 100.5 140.0 175.0 147.5 128.0 75.0 3+ ——— 102.0 150.0 150.0 114.5 129.5 64.5 4+ 2+ 109.0 152.5 181.0 151.5 148.0 94.0 2+ 1+ 102.5 -—— 151.0 150.0 146.0 | 130.5 2+ 24 108.5 168.5 165.5 159.0 153.01 102.5 3+ 4+ Case 23 Phase III, l--=----e--u- 76.0 127.5 141.5 118.5 102.0 86.0 | Negative | Negative III, 2- — 87.0 109.5 96.0 120.5 88.5 | Negative | Negative LV ==smmmmimmmimn 90.5 90.0 93.5 103.0 113.5 69.0 | Negative | Negative Case 26 Phase ITI, l-===c--mocmmm mm ceeeeeeeeo 75.0 129.0 140.5 128.0 109.0 59.5 | Negative Negative 139.0 134.0 129.0 106.0 84.5 | Negative | Negative 161.5 169.0 148.5 118.5 59.3 Trace Trace 158.0 171.5 113.0 100.0 44.5 3+ — 151.0 223.0 132.0 30.5 50.5 -— 0 157.5 136.5 126.5 101.5 21.5 4+ —— 148.5 195.0 126.5 40.0 46.0 4+ 3+ 149.5 180.0 198.5 154.0 48.0 4+ 3+ Case 29 Phase ITI, l------ 65.5 89.5 75.0 65.5 69.0 77.5 | Negative | Negative I, 2 68.5 99.0 83.0 103.0 73.0 68.0 | Negative | Negative IV mm mmm me ee ee ee 5: 74.5 45.0 7 67.5 | Negative | Negative 26 Table IV. Standard glucose tolerance tests: 28 men, Milan, Phases II, III, and IV—Con. Blood glucose levels (mg. %) by time after challenge Urine glucose Case number and test series 0 % L W 2 3 1 hours 3 hous after after hour hour hour hours hours hours challenge | challenge Case 30 Phase II, l----------ccmemommmmmmmmmmmmmm mem 77.0 11.3 83.5 85.0 78.5 82.0 | Negative --- II, 2--==-=--e--eccmmeceecemeem—eeooa- 78.5 118.0 99.5 100.5 82.5 54.0 | Negative | Negative Case 31 Phase II, l-weserrronsmsmer ene ———— 74.0 96.0 87.0 80.5 86.0 88.0 Trace -—— II, 2------msmeemmeeeceeeeeeeeemce————-- 8L.5 63.5 75.0 98.0 79.0 63.0 | Negative | Negative Case 33 Phase 76.5 125.0 128.0 100.5 100.0 92.0 | Negative -— 71.0 115.0 86.5 116.0 29.0 51.5 | Negative | Negative 74.5 149.5 115.0 90.5 94.5 71.5 | Negative | Negative 66.0 161.5 137.5 111.0 77.0 37.5 3+ 1+ 68.5 128.5 108.0 116.5 78.0 45.0 Trace Negative 83.0 106.5 90.0 82.0 93.5 81.0 | Negative Trace 86.0 103.0 105.5 88.0 103.0 | 108.0 Trace Negative 83.5 102.5 116.5 110.5 121.0 | 106.5 | Negative | Negative 76. 105.0 49.0 60.5 64.0 68.0 | Negative --- 79.5 81.0 58.0 71.0 66.5 66.0 | Negative | Negative 78.0 103.5 78.0 70.0 84.0 67.5 | Negative | Negative 81.5 109.0 110.0 86.5 72.5 41.0 1+ --- 78 111.0 76.0 91.5 97.5 90.5 | Negative | Negative 80. 138.0 128.5 86.5 84.5 45.5 5 -— 84.5 123.5 129.0 82.0 99.5 76.5 2+ 0 74.5 123.5 120.5 80.5 125.5 58.0 1+ Trace 74.0 94.0 101.0 75.0 67.5 | 102.0 1+ 1+ 75.5 85.5 98.0 76.5 85.0 72.0 3H 1+ 81.5 80.5 103.0 104.0 87.5 47.5 | Negative | Negative 79.5 124.0 113.0 78.5 109.5 60.5 | Negative | Negative 78.0 157.5 63.5 88.0 111.5 65.5 24 Trace 73.0 141.5 126.5 108.5 103.0 41.0 | Negative | Negative 83.5 152.5 126.5 114.5 100.0 53.0 1+ 1+ 54.0 121.0 109.0 118.0 153.5 79.0 Trace Negative 75.0 109.0 88.5 81.0 65.0 77.0 | Negative | Negative 68.0 87.0 51.5 80.5 90.5 44.0 | Negative | Negative 57.0 79.0 85.5 69.5 87.0 74.5 | Negative | Negative 81.0 125.0 146.0 83.0 82.5 87.5 1+ --- 81.5 129.5 112.0 82.0 83.5 90.0 Trace 0 75.0 91.5 108.5 94.5 58.0 67.0 Trace --- 74.0 119.5 119.5 131.5 65.5 74.5 -— 0 72.0 116.0 92.0 82.5 73.0 68.0 ho Negative Case 49 Phase 74.0 102.5 41.5 82.0 81.0 36.5 | Negative --- 78.5 93.0 103.5 74.5 80.0 34.5 | Negative | Negative 70.0 106.5 106.5 74.0 90.5 49.0 | Negative | Negative Case 50 Phase 86.5 95.5 85.0 71.5 84.0 73.0 | Negative -— 99.0 109.0 111.0 84.5 90.5 73.5 | Negative | Negative 97.5 143.5 90.0 102.5 71.5 | 110.5 | Negative | Negative Case 59 Phase 77.5 109.0 132.5 125.5 104.0 58.5 | Negative | Negative 72.0 120.0 118.0 103.0 929.5 87.5 | Negative | Negative 76.0 115.5 113.0 118.5 114.0 85.0 | Negative | Negative NOTE: In these standard glucose tolerance tests each person was given a challenge of 100 grams after an overnight fast, In Phases II and III there was no alteration of the insti- tutional diet. In Phase IV each person was on a high carbohydrate diet for the 3 days prior to the test. 27 Table V. Carbohydrate intake within 4 hours preceding glucose challenge: 24 men,Milan, Phase III Case number Intake in grams Mean Range 06m mmm mmm mm mee eee mmeem meee 88.7 60-109 0m mmm mm mm mmm ee ee ememn 97.6 63-123 05m mm mmm mmm mm mm ee eee 93.9 40-159 08m mm mmm mm mm mm eee 115.1 55-242 10m mmm mmm mmm mm me eee 73.1 50-110 Imm mm mmm mmm mm mm ee ee meme meee 112.0 85-150 20mm mmm mm mmm mm ee eee 90.6 48-180 2mm mmm mmm mm eee ee mmm 66.0 30-103 23m mmm mm mm me ee meee 81.4 39-137 26m mmm mm mmm mm emma 94.8 19-165 28m mm mm mm mm ee emcee 70.9 35-103 29m mmm mmm mm emma 66.9 19-138 33mm mmm mm me ee ee 108.3 79-162 36m mmm mmm mmm meme 64.9 25-120 Bm mmm mm emma 60.6 15-130 Bm mmm mm mmm ee ee meee 93.8 50-139 Alm mmm mmm me mm meeen 72.4 44-148 A5m mm mmm mmm mm een 72.8 42-119 Bm mmm mm mm mmm eee 74.8 25-149 Bm mmm mmm mmm mm emma 106.1 50-159 AB mm mmm eae 82.5 50-99 BO mmm mm mmm mm meena 94.9 64-127 50mm mmm mmm ee ee eee mmm 106.6 79-188 59m mmm mmm mm mm ee ee meme mma 60.1 35-103 28 Table VI. Clinical classification of study participants by specified tests: 24 men, Milan, Phases III and IV Phase III Tests? Phase IV Tests! Case number Age Race SGTT1 SGTT2 SGTT CGTT Ol-vemreermm me ————————————————— 44 | White 0 0 0 + Obm-=mmmmmmmmmmm meme mm mmc cc mmm em me 41] White 0 0 0 05mmmmmmmm mmm mmm me mmm mmm mmm 43 | Negro 0 0 0 0 08-----mmmmmmm mmc mem ———————— 46 | Negro 0 0 -—— -— 10=mmmmmmmmm mmm cme meme mm mmm mmm 47| Negro 0 0 0 0 llemmmmmmmmm mmm mmm mmm meme mo 48 | Negro 0 0 0 0 20==mmmmmmmmmemmmmm mmm mmm—mm—m momo 44 | White 0 0 °0 0 2]lmmmmmmmmmm mmm me mmm emm mmm me 42| White + 0 + + 23mm meme meme meme mmm mm mmm 41| White 0 0 0 26-=-mm=mmmmmemmmmm mmm —mmeeomm———mm= 43| White 0 0 ? 0 28----mmmmmmm meme mmm mmm mmmmm mmm 52| White 0 0 + + Zw 2 49 | Negro 0 0 + 33cm mmm mmm memo m mmm mmo 40 | White 0 0 0 0 3fmmmmmmmmmmmm meme mmm mmm 48 | White 0 0 -—- -—- femme meme mmm mmm mem ——— 46 | Negro 0 0 0 0 f3emmm meme emcee mmm mmm mmm meme 46 | Negro 0 0 0 0 Glhmmmm cme m mmm mem 43| White 0 0 0 0 45m mmm mmm mmm mmm mmm mmm mm 45| White 0 0 0 + Lfmmmmm mmm meme mmm mcmmmm mmm 40| Negro 0 0 0 0 f]mmmm mmm mmm meme mmm mm mmm mm mmo 43] White 0 0 0 0 L8mmmmmmm mmm mmm mm mmm mmm mmm mmm 42] White 0 0 0 0 LQmmmmmmm mmm meme mmm mmm mmm mm mm 41| White 0 0 0 0 50==mmm mmm mmm mmmmm mmm mm memo mm momo 41| White 0 0 0 0 50mm mmm meee mmm mmm mm mmm mmm 42| White 0 0 0 0 Ay sual diet. bHigh carbohydrate preparatory diet. CThe response curve was above 150 mg.% at 1 hour and 110 mg.% at 2 hours and would be classified as ‘‘probable diabetic’’ by Unger. NOTF: The SGTT (standard glucose tolerance test) and CGTT (cortisone glucose tolerance test) are defined in the text. The criteria used are those of Fajans and Conn. For the SGTT, if the response curve was above 160 mg.% at 1 hour, 140 at 1% and 120 at 2 hours the person was classified ‘‘diabetic’’ (+). Response curves lower than this but above 160, 135, and 110 mg.% at the same points were classified as ‘‘probable diabetic” (+). All others were classified as ‘‘not diabetic’’ (0). For the CGTT, levels above 160 mg.% at 1 hour and 140 mg.% at 2 hours were classified as diabetic (+) and there was no borderline class. 29 Table VII. Heights, weights, and changes in weight: 24 men, Milan, Phase III Case number WELT [TR EI |r nee OL em ie 68 1/2 175 175 0 is 67 1/2 186 184 -2 DIB cn em i 67 160 163 +3 08--memmmemmmem—m meme ——— 69 1/2 225 228 +3 JD stm em mr mm 68 164 160 -4 lleceecmeme—— eee m————————— 67 161 163 +2 20-==m mm mmmmmmeem——eeee 67 190 - Dm 68 1/2 160 157 -3 23 ne oe mm 68 1/2 184 180 -4 6 69 263 275 +12 28cm ———————————— 67 1/2 177 176 ml, 0 ro mm mm i EE 68 1/2 144 142 -2 33mm mmm 70 182 184 +2 36mmmmmmmm mmm een 68 1/2 150 154 +4 eR 69 1/2 214 217 +3 43mm mmm 68 172 169 -3 Ula yan Sa AR ARERR 69 1/2 176 176 0 45mmmmrmmm mm ————— mmm mm —————— 76 180° 176 -4 RE 68 178 - - GT mmm mn ———————— 68 1/2 167 165 -2 ER ARR om mms wn el 66 1/2 151 145 -6 LD) mmm rR 73 1/2 172 169 -3 Semin Posi i ss sn ss se 69 1/2 173 172 -1 BD ue at er rm 69 180 180 0 30 APPENDIX | TECHNICAL VARIABILITY OF BLOOD GLUCOSE DETERMINATION In any study, the reliability of measurement is an essential ingredient. For the Milan study, all blood glu- cose determinations were made by the laboratory of the Diabetes Field Research Unit of the Diabetes and Ar- thritis Branch, Division of Chronic Diseases, Bureau of State Services, U.S. Public Health Service. This laboratory has measured all of the blood glucose speci- mens of the Health Examination Survey. Blood specimens of about 3 ml. were collected at the prison in prelabeled B-D '""Vacutainers' (3204x, formula 44) containing 30 mg. of sodium fluoride. These were packed on ice within 3 hours of collection and were shipped by air mail, special delivery to the labo- ratory in Boston. Previous studies by the Health Exami- nation Survey on the effects of handling and shipping specimens had shown that these factors have no dis- cernible effect on the measurement. Tests were made in duplicate by the Somogyi-Nelson macromethod and the results were averaged. Generally, the laboratory work was performed the day after the specimens were collected. During most of the study, the same two technicians made all of the determinations, one of them measuring specimens for case numbers 1 through 29 (it varied slightly) and the other measuring the re- maining specimens. Thus, most of the Phase III speci- mens for any specific study person were measured by one laboratory technician. There are several gauges on the reliability of measurement during Phase III. The crudestis the weekly average for all specimens taken before challenge. Omit- ting case number 08, who came late into the study, and taking the value for the replicate week in the few in- stances where a specimen was missing for a given week, the average level before challenge varied from a low of 78.7 mg.% to a high of 88.7 mg.%. There was no indication of any trend with time in this average. Another gauge is the difference between specimens taken a week apart on the same individual after an overnight fast. This yielded a "'within-person'' standard deviation of 6.5 mg.%. As already noted, these two specimens were almost always measured by the same technician. An unknown part of the variation represents the biological variation of fasting blood glucose levels. The remainder is the technical variability for a single technician in two laboratory ''runs." The third gauge is given by the various control specimens and standards measured by each technician as part of the routine of each laboratory '"run''. These are primarily working devices for uncovering obvious laboratory aberrations. On the standard 100 (a concen- tration of 100 mg.% of glucose dissolved in water), technician A averaged 99.9 mg.% during Phase III and technician B averaged 99.8 mg.%. On the standard 200, they averaged 197.3 and 198.5 mg.%, respectively. The other laboratory controls are somewhat better indi- cators of technical variability. On the serum control, technician A averaged 1.12 mg.% lower than technician B. On the blood control pool, she averaged 0.56 mg.% higher. If half of the squared difference between meas- urements made each week is averaged,the figure which results can be designated as total technician varia- bility (within the same run). This average was 17.6 and 14.9 mg.% for blood and serum controls, respectively, or standard deviations of 4.2 and 3.9. Besides within- technician variability, these figures include a component of between-technician variability and between-run vari- ability. This will obviously be an underestimate of tech- nician variability. As a more accurate gauge, a series of control specimens was introduced at the prison each week. This was done by taking replicate specimens in sequence from case numbers 46 and42,and relabel- ing the blind duplicate with case numbers 12 and 14, respectively. Since these were bonafide case numbers used in Phase I, they could not be identified by the laboratory technicians as control specimens. Each week two specimens were sent to the laboratory for case number 12 (one for before challenge and one for 1 hour after challenge) and two for case number 14. Except for accidental losses, then, there were four pairs of replicate specimens in each laboratory series for a control. Because of the laboratory arrangements in force one specimen of each pair was measured by one technician, the replicate specimen of the pair by the other technician. Altogether there were 53 such com- parisons from Phase III. On the average, technician A measured the speci- mens 1.3 mg.% higher than technician B. Differences between replicate measurements ranged from 0.0 to 15.5 mg.%. This included an unknown variability arising 31 from field errors and differences in the handling of the specimens, a slight average difference between the level at which the two technicians customarily meas- ured, between-technician and between-run variability, as well as the ''pure' variability of the technicians. The total technical variability was 5.4 mg.%. It probably represents an overstatement of the technical vari- ability in our study comparison, since in most cases the same technician measured nearly all the specimens from a specific examinee. The set of specimens averaged quite close to a level of 100 mg.%. For the specimens taken after challenge, the variability was greater than for the specimens taken before challenge, which accords with the usual experience that technician variability rises with the level of the specimen meas- ured. The set of blind replicates which are most com- parable in general level with the regular laboratory controls were those taken 1 hour after challenge from case number 46. These yield a figure of 6.6 mg.% as compared with an estimate from the laboratory control specimens of 4.0 mg.%. Another series of blind replicates was obtained (from other specimens) by running aliquots from 25 specimens on an autoanalyzer at Ann Arbor to com- pare with regular determinations made at Boston during weeks 6-9 of Phase III. Besides showing that the auto- analyzer measured blood glucose concentration an average of 2.4 mg.% higher than did the technicians at the Boston laboratory, this comparison showed that during those weeks technician A was measuring 4.25 mg.% higher than technician B. The regular series of blind replicates for the same 4 weeks (an entirely different series of specimens) indicated an average technician difference of 4.23 mg.%. The almost exact agreement is, of course, quite accidental, but it does argue for the reliability of the control series intro- duced into the trials. One final gauge may be mentioned. As an experi- ment, aliquots were drawn from one of the study parti- cipants (case number 59) during the course of Phase III. They were given a dummy case number (13), frozen, and retained frozen until after the study was compieted. Twelve weeks after the end of Phase Ill they were thawed and shipped to the Boston laboratory for determination. 32 If this process introduced no serious artifacts into the measurement, this series might uncover any laboratory drift that might have occurred during the study. All told, there were 24 specimens in this series (a pair of specimens for each of 12 weeks in Phase III). Twenty-two of these specimens were measured by the same technician (technician A) both on the original aliquot and the frozen aliquot. For all specimens, except those for weeks 3, 4, and 14, the determinations on the frozen aliquots were higher than the original determinations. For 7 of the 12 weeks, the average difference between the original pair of determinations and the subsequent pair was less than 3 mg.%. The average difference was larger than this only for weeks 3, 4, 6, and 7, the largest being for week 7—38.25 mg.%. Admittedly, these data will not support a heavy load of inference but at the very least they can be said togive noevidence of a laboratory drift during Phase III. “If the differences between the original and frozen specimens can be regarded as representing the varia- bility of technician A over the entire period of Phase III, the number to assign to thatvariabilityis 5.0 mg.%. The average level of these specimens is 113.6 mg.%. It is possible to summarize the various indices of technical variability as follows. There are two measures of within-technician between-run variability, that from the fasting specimens for the same person and that from the frozen aliquots. The first standard deviation is 6.5 mg.% and the second is 5.0 mg.%. Since the first measure also includes a component arising from bio- logical variation, it probably is an overstatement of the technical variability. Then there are the measures of variability from the control specimens and the blind replicates. While these came from measurements done the same week, they were derived from different labo- ratory runs and in addition include variation arising from technician differences. The first was 4.0 mg.% and the second 5.4 mg.%. It should be reiterated that in most instances specimens for the same person were measured by one technician through all of Phase III. It seems reasonable to conclude from all the evi- dence that the effective technical variation for Phase III did not exceed 5 mg.%. APPENDIX I NOTES ON THE STATISTICS The chief focus ot the study was Phase III. During Phase III, each of 24 persons had each of 8 procedures performed in duplicate. Thus for any one procedure X, there will be two measures of blood sugar levels at 1 hour after challenge, x; and x. 2 2 Xx; + xg has a variance (x; - Xx,)” =d 1 For one procedure undertaken on one person, X; and X, may be considered as statistically independent from y, and y, from any other procedure Y yielding blood glucose values for the same person. The symmetry of these experimental arrangements leads to the nice result that comparing procedures for each person and pooling the results is exactly the same as comparing the mean levels for all 24 persons. Hence if X and ¥ are the mean blood glucose levels for pro- cedures X and Y then : ~Y has a Student's -t distri- x,y bution with 48 degrees of freedom, where 2 = = n 2 xy $2+82. $2 is, of course, Z g ,whered, = x, - x I. = (2m)? IT for any one person and n is the number of persons. In general, Se is the value given in table 2 divided by 17 (48). This test amounts to a comparison of differences between procedures against the ''within-person, within- procedure'' variability. An alternative procedure is to compare the average difference between procedures against the ''between-person, between-procedure'' vari- ability. For this purpose, form the statistic (x; + X,) - (y; + yy) = pfor each person and compute the average of these values p for all 24 persons and the variance wi? g IZM; -P of these values 5 = To where n is the number of persons. Test, then, to see whether PF is 2 1/2 s5/m/ significantly different from zero, using Student's -t distribution with 23 degrees of freedom. This was the procedure used for testing the urine glucose values. It was not used in testing blood glucose values, although it is a procedure that many analysts would prefer. The procedure used for testing differences in blood glucose levels is specially vulnerable to the situation where a few persons show large differences while most persons show almost none. To check against this possi- bility, a sign test was used. Thus, where procedure X was being compared with procedure Y, (x; + X,) - (vy; +9) will either be positive or negative. If 24 persons are compared, the null hypothesis calls for 12 differences to be positive. If 17 or more are posi- tive, there are more positive values than would be expected by chance. Here, as elsewhere, tests are made at a level of 5%. In this instance, the test will always be a one-sided test, since it is intended as a check on conclusions already drawn from a test of differences in means. The rank correlation computed in the text is the one proposed by Spearman, If X; is the rank of the & person under procedure X and Yi is his rank under s2d procedure Y and d, =x; =; thenr = 1 - n -n where n is the number of persons. r has the approxi - 2 1 -1 7 2 mate variance of S° = 7 and — is distributed as Student's - t with n - 2 degrees of freedom. The test is a one-sided one. Another test performed was for differences between variances. For this purpose a rough approximation was used by first computing the pooled variances for each person and testing these for homogeneity between persons and then performing a similar test for the pooled variance for each procedure. The statistic used was Hartley's M-statistic which is tabled in the Bio- metrika Tables for Statisticians, Volume I. Two minor issues merit consideration. The first is the handling missing data. There were 5 occasions where a replicate measurement was not available for blood glucose levels before challenge and 4 where a 33 replicate measurement was not available for blood glu- cose levels after challenge. These are, of course, trivial omissions. In these cases the internal variance can be estimated to be the same as the average for the other persons tested by the specified procedure and all tests can be performed as if the missing information was present. In no case was the actual number of de- grees of freedom less than 23 for any procedure and except for borderline tests the effect of assuming 24 degrees of freedom is negligible. The second minor issue is the handling of abnormal data. There are several instances where the blood glu- cose level reported seemed unlikely for the person and the circumstances. There is no really satisfactory way of dealing with such cases. In general the best solution is to accept the data. But there are occasions when it seems completely inadmissible to accept the data. There were five such specimens in our series. Specifically: In week 16, the fasting blood glucose for case number 23 was reported as 149.0 mg.%. This value was discarded and not replaced. In the 15 other cases where a blood specimen was taken before challenge from this person the level ranged from 62.5 to 101.0. The three other fasting specimens for this person were 72.5, 80.5, and 76.0 mg.%. In week 15, case number 36 gave a fasting blood sugar level of 140.0 mg.% and a value 1 hour after 34 challenge of 59.5 mg.%. These values were both dis- carded and not replaced. Both values are outside the range of other comparable values for this person and are very different from the paired values in week 16. It seems likely that their labels were reversed. In week 15, case number 59 had values before and after challenge of 146.5 and 78.5 mg.%, respectively. Blind replicates for these same specimens were 144.0 and 83.5 mg.%. These values are both discarded for the same reasons as in the preceding case. The pro- cedure in question was repeated on this person in week 18 and the results from this are used as replace- ments. One last comment is in order. The study called for each person to be submitted successively to all factors under study. The major breach in this design was to divide the study participants into breakfast and lunch groups,with the expectation that varying the cir- cumstances of challenge after breakfast would have the same effect as varying them after lunch. The results did not bear out these expectations. It is always difficult to decide whether it is pref- erable to be able to make a limited statement with great assurance or to attempt to learn more at the risk of decreased precision. In this instance, more information was obtained than a rigid design would have allowed at the cost of a serious loss in neatness. # U.S. GOVERNMENT PRINTING OFFICE : 1963 O — 690-807 VITALand HEALTH STATISTICS DATA EVALUATION AND METHODS RESEARCH NATIONAL A CENTER Series 2 For HEALTH Number 4 STATISTICS comparison of two methods of H(A] ITT FE Life Tables by reference to a “standard” table See inside of back cover for catalog card. Public Health Service Publication No. 1000-Series 2-No. 4 For sale by the Superintendent of Documents, U. S. Government Printing Qffice Washington, D.C. 20402 - Price 15 cents NATIONAL CENTER| Series 2 For HEALTH STATISTICS | Number 4 VITALand HEALTH STATISTICS DATA EVALUATION AND METHODS RESEARCH comparison of two methods of Constructing Abridged Life Tables by reference to a “standard” table Comparison of the revised and the prior method of constructing the abridged life tables for the United States. Washington, D.C. February 1964 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Anthony J. Celebrezze Luther L. Terry Secretary Surgeon General NATIONAL CENTER FOR HEALTH STATISTICS Forrest E. Linder, Ph.D., Director Theodore D. Woolsey, Deputy Director Oswald K. Sagen, Ph.D., Assistant Director (for Professional Relations) Walt R. Simmons, M.A., Statistical Advisor Alice M. Waterhouse, M.D., Medical Advisor James E. Kelly, D.D.S., Dental Advisor Louis R. Stolcis, M.A., Executive Officer DIVISION OF HEALTH RECORDS STATISTICS Monroe G. Sirken, Ph.D. , Chief Public Health Service Publication No. 1000-Series 2-No. 4 CONTENTS Introduction === === cme eee Method of ConsStruction=---===-mcme mcm ccccceee ms Constructing the 1959 Abridged Life Tables---------=-o--o-- Table 1. Conversion factors based on decennial life tables for the United States, 1949-51---ocmmmemmcmmmc cece Evaluation of the Abridged Life Table Methods-------------- Table 2. Differences between values of expectation of life in the complete life table and in abridged life tables, by color, sex, and age: United States, 1949-51----- Table 3. Differences between values of the probability of dying in the complete life table and in abridged life tables, by color, sex, and age: United States, 1949-51----- Appendix: Explanation of the Columns of Table A----------- Appendix Table A. Computation of abridged life table for the total population of the United States, 1959---------- SYMBOLS Data not available======mcmemm ee eeeee ee Category not applicable--=-==ecaceooaaoo- Quantity zZero==-===mmmmmce meee Quantity more than O but less than 0,05---- Figure does not meet standards of reliability or precision----------c-cuo-- COMPARISON OF TWO METHODS OF CONSTRUCTING ABRIDGED LIFE TABLES INTRODUCTION The publication of an annuai series of abridged life tables for the United States was started in 1945. After small biases were detected in the values of 1950 U.S. abridged life tables, studies were undertaken which led to the development of a revised method for constructing the U.S. abridged life tables. This report outlines the re- vised method used in constructing the abridged life tables since 1954. The construction of the life table for the total population for 1959 is shown in appendix table A. An earlier report! out- lined the method used in preparing the abridged life tables for the years 1946 to 1953 inclusive, which henceforth will be referred to as the original method. A test of the accuracy of the revised method of constructing the U.S. abridged life tables is pre- sented which involves a comparison of the 1949- 51 abridged life tables constructed by the revised method with the complete decennial 1949-51 life tables which were constructed by elaborate and laborious methods®. The 1949-51 abridged life tables constructed by the original method are also compared with those derived from the 1949-51 life tables, Comparing the abridged life tables, con- structed by original and revised methods, with the decennial life tables provides atest of the relative accuracy of these methods of constructing the U.S. abridged life tables. This report was prepared by Monroe G. Sirken, of the Division of Health Records Statistics. METHOD OF CONSTRUCTION The original and the revised methods of con- structing the U.S. abridged life tables have in common the fact that each involves reference to a standard life table. According to this method of constructing abridged life tables, certain relation- ships among the functions of the life table under construction are assumed to be the same as those of another life table already existing (referred to as the ''standard' table). In the calculation of the annual abridged life tables since 1954, the de- cennial U.S. life tables 1949-51” have been used as standard tables. When the 1959-61 decennial life tables are constructed, they will become the standard life tables in constructing the U.S. abridged life tables. The method presented here is based on an observed relationship between the probability of death ¢ g )and the age-specific deathrate Gu,). The function gq, is the proportion nx where I is X the number of survivors to exact age x in the hypothetical life table cohortand d, isthe number of the group who die before reaching exact age x + n. The function nu, is the quotient of the num- ber of deaths between exactages x and x +n during the year and the size of the living population be- tween these exact ages. The age-specific death rate may be defined either in terms of observed population data ¢,M,) or in terms of the stationary population of the life table (m,). The former M,) is the quotient of the number of deaths in a given calendar year between exact ages x and x+n and the midyear population between those exact ages. The latter (;m,) is the number of deaths (,d,) in the life table divided by the number of persons (,L,) in the stationary population of the life table between ages x to x+n. According to the revised method of con- structing the abridged life table, the relationship between gq, and 4 is given by the formula n u (1) a, = rr 1+ (ay) pu, It will be observed that formula (2) generates 2 sets of conversion constants according to whether u, is defined as ,M, the observed age-specific mortality rate, or as ,m,, the age- specific mortality rate of the stationary population of the life table. The constants «, are uséd n xX as adjustment factors to convert the observed population age-specific mortality rates into the values on .q, of the abridged life table. The constants « are used to calculate the values ny of ,L, from the values of / and d, in the abridged life table. Thus, (3) %m,= or (4) nly = ni, _ Um andy + Greville! has also suggested the use of formula (4) to calculate the L-function in the construction of the abridged life table by reference toa stand- ard table. The assumption underlying the abbreviated method of life table construction used here is that in each age interval x to x+n, the constants ay, Cu =M,m) may be regarded as having the same value in the life table under construction as 2 in the standard table. The constants « M and n xX a that have been used in the construction of n Xx the abridged life tables since 1954 are presented in table 1. They were derived by formula (2) according torelationships observed between gq, and pu, in the complete U.S. life table for the decennial period 1949-51. Until more current standard tables (U.S. life tables for the decennial period 1959-61) are constructed, these constants will be used each year to construct the U.S. abridged life tables. CONSTRUCTING THE 1959 ABRIDGED LIFE TABLES Basic sources of data used in the preparation of the U.S. life tables for 1959 were the annual mortality tabulations of the National Vital Sta- tistics Division and estimates of the population on July 1, 1959, by age, color, and sex pre- pared by the U.S. Bureau of the Census. Values of ,M,, the observed population age- specific mortality rates were obtained from the basic mortality and population data. The values of gq, were calculated by formula (1) using the set of constants «nr presented in table 1. (The nox method of calculating the values of the proba- bility of death during the first year of life and of the final age group 85 years and over is de- scribed below.) After the values of a, had been obtained, the I, and ,d, functions were computed in the conventional manner, according to the formula ndy = (1) (a); 1, = 1 a, : +n x WR Thereafter, the values of ,L, were calculated by formula (4) using the set of constants, a on n x presented in table 1. The values of 7; were ob- tained by summing the ,L, column, starting with the oldest age group. In other words, Li=T, int nlx Table 1. Conversion factors based on decennial life tables for the United States, 1949-51 Total Male Female popu- Age lagion Total White Nonwhite| Total White Nonwhite interval (years) n__ _1 aM, = 2, My l=-5mmmmmm mr ——— 18.7253 19,0755 18.5164 16.7398 17.8984 18.6698 15.7642 i 17.1188 16.1574 21.3402 9.5389 24,9787 22,5984 | 21.2327 10~15=mmmme mma -27.7680 -28.3119 -31.7154 | -15.5642| -22,0673 §{ -18.0455| -44,.7131 15m 20m mm mmm mm = 10,2732 7.1700 7.8382 .1092| 18.0825 19.9746 8.9045 20-25= mm mmm ———- 7.5326 7.2706 8.1040 4.9690 5.8762 6.3873 345398 25=30n===mmmmm . 1806 -.2167 -1.0433 1.7692 .3806 -1.3368 2,229] 30=35mm mmm -1.0910 1.2348 2.6955 549535 -3.9384 3.0035 2.7942 35-407 mmm mmm .9558 1.1954 1.6008 -1.8298 .7362 1.1471 3.3718 40-45~==~~==-~ 2.2822 2.1367 1.7854 4.0718 2.3874 1.0406 5.9329 45% 50m mm mmm m= 1.6621 1.8081 2.1524 4126 1.4901 2.1793 -.0091 50m 55m mmm www w= 2.2507 +2277 2.1262 2.9371 2.3393 2.0834 2.9772 55=60=m=mmmm—- 2.2598 2.3750 2.3389 2.5563 2.0481 1.9335 2.3614 60=65===mmm=== 2.4041 2.3848 2.368) 2.4971 2.4313 2.3863 2.6988 65-70~mmmmmn= 2.2343 2.3584 2.3431 2.4228 2.0201 2.0204 1.9844 10-75 =m mmm 2.3399 2.3872 2.3807 2.4814 2.2793 2.2374 2:7213 15-80====mmmin= 2.4376 2.5014 2.5026 2.5300 2.3645 2.3595 2.4964 80=8 5m mmm mm ee 2.5307 2.5607 2.5621 2.5478 2.4998 2.4987 2.5590 1 L «mm = X — NX nox dy l-5mmmmmmm mmm 2.4152 2.3990 2.3664 2.5044 2.4354 2.4212 2.4671 5-10=mmmmm==m= 2.6834 2.6602 2.6537 2.7082 2.7176 2.6996 2.8127 10-15 ===m=e== 2.3174 2.2879 2.2954 2.2475 2.3634 2.4046 2.1910 15-20 =mmmm=m= 2.3205 2.3074 2.3189 2.2390 2.3373 2.3663 2.2543 FO Blowin wrist 2.4423 2.4542 2.4687 2.3895 2.4178 2.4252 2.3851 DB Bimmer 2.4468 2.4757 2.4863 2.4377 2,4035 2.4006 2.4064 300 35 w mom mm 2.3839 2.3920 2.3866 2.4039 2.3135 2.3678 2.3792 | ——— 2.3421 2.3375 2.3244 2.3909 2.3507 2.3427 2.3711 40-45-=rmmm=m= 2:3293 2.3233 2.7740 2.3698 2.3411 2.3318 2.3658 45=50===m=m=mm 2.3437 2.3370 2.3287 2.3703 2.3563 2.3436 2.3834 50-55---======- 2.3536 2.3521 2.3427 2.3946 2.3609 2.3436 2.4024 55=60===mmmmm- 2.3700 2.3820 2.3725 2.4442 2.3569 2.3455 2.4361 60-65----==-=-~ 2.3980 2.4145 2.4054 2.4912 2.3731 2.3392 2.4729 65-70~=======-- 2.4055 2.4356 2.4272 2.5175 2.3640 2.3569 2.4974 70-75 =m ancene 2.4280 2.4624 2.4569 2.5313 2.3876 2.3488 2.5064 15~-80-=m=mm==-= 2.4879 2.5242 2:3223 2.5607 2.4499 2.3780 2.5321 80-85-==m=mm==- 2.5747 2.6051 2.6061 2.5985 2.5449 2.4453 2.5693 The values of the average remaining lifetime was then obtained by division &, = T, + I - Formulas (1) and (3) respectively were not used to compute the gq and L, functions for the first year of life and the final age group 85 years and over. Rather, the special treatment of these age groups used in the construction of U.S. abridged life tables for the years 1945 to 1953 inclusive was continued. The following explana- tion has been adapted and extracted from a report that describes the method used to construct these earlier tables.! For the age group 85 years and over formula (2) shows that “nr ngs Hence the assumption that the value of « M ootvhes is the same in the life table under construction as in the standard table is not useful, and some other assumption must be made. Instead, the ratio a,defined as the quotient of the value of «My based on the actual data by the corre- sponding value _m Mg, for the stationary popu- lation of the life table was assumed tobe the same in the table under construction as in the standard table. But ~My is the reciprocal of 8 the average remaining lifetime. Thus, the value of 4 can be computed by the formula fC 85 M «gs According to the standard tables (1949-51), 4=.9487119 for the total population. The values of « for the 4 subdivisions of the population by color and sex are shown below: Subdivision of the population A White males .9610759 White females .9554947 Nonwhite males .8534401 Nonwhite females .8072982 is infinite since n=o. The abridged life table for 1959 can then be computed since The value of q, the proportion of liveborn infants dying before reaching age 1, is computed from birth and death statistics, being taken as equal to the adjusted infant death rate. A method of adjusting the infant death rate for the changing number of births is described ina previous publi- cation.® The adjustment is made by allocating the deaths of infants occurring during a given year to the year in which the infants were born. The infant deaths so allocated are then related to the births occurring in the respective year of birth. The expression for computing the adjusted infant mortality rate per 1,000 live births may be written: Adjusted rate = [2g 2) x 1,000 where D - number of infant deaths occurring in the given year. f =ratio of deaths occurring in the given year among infants born in the preceding year to the total infant deaths of the given year. This is referred to as the "separation factor." E = number of births occurring in the given year. E' -number of births occurring in the pre- ceding year. The stationary population in the first year of life was obtained by the formula L,= 1, - (I-f)d,, EVALUATION OF THE ABRIDGED LIFE TABLE METHODS A set of U.S. abridged life tables, 1949-51 for subdivisions of the population by color and sex was constructed by the revised method of con- struction by reference to a standard table. Values of the constants « and « needed in the nx nix construction of these tables were derived from the complete U.S. life tables, 1939-41, which served as the standard tables. The decennial U.S. life tables 1949-51 were the criterion tables for the evaluation of the precision of the abridged life tables. The basic data used in the preparation of the U.S. abridged life tables 1949-51 were essentially the same as those which had been usedin the pre- paration of the complete U.S, life tables 1949-51. These included mortality data by age, sex, and color for the 3-year period 1949-51, extracted from the annual issues of the Vital Statistics of the United States published by the National Vital Statistics Division, and population data by age, sex, and color enumerated in the 1950 Census and published by the Bureau of the Census in U.S. Census of Population, Volume II, "Characteristics of the Population." There is close agreement (table 2) between the values of the expectation of life based on the complete life tables and those based on the re- vised abridged life table method. The abridged life table values exceed the decennial life table values at virtually all ages but the differences are small. For example, the difference between the values of the expectation of life at birth was only .01 years for the total population; it was less than .03 years for white males, white females, and nonwhite males; and .15 years for nonwhite fe- males. For each of these population groups, there is a tendency for the differences between the values of the expectation of life to increase with advancing age. Atvirtually all ages the differences are greater for nonwhite than for white persons, and within each color group, the differences are greater for females than for males. Using the same basic data, that is the popu- lation data from the 1950 Census and the mor- tality data for the 3-year period 1949-51, another set of abridged U.S. life tables 1949-51 were pre- pared by the original abridged life table method. This is the method of construction by reference to a standard table, that had been used to construct the annual abridged U.S. life tables, 1945-53. The assumptions underlying the original method are that in each age interval x to x+n, h, defined as the ratio a, + oM, and the values of the ratio Jj, = 2Lx + (ILt+ I 4, ) were as- sumed to have the same value in the life table under constructions as in the standard table. Values of the constants ,h,and ,j, needed in the construction of the abridged U.S. life tables 1949-51 by the original method were available! for the decennial U.S. life tables 1939-41 which served as the standard tables. The values of expectation of life based on the original method exceed those of the decennial life table at every age (table 2). At virtually every age, these differences are greater than the amounts by which values of expectation of life based on the revised method exceed those based on the decennial life table. Thus, for the total population the value of expectation of life at birth according to the decennial life table is exceeded by .01 years according to the revised method and it is exceeded by .15 years according to the original method. Itis noteworthy that both methods of constructing life tables by reference to a standard table slightly overstate the values of the expectation of life at every age, although the over- statement is consistently less for the revised than for the original life table method. The absolute values of difference between a, values based on the decennial life tables and on the abridged life table are virtually always smaller for the revised than for the original abridged life table method (table 3). Furthermore, the original method in most age groups under- states the values of apa tendency which is not evident for the revised method. REFERENCES Thomas N. E. Greville and Gustav A. Carlson, ‘“Method of Constructing the Abridged Life Tables for the United States, 1949,” Vital Statistics--Special Reports, Vol. 33, No. 15, June 30, 1953. 2fontoe G. Sirken and Mortimer Spiegelman, ‘Method of Con- structing the 1949-51 National, Divisional, and State Life Tables,” Vital Statistics--Special Reports, Vol. 41, No. 5, July 31, 1959. 3Monroe G. Sirken, ‘United States Life Tables, 1949-51," Vital Statistics--Special Reports, Vol. 41, No. 1, November 23, 1954. 4Thomas N. E. Greville, “On the Formula for the L-Function in a Special Mortality Table Eliminating a Given Cause of Death,” Transactions of the Society of Actuaries, Vol. VI, Meeting No. 14, April 1954. Siwao M. Moriyama and Thomas N. E. Greville, ‘Effect of Changing Birth Rates Upon Infant Mortality Rates,’ Vital Statistics-- Special Reports, Vol. 19, No. 21» November 10, 1944. 6Thomas N. E. Greville, “United States Life Tables and Actu- arial Tables, 1939-41,”’ U.S. Government Printing Office, 1947. Table 2. Differences between values of expectation of life in the complete life table and in abridged life tables, by color, sex and age: United States, 1949-51 White Nonwhite Abridged life table values Abridged life table values 0 minus 0 minus Sex and ¢ based complete life table values | © baged complete life table values age X on on complete complete Pa, Original Revised ial Original Revised abridged abridged € abridged abridged method method method method MALE 0-lrmemmm—— 66.31 .10 .00 58.91 .16 .02 1-5-mrmunn- 67.41 .10 = «01 61.06 «17 +02 5-10====m===- 63.77 .11 .00 57.6% sal .06 10-15-~-===~~ 58.98 +12 .00 52.96 +23 .07 15-20---=-~-- 54.18 .12 .01 48.23 "22 .06 20-25====== 49.52 «12 .00 43.73 v3 +07 25-30==m=m-- 44,93 «11 .00 39.49 ied +07 30-35--===~ 40.29 .12 .00 35.31 23 +07 35-40-==m=- 35.68 wld: .00 21.21 wl .09 40-45-=m=m- 31.17 wli2 .00 27.29 .22 .07 45-50====-~ 26.87 i .00 23.59 «24 .10 50-55-===== 22,83 wl) .00 206.25 22 +09 55-60=====~ 19.11 v12 .00 17.36 «19 .08 60-65-~=~~-- 15.76 .12 .00 14.91 21 +11 65-70-===== 12.75 14 .01 12.75 +27 +13 70-75-====- 10.07 «lS + 0% 10.74 3 .20 15-80===wm= 7.17 «13 .01 8.83 Jb +27 80-85----~-~ 5.88 vial .02 7.07 .54 .40 85mm mmm 4.35 .05 .05 5.38 .61 +61 FEMALE 0-l-mmwmmm- 72.03 .20 .02 62.70 .30 +15 1-5-»mmmmmn 12.77 «20 «01 64.37 .31 +15 5-10-===mm= 69.09 22 .02 60.93 .37 «19 10-15---~=-- 64.26 v2 .02 56.17 .36 .18 15-20----~-- 59.39 .21 .02 51.36 .37 .18 20-25-===== 54.56 +22 .03 46.77 .37 +19 25-30~===m=~ 49.77 22 .03 42.35 NY +19 30-35~~===~ 45.00 .22 .03 38.02 .37 .20 35-40-====- 40.28 roe .03 33.82 .36 «20 40-45------ 35.64 vee .03 29.82 35 .19 45-50---~-~ 31.12 22 .03 26.07 .38 «22 50~55==mmm= 26.76 w22 .03 22.67 40 .24 55-60------ 22.58 .23 .03 19.62 +37 24 60-65-----~ 18.64 vod .03 16.95 «39 +28 65-70------ 15.00 wl .03 14.54 45 «35 70-75------ 11.68 22 .03 12,29 .51 J4l 75-80----~~ 8.87 woe .05 10.15 +36 «57 80-85----~-- 6.59 «19 .06 8.15 74 77 85+-mmmmmm- 4.83 wll + LL 6.15 1.18 1.18 Table 3. Differences between values of the probability of dying in the complete life table and in abridged life tables, by color, sex, and age: United States, 1949-51 White Nonwhite Abridged life table values Abridged life table values Sex and minus q minus age ny based complete life table values |” x Desed complete life table values interval complete complete Life Original Revised Site Original Revised abridged abridged abridged abridged method method me thod method MALE 1» 5emmmm= .00544 .00016 .00015 .01043 .00080 .00067 5-10-====~ .00347 .00001 .00002 .00498 -.00002 .00001 10=15===~ .00354 -.00005 -.00006 .00522 -.00005 -.00006 15-20 === .00652 .00002 .00007 .01102 .00000 .00000 20-25-~-~-- .00852 .00002 .00012 .01801 -.00007 .00001 25-30-~--- .00853 -.00003 -.00001 .02168 -.00018 -.00003 30-35-»=~ .01013 -.00002 .00004 .02703 -.00013 .00028 35-40--~-- .01480 -.00008 -.00004 .03616 -.00077 -.00066 40-45-=~~ .02381 -.00009 -.00003 .05005 .00031 .00086 45-50 = == .03821 -.00021 -.00008 .07365 -.00198 -.00130 50-55==== .05963 -.00009 .00008 . 10658 -.00301 -.00110 55-60===~ .09098 -.00032 .00000 14721 -.00073 .00031 60-65-~~-~ «13163 -.00064 -.00008 .18614 .00028 .00066 65=70===~ . 18580 -.00142 -.00018 .22524 .00094 .00123 70-75==== .26348 -.00287 .00021 .27260 -.00325 -.00092 15-80-»=~ .37002 -.00838 .00025 .33636 -.00581 .00149 80-85---- 49946 -.02093 .00330 Jalbbh -.02013 -.00279 FEMALE 1-5-==mm- .00457 .00011 .00011 .00894 .00059 .00047 5-10---=~ .00246 .00013 .00002 .00396 .00002 .00004 10-15---- .00210 .00000 -.00001 .00355 .00001 -.00006 15-20---- .00312 .00001 .00002 .00846 -.00008 .00004 20-25-~-~-~ .00396 -.00001 .00001 01291 -.00009 .00003 25~30=m== .00485 -.00001 .00000 .01665 -.00009 .00001 30-35-»~~ .00657 -.00001 .00000 .02196 -.00042 -.00012 35-40--~~ .00945 -.00006 -.00003 .03100 -.00059 -.00061 40-45---- .01440 -.00005 -.00002 .04410 -.00008 .00067 45-50---- .02200 -.00012 -.00003 .06382 -.00028 -.00020 50-55-~~~ .03294 -.00016 -.00001 .08845 -.00289 -.00101 55-60---~- .05039 -.00041 -.00008 .12020 -.00165 -.00013 60-65--~-- .07812 -.00080 -.00001 13221 -.00022 .00119 65~70==== «12021 -.00219 -.00050 .18615 -.00278 -.00145 70-75---- .19465 -.00401 -.00016 .22601 -.00008 .00188 715-80-~~~ .30096 -.01071 -.00114 .28105 -.00445 .00041 80-85---- .43860 -.02056 -.00257 .34418 -.00583 .00117 APPENDIX EXPLANATION OF THE COLUMNS OF TABLE A Column 1—Age interval (Xto x+ n).—~The age interval shown in column 1 is the interval between the two exact ages indicated. For in- stance, ''20-25" means the 5-year interval be- tween the 20th and the 25th birthdays. Column 2— Population (,P,).—This column shows the estimated midyear population for the indicated age interval. Births for 1958 and 1959 were used in computing gq. Column 3—Deaths (,D,).—This column shows the number of deaths for the age interval during 1959. Columns 4 and 5— Death rates (,M,) .—The age-specific death rate shown in column 4 is the central death rate for the age interval. In column 5, these rates have been adjusted proportionately for deaths for which age was not reported on the death certificate. Column 6— Conversion factor (a n Xx y .— This column is derived from a ''standard' table, in this instance, the life table for the total popu- lation of the United States, 1949-51. These con- version factors are shown in table 1. Columns 7 and 8— Proportion dying ( a) = The number shown in column 7 is the denominator of the proportion of the cohort dying in the age interval according to formula (1), page 3. Column 8 shows the proportion of the cohort who are alive at the beginning of an indicated age interval who will die before reaching the end of that age interval. For example, for the population in the age in- terval 20-25, the proportion dying is 0.0061 —out of every 1,000 persons alive and exactly 20 years old at the beginning of the period, 6.1 will die before reaching their 25th birthday. In other words, the 9, values represent probabilities that persons who are alive at the beginning of a spe- cific age interval will die before reaching the beginning of the next age interval. The "'propor- tion dying" column forms the basis of the life table; the life table is so constructed that all other columns are derived from it. Column 9— Number surviving (1,).—This column shows the number of persons, starting with a cohort of 100,000 live births, who sur- vive to the exact age marking the beginning of each age interval. The 1, values are computed from the ,, values, which are successively ap- plied to the remainder of the original 100,000 per - sons still alive at the beginning of each age in- terval. Thus, out of 100,000 live born babies, 97,357 will complete the first year of life and enter the second; 96,948 will begin the sixthyear; 96,051 will reach 20; and 17,877 will live to age 85. Column 10-- Number dying ( d,).—This col- umn shows the number dying in each successive age interval out of 100,000 live births. Out of 100,000 persons born alive, 2,643 die in the first year of life, 409 in the succeeding 4 years, 584 in the 5-year period between exact ages 20 and 25, and 17,877 die after reaching age 85. Each figure in column 10 is the difference between two successive figures in column 9. Column 11— Conversion factor (ap, ).— n X This column is derived from a "standard table, in this instance, the life table for the total popu- lation of the United States, 1949-51. These con- version factors are shown in table 1. Columns 12 and 13— Stationary population (Ly, and T,) .—Suppose that a group of 100,000 individuals is born every year and that the pro- portions dying in each such group in each age in- terval throughout the lives of the members are exactly those shown in column 8, If there were no migration and if the births were evenly distri- buted over the calendar year, the survivors of these births would make up what is called a stationary population—stationary because in such a population the number of persons living in any given age group would never change. Thus, a census taken at any time in such a stationary community would always show the same total population and the same numerical distribution of that population among the various age groups. In such a stationary population supported by 100,000 annual births, column 9 shows the number of per- sons who, each year, reach the birthday which marks the beginning of the age interval indicated in column 1, and column 10 shows the number of persons who die each year in the indicated age interval. Column 12 shows the number of persons in the stationary population in the indicated age in- terval. For example, the figure given in the age interval 20-25 is 478,829. This means that in a stationary population supported by 100,000 annual births and with proportions dying in each age group always in accordance with column 8, a census taken on any data would show 478,829 per - sons between exact ages 20 and 25. Column 13 shows the number of persons in the stationary population in the indicated age in- terval (column 12) and all subsequent age in- tervals. For example, in the stationary popu- lation referred to in the last illustration, column 13 shows that there would be at any given moment, a total of 5,030,781 persons who have passed their 20th birthday. The population at all ages 0 and above (in other words, the total population of the stationary community) would be 6,965,532. Column 14--Avevage vemaining lifetime 8, ) .—The average remaining lifetime (also called expectation of life) at any given age is the average number of years remaining to be lived by those surviving to that age on the basis of a given set of age-specific rates of dying. In order to arrive at this value, it is first necessary to observe that the figures in column 12 can also be interpreted in terms of a single life table cohort without introducting the concept of the stationary population. From this point of view, each figure in column 13 represents the total time (in years) lived between two indicated birthdays by all those reaching the earlier birth- day among the survivors of a cohort of 100,000 live births. Thus, the figure 478,829 in the age interval 20-25 is the total number of years lived between the 20th and 25th birthdays by the 96,051 persons (column 9) who reached the 20thbirthday out of 100,000 live born babies. The corresponding figure (5,030,781) in column 13 is the total num- ber of years lived after attaining age 20 by the 96,051 persons reaching that age. This number of years divided by the number of persons (5,030,781 divided by 96,051) gives 52.4 years as the average remaining lifetime at age 20. Appendix Table A. Computation of abridged life table AGE INTERVAL Pestn |Peath rete | oo, | Deno, Estimated Deaths g adjusted oye na opulation | in 1959 | fate un=- for age 51.00 formula Period of life pop : adjusted & factor (1) (See July 1, 1959 | within not stated between two Tn: page ) SRAEE Gna within age age stated 3 interval interval years Col. 3 Col. 4 (See 1#Col.5 Col. 2 1.00054 table 1) Col. 6 xtox +n 2 Dy oe MM, AM; (1) (2) (3) (4) {5) (6) (7) OE EE aatata ...| 112,008 ve ‘en ... eis l=5-mcmmmmc cree eam 16,000 17,116 | 0.001069 0.001070 18.7253 1.02004 5=10===mmmmmm mmm 18,703 9,028 .000483 .000483 17.1188 1.00827 10-15-=-~=mmmmmec—- 16,435 7,402 .000450 .000450 | -27.7680 0.98749 15-20=====cmmmmmmam 12,850 11,931 .000928 .000929 10.2732 1.00954 20=25-=m=mmmmm mmm 10,867 13,337 .001227 .001228 7.5326 1.00925 25=-30=====mmm—————— 10,922 14,084 .001290 .001290 0.1806 1.00023 30-35===mmmmm mmm 11,928 19,734 .001654 .001655 -1.0910 0.99819 35=40==mmmmmm mmm 12,299 28,477 .002315 .002316 0.9558 1.00221 40-45 =m mmmmm mmm 11,382 41,569 .003652 .003654 2.2822 1.00834 45-50 ==m=mmmm————— 10,907 62,544 .005734 .005737 1.6621 1.00954 50=55= mmm mmm am 9,575 87,521 .009141 .009145 2.2507 1.02058 55-60-=====mmmmmem— 8,228 114,895 .013964 .013970 2.2398 1.03157 60-65-=====mmmm———— 7,133 148,102 .020763 .020773 2.4041 1.04994 65-70-===mmmm—— mmm 5:752 191,536 .033299 «033315 2.2343 1.07444 70-75 ==mmmmm mmm mmm 4,284 214,256 .050013 .050037 2.3399 1.11708 75-80-====mmmmmmmm 2,971 210,524 .070860 .070893 2.4376 1.17281 80-85===mmmmmm mem 1,520 177,601 .116843 .116898 2.5307 1.29583 85 and overl------- 860 | 174,369 | .202753 .202850 che 5 LFor method of computing values at these ages, see text on page 4. for the total population of the United States, 1959 Proports STATIONARY POPULATION porcien Average of persons alive at sel {ui Number velit pe beginning iy SY any v ng dying | Conver- In this and vente, of age t KC as in age sion all sub- ine to interval our iio inter- factor In age sequent ng dying om alive val interval intervals OES during at age interval Sum of Col. 12 for this n Col. 5 Col. 9 (Line above)— | Col. 8 (See n Col. 9— line and Col. 13 col. 7 Col. 10 (Line above) Col. 9 | tablel) | (10) (11) all below Col. 8 2 1 nx Im, Lx 7, 8 (8) (9) (10) (11) (12) (13) (14) 0.0264 100,000 2,643 "tie 97,681 6,965,532 69.7 .0042 97,357 409 2.4152 388,440 6,867,851 70.5 .0024 96,948 232 | 2.6834 484,117 6,479,411 66.8 .0023 96,716 221 2.3174 483,068 5,995,29% 62.0 .0046 96,495 vin 2.3205 481,445 5,512,226 57.1 .0061 96,051 584 | 2.4423 478,829 5,030,781 52.4 .0064 95,467 616 2.4468 475,828 4,551,952 47.7 .0083 94,851 786 2.3839 472,381 4,076,124 43.0 .0116 94,065 1,087 2.342) 467,779 3,603,743 38.3 .0181 92,9738 1,684 | 2.3293 460,967 3,135,964 33.7 .0284 91,294 2,594 2.3437 450,390 2,674,997 29.3 0448 88,700 3,974 | 2.3536 434,147 2,224,607 25.1 .0677 84,726 5,737 2.3700 410,034 1,790,460 2:.1 .0989 78,989 7,814 | 2.3980 376,207 1,380,426 17.5 +1550 71,175 11,034 2.4055 329,333 1,004,219 14.1 .2240 60,141 13,469 2.4280 268,002 674,886 11,2 .3022 46,672 14,106 2.4879 198,265 406,884 8.7 4511 32,566 14,689 2.5747 125,010 208,619 6.4 1.0000 17.877 17,877 vam 83,609 83,609 4.7 “ U.S. GOVERNMENT PRINTING OFFICE : 1964 0—718-396 SERIES 1-4. OUTLINE OF REPORT SERIES FOR VITAL AND HEALTH STATISTICS Public Health Service Publication No. 1000 GENERAL SERIES. Program descriptions, methodological research, and analytical studies of vital and health statistics. Earlier reports of this kind have appeared in “Vital Statistics—Special Reports” and in “Health Statistics from the National Health Survey,” Series A and D, PHS Publication No. 584. Series 1: Programs and collection procedures.— Reports which describe the general programs of the National Center for Health Statistics and Series 2: Series 3: Series 4: its offices and divisions, data collection methods used, definitions, and other material necessary for understanding of the techni- cal characteristics of published data. Data evaluation and methods research. ~Studies of new statistical methodology including: experimental tests of new survey meth- ods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to statistical theory. Analytical Studies.—~This series comprises reports presenting analytical or interpretive studies based on vital and health statistics. Documents and committee reports.— Final reports of major committees concerned with vital and health statistics and documents such as recommended model vital registration laws and revised birth and death certificates. SERIES 10-12. DATA FROM THE NATIONAL HEALTH SURVEY Earlier reports of the kind appearing in Series 10 have been issued as “Health Statistics from the National Health Survey,” Series B and C, PHS Publication No. 584. Series 10: Series 11: Series 12: Statistics on illness, accidental injuries, disability, use of hospital, medical, dental, and other services, and other health-related topics, based on data collected in the continuing National Health Interview Survey. Data from the Health Examination Survey based on the direct examination, testing, and measurement of national samples of the population of the United States, including the medically defined prevalence of specific diseases, and distributions of the popula- tion with respect to various physical and physiological measurements. Data from the Health Records Survey relating to the health characteristics of persons in institutions, and on hospital, medical nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents of patients, or of records of the establishments. SERIES 20-23. DATA FROM THE NATIONAL VITAL STATISTICS SYSTEM Earlier reports of this kind have been issued in “Vital Statistics—Special Reports.” Series 20: Series 21: Series 22: Series 23: Various reports on mortality, tabulations by cause of death, age, etc., time series of rates, data for geographic areas, States, cities, etc.—other than as included in annual or monthly reports. Data on natality such as birth by age of mother, birth order, geographic areas, States, cities, time series of rates, etc.—compila- tions of data not included in the regular annual volumes or monthly reports. Data on marriage and divorce by various demographic factors, geographic areas, etc.—other than that included in annual or monthly reports. Data from the program of sample surveys related to vital records. The subjects being covered in these surveys are varied includ- ing topics such as mortality by socioeconomic classes, hospitalization in the last year of life, X-ray exposure during pregnancy, etc. Catalog Card U.S. National Center for Health Statistics. Comparison of two methods of constructing abridge life tables by reference to a ‘standard’ table; comparison of the revised and the prior method of constructing the abridged life tables for the United States. Washington, U.S. Department of Health Ed- ucation, and Welfare. Public Health Service, 1964. 11p. tables. 27cm. (Its Vital and Health Statistics, Series 2, no. 4) U.S. Public Health Service. Publication no. 1000, Series 2, no. 4. 1. U.S. - Statistics, Vital - Methodology. I. Title. Cataloged by Department of Health, Education, and Welfare Library PUBLIC HEALTH SERVICE PUBLICATION NO. 1000-SERIES 2-NO. 4 U.C. BERKELEY LIBRARIES C021205942