TREASURY DEPARTMENT UNITED STATES PUBLIC HEALTH SERVICE HUGH S. GUMMING, Surgeon General DIFFICULTIES IN COMPUTING CIVIL DEATH RATES FOR 1918 WITH ESPECIAL REFERENCE TO EPIDEMIC INFLUENZA BY EDGAR SYDENSTRICKER Statistician AND MARY L. KING Statistical Clerk United States Public Health Service REPRINT No. 583 FROM THE PUBLIC HEALTH REPORTS February 13, 1920 (Pages 330-345) 2^-'^^1'^3 WASHINGTON GOVERNMENT PRINTING OFFICE 1920 zL Monograph ^^r. 'o^ iMOV id DIFFICULTIES IN COMPUTING CIVIL DEATH RATES FOR 1918, WITH ESPECIAL REFERENCE TO EPIDEMIC INFLUENZA.^ By Edgar Sydenstkicker, Statisticiaa, and Mary L. King, Statistical Clerk, United States Public Health Service. Abnormal Conditions Affecting the Distribution of Population in 1918. Various conditions are known to have existed in 1918 which caused an abnormal distribution of population in the various demographic groups commonly employed in vital statistics. Mention has been made already of them, such as (1) the withdrawal of over 4,000,000 males of certain ages from civil life for the armed forces in the War with Germany, a condition which affected some localities more than others and which varied in its influence upon the sex and age com- position of the population at different times during the period begin- ning April, 1917; (2) the movement of population, particularly of persons of industrial ages, to localities and areas where V'ar indus- tries were concentrated; (3) changes in the occupational status of the population, which were particularly important from the points of view of sex and age, since the number of women and the number of m^en above the usual wage-earning age were increased m some occu- pations; (4) a well-defined movement of negroes of certain ages from southern into northern and eastern States. It is obvious that, in expressing mortality rates and other ratios for the period in question, some account should be taken of these factors. Their effects are too important to be ignored or to be dismissed with the comforting guess that the more or less conflicting changes had "compensating" effects. 'Unfortunately, their maxi- 1 From the Statistical Office, United States Public Health Service. Acknowledgments are made to Dr. WiUiam H. Davis, chief statistician for vital statistics, Bureau of the Census, for the use of mortality data for Indiana and for computations of rates by certain methods to v^'hich specific reference is made in this paper. Bepriut from the Public Health Reports, vol. 35, No. 7, Feb. 13, 1920, pp. 330-345. 3 4 COMPUTII^G CIVIL DEATH RATES FOR 1918. mum effect came in the latter part of 1918, when, because of the pandemic of influenza, especial interest was attached to the work of the vital statistician and the epidemiologist. A review of the literature so far published on 1918 vital statistics and on the statistics of the influenza epidemic will show that little account has been taken of the abnormal factors affecting the distri- bution of population, particularly from the points of view of sex and age. In a number of instances their possible effects upon mortality rates have been noted, but rarely has there been any attempt to express these eft'ects statistically. Two very good reasons of a prac- tical nature can be given for this omission. One is the lack of reliable data upon wiiich to base corrections of the usual intercensal estimates. Local estimates of additions to or subtractions from the number of persons of different sex, age, occupation, etc., in the population of a given community or section are so crude as to be almost worthless in the great majority of instances.^ The other reason is that a decennial census will be taken as of January 1, 1920, and the natural disposition is to await the results of tliis enumeration before attempting to correct the preliminary computations based on estimated populations. Un- doubtedly the 1920 census will afford a more accurate basis for 1918 rates than estimates based on the 1910 census, but it is important to note that the 1920 enumeration comes too late to give a true picture of the abnormal situation in 1918. Already many readjustments have taken place since 1918. Practically all of the males called to the colors in the War with Germany have returned to civil life. With the cessation of munitions making and with the reduction in the manufacture of products which were in unusual demand during the war, a movement of population away from many centers of "war industries" has set in. Even in those localities where the stimulus of war industries has continued there have doubtless occurred marked changes in the sex, age, and occupation distribution of persons em- ployed because of displacements that have followed the replacements occasioned by the war. So that even with the results of the 1920 census before him, the vital statistician will be put to it to obtain reasonably accurate population bases for computing his ratios for 1918. Certain modifications of population figures as enumerated for 1920 will be necessary in estimating populations for 1918, and he will need all of the data that he can collect relating to purely local popu- lation changes and to males withdrawn from civil life. In view of the fact that it will be some time before the results of the 1920 census are known in any considerable detail, and in view of the practical necessity for as correct population bases as it is possible to get for use in computing preliminary rates, some consideration has I At the same time more use can be made than has been made of statistics of employment, for example, for determining changes in population in communities. COMPUTING CIVIL, DEATH RATES FOR 1918. 5 been given in this ofEce to the question of making tentative cor- rections of population estimates, especially for sex and age, in com- puting influenza and pneumonia mortality rates for 1918. In the following pages an attempt is made to utilize such data as are avail- able relating to the withdrawal of males of certain ages from the civil population in 1918. In order to test the probable accuracy of rates computed upon the basis of population estimates corrected, or modified, from this point of view, a method of estimating population by the use of "normal" death rates from certain causes has been applied, and the results have been compared with the mortality curve for males according to age as computed from data collected among specially canvassed "sample" populations. Correction of Population Estimates by Taking into Account the Withdrawal of Males from Civil Life in 1918. The most complete information so far published relating to the number of males withdrawn from civil life during 1917 and 1918 is furnished in the Second Keport of the Provost Marshal General.^ From these data several tables have been compiled in the belief that they may be of value to officers of health departments and others interested in vital statistics, and are presented herewith. Table I. — Number of males withdrawn from civil life in the United States from April 1, 1917, to January 1, 1919, by months, and the j^ercentages they were of total pop- ulation, of males of all ages, ana of males of ages 18-45, inclusive. Months (1917-18). Males withdrawn from civillife — Ages 18-45. Number. By months. Cumulative. Cumulative per cent of— Total popula- tion. Males. AU ages. 18-45. Total 1917— April May June July August September October . . . November. December. 1918 — January . . . February. . March April May June July August September, October . . . November. 4,178,172 4, 178, 172 100. 00 100.00 113,633 146,868 150,249 85,838 66, 172 324,248 210,392 90,395 194, 700 93,522 121, 693 169, 791 220,079 428, 466 431, 582 452,417 346,924 273,080 249, 185 8,938 113,633 260, 501 410,750 496, 588 562, 760 887,008 1,097,400 1,187,795 1,382,495 1,476,017 1,597,710 1,767,501 1,987,580 2,416,046 2,847,628 3,300,045 3,646,9e9 3,920,049 4,169,234 4,178,172 .11 .25 .40 .48 .54 .85 1.05 1.14 1.32 1.41 1.53 1.69 1.89 2.30 2.71 3.13 3.46 3.71 3.94 3.94 .21 .49 ,77 .93 1.05 1.66 2.05 2.22 2.58 2.75 2.97 3.28 3.68 4.47 5.26 6.10 6.72 7.22 7.67 7.67 100.00 .43 1.10 1.73 2.09 2.36 3.72 4.60 4.97 5.78 6.16 6.66 7.36 8.26 10.03 11.80 13.66 15.08 16.18 17.19 17.21 1 Crowder, E. H., Second Report of the Provost Marshal General on the Operations of the Selective Service System to Dec. 20, 1918, Government Printing Office, Washington, D. C.,1919. This report covers the period from May 18, 1917, the date of the selective service act, to Dec. 20, 1918, and contains also in- formation relating to voluntarv enlistoDP-ut^ ip tb*) Army, Navy, and Marine Corps for the same period. 6 COMPUTING CIVIL DEATH RATES FOR 1918. Table I-A. — Estimated number of persons of specified sex and ages in total population of the United States. Age and sex. July 1, 1917. July 1, 1918. All ages: 103,635,306 53,268,547 23,757,772 105,253,300 All ages: 54,100,19(1 Males: 24,128,687 In Table I are shown the monthly increments and total increase of the armed forces of the United States (inductions through the selec- tive service, and enUstments in the Army, Navy, and Marines) during the period April, 1917, to November, 1918, exclusive of the inductions in October and November, 1918, which were on account of the third registration. The last inductions mentioned were nearly all of stu- dents in schools and colleges, made just before the armistice was signed, and, for practical purposes, the men so inducted need not be considered as withdrawn from civil life. Taking the male population of the age group 18 to 45, inclusive, in the United States, estimated as of July 1, 1917, and July 1, 191S, as the bases, the cumulative per cent withdrawn from civil life at the end of each month from April 1, 1917, to November 30, 1918, has been computed. In Figure I the monthly increments as well as the cumulative totals and the percent- ages referred to have been plotted. It will be noted that at the time of the influenza epidemic in October and November, 1918, the civil population had been decreased by something over 4,000,000 persons. This was equivalent to nearly 4 per cent of the entire pop- ulation, nearly 8 per cent of the entire male population, and about 17 per cent of males in the ages 18 to 45, as estimated for July 1, 1918. While the figures are not absolutely accurate, particularly in that no account can be taken of the discharge of soldiers from camps or of commissioned officers,' they are sufficiently correct to show that the number of males of these ages in civil life was so considerably de- creased that any computation of specific mortalit}^ rates, for example, according to age, based on the estimated population without allowing for withdrawals on account of military duty, would be seriously erroneous. Since neither the total population nor the number of males of all ages nor the number of males of specific ages in continental United States is ordinarily used in computing mortality rates. Table I does not afford any data for practical use except the cumulative percentages by months. Similar data for States or smaller divisions are not afforded, 1 It appears that 8. 1 per cent of men actually inducted into service during the period Feb. 10-Nov. 1, 1918, were later rejected on physical examinations after reaching camp (Second Report of the Provost Marshal General, p. 420, Table 56-A). How large a percentage of enlisted men was rejected for this reason is not stated, so far as the writers are ableto ascertain. The figures giveuin Table I are therefore somewhat high. On the other hand, about 230,000 commissioned officers are not included in Table I. COMPUTING CIVIL DEATH RATES FOR 1918. but the monthly cumulative percentages for the country as a whole can be applied to the State totals (as given in Table II) or to totals for localities that can be determined from State enhstment figures and from the returns from local exemption boards which are published in the report referred to. Upon the assumption that the population NUMBER OF MEN WITHDRAWN FROM CIVIL LIFE AND PER CENT THEY WERE OF MALES OF THE AGES 18-45 IN THE TOTAL POPULATION BY MONTHS, 1917-1918 Number of RrCent Men in Armed Torca of Moles 18 -tS Apoofioo - By Months / 5.500P00 - CunUlATIVE — ^ / -19 3.000^600 - / 2.500,000 - / -10 Zpoopod - / yoopoo - /^ V>oo,oo» - /^ - 5 300,000 - - ^yir\ ^ „-''"'~'"--x AMJJASOND 1917 JFMAMJJASON 1918 Figure 1. in each locality or larger division was affected in about the same way as the population of the country at large, a rough approximation can be made of the number of withdrawals of males from civil hfe at the end of any month in the locality or section in question. This approximation, of course, can be used in connection with the percent- age distribution of males actually withdrawn (see Tables III and IV) 8 COMPUTING CIVIL DEATH RATES FOE 1918. and some crude estimate can be made of the population in diflfereRt age groups remaining in civil life at the end of any month in the period from April 1, 1917, to November 30, 1918. In Table II are given the total increments' to November, 1918, from various sources in each State. The figures for white and colored combined are available for men inducted but not for men enlisted. Since only 1.5 per cent of enlisted men were colored, however, the number of colored males for practical purposes is neghgible. Table II. — Enlistments and inductions, Apr. £, 1917, to Oct. SI, 1918, under first and second registrations, compared by States.^ Locality. Total incre- ment. United States 4,034,743 Alabama Arizona Arkansas California Colorado Connecticut Delaware r>istrict of Columbia Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky I/ouisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana ■ Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming .Maska Hawaii Porto Rico Not located. 73, 11, 65, 131, 38, 55, 7, 17, 36, 86, 20, 272, 104, 101, 66, 77, 71, 26, 51, 157, 142, 106, 56, 140, 39, 49, 5, 14, 118, 13, 410, 74, 27, 205, 84, 34, 313, 22, 54, 30, 80, 174, 19, 11, 78, 55, 55, 101, 12, 2, S, 16, 543 410 311 484 751 218 985 945 211 973 467 235 973 638 645 983 271 602 700 101 397 918 740 257 049 614 488 970 350 586 569 705 253 852 909 430 297 270 284 130 139 061 421 223 524 433 895 696 223 105 733 490 Total in- ducted. 2,666,867 1,286 59, 755 8,113 49,312 67, 067 22,858 32,539 4,993 9,631 24,916 66,841 12,566 177,483 69, 749 66,864 41,905 58,330 66, 205 15, 266 33, 867 76,567 96, 480 73, 680 43,362 92, 843 27,340 29, 807 3,164 8,404 71,390 8,862 253, 589 5S, 441 18, 595 138, 148 64,941 16, 158 201, 211 11,176 44,059 21,255 59, 878 117,395 10, 788 6, 629 58,337 28, 686 45, 355 70,982 7,923 1,962 5,466 15, 734 Total en- listed. 1,367,876 3,667,033 13, 788 3,297 15,999 64, 417 15,893 22,679 2,992 8,314 11,295 20, 132 7,901 94, 752 35. 224 34, 774 24, 740 19, 653 15,066 11,336 17,833 80,534 45, 917 33,238 13,378 47, 414 11, 709 19, 807 2,324 6,566 46, 960 4,724 156,980 16, 264 8,058 67, 704 19,968 18, 272 112,086 11,094 10. 225 8,875 20, 261 56, 666 8,633 4,594 20,187 26, 747 10, 540 30, 714 4,300 143 267 756 White. Total. 2,299,157 47. 669 11,333 47, 767 130, 565 38, 380 54, 277 6,620 13,945 23,307 52. 670 20,372 263, 481 100,394 100, 709 64,518 66, 663 42,560 26,552 42, 488 155,901 140,002 106, 407 32, 674 131,038 38, 851 48,972 5,462 14, 943 113, 487 13, 535 404,376 54,623 27, 166 197, 991 79, 215 34,362 297,905 21,979 28, 486 30, 068 62, 365 142, 555 19,344 11,201 54, 983 55, 260 50,403 101,472 12, 128 2,100 5, 733 16, 490 In- ducted. 877, 458 1, 286 1, : 33,881 8,036 31, 768 66, 148 22, 487 31, 598 3.628 5,631 12,012 32,538 12,471 168, 729 65, 170 65,935 39, 778 47,010 27,494 15, 216 24, 655 75,367 94, 085 73, 169 19, 296 83,624 27. 142 29,165 3,138 8,377 66,527 8,811 247,396 38,359 18, 508 130, 287 59, 247 16, 090 185,819 10, 885 18, 261 21,193 42, 104 85, 889 10, 711 6,607 34, 796 28, 513 39,863 70, 758 7,828 1,957 5,466 15, 734 Enlisted. Army. Navy. Marines. 437,527 9,562 1,854 11, 699 38,992 9,670 13,151 2,003 4,442 6,834 14, 160 4,955 61,938 25, 847 26,389 18, 217 13, 934 7, 570 7,290 10,144 41,985 32, 403 20, 272 9,044 29,863 7,331 14,416 1,888 4,408 28, 333 3,649 89, 031 10, 573 6,611 48,885 14, 105 10, 626 78, 671 5,436 6,505 7,083 13, 563 37, 704 5,335 3,088 10, 556 12, 761 7,359 22,349 3,554 143 267 756 3,938 1,269 4,025 23,058 5,075 9,319 919 3,500 4,375 6,382 2,450 28,264 8,313 7,832 5,907 5,163 6,782 4,025 6,913 36, 884 11, 463 10, 588 4,069 14,132 3,281 4,944 350 2,100 17,457 1,050 61, 779 5,250 1,838 14, 176 5,513 6,694 29,446 5,600 3,675 1,663 5,425 16,889 2,494 1,488 9,144 12,382 2, 625 7,569 656 254 394 52,891 288 174 275 2,367 1,148 209 70 372 86 590 496 4,550 1,064 553 616 556 714 21 776 1,665 2,051 2,378 265 3,419 1,097 447 86 58 1,170 25 6,170 441 209 4,643 350 952 3,969 58 45 129 1,273 2.073 804 18 487 1,604 ■ 556 796 90 638 Colored in- ducted. 367, 710 25,874 77 17, 544 919 371 941 1,365 4,000 12,904 34,303 95 8,754 4,579. 929 2,127 11,320 28,711 50 9,212 1,200 2,395 .511 24,066 9,219 198 642 26 27 4,863 51 6,193 20,082 87 7,861 5,694 68 15,392 291 25,798 62 17, 774 31, 506 77 22 23,641 173 5,492 224 95 5 1 Compiled from the second report of the Provost Marshal General, pp. 468 and 459. COMPUTING CIVIL DEATH RATES FOR 1918. 9 The report of the Provost Marshal General already referred to shows for each local exemption board the number of men accepted at camp, but it does not give the number of enlisted men from each locality. Here again rough approximation must be resorted to. The number of men withdrawn from a given locality by enlistments can be estimated by using the ratio of inductions to enlistments for the particular State and applying it to the inductions from the locality. This, however, will afford rather doubtful results except possibly for the large population centers. Since the males withdrawn from the civil population were almost entirely within the ages 18 to 45, and principally within the ages 21 to 31, corrections of population estimates for males of specific ages or age groups are especially important. Unfortunately, no tabula- tion by years of age of the number of men inducted into or enlisted in the armed forces has as yet been published.^ About the closest approximations that can be made are from the percentage distribu- tion of registered men actually placed in Class I for the country as a whole, ^ and from Army and Navy estimates (based on samphng) of the number of enlisted men who were of the ages under 21 and over 30.^ Since no statistics are available as to the distribution of enlisted men in the age group 21 to 30, we must assume that the dis- tribution was the same as for the selective service men who were actually placed in Class I. Without presenting the detailed compu- tations, the approximations arrived at are presented in Table III, which shows the number of men of each age and the percentage they were of totals inducted and enlisted in the various services in the United States as a whole. For reasons that are apparent, the figures thus obtained are approximations only, but it is believed that they afford reasonably accurate percentages of distribution which can be utilized in making estimates for States or smaller geograpliic divisions. Since 5-year age groups ordinarily are utilized in com- puting rates, the percentage distribution given in Table III is sum- marized by 5-year age groups in Table IV. ' The Provost Marshal General's second report gives in detail the number of men by years of age who were re^isifrerf for selective service in each State, but not the number of men by years of age who were actu. ally inducted or who had enlisted. A tabulation is presented showing, for men inducted in 1917, the propor- tions of registered males for each year of age from 21 to 30, inclusive, who were placed in Class I for the country as a whole (Id. p. 189, Table 67). The percentagea varied from 46.3 for those 21 years of age to 22.1 for those 30 years of age. If it can be assumed that these ratios held for any State (and probably they are suffi- ciently accurate for the purpose), the number of men placed in Class I can be estimated for each year of age. It is not beUeved, however, that this method would yield more accurate estimates than the one which is immediately suggested. 2 Id. p. 189, Table 67. 8 Id. p. 317. 187584°— 20 2 10 COMPUTING CIVIL DEATH RATES FOR 1918. Table III. — Numher of males withdrawn from civil life for service in the armed forces, and estimated number in each age, during 1917 and 1918, in the United States. Enlistments. Age. Total. Army. Navy. Marines. Total enlistments. Inductions. Num- ber. Per cent. Num- ber. Per cent. Num- ber. Per cent. Num- ber. Per cent. Number. Per cent. Num- ber. Per cent. All ages. 18 4,034,743 62, 210 352, 526 819,577 403, 718 374,888 335,807 294,061 268, 646 247,398 213,981 209, 067 219,317 203, 547 100.0 1.5 8.8 20.3 10.0 9.3 8.3 7.3 6.7 6.1 6.1 5.2 5.4 5.0 877,458 35, 537 201,377 73, 522 66,520 61,519 55,017 48, 515 44,013 40,512 40,012 34,510 36,011 140,393 100.0 4.0 23.0 8.4 7.6 7.0 6.3 5.5 5.0 4.6 4.6 3.9 4.1 16.0 437,527 24,611 139, 461 32, 158 29,096 26,908 24,064 21,220 19,251 17,720 17,501 15,095 15,751 54,691 100.0 5.6 31.9 7.4 6.6 6.2 5.5 4.8 4.4 4.0 4.0 3.4 3.7 12.5 52,891 2,062 11,688 4,510 4,080 3,773 3,374 2,976 2,700 2,485 2,454 2,117 2,209 8,463 100.0 3.9 22.1 8.5 7.7 7.2 6.4 5.6 5.1 4.7 4.6 4.0 4.2 16.0 1,367,876 62,210 352,526 110,190 99, 696 92, 200 82,455 72,711 65,964 60,717 59,967 51,722 53,971 1,203,547 100.0 4.6 25.8 8.1 7.3 6.7 6.0 5.3 4.8 4.4 4.4 3.8 3.9 14.9 2,666,867 100.0 19-20 21 (1) 709,387 304,022 232,688 253,352 221,350 202, 682 186,681 184,014 157,3^5 165,346 26.6 22 11.4 23 10.6 24 9.5 25 8.3 26 7.6 27 7.0 28 6.9 29 5.9 30 6.2 Over 30 1 143,429 were inducted from the third registration, presumably college students enrolled in student Army camps. They are not included here. Table IV. — Approximated percentages of males enlisted in the Army, Navy, and Marines, and inducted during 1917-18, who were in specified age groups. Age groups. Enlisted. Inducted. Army. Navy. Marines. Total. 100.0 15.5 40.8 27.7 16.0 100.0 21.6 41.6 24.3 12.5 100.0 15.0 40.8 28.2 16.0 100.0 17.5 41.0 26.6 14.9 100 15-19 20-24 58.1 25-29 35.7 30 and over 6.2 With these data before us, the following method of utilizing them in correcting population estimates for the latter part of 1918, from the single point of view of withdrawals of males from civil life, seems to be logical and practicable: For a given State, find the number of males inducted (Table II) and multiply this number by the percentages for different ages for inducted men (Table III or Table IV). The resulting figures are the estimated number of males of different ages or age groups with- drawn from civil life by the selective service law. Proceed in the same manner for men enlisted in each of the different services (Army, Navy, and Marines). Summate for each age or age group the num- ber of men withdrawn by inductions and enlistments, and subtract from the estimated male population as of July 1, 1918, in correspond- ing ages or age groups. The remainder will be an estimate of the male population of the ages specified who were in civil life at the time under consideration. COMPUTING CIVIL DEATH BATES FOR 1918. H Population Estimates on the Basis of Normal Death Rate from Selected Causes. For localities or sections where no other causes are known to have affected distribution of population according to age, occupation, etc., in an appreciable degree, a correction of the effect of withdrawals of males for military service will be sufficient. The vital statistician must, of coui-se, determine from such information as he can obtain whether or not any other important causes were at work. It is safe to say, however, that in a number of States and localities abnormal conditions other than the withdrawal of males for mili- tary service did affect the population. In any event, it is highly desirable to have some criterion by which the combined influences of the various possible conditions may be expressed statistically. Sampling by means of actual enumerations was not resorted to, except in a few instances for special purposes. Is there any other method simple enough for every day use? We are indebted to the division of vital statistics of the Bureau of the Census for a method which has been used in checking estimates arrived at by conventional procedure. A death rate is made up of two factors: The number of persons in the group considered and the number of deaths occurring among those persons from the cause or causes in question. The usual formula is ~ Number of deaths X 1,000 Population Rate If the death rate during a normal period from certain causes be used as the divisor and the number of deaths (multiplied by 1,000) from the same causes during the period for which a population estimate is desired be the dividend, the quotient will be the desired estimate of the population. To illustrate: Let the number of deaths from all causes, other than acute infections diseases, among males of the age group' 20 to 24 during the period September-December, 1918, = 300. Let the death rate from the same causes among males of the same ages during a period which is as- sumed to be normal; e. g., the average for September-December for the years 1909-1911 =2 per 1,000. Then ^P°V-°°° = 150,000 which is an estimate of the number of males in September-December, 1918. Of course this estimate involves several assumptions, impor- tant among which is that the death rate of 2 per 1,000 from the causes specified did not change from 1909-1911 to 1918. But grant- 12 COMPUTING CIVIL DEATH RATES FOR 1918. ing that this estimate is fairly accurate, the specific rate for any disease for 1918 may be computed.^ A Comparison of Rates Derived from Different Population Estimates made by the Various Methods Described. It will be of interest now to compare the results derived by the use of the various methods described. The deaths from influenza and pneumonia (all forms) during the four months' period, September 1 to December 31, 1918, among males of different ages in Indiana are used (Table V). Three sets of annual rates are computed; namely, (1) a rate based on the number of males in each age group as estimated for July 1, 1918, by the usual arithmetic method for intercensal years; (2) a rate based on the number of males in each age group as estimated in (1), but after subtracting, for certain age groups, the number of males withdrawn from civil life up to November 30, 1918; (3) a rate based on the normal death rate from all causes except pneumonia (all forms), computed according to either of the two methods used by the vital statistics division of the Bureau of the Census. For convenienc3 we may denote them as rate 1, rate 2, and rate 3. Table V. — Nuviber of deaths from influenza and pneumonia (all forms) and from all other causes among males in Indiana Sept. 1,-Dec. 31, 1918. Age groups. Influenza and pneu- monia (all forms). All other causes. 833 187 166 360 371 611 619 452 214 117 87 61 66 209 1,042 6 to 9 107 10 to 14 103 15 to 19 - 161 20 to 24 143 25 to 29.. . . - - 188 30 to 34 . 199 35to39 227 40 to 44 . . 224 45 to 49 262 60 to 54... 261 55 to 59 286 60 to 64 423 1,880 1 This method involves a good deal of arithmetical labor, and practically the same result is obtained by a shorter procedure employed by the vital statistics division of the Bureau of the Census. This procedure involves exactly the same principle as the one described above, but the actual computation of a new esti- mate of population is eliminated. For example, in checking the accuracy of the calculated death rates from influenza and pneumonia in Indiana during September-December, 1918, rough death rates were found as follows: Graduated data of 1909, 1910, and 1911 were first used to find a set of normal death rates by age and sex as in the construction of a life table. To make these rates more nearly accord with rates based on all deaths except influenza and pneumonia (all forms) to total deaths, the percentages of deaths from pneumonia (all forms) to total deaths by ages and sex in 1910 were calculated and the life-table rates reduced to corresponding amounts. (No allowance was made for influenza deaths in 1910 as this was considered an unnecessary refinement for this rough calculation.) The resulting death rates were assiuned to represent the death rates from all causes except influenza and pneumonia (all forms) as primary causes in the last four months of 1918, and rough death rates were then calculated in the following manner: The percent of total deaths in the last four months of 1918 which were from influenza and pneumonia (all forms) was divided by the per cent which were from all other causes. The quotient was then multiplied by the assumed normal annual death rate from all causes except pneumonia (all forms) based on graduated data of 1909, 1910, and 1911. The product is a rough annual death rate from influenza and pneumonia (all forms) for the last four months of 1918 among persons of a specific sex and age. COMPUTING CIVIL DEATH RATES FOE 1918. 13 In computing rates 1 and 2, the detailed population estimates are presented for purposes of illustration in Table VI. The number of males in each age group as estimated by the usual arithmetic method for intercensal years is shown in the first column. In the next five columns are shown the number of males withdrawn from civil life, approximated according to the procedure previously outlined. In the last colunm is given the estimated number of males in each age group after allowing for these withdrawals. Table VI. — Number of males of different ages in Indiana, estimated as of July 1, 1918, by arithmetic method and after deducting therefrom males withdrawn from civil life in 1917 and 1918. Male popu- lation, estimated as of July 1, 1918. Number of males withdrawn from mal life. Male-popu- lation Age groups. Total. In- ducted. EnUsted. estimates, less males withdrawn • Army. Navy. Marines. from civil Ufa. All known ages 1,459,3.53 148,043 142, 066 136,535 138,586 133,999 123,154 106,605 103,154 90,604 77,681 75,345 55,955 45,797 81,829 104,973 69,749 25, 847 8,313 1,064 1,354,380 148, 043 5 to 9 ( 142, 066 10tol4 136, 535 15tol9 5,963 54,961 34,380 9,669 4,007 10,545 7,160 4,135 1,796 3,458 2,020 1,039 ieo 434 300 170 132,623 79, 038 20to24 40,524 24,900 4,325 25 to 29 89, 774 30to34 96, 936 35 to 39 103,154 40 to 44 90,604 77,681 45 to 49 60 to 54 75,345 55 to 59 55,955 60 to 64 45,797 65 and over 81,829 In Table VII are shown the annual rates computed by the three methods. - These rates are plotted in Figure 2. Table VII. — Annual death rate per 1,000 males of different ages in Indiana from influenza and pneumonia {all forms), September-December, 1918, computed upon the bases of various estimates of population. Age groups. 1 Based on popu- lation as esti- mated by arithmetic method. 2 Same as 1, but after allowing for withdrawal of males from civil life. 3 Based upon normal death rate from all causes other than pneumonia. Under 5 16.8 3.9 3.6 7.8 8.3 14.8 17.3 13.1 7.1 4.5 3.5 3.3 4.3 7.6 16.8 3.9 3.6 8.1 14.0 20.4 20.1 13.1 7.1 4.5 3.5 3.3 4.3 7.6 20.4 5 to 9 4.7 10 to 14 3.7 15 to 19 8.2 20 to 24 13.8 25 to 29 16.8 30 to 34... . 18.4 35 to 39 12.8 40 to 44 7.2 45 to 49 4.3 50 to 54 3.8 55 to 59 3.8 60 to 64 4.0 65 and over i 8.6 14 COMPUTING CIVIL DEATH EATES FOE 1918. It will be observed that the rates are practically identical for all ages except those between 15 and 35, thus indicating the fact that certain conditions existed which disturbed the normal age distribu- tion of the male population of those ages. Which of the three rates DEATH RATES FROM INFLUENZA AND PNEUMONIA AMONG MALES SEPTEMBER-DECEMBER 1918 IN INDIANA: VARIOUS ESTIMATES OF POPULATION 5 10 15 20 25 30 35 40 45 50 55 60 65 AGE IN YEARS FiGxmE 2. is the correct one ? Or, rather, which of the three rates most nearly approximates the correct rate ? Kate 1 is much lower, particularly in the age group 20 to 24, than rates 2 and 3, suggesting the conclusion that the computation of a mortahtyrate upon the basis of an estimated COMPUTING CIVIL DEATH EATES FOR 1918. 15 popiJation without taking into account the withdrawal of males from, civil hf e is quite inaccurate. As between rates 2 and 3 , the suggestion is offered that rate 2 is too high because the withdrawal of males from civil life may have been compensated by an abnormal addition of COMPARISON OF THE RELATIVE VARIATIONS IN DEATH RATES FROM • INFLUENZA AND PNEUMONIA AMONG MALES OF DIFFERENT AGES BASE : ARITHMETIC AVERAGE OF THE SPECIFIC RATES moicEs 1 1 1 1 1 300- 250- A 1. Indiam: Upon the basis cf males estimat as of July 1 , W18 a Indiana: Same as 1 , but allowing for w of maJes from civil life in lln-WlS 3 Indiaia: Upon the basis of^ maies as est from deaths from 'normaJ'or 'constant causes ed ithdrawal nated ^iaJlu r^ — -4 enumerited populations Cirfluenza surveys of U.S.P.H.S.) — ,<> — — . 200- 1 > \ \ to ^ ^\ ISO. ,1 11. / \ \ Vv 1 \\ \ 1 1 1 ^/ 1 loo- se 1 J f V. A J i 1 ^ V^ C-l V" 5 1 IS 20 25 30 35 40 45 50 55 80 65 AGE IN YEARS FiGUEE 3, males in certain occupations. In weighing the relative accuracy of rates 2 and 3, however, we must take into consideration other possible conditions, as, for example, the demand in Indiana for males of these ages for employment in the so-called war industries. As a matter of 16 COMPUTING CIVIL DEATH RATES FOE 1918. fact, Indiana was a State wliere comparatively few war industries were located and there was a tendency toward emigration rather than toward an immigration. In order to test this assumption the various curves already com- puted may be compared with the mortality curve among males of a canvassed population. Accordingly, in Figure 3 the rates have been reduced to a relative basis, and a fourth curve — that of mortahty from influenza and pneumonia among males of different ages who were actually canvassed in the course of special influenza surveys made in various localities in the United States by the Public Health Service — has been fitted in. This fourth curve, which we may de- note as rate 4, is considered as the normal. If it be true that rate 4 can properly be regarded as a normal one for the period of the influenza epidemic, it is clearly indicated that (Table VIII and Fig. 3) the rate which most closely approximates it for Indiana is rate 2.^ Table VIII. — Comparison of the relative variations in death rates from influenza and pneumonia (all forms), computed by various methods for males of different ages in Indiana, September-December, 1918, with that in areas where special surveys were made. [Base: Arithmetic average of the specific rates.] Age groups. 1 Population estimated by usual method. 2 Same as 1, but after allowing for males with- drawn from civillife. 3 By census method. 4 Localities canvassed. Under 5 202 47 43 94 100 178 208 158 86 54 42 40 52 92 202 47 43 98 169 246 242 158 86 54 42 40 52 92 219 61 40 88 148 181 198 138 77 46 41 41 43 92 149 5to9 37 10 to 14.... ZJ 15 to 19 78 20 to 24....! 133 2Sto29 261 30 to 34 178 35 to 39 155 40to44 80 45 to 49 69 SO to 54 65 55 to 59 37 60to64 ^3 65 and over 82 A similar result has been found in comparing rates computed upon the various bases for other sections and localities. In this connection it may be noted that if the ratio be computed for each age group of the number of males and of females enumerated in the special influenza survey made in Baltimore, Md., during December, 1918, to that enumerated in the 1910 census of the same 1 The divergences of the rates in the ages 40 and over are probably due to a paucity of data, since the number of deaths and the number of persons in those ages in the special surveys were rather small. COMPUTING CIVIL DEATH RATES FOR 1918. 17 city, it mil be seen thcat a very marked depletion of males of the ages 18 to 35 occurred. This has been done in Table IX and plotted in Figure 4. While the number of persons included in the survey was relatively small (about 32,000), and the sample is not large enough to afford as great regularity in results as would be desirable, the RATIO OF NUMBER OF PERSONS OF DIFFERENT AGES ENUMERATED IN SAMPLE AREAS IN BAUIMORE IN 1918 TO THAT ENUMERATED IN THE ENTIRE CITY IN 1910 Figure 4. great divergence of the distribution of males in the ages 18 to 34 from that of males and females in the same ages in 1910 in Decem- ber, 1918, is unmistakably manifest. Here, therefore, in a center of industrial activity the depletion of the male population of military age far outweighed accretions. 18 COMPTJTIITG CIVIL DEATH RATES EOR 1918. Table IX. — Ratio of number of persons of different ages enumerated in sample areas in Baltimore in 1918 to that enumerated in the entire city in 1910. Population. Ratio, 1918 to 1910. Indices of ratios. Age period. Beth sexes. Male. Female. Both sexes. Male. Fe- male. Both sexes. Males. Fe- 1910 1918 1910 1918 1910 1918 males. 557, 790 31,697 267,897 14,677 289,893 17,020 0. 0568 0. 0548 0.0587 100 Under 5 5 to 9 10tol4 15 to 19 20 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 55 to 59 60 to 64 65t-i69 70tT74 75 and over . 51,986 49, 617 48,507 54,253 58,713 54,311 45, 604 43, 163 36,963 31,627 26,510 18, 607 14,351 10,454 6,647 6,477 3,226 2,988 2,826 2,815 2,749 2,826 2,758 2,445 2,057 1,880 1,631 1,071 936 663 417 409 26,189 24,739 23,564 25, 124 27, 373 26, 127 22, 195 21,029 17,852 15, 367 12,856 8,987 6,541 4,662 2,800 2,492 1,641 1,500 1,375 1,304 959 1,138 1,260 1,207 1,000 912 784 499 452 311 170 165 25,797 24, 878 24,943 29, 129 31,340 28,184 23,409 22, 134 19,111 16,260 13, 654 9,620 7,810 5,792 3,847 3,985 1,585 1,488 1,451 1,511 1,790 1,688 1,498 1,238 1,057 968 847 672 484 352 247 244 .0621 .0602 .0583 .0519 .0468 .0520 .0605 .0566 .0557 .0594 .0615 .0576 .0652 ■ . 0634 . 0627 . 0631 .0627 .0606 .0584 ,0519 .0350 .0436 .0568 .0574 .0560 .0593 .0610 .0555 .0691 . 0667 .0607 .0662 .0614 .0598 .0582 .0519 .0571 . 0599 .0640 .0559 .0553 .0595 .0620 .0595 .0620 .0608 .0642 .0612 109 106 103 91 82 92 107 100 98 105 108 101 115 112 110 111 Ill 107 103 91 62 77 100 101 99 104 107 98 122 117 107 117 108 105 102 91 101 105 113 98 97 105 109 105 109 107 113 108 Unless in a given locality there are known to have been considerable accretions to the male population of the ages under consideration, which offset the depletion due to withdrawals for military service, rates based on the estimated number of total males or females in each age group, after allowing for withdrawals for military service, will be more accurate than rates which do not take into account these with- drawals. Probably they will be more nearly correct than rates com- puted on the basis of "normal" death rates. At best, such corrections as can be made of population estimates for 1918 are very rough approximations. The foregoing may prove suggestive to health departments as rather simple methods for making corrections for the withdrawal of males of certain ages from civil life, and for testing rates based upon ''corrected" and "uncorrected" population estimates. ADDITIONAL COPIES OF THIS PUBLICATION MAY BE PEOCTJRED FROM THE SUPERINTENDENT OF DOCUMENTS GOVERNMENT PRINTING OFFICE "WASHINGTON, D. C. AT 5 CENTS PER COPY II '' i