U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE
Public Health Service
Influenza in the United States, 1887-1956
REVIEW AND STUDY OF ILLNESS AND MEDICAL CARE
WITH SPECIAL REFERENCE TO LONG-TIME TRENDS
Public Health Monograph No. 48, 1957 (Public Health Service Publication No. 544)
SELWYN D. COLLINS, Ph.D.
Trend and Age Variation of Mortality and Morbidity from Influenza and PneumoniaInfluenza and its frequent complication, pneumonia, and the excessively high death rates all over the world during the great influenza pandemic of 1918-19 have given rise to much clinical, laboratory, and statistical research. Rather frequent epidemics of moderate severity occurred in the last quarter of the 19th and the first half of the 20th centuries, but nothing within modern times has approached the 1918-19 pandemic, with its estimated 550,000 deaths in excess of the usual expectancy in the United States in those years and a total of 21,642,000 deaths credited to influenza and pneumonia throughout the world in the three waves of that epidemic.
In Jordan's Epidemic Influenza estimates of deaths from influenza-pneumonia are given by country and summarized by continent (pp. 228-229) as follows: North America, 1,075,685; South America, 327,250; Europe, 2,163,303; Asia, 15,757, 363; Australia and Oceania, 965,254; Africa and Madagascar, 1,353,428; total, 21,642,283.
E. Symes Thompson, in a study of influenza epidemics in Great Britain from 1510 to 1890, traces the major epidemics that occurred during that 380-year period and discusses some 20 of them. Included in this list is an epidemic in 1173. These epidemics all appear to be due to influenza and have been so described in earlier writings, but they differ markedly. Many originated on the Continent and others spread from England to Europe and Asia. In studies of influenza, the statistical measure most commonly used of the extent of an epidemic is the death rate from influenza and pneumonia in excess of the usual seasonal expectancy,
with account taken of trend also. This expectancy is measured in different ways but always to represent a base expectancy, usually in weekly rates (annual basis) exclusive of epidemic increases. Such a norm or average for the different weeks of a year may be based on a median of rates for each of the 52 weeks over a reasonable period of years, or for a long series of years if there is no trend in the rates or if correction is made for trends not associated with epidemics.
Figure 46. Excess annual death rates from influenza and pneumonia in the population of various States and groups of cities in the United States, 1887-1956.
TrendInfluenza epidemics spread so rapidly and last such a short time in a given community that weekly death rates are essential to give a realistic picture of the true excess mortality that occurs during a given epidemic, the true duration of the epidemic, and other characteristics of the disease. However, in the United States, deaths from influenza and pneumonia were not generally tabulated by weeks prior to September 1918, when the great pandemic began; therefore, excess deaths in epidemics before that time can be measured by monthly data only.
The earliest records now available for any State in the United States are deaths by months from influenza and pneumonia in Massachusetts. These monthly excess death rates (annual basis) are shown in the two top sections of figure 46 for the years 1887 to 1910, inclusive, during which time 12 small and moderate-sized outbreaks occurred; the larger ones include those with peaks in January 1890, December-January 1891-92, January 1899, and March 1900.
Figure 46 also shows influenza and pneumonia excess death rates, by months, in a group of 35 large cities in the United States from January
1910 through August 1918, during which period 3 epidemics occurred with peaks in January 1916, January 1917, and April 1918. Beginning with September 1918, data for the same group of 35 cities are shown by weeks to the end of 1921. Starting with January 1922 (fig. 46), the record is continued by weeks for a group of 90 representative cities to the end of 1942, after which time the data are shown by weeks for a group of 56 large cities to the end of 1950, at which time 6 cities were added to make 62 cities carried until early in 1956.
Figure 46 shows the high fall peak of the great pandemic, which occurred in October 1918 and was followed by two smaller secondary peaks in December 1918 and January 1919 respectively. These latter peaks are higher than any peak of influenza mortality that has occurred since that time, except the peak of 1920.
During the 70 years represented in figure 46, 1887 to 1956, inclusive, there were 41 epidemics. These ranged in size from three which showed excess deaths in only a few geographic sections, with no measurable excess for the total of all sections of the United States, to the pandemic of 1918-19, which was excessively severe in all sections but less severe in some than in others. Table 7 summarizes these epidemics with respect to total actual excess death rates from influenza and pneumonia, date of middle of the peak week, and number and dates of the weeks that were found to be above the normal expectancy of influenza-pneumonia death rates.
Figure 46 for Massachusetts and the several groups of cities in the United States shows only the excess deaths credited solely or primarily to influenza and pneumonia. But, contrary to first expectations, it was found that peaks of excess influenza-pneumonia mortality were accompanied by small peaks of deaths credited to causes other than influenza and pneumonia. The first task was to find the extent of these peaks of deaths not credited to influenza or pneumonia which, nevertheless,
occurred at approximately the same time as the deaths in the influenza epidemics. This process consisted of treating the weekly death rates from all causes for the same groups of cities in approximately the same way as those from influenza and pneumonia, to obtain death rates from all causes in excess of normal expectancy.
|Date of epidemic and sections with highest and lowest total excess||Actual
peak week or
month of peak
|Total period considered as above normal|
|Months or weeks included|
|Early 1890||34.8||January 1890||1||January 1890|
|Spring 1891||23.9||May 1891||3||April-June 1891|
|1891-92||98.5||January 1892||3||December 1891-February 1892|
|Spring 1893||35.5||April 1893||4||March-June 1893|
|1893-94||33.0||December 1893||2||December 1893-January 1894|
|Early 1895||31.0||February 1895||3||February-April 1895|
|Spring 1897||15.4||March 1897||2||February-March 1897|
|1898-99||34.7||January 1899||3||December 1898-February 1899|
|Spring 1900||47.8||March 1900||3||March-May 1900|
|Early 1901||17.1||February 1901||3||January-March 1901|
|Early 1905||13.8||February 1905||3||January-March 1905|
|1907-08||19.7||December 1907||3||December 1907-February 1908|
|Spring 1915||12.6||April 1915||2||March-April 1915|
|35 large cities:|
|1915-16||22.8||January 1916||2||December 1915-January 1916|
|Early 1917||14.0||January 1917||2||January-February 1917|
|Spring 1918||21.1||April 1918||4||January-April 1918|
|1918-1919||550.5||Oct. 16, 1918||31||Sept. 15, 1918-Apr. 19, 1919|
|90 representative cities:|
|Early 1920||99.3||Feb. 11, 1920||12||Jan. 4-Mar. 27, 1920|
|High: Mountain||159.5||Feb. 11, 1920||14||Jan. 4-Apr. 10, 1920|
|Low: Pacific||57.7||Feb. 11, 1920||11||Jan. 18-Apr. 3, 1920|
|Early 1922||18.3||Feb. 22, 1922||12||Jan. 8-Apr. 1, 1922|
|High: Pacific||36.3||Feb. 22, 1922||9||Jan. 29-Apr. 1, 1922|
|Low: South Atlantic||9.4||Mar. 29, 1922||8||Feb. 19-Apr. 15, 1922|
|1922-23||29.9||Feb. 21, 1923||17||Nov. 26, 1922-Mar. 24, 1923|
|High: West North Central||53.3||Feb. 28, 1923||18||Nov. 26, 1922-Mar. 31, 1923|
|Low: West South Central||6.7||Feb. 28, 1923||10||Jan. 14-Mar. 24, 1923|
|1925-26||25.3||Mar. 24, 1926||17||Jan. 31-May 29, 1926|
|High: West South Central||58.8||Feb. 10, 1926||20||Nov. 8, 1925-Mar. 27, 1926|
|Low: Pacific||9.3||Jan. 6, 1926||9||Dec. 27, 1925-Feb. 27, 1926|
|Spring 1928||11.6||May 9, 1928||19||Mar. 11-July 21, 1928|
|High: Middle Atlantic||20.9||May 9, 1928||19||Mar. 11-July 21, 1928|
|Low: West North Central||4.9||May 2, 1928||10||Apr. 8-June 16, 1928|
|Winter 1928-29||44.4||Jan. 9, 1929||12||Nov. 25, 1928-Feb. 16, 1929|
|High: East South Central||92.0||Jan. 9, 1929||9||Dec. 2, 1928-Feb. 2, 1929|
|Low: New England||42.3||Jan. 23, 1929||9||Dec. 30, 1928-Mar. 2, 1929|
|Winter 1930-31||16.4||Jan. 28, 1931||16||Dec. 28, 1930-Apr. 18, 1931|
|High: South Atlantic||27.2||Feb. 11, 1931||16||Dec. 28, 1930-Apr. 18, 1931|
|Low: East North Central||9.7||Feb. 25, 1931||10||Jan. 11-Mar. 21, 1931|
|Early 1932||7.4||Mar. 23, 1932||9||Feb. 14-Apr. 16, 1932|
|High: Mountain||24.1||Jan. 6, 1932||17||Dec. 13, 1931-Apr. 9, 1932|
|Low: East North Central||4.6||Mar. 2, 1932||7||Feb. 14-Apr. 2, 1932|
|Winter 1932-33||19.2||Dec. 28, 1932||11||Nov. 20, 1932-Feb. 4, 1933|
|High: West North Central||42.7||Dec. 28, 1932||15||Nov. 20, 1932-Mar. 4, 1933|
|Low: East North Central||13.8||Dec. 28, 1932||12||Nov. 6, 1932-Jan. 28, 1933|
|Winter 1934-35||5.4||Jan. 9, 1935||9||Dec. 2, 1934-Feb. 2, 1935|
|High: East South Central||28.3||Jan. 9, 1935||17||Nov. 25, 1934-Mar. 23, 1935|
|Low: Middle Atlantic||5.3||Jan. 2, 1935||9||Nov. 18, 1934-Jan. 19, 1935|
|Winter 1935-36||12.5||Feb. 26, 1936||22||Dec. 22, 1935-May 23, 1936|
|High: East South Central||61.1||Mar. 25, 1936||24||Dec. 1, 1935-May 16, 1936|
|Low: Pacific||4.5||Mar. 4, 1936||6||Feb. 9-Mar. 21, 1936|
|Winter 1936-37||18.4||Jan. 6, 1937||11||Dec. 20, 1936-Mar. 6, 1937|
|High: Mountain||68.0||Jan. 20, 1937||11||Dec. 13, 1936-Feb. 27, 1937|
|Low: Middle Atlantic||11.4||Jan. 6, 1937||7||Dec. 27, 1936-Feb. 13, 1937|
|Early 1939||5.2||Mar. 1, 1939||9||Feb. 5-Apr. 8, 1939|
|High: East North Central||11.8||Mar. 1, 1939||7||Feb. 12-Apr. 1, 1939|
|Low: Middle Atlantic||2.9||Feb. 8, 1939||7||Jan. 22-Mar. 11, 1939|
|Early 1940||1.9||Feb. 14, 1940||6||Jan. 21-Mar. 2, 1940|
|High: West South Central||13.5||Feb. 7, 1940||8||Jan. 21-Mar. 16, 1940|
|Low: South Atlantic||5.8||Jan. 31, 1940||5||Jan. 21-Feb. 24, 1940|
|Winter 1940-41||5.4||Jan. 29, 1941||10||Dec. 15, 1940-Feb. 22, 1941|
|High: Mountain||15.6||Jan. 8, 1941||6||Dec. 22, 1940-Feb. 1, 1941|
|Low: East North Central||1.9||Jan. 29, 1941||4||Jan. 19-Feb. 15, 1941|
|High: East and West South Central||9.1||Jan. 13, 1943||16||Dec. 20, 1942-Apr. 10, 1943|
|Low: New England||5.4||Apr. 14, 1943||8||Mar. 21-May 15, 1943|
|Winter 1943-44||14.4||Dec. 29, 1943||11||Nov. 28, 1943-Feb. 12, 1944|
|High: East and West South Central||25.7||Dec. 29, 1943||15||Nov. 28, 1943-Mar. 11, 1944|
|Low: Pacific||9.8||Jan. 5, 1944||11||Dec. 5, 1943-Feb. 19, 1944|
|Winter 1945-46||3.7||Dec. 26, 1945||11||Nov. 25, 1945-Feb. 9, 1946|
|High: Mountain||16.9||Feb. 20, 1946||13||Dec. 2, 1945-Mar. 2, 1946|
|Low: Pacific||2.9||Dec. 26, 1945||12||Dec. 23, 1945-Mar. 16, 1946|
|Early 1947||2.5||Mar. 26, 1946||11||Feb. 16-May 3, 1947|
|High: West North Central||9.3||Mar. 12, 1947||12||Feb. 16-May 10, 1947|
|Low: South Atlantic||2.6||Apr. 2, 1947||8||Mar. 9-May 3, 1947|
|High: East and West South Central||6.5||Feb. 25, 1948||12||Dec. 28, 1947-Mar. 20, 1948|
|Low: Pacific||4.1||Jan. 14, 1948||14||Dec. 14, 1947-Mar. 20, 1948|
|High: East and West South Central||2.9||Feb. 2, 1949||8||Dec. 26, 1948-Feb. 19, 1949|
|Low: Mountain||2.6||Mar. 16, 1949||8||Mar. 13--May 7, 1949|
|Early 1950||2.7||Mar. 22, 1950||9||Feb. 26-Apr. 29, 1950|
| High: West North Central
East and West South Central
|Mar. 29, 1950
Mar. 29, 1950
|Feb. 19-Apr. 15, 1950|
Feb. 19-May 6, 1950
| Low: Middle Atlantic
|Mar. 22, 1950
Mar. 29, 1950
Feb. 22, 1950
|Feb. 26-Apr. 29, 1950|
Feb. 12-Apr. 29, 1950
Feb. 19-Apr. 1, 1950
|Early 1951||3.8||Mar. 14, 1951||10||Feb. 11-Apr. 21, 1951|
|High: New England||6.5||Feb. 28, 1951||8||Feb. 11-Apr. 7, 1951|
|Low: East North Central||3.1||Mar. 21, 1951||11||Mar. 4-May 19, 1951|
|Early 1953||6.9||Feb. 11, 1953||13||Dec. 28, 1952-Mar. 28, 1953|
|High: Mountain||13.0||Jan. 28, 1953||14||Dec. 14, 1952-Apr. 21, 1953|
|Low: East North Central||3.4||Feb. 11, 1953||11||Dec. 28, 1952-Mar. 14, 1953|
As data on deaths from all causes were not available for all of the 90-city group in figure 46 the data for the 35 large cities included in that figure were continued from 1921 to the end of 1934, and data for the group of 56 large cities were used for the period 1935 to 1950, with the addition of 6 cities from 1951 to the
middle of 1956. Of the 6 cities added, all were in the Mountain States and 5 were under 100,000 in population, but the Mountain States were poorly represented and additional cities were badly needed for the study. The weekly excess death rates from influenza and pneumonia were then subtracted from the excess death rates from all causes for the same weeks to obtain weekly excess death rates from causes credited primarily to diseases other than influenza and pneumonia which occurred during influenza epidemics, usually with peaks in the same week as the influenza peak.
It would, of course, have been better to have
had a record of influenza-pneumonia epidemics for the same population group for the whole 70 years included in figure 46. However, methods of record-keeping and disease classification are not infrequently changed, so for such a study as the present one, one has to compromise and put together such trend data as are available.
The two end boxes of figure 47 show the maximum heights of the excess rates from influenza and pneumonia in several of the same epidemics, and the variation in the time that the rates exceeded the usual expectancy. The peak week of each epidemic is plotted in the center, with the weeks before the peak on the left and the weeks after the peak on the right. Comparison of the excess curves for the same epidemic in the two end boxes of figure 47 shows that the curves of excess rates are similar, whether they are based on the 90 cities (right box) or the 35 large cities (left box). Although some of this similarity comes from the fact that many of the same cities are included in both groups, it appears to indicate that the characteristics of the epidemics tend to be similar in different places in the United States.
Figure 48 shows excess death rates from influenza-pneumonia and from all causes in the several groups of cities, by weeks from September 1918 to the summer of 1956.
As already indicated, the periods of practically every epidemic of influenza and pneumonia that appear in the middle section of figure 48 have a small epidemic-like peak in the bottom section for causes other than influenza and pneumonia, corresponding approximately in time with the influenza peak in the middle section of the chart. There are a few summer peaks in the "other causes" when no influenza epidemic is present; investigation indicated that these small, usually 1-week peaks, were from some cause other than influenza; occasionally they were caused by deaths due to several days of extremely hot weather. The dates on the chart for these 1-week summer peaks are in italics to indicate that they are unrelated to influenza epidemics.
The next question that arises is, What are the causes other than influenza and pneumonia that show these small peaks when an epidemic of influenza occurs? The top section of figure 49 shows, in a very small scale, the weekly excess death rates from influenza and pneumonia in the whole period from September 1918 to the middle of 1956, together with all causes of
Figure 47. Comparison of the curves of weekly excess mortality from influenza-pneumonia and from all causes during the various epidemics and in different groups of cities in the United States, 1922-44.
death and 7 diseases or disease groups, 6 of which are definitely chronic. These diseases are plotted by months, with no attempt to eliminate trend or the very obvious seasonal variation. However, it is relatively easy to see that in some years the winter season of high mortality from these chronic diseases shows small peaks superimposed upon the seasonal curve. Of course, the curve for all causes shows peaks because it includes influenza and pneumonia as well as other causes. However, 5 of the 6 chronic diseases show small peaks in various years and at approximately the periods of the epidemics as indicated by excess death rates from influenza and pneumonia; cancer shows no trace of excess death rates such as appear for the other chronic diseases. Puerperal septicemia and other acute complications of pregnancy and childbirth show peaks in the early years when, as in 1918, the childbearing ages suffered high pneumonia complications with resulting high mortality rates. However, the less severe epidemics of the last half of the period covered by the chart do not seem to have influenced puerperal mortality.
What the, is the explanation of these small "epidemics" of chronic disease? Deaths, as tabulated in the United States and in most other countries, relate to the "primary cause," and the primary cause is usually defined as the cause that initiated the train of circumstances that resulted in death. Since a patient with a "bad heart," for example, is an easy victim of pneumonia, when influenza-pneumonia strikes, the prognosis is poor, and if the patient dies, the heart disease is likely to be credited as the primary cause since it would usually be first in the train of circumstances that led to the death of the patient. This would also be true for the other chronic diseases shown in figure 49 except cancer, which shows no peaks and no seasonal variation.
Influenza-Pneumonia as a Contributory CauseDeath rates from influenza and pneumonia in this study and elsewhere usually include only sole or primary causes of death. From the tabulations of deaths in the 10 cities surveyed in connection with the study of the 1928-29 epidemic, deaths in which influenza-pneumonia was classified as contributory to some other cause were also tabulated but were used as a separate group. In figure 50, death rates by age are shown separately for influenza-pneumonia as a sole or primary cause and as a contributory cause for both the epidemic period of December 1928-February 1929 and the corresponding months in the preceding nonepidemic period of 1927-28. The death rate based on influenza-pneumonia classified as a contributory or secondary cause has about the same relative age distribution as the death rate for influenza-pneumonia which was classified as the sole or primary cause of death.
Of the total excess deaths from influenza-pneumonia during the 3 months of the 1928-29 epidemic, 63.4 per 100,000 population were classified as the sole or primary cause of death, and 16.3 per 100,000, or 26 percent of the sole or primary total, were classified as contributory causes; of the total of sole, primary, and contributory influenza-pneumonia deaths of 79.7 per 100,000 population, 20 percent were contributory causes. From data on mortality in figure 38, one would expect that there would be numerous deaths charged to causes other than influenza and pneumonia in which influenza-pneumonia would be entered as a contributory cause; however, the tabulations showed that not all excess deaths from causes other than influenza-pneumonia were designated as contributory to influenza. From figure 49, one would expect that in many of these excess deaths, whether or not designated as contributory causes, the primary cause would be recorded as a chronic disease. Such diseases would usually have existed much longer than the influenza and thus would have been regarded as the primary cause in the sense that the chronic disease started the train of circumstances that led to the death. Nevertheless, the death probably would not have occurred at this particular time had influenza not intervened.
Total Excess Death RatesWith excess death rates computed by weeks, the excess deaths for a whole epidemic can be
obtained by adding the weekly excess rates for the weeks that were above seasonal expectancy. However, the weekly excess rates are computed on an annual basis and so must be reduced to an actual basis by dividing their sum by the factor by which the weekly rates (or numbers)
Figure 50. Annual age-specific mortality from influenza-pneumonia certified as the sole or primary cause of death; as a contributory cause only; and the total deaths in which influenza-pneumonia was a sole, primary, or contributory cause--whole populations of 10 cities included in a sample survey of illness from influenza-pneumonia during the 3 epidemic months of December-February, 1928-29, compared with the nonepidemic period of December-February 1927-28.
were multiplied to put them on an annual basis. This factor would be approximately 52 for weekly rates and approximately 12 for monthly rates, although it would change slightly according to the number of days in a given month.
Figure 51 shows the actual total excess death rates for the whole of each epidemic, arrayed by these total excess rates. The epidemics range from 598 excess deaths from all causes per 100,000 population during the 1918-19 pandemic to 14.5 for each of the very small epidemics of 1950 and early 1940. Corresponding rates for excess deaths charged to influenza and pneumonia were 550 and 2.5 per 100,000 population for the epidemics of 1918-19 and early 1940.
An examination of figure 51 reveals that, in general, the smaller the epidemic in total excess
death rates from all causes, the higher the proportion of that excess that is charged to causes other than influenza and pneumonia. This is more easily seen in figure 52, in which the epidemics are arrayed by the percentage of the total excess deaths that were charged to influenza and pneumonia. Thus, in the pandemic of 1918-19, influenza-pneumonia was recorded as the primary cause of death in 92 percent of the excess fatalities, whereas in each of the
small epidemics of 1947 and 1945-46, only 15 percent of all excess fatalities were charged to influenza and pneumonia.
Figure 53 shows total excess death rates for each of the 11 epidemics from 1935-36 to early 1953, inclusive, for a group of cities in each of four large geographic regions or in as many of them as had rates in excess of the usual expectancy. There is much variation in the total excess death rates for influenza and pneumonia and also for all causes. No one or two geographic sections have any monopoly on the highest or the lowest excess rates; in the 11 epidemics shown in figure 53, the South was highest in excess death rates for al causes in 5 epidemics, the Northeast in 4, and the West in 2. In the same 11 epidemics, the North Central section was lowest in excess from all causes in 7 epidemics, the Northeast in 2, and the South and West in 1 each.
As shown in figure 54, in four epidemics the South and the North Central section each had the highest proportion of total excess death rates credited to influenza and pneumonia; in three epidemics, the Northeast section had the highest proportion.
In summary, epidemics of influenza and pneumonia have been decreasing in total excess death rates in a reasonably consistent way since the great pandemic of 1918-19. Figure 55 indicates that, except for the pandemic years 1918-19, the general trend of annual death rates from influenza and pneumonia in the death registration States, including the total continental United States in 1933 and later years, has been downward. The rate of decrease was moderate during the period 1900 to about 1937 but has been considerably more rapid since that time. In 1900, the annual death rate from influenza and pneumonia was 202 per 100,000 population in the registration States; in 1955, it was only 31 per 100,000 a decrease of 85 percent in the 55 years. The 1900 rates were not out of line with those for years immediately following nor were the 1955 rates out of line for immediately preceding years. (See page 45 for an explanation of adjustments made in 1945-55 data because of changes in the International Classification of Diseases and Causes of Death made at the sixth revision conference in 1948.)
Age Variation of Mortality in Different EpidemicsSo much research has been done on the age incidence of and mortality from epidemic influenza that a brief summary of that aspect of the outbreaks seems appropriate.
Although copies of death certificates for the whole United States are received and tabulated in the National Office of Vital Statistics of the Public Health Service, tabulations by age and sex for weeks or months for the total United States, or even for a fair-sized sample, by State, city size, age, sex, and other factors needed in a thorough study of mortality during influenza epidemics would be a large undertaking. Therefore, such detailed tabulations have not been made nor have tabulations by age, sex, and month of death been made. However, the deaths from important causes (including influenza and pneumonia) are tabulated by month of death for all ages combined, and by
age and sex for the year as a whole. Thus, influenza-pneumonia death rates for all ages combined in epidemic months can be compared with similar rates for the same months in a nearby year without an epidemic. If the corresponding nonepidemic months of the 2 years had about the same rates for all ages combined, the excess for all ages would be reasonable accurate. However, the age distribution of deaths during the epidemic might be different from the age distribution during the nonepidemic period; but if the death rates from influenza and pneumonia for all ages are reasonably comparable in nonepidemic years and in nonepidemic months of epidemic years, the death rate for the whole of each of the 2 years can be computed by age, or by age and sex, and subtracted from or in other ways related to similar rates for the epidemic year to obtain a reasonable estimate of excess death rates in the epidemic year per 100,000 population.
Data from the 1928-29 influenza epidemic illustrate the method described in the preceding paragraphs. The Public Health Service canvassed samples of 10 large cities immediately after that epidemic and also obtained from the vital statistics offices of each of the 10 cities a record of all deaths from influenza-pneumonia during the 3-month period of the epidemic, with age, sex, dates of death and birth, and other characteristics, together with similar data for the 3 corresponding months of the year prior to the epidemic and similar data for the calendar years 1927 and 1930. Death rates by age were computed and plotted for both 3-month periods, and for 1928 and 1929, the epidemic years, and 1927 and 1930, the normal years (fig. 56). Rates for the 2 normal periods (solid lines in middle and right boxes) were subtracted from the rates for corresponding periods, including the epidemics (dash lines, middle and right boxes), and these
Figure 56. Excess death rates from influenza and pneumonia computed from deaths during the 3 months of the epidemic and the corresponding months of the preceding year, as compared with deaths during the 2 calendar years which included the epidemic and the calendar years immediately preceding and following those two years; epidemic of 1928-29 in 10 cities in the United States.
differences (representing excesses) were plotted in the left box, in which the rates for the 2-calendar-year periods and the solid line represents the differences between the rates for the two 30-month periods. While the two curves are not identical, there is little doubt that their differences are not significant. Thus, this less precise method gives reasonably accurate results if used with care, especially in selecting the years to be compared. By this method approximate excess death rates from influenza and pneumonia for almost any epidemic year can be computed because all the data needed for the computation are published in the annual vital statistics reports from the National Office of Vital Statistics of the Public Health Service.
In the absence of more detailed data, excess death rates from influenza and pneumonia were estimated by comparing age curves of influenza-pneumonia mortality based on whole calendar years. These data are plotted as age-specific excess death rates per 100,000 population in figure 57 on arithmetic scales; in figure 58, as percentage excesses at different ages; and
in figure 59, as excess death rates on logarithmic vertical scales.
Figure 57 shows the age curves of excess influenza and pneumonia mortality in 12 epidemics, including two early one from the Massachusetts data and the 1953 epidemic from the 62-city data. With the exception of the epidemics of 1918-19, 1920, and to a minor extent of 1922, the age curves are all similar, with high excess rates at the oldest age and some increases under 5 years, with low rates for intervening years, particularly for the ages under 35, which in 1918 included the extremely high young adult peak.
Figure 58 shows excess influenza and pneumonia death rates in still another way, namely, the percentage that the excess influenza and pneumonia mortality rate at a given age in the epidemic year is above the corresponding rate in a nearby nonepidemic year. This method gives quite different age-curves of mortality, in which the 1918-19 and 1920 excess curves are almost identical in appearance but not in actual percentage excesses at the different ages. Since the usual curve of pneumonia, which is the real cause of influenza mortality, is high at the extremes of life, the high rates at those ages in epidemic years are not a high percentage above the high rates of normal years. In the young adult ages, mortality from pneumonia
is normally low, so when there is a young adult peak in mortality in the epidemic the rates for this age group are a very high percentage above the usual low nonepidemic rates. Thus, the most prominent characteristics of the age curves of percentage excesses in the 1918-19 and 1920 epidemics are the young adult peaks, which are far higher than the percentage excesses at any other age.
Figure 59 shows for epidemic years the influenza and pneumonia excess death rates over nearby years during which no special epidemics occurred. These data are plotted on logarithmic vertical scales, which show relative rather than actual differences. Figure 60 shows age curves of total influenza and pneumonia mortality for the 3 calendar years which included the largest influenza epidemics of the last half century, and the age curves of the same diseases for 1914 and 1924, when no general influenza epidemics occurred in the United States.
Comparison of figures 59 and 60 shows that in 1918, the annual influenza and pneumonia death rates for persons under 6 years and 25-29 years of age were about the same and were of the same order of magnitude as the excess rates at 25-29 years in the epidemic of 1918-19. However, in excess death rates in the epidemic of 1918-19 (fig. 59), the excess rate at the young adult peak was considerably higher than the excess rate among children under 5 years of age. Similarly, the total influenza-pneumonia death rate among persons over 70 years of age (fig. 60) was about as high as the young adult peak (25-29 years) of total death rates, but the excess rate (fig. 59) at 70 years and over was extremely low.
In both 1919 and 1920, the total death rates from influenza and pneumonia at the young adult peaks (fig. 60) were considerably lower than the total rates at the two extremes of life. Data were not obtained for the 1919 excess death rates, except as a part of the 1918-19 epidemic, but in 1920 the excess rates at both extremes of life were above those of the young adult peak. In the 1922 epidemic, there was a very minor peak at the young adult ages, with rates at both extremes of life that were far above the young adult peak (fig. 59). Excess mortality rates represented by the three curves in the middle box were higher among children under 5 years of age than among young adults. After 1922, the usual age curve of influenza and pneumonia mortality reappears, with very little evidence of any tendency toward young adult peaks.
Figure 60. Comparison on a logarithmic vertical scale of the age curves of annual influenza and pneumonia mortality during 3 calendar years with large influenza-pneumonia excess deaths and 2 years without appreciable epidemics--1914-24.
Thus far, attention has been centered upon influenza-pneumonia mortality, because mortality records are available for all years. However, special surveys of respiratory diseases have been made after various epidemics covering from 1 or 2 to 10 or more selected localities. In addition, general illness surveys have been in progress when influenza epidemics occurred and the data on influenza, severe colds, and pneumonia have been tabulated separately for the period of the epidemic.
Figure 61 shows age curves of illness from influenza and related diseases during several major and minor epidemics. In the upper left box are age curves of these diseases in several epidemics in Baltimore. The young adult peaks in these illness rates are much less marked than in mortality, and there is much variation in the age curves in the different epidemics. The same general description applies to the data for Hagerstown and Medical Officers' Families (Army, Navy, Public Health Service, and some medical school faculty members) in the upper right box, but not so well to the Cattaraugus Country and Syracuse data or to the 1939-41 epidemics in Baltimore (lower boxes). For various reasons, including lack of personnel, bad roads, and bad weather, the visits in Cattaraugus Country and Syracuse were at less regular intervals, so that colds, even those confining the patient to bed, were not so frequently recalled by the informant and recorded as in surveys with more frequent contacts with the families when the visit of the field interviewer was nearer the date of the illness. On the other hand, in the 1939-41 Baltimore epidemics, influenza and colds were recorded as a part of an intensive survey with regular monthly visits so that colds, confining the patient to bed were reported fairly frequently and probably influenced the age curves of total cases.
Figure 62 shows on logarithmic vertical scales some of the same data as figure 61. However, in figure 62, the diseases included are limited to influenza and grippe only, except in the mill village (upper right box). The few data on the pandemic of the fall of 1918 in one small village show the characteristic young
adult peak. The effect of the logarithmic scale is to show greater variability in the age curves, although some of the additional variability may be due to smaller numbers of cases since the data include only those reported as influenza or grippe.
Figure 63 show for some of the same localities the age incidence of pneumonia and the percentages of all recorded cases of influenza and related diseases complicated by pneumonia. However, in some of the smaller surveys, pneumonia cases were too few to give age curves of any reliability.
Furthermore, in the 1918-19 epidemic, pneumonia incidence was the real source of the young adult peak in mortality, and in 1920 there was a much lower but still definite peak of a similar nature with a considerably higher peak of pneumonia complications. In the epidemics of 128-29, 1939-41, and 1943-44 in Baltimore, and of 1922-23 in Hagerstown, there were no indications of young adult peaks of pneumonia.
Figure 61. Age incidence of certain respiratory diseases during 2- to 4-month epidemic periods, as recorded by special canvasses--Baltimore and other eastern localities, 1918-44.
by pneumonia was almost as high as in 1918-19 (lower left box). However, the pneumonia complications were fewer in the Baltimore epidemics of 1928-29, 1939-41, and 1943-44, and in the Hagerstown epidemic of 1922-23. In a group of surveyed localities in 1918-19, about 6 percent of the total recorded cases were complicated by pneumonia, as compared with about 2 percent in 1928-29. It must be remembered that it is pneumonia that causes nearly all of the fatalities in influenza epidemics.
Figure 64 shows age curves similar to those in figure 634 but on logarithmic vertical scales. The main showing of both types of charts is that the young adult peak of influenza mortality is a reflection of the high peak incidence of pneumonia, which, as a complication of the influenza, is the killing disease.
Influenza and pneumonia rates by sex have not been considered in this study, but it is worth mentioning that in the 1918-19 epidemic in Baltimore the young adult peak at 25-29 years of age in the pneumonia case rate and in the percentage of cases complicated by pneumonia were both roughly 60 percent higher for males than for females.
Two rural areas in Maryland were surveyed in the 1918-19 epidemic. Figure 65 shows age curves for the total of influenza and related
Figure 62. Age incidence, (logarithmic vertical scale) of cases reported as influenza and grippe only, during the same 2- to 4-month epidemic periods as in figure 61, as recorded by special canvasses, 1918-44.
diseases for Charles County, Md., and a group of minor Maryland towns, with comparative data for the same epidemic in Baltimore. The rates in the two rural areas are definitely and consistently above the rate for the city of Baltimore.
Source: Collins, Selwyn D. "Influenza in the United States, 1887-1956." Washington, DC: US Department of Health, Education, and Welfare, Public Health Service, 1957. [Extracted from "Review and Study of Illness and Medical Care With Special Reference to Long-term Trends," Public Health Monograph No. 48, 1957, also known as Public Health Service Publication No. 544.]