Trends in Mortality Among Females in the United States, 1900–2010: Progress and Challenges

Introduction We analyzed trends in US female mortality rates by decade from 1900 through 2010, assessed age and racial differences, and proposed explanations and considered implications. Methods We conducted a descriptive study of trends in mortality rates from major causes of death for females in the United States from 1900 through 2010. We analyzed all-cause unadjusted death rates (UDRs) for males and females and for white and nonwhite males and females from 1900 through 2010. Data for blacks, distinct from other nonwhites, were available beginning in 1970 and are reported for this and following decades. We also computed age-adjusted all-cause death rates (AADRs) by the direct method using age-specific death rates and the 2000 US standard population. Data for the analysis of decadal trends in mortality rates were obtained from yearly tabulations of causes of death from published compilations and from public use computer data files. Results In 1900, UDRs and AADRs were higher for nonwhites than whites and decreased more rapidly for nonwhite females than for white females. Reductions were highest among younger females and lowest among older females. Rates for infectious diseases decreased the most. AADRs for heart disease increased 96.5% in the first 5 decades, then declined by 70.6%. AADRs for cancer rose, then decreased. Stroke decreased steadily. Unintentional motor vehicle injury AADRs increased, leveled off, then decreased. Differences between white and nonwhite female all-cause AADRs almost disappeared during the study period (5.4 per 100,000); differences in white and black AADRs remained high (121.7 per 100,000). Conclusion Improvements in social and environmental determinants of health probably account for decreased mortality rates among females in the early 20th century, partially offset by increased smoking. In the second half of the century, other public health and clinical measures contributed to reductions. The persistent prevalence of risk behaviors and underuse of preventive and medical services indicate opportunities for increased female longevity, particularly in racial minority populations.


Introduction
From the beginning of the 20th century to 2010, the life expectancy at birth for females in the United States increased by more than 32 years (1) However, new causes of death have emerged with changes in technology and the built environment (eg, the automobile and highways), emerging infections (eg, HIV), and behavior (eg, cigarette smoking). We analyzed trends in mortality rates among females at each decade from 1900 through 2010, focusing on major causes of death, and examined differences by age and by race. Historical trends may indicate future trends, contributing factors, opportunities for intervention when interventions are known, and research needs when they are not.

Methods
Data for the analysis of decadal trends in mortality rates were obtained from yearly tabulations of causes of death from published compilations and from public use computer data files. Data for 1900 through 1940 were taken from mortality information from death registration states, which included 10 states and the District of Columbia in 1900 (40.5% of the US population) and gradually expanded to include all 48 states and the District of Columbia by 1933 (2). For decennial mortality rates from 1940 through 1960, a compilation of mortality information was used (3). The Centers for Disease Control and Prevention's (CDC's) Compressed Mortality File 1968-1992 and WONDER data system (Wide-ranging Online Data for Epidemiologic Research; http://wonder.cdc.gov) were used for mortality counts and census denominators from 1970 through 2010. Two physician epidemiologists linked ICD (International Classification of Diseases) codes. Because data specific to the black population became available beginning in 1970, we summarized only all-cause trends for this population for the available period and concentrated on prior trends for "nonwhites," a category that also includes other racial groups. Because data for age-specific rates by race became available beginning in 1920, the tabulation of AADRs began in this year (Appendices I and II). We used the term "females" in this study, because all ages were included in the analysis. We refer to "major causes of death" to distinguish from slightly different classifications of "leading causes of death" used by CDC's National Center for Health Statistics.
We analyzed all-cause unadjusted death rates (UDRs) for males and females and for white and nonwhite males and females from 1900 through 2010 in decadal years to indicate mortality burden. We analyzed UDRs for black persons beginning in 1970 when the data were first made available. We also computed age-adjusted allcause death rates (AADRs) by the direct method using age-specific death rates and the 2000 US standard population (4). We analyzed several trends in mortality rates among females only. We analyzed trends in all-cause, age-specific death rates by white/nonwhite race from 1900 through 2010 in decadal years, major causes of death in 1900 and 2010, trends in all-cause AADRs from 1900 to 2010, and trends in AADRs for specific causes of death from 1920 to 2010 by white/nonwhite race. We also analyzed trends in AADRs from 1900 through 2010 for selected chronic conditions (ie, heart disease, stroke, and cancers combined), selected infectious diseases (ie, influenza and pneumonia, tuberculosis, and enteritis and diarrhea combined), and unintentional injuries (ie, unintentional motor vehicle [UI-MV] and nonmotor vehicle injuries [UI-NMV] combined).

Results
All-cause UDR for females and males and whites and nonwhites, 1900-2010 From 1900 to 2010, the UDR among females in the United States decreased from 1,646.9 per 100,000 to 787.4 per 100,000, an over-all decrease of 52.2% (Table 1). Among males, the UDR decreased from 1,791.1 per 100,000 in 1900 to 812.0 per 100,000 in 2010, an overall decrease of 54.7% (Table 1). The male UDR exceeded the female UDR in all decadal years except 2000; by 2010, the male excess had decreased to 24.6.
Decreases in all-cause UDRs from 1900 to 2010 were higher among nonwhites than among whites for females and males (Table 1). UDRs decreased by 78.5% among nonwhite females and 47.4% among white females and by 76.8% among nonwhite males and 51.2% among white males. Among females, the UDR among nonwhites exceeded that among whites by 814.1 per 100,000 in 1900; nonwhite excess deaths decreased steadily, and beginning in 1970 the white female UDR exceeded the nonwhite female UDR. Among males, the all-cause UDR among nonwhites exceeded that among whites by 790.5 per 100,000 in 1900; this excess decreased steadily, and by 2010 the UDR among white males exceeded the rate of nonwhite males by 270.5 per 100,000.
From 1970 to 2010, death rates increased by 5.5% among white females and decreased by 22.5% among black females. Rates decreased by 20.3% among white males and by 38.9% among black males.
All-cause AADR for females and males, whites and nonwhites, 1900-2010 From 1900 to 2010, the AADR among females decreased from 2,410.4 per 100,000 to 634.9 per 100,000, a decrease of 73.7% (Table 2). Among males, the AADR decreased from 2,630.8 per 100,000 in 1900 to 887.1 per 100,000 in 2010, a decrease of 66.3% ( Table 2). The male AADR exceeded the female AADR in all decades, with the greatest excess in 1970 at 570.7 per 100,000; the male excess was higher in 2010 than in 1900.
Decreases in all-cause AADRs were higher among nonwhites than among whites for females and males ( Table 2). The AADR decreased by 80.8% among nonwhite females and by 73.7% among white females and by 74.5% among nonwhite males and 66.4% among white males. Among females, the AADR among nonwhites exceeded that among whites by 914.0 in 1900; the excess decreased steadily to 5.4 per 100,000 in 2010. Among males, the all-cause AADR among nonwhites exceeded that among whites by 963.3 per 100,000 in 1900; the excess decreased inconsistently, and by 2010 the difference was 32.5 per 100,000. In 1970, the AADR from all causes was 30.2% higher among black than among white females. The difference remained at approximately this level until 2010, then fell to 19.3%. During the same period, the excess AADR among black males changed little, from 23.8% to 25.7% (Table 2).
Age-specific all-cause death rates among females, by white and nonwhite race, 1900-2010 Relative decreases in death rates decreased by increasing age ( Table 3). The greatest decreases were among ages 1 to 4 years, a decrease of 98.8% among whites and 99.3% among nonwhites. Rates of decrease were greater among nonwhite females than among white females, except for females aged 85 years or older.
Although decreases in death rates among nonwhite females were continuous in all decades from 1900 through 2010, decreases did not continue for white females in recent decades. Among white females, death rates increased from 1960 to 1970 in females aged 15 to 24, 35 to 44, and 45 to 54 years; from 1990 to 2000 in females aged 35 to 44 years; and from 2000 to 2010 in females aged 25 to 34 and 45 to 54 years (data not shown).

Major causes of death among females in 1900 and 2010
The 5 major causes of death for females in 1900 (46.3% of all deaths) were pneumonia and influenza (198.5 per 100,000), tuberculosis (187.8 per 100,000), enteritis and diarrhea (134.9 per 100,000), heart disease (133.7 per 100,000), and stroke (107.7 per 100,000). Of these causes, only heart disease and stroke were among the 5 major causes in 2010. In 2010, the 5 major causes (59.7% of all deaths) were heart disease (184.9 per 100,000), all cancers (168.2 per 100,000), stroke (49.1 per 100,000), chronic lower respiratory diseases (46.3 per 100,000), and UI-NMV (21.8 per 100,000). Direct comparison of ranked causes between the 1900 and 2010 was not possible, because data for one major cause in 2010 (ie, chronic lower respiratory diseases) were not available in 1900.
Cause-specific AADR among females (per 100,000 population) From 1900 to 2010, the greatest decreases in AADRs were among selected infectious diseases: the AADR for pneumonia and influenza decreased by 95.7%, tuberculosis by 99.9%, and enteritis and diarrhea by 99.2%; by 1950, rates of each of these conditions had already decreased by more than 85% (Table 4, Figure 1). AADRs for diseases of the heart increased by 96.5% in 1950 and then decreased rapidly by 2010 by 70.6%. AADR for stroke peaked in 1920 and subsequently decreased by 84.5%. AADRs for UI-NMV decreased by 63.9%. The AADR for 3 causes of death increased greatly during the study period (Table 4). This was in contrast to the AADR for all cancers, which increased to a peak in 1940 and decreased by 22.6% thereafter. The AADR for cancers of the respiratory system (including lung cancer) was 6.9 per 100,000 in 1950 and increased to 38.8 per 100,000 in 2010. Chronic lower respiratory conditions, first reported in 1970 at 8.1 per 100,000, increased by 369.1% to 38.0 per 100,000 in 2010. The AADR for motorvehicle injury-related deaths, first recorded in 1910 at 0.8 per 100,000, peaked at 16.0 per 100,000 in 1930 and subsequently decreased by 59.4%.

AADR among white females compared with nonwhite females, 1920-2010
From 1920 through 2010, the decrease in AADR from all causes was greater among nonwhite females (76.9%) than among white females (68.9%) ( Table 4). From 1920 through 2010, AADRs reached a peak one decade earlier for whites than for nonwhites for heart disease (1940 and 1950, respectively) and stroke (1920 and 1930, respectively), and subsequently decreased. AADRs decreased similarly for white and nonwhite females for tuberculosis, enteritis and diarrhea, and pneumonia and influenza; were slightly greater for UI-NMV deaths for nonwhite females than for white females; and were substantially greater for whites than for nonwhites for cancers ( Table 4). The increase of UI-MV deaths was far greater for nonwhite females than for white females.
AADR trends among females for major selected chronic, infectious, and unintentional injury conditions, 1900-2010 In 1900, the selected chronic, infectious, and injury conditions assessed in this analysis accounted for 53.8% of the female AADR (Table 4, Figure 2). Trends in the AADR for these selected conditions varied markedly. Rates for selected chronic conditions increased to a peak in 1940 and 1950, then decreased. Rates for the infectious conditions decreased rapidly, began to level out in 1950, and decreased slowly thereafter. Unintentional injury rates rose slightly to a maximum in 1930, declining slowly thereafter.

Discussion
During the past 110 years, mortality rates among females in the United States decreased substantially, particularly for younger females and nonwhite females. Mortality rates from infectious diseases decreased precipitously by mid-century. In contrast, heart disease mortality rates among females increased in the first half of the century; they then decreased during the second half. Mortality from respiratory system cancers and chronic respiratory disease among females also rose dramatically, starting in the mid-1960s. The early risk in UI-MV mortality quickly plateaued and then decreased.
Providing explanations for observed trends is challenging (Appendix III). Available data allow only ecological and retrospective analyses. Sources vary in accuracy, specificity, and consistency over time. Exposures commonly interact, but information on conjoint exposures is lacking. Racial categories and disease categor-ies changed during the study period. Explanations are thus hypothetical and imprecise.
We divided potential explanations into social and environmental factors, such as education, employment, poverty, sanitation, housing, and transportation; biological and behavioral factors, such as hypertension, cholesterol levels, cigarette smoking, physical activity, and diet; and preventive and therapeutic interventions, such as measles vaccination, hypertension screening, and treatment of cardiovascular disease. Changes in ICD coding may also account for some trends observed. Overall, social and environmental explanations and biological and behavioral explanations predominate in the first half of the 20th century, whereas preventive and therapeutic interventions then gained importance.
Twenty years of the 30-year increase in female life expectancy from 1900 to 2010 occurred between 1900 and 1950 (1), affected principally by social and environmental factors. During the first half of the 20th century, sanitation improved substantially, with greater benefits for blacks than for whites (5,6). Sanitation, the provision of clean drinking water and safe disposal of sewage and solid waste, affected rates of infectious and chronic diseases (6) and was associated with almost half the total decrease in mortality rates in major US cities between 1900 and 1940, three-quarters of the decrease in infant mortality rates, and almost two-thirds of the decrease in child mortality rates (5).
Decreases in infectious disease mortality rates were also probably associated with improved housing, increased education and income, and reduced poverty (7). From 1900 to 1940 the mean number of occupants per housing unit in the United States fell from 4.8 to 3.8, reducing infectious disease contagion (8). The large decrease in major infectious diseases preceded deployment of antibiotics and immunization.
Education, an established determinant of health, increased substantially during this period. From 1900 to 1950, female high school graduation rates increased from almost 30% to approximately 80% (9). Postsecondary education for females also increased (10). Between 1940 and 1970, the median years of schooling increased for nonwhite and white females (8).
There were increases in proportions of the workforce that was female, of married females working outside the home, of female compared with male earnings, and of black compared with white earnings (10), all contributing to improved economic well-being for females and nonwhites and likely to have led to improved health.
Three major nonexclusive explanations for increased heart disease mortality rates are possible. First, as understanding of diseases improved, the apparent rise may have partly resulted from changes in classifying and assigning causes of death during the first half of the century. Examples include the shift from nonspecific causes, (eg, "ill-defined conditions") to specific causes (eg, ischemic heart disease) and the 1929 reclassification of diseases of the coronary arteries from "diseases of the arteries" to a new subcategory of heart disease, "diseases of the coronary arteries and angina pectoris" (11,12). Second, the rise has also been attributed to a reduction in "competing causes" of death, most notably the reduction of deaths due to infectious and diarrheal diseases (13).
Third, cigarette smoking was a major influence on trends in female chronic disease mortality rates, particularly trends for heart disease, respiratory system cancers, and chronic lower respiratory disease (14-16). The prevalence of cigarette smoking among females rose rapidly in the 1930s, peaked from about 1965 to 1975, and decreased thereafter (15-18). This trend is consistent with the rise of cancer death rates among females in the 1960s. Smoking is also associated with particularly high relative risks for heart disease and stroke among females younger than 50 (15,16). It is plausible that the rapid rise of smoking in the first half of the century was also associated with much mortality, principally from heart disease (19). Holford et al estimated that the decrease of smoking among females between 1964 and 2012 averted 2.7 million deaths from all causes (19). However, the decrease in deaths predated the decrease in smoking among females by 5 to 10 years, suggesting that other factors contributed (13,20). Investigators have called for additional research to explain the 20th century rise and decrease of heart disease (13,20).
Other causes of heart disease and of stroke merit assessment. Estimates of blood pressure trends indicate that blood pressures decreased during the 20th century (21), which is consistent with the decrease of stroke mortality rates during the study period. Nationally representative estimates of rates of hypertension in 1960 were 26.4% among white females and 43.1% among black females (22). Between 1988 and 1991, rates had fallen to 16.7% and 28.1%, respectively (22). Control of hypertension has increased (23). Trends in hypertension control are consistent with the accelerating decrease in trends in stroke deaths after 1970.
Nationally, the mean number of kilocalories consumed (ie, acquired but not necessarily ingested) per capita per day (for males and females combined) increased from approximately 3,400 from 1909 to 1919 to 3,900 in 2000 (24). Nationally representative estimates of rates of elevated cholesterol blood levels -another risk factor for heart disease -are not available before 1960, after which mean levels among both black and white females slowly decreased (25). Criteria for overweight have also changed since 1960, but trends indicate a substantial rise for both black and white females (26). The prevalence of diabetes in the population -another risk factor for heart disease -rose from 1% in 1958 to PREVENTING CHRONIC DISEASE www.cdc.gov/pcd/issues/2018/17_0284.htm • Centers for Disease Control and Prevention 7% in 2011 (data not available for females alone) (27). During the 20th century, total physical activity decreased -a protective factor for heart disease and stroke (28). Long-term trends in overweight, physical activity, and the prevalence of diabetes do not explain the trends in heart disease and stroke without considering the effects of improved medical care (26-28). The decrease in saturated fat ingestion is consistent with declining heart disease and stroke mortality rates in the second half of the 20th century (25).
In the last half of the 20th century clinical interventions were developed and deployed for the treatment of heart disease and stroke (29). Ford and colleagues estimated that 47% of the decrease in heart disease mortality rates since 1980 are attributable to clinical treatments and 44% to changes in risk factor prevalence (30). Conversely, increases in the prevalence of overweight and diabetes may have slowed the rate of decrease in heart disease mortality rates. Various screening tests (eg, mammography, Pap tests, colonoscopy) also affect mortality associated with specific cancers. Their use in the second half of the 20th century, along with advances in treatment, may be responsible for some of the decrease of age-adjusted cancer mortality rates (31).
Since 1998, the anomalous increasing death rate among white females in certain age groups has been noted, particularly among those with a high school education or less. Deaths are largely attributed to drug and alcohol poisoning, suicide, chronic liver disease, and cirrhosis (32). We found that these increases began earlier than previously reported and affected additional age groups.
The lack of large increases in UI-MV deaths during the 20th century is notable (33). From 1900 to 1970, the number of automobiles rose from 4,100 to 6.5 million; from 1923 to 1970, the miles of federally supported highways grew five fold; miles driven per capita annually climbed almost six fold from 1930 to 2000; and from 1945 to 1970, average vehicle highway speed increased from 45 to 60 miles per hour (8). Large increased exposure to risks for UI-MV deaths, is not, however, reflected in long-term trends in UI-MV mortality rates. Deaths per mile driven reached a low peak in 1930 and gradually decreased thereafter, probably in association with safety measures (34). In the first 3 decades of the 20th century, many UI-NMV deaths were from falls, drowning, railroad injuries, and burns (2); these causes decreased greatly after 1940. However, in recent decades deaths among females from drug overdoses have risen rapidly and since 2007 have exceeded deaths from UI-MV.
Much of the decrease in mortality rates among females in the past 110 years is attributable to improvements in major social and environmental determinants of health -education, income, housing, and sanitation. The rapid decrease in mortality rates from infectious by mid-century largely preceded the widespread use of anti-biotics or immunization. The extent and specific causes of increased heart disease mortality rates among females in the first half of the century remain uncertain. The decrease of heart disease mortality rates during the second half of the century may be the result of multiple factors. The dramatic rise in mortality rates from respiratory system cancers and chronic respiratory disease among females is most likely due to cigarette smoking. The plateau in UI-MV mortality rates, despite the rapid growth of automobile use during the century, is probably a result of early safety measures.
Trends in mortality rates during the past century reflect major patterns of health determinants. Sanitary and safety improvements along with understanding of and therapies for infectious diseases led to great reductions in infectious causes of death. With increasing longevity and more sedentary lifestyles, chronic diseases increased as major causes of death. Although some of these causes, particularly heart disease and stroke, decreased as a result of behavior change and effective health care (22), decreases in mortality rates are slowing. Ongoing and expanded efforts to control underlying determinants should accelerate decreases in mortality rates and reduce inequities.       Abbreviation: -, data not available. a Age-adjusted rates are per 100,000 US 2000 standard population. See Appendix I for sources of rates. b Selected infectious disease totals are the sum of the rates for influenza and pneumonia, tuberculosis, and enteritis and diarrhea. Selected chronic condition totals are the sum of the rates for heart disease, stroke, and all cancers. They do not include chronic lower respiratory diseases. Unintentional injury totals are the sum of the rates for unintentional motor vehicle and nonmotor vehicle injuries.