Gender Disparities in Injury Mortality: Consistent, Persistent, and Larger Than You'd Think

Abstract

Objectives. The purpose of this study is to update knowledge about gender differences in injury mortality.

Methods. Data were drawn from the Web-based Injury Query System, which contains US injury mortality data from 1981 to 2007. Male-to-female rate ratios in injury mortality are calculated for key variables, and age and ethnic group comparisons are made.

Results. Boys and men were more likely than girls and women to die of injury. From 1981 to 2007, the male-to-female age-adjusted rate ratio decreased by 20% to 2.15 for unintentional injury and increased by 11% to 3.91 for violence-related injury. Excess male mortality existed in manner of death, cause of death, and within ethnic and age groups. Additionally, with rare exception, the gender disparity was greater than ethnic and age disparities in unintentional and violence-related injury mortality.

Conclusions. Gender disparities in injury mortality are consistent and persistent. Gender patterns in injury mortality do not follow typical social justice analyses of health, in that men are at greater risk. Lifestyle and behavioral risks as well as masculine socialization merit consideration.

Males are born with a numerical advantage, an advantage that decreases over time. At birth there are 105 boys for every 100 girls.¹ There would be even more, but fetal death is 7% higher for boys than girls.² The mortality gap widens immediately; by their first birthday, 21% more boys than girls die.³ Excess male demise continues throughout life, such that by age 65 years or older, there are 75 men for every 100 women.⁴

These numbers from the United States represent a pattern noted around the globe. Among 229 countries, all have more male than female births, such that the world's gender ratio at birth is 107.⁴ However, the male advantage is not maintained in most locales; 93% of those same 229 nations report more women than men older than 65 years.⁴ Moreover, in relative terms, the gender gap in premature mortality in 187 countries has widened since 1970.⁵

Men are more likely than women to die of almost every disease and illness and to die earlier. Injury, a leading cause of premature death, was no exception.^6,7 Men's higher unintentional injury, suicide, and homicide mortality rates are observed in all age groups in low-, middle-, and high-income countries.⁸ The sole exception is for homicide of children under the age of 15 years in low- and high-income countries, where the rates for girls are similar to or higher than those for boys.

Mortality risk is not stagnant. In the United States, total injury mortality rates decreased from 1979 to 1999, then increased through 2005.⁹ From 1999 to 2005, unintentional injury mortality increased only for Whites, and increased more for White women than White men (19%–20% vs 7%–15%).^10,11 The most recent reviews of gender differences in injury mortality, based on data from the 1980s,^6,7 merit being updated to consider the subsequent gender-related trends.

The purpose of this study is to update knowledge about gender differences in injury mortality, to examine the stability of the differential (over time, across types of injuries, and within 2 key population groups), and to place the resulting information in the context of other injury disparities.

METHODS

Data for the United States were used to explore the magnitude and stability of male-to-female injury ratios. Mortality data from the National Center for Health Statistics were accessed via the Web-based Injury Query System (WISQARS), a repository of fatal and nonfatal injury data.¹²

Data from 1981 to 2007, the first and last years of injury mortality data that currently are available online, were accessed. The International Classification of Diseases-9^th Revision (ICD-9)¹³ classification system was used until 1999, when the ICD-10¹⁴ was instituted. For unintentional injuries, the comparability ratio was 1.0303, which indicated an increase in deaths of 3% as a result of the ICD revision.¹⁵ Virtually all of the increase was because of shifts from natural causes (e.g., pneumonia) in ICD-9 to unintentional injuries in ICD-10. Comparability ratios for suicide and homicide were 0.996 and 0.990, respectively, shifts that were small enough that it appeared that the revision did not substantially affect mortality patterns for suicide or homicide. In addition, in the absence of any indication of a differential or systematic gender bias in coding under the 2 systems, the change in classification systems most likely had little effect on the primary variable of interest, that is, the relative risk for men and women.

Missing data were not a problem in WISQARS. WISQARS mortality data are based on death certificates provided by all 50 US states and the District of Columbia, and most states require each death certificate to report sex, the central variable of interest in this investigation, and will not register a certificate unless it is reported (Robert N. Anderson, Chief, Mortality Statistics, National Center for Health Statistics, e-mail communication, July 8, 2010). In the rare cases for which sex, race, or age are not reported, they are assigned by the National Center for Health Statistics according to established conventions¹⁶ before the data are entered in WISQARS.

Mortality data for men and women were downloaded, and male-to-female (M-to-F) rate ratios (the rate for men divided by the rate for women) were calculated to assess relative risk. M-to-F rate ratios by age and ethnic group also were calculated. To explore the magnitude of gender versus other disparities, rate ratios were calculated for ethnic groups (non-Hispanic Black, Hispanic White, American Indian and Alaska Native, and Asian and Pacific Islander vs non-Hispanic White) and age groups (20–34, 35–54, 55–74, and 75 or more years vs 0–19 years). Analyses involving ethnic groups and investigator-determined age groups were based on deaths beginning in 1990, which was when an Hispanic identifier was added to mortality data as well as the first year that user-generated age groups can be implemented in the online database. All rates and rate ratios, except for age comparisons, were age adjusted, with the standard year being the year 2000. Results were displayed graphically so that the straightforward findings were the most accessible. Statistical tests were not appropriate, because the data were for the population, not a sample.

Boys and men are referred to as men, and girls and women are referred to as women.

RESULTS

From 1981 to 2007, a total of 2 920 260 men and 1 119 669 women died of injury. Findings documented that men were more likely than women to die of injury. This pattern was observed, with rare exception, for manner of death, cause of death, and across ethnic and age groups. The gender disparity in unintentional and violence-related injury mortality was greater, with rare exception, than ethnic and age group disparities in fatal injury.

Consistency of Male Excess

To better understand the stability and trends in gender discrepancies in injury, the 27 years of fatal injury data that were currently readily available were reviewed.

Manner of death. This study researched 2 broad categories of injury, unintentional and violence-related. (Note that WISQARS referred to intentional injury as violence-related injury; this will be the term used herein.) For both men and women, mortality as a result of unintentional injury decreased from 1981 to 1991, then remained relatively stable until 2000, when it began to increase slightly, a trend that continued through 2007. Mortality as a result of violence-related injury decreased for both groups during the same time period (Figure A is available as a supplement to the online version of this article at http://www.ajph.org).

Although both groups were at lower risk of dying from injury in 2007 than in 1981, the difference in their rates resulted in a change in relative risk. As shown in Figure 1, from 1981 to 2007, the M-to-F rate ratio decreased steadily by 20% for unintentional injury mortality and increased by 11% for violence-related mortality. Thus, although men's risk of unintentional injury (relative to women's risk) decreased, men's risk of violence-related injury (relative to women's risk) increased.

FIGURE 1.

Age-adjusted male:female rate ratio, unintentional and violence-related mortality, 1981–2007.

Cause of death. Individual causes of death were explored in a similar manner (i.e., rates for men, rates for women, and rate ratios over time). Each cause of death documented an excess of men over women in each of the 27 years. (Individual graphs are available from the author.) The trend was not consistent across cause of death: the mean 1989 to 1991 and mean 2004 to 2007 M-to-F rate ratios decreased for motor vehicle crashes (from 2.77 to 2.47), falls (from 1.91 to 1.71), drowning (from 4.43 to 3.33), and fire/burns (from 1.84 to 1.64), and increased for poisoning (from 1.60 to 1.81), suffocation (from 2.41 to 2.57), and firearms (from 4.82 to 6.71).

Three causes of death that arguably have relatively little to do with exposure or the behavior of the decedent—deaths of undetermined intent, medical mistakes, and adverse reactions to drugs—also documented higher rates of mortality among men. The M-to-F rate ratio, although substantially lower than that of other causes of death, was consistently greater than 1.0 (Figure B is available as a supplement to the online version of this article at http://www.ajph.org).

Age. Gender disparities in injury risk were observed beginning at the youngest of ages. The general pattern across the study period was for gender disparity to be largest (i.e., more than 2.0) among age groups from late adolescence through age 69, with a peak in young adulthood (20–24 years).

The lowest M-to-F rate ratio by age group for unintentional injury was 1.31 (for those aged 0–4 years in 2005), and the highest was 4.38 (for those aged 25–29 years in 1981). Except for the very young (0–4 years), for whom the relative risk remained generally stable, the trend in relative risk was downward for all age groups. The largest absolute decreases were for young adulthood through middle-age: age 25–29 years (from 4.38 in 1981 to 2.16 in 2007), 30–34 years (from 4.18 in 1981 to 2.93 in 2007), 35–39 years (from 3.70 in 1981 to 2.56 in 2007), and 40–44 years (from 3.42 in 1982 to 2.35 in 2007). The M-to-F rate ratio decreased by 25% or more in 10 of the remaining 14 age groups. The M-to-F rate ratio in unintentional injury did not increase over the 27 years for any of the age groups.

As shown in Figure 2, gender disparities by age were larger for violence-related injury. For the oldest age group (85 or more years), for example, the gender rate ratio for violence-related injury was 7-fold that for unintentional injury in the same age group. The general pattern was for the gender disparity to climb steeply from birth through late adolescent and young adult age groups, to decrease and remain similar across age groups during middle-age, and to climb steeply again from age 65 years onward.

FIGURE 2.

Male:female rate ratio, unintentional and violence-related injury mortality, by age group, 2007.

The lowest M-to-F rate ratio for violence-related injury was 0.84 (for those aged 0–5 years in 2002; the rate ratio was less than 1.0 for only 6 of the 486 age-group-by-year categories), and the highest M-to-F rate ratio was 10.1 (for persons aged 75 years or older in 1999 and again in 2003). From 1981 to 2007, the M-to-F rate ratio in violence-related mortality remained roughly the same in 8 of the 18 5-year age groups and increased in the remaining 10 age groups. At times the increase in the rate ratio was particularly notable; for each of the 5-year age groups from 10 through 24 years, the rate ratio increased by approximately 40%.

Ethnicity. As shown in Figure 3, the M-to-F rate ratio was higher for violence-related (vs unintentional) injury mortality for each ethnic group. The M-to-F rate ratio was greater than 1.6 for unintentional injury and greater than 2.4 for violence-related injury each year in each ethnic group.

FIGURE 3.

Age-adjusted male:female rate ratio for (a), unintentional and (b) violence-related injury mortality, by ethnic group, 1990–2007.

The pattern of M-to-F rate ratios over time differed by ethnic group and manner of death. Among unintentional injury deaths, Hispanic Whites had the largest M-to-F rate ratio in 17 of the 18 years of data and Asians and Pacific Islanders had the lowest (just less than 2.0) each year. From 1990 to 2007, the M-to-F rate ratio dropped substantially for American Indians and Alaska Natives (from 3.0 to 2.1), decreased gradually for non-Hispanic Blacks (from 2.9 to 2.5), and remained comparatively stable for non-Hispanic Whites (from 2.3 to 2.1).

Among violence-related deaths, Asians and Pacific Islanders again had the lowest M-to-F rate ratio (∼2.5). The M-to-F rate ratio for non-Hispanic Whites remained stable near 3.5, and the M-to-F rate ratio for Native Americans, although less stable, trended downward to a similar point. M-to-F rate ratios for Hispanic Whites and non-Hispanic Blacks mirrored each another: the M-to-F rate ratio began high for Hispanic Whites and decreased (from 6.2 to 5.1), whereas for non-Hispanic Blacks, it began at 5.1 and ended at 6.0.

Other Sociodemographic Disparities in Injury

To put the gender disparity in context, injury risk associated with 2 other sociodemographic variables—age and ethnicity—were examined. As can be seen in Figure 4, in unintentional injury, the gender disparity was greater than ethnic disparities. The higher gender rate ratio was evident across all years.

FIGURE 4.

Age-adjusted male:female and ethnic group rate ratios for (a), unintentional and (b) violence-related injury mortality, age adjusted, 1990–2007.

A similar pattern was observed for violence-related injury: aside from 1 ethnic disparity (non-Hispanic Blacks vs non-Hispanic Whites) in 3 years in the early 1990s, the gender disparity was higher than all ethnic disparities in violence-related injury.

The M-to-F rate ratio in unintentional injury was similar to that for each age group with the exception of the oldest versus the youngest (older than 75 vs younger than 19 years), where the rate ratio was very large (Figure C is available as a supplement to the online version of this article at http://www.ajph.org).The M-to-F disparity in violence-related mortality almost always (in 65 of the 72 age-group-by-year comparisons) exceeded that of age groups.

DISCUSSION

Men are more likely than women to die of injury. The pattern of excess male mortality holds, with rare exception, across time, manner of injury, cause of injury, and age and ethnic groups. Men's risk of unintentional and violence-related injury overall is at least 2 and 3 times, respectively, that of women during each year of the last generation.

Moreover, the gender disparity in unintentional and violence-related injury mortality is greater than the disparity by ethnicity and age group. These findings do not imply that injury mortality is not important for women or that ethnic disparities in injury mortality are not a problem. Injury ranks highly as a cause of death for women and for all ethnic groups.¹⁷ Thus, persons interested in women's health would be expected to be concerned about injury mortality. In addition, for some specific causes of death (e.g., homicide among non-Hispanic Blacks, motor vehicle crashes among American Indians and Alaska Natives), ethnic disparities exceed gender disparities. Nonetheless, gender disparities consistently exceed age and ethnic disparities in injury mortality.

These findings might be generalizable. Data from low-, middle-, and high-income countries document that men have higher rates of unintentional injury, suicide, and homicide mortality than do women.⁸ Systematic analysis of gender differences in injury mortality in multiple and diverse countries would help document the scope and nature of the phenomenon. To my knowledge, no other such analyses have been published in the peer-reviewed literature. Adequate vital statistics systems and other components of a well-functioning data infrastructure would facilitate the likelihood of such research.

Mortality data were chosen for this investigation because they are the most reliable and widely available data. Whether the same pattern of findings (e.g., gender disparities are larger than ethnic and age group disparities) applies to nonfatal injury remains to be seen.

Information about male–female differences, as with other group differences, in injury mortality could help prevention and intervention efforts. A thorough understanding of the phenomenon would allow for the development of increasingly sophisticated public health activities and their evaluation.

Frameworks for Understanding Injury Disparities

Although understanding the basis for the gender disparity in injury mortality was beyond the scope of this study, it might be useful to place the findings in the context of a brief review of existing conceptual frameworks. Most consider biological, sociodemographic, or behavioral characteristics links to mortality.

Regarding the idea of male advantage followed by disadvantage, perhaps male excess in injury mortality is part of a general pattern within the species. No current evidence supports the idea that men are physiologically more vulnerable to injuries and their fatal outcomes.

Current public health analyses of mortality patterns tend to focus not on the biological but on the social and individual. Health disparities research, for example, typically uses a social and structural framework in which certain population groups (e.g., low socioeconomic classes, African Americans) are thought of as being at higher risk of poor health because of their position in society. The disadvantaged populations are less likely to be in the labor force, to be paid less when employed, more likely to face prejudice and discrimination, and to experience considerable role strain. If this framework were to be extended to gender, women, about whom research generates similar findings, would be expected to have higher mortality rates. However, men, not women have higher rates of injury and all mortality. Thus, a structural disadvantage framework generally is not applied and does not, at least as commonly conceptualized, seem to be directly relevant to gender disparities in injury mortality.

Nine lifestyle and individual behavior choices are considered the “actual causes of death”¹⁸ in the developed world.¹⁹ Perhaps men are more likely to be in circumstances that increase their exposure to injury risk. Some of the circumstances (e.g., occupational) might result in extended exposure (e.g., long-distance trucker). Others might be situational and, sometimes largely, of men's own making. Alcohol use is 1 example. Alcohol consumption is common in both unintentional injury and violence-related mortality. General population surveys in 10 countries indicate that men consistently exceed women in typical drinking frequencies and quantities, as well as in rates of heavy drinking episodes.²⁰ Male gender is the largest predictor by far of driving under the influence.²¹

Such behaviors also are a product of the expectations of others. Good et al. wrote, “It has long been noted that masculinity can be harmful to men's health” and observed that scholars theorize that masculine socialization predisposes young men to take excessive risks.^{22(p 39)} Risk taking, which is related positively to injury,²³ is long associated with being male; a meta-analysis of 150 studies supported this assumption.²⁴ Gender and risk taking are so closely linked, that boys and girls as young as 6 years old believe that boys are more risk-taking than girls, but that boys are at lower risk of injury than girls.²⁵ In addition, recent injury-related research indicated that, more than biological sex, the level of masculinity and the level of internalization of gender roles explains gender differences in risk taking among adolescent pedestrians.²⁶ These studies suggest that risk taking is perceived as a male-linked trait without negative consequences. However, men's risk taking is not constant over ages or contexts,²⁴ which suggests that it is situation specific. Moreover, risk taking might be related to a gender-linked difference in competitiveness.²⁴ Recent research in Australia²⁷ supported this previous work,²⁴ indicating that the gender gap in risk taking decreased across cohorts; girls became more risk taking, which might be to the detriment of injury prevention.

In reality, a combination of factors best describes the basis of gender differences in injury mortality. Li et al.²⁸ invented a technique, called “decomposition method,” to quantitatively examine the major determinants of the male–female discrepancy in injury mortality rates. When examining gender differences in the rate of fatal motor vehicle crashes per driver (women were involved in fewer), they found that 51% of the discrepancy was attributed to driver gender and 41% to exposure differences (i.e., annual average miles per driver).²⁸ Gender differences in death rates from bicycling injury yielded comparable percentages.²⁹ Risk associated with gender and exposure could be allocated via the decomposition method, but the largest difference—gender—remains a black box.

Conclusions

Whether the focus is on injury or on health disparities, a greater acknowledgment of the pervasiveness of gender disparities in injury mortality would benefit prevention efforts. Public health attempts to change lifestyle and behavior with population-level interventions, (e.g., comprehensive campaigns to prevent and reduce tobacco use, policies and programs designed to change diet and activity patterns to reduce obesity). Relatively little effort, however, is invested in modifying masculinity-linked behavior. As noted recently,³⁰ gender-based risks are, in principle, amenable to social change, and they offer untapped potential for health interventions.