Probability of males to outlive females: an international comparison from 1751 to 2020. Marie-Pier Bergeron-Boucher et al. BMJ Open, Volume 12, Issue 8, Aug 2022. https://bmjopen.bmj.com/content/12/8/e059964
Objective To measure sex differences in lifespan based on the probability of males to outlive females.
Design International comparison of national and regional sex-specific life tables from the Human Mortality Database and the World Population Prospects.
Setting 199 populations spanning all continents, between 1751 and 2020.
Primary outcome measure We used the outsurvival statistic ( φ ) to measure inequality in lifespan between sexes, which is interpreted here as the probability of males to outlive females.
Results In random pairs of one male and one female at age 0, the probability of the male outliving the female varies between 25% and 50% for life tables in almost all years since 1751 and across almost all populations. We show that φ is negatively correlated with sex differences in life expectancy and positively correlated with the level of lifespan variation. The important reduction of lifespan inequality observed in recent years has made it less likely for a male to outlive a female.
Conclusions Although male life expectancy is generally lower than female life expectancy, and male death rates are usually higher at all ages, males have a substantial chance of outliving females. These findings challenge the general impression that ‘men do not live as long as women’ and reveal a more nuanced inequality in lifespans between females and males.
Discussion
Our study reveals a nuanced inequality in lifespan between females and males, with between one and two men out of four outliving a randomly paired woman in almost all points in time across 199 populations. These results complement the picture given by the comparisons based on life expectancy, which is a summary measure with no information on variation. A blind interpretation of life expectancy differences can sometimes lead to a distorted perception of the actual inequalities. Not all females outlive males, even if a majority do. But the minority that do not is not small. For example, a sex difference in life expectancy at birth of 10 years can be associated with a probability of males outliving females as high as 40%, indicating that 40% of males have a longer lifespan than that of a randomly paired female. Not all males have a disadvantage of 10 years, which is overlooked by solely making comparisons of life expectancy. However, a small number of males will live very short lives to result in that difference. For example, more baby boys die than baby girls in most countries.
The length of the lifespan of an individual results from a complex combination of biological, environmental and behavioural factors. Being male or female does impact lifespan, but it is not the only determinant contributing to inequalities. Lifespan has been shown to be influenced by marital status, income, education, race/ethnicity, urban/rural residence, etc.33 As we only disaggregated the population by sex and because of this complex interaction, lifespan distributions of females and males overlap. This nuance is captured by the φ metric. Males with a lower education level or who are unmarried have a particularly low chance of outliving a female. But males with a university degree or who are married have a higher chance of outliving females, in particular females with a lower education level and who are single.
As previously discussed, the φ metric expresses the probability of males to outlive females among randomly paired individuals, assuming independence between populations. However, males and females in a population are generally not random pairs but often couples, whose health and mortality have been found to be positively correlated due to a strong effect of social ties on health and longevity.34 Coupled individuals also influence each other’s health,35 and this is particularly true for males, who benefit more than females from being in a stable relationship.36 The datasets used for the analysis do not permit the estimation of the probability of males outliving females for non-randomly paired individuals. However, the outsurvival statistic relates to the probability of the husbands to outlive their wives, and even though such a measure accounts for the difference in age between husband and wife, it has been shown generally to be between 30% and 40%,37–39 values that are quite close to φ .
Other measures of overlap and distance between distributions could have been used. In the online supplemental materials, we compare the outsurvival statistic with a stratification index used by Shi and colleagues20 and the KL divergence. We found that all three indicators are strongly correlated and using any one of these would not have changed the general conclusions from this article. However, unlike the other indicators, φ directly indicates when males live longer than females, which we found in a few instances.
Trends over time in φ are consistent with the reversed trends in sex differences in life expectancy40: in developed countries, the probability of males outliving females decreased until the 1970s, after which it gradually increased in all populations. Studies showed that the increase in sex differences in mortality emerged in cohorts born after 1880,10 41 which is consistent with our analysis of φ (see online supplemental materials). The increase and decrease in sex differences in life expectancy were mainly attributed to the smoking epidemic and other behavioural differences between sexes.7 13 42
The φ values are generally higher in low/middle-income countries. However, this should not be interpreted as a sign of greater gender equality in survival. Southern Asian countries had very high φ values, above 50% in the 1950s and 1960s. Studies for India showed that mortality below age 5 was higher for females than males and remained higher for females in recent years.43 44 However, females had a growing mortality advantage above age 15 years since the 1980s, ‘balancing out’ the disadvantage at younger ages. The reasons for the higher φ and decreasing trends in developing regions vary across countries. It is outside the scope of this study to provide a detailed explanation for the trends in each country.
The outsurvival statistic can be informative for public health interventions.21 Governments develop public health programmes to reduce lifespan inequalities at different levels (eg, socioeconomic status, race, sex, etc). It would be misleading to say that half of the population is disadvantaged by sex differences in lifespan. The inequalities are more nuanced. If 40% of males live longer than females, it could be argued that if a policy aiming at reducing inequalities between sexes targeted the full male population, some of the efforts and investments would be misallocated. Such a policy could be more efficient if φ approaches 0, indicating that sex would explain a large part of the lifespan inequalities within the population, whereas a φ closer to 0.5 indicates that other characteristics (eg, socioeconomic and marital statuses) are involved in creating inequalities. We showed that some subpopulations of males have a high probability (above 50%) of outliving some subpopulations of females. Males who are married or have a university degree tend to outlive females who are unmarried or do not have a high school diploma. Inequalities in lifespan between sexes are attributable to some individuals within each population and not to the whole population. Indeed, Luy and Gast12 found that male excess mortality is mainly caused by some specific subpopulations of males with particularly high mortality. Being able to better identify the characteristics of the short-lived men could more efficiently help tackle male–female inequality.
An important result of our analysis is that the smaller the SD in the age at death, the smaller the φ . The reduction of lifespan inequality observed over time has then made it less likely for males to outlive females. This is partly explained by the fact that lifespan variation reduction has been driven by mortality declines at younger ages.45 When looking at the lifespan distribution (as in figure 1, scenario D), survival improvements at younger ages narrowed the left tails of the distribution for both sexes. By reducing the left tail of female distribution, without increasing the right tail of the male distribution, the overlapping area is reduced. In other words, the number of females with shorter lifespan, easier to outlive, decreased over time. Indeed, it has been shown that mortality declined at a faster pace for females than males below age 50, especially in the first half of the 20th century.46 47 This finding implies that more efforts are required today than in the past to reduce these inequalities, for a same difference in life expectancy. While inequalities were mainly attributable to infant and child mortality, they are today increasingly attributable to older and broader age groups. Men maintained their disadvantage at younger ages, but also faced an increasing disadvantage at older ages. Men are more prone to accidents and homicides in their 20s and 30s than females, and they tend to smoke and drink more leading to higher cancer prevalence and death in their 60s. At the same time, women benefited from reduced maternal mortality and recorded faster mortality decline at older ages. Efforts in reducing lifespan inequalities must thus target diverse factors, causes and ages.13 46 48
A decrease of φ might indicate a discrepancy in the causes of death that affect males and females. External mortality due to accidents and suicide has become more relevant in shaping sex differences in survival in recent years in high-income populations.12 Another example is observed in Latin American populations, where homicides and violent deaths have had an increased burden among males in comparison with females since the 1990s.49 50 In Mexico, for example, the increase in homicide mortality, especially among men between 20 and 40 years, contributed to increasing the gap in mortality between females and males.51 This phenomenon is reflected in the decrease over time in the overlapping of lifespan distributions, directly informing healthcare systems of emerging inequalities.
However, one might ask if a wider overlapping is necessarily better for healthcare systems. On the one hand, a larger overlapping means less inequality between sexes, but on its own it does not ensure that there is more ‘health justice’. For example, if the overlapping areas are large, this still shows a situation of great uncertainty in lifespan for both groups. One health evaluator actor could even prefer a situation where there is a small gap between groups but less inequality within the groups. In the case of sex differences, there might always be between-group differences due to biological factors,2 52 but more health equity could be reached by reducing within-group inequalities. We argue that the outsurvival statistic is a new tool to evaluate health inequalities between groups within a population by uncovering underlying dynamics that are otherwise hidden when looking only at conventional indicators. Therefore, it can inform healthcare systems of the subsequent directions to reach the preferred goal.
Posts
