Socioeconomic disparities in U.S. mortality: The role of smoking and alcohol/drug abuse

Abstract

Prior studies have identified smoking as a key driver of socioeconomic disparities in U.S. mortality, but the growing drug epidemic leads us to question whether drug abuse is exacerbating those disparities, particularly for mortality from external causes. We use data from a national survey of midlife Americans to evaluate socioeconomic disparities in all-cause and cause-specific mortality over an 18-year period (1995-2013). Then, we use marginal structural modeling to quantify the indirect effects of smoking and alcohol/drug abuse in mediating those disparities. Our results demonstrate that alcohol/drug abuse makes little contribution to socioeconomic disparities in all-cause mortality, probably because the prevalence of substance abuse is low and socioeconomic differences in abuse are small, especially at older ages when most Americans die. Smoking prevalence is much higher than drug/alcohol abuse and socioeconomic differentials in smoking are large and have widened among younger cohorts. Not surprisingly, smoking accounts for the majority (62%) of the socioeconomic disparity in mortality from smoking-related diseases, but smoking also makes a substantial contribution to cardiovascular (38%) and all-cause mortality (34%). Based on the observed cohort patterns of smoking, we predict that smoking will further widen SES disparities in all-cause mortality until at least 2045 for men and even later for women. Although we cannot yet determine the mortality consequences of recent widening of the socioeconomic disparities in drug abuse, social inequalities in mortality are likely to grow even wider over the coming decades as the legacy of smoking and the recent drug epidemic take their toll.

Keywords: Alcohol abuse; BMI, Body Mass Index; Cause-specific mortality; Drug abuse; MIDUS, Midlife in the United States; MSM, Marginal Structural Model; Mortality; SAQ, Self-Administered Questionnaire; SES, Socioeconomic Status; Smoking; Socioeconomic status.

Fig. 1

Prevalence of health behavior mediators at M1 (1995–96) by cohort and SES.

Fig. 2

Percentage of SES disparity in all-cause and cause-specific mortality mediated by alcohol/drug abuse and smoking. Note: See Table S6 for detailed estimates with confidence intervals. Results for other cancer and external/alcohol/drug-related mortality are not shown because the total effect of SES is not significant.