Can lifestyle factors explain racial and ethnic inequalities in all-cause mortality among US adults?

Background: Racial and ethnic inequalities in all-cause mortality exist, and individual-level lifestyle factors have been proposed to contribute to these inequalities. In this study, we evaluate the extent to which the association between race and ethnicity and all-cause mortality can be explained by differences in the exposure and vulnerability to harmful effects of different lifestyle factors. Methods: The 1997-2014 cross-sectional, annual US National Health Interview Survey (NHIS) linked to the 2015 National Death Index was used. NHIS reported on race and ethnicity (non-Hispanic White, non-Hispanic Black, and Hispanic/Latinx), lifestyle factors (alcohol use, smoking, body mass index, physical inactivity), and covariates (sex, age, education, marital status, survey year). Causal mediation using an additive hazard and marginal structural approach was used. Results: 465,073 adults (18-85 years) were followed 8.9 years (SD:5.3); 49,804 deaths were observed. Relative to White adults, Black adults experienced 21.7 (men; 95%CI: 19.9, 23.5) and 11.5 (women; 95%CI: 10.1, 12.9) additionaldeaths per 10,000 person-years whereas Hispanic/Latinx women experienced 9.3 (95%CI: 8.1, 10.5) fewerdeaths per 10,000 person-years; no statistically significant differences were identified between White and Hispanic/Latinx men. Notably, these differences in mortality were partially explained by both differential exposure and differential vulnerability to these lifestyle factors among Black women, while different effects of individual lifestyle factors canceled each other out among Black men and Hispanic/Latinx women. Conclusions: Lifestyle factors provide some explanation for racial and ethnic inequalities in all-cause mortality. Greater attention to structural, life course, healthcare, and other factors is needed to understand determinants of inequalities in mortality and advance health equity.


Background
Long-standing and stark racial and ethnic inequalities in health and mortality are widespread in the United States (US) [1,2].It is established that mortality rates among Black Americans are higher throughout most of the life course, relative to White Americans [2,3].In contrast, mortality rates among Hispanic/Latinx Americans are lower despite lower socioeconomic status (SES), on average, relative to White Americans [4].In recent decades, research has focused on delineating the causes and etiology of racial and ethnic inequalities in mortality.A multitude of factors and pathways have been proposed and evaluated including societal in uences (e.g., government policies) [5], environmental and occupational hazards (e.g., residential segregation) [6,7], individual-level factors (e.g., SES, lifestyle factors, health insurance, and access to quality health care) [1,2,[8][9][10], genetic factors [11], and potential biases in study designs, such as those related to selective migration (e.g., the salmon bias) [12,13].However, the complex and interrelated relationships and pathways in which these variables affect health and mortality have not been systematically evaluated and a large proportion of the observed racial and ethnic inequalities remains unexplained.
Lifestyle factors (such as smoking, alcohol use, physical inactivity, and obesity) are an important driver of inequalities in health, explaining, for example, more than two thirds of the association between SES and all-cause mortality [14,15].
Although some evidence suggests that lifestyle factors are important in explaining racial and ethnic disparities [2,3,16,17], no studies have used a comprehensive approach that evaluates multiple lifestyle factors together and their mediating and/or moderating effects.Evaluating multiple lifestyle factors is important as lifestyle factors may cluster together in distinct ways that vary by race and ethnicity [18,19].Understanding the mediating or moderating effects is essential in delineating potential mechanisms such as differential exposure, whereby health-promoting or unhealthy lifestyle factors are unevenly distributed across racial and ethnic groups (a mediation hypothesis), and differential vulnerability, whereby the same lifestyle factor can be more deleterious to speci c racial and ethnic groups (a moderation hypothesis).Disentangling these two mechanisms is important given that unique policy implications can arise from them [20,21].
Overall, the extent and means by which lifestyle factors might explain racial and ethnic disparities is largely unknown.
Using a comprehensive model (Fig. 1) and a large cohort from the US, the current study aims to delineate the extent to which racial and ethnic differences in all-cause mortality can be explained by (i) differential exposure to lifestyle factors, and (ii) differential vulnerability to the harmful effects of each lifestyle factors across different race and ethnicity groups.The lifestyle factors considered were smoking, alcohol use, physical activity, and body mass index (BMI).

Participants
Data came from the National Health Interview Survey (NHIS) linked to the National Death Index (NDI) using probabilistic record matching [28].NHIS is an annual, nationally representative, cross-sectional household survey of the civilian noninstitutionalized US population.NHIS utilized a complex, multistage sample design that involved strati cation, clustering, and oversampling of speci c population subgroups.Every year approximately 35,000 households are enrolled, from which one adult is randomly selected for a face-to-face interview.An annual assessment of all lifestyle factors in su cient detail started in 1997, and NHIS data up to 2014 have been linked to the NDI.Therefore, this study included pooled NHIS data from 1997 to 2014.The NDI contains information on vital status, time of death, and time last presumed alive with follow-up to December 31, 2015.Our sample was comprised of the adults (ages ≥ 18 years) who were not missing data on the exposure, mediators, outcome, and covariates; those with complete and missing data were largely similar across a range of characteristics (Supplementary Table S1).Participants over 85 years of age at the time of NHIS administration were removed given that their exact age was not available through the public use data les.

Measures
The outcome was time to all-cause mortality, operationalized as the time from the NHIS survey to death or last presumed alive.Race and ethnicity, the independent variable of interest, was self-reported and categorized as non-Hispanic White (reference category; henceforth White), non-Hispanic Black/African American (henceforth Black), and Hispanic/Latinx.We further distinguished all other non-Hispanic racial and ethnic groups (hereafter, non-Hispanic Other) for descriptive analyses, though sample size was too small for inclusion in the main analyses.
Participants' report of the frequency and quantity of alcoholic beverage consumed in the past 12 months was converted to grams of pure alcohol consumed per day, assuming 14 grams of pure alcohol per drink.Alcohol use was categorized according to the standards of the World Health Organization [29] and included: 1) never drinkers (no drinks in the past year and less than 12 drinks in any one year or entire life), 2) former drinkers (no drinks in the past year but have had at least 12 drinks in any one year), 3) category I (men: (0, 40] grams per day; women: (0, 20] grams per day; reference category), 4) category II (men: (40,60] grams per day; women: (20,40] grams per day), 5) category III (men: >60 grams per day; women: >40 grams per day).With respect to smoking, participants were asked to report whether they 1) have smoked at least 100 cigarettes over their entire life, and 2) whether they currently smoked cigarettes.Smoking cigarettes was categorized as never smokers (reference category), former smokers, current someday smokers, and current everyday smokers.Based on self-reported height and weight, BMI was calculated and categorized according to current WHO guidelines as underweight (< 18.5kg/m 2 ), normal weight (18.5-24.99kg/m 2 ; reference category), pre-obesity (25-29.99kg/m 2 ) or obese (≥ 30kg/m 2 ) [30].With respect to physical activity, participants reported how often and for how long they performed vigorous and light-moderate leisure-time physical activities of at least 10 minutes.No timeframe (e.g., over the past year, or past month) was speci ed.The length of moderate physical activity per week was calculated, assuming that 1 minute of vigorous physical activity is equivalent to 2 minutes of moderate physical activity [31].
Physical activity was categorized as sedentary (0 minutes/week), somewhat active (< 150 minutes) or active (≥ 150 minutes; reference category), given the WHO recommendations of 150-300 minutes of moderate-intensity physical activity per week [32].
The covariates used in all models were age (continuous), sex, educational attainment, marital status, and survey year (continuous).Educational attainment was categorized as low (high school diploma or less), medium (some college but no bachelor's degree), or high (bachelor's degree or more), and was treated as a proxy for socioeconomic status; given its ubiquity in the extant literature, stability over time, and completeness of data (e.g., relative to income) in the NHIS.
Marital status was a binary variable indicating whether the individual was married or living with partner.

Statistical Analyses
Causal mediation analysis using the marginal structural approach with Aalen's additive hazard models was used, as described by Lange et al. [33][34][35].Brie y, this exible approach uses a counterfactual framework and allows for the direct parameterization of natural 'direct' and 'indirect' effects through multiple mediators and exposure-mediator interactions.The total effect of race and ethnicity on mortality was decomposed into three components (Fig. 1): 1) the average pure indirect effect through each mediator (indicating differential exposure), 2) the average indirect effect of the mediated interaction between race and ethnicity and each mediator (indicating differential vulnerability), and 3) the average 'direct' effect of race and ethnicity independent of mediators and covariates.The model simultaneously included all mediators (lifestyle factors: alcohol use, smoking, BMI, physical activity) and covariates (age, educational attainment, marital status, and survey year), and we t separate models for men and women.Aalen's additive hazard models have the advantage of directly estimating additive interactions (re ecting differential vulnerability), which are of greater importance (relative to multiplicative interactions) for public health [36].
All analyses were completed in R 4.1.3,using the timereg package (version 2.0.2) [37]; the statistical code is publicly available (see below).The timereg package does not allow for complex sampling designs and survey weights were not utilized given the analytical and computational complexity of the analyses.
In a sensitivity analysis, causal mediation models were repeated without education included as a covariate, recognizing that race and ethnicity are deeply tied to SES in the US [38], and prior research shows SES differences in effects of lifestyle factors on mortality [14,15].

Results
Participants were 465,073 adults (55% women, mean age 46.4 years [SD 17.3]), of whom 63% were non-Hispanic White, 15% non-Hispanic Black, 17% Hispanic/Latinx, and 5% non-Hispanic Other (of whom 12% were AI/AN and 53% API; see Table 1).Participants were followed an average of 8.9 years (SD 5.3) during which 49,804 deaths were observed.At the time of survey completion, 31% had never drank alcohol, 57% had never smoked, 37% had a healthy weight, and 44% were physically active.Relative to White adults, the prevalence of category II and III alcohol use and everyday smoking were lower among Black, Hispanic/Latinx, and non-Hispanic Other adults (Fig. 2).The opposite pattern was observed for obesity and sedentary physical activity, with a higher prevalence among Black and Hispanic/Latinx compared to White adults.Figure 3 presents the overall survival probability as a function of age, with the median survival probability being markedly lower in Black women (81 years, 95% con dence intervals [CI]: 80.5, 81.5) and men (74.8 years, 95%CI: 74.0, 75.5) than for other racial and ethnic groups (women: 84.5-86.5 years, men: 78.2-81.8years).
However, Black men and Hispanic/Latinx adults were also more vulnerable to the adverse effects of smoking, which resulted in 2.2 to 11.7 additional deaths per 10,000 person years.The opposite pattern was observed for physical activity, nding that greater exposure to a sedentary lifestyle was associated with 4.6 to 7.8 additional deaths per 10,000 person years among Black and Hispanic/Latinx adults, relative to White adults, and that Black women and Hispanic/Latinx adults were less vulnerable to the adverse effects of physical inactivity, resulting in 1.7 to 5.0 fewer deaths per 10,000 person years.With respect to alcohol use, Hispanic/Latinx men were similar to White men.Among Black adults and Hispanic/Latinx women, exposure to alcohol use was associated with 1.4 to 5.1 additional deaths per 10,000 person-years, relative to White men and women.But this effect was partially offset by a greater resilience (differential vulnerability) to the adverse effects of alcohol use in these groups.Lastly, with respect to BMI, differential exposure and vulnerability effects were relatively small and offset each other.
Notably, the net indirect effect through lifestyle factors was not signi cant among Black men and Hispanic/Latinx women, and did not contribute overall to racial and ethnic inequalities in these racial and ethnic groups; different levels of physical activity were associated with additional deaths, whereas a lower prevalence of smoking was associated with fewer deaths.The differences in mortality among these groups, and to a large extent among Black women and Hispanic/Latinx men too, were attributed to the effect of race and ethnicity itself which, as stated earlier, could re ect differences in unobserved life course, structural, environmental and/or other factors in uencing mortality independent of lifestyle factors and covariates.The results of the sensitivity analysis excluding education as covariate were consistent with our main analysis and did not change our conclusions (Supplementary Table S2).

Discussion
The current study sought to evaluate the mechanism and extent to which lifestyle factors contribute to racial and ethnic inequalities in mortality among US adults.Speci cally, we examined whether these inequalities can be explained by indirect effects through differential exposure and differential vulnerability to harmful effects of different lifestyle factors.
First, and consistent with the extant literature [2][3][4], we found that relative to White adults, mortality rates were higher for Black men and women, and lower for Hispanic/Latinx women.Our key nding and the novel contribution of this study was that mechanisms of differential exposure and vulnerability to lifestyle factors help to explain the disparity in mortality rates between White and Black women, and the equivalent mortality rates of White and Hispanic/Latinx men.This was, however, not the case for Black men and Hispanic/Latinx women.In other words, lifestyle factors cannot explain the observed racial and ethnic inequalities in all-cause mortality in the latter groups.This is because the net indirect effect of race and ethnicity through differential exposure and vulnerability to lifestyle factors did not contribute to the observed inequalities among Black men and Hispanic/Latinx women given individual indirect effects canceled each other out.Speci cally, additional deaths among Black men and Hispanic/Latinx women were attributed to a higher exposure to sedentary physical activity, while a lower prevalence of smoking resulted in fewer deaths, relative to White men and women.This nding that particularly highlights the differential exposure to different lifestyle factors across racial and ethnic groups is consistent with past studies [3,16,39,40].The results of the current study help to advance the extant literature through our use of a comprehensive model to decompose the effects of differential exposure and vulnerability.Our results suggest that public health interventions targeting physical inactivity among Black and Hispanic/Latinx adults are important.However, targeting lifestyle factors alone, without consideration of more fundamental forces, such as poverty, structural racism, and limited opportunity [41], will not likely improve racial and ethnic disparities in mortality observed for Black men and women.
Our ndings for the somewhat limited role of lifestyle factors in explaining racial and ethnic inequalities in mortality stand in contrast to research on socioeconomic disparities in mortality, which report that the latter inequalities are largely attributed to the net indirect effect of lifestyle factors [14,15].This difference in ndings may be because lifestyle factors putting individuals at higher health risks were found to cluster among low SES groups [14,15], in contrast to our nding of a lower prevalence of smoking among Black and Hispanic/Latinx adults which resulted in a relative protective effect.Taken together, these ndings have important public health implications in highlighting that socioeconomic and racial and ethnic inequalities in mortality in the US may arise in unique ways (e.g., racial residential segregation is likely more relevant to the Black-White mortality gap) and likely require distinctive intervention approaches.Even so, past studies have shown that SES is an important mediator of racial and ethnic inequalities in mortality [3,16,42], and reducing socioeconomic inequalities in mortality potentially by targeting the root causes of socioeconomic health inequalities may in turn also reduce racial and ethnic disparities.
In interpreting the results presented above, limitations should be considered.First, the choice of covariates is important given that causal mediation models assume no unmeasured confounders for the exposure-outcome, exposure-mediator, and mediator-outcome relationships, and no mediator-outcome confounders caused by the exposure.Mediators are

Figures Figure 1
Figures

Table 1
Participant characteristics, strati ed by sex and race/ethnicity.

Table 2
presents the results of the causal mediation analyses, controlling for all covariates and lifestyle factors.Relative to White adults, Black adults experienced 21.7 (men; 95%CI 19.9, 23.5) and 11.5 (women; 95%CI 10.1, 12.9) additional deaths per 10,000 person-years, whereas Hispanic/Latinx women experienced 9.3 (95%CI 8.1, 10.5) fewer deaths per 10,000 person-years after adjusting for covariates.Mortality was similar among White and Hispanic/Latinx men, after adjusting for covariates.

Table 2
Results of causal mediation analyses, evaluating the extent to which the association between race and ethnicity with all-cause mortality was mediated by lifestyle factors.