key: cord-0807255-pfc31r4o authors: Zhang, Yan-Bo; Chen, Chen; Pan, Xiong-Fei; Guo, Jingyu; Li, Yanping; Franco, Oscar H; Liu, Gang; Pan, An title: Associations of healthy lifestyle and socioeconomic status with mortality and incident cardiovascular disease: two prospective cohort studies date: 2021-04-14 journal: BMJ DOI: 10.1136/bmj.n604 sha: d7bd7d28bac9f04917d78706f97d7e835fc904d2 doc_id: 807255 cord_uid: pfc31r4o OBJECTIVE: To examine whether overall lifestyles mediate associations of socioeconomic status (SES) with mortality and incident cardiovascular disease (CVD) and the extent of interaction or joint relations of lifestyles and SES with health outcomes. DESIGN: Population based cohort study. SETTING: US National Health and Nutrition Examination Survey (US NHANES, 1988-94 and 1999-2014) and UK Biobank. PARTICIPANTS: 44 462 US adults aged 20 years or older and 399 537 UK adults aged 37-73 years. EXPOSURES: SES was derived by latent class analysis using family income, occupation or employment status, education level, and health insurance (US NHANES only), and three levels (low, medium, and high) were defined according to item response probabilities. A healthy lifestyle score was constructed using information on never smoking, no heavy alcohol consumption (women ≤1 drink/day; men ≤2 drinks/day; one drink contains 14 g of ethanol in the US and 8 g in the UK), top third of physical activity, and higher dietary quality. MAIN OUTCOME MEASURES: All cause mortality was the primary outcome in both studies, and CVD mortality and morbidity in UK Biobank, which were obtained through linkage to registries. RESULTS: US NHANES documented 8906 deaths over a mean follow-up of 11.2 years, and UK Biobank documented 22 309 deaths and 6903 incident CVD cases over a mean follow-up of 8.8-11.0 years. Among adults of low SES, age adjusted risk of death was 22.5 (95% confidence interval 21.7 to 23.3) and 7.4 (7.3 to 7.6) per 1000 person years in US NHANES and UK Biobank, respectively, and age adjusted risk of CVD was 2.5 (2.4 to 2.6) per 1000 person years in UK Biobank. The corresponding risks among adults of high SES were 11.4 (10.6 to 12.1), 3.3 (3.1 to 3.5), and 1.4 (1.3 to 1.5) per 1000 person years. Compared with adults of high SES, those of low SES had higher risks of all cause mortality (hazard ratio 2.13, 95% confidence interval 1.90 to 2.38 in US NHANES; 1.96, 1.87 to 2.06 in UK Biobank), CVD mortality (2.25, 2.00 to 2.53), and incident CVD (1.65, 1.52 to 1.79) in UK Biobank, and the proportions mediated by lifestyle were 12.3% (10.7% to 13.9%), 4.0% (3.5% to 4.4%), 3.0% (2.5% to 3.6%), and 3.7% (3.1% to 4.5%), respectively. No significant interaction was observed between lifestyle and SES in US NHANES, whereas associations between lifestyle and outcomes were stronger among those of low SES in UK Biobank. Compared with adults of high SES and three or four healthy lifestyle factors, those with low SES and no or one healthy lifestyle factor had higher risks of all cause mortality (3.53, 3.01 to 4.14 in US NHANES; 2.65, 2.39 to 2.94 in UK Biobank), CVD mortality (2.65, 2.09 to 3.38), and incident CVD (2.09, 1.78 to 2.46) in UK Biobank. CONCLUSIONS: Unhealthy lifestyles mediated a small proportion of the socioeconomic inequity in health in both US and UK adults; therefore, healthy lifestyle promotion alone might not substantially reduce the socioeconomic inequity in health, and other measures tackling social determinants of health are warranted. Nevertheless, healthy lifestyles were associated with lower mortality and CVD risk in different SES subgroups, supporting an important role of healthy lifestyles in reducing disease burden. Family income level, occupation, education level, and health insurance reflect different aspects of socioeconomic status (SES), thus we used these four parameters to generate an overall SES parameter in the US National Health and Nutrition Examination Survey (NHANES). There were three levels for each parameter, i.e., the family income-to-poverty ratio of ≥4, >1 to <4, and ≤1 for family income level; upper socioeconomic index, lower socioeconomic index, and unemployment for occupation; college or above, high school or equivalent, and less than high school for education level; and private health insurance, public health insurance only, and no health insurance for health insurance. Latent class analyses with different numbers of latent classes were conducted to select a reasonable model. The maximum absolute deviation between the parameter estimates in two successive iterations of the estimation procedure was set to 0.000001, which meant iteration would terminate when the difference between the parameter estimates in two successive iterations was less than 0.000001. Akaike information criterion (AIC), Bayesian information criterion (BIC), and likelihood ratio statistic G 2 were used for the model selection. The mean posterior probability, which reflected the uncertainty of posterior classification, was also used for the model selection, and a value of 0.7 or more indicated an acceptable uncertainty. Item-response probability was a posterior probability and was used for defining latent classes. Since the model with six latent classes failed to converge, we only reported information on models with five or fewer latent classes. The following figure shows that G 2 statistics, AIC, and BIC all continued to go down as more latent classes were added. However, the decrease leveled off after the three-latent-class solution. We additionally examined the mean posterior probabilities to facilitate the model selection. The following table shows the mean posterior probabilities, the prevalence of latent classes, and item-response probabilities in models with three to five latent classes. All mean posterior probabilities from the three-latent-class solution were all ≥0.77; mean posterior probabilities of latent classes 2 and 4 from the four-latent-class solution were respectively 0.68 and 0.64 which were less than 0.70; and the mean posterior probabilities of four latent classes from the five-latent-class solution were less than 0.70. Thus, the three-latent-class solution was the best in terms of the uncertainty of posterior classification. * The maximal item-response probabilities for each latent class were marked in bold. † Prevalence indicated the prevalence of each latent class. Income 1 to 3 respectively referred to the family income-to-poverty ratio of ≥4, >1 to <4, and ≤1. Occupation 1 to 3 respectively referred to upper socioeconomic index, lower socioeconomic index, and unemployment. Education 1 to 3 respectively referred to college or above, high school or equivalent, less than high school. Insurance 1 to 3 respectively referred to private health insurance, public health insurance only, and no health insurance. Additionally, we evaluated the characteristics of each latent class in each model. For the three-latent-class solution, latent class 1 was characterized by high-level family income, occupation, education, and health insurance, which could be defined as "high SES"; latent class 2 was characterized by medium-level family income, occupation, and education, as well as high-level health insurance, which could be defined as "medium SES"; latent class 3 was characterized by medium-and low-level family income, occupation, and health insurance, as well as low-level education, which could be defined as "low SES". As for the four-latent-class solution, the latent classes 1 to 3 were correspondingly similar to those in the three-latent-class solution, while latent class 4 was characterized by medium-level family income and occupation as well as high-level education and health insurance, which could be defined as "early high SES" (considering education levels could reflect ones' SES before early adulthood while occupation and income tended to reflect ones' SES in adulthood). As for the five-latent-class solution, the latent classes 1 to 3 were correspondingly similar to those in the three-latentclass solution, while latent class 4 was characterized by medium-level family income and occupation as well as low-level education and health insurance, which could be defined as "early low SES"; while latent class 5 was characterized by medium-level family income, high-level occupation and education, and low-level health insurance, which could be defined as "high socioeconomic prestige". Considering we intended to compare mortality risks among individuals with different SES (especially comparing mortality risk among individuals with high versus low SES), sufficient sample size and events were needed among each group; however, the prevalence of "low SES" class was 12% and 7% in the four-latent-class solution and five-latent-class solution, which were relatively low. Besides, latent class 4 or 5 in the four-latent-class solution and five-latent-class solution was isolated from the medium and low SES, which was not of interest to us. Above all, comprehensively considering statistics related to model selection, the uncertainty of posterior classification, meanings of latent classes, and parsimony, we chose the three-latent-class solution and divided individuals into high, medium, and low SES, and the practical definitions of which were shown in the following figure. PIR=family income-to-poverty ratio; SES=socioecnomic status. The numbers in the cells represented the percentage of participants out of the total study population. Total household income before tax, education qualification, and employment status were used to generate an overall SES parameter in the UK Biobank. We did not consider health insurance in the UK since the National Health Service is implemented in the UK, which aims to provide comprehensive, universal and free health services. We did not regroup household income and education qualification into three groups as we did in the US NHANES because of the larger sample size in the UK Biobank and failure of model convergence due to fewer observed groups if the two variables were regrouped. Participants were divided into five groups according to total household income before tax, i.e., "less than ₤18 000", "₤18 000 to 30 999", "₤31 000 to 51 999", "₤52 000 to 100 000", "greater than ₤100 000".There are seven groups according to education qualification, including "College or University degree", "A levels/AS levels or equivalent", "O levels/GCSEs or equivalent", "CSEs or equivalent", "NVQ or HND or HNC or equivalent", "Other professional qualifications", "None of the above" (equivalent to less than high school diploma). The UK Biobank only acquired employment status instead of specific occupation information at baseline, and we regrouped participants into two groups, i.e., employed (including those in paid employment or self-employed, retired, doing unpaid or voluntary work, or being full or part-time students), and unemployed. The procedure of the latent class analysis was similar to that in the US NHANES. Since the model with four latent classes failed to converge, we only reported parameters in the threelatent-class solution. In the three-latent class solution, the G 2 statistic is 2391, AIC is 2461, and BIC is 2845. Mean posterior probabilities, the prevalence of latent classes, and item-response probabilities in the three-latent class solution are shown below. As shown above, the proportion of less than ₤31 000 of total household income before tax, O/GCSEs level and less than high school (i.e., none of the above), and unemployment were relatively high in latent class 1, which could be defined as "low SES". The proportion of ₤52 000 or more of total household income before tax, college or university degree, and employment were relatively high in latent class 2, which could be defined as "high SES". ₤18 000-51 999 of total household income before tax, college or university degree and O/GCSEs level, and employment status were prevalent in latent class 3, which could be defined as "medium SES". Mean posterior probabilities of all latent classes were above 0.70, and the practical definitions of which were shown in the following figure. The numbers in the cells represented the percentage of participants out of the total study population. In the US NHANES, dietary information was obtained through 24-hour dietary recalls, and other information on socioeconomic and lifestyle factors was obtained through questionnaires. All questionnaires could be obtained through the US NHANES website (https://wwwn.cdc.gov/nchs/nhanes/default.aspx). • Family income level: Family income-to-poverty ratio which is a ratio of family income to poverty threshold is provided in the dataset. We regrouped participants into three groups (i.e., family income-to-poverty ratio of ≥4, >1 to <4, and ≤1) according to previous studies. • Education attainment: The highest grade or level of school completed or the highest degree received was asked for each participant. Participants could choose one of the following options: less than 9th grade, 9-11th grade (including 12th grade with no diploma), high school graduates or general educational development or equivalent, some college or associate degree of Arts, and college graduate or above. We regrouped participants into three groups (i.e., less than high school, high school graduates, and some college or above) according to previous studies. Indexes of Industry and Occupation (US Census Bureau). Main reasons for not working last week were also obtained, including taking care of house or family, going to school, retired, unable to work for health reasons, on layoff, disabled, and others. Except for going to school and retired, other reasons for not working denoted unemployment. Participants were regrouped into three groups, including upper socioeconomic index, lower socioeconomic index, and unemployment. • Health insurance: Participants were asked about whether they had any private health insurance, Medi-Gap, single-service plan, Medicare, Medicaid, State Children's Healthcare Plan, military health care, Indian Health Service, State Sponsored Health Plan, or other government program. We regrouped participants into three groups, i.e., those with private health insurance, those with public health insurance, and those with no health insurance. • Cigarette smoking: Participants were asked about whether they have smoked over 100 cigarettes in life, and we defined those who have smoked over 100 cigarettes in life as ever smokers. • Alcohol drinking: From 1988 to 1994, participants were asked about the monthly frequency of consuming beer and lite beer, wine, and hard liquor, and we assumed that each participant drank one drink per time. Thus, daily drinks that each participant consumed equaled to the sum of daily frequency of different alcoholic drink intakes. Since 1999, participants were asked about the frequency of alcohol drinking over the past 12 months and average alcoholic drink consumptions per day on those days that they drank. We calculated the daily consumption of alcohol by multiplying the probability of drinking alcohol on a given day and daily consumption of alcohol. • Physical activity: We assessed the level of leisure-time physical activity since information on physical activity during work was not collected in some cycles and thusly we cannot estimate the level of total physical activity in all cycles. From 1988 to 1994, participants were asked about the monthly frequency of walking mile without stop, running or jogging, riding or exercising bicycle, swimming, doing aerobics or aerobic dancing, doing other dancing, doing calisthenics, doing garden or yard work, lifting weights, and doing other activities. Intensity rating was given for each kind of activity, and information on the duration of each activity per time was unavailable. Thus, we added the monthly frequency of all leisure-time physical activities up weighted by their intensity rating. From 1999 to 2006, participants were asked about the monthly frequency of multiple leisure-time physical activities (details are shown at https://wwwn.cdc.gov/Nchs/Nhanes/1999-2000/PAQIAF.htm#Data_Processing_and_Editing). For each activity, average number of minutes spent each time and metabolic equivalent score were also obtained. Since 2007, the dataset provided information on how many days and how much time per day did participants do recreational moderate and vigorous physical activity directly, and 4 and 8 of metabolic equivalent scores were given to moderate and vigorous physical activity. We calculated metabolic equivalent times for each participant by adding time spent on each activity weighted by its metabolic equivalent score from 1999 to 2014. • Diet: Dietary information was obtained through 24-h dietary recalls, and an observation validation study showed that the 24-h dietary recall could accurately estimate intakes of energy and macronutrients. Before 2003, only one 24-h dietary recall was administered for each participant in the mobile examination center. Since 2003, two 24-h dietary recalls were administered to each participant. We only used recalls from the first day to harmonize the data from all cycles. The dietary interviews were conducted by trained investigators following the US Department of Agriculture Automated Multiple-Pass Method for the 24-h recall. First, investigators asked respondents about what they consumed yesterday, which could match to foods from the Main Food List which contained more than 2600 food items. Then, some specific categories of foods that were frequently forgotten including fruits, vegetables, cheese, bread, sweets, snacks, nonalcoholic beverages, and alcoholic beverages were additionally asked. Next, eating occasions and time, food descriptions, and food amounts of each food were obtained. Consumptions of food groups and nutrients were determined using the US Department of Agriculture Nutrient Database for Dietary Studies and Food Patterns Equivalents Database. Components and scoring standards of Healthy Eating Index score-1995 and Healthy Eating Index score-2015 are shown in supplementary table 1. In the UK Biobank, all information on socioeconomic and lifestyle factors was obtained through questionnaires. All questionnaires could be obtained through the UK Biobank website (https://biobank.ndph.ox.ac.uk/showcase/index.cgi). • Family income level: Total household income before tax was obtained through questionnaires, and participants could choose an option from "less than ₤18 000", "₤18 000 to 30 999", "₤31 000 to 51 999", "₤52 000 to 100 000", "greater than ₤100 000", "do not know", or "prefer not to answer". • Education attainment: Education qualification was obtained through questionnaires, and participants reported their education qualifications as "College or university degree", "A levels/AS levels or equivalent", "O levels/GCSEs or equivalent", "CSEs or equivalent", "NVQ or HND or HNC or equivalent", "Other professional qualifications", "None of the above" (equivalent to less than high school diploma), or "Prefer not to answer". • Employment status: The UK Biobank only acquired employment status instead of specific occupation information at baseline through questionnaires, and participants could report their employment status as in paid employment or self-employed, retired, doing unpaid or voluntary work, being full or part-time students, looking after home/family, unable to work because of sickness or disability, unemployed, none of the above, or prefer not to answer. Those who choose the first four options were grouped into the "employed" group, while others (except for the last two options, which were treated as missing values) were grouped into the "unemployed" group. • Cigarette smoking: Participants were asked about their current tobacco smoking status, including yes (on most or all days), only occasionally, no, and prefer not to answer. Those who did not smoke on most or all days would be asked about the past tobacco smoking status, including smoked on most or all days, smoked occasionally, just tried once or twice, never smoked, or prefer not to answer. Those who reported that they smoked occasionally or just tried once or twice would be asked about whether they had smoked a total of at least 100 times in lifetime. According to the information, participants who never smoked and those who previously smoked occasionally or just tried once or twice but did not smoke 100 times in lifetime were categorized into "never smoking" group, similar to the US NHANES. Others with no missing information on smoking would be viewed as ever smokers. • Alcohol drinking: Participants were asked about the frequency of drinking alcohol, i.e., (almost) daily, three or four times a week, once or twice a week, one to three times a month, special occasions only, never, and prefer not to answer. Those who reported to drink alcohol would be asked about how much red wine (glasses), white wine (glasses), beer or cider (pints), spirits or liqueurs (standard measures), fortified wine (glasses), and other alcoholic drinks (glasses) they consumed in an average month or week. We used the information to calculate the average units of alcohol each participant drank daily. • Physical activity: To coincide with the US NHANES, we assessed the level of leisure-time physical activity in the UK Biobank. We took walking for pleasure, strenuous sports, and other exercises (e.g., swimming, cycling, keep fit, bowling) into consideration, which were given 3.3, 8, and 4 metabolic equivalent scores, respectively. The frequency and duration for each time were asked for each participant. Participants could choose one of the following frequency options, i.e., once in the last 4 weeks, 2-3 times in the last 4 weeks, once a week, 2-3 times a week, 4-5 times a week, every day, do not know, and prefer not to answer. Options including a range would be substituted by the midpoint of the range, e.g., we assigned 2.5 times a week for the option "2-3 times a week". The duration for each time was also obtained by some options, i.e., less than 15 minutes, between 15 and 30 minutes, between 30 minutes and 1 hour, between 1 and 1.5 hours, between 1.5 and 2 hours, between 2 and 3 hours, over 3 hours, do not know, and prefer not to answer. Each option would be substituted by the midpoint of the range, and those who chose over 3 hours were substituted by 3 hours as recommended by other studies. We calculated metabolic equivalent times for each participant by adding time spent on each activity weighted by its metabolic equivalent score. were evaluated according to the frequency of consumption of poultry, beef, lamb/mutton, and pork and whether the participants did not eat meat anymore (according to a question about age when last ate meat). Sugar-sweetened beverage intakes were evaluated by a question "Which of the following do you NEVER eat?" Those who chose sugar or foods/drinks containing sugar were regarded as never drinking sugarsweetened beverages. In the US NHANES, a 5-point acculturation score was constructed according to the country of birth, length of time in the US, and language spoken at home. A 3-point score was assigned to the country of birth and length of time in the US, i.e., 3 points for US-born, 2 points for foreign-born and lived in the US ≥20 years, 1 point for foreign-born and lived in the US 10 to 19 years, and 0 points for foreign-born and lived in the US <10 years. A 2-point score was assigned to language spoken at home, i.e., 2 points for English only or predominantly, 1 point for both equally, and 0 points for other languages only or predominantly. These scores were summed to get a 5point acculturation score with greater values indicating more acculturated, and participants were grouped into two groups (i.e., 0-2 points and 3-5 points). In the UK Biobank, information on language spoken at home was not collected. Thus, the acculturation score was constructed according to the country of birth and length of time in the UK, and the score ranged between 0 and 3, i.e., 3 points for UK-born, 2 points for foreign-born and lived in the UK ≥20 years, 1 point for foreign-born and lived in the UK 10 to 19 years, and 0 points for foreign-born and lived in the UK <10 years. Participants were grouped into two groups, i.e., 0-1 point and 2-3 points. *Intakes between the minimum and maximum standards are scored proportionately. †Depends on different energy intake. ‡The required number of milk group servings is 3 for pregnant and breast-feeding women and for teenagers and young adults up to age 24 years. §Includes 100% fruit juice. ||Includes all forms except juice. ¶Includes legumes (beans and peas). **Includes all milk products, such as fluid milk, yogurt, and cheese, and fortified soy beverages. † †Includes legumes (beans and peas). ‡ ‡Includes seafood, nuts, seeds, soy products (other than beverages), and legumes (beans and peas). *Socioeconomic status was generated through latent class analysis using the information on family income-to-poverty ratio, occupation, education level, and health insurance in the US NHANES, and household income, education level, and employment status in the UK *Hazard ratios (95% confidence intervals) comparing those with low socioeconomic status and 0-1 healthy lifestyle factor versus those with high socioeconomic status and 3-4 healthy lifestyle factors are shown. Socioeconomic status was generated through latent class analysis using the information on family income-to-poverty ratio, occupation, education level, and health insurance in the US NHANES, and household income, education level, and employment status in the UK Biobank. Hazard ratios were adjusted for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary disorder). Only those free from CVD at baseline were included in the analysis for incident CVD. Analysis in US NHANES included the US population and study design weights to account for the complex survey design. †Participants were equally divided into three groups according to weighted healthy lifestyle score. *Hazard ratios (95% confidence intervals) comparing those with low socioeconomic status and 0-1 healthy lifestyle factor versus those with high socioeconomic status and 3-4 healthy lifestyle factors are shown. Socioeconomic status was generated through latent class analysis using the information on family income-to-poverty ratio, occupation, education level, and health insurance in the US NHANES, and household income, education level, and employment status in the UK Biobank. Only the results comparing the low with high socioeoncomic status are reported. Hazard ratios were adjusted for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary disorder). For incident cardiovascular disease, only those free from cardiovascular disease at baseline were included. Analysis in US NHANES included the US population and study design weights to account for the complex survey design. †P for interaction indicated the difference of hazard ratio comparing those with low socioeconomic status and 0-1 healthy lifestyle factor versus those with high socioeconomic status and 3-4 healthy lifestyle factors between two subgroups was statistically significant or not. ‡Data is not shown since limited events in extreme groups. Flowchart of the study 28 Supplementary fig 2. Individual-level socioeconomic status, Townsend Deprivation Index, lifestyles, and primary outcomes in the UK Biobank CI=confidence interval; COPD=chronic obstructive pulmonary disorder; CVD=cardiovascular disease; HR=hazard ratio; US NHANES= the US National Health and Nutrition Examination Survey Socioeconomic status was generated through latent class analysis using the information on family income-to-poverty ratio, occupation, education level, and health insurance in the US NHANES, and household income, education level, and employment status in the UK Biobank. Only the results comparing the low with high socioeoncomic status are reported. Model 1 controlled for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary disorder). Model 2 additionally included the healthy lifestyle score consisting of never smoking, no heavy alcohol drinking Analysis in US NHANES included the US population and study design weights to account for the complex survey COPD=chronic obstructive pulmonary disorder; CVD=cardiovascular disease; US NHANES= the US National Health and Nutrition Examination Survey Socioeconomic status was generated through latent class analysis using the information on family incometo-poverty ratio, occupation, education level, and health insurance in the US NHANES, and household income, education level, and employment status in the UK Biobank. Hazard ratios were adjusted for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary disorder). Only those free from CVD at baseline were included in the analysis for incident CVD CI=confidence interval; CVD=cardiovascular disease; HR=hazard ratio; US NHANES= the US National Health and Nutrition Examination Survey Model 2 additionally included the healthy lifestyle score consisting of never smoking, no heavy alcohol drinking, higher physical activity level, and a higher diet quality score. Analysis in US NHANES included the US population and study design weights to account for the complex survey design. Hazard ratios (HRs) were adjusted for age, sex, self-reported race, acculturation, study center Hazard ratios (HRs) were adjusted for other socioeconomic factors, age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease For incident cardiovascular disease and cancer, only those free from cardiovascular disease and cancer at baseline were included. Error bars indicate 95% confidence intervals (CIs) Associations of socioeconomic status with mortality and incident cardiovascular disease and mediation proportion of socioeconomic inequity in health attributed to lifestyle: subgroup analyses* All-cause mortality All-cause mortality CVD mortality Incident CVDIn the US National Health and Nutrition Examination Survey (US NHANES), models included US population and study design weights to account for the complex survey design. Hazard ratios (HRs) were adjusted for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary disorder). Only those free from cardiovascular disease at baseline were included in the analysis for incident myocardial infarction and stroke. Error bars indicate 95% confidence intervals (CIs). Multiplicative interaction was evaluated using HR for the product term between the healthy lifestyle score (0-1 point versus 3-4 points) andsocioeconomic status (low versus high), and the multiplicative interaction was statistically significant when its CI did not include 1. Additive interaction was evaluated using relative excess risk due to interaction (RERI) between the healthy lifestyle score (0-1 point versus 3-4 points) and socioeconomic status (low versus high), and the additive interaction was statistically significant when its CI did not include 0. In the US National Health and Nutrition Examination Survey (US NHANES), models included US population and study design weights to account for the complex survey design. Hazard ratios (HRs) were adjusted for age, sex, marital status (US NHANES only), self-reported race, acculturation, study center (UK Biobank only), body mass index, and prevalent comorbidities (including hypertension, diabetes, cardiovascular disease, cancer, and chronic bronchitis, emphysema, or chronic obstructive pulmonary 39 disorder). Only those free from cardiovascular disease at baseline were included in the analysis for incident cardiovascular disease. Error bars indicate 95% confidence intervals (CIs). Multiplicative interaction was evaluated using HR for the product term between the healthy lifestyle score (0-1 point versus 3-4 points) and socioeconomic status (low versus high), and the multiplicative interaction was statistically significant when its CI did not include 1. Additive interaction was evaluated using relative excess risk due to interaction (RERI) between the healthy lifestyle score (0-1 point versus 3-4 points) and socioeconomic status (low versus high), and the additive interaction was statistically significant when its CI did not include 0.