The influence of lifestyle factors such as diet and physical activity and of genetic and metabolic predisposition on the incidence and progression of chronic diseases is well established (1,2). Findings on the independent effects of diet and physical activity on disease outcomes, however, are conflicting. Experts suggest that these differences lie in a lack of precision in estimating the effect of small ranges of exposures and in distortions due to measurement errors (3). Furthermore, few epidemiologic cohort studies on the etiology of chronic diseases are powerful enough to distinguish racial and ethnic determinants from socioeconomic determinants of health behaviors and disease patterns.

To help shed light on these issues, a longitudinal study is needed that recruits sufficient numbers of black* and white respondents to provide subsamples for disease endpoints that are large enough to create prediction models that include lifestyle factors. The population in such a cohort should have significantly varied dietary and other lifestyle exposures, and the study should use methods of analysis that take the challenges of this type of study into account. The Adventist Health Study-2 (AHS-2), with its anticipated 30,000 black and 70,000 non-Hispanic white Seventh-day Adventist (Adventist) participants is one such study. The study's design and large numbers, the variability in lifestyles of its target populations, and the proposed novel analytic methods are promising. This paper focuses on some preliminary baseline analyses of responses from the first 56,754 respondents recruited for AHS-2. Recruitment of the balance of the targeted 100,000 participants is in progress.

*The term black is used because the Adventist population includes descendents of African American slaves, people of Afro-Caribbean background, and more recent immigrants from a number of African countries.

Adventists are a Christian Protestant denomination that counsels its members to avoid alcohol, tobacco, and pork and recommends a vegetarian diet. Church members follow these recommendations to varying degrees. The National Institutes of Health funded two previous longitudinal studies of Adventists in California: the Adventist Mortality Study (AMS) (4,5) from 1960 to 1966 and the first Adventist Health Study (AHS-1) (6,7) from 1974 to 1988. These studies indicated that the risk for cardiovascular disease, diabetes, and most cancers was lower for Adventists than for non-Adventists and that life span was 7.3 years longer for Adventist men and 4.4 years longer for Adventist women than for men and women in the general California population (7).

About 35% of Adventists who participated in AHS-1 were lacto-ovo vegetarian (i.e., ate dairy and eggs), 3% were vegan (i.e., ate no meat, dairy, or eggs), 20% ate meat less than once a week, and the remainder ate meat at rates similar to those of non-Adventists (7). This population also has low levels of smoking (1.8%) and alcohol consumption (less than 10%) (2), practices that have been major confounders in the epidemiologic study of chronic disease. Furthermore, analysis of three 24-hour recalls from approximately 350 AHS-2 subjects in a calibration substudy found that their mean exercise levels were 50% higher than the U.S. national average but showed great individual variation. Forty-five percent ate soy protein at levels similar to those seen in China and Singapore (8,9; AHS-2, unpublished data, 2006).

A number of studies indicate that Adventists have health advantages over the general, non-Adventist population (4-6,10,11). The study that first identified the relationship between vegetarianism and the prevention of type 2 diabetes, heart disease, cancer, and overall mortality involved an Adventist population (2,12). The California study showed the prevalence of diabetes to be approximately 100% higher for non-vegetarian than for vegetarian Adventists and the prevalence of hypertension to be approximately 200% higher (7). This study also found that vegetarianism, nut consumption, normal body mass index (BMI), physical activity, and not smoking contributed substantially to the observed increase in life expectancy for Adventists over other Californians (7). Because these studies lack sufficient numbers of black respondents, however, it is unclear if and how their findings translate to black Adventists. Answering these questions could add significantly to our understanding of the relationship of lifestyle and genetics to the etiology of chronic diseases.

Heart disease, the leading cause of death in the United States (13), affects blacks disproportionately. Overall, the incidence of cardiovascular disease (CVD) is higher for black than for white women, and mortality from CVD and hypertension is higher for blacks than for whites of both sexes (14,15). For AHS-1 participants with a follow-up of 12 years, however, rates of these three conditions were considerably lower than those for blacks nationally (16). Physical activity; absence of red meat in the diet; consumption of nuts, fruits, vegetables, and whole grains; and sufficient water consumption were associated with this lower risk. Given the limited sample size of the original study, however, the degree of benefit from these factors is clearer for whites than for blacks.

Rates of morbidity and mortality from cancer, the second leading cause of death in the United States, are also generally higher for blacks than for whites (17,18), and cancer survival rates are lower for blacks than for whites for almost all cancers regardless of site and stage (18,19). Although evidence that fruit and vegetable consumption protects against many cancers is convincing (2,20), results on the effects of specific nutrients and dietary patterns such as vegetarianism remain contentious and differ by type of cancer (4-6,10,11,20-25). We expect that AHS-2, with its larger black population and pertinent information on lifestyle factors, will help elucidate the degree of protection from heart disease, cancer, and other chronic diseases that lifestyle may hold for blacks.

Blacks differed significantly (P <.001) from whites in sex (more were female), age (they were younger), marital status (fewer were currently married and more were never married, divorced or separated, or widowed), education (they were less highly educated) (Table 1). Personal income was lower for black than for white men (P <.001), but household income for the two groups did not differ significantly. Black women reported earning more than white women in personal income but having less than white women in household income. Also, more blacks than whites (73% v 58%) currently worked for pay, and blacks worked more hours per week (P <.001). Additional analyses not shown in the table indicate that black women were more likely than black men to have never married, to be divorced or separated, and not to be currently married. Because of the observed age and sex differences between black and white respondents and the well-known correlation of these variables with health, we adjusted all subsequent analyses for age and sex.

More black than white respondents reported having smoked in the past (P <.001), although current smoking rates were low for both racial groups (Table 2). A higher proportion of blacks than whites also reported having ever used alcohol (P <.001), but current use was lower for blacks than for whites (5% vs 7%, P <.001). Fewer blacks than whites (24% vs 48%) were vegetarian, and 35% of blacks reported consuming meat more than 4 days a week, whereas only 18% of whites reported this behavior (P <.001). Levels of water consumption were significantly lower for blacks than for whites (P <.001).

Overall, activity levels were higher for whites than for blacks (P <.001), and more blacks than whites (29% vs 21% of women, 25% vs 18% of men) reported low levels of activity. Although differences in levels of vigorous activity between blacks and whites did not differ significantly, men of both ethnicities reported higher levels of activity than did women. Finally, the two racial groups had significantly different BMI levels (P <.001). Forty-two percent of whites had normal weight according to their BMI; 28% of blacks were of normal weight. Obesity was more prevalent among blacks than among whites (35% vs 22%).

Blacks slept significantly less than did whites: 53% of blacks and 24% of whites slept 6 or fewer hours per night (P <.001). Blacks also reported spending more hours watching television. Overall, blacks ranked their health significantly lower than did whites P <.001): 15% of blacks and 26% of whites indicated excellent health; 21% of blacks and 12% of whites indicated fair health.

For self-reported prevalence of doctor-diagnosed diseases and conditions (Table 3), significantly more blacks than whites (P <.001) reported having a doctor's diagnosis of high blood pressure, stroke, and type 2 diabetes. Prevalences of myocardial infarction, high serum cholesterol, emphysema, and all cancers were lower for blacks than for whites (P <.001).

To further explore the substantially lower overall cancer results for black Adventists, we excluded skin cancers from the analysis and found that the lower risk remained for black women (P <.001), but not for black men (Table 4). Clearly, skin cancer was much less frequent among blacks than among whites in our study. The prevalence of prostate cancer, however, was much higher for black than for white men (OR = 1.47, P <.02).

Data not displayed in the tables show that blacks tend to become Adventists at a later age than do whites: the average age of baptism was 24.3 years for blacks and 21.2 years for whites (P <.001). Blacks were less likely than whites to identify themselves as Adventists at 15 to 25 years of age (P <.001), and black parents were less likely than white parents to have been Adventists when their children were aged 0 to 15 years (P <.001).

Our analyses allowed us to demonstrate some striking contrasts between white and black Adventists that warrant further exploration. The finding of shorter sleep duration for blacks is one of these differences. Other studies have found similar results by race for women (33) and evidence of different sleep physiology between the two racial groups (34). The link between sleep and obesity also warrants further examination. Despite our finding that the habits of black Adventists were somewhat less healthful than those of their white counterparts, their health habits on available comparison measures were considerably better than those of non-Adventist blacks: rates of smoking, drinking, and meat consumption were lower, and rates of vegetarianism and water consumption were higher (35).

The differences between black and white Adventists in self-reported disease prevalence are in line with data from other sources. Prevalences of hypertension and diabetes are known to be higher for blacks than for whites. That these disease rates are lower in our study than the comparable national rates for both blacks and whites, with the most strikingly lower rates being for black men, is noteworthy (36,37). More importantly, against expectations from the national data, the self-reported prevalence of diagnosed high serum cholesterol, emphysema, and cancer (except prostate) was actually lower for black than for white Adventists (38-40).

Although black respondents reported less healthy lifestyle practices overall than did whites, their low rates of current tobacco and alcohol use were similar to those of whites. The tendency of blacks to become Adventists at a later age than do whites may contribute to their longer lifetime use of both substances. Another possible explanation for the lifestyle differences between the two groups is that a larger proportion of white respondents are multigenerational Adventists and have a longer history of adherence to, and greater cultural support for, the recommendations of the church. Even though lifestyle behaviors were poorer for blacks than for whites in this cohort, their low current rates of smoking and alcohol use and their lifestyle behaviors, which are generally better than those of the total U.S. black population, may help explain why their health outcomes were better than those reported for blacks nationally.

Although these data should be interpreted cautiously, the power implicit in the large sample size used in the analyses of this study is strong. These results are encouraging but suggest a need for further study. For instance, the importance of spirituality in black U.S. culture and its known value as a disease mediator (41, 42) raise the question of whether this life component somehow mediates stress more for these Adventists and contributes to our finding that prevalences of a number of chronic diseases were lower for black than for white Adventists.

A number of study limitations must be considered when interpreting these findings. First, the data are self-reported, and health risk behaviors and disease states may be underreported. Second, observed differences in health outcomes may result at least partially from underdiagnosis among black respondents. Evaluation and treatment are often less aggressive for blacks than for whites with similar signs and symptoms, leading many to argue that health care system barriers and racism are behind underdiagnoses and late diagnoses, which are likely to result in poorer health outcomes (43). This discrepancy in care is unlikely to account entirely for the observed differences, however, given that the education level was higher for this black cohort (35% bachelor's degree or higher) than for blacks nationally (15% bachelor's degree or higher) (44).

Also, national data indicate that 80.3% of blacks and 88.7% of whites have health insurance (45). Although we do not have insurance data for the study population, national rates are most likely lower than those of this black cohort, whose education and full-time employment levels are higher than those for blacks nationally. Again, these facts may diminish the chances that all or most of the observed differences are accounted for by underdiagnosing alone. On the other hand, studies have found that blacks have less dyslipidemia, consistently higher HDL cholesterol, and sometimes lower LDL cholesterol than do whites (46,47). Published data on the prevalence of myocardial infarction among blacks are scarce. If the incidence of myocardial infarction is similar for blacks and whites, the explanation may be that mortality from myocardial infarction is higher for blacks and, thus, prevalence is lower.

Although the overrepresentation of women in our study should be noted as a concern, church-going populations usually have more women than men (48). The more pronounced difference for blacks is not surprising, given that church life plays a more central role in black than in white culture (49). The younger age of black respondents, although a result of eligibility requirements, should be noted as a potential study limitation even though all analyses were age-adjusted. Although we found that personal income was higher for black than for white women, we did not find this difference between black and white men. The reason may be that black women continue to work at later ages than do white women and are more often single heads-of-households, possibilities that are supported by the significantly lower household incomes among black than among white women. Exploring such complex issues as whether the stronger health benefits we observed for black men as compared with black women might be associated with added stress in black women's lives will be of interest. A final source of potential bias is the differential recruitment of black and white respondents to the study. We believe, however, that the need to recruit successfully far outweighs concerns about not using the same protocol for the two racial groups, a practice that has resulted in numbers too low for meaningful analyses in other studies.

The numbers in our study were large enough to allow us to explore whether the health benefits of lifestyle practices observed for white Adventists hold for black Adventists (7). Although the results for hypertension and diabetes were poorer for blacks than for whites, health outcomes in a number of categories were actually better for black than for white Adventists and especially better for black Adventists than for blacks nationally. With a full enrollment goal for AHS-2 of approximately 100,000 participants, 30,000 of whom are black, we will have adequate numbers to further investigate the hypothesis that black participants benefit even more than whites from a healthy diet and a tobacco-free, active lifestyle.

For these analyses, we plan to examine the independent effects of components of the Adventist lifestyle and their relative effects in these two groups by looking at high and low adherers to each lifestyle practice. We will also attempt to separate effects among blacks of Caribbean descent and those of American descent. This analysis will be limited, however, because blacks of Caribbean descent make up only 15% of the black study population, and not all endpoints are represented in this small group. Finally, because this cohort is recruited through churches, exploration of the role of spirituality as either a supplementary or a confounding factor is particularly relevant, and a separately funded study is already exploring this issue.

The potential contribution of our current and future findings to the national dialogue on health disparities is substantial. Evidence from this study acknowledges that black Adventists have a broad range of lifestyles, but just as the profile of health habits for white Adventists is better than that of white non-Adventists, the profile for black Adventists is better than that of their non-Adventist counterparts. Given the challenges of many other contributors to health disparities, including racism, housing segregation, employment discrimination, limited educational opportunity, and poorer health care, the study of the relative advantages of the Adventist lifestyle for blacks holds promise for helping to close the gap in health status between blacks and whites nationally.