Skip to main content

Associations of clustered health risk behaviors with diabetes and hypertension in White, Black, Hispanic, and Asian American adults



The clustering of Big Four contributors to morbidity and mortality—alcohol misuse, smoking, poor diet, and physical inactivity—may further elevate chronic health risk, but there is limited information about their specific combinations and associated health risks for racial/ethnic minority groups. We aimed to examine patterns of clustering in risk behaviors for White, Black, Hispanic, and Asian American adults and their associations with diabetes and hypertension. As these behaviors may be socioeconomically-patterned, we also examined associations between clustering and socioeconomic status (SES).


Latent class analyses and multinomial and logistic regressions were conducted using a nationally-representative sample of United States (US) adults ages 40–70 (N = 35,322) from Waves 2 (2004–2005) and 3 (2012–2013) of the National Epidemiologic Survey on Alcohol and Related Conditions. Obesity was used as a proxy for unhealthy diet. The outcomes were diabetes and hypertension.


A relatively-healthy-lifestyle class was found only among White adults. Common patterns of unhealthy clustering were found across groups with some variations: the obese-inactive class among White, Black, and Hispanic adults (and the inactive class among Asian adults); the obese-inactive-smoking class among White, Black, and Hispanic adults; the smoking-risky-drinking class among White and Hispanic adults; and the smoking-risky-drinking-inactive class among Black and Asian adults. Positive associations of unhealthier clustering (having a greater number of risk behaviors) with lower SES (i.e., family income and education) and with health conditions were more consistent for Whites than for other groups. For racial minority groups, lower education than income was more consistently associated with unhealthy clusters. The associations between unhealthier clustering and diabetes and hypertension were less clear for Blacks and Asians than for Whites, with no significant association observed for Hispanics.


Concerted efforts to address clustered risk behaviors in most US adults, particularly in racial/ethnic minority groups given the high prevalence of unhealthy clustering, are warranted.

Peer Review reports


There has been increasing research interest in lifestyle risk behaviors that collectively increase health risk. Of particular concern are alcohol misuse, cigarette smoking, poor diet, and physical inactivity—the “Big Four” contributors to mortality [1] and the leading proximal and modifiable causes of morbidity [2]—in which their synergistic effects are suggested to be more detrimental to health than their cumulative individual effects [2]. Still, there is a paucity of information about the specific combinations of them [3], as studies have focused on quantifying the co-occurrence of these behaviors, mostly using counts of risk behaviors [4] or lifestyle indices as summary measures of healthfulness of lifestyles [5, 6].

In the current study, we aim to examine the clustering of these behaviors and its associations with two common chronic conditions, hypertension and diabetes, among White, Black, Hispanic, and Asian American adults. Hypertension is the leading single risk factor for morbidity and mortality, and a major risk factor for cardiovascular disease [7, 8], the leading cause of death in the United States (US) [9]. Type 2 diabetes is accompanied by complications like cardiovascular diseases, retinopathy, nephropathy, and cancers, and consequently associated with increased risk for premature death [5]. Although genetic predisposition partly determines individual susceptibility, these conditions largely are by-products of unhealthy lifestyles featuring health risk behaviors [10]. Continued engagement in these behaviors while having these conditions increases the risks for complications with greater morbidity and premature mortality.

Of note, race-specific information about the clustering of these behaviors is rare. To our knowledge, our recent study is the first one that has reported on the clustering of the Big Four behaviors among Whites, Blacks, and Hispanics, using the National Alcohol Survey data and validating the clusters using self-rated health as the outcome [11]. Health behaviors are influenced by the sociocultural and economic circumstances that shape decisions about them [12, 13]. Racial/ethnic minorities (excepting Asians) tend to have lower SES than Whites in the US [14]. As past research suggests that individuals of lower SES are more likely to engage in unhealthy behaviors such as tobacco use, physical inactivity, and poor nutrition [15], more unhealthy clustering of lifestyle behaviors may be observed among Black and Hispanic adults.

There is also evidence that cultural norms and expectations in ethnic minority communities influence health behaviors. For example, in addition to income, prices, and access to quality fresh food (often dictated by one’s SES), dietary patterns are also shaped by individual preferences and beliefs, and sociocultural and ethnic factors [16, 17], with social norms and modeling exerting powerful influences on food choice and consumption amounts [18,19,20,21]. Foods are often used to affirm culture and forge social bonds, and frequent kinship gathering among Blacks and Hispanics around food, where rich, traditional or cultural foods may take precedence over more healthful eating, along with community norms that may dissuade adopting healthier food options [17, 22], may also result in unhealthier diets being more pervasive in Black and Hispanic communities. Low social support for health-promoting activities in Black communities such as healthy diet or regular exercise has also been noted as a barrier to a healthy lifestyle [23, 24], and might reflect a lack of neighborhood amenities (e.g., recreational spaces, affordable and accessible fresh foods) fostering this [25, 26]. Asian cultural values that do not prioritize physical activity [27] may contribute to a sedentary lifestyle. Additionally, greater exposure to stressors associated with racial minority status such as racial discrimination [28] and lower access to health-promoting resources such as health care may lead to unhealthier lifestyles among racial minority groups. The clustering of health risk behaviors, therefore, is likely to be ethnically-patterned, and understanding race-specific patterns is critical for informing contextually-relevant interventions.

As these risk behaviors constitute pathways that lead to disparities in these conditions or their management [29], a better understanding of their clustering, common or varying among these groups, can inform appropriate intervention strategies tailored to each group. As the effects of race and SES on health are confounded in the US [30], to disentangle the respective effects of race and SES, each of which may engender constraints on health-related behaviors, we also examine whether unhealthier clustering is associated with lower SES in each group.

Three research questions are addressed: 1) What are the common and diverging patterns of clustered risk behaviors across these four racial/ethnic groups?; 2) Is unhealthier clustering associated with lower SES?; and 3) Is unhealthier clustering associated with diabetes and hypertension?



A nationally-representative sample of US adults ages 40 to 70 (N= 35,322) was drawn from Waves 2 (2004–2005) and 3 (2012–2013) of the cross-sectional National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). This sample excludes younger and older age groups because of age-related health risks and lifestyle patterns. Diabetes and hypertension become more prevalent in midlife, with their risks steeply increasing in older adulthood largely due to aging [31]. Older adults tend to reduce dietary/alcohol intake, quit smoking out of health concerns, and reduce physical activities due to age-related functional declines [32,33,34]. NESARC used multistage probability sampling, and the response rates were 87.6% at Wave 2 and 84.0% for Wave 3.


Lifestyle factors

In light of research showing beneficial effects of moderate drinking on some chronic conditions (e.g., diabetes) and adverse health effects of heavy drinking [35,36,37] and abstinence [36, 38], past-year alcohol consumption was an ordinal variable of lifetime abstinence, former drinking, drinking < 7 drinks/week, 7- < 14 drinks/week, and > 14 drinks/week (hereafter referred to as risky drinking), based on the low-risk drinking guidelines of the National Institute on Alcohol Abuse and Alcoholism [39].

Smoking status had three categories of current smoker, former smoker, and lifetime non-smoker [40,41,42] to differentiate former smoker who may have quit smoking due to health concerns from lifetime non-smoker.

With no information about diet in NESARC, obesity (a body mass index of > 30 kg/m2) [43] was used. Obesity is an indicator of a state of positive energy balance that reflects chronic overeating [44] and attributed primarily to excess caloric intake [45], and thus is a reasonable proxy for unhealthy diet. Importantly, obesity is among the most prominent risk factors for a host of debilitating and life-threatening chronic conditions such as type 2 diabetes, cardiovascular diseases, and some cancers [46].

Physical inactivity is a dichotomous variable of < 150 min of moderate-intensity or < 75 min of vigorous-intensity activity weekly, based upon the US guidelines for physical activity [47].

Health outcomes

A dichotomous measure of doctor-diagnosed and self-reported diabetes or hypertension was based upon affirmative responses to both of the two questions: “During the last 12 months, did you have [name of condition]?”; if yes, “Did a doctor or other health professional tell you that you had [name of condition]?”

Demographic variables

Race/ethnicity was assessed using two items: one for selecting 1+ categories that describe the respondent’s race among Whites, Blacks, and Asians; and another about the respondent’s being of Hispanic or Latino origin.

Marital status variable indicated being married/living with a partner versus widowed/divorced/separated/never married [11]. Marital status was associated with both health outcomes [48] and health behaviors [49, 50].

Education was a dichotomous variable of having a 4-year college or advanced degree versus less than a college degree. Family income was a ratio of family income to the corresponding survey year’s US Federal Poverty Level [51]. We used separate indicators of SES to avoid the conceptual blurring of explanatory mechanisms for SES effects that occurs with use of a composite [52]. Health insurance coverage, which may influence disease diagnosis [53], indicated having coverage (versus no coverage) in the past year.

Nativity status (US-born versus foreign-born) and ethnicity based on the respondents’ countries of origin, both of which are potential confounders of the relationship between disease conditions and health risk behaviors, were included in models for Asians (using the categories of Chinese, Filipino, Japanese, Korean, South Asian, Vietnamese, Southeast Asian other than Vietnamese, and other Asian) and Hispanics (Cuban, Mexican, Puerto Rican, South American and Central American). Though research is somewhat mixed [54], cardiovascular risk factors including diabetes and hypertension were associated with US-born status for Hispanic adults [55] and Asian adults [56, 57] living in the US, as well as their ethnicity (or national origin) [54, 58]. Being US-born was associated with health risk behaviors such as obesity [56, 57] and alcohol consumption [59, 60] in these populations, as was their national origin [60,61,62].

Statistical analyses

Latent class analysis (LCA), a semi-parametric statistical technique that groups individuals into mutually-exclusive and substantively-meaningful latent classes [63,64,65], was conducted in Mplus [66] to identify clusters of risk behaviors. Mplus is a statistical software package that can implement a wide array of statistical models, but it is primarily known for its latent variable modeling capabilities [67].

As qualitatively distinct patterns of clustering were anticipated across racial/ethnic groups, LCA was performed separately for each group to compare patterns of clustering qualitatively, not to compare latent class prevalence directly across groups [64]. Model selection was based on model fit indices and statistics (Bayesian information criterion (BIC), Akaike information criterion (AIC), sample-sized adjusted BIC (aBIC), and bootstrapped likelihood ratio tests) and practical usefulness of each class (> 5% of the sample and meaningfully differentiating an additional class) [65, 68, 69]. Our fit statistics and practical criteria, taken together, suggest a 4-class model as the most parsimonious and substantively sound for Whites and a 3-class model each for Asians, Blacks, and Hispanics (Table 2). Where AIC, BIC, and aBIC pointed to different models, BIC and aBIC were prioritized in model selection [65, 70]. We then performed logistic regressions to estimate associations between class membership and health conditions, and multinomial logistic regressions to examine the associations of class membership with SES, accounting for demographic variables. We used the 3-step method [71], which is considered superior to the standard approach of combining the latent class model and the latent regression model into a joint model [71], specifying class membership based on the maximum posterior probabilities from the best-fitting model as a nominal variable and then using the logit of this variable to estimate logistic regression models. To adjust for sampling strategy and nonresponse, sampling weights were incorporated in our model estimation.


Table 1 shows the characteristics of our sample. White adults were slightly older than other groups. Whites had the highest average income (on average 446.9% above the Federal Poverty Line), and Asians had the highest education level, with almost half (48.5%) having a 4-year college degree. Prevalence of individuals who drank more than 14 drinks per week was higher among Blacks (8.1%) than other groups. Current smoking was more prevalent among Whites (27.5%) and Blacks (27.3%) than the other two groups, and obesity more prevalent among Blacks (44.7%) and Hispanics (37.5%) than others. Asians were less likely to engage in other risk behaviors but more likely to be inactive (39.4%) than other groups. The proportion of individuals with diagnosed diabetes or hypertension, particularly the latter, was higher among Blacks (17.1% for diabetes and 25.9% for hypertension) than among other groups.

Table 1 Sociodemographic characteristics of US adults ages 40–70, NESARC Waves 2 and 3

Clustering of health risk behaviors: latent class models

As shown in Fig. 1, about three in ten White adults were in the relatively-healthy-lifestyle class characterized by low prevalence of risky drinking (4.8%), smoking (10.1%), and physical inactivity (11.3%), and relatively low prevalence of obesity (20.1%). The obese-inactive class (19%) had higher prevalence of obesity (36.8%) and inactivity (39.2%) but mostly did not engage in risky drinking (1.2%) or smoking (0.6%). The obese-inactive-smoking class (28%) had even higher prevalence of obesity (51.3%) and inactivity (49.6%), prevalence of current smoking (27.9%) somewhat higher than the overall average (25.7%; see Table 2), and very low prevalence of risky drinking (3.0%). The smoking-risky-drinking class (24%) had very high prevalence of smoking (73.1%) and relatively high prevalence of risky drinking (21.3%), and prevalence of obesity (20.5%) and inactivity (28.2%) somewhat lower than the overall and White averages.

Fig. 1
figure 1

Classes of Clustered Risk Behaviors among US adults ages 40–70, NESARC Waves 2 and 3

Table 2 Latent class analyses of clustered risk behaviors: fit indices

For Black adults, almost half were in the obese-inactive class with about 49.1% of this class likely to be obese and 36.4% inactive, but hardly likely to smoke or engage in risky drinking. The obese-inactive-smoking class (25.0%) showed higher prevalence of obesity (58.5%) and inactivity (48.5%) than for the obese-inactive class, somewhat higher prevalence of smoking (28.6%) than the overall and group averages, and virtually no risky drinking (0.8%). The smoking-risky-drinking-inactive class (28%) had very high prevalence of smoking (73.6%) and relatively high prevalence of risky drinking (22.7%), and prevalence of inactivity (36.1%) somewhat higher than the overall average (32.3%).

For Hispanic adults, over half of them were in the obese-inactive class with about 36.6% of this class likely to be obese and 39.7% physically inactive, but not likely to smoke or engage in risky drinking. Like Whites and Blacks, the obese-inactive-smoking class (24.0%) had higher prevalence of obesity (43.1%) and inactivity (43.2%) than for the obese-inactive class, relatively high prevalence of smokers (35.2%), and very low prevalence of risky drinking (3.3%). The smoking-risky-drinking class was characterized by relatively high prevalence of smoking (36.8%) and risky drinking (21.6%), and lower than national and group averages of obesity (29.3%) and inactivity (20.2%).

About half of Asian adults were in the inactive class, with about 45.1% them likely to be inactive, but with low prevalence of risky drinking (1.2%) and obesity (8.0%), and no smokers. The somewhat-inactive-some smoking class (34%) had a similar profile to that of the inactive class but had some smokers (13.4%) and lower prevalence of inactivity (30.0%) than the inactive class. Lastly, the smoking-risky-drinking-inactive class had very high prevalence of smoking (74.2%), high prevalence of inactivity (47.4%), relatively high prevalence of risky drinking (18.5%), and prevalence of obesity (25.7%) lower than the overall average (33.7%).

Associations of health conditions and SES with class membership: multinomial and logistic regressions

In multinomial logistic regression models to examine demographic and socioeconomic profiles of class membership (Table 3), the class deemed to show a healthier lifestyle than the other classes in each group was used as the referent. For Whites, all the other classes were likely to have lower income (aOR = 0.78, p < .001 for the obese-inactive class; and aOR = 0.75, p < .001 for the obese-inactive-smoking class) and education levels (aOR = 0.44, p < .001 for the obese-inactive class; and aOR = 0.14, p < .001 for the obese-inactive-smoking class) than the relatively-healthy-lifestyle class. For Blacks, the smoking-risky-drinking-inactive class had lower income (aOR = 0.85, p < .001) and education (aOR = 0.27, p < .001), and the obese-inactive-smoking class had lower education (aOR = 0.66, p < .01), than the obese-inactive class, the referent. For Hispanics, the smoking-risky-drinking class had lower education (aOR = 0.48, p < .05), but the obese-inactive-smoking class (aOR = 2.19, p < .001) and the smoking-risky-drinking class (aOR = 4.71, p < .001) had higher income, than the obese-inactive class. For Asians, the smoking-risky-drinking-inactive class had lower education (aOR = 0.06, p < .01) and income (aOR = 0.68, p < .001), but the somewhat-inactive-some-smoking class had higher income (aOR = 1.30, p < .001) than the inactive class. Compared to the obese-inactive class among Blacks and Hispanics (and the inactive class among Asians), other lifestyle classes were more likely to be male. Similarly, the obese-inactive class was less likely to be male for Whites. The smoking-risky-drinking class was younger compared to the referent among Whites and Hispanics.

Table 3 Demographic Profiles of Clustered Risk Behavior Classes among US adults ages 40–70

In logistic regression models to examine the associations between class membership and the two conditions (Table 4), the obese-inactive class was the referent for Whites as well, because of no individual with diabetes being in the relatively-healthy-lifestyle class and our intention to examine whether an additional risk behavior added to the most-commonly observed combination of obesity and inactivity was associated with higher odds of diabetes or hypertension. For Whites, the obese-inactive-smoking class was associated with both diabetes (aOR = 2.14, p < .001) and hypertension (aOR = 1.74, p < .001). The relatively-healthy-lifestyle class was associated with no odds for diabetes and lower odds for diabetes (aOR = 0.17, p < .001 for diabetes), and the smoking-risky-drinking class was also inversely associated with diabetes (aOR = 0.17, p < .001) and hypertension (aOR = 0.55, p < .001). For Blacks, the obese-inactive-smoking class was positively associated with diabetes (aOR = 1.43, p < .01) and hypertension (aOR = 1.40, p < .01), and the smoking-risky-drinking-inactive class was inversely associated with diabetes (aOR = 0.54, p < .001), compared to the obese-inactive class. For Asians, the smoking-risky-drinking-inactive class was more likely to be diabetic (aOR = 2.67, p < .05) than the inactive class. For Hispanics, there were no significant associations between class membership and either condition.

Table 4 Associations between Clustered Risk Behavior Classes and Chronic Conditions among US adults ages 40–70


Our findings partially support our hypotheses: with some exceptions, we found unhealthier clustering of risk behaviors was associated with lower SES, and with chronic conditions. Common and different patterns were found in these relationships across the four racial/ethnic groups. Commonalities include the obese-inactive class among all but Asian adults (who, instead, had the inactivity class) and the clusters that add smoking to this mix in each group. Also common is the smoking-risky-drinking class among Whites and Hispanics, with a variation seen among Blacks and Asians in the addition of inactivity to this cluster. Key differences across racial/ethnic groups, or, to be precise, between Whites and racial minority groups, include: a sizeable relatively-healthy-lifestyle class observed only among Whites; and positive associations of unhealthier clusters with diabetes and hypertension, as well as with income and education, being more consistent for Whites than for others. For racial minority groups, education than income was more consistently associated with unhealthier clusters, and the associations of unhealthier clusters with the two disease conditions were less clear for Blacks and Asians than for Whites, with no significant association observed for Hispanics. As we discuss below, both the commonalities and differences across the groups have important public health implications.

The commonality of the obese-inactive cluster among Whites, Blacks, and Hispanics in the US suggests that addressing obesity and inactivity should be a key component of lifestyle interventions for these groups. Obesity is a well-recognized health problem in the US that increases risk for morbidity and premature mortality from major illnesses including hypertension, dyslipidemia, type 2 diabetes, cardiovascular diseases, respiratory problems, and some cancers [46]. As physical inactivity is one of the primary contributors to the obesity epidemic in the US [72], the common cluster of obesity and physical inactivity is not entirely surprising. However, another common cluster we found that additionally includes smoking is notable. As reported in a study, the joint effects of smoking, physical inactivity, and obesity could increase all-cause and CVD-specific mortality by at least 7.9 years U.S. adults [73]. The obese-inactive-smoking class, comprising about one in four adults in each of these three groups, is thus of great public health concerns.

The absence of a healthy lifestyle class among racial/ethnic minority groups does not mean that there were no individuals in these groups showing all four health-promoting behaviors, but the lack of such a class consisting of at least 5% of each minority sample in our LCA. This absence can be attributed to various sociocultural and structural forces. Past research suggests that the commonality of inactivity observed in all unhealthy clusters for Blacks and Asians may be partly attributed to lower social support for regular exercise in some Black communities [24] and Asian cultural values that place a lower priority on physical activity [27]. Perhaps more fundamentally, disparities in health-promoting and deleterious resources and environments (e.g., recreational spaces, food deserts, alcohol outlets) combined with differential exposure to chronic stressors such as racial discrimination [28] and financial strain [74], may lead to disparities in health behaviors [25, 26].

Overall, positive associations between unhealthier lifestyle classes and health conditions were more consistent for Whites than for others, with the relatively-healthy-lifestyle class having lower odds of diabetes and hypertension than the obese-inactive class (and the smoking-risky-drinking class, according to our post hoc analysis using the relatively-healthy-lifestyle class as the referent; results not shown for brevity of recording). Unhealthier clustering that adds smoking to the obesity-inactivity combination for Whites and Blacks (and to inactivity in Asians) was associated with higher odds of disease condition. To the extent these classes capture each respondent’s long-term lifestyle, these findings suggest elevated health risk associated with an additional risk behavior added to the unhealthy cluster. This is consistent with past research showing poorer health associated with larger counts of risk behaviors [75] or lower healthy lifestyle index scores [5, 6].

Still, given the cross-sectional design of the present study, strictly speaking, these associations capture continued engagement in risk behaviors while having either or both conditions, rather than the causal effects of the clustered risk behaviors on the conditions. That individuals who already have diabetes or hypertension and who can risk complications are more, not less, likely to (continue to) engage in risk behaviors is a cause for public health concerns. Concerted efforts to address clustered health risk behaviors in most US adults, particularly in those whose health conditions (such as diabetes and hypertension) are adversely affected by them, are warranted.

The smoking-risky-drinking-inactive class among Blacks and the smoking-risky-drinking class among Whites were associated with lower odds of disease conditions for Whites and Blacks, compared with the obese-inactive class. At least for Whites, this may be because of the lower age of the smoking-risky drinking class than the obese-inactivity class (Table 3), given that these conditions tend to develop later in adulthood. More fine-grained analyses for different age groups among midlife or older adults may shed light on this, which we did not have sufficient statistical power to do.

We found unhealthier clustering mostly associated with lower SES. This pattern was the most consistent for Whites, with the other classes having lower income and education than the relatively-healthy-lifestyle class. This aligns well with prior studies that reported on positive associations between SES and healthier lifestyle using predominantly-White samples [2, 76]. For racial minority groups, education more than income tended to be associated with unhealthier clusters. For Blacks, for example, the smoking-risky-drinking-inactive class and the obese-inactive-smoking class had lower education (and the former had lower income as well) than the obese-inactive class. Similarly, the smoking-risky-drinking-inactive class among Asians and the smoking-risky-drinking-inactive class among Hispanics had lower education than their respective referents featuring fewer risk behaviors. Higher income, on the other hand, was positively associated with the clusters involving smoking and/or risky drinking among Asians and Hispanics, perhaps due to greater affordability of alcohol and tobacco that a higher income allows [77]. Each SES indicator measures different, often-related aspects of social stratification that may influence health [52]. Education influences health through a person’s adult occupation and income and the knowledge, health literacy and skills attained through education, which enable or motivate them to have healthier lifestyles [78]. It has been suggested that education gives individuals the ability to override the ‘default’ American lifestyle characterized by poor diet and inactivity [79]. Income can influence a wide range of material circumstances that affect health and access to health enhancing resources [78], but higher income alone may have limited salutogenic effects, particularly for racial minority groups.

That the relationship between higher SES and healthier lifestyle is more pronounced for Whites than other groups may be partly because cultural practices and social support (or the lack thereof for healthy lifestyle) in racial minority groups, which, as we noted above, may also influence lifestyle independently of SES to some degree. Unhealthy lifestyles among racial minority groups may also be attributed to psychosocial stressors such as racial inequities they experience (including low occupational achievements even at the same education level as Whites) [80], racial prejudice and unfair treatment they encounter in everyday lives [28], and overall lower subjective social status they may experience [81, 82], all of which may trigger health risk behaviors to cope with these stressors [74] or distract individuals from health-seeking behaviors [25, 26].

Our findings have important implications for future interventions. Given that the obese-inactive class and the obesity-inactive-smoking class among White, Black, and Hispanic adults together comprise a large segment of each group—47% of White, 72% of Black, and 80% of Hispanic adults (Fig. 1)—obesity and inactivity should be a central focus of lifestyle interventions for these three groups. Furthermore, in light of the synergistically adverse health effect of obesity, inactivity, and smoking on health noted above [73], as well as our current findings showing positive associations of the obese-inactive-smoking class with diabetes and hypertension for Whites and Blacks, preventive and intervention strategies for maintaining good cardio-respiratory health, particularly for these two groups, are warranted. Importantly, it should be emphasized that while clinical patient-oriented interventions are important, multi-level interventions are very likely needed for facilitating behavioral change, as health risk behaviors are related to individual, neighborhood, and environmental conditions as noted above.

In light of the consistently inverse associations between SES and unhealthy lifestyles for White adults, targeting low-SES Whites with these intervention strategies might be fruitful. Interventions to address obesity and inactivity might be directed most adults among Blacks and Hispanics, regardless of their SES. Still, given the significant negative association of the obese-inactive-smoking class and college degree for Blacks, it would be reasonable to target Blacks without a college degree with interventions addressing all three behaviors. The absence of obesity in any unhealthy lifestyle cluster and the commonality of inactivity in unhealthy clusters for Asians calls for a unique strategy to improve their health behaviors. As noted above, physical inactivity may be rooted in Asian cultural values that do not emphasize exercise, and thus interventions to address them could be effective, especially when considering the additional health risks of smoking alongside inactivity. While inactivity is also present in all unhealthy lifestyle classes among Blacks, the high prevalence of obesity among Blacks warrant strategies to address both obesity and inactivity.

We acknowledge several limitations of this study. First, gender differences in the clustering of risk behaviors were not explored in our LCAs to maximize statistical power for racial/ethnic-specific analyses and also to keep the focus on racial/ethnic differences. Still, multinomial regressions captured gendered clustering of risk behaviors (e.g., the obese-inactive class more likely to be female). Second, though reasonable [44], the use of obesity as a proxy for unhealthy diet is a limitation. Third, as our LCA models were not specific to gender, we used the cutoff for risky drinking for men (> 14 drinks weekly versus > 7 drinks weekly for women) largely because risky drinking is more pervasive among men than women [83]. Using a conservative measure for women’s heavy drinking may have underestimated the unhealthy clusters involving risky drinking among them. There may indeed be gender differences in the clustering of risk behaviors, which we did not explore in order to maximize statistical power for racial/ethnic-specific analyses and also to keep the focus on racial/ethnic differences.


The current study meaningfully contributes to the health disparities literature concerning health risk behaviors. Adding to prior studies that used quantitative summary measure of the clustering of risk behaviors [4,5,6], our study provides qualitative information on the actual clustering and for each racial/ethnic group. Our race-specific findings—e.g., the absence of a (relatively) healthy lifestyle class and more consistent associations of unhealthier clustering with education than with income among racial minorities—add important nuances to the thesis of higher SES accompanied by a healthier lifestyle, established in studies using predominantly White samples [2, 76]. In doing so, this study points to the need to better understand the complex pathways by which social determinants influence health in ways that are common and diverse across racial/ethnic groups. Additionally, the common and diverging patterns of clustering across these groups and their associations with the two chronic conditions we report have the great potential to inform multi-behavior interventions. Concerted efforts to address clustered risk behaviors in most US adults, particularly racial/ethnic minority groups and those with chronic conditions, are warranted.

Availability of data and materials

Not applicable. This study used national survey data NIAAA makes available upon request.



Akaike information criterion


Bayesian information criterion


Latent class analysis


National Epidemiologic Survey on Alcohol and Related Conditions


Socioeconomic status


United States


  1. McGinnis JM, Foege WH. Actual causes of death in the United States. JAMA. 1993;270(18):2207–12.

    Article  CAS  PubMed  Google Scholar 

  2. Noble N, Paul C, Turon H, Oldmeadow C. Which modifiable health risk behaviours are related? A systematic review of the clustering of smoking, nutrition, alcohol and physical activity ('SNAP') health risk factors. Prev Med. 2015;81:16–41.

    Article  PubMed  Google Scholar 

  3. Lacombe J, Armstrong MEG, Wright FL, Foster C. The impact of physical activity and an additional behavioural risk factor on cardiovascular disease, cancer and all-cause mortality: a systematic review. BMC Public Health. 2019;19(1):900.

    Article  PubMed  PubMed Central  Google Scholar 

  4. McAloney K, Graham H, Law C, Platt L. A scoping review of statistical approaches to the analysis of multiple health-related behaviours. Prev Med. 2013;56(6):365–71.

    Article  PubMed  Google Scholar 

  5. Schlesinger S, Neuenschwander M, Ballon A, Nöthlings U, Barbaresko J. Adherence to healthy lifestyles and incidence of diabetes and mortality among individuals with diabetes: a systematic review and meta-analysis of prospective studies. J Epidemiol Community Health. 2020;74(5):481–7.

    Article  PubMed  Google Scholar 

  6. Zhang Y, Pan XF, Chen J, Xia L, Cao A, Zhang Y, et al. Combined lifestyle factors and risk of incident type 2 diabetes and prognosis among individuals with type 2 diabetes: a systematic review and meta-analysis of prospective cohort studies. Diabetologia. 2020;63(1):21–33.

    Article  PubMed  Google Scholar 

  7. GBD 2015 Risk Factors Collaborators. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990-2015: a systematic analysis for the global burden of disease study 2015. Lancet. 2016;388(10053):1659–724.

    Article  Google Scholar 

  8. Muntner P, Hardy ST, Fine LJ, Jaeger BC, Wozniak G, Levitan EB, et al. Trends in Blood Pressure Control Among US Adults With Hypertension, 1999–2000 to 2017-2018. JAMA. 2020;324(12):1190–200.

    Article  PubMed  Google Scholar 

  9. Heron M. Deaths: leading causes for 2018. Natl Vital Stat Rep. 2021;70(4):1–115.

    PubMed  Google Scholar 

  10. Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol. 2018;14(2):88–98.

    Article  PubMed  Google Scholar 

  11. Cook WK, Kerr WC, Karriker-Jaffe KJ, Li L, Lui CK, Greenfield TK. Racial/ethnic variations in clustered risk behaviors in the U.S. Am J Prev Med. 2020;58(1):e21–9.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Cockerham WC. The sociology of health behavior and health lifestyles. In: Bird CE, Conrad P, Fremont AM, editors. Handbook of medical sociology. 5th ed. Upper Saddle River: Prentice Hall; 2000.

    Google Scholar 

  13. Abel T, Frohlich KL. Capitals and capabilities: linking structure and agency to reduce health inequalities. Soc Sci Med. 2012;74(2):236–44.

    Article  PubMed  Google Scholar 

  14. U.S. Bureau of Labor Statistic: Race, Economics, and social status. 2018.

    Google Scholar 

  15. Pampel FC, Krueger PM, Denney JT. Socioeconomic disparities in health behaviors. Annu Rev Sociol. 2010;36:349–70.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Burrowes JD. Incorporating ethnic and cultural food preferences in the renal diet. Adv Ren Replace Ther. 2004;11(1):97–104.

    Article  PubMed  Google Scholar 

  17. Tiedje K, Wieland ML, Meiers SJ, Mohamed AA, Formea CM, Ridgeway JL, et al. A focus group study of healthy eating knowledge, practices, and barriers among adult and adolescent immigrants and refugees in the United States. Int J Behav Nutr Phys Act. 2014;11:63.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Cruwys T, Bevelander KE, Hermans RC. Social modeling of eating: a review of when and why social influence affects food intake and choice. Appetite. 2015;86:3–18.

    Article  PubMed  Google Scholar 

  19. Higgs S. Social norms and their influence on eating behaviours. Appetite. 2015;86:38–44.

    Article  PubMed  Google Scholar 

  20. Robinson E, Blissett J, Higgs S. Social influences on eating: implications for nutritional interventions. Nutr Res Rev. 2013;26(2):166–76.

    Article  PubMed  Google Scholar 

  21. Robinson E, Thomas J, Aveyard P, Higgs S. What everyone else is eating: a systematic review and meta-analysis of the effect of informational eating norms on eating behavior. J Acad Nutr Diet. 2014;114(3):414–29.

    Article  PubMed  Google Scholar 

  22. James DC. Factors influencing food choices, dietary intake, and nutrition-related attitudes among African Americans: application of a culturally sensitive model. Ethn Health. 2004;9(4):349–67.

    Article  PubMed  Google Scholar 

  23. Griffith DM. "I AM a man": manhood, minority Men's health and health equity. Ethn Dis. 2015;25(3):287–93.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Joseph RP, Ainsworth BE, Keller C, Dodgson JE. Barriers to physical activity among African American women: an integrative review of the literature. Women Health. 2015;55(6):679–99.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Ravenell JE, Johnson WE Jr, Whitaker EE. African-American men's perceptions of health: a focus group study. J Natl Med Assoc. 2006;98(4):544–50.

    PubMed  PubMed Central  Google Scholar 

  26. Cheatham CT, Barksdale DJ, Rodgers SG. Barriers to health care and health-seeking behaviors faced by black men. J Am Acad Nurse Pract. 2008;20(11):555–62.

    Article  PubMed  Google Scholar 

  27. Yi SS, Roberts C, Lightstone AS, Shih M, Trinh-Shevrin C. Disparities in meeting physical activity guidelines for Asian-Americans in two metropolitan areas in the United States. Ann Epidemiol. 2015;25(9):656–660 e652.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Zemore SE, Karriker-Jaffe KJ, Keithly S, Mulia N. Racial prejudice and unfair treatment: interactive effects with poverty and foreign nativity on problem drinking. J Stud Alcohol Drugs. 2011;72(3):361–70.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Cutler DM, Lleras-Muney A. Understanding differences in health behaviors by education. J Health Econ. 2010;29(1):1–28.

    Article  PubMed  Google Scholar 

  30. Thorpe RJ Jr, Bell CN, Kennedy-Hendricks A, Harvey J, Smolen JR, Bowie JV, et al. Disentangling race and social context in understanding disparities in chronic conditions among men. J Urban Health. 2015;92(1):83–92.

    Article  PubMed  Google Scholar 

  31. Yazdanyar A, Newman AB. The burden of cardiovascular disease in the elderly: morbidity, mortality, and costs. Clin Geriatr Med. 2009;25(4):563–77 vii.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Kehlet M, Schroeder TV, Tonnesen H. The gold standard program for smoking cessation is effective for participants over 60 years of age. Int J Environ Res Public Health. 2015;12(3):2574–87.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Saint Onge JM, Krueger PM. Education and racial-ethnic differences in types of exercise in the United States. J Health Soc Behav. 2011;52(2):197–211.

    Article  PubMed  Google Scholar 

  34. Lin KC, Chi LY, Twisk JW, Lee HL, Chen PC. Trajectory stability and factors affecting trajectories over time of the longitudinal age-related change in physical performance among older people. Exp Aging Res. 2011;37(3):358–76.

    Article  PubMed  Google Scholar 

  35. Pietraszek A, Gregersen S, Hermansen K. Alcohol and type 2 diabetes. A review. Nutr Metab Cardiovasc Dis. 2010;20(5):366–75.

    Article  CAS  PubMed  Google Scholar 

  36. Roerecke M, Rehm J. Alcohol consumption, drinking patterns, and ischemic heart disease: a narrative review of meta-analyses and a systematic review and meta-analysis of the impact of heavy drinking occasions on risk for moderate drinkers. BMC Med. 2014;12:182.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Kerr WC, Yu Y. Population-level relationships between alcohol consumption measures and ischemic Heart disease mortality in U.S. time-series. Alcohol Clin Exp Res. 2007;31(11):1913–9.

    Article  PubMed  Google Scholar 

  38. Kerr WC, Ye Y. Relationship of life-course drinking patterns to diabetes, heart problems, and hypertension among those 40 and older in the 2005 U.S. National Alcohol Survey. J Stud Alcohol Drugs. 2010;71(4):515–25.

    Article  PubMed  PubMed Central  Google Scholar 

  39. National Institute on Alcohol Abuse and Alcoholism. Helping patients who drink too much: a clinician's guide. Rockville: NIH Publication; 2005.

    Google Scholar 

  40. Fine LJ, Philogene GS, Gramling R, Coups EJ, Sinha S. Prevalence of multiple chronic disease risk factors. 2001 National Health Interview Survey. Am J Prev Med. 2004;27(2 Suppl):18–24.

    Article  PubMed  Google Scholar 

  41. Funderburk JS, Maisto SA, Sugarman DE, Wade M. The covariation of multiple risk factors in primary care: a latent class analysis. J Behav Med. 2008;31(6):525–35.

    Article  PubMed  Google Scholar 

  42. Conry MC, Morgan K, Curry P, McGee H, Harrington J, Ward M, et al. The clustering of health behaviours in Ireland and their relationship with mental health, self-rated health and quality of life. BMC Public Health. 2011;11:692.

    Article  PubMed  PubMed Central  Google Scholar 

  43. Centers for Disease Control and Prevention. Defining Adult Overweight and Obesity. In: Edited by Division of Nutrition PA, and Obesity; 2012.

    Google Scholar 

  44. Spring B, Moller AC, Coons MJ. Multiple health behaviours: overview and implications. J Public Health (Oxf). 2012;34(Suppl 1):i3–10.

    Article  Google Scholar 

  45. Young LR, Nestle M. The contribution of expanding portion sizes to the US obesity epidemic. Am J Public Health. 2002;92(2):246–9.

    Article  PubMed  PubMed Central  Google Scholar 

  46. National Heart L, and Blood Institute. Managing overweight and obesity in adults: Systematic evidence review from the Obesity Expert Panel 2013.

    Google Scholar 

  47. Office of Disease Prevention and Health Promotion: 2015–2020 Physical Activity Guidelines for Americans. In: U.S. Department of Health and Human Services. 2017.

  48. Shor E, Roelfs DJ, Bugyi P, Schwartz JE. Meta-analysis of marital dissolution and mortality: reevaluating the intersection of gender and age. Soc Sci Med. 2012;75(1):46–59.

    Article  PubMed  Google Scholar 

  49. Keenan K, Ploubidis GB, Silverwood RJ, Grundy E. Life-course partnership history and midlife health behaviours in a population-based birth cohort. J Epidemiol Community Health. 2017;71(3):232–8.

    Article  PubMed  Google Scholar 

  50. Kim A, Lee JA, Park HS. Health behaviors and illness according to marital status in middle-aged Koreans. J Public Health. 2017;40(2):e99–e106.

    Article  Google Scholar 

  51. United States Department of Health & Human Services: Prior HHS Poverty Guidelines and Federal Register References. In. Edited by Secretary OoTA, Evaluation fPa. Washington, DC.

  52. Braveman PA, Cubbin C, Egerter S, Chideya S, Marchi KS, Metzler M, et al. Socioeconomic status in health research: one size does not fit all. JAMA. 2005;294(22):2879–88.

    Article  CAS  PubMed  Google Scholar 

  53. Freeman JD, Kadiyala S, Bell JF, Martin DP. The causal effect of health insurance on utilization and outcomes in adults: a systematic review of US studies. Med Care. 2008;46(10):1023–32.

    Article  PubMed  Google Scholar 

  54. Kamody RC, Grilo CM, Vásquez E, Udo T. Diabetes prevalence among diverse Hispanic populations: considering nativity, ethnic discrimination, acculturation, and BMI. Eat Weight Disord. 2021;26(8):2673–82.

    Article  PubMed  Google Scholar 

  55. Kershaw KN, Giacinto RE, Gonzalez F, Isasi CR, Salgado H, Stamler J, et al. Relationships of nativity and length of residence in the U.S. with favorable cardiovascular health among Hispanics/Latinos: The Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Prev Med. 2016;89:84–9.

    Article  PubMed  PubMed Central  Google Scholar 

  56. Echeverria SE, Mustafa M, Pentakota SR, Kim S, Hastings KG, Amadi C, et al. Social and clinically-relevant cardiovascular risk factors in Asian Americans adults: NHANES 2011-2014. Prev Med. 2017;99:222–7.

    Article  PubMed  Google Scholar 

  57. Kalra R, Patel N, Arora P, Arora G. Cardiovascular health and disease among Asian-Americans (from the National Health and nutrition examination survey). Am J Cardiol. 2019;124(2):270–7.

    Article  PubMed  Google Scholar 

  58. Ye J, Rust G, Baltrus P, Daniels E. Cardiovascular risk factors among Asian Americans: results from a National Health Survey. Ann Epidemiol. 2009;19(10):718–23.

    Article  PubMed  Google Scholar 

  59. Zemore SE. Acculturation and alcohol among Latino adults in the United States: a comprehensive review. Alcohol Clin Exp Res. 2007;31(12):1968–90.

    Article  PubMed  Google Scholar 

  60. Cook WK, Bond J, Karriker-Jaffe KJ, Zemore S. Who's at risk? Ethnic drinking cultures, foreign nativity, and problem drinking among Asian American young adults. J Stud Alcohol Drugs. 2013;74(4):532–41.

    Article  PubMed  PubMed Central  Google Scholar 

  61. Cook WK, Caetano R. Ethnic drinking cultures, gender, and socioeconomic status in Asian American and Latino drinking. Alcohol Clin Exp Res. 2014;38(12):3043–51.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Cook WK, Mulia N, Karriker-Jaffe KJ. Ethnic drinking cultures and alcohol use among Asian American adults: findings from a National Survey. Alcohol Alcohol. 2012;47(3):340–8.

    Article  PubMed  PubMed Central  Google Scholar 

  63. Magidson J, Vermunt JK. Latent Class Models. In: Kaplan C, editor. The Sage Handbook of Quantitative Methodology for the Social Sciences. Thousand Oaks: Sage Publications; 2004. p. 175–98.

    Google Scholar 

  64. Collins LM, Lanza ST. Latent class and latent transitional analysis: with applications in the social, behavioral, and health sciences. Hoboken: Wiley; 2010.

    Google Scholar 

  65. Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Model. 2007;14(4):535–69.

    Article  Google Scholar 

  66. Muthén LK, Muthén BO. Mplus User’s guide. 8th ed. Los Angeles: Muthén & Muthén; 1998-2017.

    Google Scholar 

  67. Frey BB. The SAGE encyclopedia of educational research, measurement, and evaluation. Thousand Oaks: Sage Publications, Inc.; 2018.

  68. Muthen B. Statistical and substantive checking in growth mixture modeling: comment on Bauer and Curran (2003). Psychol Methods. 2003;8(3):369–77 discussion 384-393.

    Article  PubMed  Google Scholar 

  69. Muthen B, Muthen LK. Integrating person-centered and variable-centered analyses: growth mixture modeling with latent trajectory classes. Alcohol Clin Exp Res. 2000;24(6):882–91.

    Article  CAS  PubMed  Google Scholar 

  70. Yang C-C. Evaluating latent class analysis models in qualitative phenotype identification. Comput Stat Data Anal. 2006;50(4):1090–104.

    Article  Google Scholar 

  71. Asparouhov T, Muthén B. Auxiliary variables in mixture modeling: three-step approaches using M plus. Struct Equ Model. 2014;21(3):329–41.

    Article  Google Scholar 

  72. Centers for Disease Control and Prevention: Adult obesity causes and consequences. 2016.

    Google Scholar 

  73. Borrell LN. The effects of smoking and physical inactivity on advancing mortality in U.S. adults. Ann Epidemiol. 2014;24(6):484–7.

    Article  PubMed  Google Scholar 

  74. Mulia N, Zemore SE. Social adversity, stress, and alcohol problems: are racial/ethnic minorities and the poor more vulnerable? J Stud Alcohol Drugs. 2012;73(4):570–80.

    Article  PubMed  PubMed Central  Google Scholar 

  75. Tsai J, Ford ES, Li C, Zhao G, Pearson WS, Balluz LS. Multiple healthy behaviors and optimal self-rated health: findings from the 2007 behavioral risk factor surveillance system survey. Prev Med. 2010;51(3–4):268–74.

    Article  PubMed  Google Scholar 

  76. Kino S, Bernabe E, Sabbah W. Socioeconomic inequality in clusters of health-related behaviours in Europe: latent class analysis of a cross-sectional European survey. BMC Public Health. 2017;17(1):497.

    Article  PubMed  PubMed Central  Google Scholar 

  77. Jiang H, Livingston M. The dynamic effects of changes in prices and affordability on alcohol consumption: an impulse response analysis. Alcohol Alcohol. 2015;50(6):631–8.

    Article  CAS  PubMed  Google Scholar 

  78. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Community Health. 2006;60(1):7–12.

    Article  PubMed  PubMed Central  Google Scholar 

  79. Mirowsky J, Ross CE. Education, health, and the default American lifestyle. J Health Soc Behav. 2015;56(3):297–306.

    Article  PubMed  Google Scholar 

  80. Zeng Z, Xie Y. Asian-Americans' earnings disadvantage reexamined: the role of place of education. Am J Sociol. 2004;109(5):1075–108.

    Article  Google Scholar 

  81. Gelatt J. Looking down or looking up: status and subjective well-being among Asian and Latino immigrants in the United States. Int Migr Rev. 2013;47(1):39–75.

    Article  PubMed  PubMed Central  Google Scholar 

  82. de Castro AB, Gee GC, Takeuchi DT. Examining alternative measures of social disadvantage among Asian Americans: the relevance of economic opportunity, subjective social status, and financial strain for health. J Immigr Minor Health. 2010;12(5):659–71.

    Article  PubMed  Google Scholar 

  83. Dawson DA, Goldstein RB, Saha TD, Grant BF. Changes in alcohol consumption: United States, 2001-2002 to 2012-2013. Drug Alcohol Depend. 2015;148:56–61.

    Article  PubMed  Google Scholar 

Download references


This study was supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) grants R21 AA026654 (Cook PI) and P50 AA005595-41 (Kerr PI).


This study was supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) grants R21 AA026654 (Cook PI) and P50 AA005595–41 (Kerr PI).

Author information

Authors and Affiliations



Won Kim Cook conceptualized and designed this study, participated in data analysis, and drafted and revised the paper. Libo Li conducted data analysis, drafted the methods section, and participated in revisions of the methods and results sections. Christina C. Tam constructed measures and provided feedback on drafts. Nina Mulia and William C. Kerr participated in designing the study, and provided input concerning data analysis and feedback on drafts. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Won Kim Cook.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Public Health Institute Institutional Review Board (IRB # I18–021), which granted the request of the research team for an exemption from review of this secondary analysis project of national survey data that had been collected by the National Institute of Alcohol Abuse and Alcoholism. The authors confirm that all methods were carried out in accordance with relevant guidelines and regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cook, W.K., Li, L., Tam, C.C. et al. Associations of clustered health risk behaviors with diabetes and hypertension in White, Black, Hispanic, and Asian American adults. BMC Public Health 22, 773 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: