Health Risk Behaviors in the United States: A Socio-demographic Perspective

doi:10.21203/rs.3.rs-4721393/v1

Download PDF

Research Article

Health Risk Behaviors in the United States: A Socio-demographic Perspective

https://doi.org/10.21203/rs.3.rs-4721393/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Health disparities are ongoing challenges in the United States, with one of the underlying factors being the variation in health risk behaviors across different socioeconomic and demographic communities. In this study, utilizing zip code-level data from the Centers for Disease Control and Prevention (CDC), we conducted generalized linear model regressions to explore the associations between race, poverty, education, and urban/rural status in areas. Our analysis indicated that areas with a higher prevalence of Black population and higher poverty levels are associated with a more significant number of unique health risk behaviors. Conversely, other racial groups were associated with fewer unique health risk behaviors, although they may still engage in higher levels of specific behaviors. These insights underscore the need for tailored public health strategies to address the disparities in health risk behaviors across different sub-populations in the US. Additionally, we employed Finite Mixture Modeling (FMM) to identify distinct sub-populations in the United States based on health risk behaviors, specifically binge drinking, smoking, sleep deprivation, and physical inactivity, which are essential risk factors for chronic diseases such as cancer, diabetes, and cardiovascular diseases. We identified four clusters and conducted an in-depth analysis of the demographics and socioeconomic characteristics of the identified clusters. Our findings reveal that one cluster, characterized by the worst health risk behaviors, exhibited the highest poverty rate, the least favorable health insurance profile, and the lowest frequencies of post-secondary education attainment compared to the other three clusters. This cluster, therefore, warrants targeted public health interventions.

Health Risk Behaviors

Racial Disparities

Rural-Urban Status

Socio-economic Status

Finite Mixture Model

Generalized Linear Model

Researching health risk behaviors in the United States is critical due to the significant impact these behaviors have on public health outcomes, healthcare costs, and overall quality of life. Health risk behaviors, such as smoking, excessive alcohol consumption, physical inactivity, and poor dietary choices, are significant contributors to chronic diseases like heart disease, cancer, and diabetes, which are leading causes of morbidity and mortality in the country [1–2]. The economic burden of health risk behaviors is substantial, with direct healthcare costs and indirect costs due to lost productivity amounting to billions of dollars annually [3].

Health risk behaviors exhibit considerable variation across different socioeconomic and demographic groups and at urban/rural levels. Socioeconomic status is a crucial determinant, with lower status often associated with a higher prevalence of behaviors such as smoking, poor diet, and physical inactivity. This association is partly due to limited access to resources that promote healthy lifestyles, such as quality education, healthcare services, and safe recreational facilities [4–5]. Demographic factors like race and ethnicity could also influence health risk behaviors [6]. Furthermore, the urban/rural residence plays a significant role in shaping health risk behaviors. Urban areas often show more favorable statistics in terms of physical inactivity, smoking, and alcoholic drinking compared to rural areas [7]. Urban areas have been reported to have better access to exercise facilities, which can encourage healthier behaviors [8]. Understanding these variations is crucial for developing tailored public health interventions that address the specific needs of diverse populations and effectively reduce health disparities.

In this study, we examine health risk behaviors, specifically smoking, binge drinking, sleep deprivation, and low physical activity, across various zip code areas in the United States. By grouping these zip code areas, we analyze each group's socioeconomic and demographic characteristics and make comparisons. This approach allows us to explore how different socioeconomic and demographic factors influence health risk behaviors. We aim to identify patterns and relationships between these factors and the prevalence of each health risk behavior.

Data

In this study, we utilized publicly accessible data from Places 2020 provided by the Centers for Disease Control and Prevention (CDC), which offered metrics on the prevalence of health risk behaviors, including Binge Drinking, Smoking, Sleep Deprivation, and Physical Inactivity by zip code. Additionally, we incorporated Social Determinants of Health (SDOH) data for 2020 to extract statistics on poverty levels and the demographics by zip code [9]. To further refine our analysis, we employed Rural-Urban Commuting Area (RUCA) data to classify zip codes into urban and rural areas, consolidating suburban areas with metropolitan areas to simplify the model. RUCA codes, derived from Census tract data, use standard measures of population density, urbanization levels, and commuting patterns to classify all U.S. Census tracts based on their rural or urban status and their commuting relationships with other tracts [10]. By merging suburban areas with urban ones, we aimed to reduce the complexity of the model while still capturing significant variations between distinctly rural and urban environments. This approach allowed us to understand better the interactions between social determinants of health, such as poverty and racial demographics, and their impact on health outcomes across different geographic settings. Table 1 provides detailed definitions of our variables. Notably, our study employs a cross-sectional design and utilizes crude prevalence estimates to assess the occurrence of the specific event within each zip code.

Table 1

Variable Definitions and Details
Variable	How it is derived
Binge Drinking	Number of adults aged ≥ 18 years who report having five or more drinks (men) or four or more drinks (women) on an occasion in the past 30 days divided by the number of adults aged ≥ 18 years who report having a specific number, including zero of drinks on an occasion in the past 30 days.
Current Smoking	Number of respondents aged ≥ 18 years who report having smoked ≥ 100 cigarettes in their lifetime and currently smoke every day or some days divided by the number of respondents aged ≥ 18 years who reported information about cigarette smoking.
Physical Inactivity	Number of respondents who answered “no” to the following question: “During the past month, other than your regular job, did you participate in any physical activities or exercises such as running, calisthenics, golf, gardening, or walking for exercise?” by the number of adults aged ≥ 18 years who reported any or no physical activity in the past month
Sleep Deprivation	Number of Respondents aged ≥ 18 years who report usually getting insufficient sleep (< 7 hours for those aged ≥ 18 years, on average, during a 24-hour period) divided by the number of respondents aged ≥ 18 years who report 0–24 hours of sleep.
White Population	Percentage of population reporting White alone or in combination with one or more races
Black or African American Population	Percentage of population reporting Black or African American alone.
American Indian and Alaska Native Population	Percentage of population reporting American Indian and Alaska Native alone.
Asian Population	Percentage of population reporting Asian alone.
Poverty	Percentage of population with income to poverty ratio under 1.0
Education	Percentage of population with any postsecondary education (ages 25 and over)
Uninsurance	Percentage of population with no health insurance (ages 64 and below)
^a Those who refused to answer, had a missing answer, or answered “don’t know/not sure”) have been excluded.
^b All variables are crude and age-adjusted (standardized by the direct method to the year 2000 standard U.S. population, distribution 9 with 95% confidence intervals and by demographic characteristics when feasible.

Table 2 presents a comprehensive breakdown of the data utilized in this study, categorized by race, poverty, and level of urbanization.

Table 2

Characteristics of the Zip Codes Grouped by Distributions of Races, Socioeconomic Measures, and Levels of Urbanization
Baseline characteristic	N = 26781
Race	Mean (±SD)
White	84.27 (20.21)
Black	9.37 (16.19)
Indian/Alaskan	2.83 (9.52)
Asian	3.30 (6.88)
Socioeconomic	Mean (±SD)
Poverty	13.16 (9.00)
Post-Secondary Education	56.44 (13.69)
Level of Urbanization	N (%)
Urban	19432 (72.56)
Rural	7349 (27.44)

Methods

Generalized Linear Models

In investigating the relationships between socioeconomic and demographic variables with various health risk behaviors, initial attempts were made to assess the suitability of ordinary least squares (OLS) regression assumptions for our data. However, these efforts revealed significant departures from the assumptions of homoscedasticity, as evidenced by the Breusch-Pagan test, and normality of residuals, as indicated by both Shapiro-Wilk and Kolmogorov-Smirnov tests [11]. Consequently, to address these issues, Generalized Linear Models (GLMs) were adopted as a more robust alternative to OLS, given their less restrictive assumptions. A comparative analysis among different GLM specifications, including Gaussian, Gamma, Gamma with a Log link, and Tweedie distributions, was conducted. Based on the criterion of minimizing BIC across all three health risk behaviors under investigation, the Gamma with Log link GLM emerged as the optimal choice for developing our predictive models. This approach ensures a more accurate representation of the complex relationships between socioeconomic and demographic predictors and health risk behaviors in our study population. We included race, poverty, urban/rural, and education as predictors. Notably, except rurality, which was a binary variable with 1 meaning living in urban areas, all other variables were continuous. In all GLM models, sex ratios, median age, populations of other non-White races, and multi-racial populations have been controlled.

Finite Mixture Model

Finite Mixture Modeling (FMM) with Expectation Maximization algorithm is employed to identify latent groups within our population based on their health risk behaviors. FMM provides a robust, flexible, and computationally convenient approach for model-based cluster analysis of the underlying distribution of complex datasets and its ability to handle non-normal distributions that are common in psychological and social sciences [12–13]. We used Bayesian Information Criteria (BIC) and Akaike Information Criteria (AIC) to determine the optimal number of clusters while accounting for model complexity. [12].

We developed FMM using four Generalized Linear Models (GLMs) utilizing Gamma distributions, each tailored to one of the four identified health risk behaviors. The predictors incorporated into these models were socio-economic and demographic variables, which had demonstrated significant associations with the respective health risk behaviors during the preceding GLM analysis phase. To achieve a simultaneous fit of all four models, we employed the StepFlexMix function from the Flexmix package in R. The posterior probabilities derived from this analysis were subsequently utilized for cluster assignments.

After clustering patients using FMM models, the subsequent step involved comparing urban/rural residence, socioeconomic, and demographic status across clusters. To conduct statistical comparisons, we assessed whether the data met the assumptions underlying parametric one-way ANOVA. A Kolmogorov-Smirnov test coupled with Q-Q plots was employed to test normality. Additionally, Levene's test was utilized to evaluate the homogeneity of variances. Due to the assumption being violated, Kruskal-Walli’s non-parametric equivalent to ANOVA modeling was applied [14], followed by post-hoc pairwise Dunn’s testing in cases where the null hypothesis was rejected to discern potential differences between groups. This comprehensive approach helps safeguard against erroneous conclusions drawn from analyses [15].

Finally, we performed contingency χ2 analysis and relative risk (RR) analysis on the frequencies of rural zip codes across clusters. In all χ2 analyses, we also conducted post-hoc procedures, including pairwise χ2 analysis and reporting odds ratios (OR), to compare individual conditions, ensure reliable inferences, and avoid misinterpretations [16].

GLM Analysis

Binge Drinking

Model 1 indicated that the Black and Asian populations and poverty are all negatively associated with binge drinking, suggesting that higher levels of these variables are significantly associated with lower levels of binge drinking. On the other hand, the White population, urban status, and post-secondary education are all significantly associated with higher binge drinking behaviors. However, given the pseudo-R-squared value, it is essential to consider that a substantial portion of the variability in binge drinking is likely influenced by factors not included in this model. The standard errors are relatively small, suggesting that the estimates of the coefficients are precise.

Smoking

Model 2 indicated that higher levels of White, Black, Indian/Alaskan, and poverty are associated with increased smoking. In addition, poverty shows a highly significant positive effect on smoking. However, the Asian population, urban status, and post-secondary education have significantly negative associations with smoking prevalence.

Sleep Deprivation

The model 3 indicated a significant positive association between the Black population and poverty with sleep deprivation. The results revealed a significant negative relationship between sleep deprivation and all other populations. In addition, urban status and post-secondary education were found to have a negative association with sleep deprivation.

Physical Inactivity

The model 3 indicates that the Black population and poverty have a positive association with physical inactivity. Conversely, other populations are negatively associated with physical inactivity. Similarly, urban status and post-secondary education are also negatively associated with physical inactivity. Table 3 provided models’ parameters and details.

Table 3

GLM Models Parameters
GLM Model	Model 1	Model 2	Model 3	Model 4
Dependent Variable	Binge Drinking	Smoking	Sleep Deprivation	Physical Inactivity
Independent Variable	Estimate (Std. Err)	Estimate (Std. Err)	Estimate (Std. Err)	Estimate (Std. Err)
White Population	0.0018 *** (0.000)	0.0011 *** (0.000)	-0.0018 *** (0.000)	-0.0011 *** (0.000)
Black Population	-0.0014 *** (0.000)	0.0022 *** (0.000)	0.0012 *** (0.000)	0.0020 *** (0.000)
Asian Population	-0.0008 * (0.000)	-0.0091 *** (0.000)	-0.0013 *** (0.000)	-0.0039 *** (0.000)
Indian/Alaskan Population	0.0018 *** (0.000)	0.0046 *** (0.000)	-0.0011 *** (0.000)	-0.0007 *** (0.000)
Urban Status	0.0061 ** (0.002)	-0.0094 *** (0.002)	-0.0352 *** (0.001)	-0.0335 *** (0.002)
Poverty	-0.0039 *** (0.000)	0.0054 *** (0.000)	0.0012 *** (0.000)	0.0064 *** (0.000)
Post-Secondary Education	0.0017 *** (0.001)	-0.0117 *** (0.000)	-0.0040 *** (0.000)	-0.0102 *** (0.000)
N	26781	26781	26781	26781
Pseudo R-squared	0.307	0.914	0.748	0.904
* p < 0.05
** p < 0.01
*** p < 0.001

FMM Analysis

Following the building of the FMM model, posterior probabilities were used to conduct cluster assignments on a zip code level. Table 4 illustrates the summary statistics for health risk behaviors, as well as socioeconomic and demographic measures across the clusters.

Table 4

Descriptive Statistics of the Identified Clusters
	Cluster 1 N = 3813 (14.24%)		Cluster 2 N = 11592 (43.28%)		Cluster 3 N = 1361 (5.08%)		Cluster 4 N = 10015 (37.40%)
Measure	Mean	SD (±)	Mean	SD (±)	Mean	SD (±)	Mean	SD (±)
Binge Drinking	15.33	2.48	16.75	2.42	15.97	5.32	18.03	3.14
Smoking	22.96	4.41	19.75	4.35	14.21	6.10	16.55	4.60
Physical Inactivity	29.62	4.98	25.61	5.18	22.30	8.07	22.42	5.97
Sleep Deprivation	36.45	3.48	34.26	3.24	31.53	4.88	31.70	3.77
Black	9.35	15.17	10.46	16.51	8.19	16.42	8.27	16.08
White	85.85	18.50	84.15	19.94	78.97	23.36	84.53	20.56
Indian/Alaskan	2.85	6.18	2.71	9.46	3.56	10.98	2.87	10.37
Asian	3.18	8.73	3.02	6.36	6.45	12.91	3.26	5.16
Poverty	14.17	8.49	13.20	8.09	14.54	12.68	12.56	9.51
Post-Sec Education	54.27	12.75	55.69	13.61	60.64	20.98	57.56	17.01
Uninsurance	15.66	5.13	15.04	6.11	14.77	8.43	13.29	7.22
Rural	N = 1072 (28.11%)		N = 3126 (26.97%)		N = 343 (25.20%)		N = 2808 (28.04%)

Racial profiles of clusters

Kruskal-Wallis tests revealed a significant difference in the distribution of races between the clusters (all p < 0.0005). Subsequent post-hoc pairwise Dunn's testing demonstrated significant differences in the distribution of the Black population between all pairs of clusters (all p < 0.0005) except between clusters 3 and 4 (p = 0.389). The same test indicated significant differences in the distribution of the White population between all pairs of clusters (all p < 0.0005) except between clusters 2 and 4 (p = 0.212). In addition, the post-hoc pairwise Dunn's testing exhibited significant differences in the distribution of the Asian population between all pairs of clusters (p < 0.0005). Similarly, all pairs of clusters exhibited significant differences in the distribution of the Indian/Alaskan population (p < 0.05) except between clusters 1 and 2 (p = 0.167) and between clusters 3 and 4 (p = 0.256).

Poverty profiles of clusters

In examining poverty across clusters, tests indicated a significant difference in the distribution of poverty among zip codes between clusters (p < 0.0005). Further analysis through post-hoc pairwise Dunn's testing underscored significant differences between all cluster pairs (all p < 0.05). Specifically, cluster 4 exhibited the lowest poverty at 12.56, while cluster 3 presented the highest at 14.54.

Post-Secondary Education profiles of clusters

The tests revealed a statistically significant difference in the distribution of post-secondary education between clusters (p < 0.0005). A post-hoc pairwise Dunn's test confirmed significant differences between all cluster pairs (all p < 0.05). Notably, cluster 3 demonstrated the highest average prevalence of individuals with post-secondary education, at 60.63%, whereas cluster 1 exhibited the lowest average, at 54.27%.

Uninsurance profiles of clusters

The test results indicated a statistically significant difference (p < 0.0005) in the distribution of uninsurance between all clusters and between all pairs of clusters (p < 0.0005 for all comparisons). Notably, cluster 1 recorded the highest average uninsurance prevalence at 15.66, and cluster 4 had the lowest average at 13.29.

Urban/Rural profiles of clusters

The χ2 contingency test indicated a lack of any statistically significant difference in the distributions of rural zip codes among the various clusters (p < 0.0005). However, post-hoc pairwise comparisons exhibited differences, revealing significant differences in the distribution of rural zip codes between clusters 1 and 3 (OR = 1.16, 95% CI: 1.01, 1.34, p = 0.042) and between clusters 3 and 4 (OR = 0.86, 95% CI: 0.76, 0.99, p < 0.031). The analysis reveals that cluster 3 had the lowest odds ratio of rural zip codes.

Binge drinking profiles of clusters

The results revealed significant differences (p < 0.0005) in the distribution of binge drinking between clusters. Subsequent post-hoc pairwise testing showed significant differences in binge drinking prevalence among all pairs of clusters (all p < 0.0005. Notably, cluster 4 exhibited the highest binge drinking prevalence, and cluster 1 showed the lowest prevalence.

Smoking profiles of clusters

The test results demonstrated a significant difference in the distribution of smoking between clusters (p < 0.0005). Subsequent post-hoc pairwise testing confirmed significant differences in smoking distribution between all cluster pairs (all p < 0.0005). Specifically, cluster 1 exhibited the highest prevalence of smoking, and cluster 1 showed the lowest prevalence.

Physical inactivity profiles of clusters

The findings from the test suggested a significant difference in the distribution of physical inactivity across different clusters. Subsequent Post-hoc Dunn's pairwise comparisons revealed significant differences in the physical inactivity prevalence between all pairs of clusters (all p < 0.0005), except between clusters 3 and 4 (p = 0.716). Cluster 1 exhibited the highest prevalence of physical inactivity, and cluster 3 exhibited the lowest prevalence.

Sleep Deprivation profiles of clusters

The results revealed a significant difference in the distribution of sleep deprivation across clusters (p < 0.0005). Post-hoc pair-wise analysis using Dunn's testing revealed significant differences in the distribution of sleep deprivation among all pairs of clusters (all p < 0.0005) except between clusters 3 and 4 (p = 0.445). Further examination of mean ranks indicated that cluster 1 exhibited the highest prevalence of sleep deprivation, and cluster 1 exhibited the lowest prevalence.

To our knowledge, this research is the pioneering study that investigates disparities in health risk behaviors across different zip codes within the United States through the application of finite mixture modeling. Finite mixture modeling, a statistical approach capable of identifying distinct subpopulations, provides a novel methodological framework for our analysis.

Our study did not find a binge drinking disparity against the Black population as is reflected in some studies [17] nor against the population without post-secondary education. Instead, it supported disparities against the White population and the population holding post-secondary degrees [18]. Also, our findings are a deviation from the research claiming the impact of poverty on higher prevalence of binge drinking [19] or those claiming similar patterns of drinking for races [20]. In addition, our findings exhibit a higher binge drinking in urban areas compared to rural areas. Also, we suggest a higher frequency of binge drinking in less socioeconomically disadvantaged zip codes, which is inconsistent with several research in this area [21].

Our study revealed distinct disparities in smoking behaviors across various racial groups. Despite differences observed between the Black and Indian/Alaskan populations compared to the White population [22], the White population still exhibited less favorable smoking behavior in comparison to the Asian population [23].

Our research indicates the existence of sleep deprivation disparity against the Black population [24]. Surprisingly, Sleep deprivation was found to be the only risk behavior positively associated with only the Black population. Additionally, the analysis revealed that rural zip codes experienced sleep deprivation more than urban areas and that poverty still plays a prominent role in the prevalence of this health risk behavior [25].

In our research, we observed a surprising disparity in physical inactivity among the Black population, and we observed a positive association between physical inactivity and poverty. Our finding supports the presence of a significant historical disparity against the Black population, as reported in other studies [26–28].

Furthermore, surprisingly, among the four races analyzed, the Black population was the only race that experienced disparities in three out of four health risk behaviors. In contrast, the Asian population exhibited a negative association with all four. Therefore, the Asian population is considered to have more favorable health risk behaviors than the other races being studied. In addition, poverty and rural status had positive associations with all three except for binge drinking, and surprisingly, post-secondary education had the exact opposite pattern.

Our study findings reveal that at least one sub-population exists within the US that has the highest prevalence of health risk behaviors, including smoking, sleep deprivation, and physical inactivity, compared to the other groups, as well as the higher than the total population’s poverty prevalence, highest uninsurance prevalence, and lowest prevalence of post-secondary education among all clusters. Consequently, these findings underscore the urgent need for targeted interventions to address the fourfold challenge of poverty, lack of insurance, low education levels, and the prevalence of health risk behaviors that are all considered the main risk factors for diseases such as cancer, cardiovascular diseases, and diabetes [29–31] and these chronic diseases have burdensome out-of-pocket costs, especially for economically disadvantaged households [32].

Our research introduced a novel approach by facilitating the identification of areas that require urgent, targeted interventions, which stands as a significant applied innovation of our work. We have gained precise insights into which subpopulations necessitate prioritized assistance. This methodological innovation not only enhances the precision of public health strategies but also ensures that resources are allocated effectively. Consequently, policymakers and public health officials can use our research to pinpoint where to begin interventions, thereby optimizing the impact of health programs and reducing health disparities in the most affected communities.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by both authors.

Ethics approval

This study utilized publicly available zip-code-level data provided by Center for Disease Control and Prevention and did require ethics approval.

Consent to participate

This study utilized publicly available zip-code-level data provided by Center for Disease Control and Prevention and did require consent to participate and consent to participate was not applicable.

Consent to Publish

This study utilized publicly available zip-code-level data provided by Center for Disease Control and Prevention and did require consent to publish and consent to publish was not applicable.

Mokdad, A. H., Marks, J. S., Stroup, D. F., & Gerberding, J. L. (2004). Actual causes of death in the United States, 2000. Jama, 291(10), 1238-1245; https://doi.org/10.1001/jama.291.10.1238.
Danaei, G., Ding, E. L., Mozaffarian, D., Taylor, B., Rehm, J., Murray, C. J., & Ezzati, M. (2009). The preventable causes of death in the United States: comparative risk assessment of dietary, lifestyle, and metabolic risk factors. PLoS medicine, 6(4), e1000058; https://doi.org/10.1371/journal.pmed.1000058.
Xu, X., Bishop, E. E., Kennedy, S. M., Simpson, S. A., & Pechacek, T. F. (2015). Annual healthcare spending attributable to cigarette smoking: an update. American journal of preventive medicine, 48(3), 326-333; https://doi.org/10.1016/j.amepre.2014.10.012.
Pampel, F. C., Krueger, P. M., & Denney, J. T. (2010). Socioeconomic disparities in health behaviors. Annual review of sociology, 36, 349-370; https://doi.org/10.1146/annurev.soc.012809.102529.
Jaferian, G., Ramezani, D., Polyak, E., & Wagner, M. (2024). EXPLORING BLOCKCHAIN'S HORIZONS IN EDUCATIONAL GAMING. INTED2024 Proceedings, 5050-5058; https://doi.org/10.1177/15554120231222578.
Cook, W. K., Kerr, W. C., Karriker-Jaffe, K. J., Li, L., Lui, C. K., & Greenfield, T. K. (2020). Racial/ethnic variations in clustered risk behaviors in the US. American Journal of Preventive Medicine, 58(1), e21-e29; https://doi.org/10.1016/j.amepre.2019.08.020.
Matthews, K. A. (2017). Health-related behaviors by urban-rural county classification—United States, 2013. MMWR. Surveillance Summaries, 66; https://doi.org/10.15585/mmwr.ss6605a1.
Powell, L. M., Slater, S., Chaloupka, F. J., & Harper, D. (2006). Availability of physical activity–related facilities and neighborhood demographic and socioeconomic characteristics: a national study. American journal of public health, 96(9), 1676-1680; https://doi.org/10.2105/ajph.2005.065573.
https://www.cdc.gov/places/
https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes/
Montgomery, D. C., Peck, E. A., & Vining, G. G. (2021). Introduction to linear regression analysis. John Wiley & Sons.
McLachlan, G. J., Lee, S. X., & Rathnayake, S. I. (2019). Finite mixture models. Annual review of statistics and its application, 6, 355-378; https://doi.org/10.1146/annurev-statistics-031017-100325.
Burgess-Hull, A. J. (2020). Finite mixture models with student t distributions: an applied example. Prevention Science, 21(6), 872-883; https://doi.org/10.1007/s11121-020-01109-3.
Ostertagova, E., Ostertag, O., & Kováč, J. (2014). Methodology and application of the Kruskal-Wallis test. Applied mechanics and materials, 611, 115-120; https://doi.org/10.4028/www.scientific.net/amm.611.115.
Dinno, A. (2015). Nonparametric pairwise multiple comparisons in independent groups using Dunn's test. The Stata Journal, 15(1), 292-300; https://doi.org/10.1177/1536867x1501500117.
Franke, T. M., Ho, T., & Christie, C. A. (2012). The chi-square test: Often used and more often misinterpreted. American journal of evaluation, 33(3), 448-458.
Chartier, K., & Caetano, R. (2010). Ethnicity and health disparities in alcohol research. Alcohol Research & Health, 33(1-2), 152.
Assari, S., Farokhnia, M., & Mistry, R. (2019). Education attainment and alcohol binge drinking: diminished returns of Hispanics in Los Angeles. Behavioral Sciences, 9(1), 9; https://doi.org/10.3390/bs9010009.
Glass, J. E., Rathouz, P. J., Gattis, M., Joo, Y. S., Nelson, J. C., & Williams, E. C. (2017). Intersections of poverty, race/ethnicity, and sex: alcohol consumption and adverse outcomes in the United States. Social psychiatry and psychiatric epidemiology, 52, 515-524; https://doi.org/10.1007/s00127-017-1362-4.
Fesahazion, R. G., Thorpe Jr, R. J., Bell, C. N., & LaVeist, T. A. (2012). Disparities in alcohol use: Does race matter as much as place?. Preventive Medicine, 55(5), 482-484; https://doi.org/10.1016/j.ypmed.2012.08.007
Glass, J. E., Williams, E. C., & Oh, H. (2020). Racial/ethnic discrimination and alcohol use disorder severity among United States adults. Drug and alcohol dependence, 216, 108203; https://doi.org/10.1016/j.drugalcdep.2020.108203.
Thomson, B., Emberson, J., Lacey, B., Lewington, S., Peto, R., Jemal, A., & Islami, F. (2022). Association between smoking, smoking cessation, and mortality by race, ethnicity, and sex among US adults. JAMA Network Open, 5(10), e2231480-e2231480; https://doi.org/10.1001/jamanetworkopen.2022.31480.
Garrett, B. E., Martell, B. N., Caraballo, R. S., & King, B. A. (2019). Peer reviewed: Socioeconomic differences in cigarette smoking among sociodemographic groups. Preventing Chronic Disease, 16; https://doi.org/10.5888/pcd16.180553;
Adenekan, B., Pandey, A., McKenzie, S., Zizi, F., Casimir, G. J., & Jean-Louis, G. (2013). Sleep in America: role of racial/ethnic differences. Sleep medicine reviews, 17(4), 255-262; https://doi.org/10.1016/j.smrv.2012.07.002.
Patel, N. P., Grandner, M. A., Xie, D., Branas, C. C., & Gooneratne, N. (2010). " Sleep disparity" in the population: poor sleep quality is strongly associated with poverty and ethnicity. BMC Public Health, 10, 1-11; https://doi.org/10.1186/1471-2458-10-475.
Wilson-Frederick, S. M., Thorpe Jr, R. J., Bell, C. N., Bleich, S. N., Ford, J. G., & LaVeist, T. A. (2014). Examination of race disparities in physical inactivity among adults of similar social context. Ethnicity & disease, 24(3), 363.
Crespo, C. J., Smit, E., Andersen, R. E., Carter-Pokras, O., & Ainsworth, B. E. (2000). Race/ethnicity, social class and their relation to physical inactivity during leisure time: results from the Third National Health and Nutrition Examination Survey, 1988–1994. American journal of preventive medicine, 18(1), 46-53; https://doi.org/10.1016/s0749-3797(99)00105-1.
King, A. C., Castro, C., Wilcox, S., Eyler, A. A., Sallis, J. F., & Brownson, R. C. (2000). Personal and environmental factors associated with physical inactivity among different racial–ethnic groups of US middle-aged and older-aged women. Health psychology, 19(4), 354; https://doi.org/10.1037//0278-6133.19.4.354.
Tao, Z., Shi, A., & Zhao, J. (2015). Epidemiological perspectives of diabetes. Cell biochemistry and biophysics, 73(1), 181-185; https://doi.org/10.1007/s12013-015-0598-4.
Peto, J. (2001). Cancer epidemiology in the last century and the next decade. Nature, 411(6835), 390-395; https://doi.org/10.1038/35077256; https://doi.org/10.1038/35077256.
Vasan, R. S., & Benjamin, E. J. (2016). The future of cardiovascular epidemiology. Circulation, 133(25), 2626-2633; https://doi.org/10.1161/circulationaha.116.023528;
Richard, P., Walker, R., & Alexandre, P. (2018). The burden of out of pocket costs and medical debt faced by households with chronic health conditions in the United States. PLoS One, 13(6), e0199598; https://doi.org/10.1371/journal.pone.0199598.

No competing interests reported.

Download PDF

Editorial decision: Revision requested
14 Aug, 2024
Reviews received at journal
12 Aug, 2024
Reviews received at journal
05 Aug, 2024
Reviewers agreed at journal
25 Jul, 2024
Reviewers agreed at journal
25 Jul, 2024
Reviewers invited by journal
25 Jul, 2024
Editor assigned by journal
22 Jul, 2024
Submission checks completed at journal
19 Jul, 2024
First submitted to journal
10 Jul, 2024

You are reading this latest preprint version

Health Risk Behaviors in the United States: A Socio-demographic Perspective

Status:

Version 1

Abstract

Introduction

Materials and Methods

Generalized Linear Models

Finite Mixture Model

Results

Binge Drinking

Smoking

Sleep Deprivation

Physical Inactivity

Racial profiles of clusters

Poverty profiles of clusters

Post-Secondary Education profiles of clusters

Uninsurance profiles of clusters

Urban/Rural profiles of clusters

Binge drinking profiles of clusters

Smoking profiles of clusters

Physical inactivity profiles of clusters

Sleep Deprivation profiles of clusters

Discussion

Declarations

References

Additional Declarations

Status:

Version 1