Socioeconomic Differentials, Health Risk Behaviors and C-reactive Protein Levels in US Young Adulthood

Abstract

While past researches have shown that elevated C-reactive protein (CRP) levels could result from both socioeconomic and behavioral factors, the effects of different health behaviors on the association between socioeconomic position and CRP levels remain unclear. A comprehensive statistical data analysis and model selection process using logistic regressions to evaluate the association among socioeconomic status, health behaviors and CRP levels for US young adulthood was performed in this study. The subjects in the study are nationally representative young adults participated in The National Longitudinal Study of Adolescent to Adult Health in 2018. The results show that household asset and education attainment are two important socioeconomic predictors for CRP levels, and health risk behaviors including heavy alcohol use, physical inactivity and unhealthy diet also contribute to the odds of elevated CRP levels for young adults with varying degrees.

1. Introduction

C-reactive protein (CRP) is a blood test biomarker for infection or chronic inflammatory disease such as cardiovascular disease and heart attack. CRP levels greater than 3mg/dL are regarded as high risk of cardiovascular disease (Nehring, 2019). Both socioeconomic and behavioral factors are associated with elevated CRP levels (Winkleby, 1992; Cdc.gov, 2019). Previous studies have shown increasing poverty is associated with elevated CRP levels (Nazmi, 2007) and there is an inverse association between childhood socioeconomic status and adulthood CRP levels (Liu, 2017). On the other hand, multiple health risk behaviors such as tobacco use, unhealthy diet, physical inactivity and heavy alcohol consumption were also proved to increase the risk of cardiovascular disease (Cdc.gov, 2019); however, studies that link behavioral factors to socioeconomic status and CRP levels are limited in a number of ways.

Positively associated with aging, the prevention and clinical intervention of cardiovascular disease are often narrow-scoped and have late initiation on the group of young adults. While the majority of young people are free of the disease, far fewer are free of the disease risk factors (Chung, 2015). Existing researches on investigating socioeconomic risk factors of elevated CRP levels largely relied on multivariable regression analysis models, failing to take into account the potential confounders from health risk behaviors and their possible interactions. A systematic statistical data analysis on how different health risk behaviors affect the association between socioeconomic differentials CRP levels is needed. Such study could provide new insights into the prevention and early intervention of cardiovascular disease in the population of young adulthood through adjusting health risk behaviors.

In this study, logistic regression models were used to assess the relationship among socioeconomic factors household asset and education attainment, health risk behaviors including tobacco use, alcohol consumption, exercise and diet and CRP levels. In addition, other covariates of interest include gender. The analysis mainly address the question that how individual health risk behaviors affect the association between socioeconomic factors and CRP levels in US young adulthood population.

2. Methods

2.1 Study Subjects

The samples were obtained from the In-Home interviews in Wave IV of The National Longitudinal Study of Adolescent to Adult Health in 2018. The participants were the original nationally representative of adolescents first interviewed in 1994 and 1995 in Wave I. A sample of 80 high schools and 52 middle schools from the US was selected by systematic sampling methods and implicit stratification (Harris, 2019). A subset of the cross-sectional data in Wave IV with 2860 respondents aged 24 to 32 was used for this study. The subject data were collected from a 90-minute interview followed by physical measurements and biospecimen collection procedures (Harris, 2019).

2.2 Variables

2.2.1 C-reactive protein (CRP)

CRP levels is the response variable in the study. High Sensitivity C-Reactive Protein (hsCRP) was measured individually through capillary whole blood collection during the biological specimens collection procedure of the study (Harris, 2019). Normal range of blood CRP levels is less than 1mg/dL; levels greater than 3 mg/dL suggest high risk of cardiovascular disease (Pearson, 2013), and CRP levels greater than 10 mg/dL are proved to be associated with acute infection about 80% of the time (Nehring, 2019). This study dichotomized CRP levels into two groups: normal (CRP levels less than or equal to 3 mg/dL) and high risk (CRP levels greater than 3 mg/dL but less than 10 mg/dL).

2.2.2 Socioeconomic factors

Two socioeconomic factors were examined as main-effect predictors in this study: household asset and education attainment. Household asset is a widely used socioeconomic status indicator in low- and middle-income settings (Kabudula, 2016). The index of the household asset in this study was measured based on self-report of the best estimate of the total value of the assets and the assets of everyone who lives in the household and contributes to the household budget (Harris, 2019). The variable was categorized into four groups: low (less than $10,000), low-mid ($10,000 to $49,999), mid ($50,000 to $99,999) and high ($100,000 or more). Education attainment was another important indicator of socioeconomic status. Higher education attainment is associated with less unemployment rate and higher earnings (Bls.gov., 2019). The five education levels in this study were grouped based on completion of education: less than high school, high school, some college, college and more than college.

2.2.3 Health risk behaviors

Four health risk behaviors were investigated as potential confounders in this study: tobacco use, alcohol consumption, exercise frequency and diet. Tobacco use was dichotomized as smoker and non-smoker. Alcohol consumption was grouped based on the number of days of drinking during the past 12 months: None or light (drink less than three times a month), moderate (drink one to two days a week) and heavy drinkers (drink more than three to five days a week). Exercise frequency was a continuous variable measured based on how many times did the participants participate in strenuous sports in the past seven days. Since the diets high in cholesterol, saturated and trans fats have been linked to cardiovascular disease (Cdc.gov, 2019), obesity was dichotomized as an indicator of diet habit, determined by body mass index (BMI) greater than or equal to 30 kg/ $m^2$

(Cdc.gov, 2019).

2.2.4        Other covariates

The study examined gender as a group-level effect on CRP levels.

2.3 Statistical analysis

The distributions of normal and high-risk CRP levels were examined by the socioeconomic factors household asset and education attainment. Chi-square test of independence and one-way ANOVA test were used to compare the distributions of the socioeconomic levels by the health behavioral variables. Logistic regression was used to evaluate the odds ratios of health risk behavioral exposures and CRP levels. In addition, the effect of gender was evaluated by performing interactions in the regression analysis. Comparisons of the associations between socioeconomic factors before and after adjusting for potential health behavioral confounders were made with the final models. Likelihood ratio test and Wald test were used to determine potential confounding factors and evaluate the effects of interactions. Goodness of fit of the models was tested by model residual deviance.

2.4 Model selection

The strategy of purposeful selection of explanatory variables was used for model selection with following steps: 1. Two initial main-effect models using sole predictor household asset and education attainment were fitted with statistical significance. 2. Set starting significance level 0.2, one of the relevant health risk behavioral variables was added to the two models and checked for statistical significance. 3. Constructed model with all significant predictors after step 2, and backward elimination was conducted to keep variables either significant at a more stringent level (p-value = 0.1) or showed evidence of being a relevant confounder. 4. Added covariates of interest that were not included to the model in step 2 and tested for statistical significance. 5. Checked for plausible interactions variables with significance level 0.05. 6. Conducted follow-up diagnostic investigations (Hosmer, 2013).

3. Results

Among the 2860 study subjects, 880 has high risk of cardiovascular disease with CRP levels greater than 3mg/dL (proportion = 0.31). The descriptive characteristics of variables of interest by household asset and education attainment were presented in Table 1 and Table 2. Compared with people in high socioeconomic status, people in low socioeconomic positions were more likely to have elevated CRP levels and live with unhealthy lifestyle including smoking, heavy alcohol use and unhealthy diet. Figure 1 and Figure 2 examined the association between CRP levels and all variables of interest. Figure 1 shows females were more likely to have high-risk CRP levels than males, and obesity could be an important predictor of elevated CRP levels; however, there was no distinct difference of CRP levels between smoker and non-smoker. Figure 2 shows CRP levels generally decrease with the elevating of socioeconomic positions and an apparent increasing trend on CRP levels with increasing alcohol consumption was found; however, there was no distinct pattern between exercise frequency and CRP levels. Table 3 and Table 4 show the model selection process mainly based on the strategy of purposeful selection of explanatory variables and likelihood ratio test with model residual deviance. For the model using house asset as the main effect, model 9 with alcohol use, diet, gender and the interaction between gender and diet was selected. For the model using education as the main effect, model 10 with alcohol use, exercise, diet, gender and the interaction between gender and diet was selected. Table 5 contains crude odds ratios and adjusted odds ratios of high-risk CRP levels by socioeconomic factors and health risk behavioral factors on controlling the effects of each other and the effect of gender.

Table 1: Descriptive statistics by household asset. For categorical variables, cell values are number of count with proportion in the parenthesis; for continuous variables, cell values are the mean of the group with proportion in the parenthesis.

Table 2: Descriptive statistics by education attainment.

Figure 1: Household asset, education attainment, alcohol use and exercise vs. CRP levels

Figure 2: Smoking, obesity and gender vs. CRP levels

Table 3: Results of fitting logistic regression models to predict CRP levels with main-effect predictor household asset (HA)

Table 4: Results of fitting logistic regression models to predict CRP levels with main-effect predictor education attainment (EA)

Table 5: Crude and adjusted odds ratios of high-risk CRP levels among all predictors. Household asset used high HA as reference group; education attainment used more than college as reference group. Alcohol consumption used none or light as reference group; diet used obesity as reference group gender used female as reference group.

4. Discussion

The study provides evidence that low socioeconomic status is significantly associated with elevated CRP levels in US young adulthood, and health risk behaviors contribute with varying degrees on the associations among household asset, education attainment and CRP levels. From the results, the odds of obtaining high-risk CRP levels for participants with high socioeconomic position according to household asset were 0.84 times the odds of the participants who were classified as low socioeconomic position. For education attainment, the odds ratio dropped to 0.78, indicating an even larger gap between the CRP levels of participants from the high and low social classes. In addition to the effects of social class, health behaviors also contributed significantly to the CRP levels among young adults. Alcohol use played an important rule on the elevated CRP levels in a prominent way. Heavy drinkers were around 1.26 times more likely to be in the high-risk CRP levels than those who did not drink or were light drinkers, regardless of the socioeconomic effects. Obesity was found as another important indicator of elevated CRP levels. Both models showed a odds ratio of high-risk CRP between obese and normal-weight people greater than 2. Exercise appeared to be a significant predictor in the model using education attainment as the main effect, showing the odds of getting high-risk CRP levels would slightly decrease with one more times of exercise in each week. The CRP levels also presented a great difference between male and female participants, with females were much more likely to have high-risk CRP levels than males. The mechanisms behind this effect should be further examined in biological researches. Finally, a significant interaction between gender and diet was found in both model, indicating the different distribution of obesity among male and female groups.

The study provided insights that alcohol abstinence education programs, promotion of healthy diet and exercise could be effective on reducing the socioeconomic disparities in elevated CRP levels and cardiovascular disease; however, several limitations of the study should be considered. First, both socioeconomic and health risk behavioral factors were measured based on self-report during the in-home interview. There might be systematic errors or recalling bias during the sample collection process. The socioeconomic differentials and health risk behaviors examined in this study were not representative enough, since there were many other factors such as drug use also played crucial roles on the elevated CRP levels. Second, the indicators selected to represent the health risk behaviors of smoking, alcohol use, physical inactivity and poor diet might be inaccurate in several ways. The data was collected in a cross-sectional study. Alcohol consumption was measured based on the drinking history of the past 12 months and physical activity was collected based on the activity in the past seven days, neither was a long-term and consistent measurement for human health risk behaviors. Besides, although unhealthy diet was closely associated with body mass index, using BMI as the indicator of diet habit was not sufficient since there were multiple manifestations on unhealthy diet such as malnutrition. Furthermore, the residual deviances of the two final models were relatively large, indicating some lack of fit and further improvements of the models should be considered.

Acknowledgment

This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.

References

• Bls.gov. (2019). Unemployment rates and earnings by educational attainment. (n.d.). Retrieved from https://www.bls.gov/emp/chart-unemployment-earnings-education.htm
• Cdc.gov. (2019). Behaviors That Increase Risk for Heart Disease. (n.d.). Retrieved from https://www.cdc.gov/heartdisease/behavior.htm
• Cdc.gov. (2019). Defining Adult Overweight and Obesity | Overweight & Obesity | CDC. (n.d.). Retrieved from https://www.cdc.gov/obesity/adult/defining.html
• Chung, R. J., Touloumtzis, C., & Gooding, H. (2015). Staying Young at Heart: Cardiovascular Disease Prevention in Adolescents and Young Adults. Current Treatment Options in Cardiovascular Medicine, 17(12). doi:10.1007/s11936-015-0414-x
• Harris, K.M., C.T. Halpern, E. Whitsel, J. Hussey, J. Tabor, P. Entzel, and J.R. Udry. (2009). The National Longitudinal Study of Adolescent to Adult Health: Research Design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design
• Kabudula, C. W., Houle, B., Collinson, M. A., Kahn, K., Tollman, S., & Clark, S. (2016). Assessing Changes in Household Socioeconomic Status in Rural South Africa, 2001–2013: A Distributional Analysis Using Household Asset Indicators. Social Indicators Research, 133(3), 1047-1073. doi:10.1007/s11205-016-1397-z
• Liu, R. S., Aiello, A. E., Mensah, F. K., Gasser, C. E., Rueb, K., Cordell, B., Burgner, D. P. (2017). Socioeconomic status in childhood and C reactive protein in adulthood: A systematic review and meta-analysis. Journal of Epidemiology and Community Health, 71(8), 817-826. doi:10.1136/jech-2016-208646
• Nazmi, A., & Victora, C. G. (2007). Socioeconomic and racial/ethnic differentials of C-reactive protein levels: A systematic review of population-based studies. BMC Public Health, 7(1). doi:10.1186/1471-2458-7-212
• Nehring, S. M. (2018, November 13). C Reactive Protein (CRP). Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK441843/
• Pearson, T. A., Mensah, G. A., Alexander, R. W., Anderson, J. L., Cannon, R. O., Criqui, M., . . . American Heart Association. (2003, January 28). Markers of inflammation and cardiovascular disease: Application to clinical and public health practice: A statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/12551878
• Winkleby, M. A., Jatulis, D. E., Frank, E., & Fortmann, S. P. (1992). Socioeconomic status and health: How education, income, and occupation contribute to risk factors for cardiovascular disease. American Journal of Public Health, 82(6), 816-820. doi:10.2105/ajph.82.6.816