| Background Metabolic diseases are a series of dysregulated processes of body metabolism that affect insulin resistance,glucose homeostasis,lipid metabolism,and the normal function of proinflammatory immune cells and cytokines.Common metabolic disorders include hypertension,type 2 diabetes,dyslipidemia,and metabolic syndrome.The prevalence of metabolic diseases has been increasing in recent years,and health disparities have emerged across populations from different urban and rural areas,ethnic groups,and economic classes.Health disparities are defined as differences in disease morbidity,prevalence,and outcomes among different social attributes related populations.Health disparity would further increase the disease burden by inducing health inequities and social instability.Therefore,it has become an important public health task to describe the health disparity,identify its causes,and develop intervention strategies to reduce them.In Western countries,researchers have mainly focused on the differences of metabolic diseases between races and economic classes,while in China ethnicity is an important population characteristic.And ethnic differences in the health risk factors such as culture,environment,and health behaviors have generated significant health disparity for metabolic diseases.Most of the previous studies focused on the description of ethnic disparities,but few studies have focused on the causes and made reasonable recommendations on how to reduce the disparities.Specifically,although it is well known that health behaviors play an important role in the development of metabolic diseases,it is still unclear whether these factors are responsible for the ethnic disparities,and the magnitude of their contribution is never been studied before.In addition,although intervening in health risk factors might help to reduce ethnic disparities,intervention strategies with well-designed prioritization are still in lack.Finally,ethnic disparities are often caused by multiple factors,and these factors are often related to each other.However,no studies have systematically compared the performance of related statistical methods under this application scenario with multiple related mediators.Therefore,to address these issues,this study will focus on revealing ethnic disparity in metabolic diseases among populations in the Southwest region,exploring health behavior-related reasons for the health disparities,and evaluating the potential benefits of hypothetical interventions to reduce the health disparities.The findings will help to reduce the social burden of metabolic diseases and inform how to develop prioritized health intervention strategies.Objective The purpose of this study is to reveal population-based differences in the prevalence of common metabolic diseases between Han and ethnic minorities in Southwest China,to identify the contribution of health behavior factors to the ethnic disparities,and to evaluate the potential benefits of hypothetical interventions on reducing the health disparities.In addition,as health disparity research often involves multiple interrelated mediators,this study will also conduct a simulation study to evaluate the performance of three sequential mediation analysis methods under different scenarios.Specifically,the aims of the study are as follows:(1)To compare the accuracy of estimated mediation effects for three sequential mediation analysis methods,which would help to make a reasonable choice in studies that focused on the contributing factors and the corresponding effect for ethnic health disparities.(2)To describe the difference of the prevalence for common metabolic diseases and the difference of distributions for health risk factors between the Han and Minorities in southwest China,and further to investigate the contribution of health behavior factors to the ethnic disparities.This would help to reveal critical causes that contributed to the ethnic disparities.(3)To estimate the potential benefits of hypothetical interventions on different health intervenable factors on reducing ethnic disparities in metabolic diseases,and to provide a reference for developing prioritized health intervention strategies.Methods Based on the baseline survey of the China Multi-Ethnic cohort(CMEC)study conducted from June 2018 to September 2019 from,this study conducted three main analyses,as follows:(1)To compare the performance of the three statistical sequential mediation methods(regression estimation method,inverse probability weighting method,and natural effects model method),simulated data sets were generated and simulation scenarios(including different confounding effects,confounder-mediator association,mediator-mediator association,number of mediators,and mediating effect)were set.In each simulation scenario,absolute errors of estimated mediation effects were calculated,and the method with lower absolute errors would be used in the subsequent empirical study.(2)To describe the differences in the prevalence of metabolic diseases between the Han and the Minority,four common metabolic diseases(hypertension,diabetes,dyslipidemia,and metabolic syndrome)were chosen as the outcome variables.And then five health behavior factors(physical activity when not sedentary,alcohol consumption,smoking,body mass index,and DASH dietary index)were selected as mediators.The natural effects model was used to estimate mediation proportion of the five mediators,which was regarded as relative contribution to the health disparity.(3)To investigate potential benefits of hypothetical interventions on health intervenable factors,the Monte Carlo simulation based parametric g-formula method was used to replace the distribution of highly distributed health risk factors(body mass index,low education level,low economic level,DASH dietary index,smoking status,drinking status,mean sleep duration,insomnia status,non-sedentary status,and non-sedentary status)in the Minority/Han with corresponding distribution from the control ethnic group.Then the percentage of variation in prevalence of metabolic diseases attributed to a intervened health factors between ethnic groups(Attribute rate,AR)was calculated.Results(1)By comparing the absolute errors of the eatimated mediation effect among the regression based estimation,inverse probability weighting,and the natural effects model under different scenarios,this study found that the absolute errors of the inverse probability weighting and the natural effects model did not increase significantly with the increase of the mean size of confounding factors and the strength of confoundingmediating association,while the error by regression based estimation increase obviously.The increase in absolute error was similar for all three methods when the strength of the inter-mediator association and the strength of the mediator effect increased.When the number of mediator increased,the estimating errors of the three methods were similar in the two-mediator scenario;while in the three-mediator scenario,the estimation errors of the regression estimation method increased more obviously.In addition,the absolute error was similar for inverse probability weighting and natural effects model method when the mean size of confouners,confounder-mediator association,inter-mediator association strength,number of mediators,and mediator effect hold constant,while the absolute error was higher for some mediators from the regression based estimation.(2)By describing the prevalence of the four metabolic diseases and the distribution of health risk factors,this study found that the prevalence of hypertension(48.5%),dyslipidemia(26.4%),and metabolic syndrome(18.0%)in the Minority population was significantly higher than these among the Han population(46.4%,23.0%,and 13.9%),while the prevalence of diabetes was similar(6.4% among ethnic minorities vs.6.0%).Compared with the Han population,the Minority population has lower DASH dietary scores,a higher proportion of population with low education and low economic income,a lower body mass index,a lower proportion of smoking and drinking population,a lower proportion of population who slept less than 7 hours,a lower proportion of insomnia population,and a higher physical activity equivalents level during the non-sedentary time.For the three metabolic diseases with significant ethnic disparity,the estimated mediation proportion of health behaviors for hypertension is: DASH dietary index(51.46%)> alcohol consumption(20.06%)> smoking(19.17%)> body mass index(8.99%)> non-sedentary physical activity equivalents(0.30%).For dyslipidemia: DASH dietary index(60.22%)> smoking(26.40%)> alcohol consumption(7.91%)> non-sedentary physical activity equivalents(5.45%).And for metabolic syndrome: DASH dietary index(68.58%)> nonsedentary physical activity equivalent(15.11%)> smoking(12.22%)> alcohol consumption(4.07%).(3)Hypothetical intervention effects for 9 health interference factors were estimated using the parametric g-formula.To reduce the ethnic disparity for hypertension,the ranked potential benefits for these factors were: DASH dietary index 0.552(0.493,0.612)> low education 0.064(0.053,0.075)> low economic level 0.028(0.024,0.032).For reducing the disparity of dyslipidemia,the rank was: DASH dietary index 0.201(0.186,0.217)> low education 0.019(0.014,0.023)> low economic level 0.000(0.000,0.004).And for metabolic syndrome was: DASH dietary index 0.203(0.191,0.214)> low education 0.073 (0.069,0.078)> low economic level 0.000(0.000,0.000).For further reduing the disease prevalence among the Han population,the ranked potential benefits of these factors for hypertension were: smoking status 0.176(0.158,0.194)> drinking status 0.131(0.120,0.143)> non-sedentary physical activity 0.037(0.032,0.042)> body mass index 0.036(0.015,0.057)> average sleep time less than 7 hours 0.000(0.000,0.001)= Insomnia status 0.000(0.000,0.005).For dyslipidemia was: smoking status 0.074(0.069,0.079)> non-sedentary physical activity 0.033(0.031,0.036)> alcohol consumption status 0.021(0.018,0.024)> mean sleep duration less than 7 hours 0.000(0.000,0.001)= insomnia status 0.000(0.000,0.000)= Body mass index 0.000(0.000,0.000).And for metabolic syndrome as: non-sedentary physical activity 0.044(0.042,0.047)> smoking status 0.022(0.020,0.023)> alcohol consumption status 0.002(0.000,0.003)> mean sleep duration less than 7 hours 0.000(0.000,0.001)= insomnia status 0.000(0.000,0.000)= Body mass index 0.000(0.000,0.000).Conclusion The prevalence of hypertension,dyslipidemia,and metabolic syndrome was higher among the Minority population when compared with the Han population in the Southwest region of China,while the prevalence of diabetes was similar among the two ethnic groups.Adherence to the DASH dietary pattern is an important mediator for ethnic disease disparity.Promoting adherence to the DASH dietary pattern among the Minority population is most likely to reduce ethnic disparity of the three diseases.Smoking cessation interventions among the Han population might be the most effective way to further reduce the prevalence of hypertension and dyslipidemia.Increasing the time of non-sedentary physical activity among the Han population might be the most effective way to reduce the prevalence of metabolic syndrome.In addition,in the sequential multimediatior scenario,the estimation error was larger for the regression based method when strong confounding effect or high-dimensional mediator exist.The inverse probability weighting and natural effects model methods have similar estimation errors in most scenarios. |
PDF Full Text Request