Application Of Structural Equation Modeling To Explore The Relationship Between Marital Status And Blood Pressure

Background:Hypertension is one of the most important risk factors for cardiovascular diseases. The prevalence of hypertension was remarkable increasing according to the figure from the three national hypertension survey in China. Data from the 2002 National Nutrition and Health Survey showed that 18.8%of adult Chinese was hypertension. Moreover, Survey from WHO found that every year there are approximately 17 million deaths attributed to hypertension. In 2003, the figure reached 0.27 million for United States. Thus, Hypertension has been a major worldwide public health problem. Hypertension was caused by several factors and numbers of studies have proved that hypertension was correlated with obesity, smoking, drinking, unhealthy lifestyles, low education and income, occupation, psychology factors, social environment, dyslipidemia and family history of hypertension. In addition, early prospective epidemiological studies have described a positive association of the social psychology indicators (eg, helplessness, social network, depression, etc.) with hypertension. Although the relationship between marital status and social psychology indicators was declared, studies about marital status and hypertension were lack in China. With the views transmitting from traditional medical model to bio-psycho-social medical model, several new social and psychological indexes occur and most of those cannot be directly or incorrectly measured. Since the traditional analysis method cannot describe the effect of these potential variables on outcomes, the structural equation model with multiple dependent variables and multiple independent variables will be a suitable statistical approach to solve this problem. In this study, we will explore the association between marital statuses with blood pressure using the structural equation model. Currently, the structural equation model is rarely applied in the field of epidemiology of chronic disease. Our study, as a useful attempt of this method, will further discuss the application from different aspects.Methods:In this study, the data of control group in INTER-HEART study were selected. The measurements of blood pressure of all subjects were conducted in a quiet environment. The blood pressures of each subject were measured twice, and the average was recorded. Hypertension was defined according the WHO criteria (SBP>=140mmHg and/or DBP>=90mmHg and/or currently on medicine). During analysis, the marital status was separated into two groups according to present of misfortune during marriage after exclusion of never married:misfortune marriage (widowed, separated, and divorced) and normal marriage (currently married common law/living with partner). The collected data mainly related nine risk factors of hypertension:age, sex, body mass index, waist-hip ratio, smoking, diabetes mellitus, ApoB/ApoAl, TG, HDL-C, TC, physical activity, amount of vegetable and fruit consumed per week, amount of meat consumed per week, social psychology index, education, yearly income of family, family history of diseases (hypertension, diabetes, MI, angina and stroke). Multivariate Logistic regression model was used to explore the association of marital status and hypertension by adjusting the above risk factors. In process of adjustment, the nine risk factors were distinguished in different combinations according to the degrees of these risk factors associated with hypertension. Different combinations were used to adjust in different model.In addition, Structural equation modeling was applied to test the implied latent variable mediation model. First, in order to fit the model, we need to perform data cleaning as appropriate. Second, to find the latent variable using exploratory and confirmtary factor analysis. Finally, complete the model by model specification, estimation, evaluation of fit, model modification and interpretation. All statistical analysis was performed using SAS(＠9.13 and LISREL＠8.70 software and P<0.05 was considered statistically significant.Results:Of the 2937 subjects in this study,2039 (69.5%) were men. Totally, there were 188 misfortune marriages [169(5.75%) for widowed,5(0.17%) for divorced andl4 (0.48%) for desperate]. The median age for the whole population was 62(range:28-87, IQR:25-75). To describe the distribution of different risk factors across age, we divided the participant into 5 groups (≤45,46-55,56-65,66-75 and>75). The proportion of people aged older than 65 among misfortune marriage group was higher than that among normal marriage group. Among the female who aged between 56 and 65, persons who have misfortune marriage had a higher systolic blood pressure than those who have normal marriage (123.70±10.53mmHg versus 127.53±17.93mmHg, respectively.) the same pattern was observed in diastolic blood pressure (77.40±5.66mmHg versus 79.11±9.09mmHg, respectively). Although the 95%confidence intervals for the marital status associated with systolic blood pressure>=140mmHg did not cross 1 after using several analytical strategies, the point estimated odds ratio was all greater than 1. While doing subgroup analysis, we also found an OR of 1.01 (95%CI:0.70-1.62) for men and 1.33(95%CI:1.15-2.12) for female. With respect to diastolic blood pressure, the OR and its 95%confidence interval for marital status were 1.06(0.85-1.32) after adjusting the over risk factors. The estimated OR was 1.01(95%CI:0.70-1.62) among the male population and 1.13 (95%CI:0.92-1.54) among female population.From the result of SEM, marital status was not only associated with blood pressure directly, but also indirectly through socio-economic status. Four pathways were observed in the whole population with the direct path coefficient of 0.14 and 0.168 for the entire path coefficient. Among the male population, there are seven pathways and the direct path coefficient was 0.14 and 0.168 for the entire path coefficient. While only 3 pathways were found in female population with the direct path coefficient of 0.23 and 0.254 for the entire path coefficient. In male population, marital status can affect lifestyle while this association did not exist among female. Furthermore, a negative association was found between lifestyle and obesity.Conclusion:Misfortune marriage was a risk factor for hypertension in female population through systolic blood pressure, while this association was not observed in male. Marital status was not only associated with blood pressure directly, but also indirectly through socio-economic status.