Abstract
OBJECTIVE: To explore the direct and indirect associations of dietary patterns with hypertension using structural equation modelling (SEM). DESIGN: Factor analysis with varimax rotation was used to classify different dietary patterns and SEM was employed to investigate the associations of dietary patterns with hypertension. Total cholesterol to HDL-cholesterol (TC:HDL-C) ratio and LDL-cholesterol to HDL-cholesterol (LDL-C:HDL-C) ratio were used as observed indicator variables of the lipid latent variable. Waist circumference, body fat percentage and BMI, which were associated with hypertension, were used as observed indicator variables of the obesity latent variable. SETTING: International Chronic Disease Cohort (ICDC) that began in 2005 with the purpose of describing the frequency and determinants of chronic diseases in Jilin Province, China. PARTICIPANTS: A total of 1492 adults (40-79 years) were enrolled in the baseline study from August 2010 to August 2011. RESULTS: Hypertension prevalence in our study population was 34·9 %. It was found that the wine pattern, condiment pattern, obesity latent variable, lipid latent variable, glucose, age and family history of hypertension were factors that had an association with hypertension via SEM, and the corresponding coefficients were 0·056, 0·011, 0·230, 0·281, 0·098, 0·232 and 0·116, respectively. CONCLUSIONS: The wine pattern and lipid latent variable had positive direct associations with hypertension. The condiment pattern had a positive indirect association with hypertension via the obesity latent variable. The vegetables pattern, modern pattern and snack pattern were not associated with hypertension.