Identification of Dietary Pattern and Its Relationship to Cardiovascular Risk Factors A Finite Mixture Model Approach

Abstract

Cardiovascular diseases (CVD) is the leading cause of mortality and morbidity worldwide, 80% of these deaths are attributed by lower middle-income countries (LMIC). The association of diet and CVD risk factors are well studied, but these evidences are comparatively low in LMIC. The relation of individual food items with CVD risk factors has been reported adequately in Indian literature, but there is an inadequacy in literatures reporting complex dietary pattern prevailing in Indian community. In majority of the literature, the diet was predefined as vegetarian or non-vegetarian and compared with health outcomes. The current research aimed to bring out the prevailing mixed and complex diet structures, and to relate the identified diet patterns with health outcomes. The study data is a cross sectional data of the adult follow up of Vellore birth cohort (VBC) with an average age of 28 years (±1.2 SD). The birth cohort has a five-decade follow-up covering six difference phases of life starting from infancy to adulthood, which included rural and urban areas of Vellore, representing a wide range of socioeconomic strata. The current research analysed the phase V defining the young adult period of the cohort, comprising 1161 males and 1057 females. The information on anthropometry, biochemical, physical activity and diet intake were recorded. Trained field workers and standardized protocols were used to record the information from participants. The analysis was done separately for men and women due to the gender specific differences in CVD risk factors and demographic profiles. The study concludes an existing complex and mixed diet pattern prevailing in the young adult cohort of south India, and this complex nature of diet can be explained better using MBC compared to the other conventional techniques. Though significant proportional differences reported for many risk factors, the associations remained significant only for underweight, overweight, abdominal obesity and metabolic syndrome for VHMD compared to MMD, when the models are unadjusted.