Association between empirically derived dietary patterns with blood lipids, fasting blood glucose and blood pressure in adults - the India migration study

Background: Dietary patterns (DPs) in India are heterogenous. To date, data on association of indigenous DPs in India with risk factors of nutrition-related noncommunicable diseases (cardiovascular disease and diabetes), leading causes of premature death and disability, are limited. We aimed to evaluate the associations of empirically-derived DPs with blood lipids, fasting glucose and blood pressure levels in an adult Indian population recruited across four geographical regions of India.

Methods: We used cross-sectional data from the Indian Migration Study (2005-2007). Study participants included urban migrants, their rural siblings and urban residents and their urban siblings from Lucknow, Nagpur, Hyderabad and Bangalore (n = 7067, mean age 40.8 yrs). Information on diet (validated interviewer-administered, 184-item semi-quantitative food frequency questionnaire), tobacco consumption, alcohol intake, physical activity, medical history, as well as anthropometric measurements were collected. Fasting-blood samples were collected for estimation of blood lipids and glucose. Principal component analysis (PCA) was used to identify major DPs based on eigenvalue> 1 and component interpretability. Robust standard error multivariable linear regression models were used to investigate the association of DPs (tertiles) with total cholesterol (TC), low density lipoprotein-cholesterol (LDL-C), high density lipoprotein-cholesterol (HDL-C), triglycerides, fasting-blood glucose (FBG), systolic and diastolic blood pressure (SBP and DBP) levels.

Results: Three major DPs were identified: 'cereal-savoury' (cooked grains, rice/rice-based dishes, snacks, condiments, soups, nuts), 'fruit-vegetable-sweets-snacks' (Western cereals, vegetables, fruit, fruit juices, cooked milk products, snacks, sugars, sweets) and 'animal food' (red meat, poultry, fish/seafood, eggs) patterns. High intake of the 'animal food' pattern was positively associated with levels of TC (β = 0.10 mmol/L; 95% CI: 0.02, 0.17 mmol/L; p-trend = 0.013); LDL-C (β = 0.07 mmol/L; 95% CI: 0.004, 0.14 mmol/L; p-trend = 0.041); HDL-C (β = 0.02 mmol/L; 95% CI: 0.004, 0.04 mmol/L; p-trend = 0.016), FBG: (β = 0.09 mmol/L; 95% CI: 0.01, 0.16 mmol/L; p-trend = 0.021) SBP (β = 1.2 mm/Hg; 95% CI: 0.1, 2.3 mm/Hg; p-trend = 0.032); DBP: (β = 0.9 mm/Hg; 95% CI: 0.2, 1.5 mm/Hg; p-trend = 0.013). The 'cereal-savoury' and 'fruit-vegetable-sweets-snacks' patterns showed no association with any parameter except for a positive association with diastolic blood pressure for high intake of 'fruits-vegetables-sweets-snacks' pattern.

Conclusion: Our results indicate positive associations of the 'animal food' pattern with cardio-metabolic risk factors in India. Further longitudinal assessments of dietary patterns in India are required to validate the findings.