Dietary patterns and diabetes mellitus among people living with and without HIV: a cross-sectional study in Tanzania

{"Background": "Due to the complexity of human diets, linking single foods to health outcomes is difficult. Dietary pattern analysis offers a more holistic approach. In sub-Saharan Africa (SSA), data on dietary patterns and their association with non-communicable diseases like diabetes are limited, especially among people living with HIV. This study aimed to identify dietary patterns, factors associated with them, and their association with prediabetes and diabetes among adults with and without HIV in Mwanza, Tanzania.", "Methods": "This analysis was part of the REEHAD study, nested within the CICADA prospective cohort. Participants were enrolled between May 2019 and March 2021. Diet was assessed using a food frequency questionnaire (FFQ) adapted for the Tanzanian context, with 350+ food items aggregated into 30 food groups. Dietary patterns were derived using PCA (data-driven) and RRR (using BMI and waist circumference as response variables). Diabetes was defined by 2h OGTT (normal <7.8 mmol/L, prediabetes 7.8\u201311.0 mmol/L, diabetes \u226511.1 mmol/L). Multinomial logistic regression was used to assess associations between dietary patterns and prediabetes/diabetes, adjusting for age, sex, SES, education, physical activity, and HIV status. Of 1,174 REEHAD participants, 572 (49%) with diet data were included.", "Key Results": "The mean age was 42.6 (\u00b111.7) years, 59% were female, and 63% were people living with HIV on ART. Prediabetes prevalence was 25% and diabetes 4%. PCA identified two patterns: vegetable-rich pattern (VRP: fruits, bananas, potatoes, green/cruciferous/dark orange vegetables, tomatoes) and vegetable-poor pattern (VPP: artificially sweetened beverages, red meat, milk, chips, alcohol). RRR identified one pattern: carbohydrate-dense pattern (CDP: refined grains). Males had lower adherence to VRP (OR 0.4, 95% CI 0.3\u20130.7) but higher to VPP (OR 3.9, 95% CI 2.3\u20136.3) and CDP (OR 1.7, 95% CI 1.1\u20132.8). Higher SES was associated with higher VRP (OR 2.3, 95% CI 1.3\u20134.1) and VPP (OR 5.4, 95% CI 2.9\u201310.2) but lower CDP (OR 0.3, 95% CI 0.2\u20130.6). People living with HIV had higher VRP adherence (OR 2.0, 95% CI 1.3\u20133.2) and lower CDP adherence (OR 0.5, 95% CI 0.31\u20130.8). In fully adjusted models, high VRP adherence was associated with increased prediabetes risk (ARRR 2.12, 95% CI 1.24\u20133.66) but decreased diabetes risk (ARRR 0.26, 95% CI 0.07\u20130.95). High CDP adherence was associated with increased prediabetes risk (ARRR 1.92, 95% CI 1.12\u20133.29). VPP showed no significant association with either prediabetes or diabetes. Using HbA1c, no significant associations were observed.", "Clinical Implications": "The study reveals a complex relationship between diet and glycemic status in Tanzania. While VRP appeared protective against frank diabetes, its paradoxical association with higher prediabetes risk may reflect confounding by SES\u2014wealthier individuals consume both healthy and unhealthy foods. CDP, dominated by refined grains, was consistently linked to higher prediabetes risk. The lack of association between VPP and diabetes suggests that traditional risk factors from high-income settings may not fully apply in SSA. These findings support promoting vegetable-rich diets, particularly among men who showed lower adherence, and highlight the need for longitudinal studies to clarify causal pathways."}