Quantifying the influence of location of residence on blood pressure in urbanising South India: a path analysis with multiple mediators

Abstract

Abstract Objectives: The current study aims to estimate the causal effect of increasing levels of urbanisation on mean SBP, and to decompose the direct and indirect effects via hypothesised mediators. Methods: We analysed data from 5, 840 adults (≥ 18 years) from the Andhra Pradesh Children and Parents study (APCAPS) conducted in 27 villages in Telangana, South India. The villages experienced different amounts of urbanisation during preceding decades and ranged from a rural village to a medium sized town. We estimated urbanisation levels of surveyed villages by combining remote sensing data of night-time light intensity (NTLI), measured by unitless digital numbers, with satellite imagery and ground surveying of village boundaries. We performed mediation analysis using linear mixed-effects models with SBP as the outcome, log-transformed continuous NTLI as the exposure, and three composite mediators summarising information on (i) socio-demographics (e.g., occupation and education); (ii) lifestyle and mental health (e.g., diet and depression); (iii) metabolic factors (e.g., fasting glucose and triglycerides). All models fitted random intercepts to account for clustering by villages and households and adjusted for confounders. Results: The NTLI range across the 27 villages was 62 to 1081 (4.1 to 7.0 on the log scale). Mean SBP was 122.7 mmHg (±15.7) among men and 115.8 mmHg (±14.2) among women. One unit (integer) log-NTLI increase was associated with a rise in mean SBP of 2.1 mmHg (95% CI 0.6, 3.5) among men and 1.3 mmHg (95% CI 0.0, 2.6) among women. We identified a positive indirect effect of log-NTLI on SBP via the metabolic pathway, where one log-NTLI increase elevated SBP by 4.6 mmHg (95% CI 2.0, 7.3) among men and by 0.7 mmHg (95% 0.1, 1.3) among women. There was a positive indirect effect of log-NTLI on SBP via the lifestyle and mental health pathway among men, where one log-NTLI increase elevated SBP by 0.7 mmHg (95% CI 0.1, 1.3). Observed negative direct effects of log-NTLI on SBP and positive indirect effects via the socio-demographic pathway among both genders; as well as a positive indirect effect via the lifestyle and mental health pathway among women, were not statistically significant at the 5% level. The sizes of effects were approximately doubled among participants ≥40 years of age. Conclusion: Our findings offer new insights into the pathways via which urbanisation level may act on blood pressure. Large indirect effects via metabolic factors, independent of socio-demographic, lifestyle and mental health factors identify a need to understand better the indirect effects of environmental cardiovascular disease (CVD) risk factors that change with urbanisation. We encourage researchers to use causal methods in further quantification of path-specific effects of place of residence on CVDs and risk factors. Available evidence-based, cost-effective interventions that target upstream determinants of CVDs should be implemented across all socio-demographic gradients in India.