Spatiotemporal effects and associations between deforestation and alcohol and tobacco use in Indonesia

PROJECT SUMMARY/ABSTRACT Deforestation is a critical global issue that leads to extensive macro-level changes in our environment. While past literature has linked sudden or expansive environmental changes to substance use, a knowledge gap exists on how deforestation may influence tobacco and alcohol consumption. The objective of this K99/R00 is to establish clearer linkages and quantify effects and associations between deforestation and tobacco and alcohol use, using Indonesia as the geographic focal area. The K99 phase is designed to achieve this objective and augment the candidate’s prior research experience by providing training in: (a) the determinants and epidemiology of substance use and mental health, (b) satellite remote sensing and image classification, (c) advanced methods for environmental epidemiology, and (d) causal designs and software for reproducibility and accessibility. This training is critical to the candidate’s long-term goal of becoming a leading epidemiologist who builds and uses rigorous statistical approaches to examine environmental exposures and social and behavioral health outcomes in the United States and in low- and middle-income countries. The proposed research will draw upon satellite imagery to detect changes in forest cover and composition and 10 years of data (2012-2022) from a nationally and sub-nationally representative, repeat cross-sectional survey. Aim 1 (K99 phase) will examine associations between forms of deforestation (wildfire and logging) and tobacco and alcohol use in the general population and among potentially vulnerable subgroups. It will subsequently estimate tobacco and alcohol use under different forest loss scenarios to estimate what would be the population-weighted burden of use attributable to incremental increases in deforestation, compared to the counterfactual of stably forested areas. Aim 2 (R00 phase) will develop a new causal inference method for estimating the “pure” effect of deforestation on tobacco and alcohol use. Economic incentives from palm oil production introduce unmeasured confounding and lead to differential exposure probability when illegal wildfires are set to clear land. A longitudinal extension of differential comparison design, also known as “negative or secondary controls,” will be created in a simulation study to contrast three groups (wildfire zones with and without palm oil plantations and stably forested zones) and parse the “pure” effect of deforestation. The new method will then be applied to the real data for estimation. A software package and tutorial will be developed to facilitate use and accessibility. Aim 3 (R00 phase) will examine which combination of factors clustered within regencies are most important in tobacco and alcohol use and model joint effects in probit Bayesian kernel machine regression. The proposed research aligns with NIEHS’ Mission, Vision, and Strategic Plan and cross-cuts all themes, specifically priority areas of: Co-exposures, Creating Knowledge from Data, Environmental Health Disparities and Environmental Justice, Emerging Environmental Health Issues, Partnerships for Action, and Promotion of Collaborative Science.