Association between PM_2.5 and daily pharmacy visit tendency in China: A time series analysis using mobile phone cellular signaling data

While much is known about the effects of PM_2.5 pollution on severe health outcomes, less is known about the heterogeneous impacts of air pollution on less severe health effects experienced by large numbers of people, such as those resulting in pharmacy visits. Based on anonymized daily pharmacy visits extracted from about 136,000 mobile phone station macrocells and 28,170 pharmacy locations, we used the generalized additive model and meta-analysis to investigate the effects of PM_2.5 exposure on pharmacy visit tendencies (PVTs), as a proxy variable of health facility utilization, across regions, groups and social contexts. Overall, PVTs increased by 0.338% (95% CI: 0.335%–0.342%) for per 10 μg/m³ increases in PM_2.5. The exposure-response curve was nonlinear, having a sharp slope below 60 μg/m³ and then flattening. Sensitivity analyses showed that results were robust to modeling choices and existence of noise in captured PVT data. At pharmacy level, higher percentage increases in PVTs were observed in near-residency pharmacies (0.474%), designated medical insurance pharmacies (0.360%), and pharmacies in urban areas (0.457%) than their counterparts (0.156%, 0.321%, 0.180%). This indicates that social contexts, including inconvenient access to the pharmacy and limited social medical insurance coverage, inhibited residents from utilizing health facilities for the health care needed. Interactions of social contexts led to strong heterogeneity among cities, between rural and urban regions as well as across communities, which call for differentiated health resources allocations to ensure health equity.