LAUSR

Abstract Background China periodically has severe ambient particulate matter (PM) pollution, which could cause compromised pulmonary immunity in the population and result in high cardio-respiratory morbidity and mortality. Acknowledging the highly prevalent latent tuberculosis in elderly people, we hypothesise that they are more prone to tuberculosis reactivation in the presence of air pollution. Methods We examined the spatial and temporal association between PM 2·5 concentration and the number of reported cases of tuberculosis in elderly people aged 65 years and older in Hong Kong. Tuberculosis notification data in 2005–07 were collected from the Chinese Government's TB and Chest Service. Monthly and annual PM 2·5 concentration by geographical subdistrict (DCCA) for the corresponding years were derived from satellite images. Age–gender standardised notification rate (SNR) of tuberculosis cases in elderly people was calculated per DCCA/time. Their temporal association with PM 2·5 was examined by dynamic linear regression in R, and spatial association by spatial regression in GeoDa. Findings Among 17 294 new cases of tuberculosis reported in 2005–07, 6879 (40%) cases were in elderly people (2·69 men for every woman). The mean monthly SNR for elderly men was 35·4 per 100 000 persons (range 25·2–45·3), and the mean of mean PM 2·5 in DCCA was 30·5 μg/m 3 (16·1–47·1). Temporally, the SNR for elderly men was associated with PM 2·5 in lag 0, lag −1 (ie, moving PM 2·5 1 month earlier), and lag −4 to −6 months, with the most significant positive association at lag −6 months. However, no temporal association between the SNR for elderly women and PM 2·5 was observed. Spatially, the SNR was associated with PM 2·5 for elderly men in all 3 years but only for women in 2005. Interpretation We detected spatial and temporal associations between cases of tuberculosis in elderly men and PM 2·5 . Tuberculosis reactivation might have occurred after exposure to high PM 2·5 in the environment, as inferred from the lag interval of 6 months revealed in our results. The discrepancy between sexes and the effects of potential confounders such as sunshine duration, income level, and locality would need to be further investigated. Funding Research Grants Council under Theme-based Research Scheme (T32-102/14N).