Spatio-temporal patterns of pneumonia in Bhutan: A Bayesian analysis

Abstract Pneumonia is one of the top 10 diseases by morbity in Bhutan. This study aimed to investigate the spatial and temporal trends and risk factors of pneumonia in Bhutan. A multivariable Zero-inflated Poisson regression using a Bayesian Markov chain Monte Carlo simulation was undertaken to quantify associations of age, sex, rainfall, maximum temperature and relative humidity with monthly pneumonia incidence and identify underlying spatial structure of the data. Overall pneumonia incidence was 96.5 and 4.57 per 1,000 populations over nine years in people aged < 5 years and ≥ 5 years, respectively. Children < 5 years or being a female are more like to get pneumonia than ≥ 5 years and males. A 10mm increase in rainfall and 1°C increase in maximum temperature was associated with a 7.2% (95% (credible interval [CrI] 0.7%, 14.0%) and 28.6% (95% CrI 27.2%, 30.1%) increase in pneumonia cases. A 1% increase in relative humidity was associated with a decrease in the incidence of pneumonia by 8.6% (95% CrI 7.5%, 9.7%). There was no evidence of spatial clustering after accounting for the covariates. Seasonality and spatial heterogeneity can partly be explained by the association of pneumonia risk to climatic factors including rainfall, maximum temperature and relative humidity.