A time series analysis of the short-term association between climatic variables and acute respiratory infections in Singapore.

Abstract

Acute respiratory infections (ARIs) are among the most common human illnesses globally. Previous studies that examined the associations between climate variability and ARIs or ARI pathogens have reported inconsistent findings. Few studies have been conducted in Southeast Asia to date, and the impact of climatic factors are not well-understood. This study aimed to investigate the short-term associations between climate variability and ARIs in Singapore. We obtained reports of ARIs from all government primary healthcare services from 2005 to 2019 and analysed their dependence on mean ambient temperature, minimum temperature and maximum temperature using the distributed lag non-linear framework. Separate negative binomial regression models were used to estimate the association between each temperature (mean, minimum, maximum temperature) and ARIs, adjusted for seasonality and long-term trend, rainfall, relative humidity, public holidays and autocorrelations. For temperature variables and relative humidity we reported cumulative relative risks (RRs) at 10th and 90th percentiles compared to the reference value (centered at their medians) with corresponding 95% confidence intervals (CIs). For rainfall we reported RRs at 50th and 90th percentiles compared to 0mm with corresponding 95% CIs. Statistically significant inverse S-curve shaped associations were observed between all three temperature variables (mean, minimum, maximum) and ARIs. A decrease of 1.1°C from the median value of 27.8°C to 26.7°C (10th percentile) in the mean temperature was associated with a 6% increase (RR: 1.06, 95% CI: 1.03 to 1.09) in ARIs. ARIs also increased at 23.9°C (10th percentile)compared to 24.9°C of minimum temperature (RR: 1.11, 95% CI: 1.07 to 1.16). The effect of maximum temperature for the same comparison (30.5°C vs 31.7°C) was non-significant (RR: 1.02, 95% CI: 0.99 to 1.05). An increase in ambient temperature to 28.9 °C (90th percentile) was associated with an 18% decrease (RR: 0.82, 95% CI: 0.80 to 0.83) in ARIs. Similarly, ARIs decreased with the same increase to 90th percentile in minimum (RR: 0.84, 95% CI: 0.80 to 0.87) and maximum (RR: 0.89, 95% CI: 0.86 to 0.93) temperatures. Rainfall was inversely associated with ARIs and displayed similar shape in all three temperature models. Relative humidity, on the other hand, exhibited a U-shaped relationship with ARIs. Our findings suggest that lower temperatures increase the risk of ARIs. Anticipated extreme weather events that reduce ambient temperature can be used to inform increased healthcare resource allocation for ARIs.