Background Extreme temperatures are triggering and exacerbating hospital admissions and health burdens; however, it is still understudied. Therefore, we evaluated the effects of the average temperature on overall hospitalisation and the average length of hospital stay.

Methods Daily area-specific age-sex stratified hospitalisation records from 2006 to 2020 were collected from the National Health Research Institutes of Taiwan. The distributed lag non-linear model was used to estimate the area-specific relative risk (RR) and 95% CI associated with daily average temperature. Overall cumulative RR was pooled from area-specific RRs using random effects meta-analysis. Temperature effects of extreme high and low thresholds were also evaluated based on the 99th (32°C) and 5th (14°C) percentiles, respectively.

Results Our findings suggested that the elderly (age ≥65 years) are vulnerable to temperature effects, while differential gender effects are not explicit in Taiwan. A higher risk of in-patient visits was seen among the elderly during extreme low temperatures (RR 1.08; 95% CI 1.04 to 1.11) compared with extreme high temperatures (RR 1.07; 95% CI 1.05 to 1.10). Overall, high-temperature extremes increased the risk of hospitalisation with an RR of 1.05 (95% CI 1.03 to 1.07) among the all-age-sex population in Taiwan. Additionally, lag-specific analysis of the study revealed that high-temperature effects on in-patient visits are effective on the same day of exposure, while cold effects occurred after 0–2 days of exposure. The average length of hospital stays can also increase with high-temperature extremes among age group 41–64 years and the elderly.

Conclusion Public health preparedness should consider the increased load on health facilities and health expenditures during extreme temperatures.