Heatwaves and diabetes in Brisbane, Australia: a population-based retrospective cohort study

Available data on the effects of heatwaves on hospitalizations for diabetes and the post-discharge status of diabetics are scarce. This study aimed to assess the effects of heatwaves on hospitalizations and post-discharge deaths for diabetes, and to identify the individual- and community-level characteristics [i.e. age, gender, Socio-economic Indexes for Areas (SEIFA), and normalized difference vegetation index (NDVI)] that modified heatwave effects.

Health data were extracted from a cohort study which included patients in Brisbane, Australia, who were hospitalized due to diabetes from 1st January 2005 to 31st December 2013, and died within 2 months after they were discharged. Data on community-level modifiers, including SEIFA and NDVI (i.e. urban vegetation), were obtained from Australian Bureau of Statistics and Australian Bureau of Meteorology, respectively. Case-crossover design was used to quantify the effects of heatwaves on hospitalizations and post-discharge deaths due to diabetes. Four heatwave definitions incorporating both intensity (i.e. 90th, 95th, 97th and 99th percentiles of mean temperature distribution) and duration (2 days), as well as excess heat factor (EHF), were used. A case-only design was adopted to identify the modifiers of heatwave effects.

There were 10 542 hospitalizations for diabetes, and 513 patients died due to diabetes within 2 months after discharge. During low-intensity heatwave days (i.e. 90th percentile & 2 days), we did not observe a significant increase in hospitalizations for diabetes [9% at lag 0; 95% confidence interval (CI): -3%, 23%; P = 0.146], but we observed a significant increase in post-discharge deaths (46% at lag 2; 95% CI: 3%, 107%; P = 0.036). During middle-intensity heatwave days (i.e. 95th percentile & 2 days), hospitalizations for diabetes increased by 19% at lag 0 (95% CI: 2%, 39%; P = 0.026), and post-discharge deaths increased by 64% at lag 0 (95% CI: 6%, 154%; P = 0.027). During high-intensity heatwave days (i.e. 97th percentile & 2 days), hospitalizations for diabetes increased by 37% at lag 1 (95% CI: 11%, 69%; P = 0.004) and post-discharge deaths increased by 137% at lag 1 (95% CI: 39%, 303%; P = 0.002). When heatwave intensity increased to 99th percentile, we did not observe a significant increase in hospitalizations (-1% at lag 0; 95% CI: -38%, 59%; P = 0.870) or post-discharge deaths (79% at lag 0; 95% CI: -39%, 431%; P = 0.301). When we used EHF to define heatwaves, we observed significant increases of hospitalizations (7%; 95% CI: 1%, 15%; P = 0.039) and post-discharge deaths (68%, 95% CI: 10%, 158%; P = 0.017) during heatwave days, compared with non-heatwave days. Children and male diabetics were particularly vulnerable to heatwave effects, but we did not find any significant modification effect of SEIFA or NDVI on the associations of heatwaves with hospitalizations and post-discharge deaths due to diabetes.

Heatwaves may lead to hospitalizations of diabetics and their premature deaths. Heat-related diabetes burden in children may increase as climate warms and with increasing obesity rates in adolescents.