The impact of temperature on anxiety and depression during spring and summer in Flanders, Belgium: a case-crossover study

The role of the environment in depression and anxiety has been poorly studied in primary care. Our objective was to assess the effect of temperature on depression and anxiety and prescription of psychopharmaceuticals in adults with (vulnerable) and without (non-vulnerable) a two-year history of mental health problems in primary care. In this time-stratified case-crossover study set in Flanders, Belgium, we used consultation data from Intego from 2012 to 2021. The primary outcomes were defined as (1) the first registration of depression, (2) the first registration of anxiety disorder, (3) the first antidepressant and (4) the first benzodiazepine prescription during the study period. Daily maximum temperatures from April to September were adjusted for relative humidity. Temperature was modelled using distributed lag non-linear models with a lag of 21 days. Associations were analysed using conditional Poisson regression, adjusted for weekday, month and year. We compared moderate and extreme temperatures defined with percentiles P 1 and P 5 for summertime cold and P 95 and P99 for summertime heat. We identified 34,166 events (any outcome) in the vulnerable and 133,866 events in the non-vulnerable population. Median maximum temperature was 20·6 °C. In the non-vulnerable group, we found an overall association of colder temperatures ( P 1) with depression (IRR 2·42, 95% CI [1·14,5·15]). In the other group, we found positive associations of colder temperatures with depression at 14-day lags ( P 1: IRR 1·85 [1·06,3·25]). Further analyses for both patient subgroups revealed a positive heat association on any outcome for the low socioeconomic status (SES) category of the vulnerable group ( P 99: IRR 2·58 [1·03,6·47]). Colder summertime temperatures increased depression rates in a vulnerable population. Additionally, we found that unusually high temperatures increased the risk of any mental health outcome in the low SES group of this population.