A Case-Crossover Study of Extreme Heat and Psychiatric Emergency Encounters Among Vulnerable Pregnant People

Abstract

Background: Extreme ambient heat has been linked to maternal and neonatal mortality and morbidity. However, less is known regarding the impact of extreme ambient heat on mental health outcomes, particularly among socially vulnerable pregnant people. Objectives: We aimed to estimate the effect of ambient heat exposure on psychiatric emergency services encounters among pregnant people served by a public program in the metropolitan Boston area. We hypothesised that days of extreme heat would have higher odds of an encounter with psychiatric emergency services when compared to normal temperature days. Methods: Using electronic health records from the Boston Emergency Services Team program, we identified 861 geocodable encounters for pregnant people who utilised psychiatric emergency services between 2005–2009 and 2017–2021. Using a time-stratified case-crossover design, we matched each case (psychiatric emergency services encounter) with 3–5 control dates (matched by day-of-week and month). We obtained daily temperature (degrees Celsius) measures (mean, minimum, maximum) at an 800 m gridded resolution for cases/controls. Extreme heat exposure was defined at the 95th percentile, and exposure-response relationships were analysed using distributed lag nonlinear models with a 5-day lag period. We used conditional logistic regression models to estimate the association between ambient temperature exposure and odds of psychiatric emergency services encounters, controlling for U.S. federal holidays, relative humidity, and week of case/control date. Effect measure modification by time was examined (e.g., 2005–2009, 2017–2021). Results: Among socially vulnerable pregnant people, we found no evidence of an association between days of extreme heat and the odds of psychiatric emergency services encounters, compared to normal temperature days. Conclusions: This study does not support our a priori hypothesis. Future research is needed to confirm to what extent this association is truly null, as opposed to being obscured by selection bias into psychiatric emergency services, among socially vulnerable pregnant people.