The effect of extreme temperature on emergency admissions across vulnerable populations in England: an observational study

Abstract

Background Climate change poses an unfamiliar challenge to population health and health-systems resilience. Although previous studies have estimated morbidity attributable to heat or cold across cities, we provide, to our knowledge, the first large-scale, population-wide assessment of the effect of extreme temperatures on excess emergency admissions in England and among vulnerable populations, who could be disproportionately affected. Methods In this observational study, we combined all daily inpatient admissions during 2001–12 in England with meteorological data using inverse distance weighting. We exploited random daily variation in temperature experienced by hospitals and a 30-day lag period, using a distributed lag model with multiple fixed-effects controlling for seasonal factors, to examine interaction effects across diseases with age and the indices of multiple deprivation. Findings We analysed 29 371 084 emergency admissions. A day with temperature above 30°C was associated with 3·5 more admissions per hospital (SE 0·45), relative to a 10–15°C day. This increased to 14·1 excess admissions per hospital (SE 5·56) over the following 30 days, generating 786 excess admissions across England per heatwave day. A day under –5°C generated 3·8 more admissions per hospital (SE 0·33), or 966 excess admissions across England. This increased to 62·3 admissions per hospital (SE 4·83) over the following 30 days. These effects were heterogeneous across age and deprivation level. Populations older than 74 years were up to 8 times more affected by extreme temperatures. Individuals living in low-employment and low-income areas were 2–10 times more likely to be admitted during a temperature shock. These results were statistically significant (p<0·0001) and passed several robustness and falsification tests. Interpretation To our knowledge, this is the first study to determine heterogeneous morbidity risks of temperature across an entire population using a robust, near-causal methodology. Population health is negatively affected by hot and cold temperature shocks, with older people and those living in low-employment and low-income areas disproportionately affected. Findings were unadjusted for pollution and possible individual protective behaviours, providing underestimated—yet policy-applicable—risks of temperature. A heterogeneous lag effect indicates the importance of considering vulnerable populations in policy making to ensure mitigation and resilience policies are not restricted to immediate temperature shocks only and are sensitive to vulnerable populations, mitigating inequalities in access to health care.