Back

Modification of temperature-morbidity associations by social determinants of health

Scovronick, N.; Zhang, D.; McCann, Z. H.; D'Souza, R.; Lane, M.; Zhang, R.; Philipsborn, R.; Ebelt, S.; Chang, H. H.

2026-03-24 epidemiology
10.64898/2026.03.20.26348922 medRxiv
Show abstract

Exposure to high ambient temperature is responsible for more than 11,000 deaths and over 230,000 disability-adjusted life-years in the United States each year. However, which individuals and populations are most at risk, and why, is still not well understood. In 2015, a subset of "Z" diagnosis codes (or "Z-codes") were introduced as a standardized option for healthcare providers to document the social needs and conditions of their patients. To assess heat-related risk across social determinants of health (SDoH), we leverage these codes using a dataset of patient-level emergency department (ED) visits from seven US states. Using a bi-directional, time-stratified, case-crossover design and conditional logistic regression, we compared hospital encounters for seven different health outcomes with SDoH Z-codes at discharge to a reference group matched on age, sex, race, ethnicity, year and hospital. We investigated the following Z-code domains: inadequate housing (Z59.0, Z59.1), poverty-related (Z56.0, Z59.5-Z59.7), living alone (Z60.2), institutional living (Z65.1, Z59.3), and other problems with the social environment (other Z60 sub-codes). We calculated cumulative odds ratios (ORs) for a 3-day lag change in temperature from the 95th to 50th percentile, using ZIP code-specific temperature percentiles. Among 60,557,958 ED visits with available demographic and meteorological data, 461,468 (0.8%) included a SDoH Z-code. Across temperature metrics and outcomes, patients with SDoH Z-codes consistently showed higher associations with heat than the matched reference group without SDoH Z-codes. The largest difference was for acute kidney injury, with a ratio of ORs of 1.21 (1.10,1.33) for daily mean temperature. Notable subgroup findings included elevated kidney-related risks in patients with inadequate housing or poverty-related SDoH, increased mental health risks among those living alone, and elevated cardiovascular risks in people with other problems related to the social environment.

Matching journals

The top 11 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.