Ambient temperature as a proxy indicator for carbon monoxide poisoning risk: deriving exploratory reference points for public health monitoring in Chengdu, China.
Frontiers in public health · 2026-01-01 · Observational study
Abstract
OBJECTIVE: Low ambient temperature is associated with increased carbon monoxide (CO) poisoning incidents, likely through behavioral pathways such as heating device use and reduced ventilation. This study examines the association between daily mean temperature and CO poisoning emergency medical service (EMS) responses in Chengdu, explores potential population susceptibility, estimates the attributable burden, and proposes exploratory temperature reference points for public health monitoring. METHODS: CO poisoning EMS records from Chengdu (January 1, 2013-December 31, 2025; 4,352 cases across 1,894 days) were matched with daily meteorological and air pollutant data. A time-stratified case-crossover design combined with distributed lag nonlinear models (DLNMs; max lag = 5 days) was employed. Models adjusted for relative humidity, PM2.5, wind speed, precipitation, day of week, and holidays, using median temperature (10.5 °C) as reference. Stratified analyses covered age, sex, season, and humidity. Attributable fractions were estimated via Monte Carlo simulations. RESULTS: A reverse "J"-shaped association was observed, with a cumulative lag 0-5 days minimum morbidity temperature (MMT) of 14.9 °C. Extreme cold (2.9 °C, 2.5th percentile) vs. reference was associated with a cumulative RR of 5.12 (95% CI: 1.74-15.09), peaking at lag 0 (RR = 1.79, 95% CI: 1.47-2.17). Subgroup analyses suggested potential vulnerability among adults aged 20-39 and ≥60 years. Wind speed stratification indicated a 69% higher extreme cold risk on low-wind days (RR ratio = 1.69), though formal interaction testing was non-significant (p = 0.94). Non-optimal temperatures were statistically coincident with an estimated 32.9% of EMS responses. Nominal exploratory reference points of 7.3 °C and 6.0 °C were derived; bootstrap validation (1,000 resamples) yielded median estimates of 7.0 °C (95% CI: 6.0-8.0 °C) and 6.0 °C (95% CI: 4.3-6.7 °C). CONCLUSION: Low ambient temperature serves as a measurable proxy indicator for conditions associated with elevated CO poisoning risk in Chengdu. The proposed exploratory temperature reference points (7.3 °C and 6.0 °C) may inform targeted safety messaging during winter, though prospective validation is required before operational implementation in warning systems.