SummaryBackgroundThe associations of ambient temperature with acute myocardial infarction (AMI) have seldom been examined based on the time of symptom onset.MethodsWe conducted a time-stratified case-crossover study among 1,046,773 eligible AMI patients from 2,093 hospitals in 324 Chinese cities from January 1, 2015 to June 30, 2021, after excluding those transferred from other hospitals or having not reported the time of symptom onset. Hourly exposure to ambient temperature was calculated as multiple moving 24-h averages (days) before hourly onset of AMI symptoms. Conditional logistic regression and distributed lag non-linear models with a duration of 0–21 days were used to estimate the cumulative associations of non-optimum temperature with AMI onset and the corresponding disease burden nationally. Subgroup analyses by region and period were conducted. Specifically, cities with and without centralized heating system were classified into heating and non-heating regions, respectively. The whole year in heating region was divided into heating and non-heating periods based on the duration of centralized heating in each city.FindingsAlmost monotonically increasing risks were observed for both overall AMI and its two subtypes when ambient temperature declined. The effects of extremely low temperature occurred immediately on the concurrent day, and lasted up to almost 3 weeks. The excess risks of AMI onset associated with non-optimum ambient temperatures were observed during the whole year in the non-heating region and non-heating period in the heating region, but not during heating period. Specifically, odds ratios of AMI onset associated with extremely low temperature cumulated over 0–21 days were 1.24 (95% CI: 1.13–1.37), 1.46 (95% CI: 1.20–1.76), and 1.62 (95% CI: 1.46–1.81) in the heating region during non-heating period, in the non-heating region during winter and non-winter period, respectively. The heat effects on AMI onset were very modest and transient. Totally, 13.26% of AMI cases could be attributable to non-optimum temperatures nationally. The burden of AMI attributable to non-optimum temperature was much smaller in heating region than in non-heating region. Somewhat stronger effects were observed in females and patients aged older than 65.InterpretationThis nationwide study provided robust evidence that non-optimum ambient temperature may significantly trigger AMI onset, and for the first time estimated the disease burden after accounting for spatial and seasonal heterogeneity. Centralized heating might substantially mitigate AMI burden due to non-optimum temperature.FundingShanghai International Science and Technology Partnership Project, National Natural Science Foundation of China, Talent Training Program of Zhongshan Hospital, Fudan University.