Distinct temporal effects of ambient temperature on acute aortic dissection and aneurysm rupture: a time-stratified case-crossover study

Growing evidence links temperature fluctuations to cardiovascular events, yet their effects on aortic emergencies (AEs), including acute aortic dissection (AAD) and aortic aneurysm rupture, remain incompletely characterized.1 Despite previous studies linking cold spells to incident AAD,2–4 evidence for aortic aneurysm rupture remains unexplored. This gap is notable given their rising prevalence in an ageing population.5 Furthermore, few studies have linked large-scale clinical registries with meteorological data, hindering precise identification of high-risk subgroups essential for climate-informed prevention and resource allocation. To address these gaps, we analysed the effects of temperature on incident AEs using Tokyo CCU Network Database, a prospective registry encompassing 76 major hospitals in the Tokyo metropolitan area.6,7 We identified 10 435 patients admitted with AEs between January 2013 and December 2019, after excluding those with missing onset or clinical information (n = 431) and infectious or inflammatory aetiologies (n = 34). Disease subtypes were classified using imaging and clinical findings. Meteorological parameters were obtained from the Tokyo Meteorological Station, a representative regional observatory. We employed a time-stratified case-crossover design to analyse the association between ambient temperature and AE onset.2–4 Case days were onset days, and control days were the same weekday within the same month. This design adjusted for individual characteristics and long-term or seasonal trends. Associations were estimated using conditional logistic regression with a distributed lag non-linear model up to 6 days.8 Degree of freedom (df) was set to three based on the Akaike information criterion value, with knots evenly placed using the scale of the log of lag. Following covariates were included in the model for adjustment: a natural cubic spline of relative humidity (df = 3) and a binary variable of public holiday.4 The temperature with the lowest risk was set as reference. Subgroup analyses evaluated heterogeneity across age, sex, and clinical subtypes as described previously.2,4 For sensitivity analysis, a natural cubic spline of temperature change from neighbouring day (TCN) was additionally included as a covariate. All P-values were two-sided; values