IntroductionAtrial fibrillation is one of the most common arrhythmias, currently prevalent in approximately 1-3 % of the population but predicted to increase substantially due to an aging population. It is associated with a higher risk of stroke, dementia, and mortality, as well as reduced quality of life. A seasonal pattern of hospital admissions for atrial fibrillation has been observed, with increased cases during the colder months. Previous studies suggest associations between short-term decreases in ambient temperature and acute onset of atrial fibrillation episodes, but the evidence is limited.MethodWe screened 8,899 randomly selected 75-yearolds living in Stockholm for atrial fibrillation using home-based short-term ambulatory 1-lead ECG-measurements 2-4 times a day for 14 days. Screenings were carried out in 2012-2013 and 2016-2018. We used generalized estimating equations to quantify the association between ambient temperature obtained from a fixed monitoring station and risk of atrial fibrillation, adjusting for day of week, time of day and relative humidity. We explored different exposure windows and susceptible subgroups.ResultsAmong 218 participants with 469 atrial fibrillation episodes during the screening period we did not observe a statistically significant association between 24-hour prior ambient temperature and atrial fibrillation [-1% (95% CI -2%; 1%) per 1°C). We explored different exposure windows ranging from 6 to 48 hours as well as lagged exposures 12-24 hours and 24-48 hours with similar suggestive but non-significant inverse associations. A more pronounced and statistically significant association was observed for participants with diabetes [-5% (95% CI -9%; -1%) per 1°C). We did not observe any difference in the association by sex or hypertension.ConclusionOverall, short-term temperature was not strongly associated with triggering of atrial fibrillation episodes. However, there was some evidence to suggest an association between a drop in temperature and increased risk of atrial fibrillation among participants with diabetes.