Objective: To investigate the association between ambient temperature and preterm birth (PTB) and to estimate the population attributable fraction (PAF) of PTBs due to low and high temperatures.
Design: Time-stratified case-crossover design.
Setting: Japan (46 prefectures, excluding Okinawa), 2011-2020.
Sample: 214 050 PTBs registered in the Japan Perinatal Registry Network database among 1 908 168 singleton live births.
Methods: A quasi-Poisson regression model with a distributed lag nonlinear model was employed to assess the associations between daily mean temperature and PTBs for a lag of 0-27 days in each prefecture. A random effects meta-analysis was conducted by combining effect estimates from the 46 prefectures to estimate pooled relative risks (RRs). The PAFs of the PTBs due to below or above the mean of the 46 median temperatures (16.0°C) were calculated.
Main outcome measures: Preterm singleton live births.
Results: The association between daily mean temperature and PTB risk exhibited a U-shaped curve. The adjusted RRs were 1.15 (95% confidence interval [CI] 1.05-1.25) at the mean of the 1st percentiles (0.8°C) and 1.08 (95% CI 1.00-1.17) at the mean of the 99th percentiles (30.2°C) of 46 prefectures, with 16.0°C as the reference temperature. Approximately 2.3% (95% CI 0.6-4.0) of PTBs were attributable to low temperatures.
Conclusions: Both low and possibly high temperatures were associated with an increased risk of PTBs. These findings may help to inform preventive measures for pregnant women.
Keywords: attributable fraction; cold; heat; preterm birth; temperature.
© 2023 John Wiley & Sons Ltd.