In this issue of JAMA Pediatrics, Ye et al 1 report an analysis of more than 1 million births in the Greater Sydney region of New South Wales, Australia, revealing a significant association between extreme heat exposure in the third trimester and preterm birth (PTB). Individuals experiencing temperatures over the 95th percentile for their residential location during their third trimester had 61% higher odds of PTB (adjusted odds ratio [OR], 1.61; 95% CI, 1.55-1.67). There was no significant association of extreme heat in the other 2 trimesters with PTB. Ye and colleagues also observed significant interaction between heat and greenness on the outcome of PTB; individuals with extreme heat exposure had lower odds of preterm birth if they resided in greener areas (interaction P < .05). We commend the authors' study and its important implications. As the world's temperatures increase, it is critical not just to document health impacts but to identify targets of intervention, such as greening, that promise to mitigate some of the negative effects of climate change.The use of publicly available data from global climate reanalysis products, merged with an enormous, local health registry dataset, is novel and impressive. Geographic information systems analyses have become more readily available and are ideal for linkage with population-based datasets, which do not suffer from referral and selection bias inherent to hospitalbased and payer databases, respectively. The association of higher temperatures with higher odds of PTB is in the expected direction. However, the magnitude of the association is higher than what is often observed in epidemiologic studies of other exposures with PTB. Maternal smoking is one of the strongest risk factors for PTB. 2 ORs for PTB among individuals who smoke, compared with those who do not, range from 1.2 in a meta-analysis of 20 cohort studies across the world to 1.59 in US population-based data of over 25 million births. 3,4 We suspect that the large effect estimate of heat exposure with PTB (OR, 1.61) reported in this study may be a result of methodologic challenges of characterizing the relevant etiologic period inherent to PTB studies and highlights the importance of methods to overcome these challenges.The investigators used a gridded spatiotemporal layer at 0.1°× 0.1°resolution, described as 9 × 10 3 m, which would translate to approximately 8.1 × 10 7 m 2 when squared, which is how the data are downloaded. 5 Addresses were collected within local statistical areas with a median area of 8.3 × 10 6 m 2 and because the gridded temperature predictions did
Related articleOpinion EDITORIAL jamapediatrics.com