Invited Perspective: Climate Change and Reproductive Health—the Perils of Oversimplification

Abstract

This year has borne witness to unprecedented levels of heat that have shattered records around the world. 1,2 Because of their unique physiology, 3 pregnant individuals and fetuses face heightened risk of thermal stress (i.e., overheating), amplifying the likelihood of complications during pregnancy and childbirth. In this light, the effects of climate change on reproductive health demand urgent and comprehensive attention. Several studies have already looked at whether or not hotter ambient temperatures are related to adverse birth outcomes, with mixed results (e.g., the evidence looks relatively consistent for preterm birth but mixed for birth weight). 4 Yet, many of them are likely oversimplifying participants’ thermal stress exposure because human thermophysiology is not just a function of air temperature but, rather, reflects a combination of several climatic variables (e.g., temperature, solar radiation, humidity, wind) and human factors (e.g., activity, clothing). 5 One thermal index that tries to incorporate the totality of the body’s thermal response is the Universal Thermal Climate Index (UTCI). 6 To date, the UTCI has been used mostly in thermal–health warning systems and weather forecasting, 7 and its application in environmental epidemiology has been rare. However, in this issue of Environmental Health Perspectives, Nyadanu et al. use it to examine the associations between biothermal stress and the risks of small- and large-for-gestational-age (SGA and LGA, respectively). 8 Although the use of the UTCI is certainly a key strength of the study, the authors’ analytic approach should also be commended for not oversimplifying the complexities of the research. Part of the reason reproductive health is challenging to study in the context of climate change is that environmental exposures often have long-lagged effects (e.g., months). This makes it difficult to identify causal culprits. This complexity is heightened when looking at a) thermal exposures, because temperature–health effects have been shown to be either U- or J-shaped 4,9 ; and b) outcomes, such as birth weight, where both extremes (SGA and LGA) have been shown to predict later morbidity and mortality. 10 If any of these aspects are oversimplified in our models—if, for instance, we a) look at whole-pregnancy or trimester-specific exposures (when the biologically relevant window is more acute and would require an analysis involving temporally fine lags of the exposure); b) treat the exposure response as linear (when the true relationship is U-shaped); or c) treat birth weight as a continuous variable and look at only the change in its mean (when the tails are relevant)—then we can end up with biased results, or even null findings when there is a true association. Nyadanu et al. attempt to tackle these complexities by fitting distributed lag nonlinear models (DLNMs) 11,12 to a large population-based cohort from Western Australia with temporally resolved UTCI exposure. These DLNMs allowed the authors to flexibly look at time-lagged associations (weekly and monthly) with SGA and LGA while also allowing for nonlinear exposure–response curves. In doing so, they found that both extremes of biothermal stress were associated with increased risks of SGA and LGA. They also identified late pregnancy and preconception as potentially critical exposure windows. Their finding that the thermophysiological response to extreme cold is associated with adverse birth outcomes is particularly noteworthy because most studies in this field have focused on heat. 4 Although the focus on heat is not surprising given that average temperatures worldwide are soaring to alarming levels, 1,2 the climate crisis has made confronting extreme cold a regular occurrence in many places. As climate change continues to increase temperature variability and extremes, 13 it is imperative not to overlook the effects of extreme cold on reproductive health. Beyond the findings by Nyadanu et al., there is still much we can learn about the effects of biothermal stress, and I offer some future directions. First, the authors looked at absolute values of UTCI as their exposure, but this may miss part of picture given that other work has shown that temperature variability (not absolute temperature) is associated with health. 14 Both of these dimensions (i.e., the mean and variation) should be explored for a more comprehensive assessment. Second, the authors looked at birth outcomes as a proxy for fetal and newborn health, but doing so misses impacts during pregnancy. For example, it ignores the effects on pregnancy loss, which is not only a relevant health outcome but also a potential source of bias when examining outcomes in live-born children. 15 Thus, it would be worthwhile for future studies to examine health outcomes prior to birth, such as pregnancy loss or fetal biometry as measured by ultrasound. 16 Finally, the preconception period is often neglected when it comes to exploring critical exposure windows, but this work points to its potential relevance—exposures before conception (not just during pregnancy) can also affect fetal growth, possibly through epigenetic mechanisms, although these are not well understood. 17 Thus, to improve reproductive health research moving forward, preconception exposures should be considered in future studies. In summary, this study by Nyadanu et al. is important for two reasons. First, the authors make a compelling argument that extremes of thermal stress (both hot and cold) may be responsible for adverse birth outcomes, emphasizing the need to prioritize reproductive health within climate change mitigation strategies. Second, their study is an excellent example of how to ask complex questions without oversimplifying. Collaborations between climate scientists and epidemiologists are paramount in this regard because they ensure not only the use of the most up-to-date thermophysiologically relevant metrics but also the implementation of appropriate statistical models to flexibly look at these associations. It is important that we do not oversimplify in our own climate research. Otherwise, we may end up paying the cost with ineffective mitigation and adaptation policies.