Preplanned Studies: Disparities of Heatwave-Related Preterm Birth in Climate Types — China, 2012–2019

Recent systematic reviews have identified a need for studies investigating the association between high temperatures and preterm birth (PTB) across different climate types (1-2). Previous research has suggested that the association between extreme heat and PTB may vary depending on the climate (3-4), and there is also variability in the definition of heatwaves used in different studies (5). Furthermore, recent studies have revealed differences in the dominant subtypes of heat episodes across regions in China (6). To enhance our understanding of this topic, we conducted a large, nationally representative case-crossover study using data from China’s national maternal surveillance system encompassing 5,446,088 participants from 2012 to 2019. Our study aimed to examine the risk of PTB associated with 18 different definitions of heatwaves in various climate types according to the Köppen-Geiger classification. Our findings indicate that pregnant women in the arid-desert-cold climate type faced a higher risk of PTB when exposed to compound heatwaves, while those in the tropical monsoonal climate type experienced an increased risk with daytime-only heatwaves. These results provide valuable evidence for the development of targeted strategies for heat-PTB prevention in China, taking into account the disparities in heatwave-related PTB among different climate types.

We obtained data on singleton live births from China’s National Maternal Near Miss Surveillance System (NMNMSS) for the period between January 1, 2012, and December 31, 2019, as data for 2020–2022 were not available during the study period. The data included information from 438 health facilities in 325 counties or districts across China. We applied four exclusion criteria and extracted a final analytic sample of 5,446,088 births in the warm season (April to October), as described elsewhere (7). The NMNMSS was approved by the Ethics Committee of West China Second University Hospital, Sichuan University, China (Protocol ID: 2012008), and adhered to the principles of the Declaration of Helsinki. The ethical approval (Protocol ID: 2012008) also authorized the use of NMNMSS data for subsequent studies, including the current study, on maternal health.

We defined PTB as births occurring before 37 completed weeks of gestation. To assign climate types for eligible birth records, we used the addresses of the delivery health facilities for each pregnant woman, as residential addresses were not available in the NMNMSS. Climate types were classified based on the updated Köppen-Geiger climate classification. We obtained climate classification data from the 1 km global Köppen-Geiger raster product for the time period 1981–2010 from Climatologies at high resolution for the earth’s land surface areas (CHELSA) (8). Daily maximum temperature (Tmax), minimum temperature (Tmin), relative humidity, and fine particulate matter levels with an aerodynamic diameter less than or equal to 2.5 (PM_2.5) were extracted. To assign exposure, we calculated the mean grid from a zone with a 25-km radius around each pregnant woman’s delivery address. Detailed information on the calculation method can be found elsewhere (7). We defined 18 types of heatwaves, categorized into three distinct types: daytime-only (Tmax exceeds thresholds only), nighttime-only (Tmin exceeds thresholds only), and compound (both Tmax and Tmin exceed thresholds). We used six indexes, namely 75th-D2, 75th-D3, 75th-D4, 90th-D2, 90th-D3, and 90th-D4, which represent periods of equal to or more than two, three, or four consecutive days above the daily temperature thresholds at the 75th or 90th percentiles. (Supplementary Material).

In this multisite study, a space-time-stratified case-crossover design was used to examine the relationship between heatwave events and PTB. This design allowed each participant to serve as her own control and compared exposure on case days to control days (7). Time-invariant individual level confounders, as well as long-term and seasonal trends, were controlled for in the design. Conditional logistic regression models were employed to assess the association between heatwave events and PTB. The models were adjusted for the moving average of relative humidity and PM_2.5 in the last gestational week (lag06), calculated across the time window, using a natural cubic spline with 3 df. The analysis explored the variation in climate zones using an interaction term between the heatwave exposure variable and the category variable for climate types. The reference group for this analysis was the tropical monsoonal (Am) climate type. The significance of the interaction term was tested using a two-sided P-value of <0.05. Each of the three types of heatwave definitions and six indexes were modeled individually. Additionally, the lag effects of the final week prior to delivery were investigated. We estimated the odds ratio employing the maximized model goodness of fit in the seven lag days (lag0, lag1, lag2, lag3, lag4, lag5, lag6) for each heatwave definition (9).

All analyses were performed using R software (version 4.1.1, R Project for Statistical Computing, Vienna, Austria). The “survival” package (version 3.2.11) and the “splines” package (version 4.1.1) were employed for conducting the conditional logistic regression analysis.

In the final analytic sample, which included a total of 5,446,088 participants, we observed coverage of 10 different Köppen-Geiger climate types. These types encompassed two tropical [Tropical-monsoon (Am) and Tropical-savannah (Aw)], two arid [Arid-steppe-cold (BSk) and Arid-desert-cold (BWk)], three temperate [Temperate-fully humid-hot summer (Cfa), Temperate-dry winter-hot summer (Cwa), and Temperate-dry winter-warm summer (Cwb)], and three cold [Cold-fully humid-hot summer (Dfa), Cold-dry winter-hot summer (Dwa), and Cold-dry winter-warm summer (Dwb)] climate types (Supplementary Table S1). The majority of participants (68.93%) resided in temperate climate types (Table 1). The rate of PTB did not vary significantly among tropical, arid, temperate, and cold climate types (Table 1). However, pregnant women living in tropical climate types experienced higher exposure to compound and nighttime heat waves during the study period. Conversely, women in arid climate types experienced less exposure to compound heat waves (Table 2).

Pregnant women in arid BWk climate type endure a higher risk of PTB {adjusted odds ratio (AOR) range, 1.55 [95% confidence interval (CI): 1.21–1.97] to 2.11 (95% CI: 1.35–3.31)} than tropical Am climate type during exposure to compound heat waves in the 90th-D3 and 90th-D4 indexes (Figure 1). When exposed to daytime-only heat waves, pregnant women in tropical Am climate type also face a higher risk of PTB [AOR range, 1.25 (95% CI: 1.03–1.51) to 1.37 (95% CI: 1.05–1.77)] than other climate types in the 75th-D2, 75th-D3, 90th-D3, and 90th-D4 indexes. The risk of PTB for pregnant women in arid BWk climate type is associated with exposure to nighttime-only heat waves in the 90th-D4 index [AOR, 1.23 (95% CI: 1.00–1.51)], with no significant difference compared with tropical Am pregnant women.

Figure 1. 

AORs of preterm birth associated with heat waves during the last week before delivery among climate types.

Note: 18 definitions of heat waves with three types, daytime-only (only daily maximum temperature exceeds thresholds), nighttime-only (only daily minimum temperature exceeds thresholds), and compound (both daily maximum and minimum temperature exceeds thresholds) heat waves, and six indexes, 75th-D2, 75th-D3, 75th-D4, 90th-D2, 90th-D3, and 90th-D4 (periods equal to or more than two, three, or four consecutive days above the daily thresholds of temperature as 75th or 90th percentiles). Climate types and descriptions followed the updated Köppen-Geiger climate classification. All models adjusted moving average of relative humidity and PM_2.5 in the last gestational week (lag06), calculated across the time window, using a natural cubic spline with 3 df. We examined the climate zonal variation with an interaction term of heatwave exposure variable and climate types’ category variable.

Abbreviation: Ref=reference; AOR=adjusted odds ratios; CI=confidence interval; Am=Tropical-monsoon; Aw=Tropical-savannah; BSk=Arid-steppe-cold; BWk=Arid-desert-cold; Cfa=Temperate-fully humid-hot summer; Cwa=Temperate-dry winter-hot summer; Cwb=Temperate-dry winter-warm summer; Dfa=Cold-fully humid-hot summer; Dwa=Cold-dry winter-hot summer; Dwb=Cold-dry winter-warm summer.

* Statistically significant.