Background and objectives In recent decades there has been a change in temperature and certain weather events patterns due to climate change. Future climate is predicted to have more intense, longer lasting and frequent extreme heat episodes due to global warming. There is growing concern in the scientific community about the effects that these weather events can cause on human health. Similarly, prematurity at birth is a major cause of infant death worldwide and is associated with the onset of comorbidities during the first years of life and sequelae in adulthood. Given the importance of the perinatal period for the normal development of the baby throughout his life, in recent decades the potential risk factors that can affect pregnant women’s health have been studied. However, scientific evidence regarding the effect of exposure to elevated temperatures on pregnant women is limited. The aim of this work is to examine the relationship between exposure to elevated temperatures and the increase of premature births in the metropolitan area of Valencia during the warm season of 2006-2010. Methodology Data of births occurred during the warm season (May-September) of 2006-2010 in the metropolitan area of the city of Valencia (N = 18,026) was collected. We gathered meteorological data registered at Manises Airport station, along with daily levels of nitrogen dioxide, ozone and particulate matter of different monitors located in the study area. We conducted a time series analysis using generalized additive models with quasi-Poisson regression. As confounding variables, we introduced the terms of seasonality and long-term trend, day of week, holidays and other meteorological variables (atmospheric pressure and relative humidity). Model selection was performed based on the generalized cross-validation criterion. Flexible functions were included with natural splines in both dimensions of the relationship and lag structure. Using nonlinear distributed lag models we estimated the effect of temperature on the risk of prematurity through three different indicators (daily apparent maximum temperature, minimum temperature and mean temperature) and for heat wave episode in a time window of up to three weeks. The median value of the complete annual series was established as reference, and lag-specific and accumulative relative risks in the percentiles 50, 90 and 99 of the warm season series were estimated. The association between temperature and risk of preterm birth was also evaluated in each temperature range between the selected percentiles. Heat wave effect was assessed based on the definition of EuroHeat using the percentile cutoffs of both the monthly and annual temperature distributions. We applied as denominator the pregnancies at-risk corrected for its gestational age distribution in each day. We evaluated whether the main effect of temperature was modified by introducing the levels of air pollutants in the model. Results We observed an immediate effect with an increased risk of preterm birth up to 30% 2-3 days after exposure to maximum apparent temperatures above the 90th percentile. Similar estimates were obtained for the average temperature. For minimum temperature, a milder and long-lasting effect was obtained with an increase up to 10% in the risk 3-6 days after exposure to extreme values. The results show differences in the distribution pattern of the overall effect of temperature on each of the intervals and temperature indicators evaluated, with an immediate effect for milder interval of maximum apparent temperatures, whereas a more delayed effect with lower magnitude for the extreme interval. The immediate effect observed in the main analysis of minimum temperature was only obtained in the milder interval. And three different patterns were obtained in the case of average temperature, with an immediate effect in the milder interval, and a more delayed but similar magnitude for more extreme exposures 3 weeks before childbirth, whereas no changes in preterm risk was observed for values in the intermediate interval. No consistent effects for the exposure to heat wave episode were obtained, but our results suggested a possible effect of the days with most extreme persistent temperatures during the warm season. The introduction of pollutants into the model did not alter the main effect of temperature. Conclusions The exposure to extreme temperatures was associated with an increased risk of preterm birth in the next three weeks. The risk of preterm birth varied in magnitude and delay according to the range of temperatures and the indicator. Further research is needed to advance knowledge about the effect of extreme temperatures on the health of pregnant woman. Due to the specific meteorological, environmental and socio-demographic conditions, and the adaptation that local people experience to the usual climatic conditions, studies should be performed in specific locations. Hence, conclusive evidences should be obtained with the aim of making proposals to stakeholders in the field of public health, with the ultimate goal of designing specific prevention and mitigation plans for the potentially vulnerable population group of pregnant women.
