American Journal of Epidemiology © The Author 2016. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. Vol. 183, No. 10 DOI: 10.1093/aje/kwv295 Advance Access publication: March 31, 2016 Original Contribution Association Between High Ambient Temperature and Risk of Stillbirth in California * Correspondence to Dr. Rupa Basu, Air and Climate Epidemiology Section, Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, 1515 Clay Street, 16th Floor, Oakland, CA 94612 (e-mail: Rupa.Basu@oehha.ca.gov). Initially submitted July 1, 2015; accepted for publication October 19, 2015. Recent studies have linked elevated apparent temperatures with adverse birth outcomes, such as preterm delivery, but other birth outcomes have not been well studied. We examined 8,510 fetal deaths (≥20 weeks’ gestation) to estimate their association with mean apparent temperature, a combination of temperature and humidity, during the warm season in California (May–October) from 1999 to 2009. Mothers whose residential zip codes were within 10 km of a meteorological monitor were included. Meteorological data were provided by the California Irrigation Management Information System, the US Environmental Protection Agency, and the National Climatic Data Center, while the California Department of Public Health provided stillbirth data. Using a time-stratified case-crossover study design, we found a 10.4% change (95% confidence interval: 4.4, 16.8) in risk of stillbirth for every 10°F (5.6°C) increase in apparent temperature (cumulative average of lags 2–6 days). Risk varied by maternal race/ethnicity and was greater for younger mothers, less educated mothers, and male fetuses. The highest risks were observed during gestational weeks 20–25 and 31–33. No associations were found during the cold season (November–April), and the observed associations were independent of air pollutants. This study adds to the growing body of literature identifying pregnant women and their fetuses as subgroups vulnerable to heat exposure. ambient temperature; California; case-crossover studies; environmental factors; fetal death; mortality; stillbirth Abbreviations: CI, confidence interval; PM2.5, particulate matter less than 2.5 µm in diameter; ppb, parts per billion; ppm, parts per million; SES, socioeconomic status. Common causes of stillbirth include obstetrical conditions, placental abnormalities, fetal genetic/structural abnormalities, infection, umbilical cord abnormalities, hypertensive disorders, and other maternal medical conditions (9), but important risk factors for stillbirth remain largely unknown. In this study, we examined the relationship between apparent temperature and stillbirth in California during the warm season from 1999 to 2009. California’s large and diverse population and its vast network of meteorological and air pollution monitors provide an ideal setting in which to study temperature-related health outcomes while targeting vulnerable subgroups. We considered potential seasonal differences, as well as variations in associations by maternal demographic factors such as age, education, race/ethnicity, and fetal sex. We also examined potential confounding and effect modification according to a number of criteria air pollutants. The relationship between apparent temperature and adverse birth outcomes has only recently been explored. Most previous studies have linked temperature or seasonality with low birth weight and preterm birth in the United States and worldwide (1–3). Other adverse birth outcomes, such as stillbirth (also known as fetal death), have only rarely been considered, and only outside of the United States; both cold temperatures in Sweden (4) and hot temperatures in Australia (5) have been examined. Although there is no standardized definition of stillbirth, it is typically defined as death of a fetus at ≥20 weeks of gestation or ≥28 weeks of gestation, depending on geographic location. The rate of stillbirth has been fairly consistent in the past decade, and in California in 2010 it was reported to be 5.1/1,000 births (6). Stillbirth may be associated with higher monetary costs than livebirth (7) and can contribute to symptoms of depression and anxiety that may continue through subsequent pregnancies (8). 894 Am J Epidemiol. 2016;183(10):894–901 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 Rupa Basu*, Varada Sarovar, and Brian J. Malig Temperature and Stillbirth in California 895 Identification of cases Exposure classification The study population consisted of stillbirths occurring between January 1, 1999, and December 31, 2009, in California. Fetal Death Statistical Master Files data were provided by the Office of Health Information Research of the California Department of Public Health (16). The State of California defines stillbirth as the death of a fetus whose development has advanced to the 20th week of gestation or beyond, prior to complete expulsion or extraction from the mother (California Code of Regulations, Title 17, Section 916). The following information was abstracted from fetal death certificates: sex of the fetus, date of stillbirth, gestational age of the fetus, date of the last menstrual period, maternal residential zip code, maternal age, maternal education, and maternal race/ethnicity. When information on the last menstrual period was missing from the original data, it was estimated by taking the difference between the date of delivery and gestational age given in days. To avoid the potential for the fixed cohort bias (17), we limited the data to a cohort conceived between 20 weeks prior to January 1, 1999, and 44 weeks prior to December 31, 2009. Meteorological data were obtained from 3 different sources: the California Irrigation Management Information System (10), the US Environmental Protection Agency (11), and the National Climatic Data Center (12). As described previously (13), we calculated mean apparent daily temperature, a combination of temperature and relative humidity, in degrees Celsius (°C) and then converted the data to degrees Fahrenheit (°F). Since we were interested in warm-season associations, we limited our main analysis to May through October 1999– 2009. We considered the number of stillbirths with maternal zip codes located within 10 km, 10–20 km, and 20 km of a meteorological monitor in the same climate zone to maximize the number of cases as well as refine our exposure assessment to reduce potential misclassification. Sixteen climate zones have been previously defined by the California Energy Commission, based on weather, energy use, and other factors related to climate (14), as shown in Figure 1. Using data provided by the California Air Resources Board (15), we evaluated potential confounding by maximum 1-hour measurements for carbon monoxide (CO), nitrogen dioxide (NO2), ozone (O3), and sulfur dioxide (SO2) in separate models along with apparent temperature. These analyses were restricted to case mothers living within 5 km of a carbon monoxide or nitrogen dioxide monitor, within 10 km of an ozone monitor, and within 20 km of a sulfur dioxide monitor. Mean Apparent Temperature for Cases (°F) 51.7–58.7 58.8–65.4 65.5–72.0 72.1–78.7 78.8–85.4 1 11 16 2 12 3 13 4 14 5 9 6 8 0 30 60 120 180 240 10 7 Miles 15 14 Figure 1. Mean apparent temperature for stillbirth cases (°F) in the 16 climate zones defined by the California Energy Commission, California, May–October 1999–2009. °C = (°F − 32) × 5/9. Am J Epidemiol. 2016;183(10):894–901 Statistical analysis The time-stratified case-crossover method was used to analyze the data (18). This study design is ideal for short-term exposures with acute relationships, and it controls for both known and unknown confounders by using each woman as her own control. Since control periods were selected every seventh day during the same month as the stillbirth (i.e., case period), we were able to inherently control for day of the week and long-term time trends. There could be a maximum of 4 control periods per case, with a minimum of 7 days and a maximum of 28 days before or after the case period occurred. We conducted univariate conditional logistic regression analysis, using a continuous measure of apparent temperature, to obtain an effect estimate. These analyses were performed using the PHREG procedure in SAS statistical software (SAS Institute, Inc., Cary, North Carolina) (19). We calculated an odds ratio and corresponding 95% confidence interval for each estimate. The results were scaled to represent percent change in mortality risk per 5.6°C (equivalent to 10°F) increase in apparent temperature, using the calculation (odds ratio − 1) × 100%. We combined the estimates calculated for each climate zone with at least 50 stillbirth cases into 1 overall estimate for California via meta-analytical techniques in a random-effects model, using the rmeta package for R statistical software (R Foundation for Statistical Computing, Vienna, Austria) (20, 21). Because previous literature has suggested that it may take an average of 48 hours in the third trimester or 70 hours during an entire pregnancy (with median 7 hours) for a deceased fetus to be expelled from the womb (22, 23), we considered several lag periods for apparent temperature, beginning with lag 2 (2 days prior to the date on which fetal death was recorded). We considered previous days 3, 4, 5, and 6 and cumulative lag periods of 2–3, 2–4, 2–5, and 2–6 days (an average of 2, 3, 4, 5, and/or 6 days prior), as was done in a previous study of air pollution and stillbirth (24). For these analyses, cases were restricted to days for which apparent Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 METHODS 896 Basu et al. Table 1. Demographic Characteristics of the Study Population by Climate Zone, California, May–October 1999–2009 Maternal Ethnicity, % California Climate Zonea Non-Hispanic White Non-Hispanic Black Hispanic Maternal Age Group, % Non-Hispanic Asian <25 Years 25–34 Years ≥35 Years Maternal Education, % High School or Less At Least Some College Fetal Sex, % Male 15 73 0 20 7 47 47 7 60 40 40 2 121 48 2 46 2 30 46 24 64 36 49 3 743 22 20 38 18 31 46 22 56 44 57 4 389 25 3 46 25 23 53 24 51 49 51 5 93 26 5 65 4 33 55 12 64 36 48 6 342 42 4 46 7 23 46 31 48 52 50 7 506 26 12 51 10 37 43 20 57 43 53 8 1,389 10 26 57 6 34 46 20 65 35 53 9 1,530 19 5 67 10 27 47 26 66 34 50 10 886 25 14 54 5 36 46 18 64 36 54 54 11 181 63 2 25 7 37 46 17 54 46 12 1,166 35 16 37 11 36 44 20 63 37 53 13 778 22 12 59 5 43 41 16 73 27 53 14 230 35 22 40 2 45 41 14 69 31 53 15 156 15 5 77 3 33 36 31 64 36 54 16 17 59 0 35 0 35 41 24 71 29 65 8,542 24 13 52 9 33 46 21 63 37 52 Climate zones 1 and 16 did not have enough cases to be included in any analysis. temperature data were available for all of the lags considered. We tested for nonlinearity by evaluating the significance (P < 0.05) of an additional squared term for apparent temperature added to our models. We stratified the results by maternal demographic factors and fetal sex (male, female) to assess potential effect modification in separate models. Specifically, we analyzed maternal age (<25, 25–34, or ≥35 years), highest level of maternal education completed at the time of giving birth (high school graduation or less, at least some college), and maternal race/ ethnicity (Hispanic, non-Hispanic white, non-Hispanic black, or non-Hispanic Asian). We also examined whether effect estimates differed by length of gestation (20–25, 26–30, 31–33, 34–36, or 37–44 weeks) or between preterm (20–36 weeks’ gestation) and full-term (37–44 weeks’ gestation) births. For all pollutants examined, lag times of an average of 2–6 days were used. To investigate confounding by air pollutants, we conducted bivariate analyses using a subset of the data in which we had both the relevant data for each pollutant and apparent temperature. We evaluated potential effect modification by the same pollutants by adding an interaction term with apparent temperature and each pollutant in separate models. Finally, we compared our results for women living within 10 km of a meteorological monitor with results for women living within 20 km of a monitor and women living between 10 km and 20 km to see whether obtaining more cases by going further from the monitor influenced the effect estimates. We also considered the cold season of November through April for 1999–2009 to compare the findings with our warm-season results. Prior to initiation of this study, the California Committee for the Protection of Human Subjects approved the research protocol. RESULTS A total of 26,355 stillbirths occurred in California between 1999 and 2009. Of these, the mothers of 22,580 stillborn infants had a meteorological monitor within 20 km of their residence. During the warm season, there were 8,542 stillbirths with monitors within 10 km of maternal residential zip codes; descriptive statistics are provided in Table 1. Since climate zones 1 and 16 did not have at least 50 stillbirth cases for data analyses, we excluded them. Our final study population of 8,510 stillbirths in 14 California climate zones comprised participants who were primarily Hispanic (52%), were most commonly between 25 and 34 years of age (46%), and were mostly not educated beyond high school (63%). Most fetuses were male (52%). Table 2 gives the apparent temperature and air pollution exposures for lag 2–6 of the case periods. For the entire study population during the warm season, average apparent temperatures were 68.4°F (5th and 95th percentiles, 55.9°F, 82.2°F). The average ambient carbon monoxide concentration was 1.0 parts per million ( ppm) (5th and 95th percentiles, 0.3 ppm, 2.5 ppm), the nitrogen dioxide concentration was 34.0 parts per billion ( ppb) (5th and 95th percentiles, 10.5 ppb, 69.5 ppb), the ozone concentration was 57.0 ppb (5th and 95th percentiles, 29.5 ppb, 96.5 ppb), and the sulfur dioxide concentration was 4.0 ppb (5th and 95th percentiles, 0.5 ppb, 11.5 ppb). Pearson correlation coefficients Am J Epidemiol. 2016;183(10):894–901 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 1 Total a No. of Stillbirths Temperature and Stillbirth in California 897 Table 2. Mean Apparent Temperature and Mean Air Pollutant Concentrations (Cumulative Average of Lags 2–6 Days) for Stillbirth Cases, by Climate Zone, California, May–October 1999–2009 Climate Zonea Apparent Temperature, °Fb Case-Control Difference in Apparent Temperature,c °F 5th and 95th Percentiles 1 51.7 44.4, 55.9 2.6 2 59.4 51.2, 67.9 3.8 3 59.3 51.8, 66.3 4 63.3 54.2, 72.1 5 57.9 6 Nitrogen Dioxide, ppb 5th and 95th Percentiles Mean NA NA 0.6 0.3, 1.2 3.3 0.7 0.2, 1.8 4.4 0.9 0.4, 1.8 52.3, 63.5 2.5 0.4 65.2 57.3, 72.7 3.1 7 67.7 58.9, 76.4 8 69.0 60.1, 78.2 9 69.8 10 11 Ozone, ppb Sulfur Dioxide, ppb 5th and 95th Percentiles Mean NA NA NA NA 0 0, 0 18 11, 34 37 27, 58 5 1, 15 18 6, 38 38 22, 54 3 1, 8 28 13, 53 45 28, 68 1 0, 1 0.2, 0.7 13 5, 24 38 28, 49 7 0, 23 0.7 0.1, 1.7 24 10, 56 52 39, 67 4 1, 12 3.6 0.9 0.4, 1.7 24 12, 41 50 37, 65 6 2, 12 3.7 1.1 0.2, 3.2 37 18, 64 50 30, 73 6 1, 14 58.7, 80.5 4.8 1.2 0.4, 2.4 47 24, 82 66 41, 95 3 0, 7 71.7 59.3, 84.2 5.3 1.1 0.4, 2.3 41 16, 71 75 45, 112 3 0, 8 71.0 58.0, 83.5 6.2 0.6 0.2, 1.5 25 12, 44 61 41, 80 2 0, 5 12 68.0 55.7, 79.4 5.7 0.7 0.3, 1.8 25 11, 50 59 37, 87 4 1, 9 13 73.6 58.2, 87.4 6.5 0.8 0.2, 2.0 33 15, 60 76 50, 102 4 1, 6 14 70.4 56.0, 82.7 6.0 0.7 0.2, 1.4 43 25, 62 74 51, 105 2 0, 5 15 85.4 67.4, 102.5 6.5 0.9 0.0, 1.5 31 13, 58 72 44, 100 1 0, 7 16 58.7 39.3, 79.1 4.2 0.7 0.7, 0.7 14 11, 17 67 50, 103 NA NA Overall 68.4 55.9, 82.2 4.7 1.0 0.3, 2.5 34 11, 70 57 30, 97 4 1, 12 Mean 5th and 95th Percentiles Mean 5th and 95th Percentiles Abbreviations: NA, not available; ppb, parts per billion; ppm, parts per million. a Climate zones 1 and 16 did not have enough cases to be included in any analysis. b °C = (°F − 32) × 5/9. c Difference between absolute values. for correlations between apparent temperature and the gaseous pollutants are shown in Table 3. The greatest correlation between apparent temperature and any gaseous pollutant was with ozone at r = 0.44. Carbon monoxide and nitrogen dioxide had the greatest correlation at r = 0.73 but were never included in the same model, as only one gaseous pollutant at a time was included in the model with apparent temperature. We did not find any consistent evidence of nonlinearity in the apparent warm-season temperature-stillbirth association; thus, all reported results were based on linear models. Figure 2 depicts the meta-analysis results for all lags examined during the warm season. We chose to focus on the cumulative average of lags 2–6 days to account for possible lag times from exposure to fetal death and from death to expulsion. For every 10°F increase in apparent temperature (average of lags 2–6 days), we found a 10.4% (95% confidence interval (CI): 4.4, 16.8) increase in the excess risk of fetal death during the warm season. We found some variability according to maternal demographic factors, including age, education, and race/ethnicity, as well as by fetal sex (Figure 3). Younger mothers under age 25 years had an increased risk of stillbirth associated with apparent temperature during the warm season (11.8%, 95% CI: 1.5, 23.2) in comparison with mothers aged 25–34 years (10.1%, 95% CI: 0.8, 20.2) and mothers aged ≥35 years Table 3. Pearson Correlation Coefficients for Correlations Between Mean Apparent Temperature and Mean Air Pollutant Concentrations Among Stillbirth Cases (Cumulative Average of Lags 2–6 Days), California, May–October 1999–2009 Apparent Temperature (n = 8,542) Carbon Monoxide (n = 7,389) Nitrogen Dioxide (n = 7,710) Ozone (n = 8,280) 1 −0.03 0.13 0.44 0.01 0.73 0.05 0.30 1 0.28 Apparent temperature Carbon monoxide Nitrogen dioxide Ozone Sulfur dioxide Am J Epidemiol. 2016;183(10):894–901 1 1 Sulfur Dioxide (n = 4,785) 0.26 −0.08 1 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 Mean Carbon Monoxide, ppm % Change 898 Basu et al. 18 16 14 12 10 8 6 4 2 0 2 3 4 5 6 23 24 25 26 Figure 2. Estimated percent change in stillbirth risk associated with a 10°F (5.6°C) increase in mean apparent temperature, by lag time, California, May–October 1999–2009. Bars, 95% confidence intervals. Variable Race/Ethnicity Non-Hispanic white Hispanic Non-Hispanic black Non-Hispanic Asian Age, years <25 25−34 >34 Education High school or less Some college Fetal Sex Male Female −10 0 10 20 30 % Change Figure 3. Estimated percent change in stillbirth risk associated with a 10°F (5.6°C) increase in mean apparent temperature (cumulative average of lags 2–6 days), by maternal race/ethnicity, maternal age, maternal education, and fetal sex, California, May–October 1999–2009. Bars, 95% confidence intervals. Am J Epidemiol. 2016;183(10):894–901 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 Lag Time, days (4.9%, 95% CI: −8.8, 20.6). Greater risk was also found for mothers who had a high school education or less (10.6%, 95% CI: 2.9, 18.8) as compared with those who had completed at least some college (7.4%, 95% CI: −4.9, 21.4). While most of the maternal racial/ethnic groups examined had similar risks (ranging from a 9.4% increase to a 10.8% increase per 10°F), Asians had a lower risk, with a 4.8% increase (95% CI: −16.5, 31.5), while risk for Hispanics was statistically significantly increased (10.5%, 95% CI: 2.1, 19.5). Male fetuses exhibited greater risk, with an estimate of 13.3% (95% CI: 4.8, 22.4), than did female fetuses (9.8%, 95% CI: −1.9, 22.9). Additionally, we examined gestational week, distance from a meteorological monitor, and season as effect modifiers and evaluated air pollutants as both possible confounders and effect modifiers. Although the risks for preterm and full-term births were similar (not shown), the association was not Temperature and Stillbirth in California 899 60 50 % Change 40 30 20 10 0 –10 –20 –30 20–36 20–25 26–30 31–33 34–36 37–44 Figure 4. Estimated percent change in stillbirth risk associated with a 10°F (5.6°C) increase in mean apparent temperature, by gestational period, California, May–October 1999–2009. Bars, 95% confidence intervals. significant for full-term births. As shown in Figure 4, the associations for stillbirth were most elevated during gestational weeks 20–25 and gestational weeks 31–33. Associations were also more pronounced for mothers who lived within 10 km of a meteorological monitor, as was used in our study, compared with those living within 10–20 km of a monitor (not shown). Finally, no confounding or consistent effect modification was found for any air pollutant examined (not shown). There were no statistically significant associations for the cold season, and the effect estimates ranged from −0.93% for an average lag of 2–3 days to 2.74% for a lag of 6 days, using the various lag periods that were considered for the main analysis. DISCUSSION We found positive associations between apparent temperature and stillbirth during the warm season in California. Although we had approximately 8,500 cases in our main analyses, the numbers in our stratified analyses, such as those stratified by gestational week, were relatively small. We observed higher temperature-related stillbirth risks for preterm births, with the greatest risks observed at 20–25 weeks of gestation, which is also the window of highest overall stillbirth risk in this population, as well as 31–33 weeks of gestation. In a previous study, increasing temperature increased the hazard ratio for stillbirth specifically for preterm births but not for full-term births (5). Another study examining stillbirths found that among full-term stillbirths, the greatest risk of small size for gestational age occurred between 31 and 33 weeks of gestation (22). Younger mothers and less educated mothers had greater risks of stillbirth, while whites, blacks, and Hispanics also exhibited greater risk than Asians in our study. Previous investigators found general disparities in birth outcomes by race/ ethnicity for black and Hispanic infants (25) and specifically during periods of heat exposure (26). Some of the racial/ ethnic disparity may be due to lack of access to prenatal care for some Hispanic women or to nutrient deficiencies in AfricanAmerican women (5, 25). Race/ethnicity may also serve as a proxy for socioeconomic status (SES) in the United States, Am J Epidemiol. 2016;183(10):894–901 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 Gestational Weeks specifically for heat and health outcomes, as described in a recent review (27). While older maternal age has been generally associated with increased risk of stillbirth (28), younger mothers generally have lower SES, have less education, and exhibit more risky behavioral patterns (29). Thus, these maternal demographic factors may all be markers of SES, implying that pregnant women with lower SES are at greater risk from outdoor heat exposure and may spend more time outdoors. Older mothers, in particular, could have other competing risk factors that overpower the risk from ambient temperature exposure. There was a greater proportion of stillbirths among male fetuses in our study, and they appeared to be at slightly greater risk from heat exposure than female fetuses. Males are typically at higher risk of stillbirth, with differences in development and metabolic rates being posited as reasons why they would be physiologically more vulnerable to stressors (30). Maternal stressors during pregnancy have been linked to increased male fetal losses and reduced male:female birth sex ratios, and stress from temperature exposure appears to act similarly (31). Increased male fetal loss with stress may represent a biological unwillingness to support less resilient offspring during times of hardship, an adaptation possibly favored by natural selection (31). To our knowledge, only 2 previous studies have examined the association between temperature and stillbirth (4, 5). In Brisbane, Australia, Strand et al. (5) found that risk of stillbirth increased with higher temperature exposures during the prior week up to temperatures of 21°C and that there was no increased risk at the highest temperatures. In a historical cohort study in Uppsala, Sweden, Bruckner et al. (4) found that colder average temperatures during the entire gestation period were linked to increased fetal death, using data collected at a time when indoor heating and shelter were less protective than they are now. In a recent review, Strand et al. (1) summarized data on the association between seasonality and stillbirth and found evidence of a link during the winter months in New York, Minnesota, and Switzerland. A higher rate of stillbirth was also found in the fall in Minnesota and in the spring in Switzerland (1). Only 1 previous study in this review found a peak of stillbirth during the summer in Australia (32), where the climate is similar to that of California. Recent studies of air pollutants and stillbirth have been conducted in California, New Jersey, Taiwan, South Korea, and England. Depending on the location and the window of pregnancy exposure, fine particulate matter ( particulate matter less than 2.5 µm in diameter (PM2.5)) (33), particulate matter less than 10 µm in diameter (PM10) (34, 35), ozone (33), nitrogen dioxide (33, 36), carbon monoxide (36), and sulfur dioxide (34, 36) have all been reported to be associated with stillbirth, while black smoke was not found to be related in 1 study (37). In a previous short-term study, sulfur dioxide and carbon monoxide exposures were significantly related to stillbirth, with elevated risks from nitrogen dioxide and PM2.5 being reported as well (24). The lack of studies on both temperature and air pollutants indicates that more work is needed to better characterize these relationships. Various mechanisms have been postulated to support the association between apparent temperature and stillbirth. If sweating efficiency is inadequate, the body cannot cool down and can become dehydrated. With dehydration, uterine blood 900 Basu et al. of heat for pregnant women and their fetuses and help them be more vigilant about providing adequate care, especially during the warm season. Furthermore, the emotional and financial burdens placed on families by stillbirth, as well as the economic stress on the health-care system, can be partially relieved. Since research on temperature and stillbirth is still relatively new, further studies in other locations are warranted. Databases with more information on maternal risk factors and medical conditions would be beneficial. ACKNOWLEDGMENTS Author affiliations: Air and Climate Epidemiology Section, Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, Oakland, California (Rupa Basu, Brian J. Malig); and Division of Biostatistics, School of Public Health, University of California, Berkeley, Berkeley, California (Varada Sarovar). The opinions expressed in this article are those of the authors and do not represent those of the California Environmental Protection Agency or the Office of Environmental Health Hazard Assessment. Conflict of interest: none declared. REFERENCES 1. Strand LB, Barnett AG, Tong S. The influence of season and ambient temperature on birth outcomes: a review of the epidemiological literature. Environ Res. 2011;111(3): 451–462. 2. Beltran AJ, Wu J, Laurent O. Associations of meteorology with adverse pregnancy outcomes: a systematic review of preeclampsia, preterm birth and birth weight. Int J Environ Res Public Health. 2014;11(1):91–172. 3. Chodick G, Flash S, Deoitch Y, et al. Seasonality in birth weight: review of global patterns and potential causes. Hum Biol. 2009;81(4):463–477. 4. Bruckner TA, Modin B, Vågerö D. Cold ambient temperature in utero and birth outcomes in Uppsala, Sweden, 1915–1929. Ann Epidemiol. 2014;24(2):116–121. 5. Strand LB, Barnett AG, Tong S. Maternal exposure to ambient temperature and the risks of preterm birth and stillbirth in Brisbane, Australia. Am J Epidemiol. 2012;175(2): 99–107. 6. California Department of Public Health. Table 3-7. Fetal deaths and mortality ratios from selected causes of death, California, 2010 (by place of residence) [database]. http://www.cdph.ca. gov/data/statistics/Documents/VSC-2010-0307.pdf. Accessed September 23, 2015. 7. Gold KJ, Sen A, Xu X. Hospital costs associated with stillbirth delivery. Matern Child Health J. 2013;17(10):1835–1841. 8. Blackmore ER, Côté-Arsenault D, Tang W, et al. Previous prenatal loss as a predictor of perinatal depression and anxiety. Br J Psychiatry. 2011;198(5):373–378. 9. Stillbirth Collaborative Research Network Writing Group. Causes of death among stillbirths. JAMA. 2011;306(22):2459–2468. 10. California Irrigation Management System. California Irrigation Management System (CIMIS). CIMIS overview. Davis, CA: Office of Water Use Efficiency, California Department of Am J Epidemiol. 2016;183(10):894–901 Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 flow to the fetus can decrease and induce labor, often prematurely (38). Heat exposure may cause damage to cells, the placenta, and vascular systems, resulting in insufficient fetal nutrition (39). Khamis et al. (40) found that applying heat to the abdominal wall during labor increased uterine activity. Further evidence comes from heat-induced miscarriages in a study of pregnant women who used a hot tub (39). Women who used hot tubs more frequently and at greater water temperature were at increased risk, particularly during early stages of gestation. Heat has also been shown to be teratogenic in animals (41) and has been linked to birth defects such as congenital cataracts from outdoor heat exposure (42) or heart defects during extreme heat events (43), which could lead to unviability of the fetus. Recent evidence suggests that maternal vitamin D deficiency may result in shorter gestation periods (44), lower birth weights (45), greater risk of fetal musculoskeletal problems (46), and lower offspring serum calcium concentrations and bone mass (45). Even in areas where there is adequate sun exposure, mothers may be deficient in vitamin D (47). There were several limitations in our study. We were restricted by the birth certificate database with regard to the variables that we could evaluate as effect modifiers. We relied on maternal race/ethnicity, age, and education as markers of SES but did not have information on health insurance or use of air conditioning. Typically, persons with lower SES use air conditioners less because of concerns about cost. However, many houses in the more expensive coastal areas of California, particularly in Northern California, do not have air conditioners, since they have not been needed because of milder climates. We also could not examine medical conditions during pregnancy, so we did not have the ability to exclude mothers with these prior risk factors in our study. We were not able to examine confounding by PM2.5, since it is only measured every third or sixth day, and our control periods for data analysis were selected daily. However, in a previous California study of air pollution and stillbirth that used the same data (34), air pollutants were found to have long-term associations with stillbirth as measured by full-pregnancy and trimester average exposures, and we only considered short-term relationships in this study. In another short-term study, Faiz et al. (24) reported associations with PM2.5, but the findings were not significant. Finally, we based exposure on residential zip codes, since we did not have information on occupation or occupational addresses. However, we attempted to refine exposure assessment by limiting residential zip codes to those within 10 km of a meteorological monitor, which was found to make a difference in comparison with the estimates including mothers who lived within 20 km of a meteorological monitor. We did not have information on where the majority of time was spent, but mothers who spent more time outdoors during the warm season could inevitably have had greater heat exposure. There is a growing body of evidence that pregnant women and their fetuses are vulnerable to seasonal influences and ambient temperatures (1). To our knowledge, our study is the first large-scale epidemiologic study conducted anywhere in the United States to have specifically examined the association between apparent temperature and stillbirth. Combined with evidence from other studies, the findings from this study will inform health-care practitioners about the increased risk Temperature and Stillbirth in California 901 11. 12. 13. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Am J Epidemiol. 2016;183(10):894–901 29. McDonald JA, Suellentrop K, Paulozzi LJ, et al. Reproductive health of the rapidly growing Hispanic population: data from the Pregnancy Risk Assessment Monitoring System, 2002. Matern Child Health J. 2008;12(3):342–356. 30. Mondal D, Galloway TS, Bailey TC, et al. Elevated risk of stillbirth in males: systematic review and meta-analysis of more than 30 million births. BMC Med. 2014;12:220. 31. Catalano R, Bruckner T, Smith KR. Ambient temperature predicts sex ratios and male longevity. Proc Natl Acad Sci U S A. 2008;105(6):2244–2247. 32. Barnett AG, Dobson AJ. Analysing Seasonal Health Data. (Statistics for Biology and Health). Berlin, Germany: Springer Berlin GmbH; 2010. 33. Green R, Sarovar V, Malig B, et al. Association of stillbirth with ambient air pollution in a California cohort study. Am J Epidemiol. 2015;181(11):874–882. 34. Hwang BF, Lee YL, Jaakkola JJ. Air pollution and stillbirth: a population-based case-control study in Taiwan. Environ Health Perspect. 2011;119(9):1345–1349. 35. Kim OJ, Ha EH, Kim BM, et al. PM10 and pregnancy outcomes: a hospital-based cohort study of pregnant women in Seoul. J Occup Environ Med. 2007;49(12):1394–1402. 36. Faiz AS, Rhoads GG, Demissie K, et al. Ambient air pollution and the risk of stillbirth. Am J Epidemiol. 2012; 176(4):308–316. 37. Pearce MS, Glinianaia SV, Rankin J, et al. No association between ambient particulate matter exposure during pregnancy and stillbirth risk in the north of England, 1962–1992. Environ Res. 2010;110(1):118–122. 38. Stan C, Boulvain M, Hirsbrunner-Amagbaly P, et al. Hydration for treatment of preterm labour. Cochrane Database Syst Rev. 2002;(2):CD003096. 39. Li DK, Janevic T, Odouli R, et al. Hot tub use during pregnancy and the risk of miscarriage. Am J Epidemiol. 2003;158(10): 931–937. 40. Khamis Y, Shaala S, Damarawy H, et al. Effect of heat on uterine contractions during normal labor. Int J Gynaecol Obstet. 1983;21(6):491–493. 41. Graham JM Jr, Edwards MJ, Edwards MJ. Teratogen update: gestational effects of maternal hyperthermia due to febrile illnesses and resultant patterns of defects in humans. Teratology. 1998;58(5):209–221. 42. Van Zutphen AR, Lin S, Fletcher BA, et al. A population-based case-control study of extreme summer temperature and birth defects. Environ Health Perspect. 2012;120(10):1443–1449. 43. Agay-Shay K, Friger M, Linn S, et al. Ambient temperature and congenital heart defects. Hum Reprod. 2013;28(8): 2289–2297. 44. Morley R, Carlin JB, Pasco JA, et al. Maternal 25hydroxyvitamin D and parathyroid hormone concentrations and offspring birth size. J Clin Endocrinol Metab. 2006;91(3): 906–912. 45. Harvey NC, Holroyd C, Ntani G, et al. Vitamin D supplementation in pregnancy: a systematic review. Health Technol Assess. 2014;18(45):1–190. 46. Palermo NE, Holick MF. Vitamin D, bone health, and other health benefits in pediatric patients. J Pediatr Rehabil Med. 2014;7(2):179–192. 47. Karras SN, Anagnostis P, Annweiler C, et al. Maternal vitamin D status during pregnancy: the Mediterranean reality. Eur J Clin Nutr. 2014;68(8):864–869. Downloaded from https://academic.oup.com/aje/article-abstract/183/10/894/1739797 by guest on 09 June 2020 14. Water Resources; 2014. http://www.cimis.water.ca.gov/cimis/ welcome.jsp. Accessed September 23, 2015. United States Environmental Protection Agency. AQS Data Mart [database]. Washington, DC: Environmental Protection Agency; 2014. https://aqs.epa.gov/aqsweb/documents/ data_mart_welcome.html. Accessed September 23, 2015. National Climatic Data Center, National Oceanic and Atmospheric Administration. Climate Resources [database]. (Data for 2012). http://www.ncdc.noaa.gov/oa/climate/ climateresources.html. Accessed September 23, 2015. Basu R, Feng WY, Ostro BD. Characterizing temperature and mortality in nine California counties. Epidemiology. 2008; 19(1):138–145. California Energy Commission. California Statewide Residential Appliance Saturation Study: Final Report. (Report no. 400-04-009). Sacramento, CA: California Energy Commission; 2008. http://www.energy.ca.gov/appliances/rass/. Accessed September 23, 2015. Air Quality Data Section, Air Resources Board, California Environmental Protection Agency. California Air Quality Data [database]. Sacramento, CA: California Environmental Protection Agency; 2011. http://www.arb.ca.gov/aqd/aqdcd/ aqdcddld.htm. Accessed September 23, 2015. California Department of Health Services. Birth Statistical Master Files, 1999–2009. Sacramento, CA: Office of Vital Statistics, California Department of Health Services; 2009. Strand LB, Barnett AG, Tong S. Methodological challenges when estimating the effects of season and seasonal exposures on birth outcomes. BMC Med Res Methodol. 2011;11:49. Levy D, Lumley T, Sheppard L, et al. Referent selection in case-crossover analyses of acute health effects of air pollution. Epidemiology. 2001;12(2):186–192. SAS Institute, Inc. SAS Statistical Software, Version 9.3. Cary, NC: SAS Institute, Inc.; 2014. R Development Core Team. R: A Language and Environment for Statistical Computing. Version 2.6.2. Vienna, Austria: R Foundation for Statistical Computing; 2008. https://cran. r-project.org/doc/manuals/r-release/fullrefman.pdf. Accessed September 23, 2015. Lumley T. rmeta: Meta-analysis. Vienna, Austria: R Foundation for Statistical Computing; 2012. https://cran.r-project.org/web/ packages/rmeta/index.html. Accessed September 23, 2015. Gardosi J, Mul T, Mongelli M, et al. Analysis of birthweight and gestational age in antepartum stillbirths. Br J Obstet Gynaecol. 1998;105(5):524–530. Genest DR, Williams MA, Greene MF. Estimating the time of death in stillborn fetuses: I. Histologic evaluation of fetal organs; an autopsy study of 150 stillborns. Obstet Gynecol. 1992;80(4):575–584. Faiz AS, Rhoads GG, Demissie K, et al. Does ambient air pollution trigger stillbirth? Epidemiology. 2013;24(4):538–544. Lorch SA, Kroelinger CD, Ahlberg C, et al. Factors that mediate racial/ethnic disparities in US fetal death rates. Am J Public Health. 2012;102(10):1902–1910. Basu R, Malig B, Ostro B. High ambient temperature and the risk of preterm delivery. Am J Epidemiol. 2010;172(10):1108–1117. Gronlund CJ. Racial and socioeconomic disparities in heat-related health effects and their mechanisms: a review. Curr Epidemiol Rep. 2014;1(3):165–173. Huang L, Sauve R, Birkett N, et al. Maternal age and risk of stillbirth: a systematic review. CMAJ. 2008;178(2):165–172.