This study demonstrated that the number of hours worked per week during the first trimester of pregnancy was associated with pregnancy complications such as TA (partially) and PTB. Compared with women who worked 40 hours or less per week, women who worked 71 hours or more per week had a three-fold higher risk of experiencing TA, women who worked 51-70 hours per week had 2.5 times higher risk of experiencing PTB, and women who worked 71 hours or more had 4.2 times higher risk of experiencing PTB even after adjusting for medical specialty, maternal age, and current household income.
The results of related published studies have not been consistent. For example, Pompeii et al. [
6] observed a protective effect of working hours on PTB. Several studies [
7‐
11,
17] reported that long working hours increased the risk of PTB, whereas others failed to identify any association in this regard [
5]. Inconsistent results across studies may be primarily attributable to the small effect size of the number of working hours on PTB, different cut-off points for working hours per week, data collection at different time points during pregnancy (e.g., different trimesters), or the confounding effects of occupation and different job characteristics. In terms of the effect of occupation, there was only one study that focused on women physicians, published in 1990s. This study found that residents who worked ≥100 hours per week during the first trimester of pregnancy had a higher risk of preterm delivery compared with residents who worked < 100 hours per week (9.8% vs. 4.6%,
P = 0.012). Taken together, this previous study and our research adds to the literature showing that longer working hours may increase the risk of preterm delivery among physicians.
The prevalence of TA and PTB in our sample was 15% and 12%, respectively, which were close to the rates in the general population (i.e., those of TA and PTB have been reported as 16-25% [
19] and 11% [
20], respectively). It has previously been recognized that maternal age is associated with adverse pregnancy complications. In our sample, the mean maternal age during the first pregnancy was 31 ± 4 years, which is slightly older than that in Japanese general population (i.e., mean = 29.7 years based on the 2010 Vital statistics reported by the Japanese Ministry of Health, Labour, and Welfare). Thus, the comparable prevalence found by the present study and previous reports [
19,
20] may be explained by the small difference between the maternal age of between our sample and general population. Alternatively, our sample may have been healthier than the general population.
This study also has several limitations. First, our sampling methods may have resulted in sampling bias and selection bias. Sampling bias may have arisen from our inclusion of only private medical school alumnae, and from the low participation rate (1,684 of the 9,544 subjects who were initially recruited for this study actually participated). This sampling bias may undermine the external validity of our results, and our findings may not be generalizable to all women physicians in Japan. Selection bias may have been caused by a tendency of participants who were frustrated with poor working conditions to over-report the number of hours worked per week. Because such a selection bias would undermine internal validity, our results should be interpreted cautiously. Second, we measured the number of hours worked per week during the first trimester by asking “On average, how many hours per week did you work when you initially became aware of your first pregnancy?” Although, menstruation stops after a woman gets pregnant, some women continue to experience some hormone-driven or abnormal bleeding while pregnant, and mistakenly perceive it as menstruation. Thus, some women may have answered this question with reference to their second trimester, which may have caused misclassification. Third, self-reported data may be subject to recall bias. Physicians may be aware that long working hours may be detrimental to pregnancy, and they may have unconsciously added hours to their first trimester work experience after experiencing pregnancy complications. To examine the extent of recall bias, we performed sensitivity analyses by excluding women who were ≥45 years of age from the analyses, but the results did not change. According to the additional analysis, subjects with TA or PTB were more likely to work longer hours (i.e., 60 or 65 h per week, respectively) compared with those without complications (i.e., 50 h per week). Additionally, the self-reported rates of TA and PTB were similar to those in previous studies, suggesting that recall bias may not be critical. Fourth, although the outcome variables (i.e., TA and PTB) in this study were defined based on medical diagnosis, our use of a self-report measure may have resulted in the over-reporting of unspecified symptoms, such as vaginal bleeding, as TA. Thus, the frequency of TA may have been overestimated. Fifth, we collected data regarding the first trimester of pregnancy, which is when the rates of fetal growth and development are considered to be highest [
21]. However, previous research has suggested that the last trimester is more strongly associated with PTB [
22]. Future studies should investigate the impact of number of hours worked on pregnancy outcome according to trimester. Sixth, we did not measure lifestyle variables, such as alcohol intake, caffeine consumption, or smoking. However, these factors have been reported to cause short fetal crown-to-rump length, which has been associated with babies who are small for their gestational age [
23], but not with PTB. However, we still did not adjust for a lifestyle factor of physical activity or even vaginal infection which may contribute to preterm birth but was not included in the analysis. Seventh, the demands placed upon physicians can be detrimental and possibly dangerous to the health of the fetus and the mother. For example, a previous report [
18] highlighted the reproductive risks related to occupational exposure to anesthetics. Therefore, one analysis, excluded women whose specialty included anesthesiology but the results did not change. Finally, unmeasured factors, such as stress, fatigue, or the psychological burden related to long working hours, may have affected our results.