For thousands of years, people have believed that natural phenomena such as the lunar cycle, weather conditions, or seasonality influence childbirth. As such, there have been many papers discussing the relationship between natural phenomena, seasonality, and obstetric problems [1, 2, 3]. For example, “The Christmas effect” is one of the most famous rumors relating to the association between seasonality and obstetrics, which suggests that the rate of conception is highest around the December holidays [4]. In addition, there is an empirical finding known to experienced obstetricians and midwives that bad weather conditions induce labor onset. In fact, previous studies have evaluated the relationship between deliveries and environmental factors, such as barometric pressure, temperature, humidity, wind, and, recently, particulate matter 2.5 [1, 5, 6, 7]. Some of these studies showed that temperature and barometric pressure might influence deliveries [5, 6]. Furthermore, because of the recent global attention to climate change, the number of studies focusing on the relationship between climate change and pregnancy has increased [8, 9]. Thus, the relationship between pregnancy and various atmospheric factors should be further investigated.

In the same way, there is considerable information available online relating to the empirical finding that weather conditions influence not only the timing of deliveries but also the timing of labor onset. However, while most previous reports have focused mainly on the relationship between the timing of deliveries and natural phenomena, the timing of labor onset has not been assessed extensively. Although the mechanism and key factors of labor onset are yet to be clarified, the understanding and expectation of labor onset will enable obstetricians and obstetric staff to provide appropriate management for pregnant women. Therefore, the relationship between labor onset and atmospheric conditions is worth investigating. The present study aimed to assess the relationship between weather conditions and labor onset and consider the underlying mechanisms.

During the 8-year study period, we recruited a total of 1321 pregnant women who were hospitalized for labor onset. The flow diagram of study enrollment is presented in Fig. 1. Finally, 1178 women were enrolled for analysis (Fig. 1). No women were hospitalized for labor onset at post-term because elective labor induction is routinely offered before 42 weeks’ gestation at our institute. There were no cases of multiple pregnancies hospitalized for labor onset at 37–41 weeks’ gestation because elective Cesarean section is routinely performed around 37 weeks of gestation. Participant characteristics are detailed in Table 1.

Fig. 1.

Flow diagram of study enrollment. A total of 1178 women were eligible and analyzed.

The relationship between atmospheric variables and the percentage of days is shown in Table 2. There was no significant difference in the percentage of days in relation to barometric pressure, vapor pressure, or temperature difference. However, we observed a significant difference in the percentage of days when participants were divided according to total solar irradiation, with low total solar irradiation associated with a significantly higher rate than the middle/high group.

Moreover, the relationship between atmospheric variables and the number of laboring women is shown in Table 3. The number of laboring women was significantly higher for the low total solar irradiation group than for the middle/high group. In contrast, barometric pressure, vapor pressure, and temperature difference did not significantly affect the number of laboring women.

In addition, the backgrounds of women in the low and middle/high groups of total solar irradiation are shown in Table 4. There is no difference in the woman characteristics in both groups.

The outcomes of total solar irradiation were significantly different; therefore, we further investigated the seasonal effects on the percentage of days and the number of laboring women, the results of which are listed in Supplementary Tables 1a,b,2a,b. The levels of total solar irradiation (MJ/m${}^2$) in each season were 19.0 $\pm$ 7.8 (Spring, March–May), 17.9 $\pm$ 7.6 (Summer, June–August), 11.3 $\pm$ 5.4 (Autumn, September–November), and 11.0 $\pm$ 4.1 (Winter, December–February; shown as mean $\pm$ standard deviation). At first, we assessed these outcomes in the warmer seasons (from spring to summer) and the colder seasons (from autumn to winter). The percentage of days was significantly increased for the women who received low amounts of total irradiation (Supplementary Table 1a) in both warmer and colder seasons. In addition, in the warmer seasons, the number of laboring women who received low amounts of total irradiation also significantly increased (Supplementary Table 1b). Finally, we performed the same analysis to assess the percentage of days and women within each season (Supplementary Table 2a,b). In the spring and summer seasons, these outcomes were significantly increased women who received low amounts of total solar irradiation. In the autumn and winter seasons, these outcomes intended to be higher than those of women who received middle/high amounts of total solar irradiation.

The effect of nature on our health has been of interest to humans for centuries. Studies have revealed that some natural phenomena affect specific aspects of human health; for example, heart failure is exacerbated by cold temperatures [11] and the rate of depression increases with decreased sunlight during the winter season [12]. In the field of obstetrics, it is believed that natural phenomena such as the lunar cycle and weather conditions influence the occurrence of childbirth, although its validity is controversial. Previous reports have mainly focused on the relationship between natural phenomena, the timing of childbirth and incidences of preterm birth, premature membrane rupture, and other obstetric complications [1, 2, 3]. Morton-Pradhan et al. [1] reported that there is no significant correlation between birth rate and the stage of the lunar cycle or atmospheric conditions (including barometric pressure, dew point, and temperature) in the state of Arizona. Arliss et al. [2] also concluded that there is no obvious relationship between lunar cycle and birth rate in the state of North California from a study including more than 100,000 births from official birth certificate data of the USA. Muresan et al. [6] reported the incidence of weekly premature births significantly increased with average temperature and temperature variation, through retrospective analysis of almost 1800 eligible women from the obstetrical database of their own hospital. However, there are few studies on the relationship between natural phenomena and labor onset, compared with studies about birth rate or the incidence of premature birth. Akutagawa et al. [5] conducted a retrospective analysis of the relationship between barometric pressure and obstetric events including labor onset, premature membrane rupture, and childbirth at their hospital, including almost 2000 participants, which led them to conclude that, although the number of spontaneous deliveries increased on days when barometric pressure was low, there was no significant relationship between barometric pressure and the number of pregnant women hospitalized for labor onset. Polansky et al. [13] also reported no relationship between labor onset and barometric pressure. Based on our result and previous reports, labor onset does not seem to be affected by barometric pressure. Further, low mean vapor pressure and low temperature difference between the warmest and coldest temperatures in a day did not have significant effects on labor onset. No previous report has focused on the effects of these factors on labor onset; therefore, further studies are needed for making a conclusion.

Moreover, the relationship between labor onset and total solar irradiation has not been reported. However, our result showed that total solar irradiation had a significant effect on labor onset; days of low solar irradiation were associated with increased numbers of women hospitalized due to labor onset. Based on our results, we speculated some hypothesis to answer why low amounts total solar irradiation affects labor onset in the consideration to several hormones regulated by sunlight. First, recent reports have stated the relationship between hormones associated with vitamin D and perinatal outcomes [12, 14, 15]. Vitamin D is one of the famous hormones synthesized by sunlight exposure. Shibata et al. [14] found that threatened premature delivery within 3rd trimester was associated with lower serum 25-hydroxyvitamin D (25-(OH)D) levels in Japanese population, which indicates that longitudinal 25-(OH)D deficiency, associated with lower sun exposure, may start the mechanism of uterine contraction. Because our report showed that the short-term effect of low amounts of total solar irradiation in a single day was related to labor onset, further information may be interesting for the obstetrical effects of daily change of serum levels of vitamin D or calcium. The other hormone that may play a role is melatonin, which is a regulator of circadian rhythm and a scavenger of free radicals. As it is well-known that the production of melatonin is regulated by solar irradiation and has the diurnal variation, Sharkey et al. [16] insisted that the pigment was involved in the induction of parturition, similar to oxytocin. They also referred that melatonin could induce contraction of the human myometrium and might bind strongly to myometrial membranes during laboring. One report demonstrated that exposure of the lamp light to eyes to suppress melatonin secretion caused impaired uterine contraction [17]. Trap et al. [18] reported that labor onset often starts at night, when the melatonin level is generally high, and they mentioned the possibility of the positive effects of circadian rhythm on labor onset. According to these reports, we can hypothesize that higher levels of melatonin on days with low solar irradiation may induce cyclic uterine contraction thus initiating labor onset. Recently, it has been revealed that decreased progesterone and increased inflammation cytokines such as prostaglandin are important for labor onset [19]. Therefore, it may be interesting in investigating the association between these sunlight-related hormones, progesterone and inflammation cytokines.

Furthermore, we assessed the seasonal effect of total solar irradiation on labor onset. In each season, particularly spring and summer, the percentage of days and the number of laboring women receiving low amounts of solar irradiation were higher than those receiving middle/high amounts of solar irradiation. According to the results, the assumption is that labor onset tends to be more affected by the days of low solar irradiation during the hot seasons with higher solar irradiation on average than during the cold seasons with lower solar irradiation on average. In view of the relationship between sunlight and seasonality and obstetrical outcomes, Algert et al. [20] reported that morbidity from pregnancy hypertension is lower in spring and summer; during these seasons, higher the levels of sunlight, lower the monthly hypertension rates. According to their report and our result, we think that the effect of seasonality, in terms of total solar irradiation, on pregnant women is worth investigating. To the best of our knowledge, this is the first epidemiological report to investigate the relationship between total solar irradiation and labor onset. Because the above discussions are just assumptions, further investigations are warranted to confirm the results through which low total solar irradiation can induce labor onset.

There are some limitations of this study which should be acknowledged. The first was our sample size; according to our database, only 1321 women were hospitalized for labor onset during the study period, although there are approximately 400–500 births annually in this hospital. The other main reasons for obstetric hospitalization included scheduled cesarean section, induction of labor, rupture of membranes, threatened preterm birth, and maternal complications, such as hypertensive disorder of pregnancy and gestational diabetes mellitus; women hospitalized for these reasons were naturally ineligible for inclusion in this analysis. Moreover, we could not precisely assess whether the women excluded also experienced labor onset when hospitalized. Thus, this study was affected by sample size bias and selection bias. Previous single-center studies with almost 1000–2000 eligible participants, such as those of Akutagawa et al. [5] and Muresan et al. [6], have revealed obvious effects of weather conditions on the incidence of obstetrical problems. We believe that the influence of weather on inhabitants of a single area can be assessed more precisely in single-center studies than in large-area analyses, even if the study population is small. To obtain more reliable data and minimize sample size bias and selection bias, neighboring multicenter prospective analyses with larger numbers of participants are necessary. Second, we could not assess the precise time of labor onset because we defined that it as when the woman was admitted to hospital. However, this is unlikely to influence our results significantly because eligible women were limited to those residing in Tokyo who could visit the hospital within an hour or so. Third, we did not evaluate the effects of solar irradiation changes within a single day or between days compared with that of weekly solar irradiation with respect to labor onset. Fourth, because we did not assess the location and behavior of pregnant women when cyclic uterine contractions were felt, for example, staying or sleeping at home, or walking outside, and the duration of their outdoor activity, we cannot conclude that the relationship between labor onset and solar irradiation was a direct causal relationship. Fifth, according to our result, it would be worthwhile to investigate the relationship between preterm labor onset and each atmospheric condition. However, our database included only 32 women with preterm labor onset during the study period who were excluded from our primary analysis; our preliminary analysis revealed no statistical difference among the percentage of days, the number of laboring women and each atmospheric condition (Supplementary Table 3a,b). Since the preterm birth rate in Japan is low (5.6%) [21] and our primary interest is the relationship between labor onset and atmospheric conditions, solely with respect to pregnant women hospitalized for labor onset, the sample size of laboring women at preterm in this study was quite small. Therefore, this relationship requires further research with larger sample sizes. Finally, we used a univariable simplistic statistical approach in this study because we hypothesized that atmospheric factors were independent of general background characteristics, such as age and parity; thus, we did not evaluate obvious confounding factors. Researchers in further studies should investigate the effect of weather change on obvious factors related to labor onset, such as premature membrane rupture or cervical ripening.

There are a lot of unproved empirical findings relating to the obstetric area, and the precise mechanism of labor onset is still unclear. The present study demonstrates that pregnant women with hospitalized due to labor onset tend to be hospitalized in a day with low total solar irradiation. Weather changes may be an important aspect of research for the mechanism of labor onset.

The datasets analyzed in the present study are available from the corresponding author on reasonable request.

SH, TT, HO and MS contributed to the study design and conception, material preparation, and data collection. Data analysis was performed by SH and TT. The first draft of this text, figures, and tables was written and created by SH, and TT, HO and MS commented on the first version of this manuscript. SH, TT, HO and MS approved the final manuscript. All authors read and approved the final manuscript.

All subjects gave their informed consent for inclusion in this study by posting the optout at the outpatient office of our department. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of National Center for Global Health and Medicine (approval number: NCGM-G-003245-00).

The Authors would like to thank Enago (www.enago.jp) for the English language review.

This work was supported by the Grants-in-Aid for Research from the National Center for Global Health and Medicine [20A3002].

The authors declare no conflict of interest.