The present study was based on a cross-sectional survey in October-December 2021. A total of 30 kindergartens in Chengdu, a provincial capital in Southwest China, were selected by a cluster random sampling method. Data were collected by a validated questionnaire administered by the mothers of preschool children via WJX (an online survey platform, https://www.wjx.cn). Overall, 4,360 mothers participated in the survey and were finally included in the analysis.

The study protocol was approved by the Ethics Committee of Soochow University (Approval NO. SUDA20210820H01). The questionnaire opens with the statement “answering the questionnaire constitutes consent to participate” in the study.

Information on sleep duration in pregnancy was obtained via the following question: “For how long did you sleep including siesta per day during pregnancy? (a) <7 h/day, (b) 7–8 h/day, (c) >8 h/day” (16, 17).

The PTB status was determined by the following question in the questionnaire: “How many gestational weeks did you have when you delivered your child? (a) ≥37 weeks, (b) <37 weeks”. Gestational weeks of <37 weeks were defined as PTB (1).

Gender of baby (male and female), maternal age at delivery (<25, 25–29, 30–34, and ≥35 years), educational level of mother (lower secondary education or below, high school or secondary vocational education, associate degree, bachelor's degree or above), educational level of father (lower secondary education or below, high school or secondary vocational education, associate degree, bachelor's degree or above), diabetes history (yes and no), gestation diseases including gestational hyperglycemia (yes and no), gestational hypertension (yes and no), anemia during pregnancy (yes and no) and anxiety or depression during pregnancy (yes and no), and pregnancy behaviors including smoking or passive smoking during pregnancy (0, 1–2, 3–4, and ≥5 days/week), alcohol drinking during pregnancy (0, 1–2, 3–4, and ≥5 days/week), prenatal education (0, 1–2, 3–4, and ≥5 times/week), exercise during pregnancy (0, <20, 20–40, and >40 min/day), and folic acid supplementation daily during pregnancy (yes and no) were all collected by self-report via the questionnaire (18, 19).

The basic characteristics of the participants by PTB status and sleep duration in pregnancy are presented as percentages (%) and were compared using the chi-square test or Fisher's exact test.

We used logistic regression to evaluate the association of sleep duration in pregnancy with PTB. Three models were constructed: model 1 was the basic model adjusted for sex of the baby, maternal age at delivery, educational level of mother and educational level of father; model 2 was adjusted as model 1 plus diabetes history and gestation diseases, including gestational hyperglycemia, gestational hypertension, anemia during pregnancy and anxiety or depression during pregnancy; and model 3 was adjusted as model 2 plus pregnancy behaviors, including smoking or passive smoking during pregnancy, alcohol drinking during pregnancy, prenatal education, exercise during pregnancy and folic acid supplementation daily during pregnancy. The P-trend was calculated by modelling the independent variable as a continuous variable in the regression models.

Effect modification of the association between sleep duration in pregnancy and PTB by gender of baby, gestational hyperglycemia, anemia during pregnancy and folic acid supplementation daily during pregnancy was examined by stratified analyses, and significance of interactions was evaluated on the first-degree multiplicative models for each stratification variable separately.

The statistical analyses were performed using SPSS 26.0 statistical software (IBM Corp., USA). All tests were two-tailed, and a P value <0.05 was considered to indicate statistical significance.

In this cross-sectional study, we observed that short sleep duration in pregnancy was independently associated with increased odds of PTB among Chinese women. The association between sleep duration in pregnancy and PTB was independent of confounding factors and similar across subgroups stratified by gender of baby, gestational hyperglycemia, anemia during pregnancy and folic acid supplementation daily during pregnancy.

Several studies have demonstrated that insufficient sleep is potentially associated with an increased risk of preterm birth among pregnant women, which is consistent with our conclusion. A study from India showed that sleep duration in full-term women was longer than that in preterm women (20). In a cohort study, compared to pregnant women with sleep duration >8 h/day in the nighttime, those with sleep duration ≤5 h/day were more inclined to give preterm birth (21). In a case‒control study based on pregnant women in the first 6 months of pregnancy, compared to women with sleep duration in pregnancy of 7–8 h/day, those with sleep duration 6 h/day were at higher risk of PTB (22). In two prospective studies and one meta-analysis, sleep duration of <7 h/day significantly increased the odds of PTB (11, 13, 23).

Other studies have shown that both insufficient and excessive sleep duration increases the risk of preterm birth. A dose-response meta-analysis demonstrated a strong association between extreme sleep duration during pregnancy and preterm birth, characterized by a U-shaped relationship (24). Meanwhile, a Mendelian randomized analysis revealed that pregnant women sleeping less than 5 h per night or more than 10 h per night exhibited a higher risk of preterm birth (25).

However, a study from the USA showed no associations between the sleep duration of pregnant women at night and PTB, whether in the first, second or third trimester (12). Generally, the current research on preterm birth and sleep duration is inconclusive, which is consistent with the findings of another review (26).

Different study designs, study populations and methods of data collection may explain inconsistent findings.

The underlying mechanism of sleep duration in pregnancy with PTB remains unclear. Most scholars believe that the immune system plays a mediating role between them (12, 27–30). Sleep disorders, either poor sleep quality or short sleep duration, interfere with the normal immune process, elevated levels of proinflammatory cytokines and enhanced inflammatory responses (30).

The dynamic equilibrium between Th1 (cell-mediated immunity) and Th2 (humoral immunity) at the fetal-maternal interface plays a key role in women's successful pregnancy. Th1 cells produce proinflammatory cytokines such as interleukin-1, 2, 6, 8, 12 and tumor necrosis factor α. Th2 produced anti-inflammatory cytokines such as interleukin-4, 5, 10, and 13 (31–33). During normal pregnancy, Th2 activity was much stronger than Th1 activity, which had potential protective effects in the fetal-maternal relationship. For instance, the injection of IL-10, a Th2-type cytokine, prevents fetal wastage in mice prone to fetal resorption via conspicuous downregulatory effects on Th1-type cytokines (34). However, immune dysregulation by a progressive shift toward Th1 predominance may initiate and intensify the cascade of proinflammatory cytokine release, including IL-1, 2, 6, 8 and TNF-α (31), which induces the production of prostaglandins, an important inducer of uterine contraction (35, 36).

In addition, other physical and psychological conditions, such as gestational diabetes (37, 38), gestational hypertension (39, 40), obesity (41, 42), poor psychological status including depression (43, 44) and hyperhomocysteinaemia (45, 46), may also play a role in the association between sleep duration in pregnancy and PTB.

The main strengths of our study include a relatively large sample size and consideration of adjustment for a number of potential confounders. Meanwhile, the findings of the existing research are inconsistent. Our study provides additional reference evidence for research in this field and is meaningful. Sleep duration was collected by subjective estimation of the participants instead of objective measurement, which may lead to self-reporting bias. The gestational status in our study was retrospectively estimated; thus, recall bias might have occurred. Quality of sleep, such as sleep-disordered breathing, was not collected, which limited further exploration (47).