Relationship between pre-pregnancy body mass index and postpartum depression, and the mediation effects of cesarean delivery, low birth weight, child anomaly, and breastfeeding: a prospective birth cohort in the Japan environment and children’s study Authors: Yasuaki Saijo, Yukihiro Sato, Hiroyuki Shiotsuki, Tomoki Ishikawa, Eiji Yoshioka, Kentaro Nakanishi, Yasuhito Kato, Ken Nagaya, Satoru Takahashi, Yoshiya Ito, Mariko Itoh, Hiroyoshi Iwata, Takeshi Yamaguchi, Reiko Kishi, Michihiro Kamijima, Michihiro Kamijima, Shin Yamazaki, Maki Fukami, Chiharu Ota, Koichi Hashimoto, Kenichi Sakurai, Shuichi Ito, Ryoji Shinohara, Hidekuni Inadera, Takeo Nakayama, Ryo Kawasaki, Yasuhiro Takeshima, Hideki Nagashima, Narufumi Suganuma, Mayumi Tsuji, Kimitoshi Nakamura Published: 2026 ABSTRACT BackgroundWhereas underweight and obesity have been reported as risk factors for postpartum depression (PD), in Japan, younger women tend to express a stronger preference to have a lower body mass index (BMI). In this study, we aimed to examine the association between pre-pregnancy BMI categories and PD among Japanese pregnant women, and conducted an exploratory analysis of the mediating roles of adverse birth outcomes.MethodsThis prospective study included 91,640 pregnant women from the Japan Environment and Children Study. Pre-pregnancy BMI was classified into seven categories: ≤ 16.9 kg/m2 (severe–moderate underweight), 17.0‒18.4 kg/m2 (mild underweight), 18.5‒19.9 kg/m2 (lower-normal), 20.0‒22.9 kg/m2 (mid-normal: reference), 23.0‒24.9 kg/m2 (upper-normal), 25.0‒29.9 kg/m2 (overweight), and ≥ 30 kg/m2 (obesity). PD was assessed using the Japanese version of the Edinburgh Postnatal Depression Scale at 1 and 6 months after childbirth (1mo and 6mo, respectively). Potential mediators included adverse birth outcomes, including cesarean delivery, low birth weight, child anomalies, and non-initiation of breastfeeding by 1mo. Logistic regression models estimated adjusted odds ratios, and a causal mediation analysis was conducted to decompose the total effect of pre-pregnancy BMI into natural direct and indirect effects through the candidate mediators.ResultsSevere–moderate underweight, overweight, and obesity were significantly associated with PD at 1mo and 6mo (1mo: OR = 1.21, 95% confidence interval [CI]: 1.08‒1.35; 6mo: OR = 1.16, 95% CI: 1.03‒1.31), and upper-normal BMI was significantly associated with PD at 1mo (OR = 1.09, 95% CI: 1.02‒1.17). Significant mediating effects of low birth weight were observed among lower BMI categories, child anomalies among higher BMI categories, and non-initiation of breastfeeding among both BMI categories.ConclusionsPre-pregnancy severe‒moderate underweight, overweight, and obesity were significant risk factors for PD at 1mo and 6mo, and an upper-normal BMI was a significant risk factor at 1mo. Low birth weight, child anomalies, and non-initiation of breastfeeding were possible mediators. Pregnant women with severe to moderate underweight, overweight, obesity, and upper-normal BMI are important targets for PD prevention, and possible mediators may indicate PD risk among high-risk groups.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12884-026-09332-6. Background Postpartum depression (PD) is a type of depression that occurs within 12 months after childbirth [1, 2], with a reported prevalence ranging between 14.0% and 17.7% according to recent meta-analyses [3–5]. PD affects not only mothers’ subsequent mental and physical health, and social and occupational functioning, but also the mother–child relationship, child health, and partner interaction [6]. Both underweight and obesity have been identified as risk factors for depression in the general population [7, 8], and a meta-analysis has also reported associations between pre-pregnancy underweight, overweight, obesity, and PD [9]. In Japan, younger women tend to express a stronger desire to be slim than older women [10]. According to the 2023 National Health and Nutrition Survey of Japan, the prevalence of underweight among women aged 15–19, 20–29, and 30–39 years was 18.3%, 24.4%, and 17.9%, respectively [11]. The greater proportion of underweight young women suggests a larger number of severely underweight young women in Japan, and severely underweight pregnant women have been reported to have a higher risk of adverse birth outcomes [12]. However, to the best of our knowledge, no studies have investigated the association between the severity of pre-pregnancy underweight, defined by body mass index (BMI), and PD. Abnormal pre-pregnancy BMI has been associated with adverse birth outcomes [12–16] and non-initiation of breastfeeding [17–19]. Cesarean delivery, identified as a risk factor for postpartum depression (PD), can elevate interleukin-6 levels, a cytokine associated with depression [1, 20, 21]. Low birth weight, often related to prematurity, has also been linked to PD and may increase maternal stress [22]. Mothers with infant illness are likely to experience PD [1], with neonatal congenital anomalies significantly associated with PD [23]. Initiation of breastfeeding was significantly linked to a lower risk of PD [24], and feeding patterns affected postpartum depression through parenting self-efficacy [25]. Therefore, these findings suggest that cesarean delivery, low birth weight, child anomalies, and breastfeeding may mediate the association between abnormal pre-pregnancy BMI and PD; however, the mediating roles of these factors have not been fully elucidate. The objective of this epidemiological study was to determine the associations between pre-pregnancy BMI categories (severe–moderate and mild underweight, lower-normal, upper-normal, overweight, and obesity) and PD among the general Japanese population, characterized by a relatively low BMI distribution, and conducted an exploratory analysis of the potential mediating roles of adverse birth outcomes and non-initiation of breastfeeding. Materials and methods Participants The present study was based on data obtained from the Japan Environment and Children’s Study (JECS), an ongoing nationwide prospective birth cohort study conducted in 15 Regional Centers located across all geographical areas of Japan (Hokkaido, Miyagi, Fukushima, Chiba, Kanagawa, Koshin, Toyama, Aichi, Kyoto, Osaka, Hyogo, Tottori, Kochi, Fukuoka, and South Kyushu/Okinawa), and the study recruited approximately 100,000 pregnant women between January 2011 and March 2014 [26, 27]. In this study, longitudinal data from JECS were analyzed. Of 103,040 pregnancy registrations, after excluding second or later pregnancies in the same women, miscarriages and stillbirths, participants with missing pre-pregnancy BMI data, and those missing both 1- and 6-month PD data, 91,640 unique pregnant women were included in the analysis (Fig. 1).Fig. 1Flowchart of the study population selection Flowchart of the study population selection The JECS protocol was reviewed and approved by the Ministry of the Environment's Institutional Review Board on Epidemiological Studies and the Ethics Committee of all participating institutions. The study was conducted in accordance with the Declaration of Helsinki and other national regulations. Written informed consent was obtained from all participants. Pre-pregnancy BMI Maternal height and pre-pregnancy weight were transcribed from medical records; if unavailable, self-reported data were used. Pre-pregnancy BMI was calculated as maternal pre-pregnancy weight (kg) divided by the square of maternal height (m2), obtained from medical record transcriptions or self-reports. Pre-pregnancy BMI was classified into seven categories: ≤ 16.9 kg/m2 (severe–moderate underweight), 17.0‒18.4 kg/m2 (mild underweight), 18.5‒19.9 kg/m2 (lower-normal), 20.0‒22.9 kg/m2 (mid-normal: reference), 23.0‒24.9 kg/m2 (upper-normal), 25.0‒29.9 kg/m2 (overweight), and ≥ 30 kg/m2 (obesity) [12, 28, 29]. The mid-normal pre-pregnancy BMI group served as the reference group in the analyses. PD PD was assessed using a self-administered questionnaire, namely the Japanese version of the Edinburgh Postnatal Depression Scale (EPDS) at 1 and 6 months after childbirth (1mo and 6mo, respectively). The EPDS is a well-validated 10-item self-report scale designed to measure PD [30]. A cut-off score of 9, specifically validated for the Japanese population to account for cultural differences in symptom expression and reporting, demonstrated optimal sensitivity (75%) and specificity (93%) for identifying clinically diagnosed depression in Japanese mothers [23, 31]. Therefore, a score of ≥ 9 was used as the cutoff for PD in this study. Mediator variables The mediators were adverse birth outcomes, including cesarean delivery, low birth weight, child anomaly [12–16], and non-initiation of breastfeeding at 1mo [17–19]. Information on cesarean delivery, birth weight, and child anomalies was collected from medical record transcriptions. Non-initiation of breastfeeding was defined as not initiating breastfeeding by 1mo, as reported in maternal questionnaire responses collected at 1mo. Adjusted variables Adjusted variables included age, parity, smoking, alcohol consumption, marital status, employment status, education, household income, psychiatric disorders, and physical activity. A directed acyclic graph was used to identify potential confounders for inclusion in the statistical model and was constructed using Dagitty, which also identified the mediators [32] (Additional file 1). A questionnaire was distributed to participants during the first trimester (T1 questionnaire); if participation was delayed, it was administered in the second or third trimester. The questionnaire included items on marital status, employment status, smoking status, alcohol consumption, and medical history. Marital status was classified as married, unmarried, divorced, or bereaved. Maternal employment status was classified as employed, homemaker, unemployed, or student. Maternal smoking status was categorized as never smoking, quit smoking before pregnancy, or quit smoking during early pregnancy/continued smoking. Maternal alcohol consumption was classified as non-drinker, former drinker, or current drinker. The questionnaire also included items on the history of depression, dysautonomia, anxiety disorder, and schizophrenia; a history of psychiatric disorder was defined as a positive history of at least one of these conditions. The T2 questionnaire, distributed during the second or third trimesters, included questions on maternal educational attainment, categorized as ≤ 9 years (junior high school), 10–12 years (high school), 13–15 years (technical junior college, technical/vocational college, or associate degree), or ≥ 16 years (bachelor’s degree or postgraduate degree). This questionnaire also included questions on annual household income, categorized as ≤ 199, 200–399, 400–599, 600–799, 800–999, and ≥ 1000 thousand yen. Maternal age and parity were transcribed from medical records. Maternal age was categorized as ≤ 19, 20–24, 25–29, 30–34, 35–39, or ≥ 40 years, and parity was categorized as 0 or ≥ 1. Pre-pregnancy maternal physical activity was assessed using the Japanese version (short and self-administered) of the International Physical Activity Questionnaire (IPAQ) in the T1 questionnaire, and physical activity level was calculated as Met·min/day (metabolic equivalent task minutes per day) [33–35]. Physical activity, as defined in the IPAQ, includes all time spent in work-related, household, and leisure-time activities. Physical activity was categorized into quartiles for analysis. Statistical analysis Logistic regression models were used to estimate adjusted odds ratios (ORs) for the associations between BMI categories and PD at 1mo and 6mo. An exploratory causal mediation analysis was then conducted within a counterfactual framework to decompose the total effect of pre-pregnancy BMI into natural direct effects (NDEs) and natural indirect effects (NIEs) through candidate mediators [36]. Using this method, we estimated NDEs and NIEs of BMI for each of four mediators (cesarean delivery, low birth weight, child anomalies, and non-initiation of breastfeeding), while accounting for confounding. Each mediator was entered separately into the statistical model. We used the k-nearest neighbor (kNN) imputation method in the R package ‘VIM’ (V.4.5.2; R Foundation for Statistical Computing, Vienna, Austria) [37], introducing all the aforementioned outcomes, mediators, and adjusted variables, because the dataset contained missing values. The kNN method is a widely accepted single imputation method, and its validity has been established [38]. Two-sided P values < 0.05 were considered statistically significant. Logistic regression and mediation analyses were performed using Stata statistical software for Windows (version 19.0; StataCorp, College Station, TX, USA). The mediation analyses were performed using the ‘mediate’ command [39]. Results Table 1 presents the participants' characteristics. The proportions of BMI ≤ 16.9, 17.0‒18.4, 18.5‒19.9, 20.0‒22.9, 23.0‒24.9, 25.0‒29.9, and ≥ 30.0 kg/m2 were 2.7%, 12.9%, 24.4%, 38.4%, 10.5%, 8.4%, and 2.6%, respectively. The prevalence of PD at 1mo and 6mo was 14.1% and 10.9%, respectively. The prevalences of 1mo and 6mo PD within each BMI category are presented in Supplemental Table 1.Table 1Characteristics (N = 91,640)N%Age (years) ≤2410,45911.4 25‒2926,95229.4 30‒3431,96634.9 35‒3919,02220.8 ≥403,2393.5 Missing20.0Parity 038,14241.6 ≥151,25055.9 Missing2,2482.5Smoking Never smoking52,80157.6 Quit smoking before pregnancy21,10623.0 Quit smoking during early pregnancy/continued smoking16,39117.9 Missing1,3421.5Alcohol Non-drinker30,10032.9 Former drinker57,46262.7 Drinker2,4892.7 Missing1,5891.7Marital statusMarried86,50894.4 Unmarried3,2743.6 Divorced or bereavement7740.8 Missing1,0841.2Employment status Employed59,70465.2 Homemaker or unemployed28,11730.7 Student5320.6 Missing3,2873.6Household income (Ten thousand yen) ≤1994,7045.1 200‒39928,97831.6 400‒59927,89130.4 600‒79913,50914.7 800‒8995,6326.2 ≥1,0003,6023.9 Missing7,3248.0Educational attainment Junior high school4,2104.6 High school28,25630.8 Technical junior college, technical/vocational college, or associate degree38,11241.6 Bachelor's degree, or postgraduate degree19,70121.5 Missing1,3611.5Psychiatric diseases Negative83,83291.5 Positive7,1217.8 Missing6870.8IPAQ 1 st quartile(0‒ < 28.4)23,03525.1 2nd quartile(28.4‒ < 125.4)21,81423.8 3rd quartile(125.4‒ < 438.9)22,16624.2 4th quartile(483.9‒12,589.7)22,33724.4 missing2,2882.5Child anomaly Negative81,26888.7 Positive8,3119.1 Missing2,0612.3Cesarean Negative73,71580.4 Positive17,70319.3 Missing2220.2Low birth weight Negative83,74391.4 Positive7,8448.6 Missing530.1Breastfeeding Initiated 1 month postpartum89,46397.6 Not initiated by 1 month postpartum1,3391.5 Missing8380.9Body mass index (kg/m2) ≤ 16.92,8052.7 17.0‒18.413,37912.9 18.5‒19.925,17124.4 20.0‒22.939,73838.4 23.0–24.910,87910.5 25.0–29.98,6828.4 ≥ 30.02,7112.6Postpartum depression at 1mo Postpartum depression negative (≤ 8)76,97684.0 Postpartum depression positive (≥ 9)12,93814.1 Missing1,7261.9 Postpartum depression at 6 months Postpartum depression negative (≤ 8) 75,850 82.8 Postpartum depression positive (≥ 9) 10,007 10.9 Missing 5,783 6.3Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityIPAQ International Physical Activity Questionnaire Table 1Characteristics (N = 91,640)N%Age (years) ≤2410,45911.4 25‒2926,95229.4 30‒3431,96634.9 35‒3919,02220.8 ≥403,2393.5 Missing20.0Parity 038,14241.6 ≥151,25055.9 Missing2,2482.5Smoking Never smoking52,80157.6 Quit smoking before pregnancy21,10623.0 Quit smoking during early pregnancy/continued smoking16,39117.9 Missing1,3421.5Alcohol Non-drinker30,10032.9 Former drinker57,46262.7 Drinker2,4892.7 Missing1,5891.7Marital statusMarried86,50894.4 Unmarried3,2743.6 Divorced or bereavement7740.8 Missing1,0841.2Employment status Employed59,70465.2 Homemaker or unemployed28,11730.7 Student5320.6 Missing3,2873.6Household income (Ten thousand yen) ≤1994,7045.1 200‒39928,97831.6 400‒59927,89130.4 600‒79913,50914.7 800‒8995,6326.2 ≥1,0003,6023.9 Missing7,3248.0Educational attainment Junior high school4,2104.6 High school28,25630.8 Technical junior college, technical/vocational college, or associate degree38,11241.6 Bachelor's degree, or postgraduate degree19,70121.5 Missing1,3611.5Psychiatric diseases Negative83,83291.5 Positive7,1217.8 Missing6870.8IPAQ 1 st quartile(0‒ < 28.4)23,03525.1 2nd quartile(28.4‒ < 125.4)21,81423.8 3rd quartile(125.4‒ < 438.9)22,16624.2 4th quartile(483.9‒12,589.7)22,33724.4 missing2,2882.5Child anomaly Negative81,26888.7 Positive8,3119.1 Missing2,0612.3Cesarean Negative73,71580.4 Positive17,70319.3 Missing2220.2Low birth weight Negative83,74391.4 Positive7,8448.6 Missing530.1Breastfeeding Initiated 1 month postpartum89,46397.6 Not initiated by 1 month postpartum1,3391.5 Missing8380.9Body mass index (kg/m2) ≤ 16.92,8052.7 17.0‒18.413,37912.9 18.5‒19.925,17124.4 20.0‒22.939,73838.4 23.0–24.910,87910.5 25.0–29.98,6828.4 ≥ 30.02,7112.6Postpartum depression at 1mo Postpartum depression negative (≤ 8)76,97684.0 Postpartum depression positive (≥ 9)12,93814.1 Missing1,7261.9 Postpartum depression at 6 months Postpartum depression negative (≤ 8) 75,850 82.8 Postpartum depression positive (≥ 9) 10,007 10.9 Missing 5,783 6.3Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityIPAQ International Physical Activity Questionnaire Characteristics (N = 91,640) Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesity IPAQ International Physical Activity Questionnaire Table 2 presents adjusted ORs for PD at 1mo and 6mo. Severe–moderate underweight had significantly higher ORs for PD at both time points (1mo: OR = 1.21, 95% confidence interval [CI]: 1.08‒1.35; 6mo: OR = 1.16, 95% CI: 1.03‒1.31). Upper-normal BMI was significantly associated with a higher OR for PD only at 1mo (OR = 1.09, 95% CI: 1.02‒1.17). Overweight and obesity had significantly higher ORs for PD at both time points (overweight—1mo: OR = 1.18, 95% CI: 1.10‒1.27; 6mo: OR = 1.15, 95% CI: 1.07‒1.24: obesity—1mo: OR = 1.35, 95% CI: 1.21‒1.51; 6mo: OR = 1.20, 95% CI: 1.06‒1.36).Table 2Odds ratios for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95% CIPOR95% CIPBody mass index ≤ 16.91.211.08to1.350.0011.161.03to1.310.013 17.0‒18.41.030.97to1.100.2690.960.90to1.030.242 18.5‒19.91.000.95to1.050.8570.970.92to1.030.320 20.0‒22.91.001.00 23.0‒24.91.091.02to1.170.0081.030.96to1.110.385 25.0‒29.91.181.10to1.27< 0.0011.151.07to1.24< 0.001 ≥ 30.01.351.21to1.51< 0.0011.201.06to1.360.004Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratio Table 2Odds ratios for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95% CIPOR95% CIPBody mass index ≤ 16.91.211.08to1.350.0011.161.03to1.310.013 17.0‒18.41.030.97to1.100.2690.960.90to1.030.242 18.5‒19.91.000.95to1.050.8570.970.92to1.030.320 20.0‒22.91.001.00 23.0‒24.91.091.02to1.170.0081.030.96to1.110.385 25.0‒29.91.181.10to1.27< 0.0011.151.07to1.24< 0.001 ≥ 30.01.351.21to1.51< 0.0011.201.06to1.360.004Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratio Odds ratios for postpartum depression (N = 91,640) Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesity Adjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activity The k-nearest neighbors imputation method was used to impute missing values CI Confidence interval, PD Postpartum depression, OR Odds ratio In the mediation analysis of cesarean delivery (Table 3), no significant NIE was observed. In the mediation analysis of low birth weight (Table 4), severe–moderate underweight had significantly higher NIE ORs for PD at both time points (1mo: OR = 1.014, 95% CI: 1.008‒1.021; 6mo: OR = 1.007, 95% CI: 1.000‒1.014). Although the total effects (TEs) were not significant, mild underweight had significantly higher NIE ORs for PD at both time points, and lower-normal BMI had a significantly higher NDE OR for PD at 1mo. In the mediation analysis of anomalies (Table 5), obesity was associated with significantly higher ORs for PD at both time points (1mo: OR = 1.002, 95% CI: 1.000‒1.004; 6mo: OR = 1.003, 95% CI: 1.000‒1.005). Although the TEs were not significant, mild underweight was associated with significantly higher NIE ORs for PD at both time points. In the mediation analysis of non-initiation of breastfeeding (Table 6), severe–moderate underweight was significantly associated with higher NIE ORs for PD at both time points (1mo: OR = 1.009, 95% CI: 1.004‒1.013, 6mo: OR = 1.005, 95% CI: 1.002‒1.009). Overweight and obesity also had significantly higher ORs for PD at both time points (overweight—1mo: OR = 1.009, 95% CI: 1.006‒1.012; 6mo: OR = 1.006, 95% CI: 1.003‒1.008; obesity—1mo: OR = 1.013, 95% CI: 1.008‒1.019; 6mo: OR = 1.008, 95% CI: 1.004‒1.013). Upper-normal BMI was significantly associated with a higher NIE OR for PD at 1mo (OR = 1.006, 95% CI: 1.003‒1.008), but it was also significantly associated at 6mo without the TE significance.Table 3Mediation analysis of cesarean delivery for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95% CIPOR95% CIPNatural indirect effect ≤ 16.90.9980.996to1.0000.0680.9990.997to1.0010.213 17.0–18.40.9980.996to1.0000.0550.9990.997to1.0010.202 18.5‒19.90.9990.997to1.0000.0560.9990.997to1.0010.203 20.0‒22.91.0001.000 23.0‒24.91.0021.000to1.0040.0551.0020.999to1.0040.202 25.0‒29.91.0041.000to1.0080.0521.0030.998to1.0080.199 ≥ 30.01.0081.000to1.0160.0511.0060.997to1.0150.198Natural direct effect Body mass index≤ 16.91.2011.082to1.3330.0011.1581.032to1.3000.01217.0–18.41.0350.976to1.0980.2480.9630.902to1.0280.25618.5‒19.90.9970.950to1.0460.8990.9750.925to1.0270.33820.0‒22.91.0001.00023.0‒24.91.0861.021to1.1570.0091.0300.961to1.1030.40425.0‒29.91.1671.092to1.248 < 0.0011.1391.060to1.225< 0.001≥ 30.01.3241.190to1.474 < 0.0011.1841.051to1.3340.005Total effect Body mass index≤ 16.91.1991.080to1.3310.0011.1571.031to1.2980.01317.0‒18.41.0330.974to1.0960.2730.9620.901to1.0260.24018.5‒19.90.9960.949to1.0440.8540.9740.924to1.0260.31920.0‒22.91.0001.00023.0‒24.91.0891.023to1.1590.0081.0310.963to1.1050.38025.0‒29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3351.200to1.486< 0.0011.1911.058to1.3410.004Mediator: cesarean deliveryBody mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratioTable 4Mediation analysis of low birth weight for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95%CIPOR95%CIPNatural indirect effect Body mass index≤ 16.91.0141.008to1.021 < 0.0011.0071.000to1.0140.04817.0‒18.41.0061.003to1.009 < 0.0011.0031.000to1.0060.04818.5‒19.91.0021.001to1.004 < 0.0011.0011.000to1.0020.05420.0‒22.91.0001.00023.0‒24.91.0000.999to1.0010.7931.0001.000to1.0010.79425.0‒29.91.0010.999to1.0020.3191.0001.000to1.0010.361≥ 30.01.0010.999to1.0030.1591.0011.000to1.0020.234Natural direct effect Body mass index≤ 16.91.1821.064to1.3130.0021.1491.024to1.2900.01817.0‒18.41.0270.969to1.0890.3700.9590.899to1.0240.21018.5‒19.90.9930.947to1.0420.7790.9730.923to1.0250.30120.0‒22.91.0001.00023.0‒24.91.0881.023to1.1590.0081.0310.962to1.1050.38325.0‒29.91.1711.095to1.252 < 0.0011.1431.063to1.228< 0.001≥ 30.01.3341.199to1.484 < 0.0011.1911.057to1.3400.004Total effect Body mass index≤ 16.91.1991.080to1.3320.0011.1571.031to1.2990.01317.0‒18.41.0340.975to1.0960.2700.9620.901to1.0270.24318.5‒19.90.9960.949to1.0440.8570.9740.924to1.0260.32120.0‒22.91.0001.00023.0‒24.91.0891.023to1.1590.0081.0310.962to1.1050.38225.0‒29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3361.200to1.486< 0.0011.1911.058to1.3410.004Mediator: low birth weightBody mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratioTable 5Mediation analysis of child anomaly for postpartum depression (N = 91,640)PD at 1moPD at 6moOR95%CIpOR95%CIpNatural indirect effect Body mass index≤16.91.0010.999to1.0020.3231.0010.999to1.0030.31917.0–18.41.0011.000to1.0020.0251.0011.000to1.0030.01918.5–19.91.0001.000to1.0010.5821.0000.999to1.0010.58020.0–22.91.0001.00023.0–24.91.0000.999to1.0010.9271.0000.999to1.0010.92725.0–29.91.0000.999to1.0010.5021.0000.998to1.0010.501≥ 30.01.0021.000to1.0040.0271.0031.000to1.0050.021Natural direct effect Body mass index≤16.91.1981.079to1.3300.0011.1561.030to1.2970.01417.0–18.41.0320.973to1.0940.2890.9600.900to1.0250.22418.5–19.90.9950.949to1.0440.8500.9730.923to1.0260.31720.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0320.963to1.1050.37725.0–29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3321.197to1.482< 0.0011.1881.055to1.3370.004Total effect Body mass index≤16.91.1991.080to1.3310.0011.1571.031to1.2980.01317.0–18.41.0330.975to1.0960.2730.9620.901to1.0260.24118.5–19.90.9960.949to1.0440.8560.9740.924to1.0260.32120.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0320.963to1.1050.37625.0–29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3351.199to1.485< 0.0011.1911.058to1.3410.004Mediator: child anomalyBody mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obeseAdjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activityThe k-nearest neighbor imputation method used for missing value imputationsCI Confidence interval, PD Postpartum depression, OR Odds ratioTable 6Mediation analysis of non-initiation of breastfeeding for postpartum depression (N = 91,640)PD at 1moPD at 6moOR95%CIpOR95%CIpNatural indirect effect Body mass index≤16.91.0091.004to1.013 < 0.0011.0051.002to1.0090.00217.0–18.41.0000.998to1.0010.8371.0000.999to1.0010.83718.5–19.90.9990.998to1.0010.3491.0000.999to1.0000.35220.0–22.91.0001.00023.0–24.91.0061.003to1.008< 0.0011.0031.002to1.005< 0.00125.0–29.91.0091.006to1.012< 0.0011.0061.003to1.008< 0.001≥ 30.01.0131.008to1.019< 0.0011.0081.004to1.013< 0.001Natural direct effect Body mass index≤16.91.1911.073to1.3230.0011.1521.027to1.2930.01617.0–18.41.0340.975to1.0960.2640.9620.902to1.0270.24718.5–19.90.9960.950to1.0450.8790.9740.924to1.0270.33220.0–22.91.0001.00023.0–24.91.0831.017to1.1520.0131.0280.959to1.1010.44125.0–29.91.1631.088to1.244< 0.0011.1371.057to1.2230.001≥ 30.01.3211.187to1.470< 0.0011.1831.051to1.3320.005Total effect Body mass index≤16.91.2021.082to1.3340.0011.1581.032to1.3000.01317.0–18.41.0340.975to1.0960.2660.9620.902to1.0270.24618.5–19.90.9960.949to1.0450.8600.9740.924to1.0270.32520.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0310.962to1.1050.38525.0–29.91.1731.097to1.255< 0.0011.1431.063to1.229< 0.001≥ 30.01.3391.203to1.490< 0.0011.1931.060to1.3440.004Mediator: non-initiation of breastfeedingBody mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obeseAdjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activityThe k-nearest neighbor imputation method used for missing value imputationsCI Confidence interval, PD Postpartum depression, OR Odds ratio Table 3Mediation analysis of cesarean delivery for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95% CIPOR95% CIPNatural indirect effect ≤ 16.90.9980.996to1.0000.0680.9990.997to1.0010.213 17.0–18.40.9980.996to1.0000.0550.9990.997to1.0010.202 18.5‒19.90.9990.997to1.0000.0560.9990.997to1.0010.203 20.0‒22.91.0001.000 23.0‒24.91.0021.000to1.0040.0551.0020.999to1.0040.202 25.0‒29.91.0041.000to1.0080.0521.0030.998to1.0080.199 ≥ 30.01.0081.000to1.0160.0511.0060.997to1.0150.198Natural direct effect Body mass index≤ 16.91.2011.082to1.3330.0011.1581.032to1.3000.01217.0–18.41.0350.976to1.0980.2480.9630.902to1.0280.25618.5‒19.90.9970.950to1.0460.8990.9750.925to1.0270.33820.0‒22.91.0001.00023.0‒24.91.0861.021to1.1570.0091.0300.961to1.1030.40425.0‒29.91.1671.092to1.248 < 0.0011.1391.060to1.225< 0.001≥ 30.01.3241.190to1.474 < 0.0011.1841.051to1.3340.005Total effect Body mass index≤ 16.91.1991.080to1.3310.0011.1571.031to1.2980.01317.0‒18.41.0330.974to1.0960.2730.9620.901to1.0260.24018.5‒19.90.9960.949to1.0440.8540.9740.924to1.0260.31920.0‒22.91.0001.00023.0‒24.91.0891.023to1.1590.0081.0310.963to1.1050.38025.0‒29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3351.200to1.486< 0.0011.1911.058to1.3410.004Mediator: cesarean deliveryBody mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratio Mediation analysis of cesarean delivery for postpartum depression (N = 91,640) Mediator: cesarean delivery Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesity Adjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activity The k-nearest neighbors imputation method was used to impute missing values CI Confidence interval, PD Postpartum depression, OR Odds ratio Table 4Mediation analysis of low birth weight for postpartum depression (N = 91,640)PD at 1 monthPD at 6 monthsOR95%CIPOR95%CIPNatural indirect effect Body mass index≤ 16.91.0141.008to1.021 < 0.0011.0071.000to1.0140.04817.0‒18.41.0061.003to1.009 < 0.0011.0031.000to1.0060.04818.5‒19.91.0021.001to1.004 < 0.0011.0011.000to1.0020.05420.0‒22.91.0001.00023.0‒24.91.0000.999to1.0010.7931.0001.000to1.0010.79425.0‒29.91.0010.999to1.0020.3191.0001.000to1.0010.361≥ 30.01.0010.999to1.0030.1591.0011.000to1.0020.234Natural direct effect Body mass index≤ 16.91.1821.064to1.3130.0021.1491.024to1.2900.01817.0‒18.41.0270.969to1.0890.3700.9590.899to1.0240.21018.5‒19.90.9930.947to1.0420.7790.9730.923to1.0250.30120.0‒22.91.0001.00023.0‒24.91.0881.023to1.1590.0081.0310.962to1.1050.38325.0‒29.91.1711.095to1.252 < 0.0011.1431.063to1.228< 0.001≥ 30.01.3341.199to1.484 < 0.0011.1911.057to1.3400.004Total effect Body mass index≤ 16.91.1991.080to1.3320.0011.1571.031to1.2990.01317.0‒18.41.0340.975to1.0960.2700.9620.901to1.0270.24318.5‒19.90.9960.949to1.0440.8570.9740.924to1.0260.32120.0‒22.91.0001.00023.0‒24.91.0891.023to1.1590.0081.0310.962to1.1050.38225.0‒29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3361.200to1.486< 0.0011.1911.058to1.3410.004Mediator: low birth weightBody mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesityAdjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activityThe k-nearest neighbors imputation method was used to impute missing valuesCI Confidence interval, PD Postpartum depression, OR Odds ratio Mediation analysis of low birth weight for postpartum depression (N = 91,640) Mediator: low birth weight Body mass index (kg/m2): ≤ 16.9, severe–moderate underweight; 17.0‒18.4, mild underweight; 18.5‒19.9, lower-normal; 20.0‒22.9, mid-normal; 23.0‒24.9, upper-normal; 25.0‒29.9, overweight; ≥ 30, obesity Adjusted for age, parity, smoking, alcohol consumption, marital status, employment, education, household income, psychiatric disorders, and physical activity The k-nearest neighbors imputation method was used to impute missing values CI Confidence interval, PD Postpartum depression, OR Odds ratio Table 5Mediation analysis of child anomaly for postpartum depression (N = 91,640)PD at 1moPD at 6moOR95%CIpOR95%CIpNatural indirect effect Body mass index≤16.91.0010.999to1.0020.3231.0010.999to1.0030.31917.0–18.41.0011.000to1.0020.0251.0011.000to1.0030.01918.5–19.91.0001.000to1.0010.5821.0000.999to1.0010.58020.0–22.91.0001.00023.0–24.91.0000.999to1.0010.9271.0000.999to1.0010.92725.0–29.91.0000.999to1.0010.5021.0000.998to1.0010.501≥ 30.01.0021.000to1.0040.0271.0031.000to1.0050.021Natural direct effect Body mass index≤16.91.1981.079to1.3300.0011.1561.030to1.2970.01417.0–18.41.0320.973to1.0940.2890.9600.900to1.0250.22418.5–19.90.9950.949to1.0440.8500.9730.923to1.0260.31720.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0320.963to1.1050.37725.0–29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3321.197to1.482< 0.0011.1881.055to1.3370.004Total effect Body mass index≤16.91.1991.080to1.3310.0011.1571.031to1.2980.01317.0–18.41.0330.975to1.0960.2730.9620.901to1.0260.24118.5–19.90.9960.949to1.0440.8560.9740.924to1.0260.32120.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0320.963to1.1050.37625.0–29.91.1721.096to1.253< 0.0011.1431.063to1.229< 0.001≥ 30.01.3351.199to1.485< 0.0011.1911.058to1.3410.004Mediator: child anomalyBody mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obeseAdjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activityThe k-nearest neighbor imputation method used for missing value imputationsCI Confidence interval, PD Postpartum depression, OR Odds ratio Mediation analysis of child anomaly for postpartum depression (N = 91,640) Mediator: child anomaly Body mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obese Adjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activity The k-nearest neighbor imputation method used for missing value imputations CI Confidence interval, PD Postpartum depression, OR Odds ratio Table 6Mediation analysis of non-initiation of breastfeeding for postpartum depression (N = 91,640)PD at 1moPD at 6moOR95%CIpOR95%CIpNatural indirect effect Body mass index≤16.91.0091.004to1.013 < 0.0011.0051.002to1.0090.00217.0–18.41.0000.998to1.0010.8371.0000.999to1.0010.83718.5–19.90.9990.998to1.0010.3491.0000.999to1.0000.35220.0–22.91.0001.00023.0–24.91.0061.003to1.008< 0.0011.0031.002to1.005< 0.00125.0–29.91.0091.006to1.012< 0.0011.0061.003to1.008< 0.001≥ 30.01.0131.008to1.019< 0.0011.0081.004to1.013< 0.001Natural direct effect Body mass index≤16.91.1911.073to1.3230.0011.1521.027to1.2930.01617.0–18.41.0340.975to1.0960.2640.9620.902to1.0270.24718.5–19.90.9960.950to1.0450.8790.9740.924to1.0270.33220.0–22.91.0001.00023.0–24.91.0831.017to1.1520.0131.0280.959to1.1010.44125.0–29.91.1631.088to1.244< 0.0011.1371.057to1.2230.001≥ 30.01.3211.187to1.470< 0.0011.1831.051to1.3320.005Total effect Body mass index≤16.91.2021.082to1.3340.0011.1581.032to1.3000.01317.0–18.41.0340.975to1.0960.2660.9620.902to1.0270.24618.5–19.90.9960.949to1.0450.8600.9740.924to1.0270.32520.0–22.91.0001.00023.0–24.91.0891.023to1.1590.0081.0310.962to1.1050.38525.0–29.91.1731.097to1.255< 0.0011.1431.063to1.229< 0.001≥ 30.01.3391.203to1.490< 0.0011.1931.060to1.3440.004Mediator: non-initiation of breastfeedingBody mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obeseAdjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activityThe k-nearest neighbor imputation method used for missing value imputationsCI Confidence interval, PD Postpartum depression, OR Odds ratio Mediation analysis of non-initiation of breastfeeding for postpartum depression (N = 91,640) Mediator: non-initiation of breastfeeding Body mass index (kg/m2);≤ 16.9, severe–moderate underweight; 17.0–18.4, mild underweight; 18.5–19.9, lower-normal; 20.0–22.9, mid-normal; 23.0–24.9, upper-normal; 25.0–29.9, overweight;≥ 30, obese Adjusted for age, parity, smoking, alcohol consumption, marital status, job, education, household income, psychiatric diseases, and physical activity The k-nearest neighbor imputation method used for missing value imputations CI Confidence interval, PD Postpartum depression, OR Odds ratio Discussion In this study, we found that pre-pregnancy severe–moderate underweight, overweight, and obese were significantly associated with PD at 1mo and 6mo, and upper-normal BMI was significantly associated with PD at 1mo. Although the effects were relatively small, significant mediating effects were observed for low birth weight among lower BMI categories, child anomalies among higher BMI categories, and non-initiation of breastfeeding among both lower and higher BMI categories. To the best of our knowledge, this is the first study to elucidate the independent associations of severe–moderate underweight and upper-normal BMI with PD and the possible mediating roles of low birth weight, anomalies, and non-initiation of breastfeeding. Our results are generally consistent with those of a recent meta-analysis reporting an increased risk of PD among women who were underweight, overweight, or obese before pregnancy [9]. In the Japanese population, where underweight is relatively common among young women [11], more severe underweight has been associated with adverse perinatal outcomes [12]. Accordingly, we observed a substantial association between severe–moderate underweight and PD, whereas mild underweight was not significantly associated with PD. In addition, upper-normal BMI, as well as overweight and obesity, were associated with PD, although these associations were evident only at 1mo. Regarding the relationship between pre-pregnancy overweight and obesity and PD, hypothalamic–pituitary–adrenal axis dysregulation, leptin insufficiency, and leptin resistance have been proposed as potential biological mechanisms [9, 40, 41]. Obesity during pregnancy is often stigmatized and may contribute to poorer mental health among pregnant women [42, 43]. In the present study, upper-normal BMI was significantly associated with PD at 1mo. As previously stated, since younger women in Japan have a stronger desire to be slim than older women [10], we speculate that the stigmatization of upper-normal BMI may linked to the increased risk. Pre-pregnancy underweight is linked to nutrient deficiency, and low levels of omega-3 polyunsaturated fatty acids have been associated with an increased risk of PD [44]. A systematic review of micronutrient levels reported that vitamin D deficiency had the strongest evidence of an association with PD and that lower levels of iron, folate, and zinc were potential risk factors for PD [45]. In the present study, severe‒moderate underweight had a significant association, whereas mild underweight was not. One possible explanation is that, in Japan, a strong desire to be slim may be perceived as fulfilled among women with mild underweight, which can play a protective role in PD. Women who are underweight, overweight, or obese before pregnancy have a higher risk of pregnancy complications than women with a normal BMI [12, 13, 46], and these complications may increase the risk of PD. In the present study, low birth weight, child anomalies, and non-initiation of breastfeeding had significant NIE ORs for PD, indicating mediating effects on the pathway between pre-pregnancy BMI and PD. Low birth weight was more common among underweight women, and severe–moderate underweight women had higher ORs for low birth weight and small for gestational age than mildly underweight women. This may explain the stronger mediating effects observed in the severe–moderate underweight group. Women with obesity have a higher risk of child anomalies [15, 47]. Thus, the significant indirect mediation effects of obesity in the presence of significant total effects were plausible; however, the point estimates of the ORs at 1mo and 6mo were 1.002 and 1.003, respectively. Therefore, the mediation effects appear to be small in magnitude. Being underweight or obese has been significantly associated with non-initiation of breastfeeding [17–19]. The significant indirect mediation effects of severe‒moderate underweight, overweight, and obesity in the presence of significant total effects are therefore plausible. In addition, upper-normal BMI had a significant NIE in the presence of a significant total effect at 1mo. As the effect of upper-normal BMI on breastfeeding initiation remains unclear, further studies are needed to elucidate the mediating role of breastfeeding in this population. Cesarean delivery showed no significant indirect mediation effect. Obesity has been identified as a risk factor for cesarean delivery [48–50], and cesarean delivery has also been reported as a risk factor for PD in a recent meta-analysis [51]. In contrast, a recent Japanese study reported no significant association [50], and another analysis using the same cohort data reported a marginal OR (1.10; 95% CI, 1.00–1.21) for PD at 1mo and no significant association at 6mo [52]. The reasons for the non-significant NIE for cesarean delivery were unknown, but it may reflect the relatively weak association between cesarean delivery and PD in this cohort. Some mediating factors had significant NIE ORs despite non-significant total effects; however, all corresponding point-estimate ORs, except for the low birth weight OR for 1mo PD among mild underweight participants, were < 1.005. Therefore, these effects appear to be small in magnitude, if present at all. Upper-normal BMI showed statistical significance at 1 month but not at 6 months, which may be due to the lower prevalence of PD at 6 mo. In the mediation analysis of non-initiation of breastfeeding, the NIE of upper-normal BMI was statistically significant for both 1 and 6 months. However, a significant NIE in the absence of a significant TE has limited interpretability. This study has several limitations. First, PD positivity was defined based on a self-administered questionnaire and was not clinically diagnosed. However, we used a validated screening questionnaire that has been widely employed in large-scale epidemiological studies. Second, some maternal pre-pregnancy BMI values were calculated using self-reported data when maternal height and pre-pregnancy weight were unavailable in the medical records. However, 99.82% of the weight data were obtained from the medical records, the influence of self-report weight, which tends to be underestimated compared with actual weight. seemed limited. Third, because gestational weight gain is also related to PD; however, because it occurs after the main exposure (pre-pregnancy BMI), it was not included as an adjusted variable. Fourth, in the exploratory mediation analysis, potential confounding between each mediator and postpartum depressive symptoms, as well as interactions among the mediators, were not considered. This was due to the relatively low odds ratios of the natural indirect effects observed, which limited the feasibility of more complex model analyses. Furthermore, other BMI-related perinatal complications, such as gestational diabetes and pre-eclampsia, were not considered in this study. Fifth, in the 1mo postpartum questionnaire, the start of breastfeeding and depressive symptoms were assessed simultaneously. If a respondent indicated that breastfeeding had started, it is assumed to have begun prior to the 1mo postpartum depression (PD) assessment. However, given the potential overlap in timing between these factors, the interpretation of breastfeeding as a mediator for 1mo PD should be approached with caution. In contrast, its mediating role for 6mo PD had no temporal concerns. Sixth, due to the relatively large sample size of this study, even small natural indirect effects reached statistical significance. However, these small effects may have limited clinical significance. Seventh, although the variable 'psychiatric disorders' was included as an adjustment factor, the severity of depressive symptoms was not measured, which may have led to some residual confounding. Eighth, participants who did not complete both the 1mo and 6mo questionnaires were excluded. These individuals may have experienced more depressive symptoms, which could have reduced the statistical power of the study. Finally, generalizability to populations other than the Japanese is limited because Japanese younger women tend to express a stronger desire to be slim [10], and there are the greater proportion of underweight young women [11]; however, since previous studies have suggested that a substantial proportion of young women in China and South Korea are underweight [53, 54], the findings of this study may be generalizable to these populations. Conclusions Pre-pregnancy severe‒moderate underweight, overweight, and obesity were significant risk factors for PD at 1mo and 6mo, and upper-normal BMI was a significant risk factor for PD at 1mo. Furthermore, possible mediating roles for PD were observed in the exploratory analysis as follows: low birth weight among the lower BMI categories, child anomalies among the higher BMI categories, and non-initiation of breastfeeding among both higher and lower BMI categories. In addition to overweight and obesity, severe‒moderate underweight and upper-normal BMI represent important targets for PD prevention. Moreover, possible mediators, such as low birth weight, child anomalies, and non-initiation of breastfeeding, may serve as indicators of PD risk among high-risk groups. These findings suggest that promoting the initiation of breastfeeding may be particularly important among women with abnormal BMI. Furthermore, women who experience low birth weight, infant abnormalities, or difficulty initiating breastfeeding may benefit from additional support to reduce the risk of postpartum depression. Supplementary Information Additional file 1: Directed acyclic graph illustrating the hypothesized causal and mediation relationships. Additional file 2: Supplemental Table 1 Prevalence of postpartum depression. Additional file 1: Directed acyclic graph illustrating the hypothesized causal and mediation relationships. Additional file 2: Supplemental Table 1 Prevalence of postpartum depression.