Abstract
Purpose
Labor induction using medications such as synthetic oxytocin (synOT) and prostaglandins (PG) has become increasingly common; however, its implications for mother-to-infant bonding remain unclear. These agents may plausibly influence early bonding by altering peripartum oxytocin-related physiology and by affecting early breastfeeding establishment. We examined whether labor induction and specific induction agents are associated with postpartum mother-to-infant bonding trajectories and evaluated breastfeeding mode as a potential mediator.
Methods
We analyzed data from the Japan Environment and Children’s Study (JECS). Among 104,059 fetal records, we included 58,384 mother–infant dyads with complete bonding data at 1, 6, and 12 months postpartum (56.1%; complete-case analysis), restricted to liveborn, term, singleton vaginal deliveries. Mother-to-infant bonding was assessed as mothers’ self-reported bonding-related feelings using the Mother-to-Infant Bonding Scale (MIBS). Induction status and induction agents (synOT, PG) were obtained from transcribed medical records. Linear mixed-effects models tested associations with bonding trajectories, and mediation analyses evaluated whether 1-month feeding mode (exclusive formula vs. any breastfeeding) accounted for observed associations.
Results
Induction was associated with a time-varying pattern in mother-to-infant bonding (time × induction: β = -0.057, 95% CI [-0.067, -0.046], p < 0.001), although contrasts at 1, 6, and 12 months were not significant (β = 0.013, p = 0.109; β = 0.005, p = 0.482; β = 0.000, p = 0.953). SynOT was associated with poorer bonding at 1 and 6 months (β = 0.027, p = 0.004; β = 0.016, p = 0.024), but not at 12 months (β = 0.010, p = 0.205). Mediation via feeding was small for induction (indirect effect = 0.001, 95% CI [0.0002, 0.001]) and not supported for synOT (indirect effect = 0.0004, 95% CI [-0.00009, 0.001]).
Conclusion
Labor induction and synOT exposure were associated with poorer early postpartum mother-to-infant bonding, with differences attenuating over the first postpartum year. Exclusive formula feeding was associated with poorer bonding and may contribute to early postpartum differences, supporting the clinical value of proactive breastfeeding support and attention to maternal wellbeing while reassuring families that early challenges often ease over time.
Keywords: Labor induction, Synthetic oxytocin, Breastfeeding, Mother-to-infant bonding, Japan Environment and Children’s Study
Highlights
Labor induction was associated with a different trajectory of mother-to-infant bonding across the first postpartum year.
Synthetic oxytocin use for labor induction was associated with poorer early mother-to-infant bonding compared with no oxytocin use.
Bonding improved over time in all mothers, and the between-group differences attenuated by 12 months postpartum.
Breastfeeding protected early bonding problems following induction.
Early breastfeeding support and maternal wellbeing care may reduce short-term bonding difficulties after induction.
Introduction
Parents experience profound physical, psychological, and social changes during the perinatal period that shape the bonding emotions they develop toward their infant. Mother-to-infant bonding refers to the mother’s subjective emotional experience toward her infant in the early postpartum period. This caregiving-related emotional experience is considered important for sensitive caregiving and has been discussed as relevant to infants’ subsequent socio-emotional development [25]. Although bonding disturbance has not been formally established as a diagnostic entity in major classification systems such as the ICD or DSM [6], the concept of mother-to-infant bonding is well recognized in perinatal research and clinical practice. Indeed, early bonding has been a central focus of obstetric, neonatal, and pediatric nursing care for decades [25]. For example, a perinatal mental health care manual issued by the Japan Association of Obstetricians and Gynecologists explicitly addresses the formation and disturbance of mother-to-infant bonding and describes screening approaches within perinatal care [21]. Impaired mother-to-infant bonding can hinder appropriate parenting behaviors and increase the risk of long-term behavioral problems in children. Alongside psychosis and suicidal depression, mother-to-infant bonding failure has been recognized as one of the most urgent challenges in perinatal psychiatry [7], highlighting the importance of clarifying factors that foster or hinder poor mother-to-infant bonding.
Mother-to-infant bonding is shaped by experiences during pregnancy, childbirth, and postpartum [40]. Delivery mode influences bonding, with higher mother-to-infant bonding reported after vaginal than caesarean delivery [12, 44]. Labor induction and augmentation methods—particularly pharmacological methods using synthetic oxytocin (synOT) or prostaglandins (PG)—have drawn attention for their potential impact on maternal caregiving and mental health. Induction involves initiating contractions before spontaneous onset, typically via synOT or PG. SynOT and PG differ in their typical indications and timing of use during labor, which may imply distinct pathways to early postpartum outcomes. SynOT is commonly administered to stimulate or augment uterine contractions, whereas PG are often used for cervical ripening and the initiation of labor when the cervix is unfavorable. Biologically, exogenous oxytocin may interfere with endogenous patterns of oxytocin signaling and receptor sensitivity relevant to lactation and affiliative processes [17, 30, 31], while PG may influence lactation-related pathways through downstream hormonal and inflammatory processes [22]. In addition, pharmacological stimulation of synOT may increase pain and physiological stress via stronger or more frequent contractions [10], potentially shaping childbirth experiences and early postpartum interaction [10, 13].
Existing evidence regarding the association between labor induction and postpartum mother-to-infant bonding is mixed. PG-based induction has been linked to poorer bonding at hospital discharge [43], while other studies report only indirect effects via negative birth experiences [13] or even positive associations for synOT [14]. Discrepancies may reflect heterogeneity in analytic approaches, failure to distinguish medication types, and variation in mother-to-infant bonding measurement timing. Studies often pool induction agents despite their different pharmacological actions, and few examine longitudinal trajectories [22] showed that only PG predicted reduced breastfeeding when modeled with synOT, underscoring the need for drug-specific, multi-time-point analyses. Overall, evidence on how induction—and specific agents—relates to mother-to-infant bonding remains mixed, and longitudinal medication-specific studies that also evaluate early feeding as a mediator are scarce. Clarifying these patterns is clinically relevant for counselling and targeting early postpartum support as induction becomes increasingly common.
Breastfeeding
Breastfeeding may mediate the relationship between labor inductions and mother-to-infant bonding. Intrapartum interventions (e.g., labor induction, epidural analgesia, and mode of birth) have been associated with shorter breastfeeding duration [1], increased bottle-feeding [17], and exclusive formula use [18, 28]. Breastfeeding difficulties, in turn, predict poorer bonding [33] and reduced maternal sensitivity [19, 39], and have also been associated with higher postpartum depressive symptoms, particularly when breastfeeding cessation is driven by physical difficulty or pain [9].
Using data from the Japan Environment and Children’s Study (JECS), a large-scale prospective nationwide birth cohort, we investigated whether labor induction and specific medications (synOT, PG) are associated with mother-to-infant bonding development over 1, 6, and 12 months postpartum, and whether breastfeeding mediates these associations. As mother-to-infant bonding is known to improve over time during the postpartum period, we further examined whether any adverse effects of induction diminish across follow-up. Because our observational design cannot directly test biological or behavioral mechanisms, we focused on estimating associations. Specifically, we modelled synOT and PG concurrently to compare their associations with bonding outcomes while accounting for overlapping exposure. We hypothesized that labor induction would be associated with poorer early postpartum mother-to-infant bonding, and that less breastfeeding (i.e., greater formula feeding) at 1 month postpartum would be associated with poorer bonding. We did not expect the association to be fully explained by feeding alone, given the multifactorial nature of mother-to-infant bonding and the potential for pathways beyond feeding. Given limited prior evidence, analyses of medication-specific effects were exploratory.
Materials and methods
Study design and participants
Japanese pregnant women were recruited between January 2011 and March 2014 through the Japan Environment and Children’s Study (JECS), a large-scale prospective nationwide birth cohort study funded by the Ministry of the Environment. The eligibility criteria for participants are as follows [23]: participants must reside in the Study Areas at the time of recruitment and are expected to continue residing in Japan for the foreseeable future, their expected delivery date should be between August 1, 2011, and mid-2014; and they must be capable of participating in the study without difficulty. These criteria did not specify restrictions based on gestational month at recruitment. The JECS protocol was reviewed and approved by the Ministry of the Environment’s Institutional Review Board on Epidemiological Studies (Ethical Number: No.100910001) and the Ethics Committees of all participating institutions. All participants provided written informed consent.
This study utilized the jecs-ta-20190930 and jecs-qa-20210401 datasets. These datasets include questionnaires, medical records, and biological sample data collected at multiple time points from pregnancy through the postpartum period. For detailed study design, please refer to other publications [23, 27]. The jecs-ta-20190930 and jecs-qa-20210401 datasets included data for 103,057 pregnancies, including pregnancies with multiple fetuses and stillbirths. The analytical sample was restricted to liveborn, term, vaginal deliveries to reduce major confounding and to better isolate associations with induction agents. Exclusion criteria for this study were multiple participations from the same mother (second or subsequent pregnancies), pregnancies with multiple fetuses, stillbirths, preterm birth (gestational age less than 37 weeks), non-cephalic presentation, congenital anomalies, synOT administration for purposes other than induction, cesarean section, and missing responses to induction and any mother-to-infant bonding-related questions in the maternal questionnaires conducted at one month, six months, and one year postpartum. We conducted a complete-case analysis, participants with missing mother-to-infant bonding-related question data at any of the relevant time points were excluded. A total of 58,384 participants who did not meet any of the exclusion criteria were included in the analysis (Fig. 1).
Fig. 1.
Flowchart of participant selection
Measures
Labor induction
Information on mode of delivery, presence or absence of labor induction, and medications used was obtained following standard operating procedures. Physicians, midwives, nurses, and/or Research Co-ordinators transcribed relevant information from medical records [27]. In this study, induction included sodium prasterone sulfate, oxytocin, PGE₂, PGF₂α, gemeprost, and other agents. The use of multiple agents was permitted, and the distribution of their usage for labor induction among the study participants was as follows: sodium prasterone sulfate (n = 1477), oxytocin (n = 12,277), PG including PGE₂ and PGF₂α (n = 4671), gemeprost (n = 16), and other agents (n = 781). Oxytocin and PG were used in combination in 2,132 participants.
For primary analyses, mothers were classified as induction vs. no induction. For medication-specific analyses, synOT and PG use were coded as binary variables and modeled simultaneously, with other medications as covariates.
Mother-to-infant bonding
The Mother-to-Infant Bonding Scale (MIBS) is a self-reported questionnaire assessing mothers’ bonding feelings toward their infants [38, 42]. The Japanese version has demonstrated acceptable reliability and validity and has been used in perinatal research and clinical practice in Japan [21, 42]. At 1 and 6 months postpartum, JECS administered the 5-item MIBS,at 12 months, the 10-item version was administered. For longitudinal comparability, we analyzed the five items common to all time points. At 12 months, the 5-item score was highly correlated with the full 10-item score (r = 0.897, p < 0.001), supporting its use as an index of individual differences in bonding in this cohort. The items capture both positive feelings (e.g., “Feel loving towards my baby”,reverse-scored) and negative bonding-related feelings (e.g., “Wish I did not have my baby”). Scores range from 5 to 20, with higher scores indicating poorer bonding.
Covariates
Potential confounders, selected based on prior JECS analyses with similar perinatal exposures and outcomes [26], included maternal age, parity, marital status, education, annual income, physical/psychological abuse during pregnancy, positive feelings toward pregnancy, antenatal depressive symptoms (K6 ≥ 13), sex of child, neonatal transport, neonatal complications, and history of mental/developmental disorders. Neonatal transport was defined as transfer of the newborn to another medical facility after delivery. These covariates were also chosen because a literature review study on mother-to-infant bonding has linked bonding to maternal sociodemographic characteristics (e.g., age, parity, marital status, education, and income), maternal psychological factors during pregnancy (e.g., pregnancy-related feelings and depressive symptoms), and child/perinatal factors (e.g., child sex and neonatal adversity) [40], additionally, we included epidural anesthesia as an intrapartum pain-management factor that may correlate with labor induction and early postpartum outcomes.
Participant age was treated as a continuous variable. Parity was categorized as primiparous (first-time mothers) or multiparous (≥ 2 prior deliveries). Marital status, initially assessed with four options (married, unmarried, divorced, widowed), was recoded into a binary variable (married vs. not). Education history was classified into four categories: junior high school, high school, technical junior college, technical or vocational college, or associate’s degree, and bachelor’s or graduate degree. Annual income was assessed on a 9-point scale and was treated as a continuous variable. Physical or psychological abuse from a spouse during pregnancy was assessed with separate items, categorized into binary variables indicating whether participants had experienced any form of violence or not. Positive emotional feelings toward pregnancy were assessed with options: “very happy,” “unexpected but happy,” “unexpectedly confused,” “worried,” “no particular feelings,” and “other.” These were recorded into binary variables indicating whether participants selected the first two options or not. Depression symptoms during pregnancy were assessed using the K6 scale, which has been validated in Japanese populations [16]. Participants scoring ≥ 13 were classified as having significant depressive symptoms, and those scoring < 13 as not. History of mental or developmental disorders was assessed by inquiring about any previous diagnoses among depression, anxiety disorders, autonomic dysfunction, schizophrenia, attention-deficit/hyperactivity disorder, learning disabilities, autism spectrum, and other developmental disorders. Participants were categorized into those with any history of these disorders and those without. Neonatal transport and neonatal complications were analyzed as binary variables (yes/no).
Statistical analysis
All statistical analyses were performed using R (version 4.3.3). The significance level was set at α < 0.05 (two-tailed). Because the bonding scores were positively skewed, we applied a log transformation to reduce skewness and improve approximate normality. Log transformation is commonly recommended for positively skewed variables [5, 29]. To examine the associations between induction and mother-to-infant bonding development, linear mixed-effect models (LMMs) were employed. MIBS, measured at three time points (one month, six months, and twelve months postpartum), served as the dependent variable. The independent variables included time point, induction status, and their interaction. Covariates were simultaneously modeled as fixed effects. The LMMs incorporated a random intercept and a random slope for time within individual participants to account for the repeated measures, as well as a random intercept and a random slope for recruitment site to account for clustering within sites. When the interaction term was found to be statistically significant, simple slope analyses were conducted. These included examining group differences in mother-to-infant bonding at each time point and assessing changes in mother-to-infant bonding across time points within each group.
To evaluate medication-specific associations, we conducted a follow-up model replacing the binary induction indicator with indicators for synOT and PG, including their interactions with time. synOT and PG were entered simultaneously to account for overlap in exposure (i.e., some participants received both agents) and to estimate the association of each agent with bonding outcomes conditional on the other. Follow-up analyses replaced the binary induction variable with synOT and PG (and their interactions with time) as fixed effects. Indicators of other labor medications were included as covariates.
Additionally, mediation analyses tested whether feeding mode at 1 month postpartum (formula vs. breastfeeding/mixed feeding) mediated the association between induction or synOT use and mother-to-infant bonding, using bootstrap marginal mediation with 1,000 resamples [2].
Results
The demographic characteristics of the included participants are presented in Table 1. Maternal background information was collected through questionnaire surveys conducted during early and mid-to-late pregnancy. The final analytical sample comprised 58,384 participants, and because we conducted a complete-case analysis requiring non-missing MIBS data at 1, 6, and 12 months postpartum, the sample size was identical at each time point. Because of substantial skewness in raw scores of MIBS (1 month: 1.411; 6 months: 1.962; 12 months: 1.609), log transformation was applied. This reduced skewness (1 month: 0.689; 6 months: 1.193; 12 months: 0.914), improving score distribution for analysis.
Table 1.
Characteristics of variables of this study
| Characteristic | With Induction, n = 16,1231 | Without Induction, n = 42,2611 |
|---|---|---|
| Maternal age (years) | 31.13 (4.96) | 31.00 (4.87) |
| Parity | ||
| Multiparous | 6,807 (43%) | 25,669 (62%) |
| Primiparous | 8,862 (57%) | 15,609 (38%) |
| Marital status | ||
| Married | 15,233 (95%) | 40,205 (96%) |
| Not married | 747 (4.7%) | 1,684 (4.0%) |
| Mother’s highest level of education | ||
| Junior high school | 631 (3.9%) | 1,684 (4.0%) |
| High school | 5,037 (32%) | 12,632 (30%) |
| Technical junior college, Technical or vocational college, or Associate’s degree | 6,827 (43%) | 17,819 (43%) |
| Bachelor’s or graduate degree | 3,495 (22%) | 9,745 (23%) |
| Annual household income | ||
| < ¥2,000,000 | 791 (5.3%) | 1,973 (5.0%) |
| ¥2,000,000-¥4,000,000 | 5,061 (34%) | 13,398 (34%) |
| ¥4,000,000-¥6,000,000 | 4,918 (33%) | 13,302 (34%) |
| ¥6,000,000-¥8,000,000 | 2,451 (16%) | 6,301 (16%) |
| ¥8,000,000-¥10,000,000 | 1,077 (7.2%) | 2,634 (6.7%) |
| ¥10,000,000-¥12,000,000 | 360 (2.4%) | 965 (2.5%) |
| ¥12,000,000-¥15,000,000 | 152 (1.0%) | 356 (0.9%) |
| ¥15,000,000-¥20,000,000 | 83 (0.6%) | 194 (0.5%) |
| ≥ ¥20,000,000 | 53 (0.4%) | 115 (0.3%) |
| History of emotional or physical abuse from partner | 2,056 (13%) | 5,716 (14%) |
| History of psychiatric or developmental disorders | 1,324 (8.3%) | 3,061 (7.3%) |
| Positive emotional feelings toward pregnancy | 14,637 (93%) | 38,229 (92%) |
| High K6 Score during pregnancy (≥ 13) | 462 (2.9%) | 1,196 (2.9%) |
| Child’s sex | ||
| Female | 8,029 (50%) | 20,682 (49%) |
| Male | 8,094 (50%) | 21,579 (51%) |
| Neonatal transport | 530 (3.3%) | 891 (2.1%) |
| Neonatal complications | 1,034 (6.5%) | 2,000 (4.8%) |
| Epidural anesthesia | 802 (5.0%) | 409 (1.0%) |
| Infants feeding practicing | ||
| Breastfeeding | 6,131 (38%) | 19,837 (47%) |
| Formula | 226 (1.4%) | 427 (1.0%) |
| Mix | 9,731 (60%) | 21,892 (52%) |
| MIBS at one month postpartum | 6.58 (1.45) | 6.39 (1.35) |
| MIBS at six months postpartum | 5.99 (1.20) | 5.90 (1.15) |
| MIBS at twelve months postpartum | 6.18 (1.30) | 6.14 (1.27) |
MIBS mother-to-infant bonding scale
1Mean (SD); n (%)
Induction on mother-to-infant bonding
First, the association between induction and MIBS was examined using LMMs (Table 2). There was a significant main effect of time, indicating that MIBS scores decreased across the postpartum period (i.e., bonding improved over time). In contrast, the main effect of induction was not significant, suggesting that, averaged across all assessment points, overall MIBS scores did not differ between mothers who underwent induction and those who did not. The interaction between time and induction was significant, indicating that the trajectory of MIBS change over time differed by induction status (i.e., the magnitude of group differences varied across follow-up).
Table 2.
Associations of MIBS with induction and synOT/PG administration adjusting for covariates
| 95% CI | |||||
|---|---|---|---|---|---|
| Fixed effects | β | t | p | lower | higher |
| Model with induction | |||||
| Time | −0.103 | −16.207 | < 0.001 | −0.119 | −0.087 |
| Induction | 0.005 | 0.823 | 0.411 | −0.006 | 0.016 |
| Time x induction | −0.057 | −10.650 | < 0.001 | −0.067 | −0.046 |
| Model with synOT/PG | |||||
| Time | −0.104 | −16.177 | < 0.001 | −0.120 | −0.088 |
| SynOT | 0.016 | 2.509 | 0.012 | 0.003 | 0.028 |
| PG | −0.007 | −0.714 | 0.475 | −0.025 | 0.012 |
| Time x SynOT | −0.060 | −10.186 | < 0.001 | −0.072 | −0.049 |
| Time x PG | −0.019 | −2.094 | 0.036 | −0.036 | −0.001 |
MIBS mother-to-infant bonding scale, synOT synthetic oxytocin, PG prostaglandin
Subsequent simple slopes analyses were conducted by stratifying participants into induction and non-induction groups. In both groups, MIBS scores significantly decreased over time (Induction group: β = −0.19, 95% CI = [−0.214, −0.166], p < 0.001; Non-induction group: β = −0.104, 95% CI = [−0.118, −0.089], p < 0.001). Although both groups exhibited a significant reduction in the scores (indicating improved bonding) across time points, the induction group showed a steeper decline, indicating faster improvement across time points. Time-specific contrasts showed no statistically significant group differences at 1, 6, or 12 months postpartum (1 month: β = 0.013, 95% CI [−0.003, 0.030], p = 0.109; 6 months: β = 0.005, 95% CI [−0.008, 0.017], p = 0.482; 12 months: β = 0.000, 95% CI [−0.014, 0.013], p = 0.953). This pattern is consistent with the significant interaction, in that the between-group difference varies over time, even though the time-specific contrasts do not reach statistical significance.
Medication-specific analyses
SynOT, PG, and other medications were included in the model in place of the induction variable (Table 2). The main effect of synOT was significant, whereas the main effect of PG was not, indicating that, on average across assessment points, mothers who received synOT reported higher (worse) MIBS scores than those who did not receive synOT. Additionally, both the time × synOT and time × PG interactions were statistically significant, suggesting that the magnitude of the association varied across follow-up and/or that the rate of change in MIBS scores differed by exposure status.
Simple slopes analyses under the significant synOT-time interaction revealed that MIBS scores significantly decreased over time in both synOT-administered and non-administered groups (synOT group: β = −0.201, 95% CI = [−0.226, −0.177], p < 0.001; Non-synOT group: β = −0.106, 95% CI = [−0.120, −0.091], p < 0.001). The decline in MIBS scores was steeper in the oxytocin group compared to the non-oxytocin group. Similarly, simple slopes analyses under the significant PG × time interaction showed that MIBS scores declined significantly over time in both prostaglandin-administered and non-administered groups (PG group: β = −0.155, 95% CI = [−0.184, −0.125], p < 0.001; non-PG group: β = −0.116, 95% CI = [−0.131, −0.101], p < 0.001). Although the difference in slope was smaller than that observed in the synOT interaction, the decline was slightly steeper in the PG group.
Time-specific analyses demonstrated that the main effect of oxytocin was significant at one month postpartum (β = 0.027, 95% CI = [0.009, 0.045], p = 0.004), and remained significant at six months postpartum (β = 0.016, 95% CI = [0.002, 0.030], p = 0.024). However, at twelve months postpartum, the main effect of oxytocin was not significant (β = 0.010, 95% CI = [−0.005, 0.025], p = 0.205), indicating attenuation of the early negative association over time. In contrast, time-specific analyses for PG indicated that its effect was not significant at any postpartum timepoint (1 month: β = −0.007, 95% CI = [−0.034, 0.020], p = 0.626; 6 months: β = −0.004, 95% CI = [−0.025, 0.017], p = 0.712; 12 months: β = −0.010, 95% CI = [−0.033, 0.013], p = 0.393). Overall, these findings suggest that synOT use was linked to poorer bonding in the early postpartum period, with the difference diminishing later in the first year.
Mediation analyses
Feeding mode was assessed at one month postpartum with three options: exclusively breastfeeding, exclusively formula feeding, and mixed feeding. Before recoding feeding into a dummy variable, 653 women (1.1%) reported exclusive formula feeding, whereas 57,591 women (98.9%) reported any breastfeeding (breastfeeding or mixed feeding). A dummy variable was created where formula feeding was coded as 1 and breastfeeding or mixed feeding was coded as 0, serving as the mediator in the analysis (Fig. 2).
Fig. 2.
Mediation analysis of the relationship between A labor induction and B intrapartum synthetic oxytocin (synOT) administration and MIBS at 1 month postpartum via formula feeding. Path coefficients are standardized. **p <.001, *p <.05
In the model with induction as the independent variable, the path from induction to formula feeding was significant (β = 0.258, 95% CI = [0.075, 0.441], p = 0.006), indicating that mothers who underwent induction were more likely to report formula feeding. The path from formula feeding to MIBS was also significant (β = 0.18, 95% CI = [0.114, 0.250], p < 0.001), suggesting poorer bonding among mothers who reported formula feeding. Additionally, the indirect effect did not cross zero (indirect effect: 0.001, 95% CI [0.0002, 0.001]), indicating that early feeding practice accounted for part of the association between induction and bonding. The proportion of the total effect that was mediated through feeding mode was 15.84%, suggesting that approximately one-tenth of the association between induction and MIBS scores was explained by differences in feeding practices.
In the model using synOT administration as the independent variable, the association between synOT administration and the likelihood of formula feeding was not statistically significant (β = 0.159, 95% CI = [−0.045, 0.363], p = 0.127). Consistently, the indirect effect via formula feeding crossed zero (indirect effect: 0.0004, 95% CI [−0.00009, 0.001]), providing no clear evidence of mediation in this adjusted model. The estimated proportion of the total effect attributable to the indirect pathway through formula feeding was 4.89%, suggesting a small contribution, although it was not statistically supported. PG administration was not examined as a separate independent variable in the mediation analysis, as its main effect on MIBS at one month postpartum was not significant, in contrast to the significant effect observed for synOT.
Discussion and conclusion
This study revealed that mother-to-infant bonding improved over the first postpartum year in both mothers who underwent induction and those who did not. Although the induction group showed a steeper rate of improvement, time-specific differences in bonding between the induction and non-induction groups at 1, 6, and 12 months postpartum were not statistically significant. In contrast, synOT administration was independently associated with poorer bonding at 1 and 6 months postpartum, with attenuation by 12 months. Induction was associated with a higher likelihood of exclusive formula feeding at 1 month, and formula feeding partly mediated the association between induction and poorer early bonding. In the fully adjusted model, mediation via feeding was not statistically supported for synOT.
These findings extend prior work indicating that labor and delivery–related factors influence mother-to-infant bonding [40]. In our study, all deliveries were vaginal; therefore, our results specifically suggest that labor induction and induction agents within vaginal deliveries may be associated with less favorable early postpartum bonding, consistent with prior reports on induction-related exposures and postpartum bonding [13, 43]. By identifying breastfeeding as a mediator, this study provides novel epidemiological evidence connecting labor inductions with mother-to-infant bonding via feeding practices.
Potential mechanisms include receptor desensitization from elevated synOT exposure during labor [11, 31]. Mother-to-infant bonding may require both reduced threat-related negative affect and enhanced prosocial motivation and reward responses to infant cues [37]. Oxytocin has been proposed to facilitate these processes by attenuating amygdala-related threat responses while engaging reward-related circuitry (e.g., the ventral tegmental area and nucleus accumbens), thereby increasing the rewarding value of infant-related stimuli and supporting caregiving behavior [32, 37]. Oxytocin may also regulate stress physiology via the hypothalamic–pituitary–adrenal axis and downstream stress hormones (e.g., cortisol) [3] and may increase sensitivity to social cues such as facial expressions and infant crying [32]. In this context, labor induction could plausibly interact with oxytocin-related pathways and contribute to less favorable early postpartum bonding. Labor induction may also increase the likelihood of formula feeding, potentially via effects on milk ejection or prolactin regulation [22]. Because breastfeeding difficulties are linked to poorer bonding [33] and breastfeeding-related interactions may enhance maternal sensitivity [24], disrupted early feeding may constitute an indirect pathway to less favorable bonding outcomes. Breastfeeding difficulties and early cessation—particularly when driven by physical difficulty or pain—have also been associated with higher postpartum depressive symptoms [9]. In addition, reduced skin-to-skin contact, which can facilitate oxytocin release, may further contribute to these indirect pathways [35].
However, oxytocin’s role is not deterministic. In our study, the initial bonding gap narrowed over time, consistent with research showing that mother-to-infant bonding difficulties often remit during the first postpartum year [8]. This pattern may reflect the typical postpartum trajectory of bonding, whereby bonding-related symptoms decreases as mothers physically recover, gain caregiving experience, and infant routines stabilize. If labor induction primarily affects early postpartum processes (e.g., acute pain/stress around birth and early feeding establishment), its influence may be most salient in the first weeks and may attenuate as these proximal barriers resolve. Future studies with more frequent assessments across the early postpartum period would help to clarify the timing and mechanisms of any transient effects. Experience-based interactions can support bonding independently of hormonal pathways [15, 20], highlighting the dynamic and compensatory nature of maternal–infant relationships.
Postpartum care practices and family support structures vary across countries and may shape early feeding and mother–infant interactions. In Japan, a distinctive practice is satogaeri bunben (returning to the maternal family home around childbirth), which is intended to provide practical and emotional support from family members [41]. However, available evidence suggests that satogaeri bunben does not necessarily reduce the incidence of postnatal depression [36, 41], and postnatal depression rates in Japan have been reported to be broadly comparable to those in Western settings [36, 41] At the same time, core caregiving responses to infant distress may be partly culture-common. In a cross-national study of mothers across 11 countries including Japan, mothers showed similar priorities in responding to their own infants’ cries, and complementary neuroimaging work reported convergent activation in regions implicated in action preparation and vocal communication [4]. Together, these observations suggest that while perinatal support systems differ across contexts, some components of maternal responsiveness may be universal. Because the present study was conducted within the Japanese healthcare and family-support context, future cross-national studies would be valuable to clarify the extent to which the observed induction–feeding–bonding associations generalize to other care settings.
Clinical implications
For healthcare providers, these findings underscore the need for targeted breastfeeding and emotional support in mothers undergoing inductions, while also reassuring families that early challenges do not determine long-term relationships. As maternal age and labor intervention rates rise globally [34], integrated medical, psychological, and practical support is warranted.
Strengths and limitations
Strengths include the use of a large, prospective cohort, detailed medication-specific analyses, and the identification of breastfeeding as a mediator. Limitations include the absence of dosage data for labor-inducing drugs, the use of the 5-item MIBS without clinical cutoffs, lack of information on mothers’ feeding decision-making, and the observational design, which precludes causal inference. Unmeasured confounding may also have influenced results. Further experimental and longitudinal work is needed to clarify mechanisms. A further limitation is that feeding practices were assessed at 1 month postpartum rather than during the birth hospitalization. In-hospital feeding would represent a more proximal measure of early lactation potentially affected by induction agents; therefore, mediation estimates based on 1-month feeding may not fully capture immediate postpartum pathways. Another limitation is that feeding practice at 1 month postpartum was dichotomized as any breastfeeding versus exclusive formula feeding. This recoding does not distinguish exclusive breastfeeding from mixed feeding and may therefore obscure dose–response relationships or differential associations by feeding intensity. Future studies with more detailed feeding classifications (e.g., exclusive breastfeeding, mixed feeding, and no breastfeeding) and repeated assessments across the early postpartum period are needed.
Conclusion
Induction was associated with poorer early mother-to-infant bonding, and this association was partly mediated by breastfeeding difficulties; mediation for synOT via feeding was not supported. For women who undergo induction, the findings suggest a modestly elevated risk of early bonding difficulties, underscoring the value of proactive feeding support and attention to maternal wellbeing soon after birth. Importantly, bonding scores improved over time in both induction and non-induction groups, which supports a reassuring message that early postpartum difficulties often ease as mothers recover and gain caregiving experience. Clinically, this highlights the importance of nonjudgmental support that helps families avoid undue pressure or self-blame while providing timely practical and emotional assistance. Although the underlying mechanisms remain incompletely understood and induction itself is not always modifiable, early feeding support is a practical and potentially modifiable target. Such support may reduce early formula feeding, ease breastfeeding-related distress, and thereby help improve early bonding and maternal wellbeing in the weeks after birth. Unmeasured confounding may have influenced our results; therefore, further experimental and longitudinal studies are needed to clarify underlying mechanisms.
Acknowledgements
This study was funded by the Ministry of the Environment, Japan. The findings and conclusions of this article are solely the responsibility of the authors and do not represent the official views of the above government. We would like to express our sincere gratitude to all the participants who took part in this study. We also wish to thank the dedicated staff members involved in data collection for their valuable support and contributions. Members of the JECS Group as of 2025: Michihiro Kamijima (Principal Investigator, Nagoya City University, Nagoya, Japan), Shin Yamazaki (National Institute for Environmental Studies, Tsukuba, Japan), Maki Fukami (National Center for Child Health and Development, Tokyo, Japan), Reiko Kishi (Hokkaido University, Sapporo, Japan), Chiharu Ota (Tohoku University, Sendai, Japan), Koichi Hashimoto (Fukushima Medical University, Fukushima, Japan), Kenichi Sakurai (Chiba University, Chiba, Japan), Shuichi Ito (Yokohama City University, Yokohama, Japan), Ryoji Shinohara (University of Yamanashi, Chuo, Japan), Hidekuni Inadera (University of Toyama, Toyama, Japan), Takeo Nakayama (Kyoto University, Kyoto, Japan), Ryo Kawasaki (Osaka University, Suita, Japan), Yasuhiro Takeshima (Hyogo Medical University, Nishinomiya, Japan), Hideki Nagashima (Tottori University, Yonago, Japan), Narufumi Suganuma (Kochi University, Nankoku, Japan), Mayumi Tsuji (University of Occupational and Environmental Health, Japan), and Kimitoshi Nakamura (Kumamoto University, Kumamoto, Japan).
Authors’ contributions
The study conception and design were performed by Daiki Hiraoka. Material preparation, data collection and analysis were performed by Daiki Hiraoka, Asuka Tachi, Sayaka Kato, Yuki Ito, Yukihiro Ohya, Shinji Saitoh, Mayumi Sugiura-Ogasawara, and Michihiro Kamijima. The first draft of the manuscript was written by Daiki Hiraoka, Asuka Tachi, and Yuki Takahashi, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding
This study was funded by the Ministry of the Environment, Japan. The findings and conclusions of this article are solely the responsibility of the authors and do not represent the official views of the above government.
Data availability
Data are unsuitable for public deposition due to ethical restrictions and legal framework of Japan. It is prohibited by the Act on the Protection of Personal Information (Act No. 57 of 30 May 2003, amendment on 9 September 2015) to publicly deposit the data containing personal information. Ethical Guidelines for Medical and Health Research Involving Human Subjects enforced by the Japan Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Health, Labour and Welfare also restricts the open sharing of the epidemiologic data. All inquiries about access to data should be sent to: jecs-en@nies.go.jp. The person responsible for handling enquiries sent to this e-mail address is Dr Shoji F. Nakayama, JECS Programme Office, National Institute for Environmental Studies.
Declarations
Ethics approval and consent to participate
The protocol was reviewed and approved by the Ministry of the Environment’s Institutional Review Board on Epidemiological Studies (Ethical Number: No.100910001) and the Ethics Committees of all participating institutions. All participants provided written informed consent.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Daiki Hiraoka, Email: hiraoka@med.nagoya-cu.ac.jp, Email: hiraoka@mail.kobe-c.ac.jp.
the Japan Environment & Children’s Study (JECS) Group:
Shin Yamazaki, Maki Fukami, Reiko Kishi, Chiharu Ota, Koichi Hashimoto, Kenichi Sakurai, Shuichi Ito, Ryoji Shinohara, Hidekuni Inadera, Takeo Nakayama, Ryo Kawasaki, Yasuhiro Takeshima, Hideki Nagashima, Narufumi Suganuma, Mayumi Tsuji, and Kimitoshi Nakamura
References
- 1.Bai DL, Wu KM, Tarrant M. Association between intrapartum interventions and breastfeeding duration. J Midwifery Women’s Health. 2013;58(1):25–32. [DOI] [PubMed] [Google Scholar]
- 2.Barrett TS, Lockhart G, Cruz RA. Marginal mediation analysis: a practical statistical framework for interpretable mediation effects. 2019. https://osf.io/preprints/psyarxiv/fgm8t/.
- 3.Bell AF, Erickson EN, Carter CS. Beyond labor: the role of natural and synthetic oxytocin in the transition to motherhood. J Midwifery Women’s Health. 2014;59(1):35-42: quiz 108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bornstein MH, Putnick DL, Rigo P, Esposito G, Swain JE, Suwalsky JTD, et al. Neurobiology of culturally common maternal responses to infant cry. Proc Natl Acad Sci USA. 2017;114(45):E9465–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Box GEP, Cox DR. An analysis of transformations. J R Stat Soc B Stat Methodol. 1964;26(2):211–43. [Google Scholar]
- 6.Brockington I. Maternal rejection of the young child: present status of the clinical syndrome. Psychopathology. 2011;44(5):329–36. [DOI] [PubMed] [Google Scholar]
- 7.Brockington I, Butterworth R, Glangeaud-Freudenthal N. An international position paper on mother-infant (perinatal) mental health, with guidelines for clinical practice. Arch Womens Ment Health. 2017;20(1):113–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Brockington I, Oates J, George S, Turner D, Vostanis P, Sullivan M, et al. A screening questionnaire for mother-infant bonding disorders. Arch Womens Ment Health. 2001;3(4):133–40. [Google Scholar]
- 9.Brown A, Rance J, Bennett P. Understanding the relationship between breastfeeding and postnatal depression: the role of pain and physical difficulties. J Adv Nurs. 2016;72(2):273–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Buckley S, Uvnäs-Moberg K, Pajalic Z, Luegmair K, Ekström-Bergström A, Dencker A, et al. Maternal and newborn plasma oxytocin levels in response to maternal synthetic oxytocin administration during labour, birth and postpartum - a systematic review with implications for the function of the oxytocinergic system. BMC Pregnancy Childbirth. 2023;23(1):137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cadwell K. Intrapartum administration of synthetic oxytocin and downstream effects on breastfeeding: elucidating physiologic pathways. Ann Nurs Res Pract. 2017;2(3):1024. [Google Scholar]
- 12.Cetisli NE, Arkan G, Top ED. Maternal attachment and breastfeeding behaviors according to type of delivery in the immediate postpartum period. Rev Assoc Med Bras (1992). 2018;64(2):164–9. [DOI] [PubMed] [Google Scholar]
- 13.Döblin S, Seefeld L, Weise V, Kopp M, Knappe S, Asselmann E, et al. The impact of mode of delivery on parent-infant-bonding and the mediating role of birth experience: a comparison of mothers and fathers within the longitudinal cohort study DREAM. BMC Pregnancy Childbirth. 2023;23(1):285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Edwards H, Buisman-Pijlman FT, Esterman A, Phillips C, Orgeig S, Gordon A. Exogenous oxytocin administered to induce or augment labour is positively associated with quality of observed mother-infant bonding. Compr Psychoneuroendocrinol. 2024;20(100262):100262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fleming AS, Rosenblatt JS. Maternal behavior in the virgin and lactating rat. J Comp Physiol Psychol. 1974;86(5):957–72. [DOI] [PubMed] [Google Scholar]
- 16.Furukawa TA, Kawakami N, Saitoh M, Ono Y, Nakane Y, Nakamura Y, et al. The performance of the Japanese version of the K6 and K10 in the World Mental Health Survey Japan. Int J Methods Psychiatr Res. 2008;17(3):152–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.García-Fortea P, González-Mesa E, Blasco M, Cazorla O, Delgado-Ríos M, González-Valenzuela MJ. Oxytocin administered during labor and breast-feeding: a retrospective cohort study. J Matern Fetal Neonatal Med. 2014;27(15):1598–603. [DOI] [PubMed] [Google Scholar]
- 18.Gu V, Feeley N, Gold I, Hayton B, Robins S, Mackinnon A, et al. Intrapartum synthetic oxytocin and its effects on maternal well-being at 2 months postpartum. Birth. 2016;43(1):28–35. [DOI] [PubMed] [Google Scholar]
- 19.Hahn-Holbrook J, Little EE, Abbott M. Mothers are more sensitive to infant cues after breastfeeding compared to bottle-feeding with human milk. Horm Behav. 2021;136:105047. [DOI] [PubMed] [Google Scholar]
- 20.Insel TR, Harbaugh CR. Lesions of the hypothalamic paraventricular nucleus disrupt the initiation of maternal behavior. Physiol Behav. 1989;45(5):1033–41. [DOI] [PubMed] [Google Scholar]
- 21.Japan Association of Obstetricians and Gynecologists. Ninpu sanpu mental health care manual: sango care e no kireme no nai shien ni mukete [Perinatal mental health care manual: toward seamless support for postpartum care]. Chugai Igakusha; 2021.
- 22.Jordan S, Emery S, Watkins A, Evans JD, Storey M, Morgan G. Associations of drugs routinely given in labour with breastfeeding at 48 hours: analysis of the Cardiff Births Survey: PPH prophylaxis and breastfeeding. BJOG. 2009;116(12):1622–9; discussion 1630-2. [DOI] [PubMed] [Google Scholar]
- 23.Kawamoto T, Nitta H, Murata K, Toda E, Tsukamoto N, Hasegawa M, et al. Rationale and study design of the Japan environment and children’s study (JECS). BMC Public Health. 2014;14:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kim P, Feldman R, Mayes LC, Eicher V, Thompson N, Leckman JF, et al. Breastfeeding, brain activation to own infant cry, and maternal sensitivity. J Child Psychol Psychiatry Allied Discip. 2011;52(8):907–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kinsey CB, Hupcey JE. State of the science of maternal–infant bonding: a principle-based concept analysis. Midwifery. 2013;29(12):1314–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kunimi Y, Minami M, Muchanga SMJ, Eitoku M, Hayashi K, Fujieda M, et al. Exogenous oxytocin used to induce labor has no long-term adverse effect on maternal-infant bonding: findings from the Japan Environment and Children’s Study. J Affect Disord. 2022;299:37–44. [DOI] [PubMed] [Google Scholar]
- 27.Michikawa T, Nitta H, Nakayama SF, Yamazaki S, Isobe T, Tamura K, et al. Baseline profile of participants in the Japan environment and children’s study (JECS). J Epidemiol. 2018;28(2):99–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.OlzaFernández I, Marín Gabriel M, MalalanaMartínez A, Fernández-CañadasMorillo A, López Sánchez F, Costarelli V. Newborn feeding behaviour depressed by intrapartum oxytocin: a pilot study: newborn feeding behavior depressed by intrapartum oxytocin. Acta Paediatr. 2012;101(7):749–54. [DOI] [PubMed] [Google Scholar]
- 29.Osborne J. Notes on the use of data transformations. Pract Assess Res Eval. 2002;8(1):1–7. [Google Scholar]
- 30.Perkeybile AM, Kenkel WM, Yee JR, Waugh RE, Lillard TS, Ferris CF, et al. Oxytocin treatment at birth accelerates an epigenetic shift in the oxytocin receptor gene in the maternal brain. BMC Pregnancy Childbirth. 2025;25(1):777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Robinson C, Schumann R, Zhang P, Young RC. Oxytocin-induced desensitization of the oxytocin receptor. Am J Obstet Gynecol. 2003;188(2):497–502. [DOI] [PubMed] [Google Scholar]
- 32.Rocchetti M, Radua J, Paloyelis Y, Xenaki L-A, Frascarelli M, Caverzasi E, et al. Neurofunctional maps of the “maternal brain” and the effects of oxytocin: a multimodal voxel-based meta-analysis: oxytocin in mother-infant interaction. Psychiatry Clin Neurosci. 2014;68(10):733–51. [DOI] [PubMed] [Google Scholar]
- 33.Roth MC, Humphreys KL, King LS, Gotlib IH, Robakis TK. Breastfeeding difficulties predict mothers’ bonding with their infants from birth to age six months. Matern Child Health J. 2021;25(5):777–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sauer MV. Reproduction at an advanced maternal age and maternal health. Fertil Steril. 2015;103(5):1136–43. [DOI] [PubMed] [Google Scholar]
- 35.Shorey S, Asurlekar AR, Chua JS, Lim LHK. Influence of oxytocin on parenting behaviors and parent–child bonding: a systematic review. Dev Psychobiol. 2023. 10.1002/dev.22359. [DOI] [PubMed] [Google Scholar]
- 36.Suzuki S. Single center analysis of Japanese homecoming delivery and postpartum depression. JMA J. 2022;5(3):362–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Szymanska M, Schneider M, Chateau-Smith C, Nezelof S, Vulliez-Coady L. Psychophysiological effects of oxytocin on parent-child interactions: a literature review on oxytocin and parent-child interactions. Psychiatry Clin Neurosci. 2017;71(10):690–705. [DOI] [PubMed] [Google Scholar]
- 38.Taylor A, Atkins R, Kumar R, Adams D, Glover V. A new mother-to-infant bonding scale: links with early maternal mood. Arch Womens Ment Health. 2005;8(1):45–51. [DOI] [PubMed] [Google Scholar]
- 39.Tharner A, Luijk MPCM, Raat H, Ijzendoorn MH, Bakermans-Kranenburg MJ, Moll HA, et al. Breastfeeding and its relation to maternal sensitivity and infant attachment. J Dev Behav Pediatr. 2012;33(5):396–404. [DOI] [PubMed] [Google Scholar]
- 40.Tichelman E, Westerneng M, Witteveen AB, van Baar AL, van der Horst HE, de Jonge A, et al. Correlates of prenatal and postnatal mother-to-infant bonding quality: a systematic review. PLoS One. 2019;14(9):e0222998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Yoshida K, Yamashita H, Ueda M, Tashiro N. Postnatal depression in Japanese mothers and the reconsideration of “Satogaeri bunben.” Pediatr Int. 2001;43(2):189–93. [DOI] [PubMed] [Google Scholar]
- 42.Yoshida K, Yamashita H, Conroy S, Marks M, Kumar C. A Japanese version of mother-to-infant bonding scale: factor structure, longitudinal changes and links with maternal mood during the early postnatal period in Japanese mothers. Arch Womens Ment Health. 2012;15(5):343–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Zanardo V, Bertin M, Sansone L, Felice L. The adaptive psychological changes of elective induction of labor in breastfeeding women. Early Hum Dev. 2017;104:13–6. [DOI] [PubMed] [Google Scholar]
- 44.Zanardo V, Soldera G, Volpe F, Giliberti L, Parotto M, Giustardi A, et al. Influence of elective and emergency cesarean delivery on mother emotions and bonding. Early Hum Dev. 2016;99:17–20. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data are unsuitable for public deposition due to ethical restrictions and legal framework of Japan. It is prohibited by the Act on the Protection of Personal Information (Act No. 57 of 30 May 2003, amendment on 9 September 2015) to publicly deposit the data containing personal information. Ethical Guidelines for Medical and Health Research Involving Human Subjects enforced by the Japan Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Health, Labour and Welfare also restricts the open sharing of the epidemiologic data. All inquiries about access to data should be sent to: jecs-en@nies.go.jp. The person responsible for handling enquiries sent to this e-mail address is Dr Shoji F. Nakayama, JECS Programme Office, National Institute for Environmental Studies.