Peripartum Magnesium Sulfate and Post-Delivery Depression: A Nationwide Cohort Study With 13-Year Follow-Up
Article information
Abstract
Objective
Although its potential antidepressant effects have been suggested, evidence on long-term outcomes for post-delivery depression remains limited. We evaluated whether peripartum intravenous magnesium sulfate (MgSO₄) is associated with a reduced risk of postpartum depression (PPD) within 12 months and its long-term trajectory over 13 years.
Methods
This nationwide retrospective cohort study included all women in South Korea who delivered between 2010 and 2014, excluding those with prior depression. Exposure was defined as intravenous MgSO₄ administration during delivery hospitalization. Propensity score (PS) matching (1:5) was used to balance baseline characteristics. We employed Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for both 12-month PPD and long-term depression (up to 13 years).
Results
Among 1,583,253 eligible mothers, 9,756 (0.6%) received MgSO₄. In the PS-matched cohort (9,682 exposed; 35,140 unexposed), there was no significant association between MgSO₄ and the primary outcome of 12-month PPD (HR, 1.11; 95% CI, 0.97–1.26; p=0.114). Similarly, the risk of long-term post-delivery depression showed no significant difference (HR, 1.02; 95% CI, 0.97–1.07; p=0.388). These findings were consistent in the dose-response analysis for 12-month PPD (HR, 1.10 per 10 g increase; 95% CI, 0.96–1.24; p=0.144) and in subgroup analyses stratified by preeclampsia status.
Conclusion
Peripartum MgSO₄ exposure was not associated with a reduced risk of either 12-month PPD or long-term post-delivery depression. Despite its biological plausibility, routine obstetric use of MgSO₄ does not appear to confer preventive mental health benefits for mothers in the general obstetric population.
INTRODUCTION
Magnesium sulfate (MgSO₄) is a cornerstone of obstetric practice, utilized for seizure prophylaxis in women with preeclampsia and for fetal neuroprotection in cases of anticipated preterm birth [1,2]. Beyond its established obstetric indications, magnesium acts as a natural antagonist of the N-methyl-D-aspartate (NMDA) receptor and a regulator of calcium influx, mechanisms that play a critical role in the pathophysiology of mood disorders [3,4].
Preclinical and clinical studies have proposed that magnesium may exert antidepressant and anxiolytic effects by modulating NMDA receptor activity, calcium homeostasis, and stress-related neuroendocrine pathways [5-7]. Recent clinical trials have explored the mood-stabilizing properties of magnesium supplementation [8]. Similarly, observational studies, such as the work by Pourmirzaiee et al. [9], have investigated the association between maternal serum magnesium levels and peripartum mental health. However, the specific impact of acute peripartum intravenous MgSO₄ administration on subsequent depressive risk has not been established.
Postpartum depression (PPD) is traditionally defined as a major or minor depressive episode occurring within 12 months of delivery [10]. It affects an estimated 10%–20% of the general postpartum population and remains one of the most common non-psychotic mental disorders following childbirth [11]. However, there is growing clinical interest in the longitudinal mental health trajectory of mothers beyond the first year. Maternal depression remains a significant public health concern even as children reach school age; studies indicate that 10%–15% of mothers of young children (ages 5–10) continue to experience clinically significant depressive symptoms, often linked to the chronic psychosocial stressors of sustained caregiving [12]. In this context, evaluating the “post-delivery depression” trajectory allows for a more comprehensive understanding of maternal affective resilience.
This study is based on the hypothesis that the acute neuroprotective and neurochemical stabilization provided by peripartum MgSO₄—specifically through NMDA receptor modulation and regulation of the hypothalamic-pituitary-adrenal (HPA) axis—might influence the early onset of PPD and potentially alter the long-term affective landscape [5,13]. While we agree that the direct biological impact of an acute infusion is most plausibly observed within the traditional 12-month PPD window, the extensive follow-up allows us to explore whether such early pharmacological intervention correlates with long-term psychological resilience throughout the child-rearing period.
Our study utilizes a large-scale, nationwide cohort from South Korea to evaluate the association between inpatient MgSO₄ administration and the subsequent risk of depression. The primary aim is to assess the risk of PPD within the first 12 months, followed by an exploratory analysis of the long-term risk over a 13-year period. By doing so, we aim to provide a rigorous assessment of how routine peripartum pharmacological intervention may influence both immediate and extended maternal mental health outcomes.
METHODS
Study design and ethical approval
This retrospective cohort study utilized population-level administrative health data and adhered to the STROBE reporting guidelines [14]. The study protocol received approval from the Institutional Review Board of Seoul National University Bundang Hospital (X-2407-910-902) and the National Health Insurance Service (NHIS) (NHIS-2025-02-1-010), which also provided data access. Because the dataset comprised de-identified information originally collected for healthcare administration, informed consent was not required in accordance with Korean law and the ethical principles outlined in the Declaration of Helsinki (1975, revised 2008).
Data source
We accessed data from the NHIS, a government-operated, single-payer system that provides universal health coverage to all South Korean citizens and eligible residents. The NHIS database provides comprehensive, integrated longitudinal information from both inpatient and outpatient claims, including medical diagnoses, procedures, prescriptions, healthcare utilization, mortality, and demographic characteristics. Diagnostic information is recorded using the International Classification of Diseases, 10th Revision (ICD-10). Although the majority of healthcare providers in South Korea are privately employed, standardized reporting and reimbursement protocols enforced by the government contribute to uniform data quality across institutions [15].
Study population
We identified all women who delivered in South Korea between January 1, 2010, and December 31, 2014. The date of delivery was defined as the index date. For women with multiple deliveries during the study period, only the first delivery was included to ensure independence of observations. To focus on incident cases of post-delivery depression, we excluded women with a pre-existing history of depression (identified by ICD-10 codes F32, F33, and F34.1 or related antidepressant prescriptions) documented during the lookback period. This lookback period extended from one year prior to the last menstrual period (LMP) until the index date, providing a washout period of approximately 22 months to ensure that the identified cases were new-onset episodes.
The baseline period for assessing maternal characteristics and comorbidities was defined as the one-year period preceding the LMP. The follow-up period began on the index date and continued until the first diagnosis of post-delivery depression, death, or the study end date (December 31, 2023), whichever occurred first. To minimize the confounding effect of subsequent obstetric interventions, subjects were censored at the time of their next pregnancy or delivery during the follow-up period. Both singleton and multiple births were included, with the type of pregnancy adjusted as a covariate. Women with incomplete delivery-related information or demographic data were also excluded from the analysis.
Exposure: peripartum MgSO₄ exposure
Women who were administered intravenous MgSO₄ during their delivery hospitalization were classified as the Mg group. Those with no record of MgSO₄ administration during the same hospital stay were assigned to the non-Mg group. To evaluate the dose-response relationship, the total cumulative dose of MgSO₄ administered during the hospitalization was also retrieved and analyzed as a continuous variable (per 10 g increase).
Study outcome: 12-month PPD and long-term depression
The primary outcomes were evaluated at two distinct time points to align with clinical definitions and explore long-term trajectories: 1) PPD, defined as the first clinical diagnosis within 12 months of the index delivery, and 2) long-term post-delivery depression, defined as the first diagnosis over the entire 13-year follow-up period.
The diagnosis was identified using ICD-10 codes F32, F33, and F34.1. All psychiatric diagnoses were tracked from the index date through December 31, 2023. During the followup period, subjects were censored at the time of subsequent pregnancy, death, or at the end of the study period to ensure that the depression risk was specifically associated with the index delivery. In South Korea, a diagnosis of depression is eligible for healthcare reimbursement only when formally registered in the NHIS database by a licensed psychiatrist or physician, which ensures the clinical reliability of the diagnostic ascertainment.
Covariates
Baseline maternal and demographic variables were assessed during the 12-month period immediately preceding the index delivery date. Household income was divided into quartiles based on health insurance premium levels at the time of delivery, with an additional category for recipients of the medical aid program—a government-funded scheme offering full coverage to low-income individuals. Maternal employment status and residential area (urban or rural) were also assessed.
Maternal baseline health was assessed using the Charlson Comorbidity Index, assessed over the 12-month baseline period. We additionally adjusted for pregnancy-related conditions such as placental disorders, complications of labor and delivery, other relevant medical or surgical comorbidities, and preeclampsia. Specifically, preeclampsia was included as a primary baseline covariate to address potential confounding by indication, as it is a major clinical indication for MgSO₄. Psychiatric conditions diagnosed before delivery—including bipolar disorder, anxiety, insomnia, substance use, schizophrenia, and attention-deficit/hyperactivity disorder—were included as covariates due to their established association with post-delivery depression [16]. Diagnostic codes for these variables are provided in Supplementary Table 1.
Delivery-related variables included the type of pregnancy (singleton or multiple). Additionally, subsequent pregnancies during the follow-up period were treated as censoring events rather than covariates to ensure the methodological rigor of the primary outcome assessment.
Statistical analysis
Baseline characteristics were described as frequencies and proportions for categorical variables and as means with standard deviations (SDs) for continuous variables. To address baseline imbalances and minimize confounding by indication, propensity scores (PS) were estimated using a logistic regression model. This model incorporated maternal characteristics and pre-existing psychiatric comorbidities assessed during the 12-month pre-delivery period.
One-to-five nearest-neighbor matching without replacement was performed using a caliper width of 0.1.17 The final matching ratio for some subjects was less than 1:5 because of the strict caliper restriction, which was prioritized to ensure a high quality of balance between the groups. Covariate balance before and after matching was evaluated using absolute standardized mean differences (ASDs), with values below 0.1 indicating adequate balance.
To evaluate the association between MgSO₄ exposure and the risk of depression, we performed time-to-event analyses using Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). To ensure the clinical relevance of our findings and facilitate comparison with existing literature, we evaluated outcomes at two distinct time points: 1) 12-month PPD (depression within 12 months of delivery) and 2) long-term post-delivery depression (over the 13-year follow-up period). The follow-up period for each participant began on the index delivery date and continued until the earliest occurrence of a depression diagnosis, death, or the end of the study (December 31, 2023). To isolate the impact of the index delivery intervention, each woman was censored at the beginning of a subsequent pregnancy during the follow-up period. Incidence rates were calculated as the number of events per 1,000 person-years to account for varying follow-up durations. Both the mean and median follow-up durations were calculated to provide a stable measure of the observation period.
Multivariable Cox regression analyses were performed in the full cohort to verify the findings observed in the PS-matched cohort. These analyses, along with the dose-response models (evaluating cumulative MgSO₄ exposure as a continuous variable per 10 g increase) and subgroup analyses stratified by preeclampsia, were consistently applied to both the 12-month PPD and long-term outcomes. Proportional hazards assumptions were verified using Schoenfeld residuals, and multicollinearity was assessed using variance inflation factors (VIF<2.0). All statistical analyses were performed using R software (version 4.3.2; R Foundation for Statistical Computing), with two-sided p values <0.05 considered statistically significant.
RESULTS
Study population
From January 1, 2010, to December 31, 2014, we identified 1,992,128 live births associated with 1,630,065 mothers in the NHIS database. After excluding 39,639 cases (34,250 births and 5,389 mothers) with missing or incomplete delivery-related information, 1,957,878 births from 1,624,676 mothers were eligible for inclusion. Among these, only the first delivery for each mother was retained to ensure independence of observations. We further excluded 41,423 mothers with a documented diagnosis of depression before delivery. The final analytic sample consisted of 1,583,253 mothers.
Among the final cohort, 9,756 mothers (0.6%) were classified into the Mg group based on maternal administration of intravenous MgSO₄ during the delivery hospitalization. The remaining 1,573,497 mothers (99.4%) whose mothers had no record of inpatient MgSO₄ use were assigned to the non-Mg group.
PS matching was conducted using a 1:5 nearest-neighbor algorithm without replacement, resulting in a matched cohort of 44,822 mothers—9,682 in the Mg group and 35,140 in the non-Mg group. A flow diagram of the study selection process is presented in Figure 1, and baseline characteristics before and after matching are summarized in Table 1. After PS matching, covariate balance was achieved for all variables, with ASDs below 0.1, except for preeclampsia, which remained slightly imbalanced (ASD=0.170). Nonetheless, overall balance was acceptable. Supplementary Figure 1 shows the distribution of PS values before and after matching, demonstrating improved alignment between groups. The mean follow-up duration for post-delivery depression outcome assessment was 12.5 years (SD, 1.5 years).
Study flow diagram showing inclusion, exclusion, and PS matching. PS, propensity score; Mg, magnesium sulfate.
Comparison of baseline characteristics between the Mg and non-Mg groups before and after PS matching
Primary outcome: 12-month PPD
Table 2 summarizes the time-to-event analysis for the primary outcome. In the PS-matched cohort, the incidence of 12-month PPD was 31.0 per 1,000 person-years (300 of 9,682 mothers, 3.1%) in the Mg group and 27.0 per 1,000 person-years (949 of 35,140 mothers, 2.7%) in the non-Mg group. There was no statistically significant difference in the risk of 12-month PPD between the two groups (HR, 1.11; 95% CI, 0.97–1.26; p=0.114).
Association of postpartum depression and magnesium sulfate across PS-matching and multivariable analyses
In the multivariable-adjusted analysis of the entire cohort (Table 3), the Mg group was not significantly associated with 12-month PPD (Model 1: HR, 1.13; 95% CI, 0.97–1.31; p=0.117). Similarly, the dose-response analysis showed that a 10 g increase in cumulative MgSO₄ exposure did not influence the risk of 12-month PPD (Model 2: HR, 1.10; 95% CI, 0.96–1.24; p=0.144). Adjusted HRs with 95% CIs for all covariates in Model 1 are presented in Supplementary Table 2.
Multivariable Cox regression analyses for the risk of 12-month PPD and long-term depression in the entire cohort
Long-term post-delivery depression
Regarding long-term outcomes over the 13-year period, the incidence of depression in the matched cohort was 21.4 per 1,000 person-years (2,112 of 9,682 mothers, 21.8%) in the Mg group and 20.2 per 1,000 person-years (6,966 of 35,140 mothers, 19.8%) in the non-Mg group. Cox regression analysis showed no significant difference in the long-term risk between the groups (HR, 1.02; 95% CI, 0.97–1.07; p=0.388). In the entire cohort, neither the inpatient MgSO₄ administration (Model 3: HR, 1.02; 95% CI, 0.97–1.07; p=0.388) nor the cumulative dosage (Model 4: HR, 1.00; 95% CI, 0.97–1.02; p=0.885) was associated with the long-term trajectory of post-delivery depression.
Subgroup analyses
Subgroup analyses based on the presence of preeclampsia are presented in Table 4. For the primary 12-month PPD outcome, no significant associations were found in either the preeclampsia group (HR, 0.99; 95% CI, 0.98–1.01; p=0.462) or the non-preeclampsia group (HR, 1.01; 95% CI, 1.00–1.03; p=0.180). Consistent results were observed for long-term depression risk in both the preeclampsia (HR, 0.98; 95% CI, 0.91–1.06; p=0.683) and non-preeclampsia subgroups (HR, 1.03; 95% CI, 0.97–1.13; p=0.244).
Subgroup analyses by preeclampsia status for the risk of 12-month PPD and long-term depression
DISCUSSION
In this nationwide, population-based cohort study of over 1.5 million mothers in South Korea, we investigated whether peripartum exposure to intravenous MgSO₄ during delivery hospitalization was associated with the long-term risk of post-delivery depression. Contrary to our initial hypothesis, MgSO₄ exposure was not significantly associated with a reduced risk of post-delivery depression over a 13-year follow-up period. This null finding was consistent across multiple analytic strategies, including PS-matched analysis using a Cox proportional hazards model, multivariable regression in the full cohort, and dose-response modeling based on cumulative MgSO₄ dosage. Furthermore, focused analysis of the early postpartum period (within 12 months) and subgroup analyses stratified by preeclampsia status also yielded no significant preventive benefit.
Several observational studies have reported an inverse relationship between serum magnesium levels and depressive symptoms, including PPD. In both general and perinatal populations, lower serum magnesium concentrations have been linked to a greater risk of mood disturbances [7,18,19]. Mechanistically, magnesium is thought to exert antidepressant and anxiolytic effects via modulation of NMDA receptors, regulation of calcium homeostasis, and attenuation of HPA hyperactivity [13]. However, translating these biochemical observations into clinical interventions has proven challenging. Our findings align with the results of a 2017 randomized controlled trial by Fard et al. [8], which found that magnesium supplementation did not provide a definitive preventive benefit against PPD compared to a placebo.
Our investigation differs from these prior studies in several important respects: while most examined baseline serum magnesium levels or oral supplementation in relatively small samples with short follow-up durations [7,8,20], we assessed the long-term association of a potent intravenous intervention in a large, nationwide cohort. Our findings suggest that acute peripartum exposure to MgSO₄, even at therapeutic doses, may not provide lasting protection against post-delivery depression. This underscores the critical distinction between a patient’s baseline biochemical magnesium status, chronic nutritional supplementation, and brief medical interventions, which likely differ in their biological impact on the postpartum neuroendocrine and affective landscape [5,8,13].
One plausible explanation for the null findings is that the pharmacokinetics of intravenous MgSO₄ may not be conducive to the sustained central nervous system effects necessary for modulating postpartum mood. Although magnesium can influence NMDA receptor activity and HPA axis regulation, its acute administration during labor may result in only transient elevations in serum magnesium, with limited penetration across the blood–brain barrier in otherwise healthy obstetric patients without pre-existing deficiencies [21]. Unlike chronic oral supplementation, which may gradually normalize central magnesium levels over time, peripartum intravenous MgSO₄ is typically administered over a short duration and may not achieve the neurochemical stability required to provide lasting protection against post-delivery depression [22].
Regarding data integrity, we noted a slight decrease in the sample size for index deliveries occurring in 2014. This was not a result of data loss, but rather a deliberate and conservative filtering strategy to ensure a robust and consistent follow-up duration for all participants. By prioritizing diagnostic ascertainment and full claims integration through 2023, we ensured that the final inclusion year remained as clinically reliable as the preceding years, maintaining the integrity of the long-term follow-up.
This study has several limitations. First, despite the use of PS matching and extensive covariate adjustment, the possibility of residual confounding cannot be excluded, particularly from unmeasured factors such as social support, maternal stress, or psychiatric family history. Second, post-delivery depression was identified using administrative diagnostic codes, which may underestimate subclinical or untreated cases. However, this limitation would likely bias the results toward the null. Third, although we utilized a washout period starting from one year prior to the LMP to exclude pre-existing depression, this clinical window may not capture remote psychiatric histories occurring many years before the index pregnancy. Fourth, we did not have access to serum magnesium levels or information on baseline magnesium deficiency, which may modify the potential mood-related effects of MgSO₄. Lastly, while our findings are generalizable to a population-based cohort in South Korea, they may not be directly applicable to healthcare settings with different obstetric practices or patient demographics.
In conclusion, our findings from a large, population-based cohort study suggest that peripartum exposure to intravenous MgSO₄ is not associated with a reduced risk of either 12-month PPD or long-term post-delivery depression. Despite biologic plausibility and prior interest in magnesium as a mood-modulating agent, routine obstetric use of MgSO₄ may not confer sustained mental health benefits in otherwise healthy mothers. While our findings do not support the use of intravenous MgSO₄ as a preventive intervention for depression in the general obstetric population, future studies should explore whether personalized approaches—based on biochemical or psychosocial risk profiles—might identify individuals more likely to benefit from targeted magnesium supplementation.
Supplementary Materials
The Supplement is available with this article at https://doi.org/10.30773/pi.2025.0427.
The ICD-10 codes of maternal conditions
Adjusted HRs with 95% CIs for all covariates in Model 1
Adjusted HRs with 95% CIs for all covariates in Model 1
Notes
Availability of Data and Material
The data that support the findings of this study are available on request from the corresponding author, [TKO].
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Hey-ran Choi, Hye Yoon Park, Tak Kyu Oh. Data curation: In-Ae Song. Formal analysis: Tak Kyu Oh. Investigation: Hey-ran Choi. Methodology: Hey-ran Choi, Tak Kyu Oh. Project administration: Tak Kyu Oh, In-Ae Song. Resources: Hye Yoon Park. Software: Hye Yoon Park. Supervision: Tak Kyu Oh, In-Ae Song. Validation: Hey-ran Choi. Visualization: Tak Kyu Oh, Hye Yoon Park, In-Ae Song. Writing—original draft: Hey-ran Choi. Writing—review & editing: Tak Kyu Oh, Hye Yoon Park, In-Ae Song.
Funding Statement
None
Acknowledgments
None