Abstract
Background
To develop a simple preoperative risk score identifying women requiring invasive hemostatic interventions during cesarean delivery.
Methods
This retrospective cohort included women undergoing cesarean delivery at a Hainan maternity hospital. The primary outcome was intraoperative use of intrauterine balloon tamponade and/or uterine artery ligation. Continuous variables were compared using Student’s t test or Mann–Whitney U test, and categorical variables using chi-square or Fisher’s exact test. Multivariable logistic regression identified independent risk factors, which were translated into an integer-based risk score. Model performance was assessed using discrimination and calibration metrics. Women were classified into three risk categories. For each risk group, mean scores were calculated and radar-chart personas was created.
Results
Among 8,687 cesarean deliveries, 386 (4.44%) required invasive hemostatic interventions. These women had greater blood loss, larger hemoglobin declines, higher transfusion rates, and more frequent peripartum hysterectomy. Independent predictors included placenta previa, low-lying placenta, severe preoperative anemia, prior postpartum hemorrhage, multiple gestation, preeclampsia, large leiomyoma, chorioamnionitis, and low platelet count. Based on these factors, we derived a preoperative risk score demonstrated good discrimination (AUC 0.7642, 95% CI: 0.7430–0.8305). The estimated probability of invasive hemostasis increased from about 2% in the low-risk group to (2.80 ~ 5.84)% in the medium-risk group and approximately 8% in the high-risk group. Although comprising only 8.28% of the cohort, the high-risk group accounted for 44% of all procedures.
Conclusions
A simple preoperative risk score using routinely available variables can stratify the likelihood of invasive uterus-sparing hemostatic interventions during cesarean delivery and may help optimize perioperative preparedness.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12884-026-09597-x.
Keywords: Postpartum hemorrhage, Cesarean delivery, Risk score, Invasive hemostatic interventions, Preoperative risk stratification
Introduction
Postpartum hemorrhage (PPH) remains a leading cause of severe maternal morbidity and mortality, with cesarean delivery being a major setting for life-threatening bleeding [1]. Intraoperative hemorrhage can progress rapidly. If not promptly controlled, it may lead to massive transfusion, intensive care unit admission, and peripartum hysterectomy [2]. Therefore, timely recognition and effective hemostatic strategies are critical [3].
In clinical practice, intraoperative hemorrhage management follows a stepwise approach [4]. First-line measures include uterine massage, uterotonic agents, and tranexamic acid [5]. If these fail, uterus-sparing invasive procedures such as intrauterine balloon tamponade, uterine compression sutures, and uterine artery ligation are employed, with hysterectomy as a final resort [6]. In addition, invasive measures are disproportionately used in patients with the greatest blood loss. Consequently, their application tracks hemorrhage severity [6, 7].
However, existing risk tools broadly predict PPH or transfusion [8], but they rarely address the need for invasive hemostatic procedures as a distinct outcome. Escalation decisions often rely on real-time clinical impression rather than structured preoperative risk estimation [7, 9]. Overly aggressive use in low-risk patients may cause overtreatment, whereas delayed escalation in high-risk patients may increase blood loss and reduce uterine preservation [10].
Therefore, we developed a preoperative point-based risk score for the need of invasive hemostatic interventions during cesarean delivery, specifically intrauterine balloon tamponade and/or uterine ascending artery ligation. This tool may enhance preparedness, facilitate timely escalation, and avoid unnecessary procedures in low-risk patients, contributing to more effective PPH management.
Methods
Study design and participants
This retrospective observational cohort study included women who underwent cesarean delivery at a tertiary women’s and children’s hospital in Sanya, Hainan province, between January 2020 and December 2023. Eligible women underwent cesarean delivery after 28 weeks of gestation for any obstetric indication. We excluded cases with preoperative stillbirth or incomplete clinical, ultrasound, and laboratory data. The study protocol was approved by the institutional ethics committee, which waived written informed consent due to the retrospective nature and use of anonymized data. The process of inclusion, exclusion, and grouping of the patients is shown in Fig. 1.
Fig. 1.
Flow chart for study design
The hospital functions as a regional referral center for high-risk obstetric care and receives referrals from secondary hospitals, local maternity units and community facilities across Sanya, Hainan province. Women with conditions associated with increased cesarean complexity and hemorrhage risk, such as placenta previa, suspected placenta accreta spectrum, hypertensive disorders of pregnancy, multiple gestation, or anticipated major obstetric hemorrhage, are commonly referred to this center. This referral structure is relevant when interpreting both the cesarean case mix and the frequency of invasive hemostatic procedures observed in the present study.
Data collection and variable definitions
Trained investigators extracted data from electronic medical records, including: (1) maternal characteristics: age, pre-pregnancy body mass index (BMI), obstetric history; (2) pregnancy characteristics: gestational age, parity, comorbidities; (3) preoperative and postoperative laboratory parameters: hemoglobin, hematocrit, platelet count; (4) ultrasound findings: amniotic fluid index, estimated fetal weight (EFW), placental location; (5) intraoperative and postoperative factors: estimated blood loss, uterotonic use, tranexamic acid, blood transfusion, invasive hemostatic interventions, hysterectomy, hospital stay, neonatal outcomes.
Preeclampsia was defined according to the American College of Obstetricians and Gynecologists (ACOG) criteria [11], and gestational diabetes mellitus (GDM) and diabetes mellitus (DM) were defined according to the American Diabetes Association (ADA) criteria [12]. A large leiomyoma was defined as being at least 10 cm in diameter when identified at anatomy scan [8]. When multiple leiomyomas were present, the diameter of the largest leiomyoma was used. Chorioamnionitis was defined as fever of at least 38.0 °C plus fetal tachycardia (more than 160 beat per min) or purulent fluid from cervical os [13]. Placenta covering the internal cervical os was defined as placenta previa, and placenta lower than 2 cm away from the internal cervical os was defined as low-lying placenta [14]. Women without placenta previa or low-lying placenta were considered to have normally inserted placenta. The commonly used definition of postpartum hemorrhage is blood loss of 500 mL or more within the first 24 h after birth [15].
The primary outcome was the intraoperative use of intrauterine balloon tamponade [9] and/or uterine artery ligation [6] during cesarean delivery. Invasive hemostatic interventions included: (1) intrauterine balloon tamponade; (2) uterine artery ligation; or (3) their combined use. Women receiving any of these procedures were assigned to the invasive hemostasis group. Those managed with only uterotonic drugs (with or without tranexamic acid), uterine massage, and routine intraoperative care, but no balloon tamponade or surgical ligation, were classified as the non-invasive hemostasis group. We restricted the primary composite outcome to these two interventions because they represented the most commonly used and most consistently documented uterus-sparing invasive hemostatic procedures at our institution during the study period. The B-Lynch suture and re-laparotomy was rarely employed, while internal iliac artery ligation and interventional radiologic techniques were not routinely performed. Consequently, these procedures were not included in the prespecified primary endpoint. Other procedures such as hysterectomy, if present, were recorded descriptively but were not used to redefine the primary outcome.
Institutional management of cesarean delivery
At our institution, routine cesarean deliveries are performed by attending obstetricians, while complex cases (placenta previa, suspected accreta, repeat cesarean with dense adhesions and multiple gestation) are managed by or under the direct supervision of chief or associate chief physicians. Elective and emergency deliveries follow the same hemorrhage prevention and escalation protocols, though emergencies involve faster decision-making. Staffing is adequate for the full range of surgical care during regular hours, with a reasonably reduced but still capable team on-site at night and on weekends. These organizational features may affect the threshold for escalating to invasive hemostatic procedures and are thus important for interpreting our findings.
Prophylactic uterotonics (oxytocin intramuscularly after cord clamping) are routinely given during cesarean delivery unless contraindicated. Additional uterotonics (carbetocin, carboprost, ergonovine) are used based on uterine tone, bleeding severity, and contraindications. The institutional PPH prevention protocol includes preoperative risk assessment, active blood loss monitoring, timely uterotonics, blood product availability for high-risk cases, and escalation to balloon tamponade or surgical devascularization if medical management fails. In addition, myomectomy at cesarean delivery was not routinely performed in our center unless required for uterine incision closure.
Statistical analysis
Baseline characteristics and clinical outcomes were summarized separately for the two groups. Continuous variables are presented as mean ± SD, and categorical variables as number (%). Between-group differences were assessed using Student’s t-test or Mann-Whitney U test for continuous variables, and using chi-square test or Fisher’s exact test for categorical variables as appropriate. In particular, we compared estimated blood loss, perioperative hemoglobin change, transfusion, and peripartum hysterectomy between groups to describe how the need for invasive hemostasis related to bleeding severity and adverse outcomes.
A multivariate logistic regression model was constructed with the primary outcome as the dependent variable. Preoperative variables (listed in Table 1) were entered into the model using stepwise selection. A risk score [16] was derived based on the final model. Continuous variables were categorized to balance clinical interpretability and model discrimination. The total risk score for each patient was calculated by summing the points for all applicable factors, and a lookup table was constructed to map total scores to predicted probabilities. The derivation process of the risk score function was described in Supplemental Material 1. Based on the score distribution, patients were classified into three risk categories: low-risk (0 ~ 3 points), medium-risk (4 ~ 6 points), and high-risk (≥ 7 points). Cut-off points were chosen based on the distribution of the score, the shape of the risk curve, and the need to maintain sufficient sample size within each category. The discriminative ability of the logistic model and the derived risk score was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC) with 95% CIs through 100 iterations of random data splitting with a 7:3 ratio (70% training and 30% testing).
Table 1.
Baseline characteristics of patients in non-invasive and invasive hemostasis groups
| Variables | Data Type | Overall (n = 8687) | Non-invasive hemostasis group (n = 8301) | Invasive hemostasis group (n = 386) | P-value |
|---|---|---|---|---|---|
| Age (years) | Mean (SD) | 30.58 (4.72) | 30.53 (4.70) | 31.83 (4.93) | < 0.001 |
| Height (cm) | Mean (SD) | 159.03 (5.26) | 158.98 (5.24) | 160.19 (5.43) | < 0.001 |
| Pre-pregnancy BMI (kg/m2) | |||||
| Underweight (< 18.5) | N (%) | 1158 (13.33) | 1118 (13.47) | 40 (10.36) | 0.093 |
| Normal (18.5–24.9) | N (%) | 5618 (64.67) | 5374 (64.74) | 244 (63.21) | 0.576 |
| Overweight (25.0-27.9) | N (%) | 1476 (16.99) | 1403 (16.90) | 73 (18.91) | 0.338 |
| Obesity (≥ 28.0) | N (%) | 430 (4.95) | 401 (4.83) | 29 (7.51) | 0.024 |
| Prior cesarean deliveries | N (%) | 3383 (38.94) | 3229 (38.90) | 154 (39.90) | 0.905 |
| Number of prior cesarean deliveries | Mean (SD) | 0.44 (0.59) | 0.44 (0.59) | 0.47 (0.62) | 0.475 |
| Number of full-term deliveries | Mean (SD) | 0.55 (0.67) | 0.54 (0.67) | 0.63 (0.73) | 0.024 |
| Number of preterm deliveries | Mean (SD) | 0.02 (0.24) | 0.03 (0.24) | 0.02 (0.18) | 0.146 |
| Number of prior miscarriages | Mean (SD) | 0.11 (0.41) | 0.11 (0.41) | 0.15 (0.45) | 0.061 |
| Number of ectopic pregnancies | Mean (SD) | 0.04 (0.22) | 0.04 (0.21) | 0.06 (0.31) | 0.201 |
| Prior PPH | N (%) | 110 (1.27) | 93 (1.12) | 17 (4.40) | < 0.001 |
| History of myomectomy | N (%) | 93 (1.07) | 83 (1.00) | 10 (2.59) | 0.007 |
| History of salpingectomy for ectopic pregnancy | N (%) | 206 (2.87) | 193 (2.82) | 13 (4.10) | 0.244 |
| IVF | N (%) | 460 (5.30) | 416 (5.01) | 44 (11.40) | < 0.001 |
| Gestational age at birth (days) | Mean (SD) | 271.84 (10.32) | 272.12 (9.93) | 265.89 (15.38) | < 0.001 |
| Gravida | Mean (SD) | 2.34 (1.30) | 2.32 (1.28) | 2.68 (1.51) | 0.144 |
| Parity | Mean (SD) | 1.55 (0.71) | 1.54 (0.71) | 1.67 (0.73) | 0.001 |
| Multiple gestation | N (%) | 221 (2.54) | 173 (2.08) | 48 (12.44) | < 0.001 |
| Weight gain during pregnancy (kg) | Mean (SD) | 13.29 (4.36) | 13.30 (4.36) | 13.06 (4.27) | 0.228 |
| GDM | N (%) | 1321 (15.21) | 1247 (15.02) | 74 (19.17) | 0.032 |
| DM in pregnancy | N (%) | 294 (3.38) | 278 (3.35) | 16 (4.15) | 0.483 |
| Preeclampsia | N (%) | 452 (5.20) | 413 (4.98) | 39 (10.10) | < 0.001 |
| Gestational hypertension | N (%) | 166 (1.91) | 154 (1.86) | 12 (3.11) | 0.117 |
| Chronic hypertension in pregnancy | N (%) | 18 (0.21) | 16 (0.19) | 2 (0.52) | 0.423 |
| PROM | N (%) | 1754 (20.19) | 1694 (20.41) | 60 (15.54) | 0.024 |
| Large leiomyoma | N (%) | 429 (4.94) | 387 (4.66) | 42 (10.88) | < 0.001 |
| Suspected abruption | N (%) | 47 (0.54) | 43 (0.52) | 4 (1.04) | 0.316 |
| Antepartum hemorrhage | N (%) | 116 (1.34) | 80 (0.96) | 36 (9.33) | < 0.001 |
| Chorioamnionitis | N (%) | 257 (2.96) | 242 (2.92) | 15 (3.89) | 0.344 |
| Preoperative hemoglobin (g/L) | Mean (SD) | 114.19 (14.70) | 114.51 (14.54) | 107.30 (16.34) | < 0.001 |
| Preoperative Hct (%) | Mean (SD) | 35.60 (4.13) | 35.70 (4.08) | 33.37 (4.51) | < 0.001 |
| Preoperative blood platelet (109/L) | Mean (SD) | 231.67 (58.30) | 232.32 (58.22) | 217.67 (58.48) | < 0.001 |
| EFW (g) | Mean (SD) | 3248.16 (550.23) | 3244.64 (529.43) | 3323.72 (884.15) | 0.932 |
| Amniotic fluid index (cm) | Mean (SD) | 9.98 (4.35) | 10.00 (4.34) | 9.42 (4.41) | 0.018 |
| Placenta previa | N (%) | 162 (1.86) | 89 (1.07) | 73 (18.91) | < 0.001 |
| Low-lying placenta | N (%) | 55 (0.63) | 40 (0.48) | 15 (3.89) | < 0.001 |
Abbreviations: PPH Postpartum hemorrhage, BMI Body mass index, EFW Estimated fetal weight, Hct Hematocrit, GDM Gestational diabetes mellitus, DM Diabetes mellitus, IVF In-vitro fertilization, PROM Premature rupture of membranes
Data expressed as mean (SD) or number (%). P-value: statistical significance at p < 0.05
To characterize patients across risk strata, we developed multidimensional personas using radar charts, with continuous variables normalized to the [0,1] range before grouping.
Missing data were rare. Cases with missing values in the outcome or key predictors were excluded from the primary analysis. A sensitivity analysis using multiple imputation by chained equations was performed. All analyses were conducted using R software (version 4.5.2). A two-sided P value < 0.05 was considered statistically significant.
Results
Baseline characteristics
A total of 8,687 women undergoing cesarean delivery were included, of whom 386 (4.44%) required at least one invasive hemostatic intervention intraoperatively. Among these, 169 (43.78%) underwent uterine artery ligation alone, 140 (36.27%) intrauterine balloon tamponade alone, and 77 (19.95%) both procedures. Baseline characteristics according to invasive hemostasis use are shown in Table 1. Women requiring invasive interventions were older, had higher obesity prevalence, and were more likely to have a history of cesarean delivery, myomectomy, or prior postpartum hemorrhage (PPH). They also exhibited higher rates of multiple pregnancy, large uterine fibroids, chorioamnionitis, and preeclampsia, as well as lower preoperative hemoglobin and platelet levels and higher frequency of placenta previa or low-lying placenta (all P < 0.05).
Intraoperative and postoperative outcomes
Intraoperative blood loss and clinical outcomes differed markedly between groups (Supplemental Table 1). The invasive intervention group had substantially greater estimated blood loss, larger hemoglobin decreases, and higher transfusion rates than the non-invasive group (all P < 0.001). Additional severe surgical outcomes showed a similar pattern. Peripartum hysterectomy was performed in 2 women (0.52%) in the invasive hemostasis group and in none in the non-invasive group (P < 0.001). These findings further support that invasive hemostatic interventions were predominantly used in women with more severe hemorrhage and a higher risk of adverse outcomes.
Risk factors of invasive hemostatic interventions
Multivariable analysis identified multiple factors significantly associated with invasive hemostasis (Supplemental Table 2). Increased risk was associated with older maternal age (OR = 1.0391 [95%CI: 1.0150–1.0637]), obesity (OR = 1.6889 [95%CI: 1.0840–2.5418]), prior PPH (OR = 3.7668 [95%CI: 2.0319–6.5711]), history of myomectomy (OR = 2.2538 [95%CI: 1.0429–4.4114]), multiple gestation (OR = 3.5862 [95%CI: 1.9340–6.5486]), preeclampsia (OR = 1.9194 [95%CI: 1.2820–2.8038]), large leiomyoma (OR = 2.1404 [95%CI: 1.4409–3.1085]), chorioamnionitis (OR = 2.3797 [95%CI: 1.3016–4.0464]), placenta previa (OR = 20.7094 [95%CI: 14.2778–29.9929]), and low-lying placenta (OR = 11.4533 [95%CI: 5.8929–21.2385]). Higher gestational age at birth (OR = 0.9834 [95%CI: 0.9719–0.9951]), preoperative hemoglobin level (OR = 0.9645 [95%CI: 0.9564–0.9726]), and platelet count (OR = 0.9957 [95%CI: 0.9937–0.9976]) were associated with lower risk. EFW showed a small but significant positive association with the invasive hemostatic intervention risk (OR = 1.0003 [95%CI: 1.0000-1.0005]).
Risk score for invasive hemostatic intervention
To translate these risk factors into a clinically usable tool, we derived a preoperative point-based risk score based on the regression coefficients (Table 2). High-risk contributors included placenta previa (8 points), preoperative hemoglobin < 70 g/L (7 points), and low-lying placenta (6 points). Intermediate contributors included prior PPH, multiple gestation, advanced maternal age, preeclampsia, large leiomyoma, chorioamnionitis, history of myomectomy, lower platelet count, and moderate anemia. Obesity, age 25 ~ 29 years, gestational age 238 ~ 258 days, and EFW 3500 ~ 3900 g contributed 1 point each.
Table 2.
Risk score table of invasive hemostatic interventions
| Risk Factor | Category | Wi, j | βi | D | Points |
|---|---|---|---|---|---|
| Age (years) | |||||
|
< 25 25-29 30-34 35-39 ≥ 40 |
15.00 27.00 32.00 37.00 52.00 |
0.0383 |
0.0000 0.4601 0.6518 0.8435 1.4185 |
0 1 2 2 4 |
|
| Obesity (BMI ≥ 28.0) | |||||
|
0 1 |
0.00 1.00 |
0.5241 |
0.0000 0.5241 |
0 1 |
|
| Prior PPH | |||||
|
0 1 |
0.00 1.00 |
1.3262 |
0.0000 1.3262 |
0 3 |
|
| History of myomectomy | |||||
|
0 1 |
0.00 1.00 |
0.8126 |
0.0000 0.8126 |
0 2 |
|
| Gestational age at birth (days) | |||||
|
< 238 238-258 ≥ 259 |
220.00 248.00 273.00 |
-0.0168 |
0.8895 0.4196 0.0000 |
2 1 0 |
|
| Multiple gestation | |||||
|
0 1 |
0.00 1.00 |
1.2771 |
0.0000 1.2771 |
0 3 |
|
| Preeclampsia | |||||
|
0 1 |
0.00 1.00 |
0.6520 |
0.0000 0.6520 |
0 2 |
|
| Large leiomyoma | |||||
|
0 1 |
0.00 1.00 |
0.7610 |
0.0000 0.7610 |
0 2 |
|
| Chorioamnionitis | |||||
|
0 1 |
0.00 1.00 |
0.8670 |
0.0000 0.8670 |
0 2 |
|
| Preoperative hemoglobin (g/L) | |||||
|
< 70 70-84 85-109 ≥ 110 |
50.00 77.00 97.00 120.00 |
-0.0361 |
2.5298 1.5540 0.8312 0.0000 |
7 4 2 0 |
|
| Preoperative blood platelet (109/L) | |||||
|
< 60 60-79 80-99 ≥ 100 |
48.00 69.50 79.50 300.00 |
-0.0043 |
1.0922 0.9990 0.9556 0.0000 |
3 3 2 0 |
|
| EFW (g) | |||||
|
< 3500 3500-3900 ≥ 4000 |
2700.00 3700.00 5000.00 |
0.0003 |
0.0000 0.2573 0.5919 |
0 1 2 |
|
| Placenta previa | |||||
|
0 1 |
0.00 1.00 |
3.0306 |
0.0000 3.0306 |
0 8 |
|
| Low-lying placenta | |||||
|
0 1 |
0.00 1.00 |
2.4383 |
0.0000 2.4383 |
0 6 |
BMI Body mass index, EFW Estimated fetal weight, PPH Postpartum hemorrhage
P-value: statistical significance at p < 0.05
Predicted risk increased with total score (Fig. 2), and the corresponding predicted probability for each total score is shown in Supplemental Table 3. Scores below 3 corresponded to low and stable risk, scores between 3 and 6 to a transitional range with gradual risk increase, and scores above 6 to a more pronounced risk increase. Using cutoffs of 3 and 6, women were classified into low-, medium-, and high-risk groups. Compared with the low-risk group (invasive hemostasis probability about 2%), the medium-risk group had a threefold higher risk (2.80%~5.84%), and the high-risk group had an approximately fourfold higher risk (approximately 8%). Despite the low overall prevalence of invasive procedures (4.44%), the score concentrated events effectively: the low-risk group comprised 57.58% of the population but accounted for only 21% of invasive cases; the medium-risk group comprised 34.13% of the population and accounted for 35% of cases; and the high-risk group comprised 8.28% of the population but accounted for 44% of cases. The AUC for the prediction of invasive hemostatic interventions was 0.7642 (95% CI: 0.7430–0.8305), demonstrating good discriminative performance (Supplemental Fig. 1).
Fig. 2.
Predicted probability of requiring invasive hemostatic interventions during cesarean delivery according to the preoperative risk score
Personas across risk groups
Radar-chart personas (Fig. 3) demonstrated a clear gradient across risk strata. The low-risk group was characterized by younger maternal age, higher preoperative hemoglobin, longer gestational age, and lower prevalence of complications. The medium-risk group represented a transitional phenotype with modestly increased risk features. The high-risk group showed a complex profile characterized by older age, lower hemoglobin, shorter gestation, higher EFW, and substantial burden of placental abnormalities, multiple gestation, and prior obstetric complications. Detailed characteristics across risk groups are provided in Supplemental Table 4.
Fig. 3.
Radar plot of personas across different risk groups. Abbreviations: EFW, Estimated fetal weight; PPH, Postpartum hemorrhage; Plt, Preoperative blood platelet; Hb, Preoperative hemoglobin
Sensitivity analysis using multiple imputation yielded similar results (Supplemental Material 2).
Discussion
A total of 8,687 women undergoing cesarean delivery were included in the analysis, corresponding to an overall cesarean delivery rate of 43.57%. This rate is comparable to previously reported rates from major tertiary centers in China, which ranged from 40.6% in Shanghai [17] and from 38.77% in Guangzhou [18]. The slightly higher rate in our cohort likely reflects our hospital’s role as a referral center for high-risk obstetric care in Sanya, Hainan Province. Of the 8,687 women included, 386 (4.44%) required required invasive uterus-sparing hemostatic interventions (intrauterine balloon tamponade and/or uterine artery ligation). Women receiving these procedures had substantially greater blood loss, hemoglobin declines, transfusion rates, and peripartum hysterectomy, confirming that invasive hemostasis was used in cases with more severe hemorrhage. Using routinely available preoperative variables, we developed a point-based risk score to predict the need for invasive hemostasis. Placenta previa, low-lying placenta, severe preoperative anemia, and other maternal and pregnancy factors emerged as key predictors. Stratified into three risk categories, the estimated probability of invasive hemostasis increased from approximately 2% in the low-risk group to 2.80 ~ 5.84% in the medium-risk group and about 8% in the high-risk group, with the high-risk group representing only 8.28% of the cohort yet accounting for 44% of all invasive procedures.
The predictors in our score align with established PPH risk factors, including abnormal placental location, multiple pregnancy, hypertensive disorders, intrauterine infection, macrosomia, prior PPH, and impaired hematologic reserves [3]. Notably, placenta previa, low-lying placenta, preoperative anemia, and thrombocytopenia received the highest point allocations, consistent with evidence linking these factors to severe hemorrhage and failure of conservative measures [7, 19, 20]. Importantly, the score captures cumulative risk: aggregation of factors such as advanced maternal age, multiple pregnancy, or prior PPH can elevate risk even in the absence of placental or hematologic abnormalities.
Previous risk stratification models for PPH have primarily targeted severe PPH [21], transfusion need [22], or composite severe maternal outcomes, and most were developed for mixed cohorts of vaginal and cesarean births [8]. While some studies have evaluated outcomes of interventions like intrauterine balloon tamponade or uterine ligation [23, 24], none have specifically predicted the preoperative probability of requiring invasive uterus-sparing procedures during cesarean section. To our knowledge, this is among the first studies to develop a practical, point-based risk score based solely on preoperative variables for this specific endpoint, offering actionable insight for surgical planning.
An important finding is that although the high-risk group accounted for a disproportionately large share of invasive hemostatic procedures, 56% of events still occurred among women classified as low or medium risk. This indicates that the score is useful for risk enrichment but does not identify all women who may ultimately require invasive hemostasis. In practice, the score may help identify a subgroup in whom heightened preparedness is especially warranted—for example, ensuring senior review, blood product readiness, and immediate access to hemostatic devices—but it should not be used to reduce vigilance in women with lower scores. This residual event burden outside the high-risk category likely reflects the multifactorial and dynamic nature of obstetric hemorrhage, including intraoperative findings, uterine response to medical treatment, and acute clinical deterioration that may not be fully predictable from preoperative variables alone. Universal preparedness for postpartum hemorrhage therefore remains essential during cesarean delivery.
Current PPH management protocols emphasize stepwise escalation to mechanical and surgical interventions when medical therapy fails [4]. Intrauterine balloon tamponade and uterine artery ligation serve as effective second-line uterus-sparing rescue measures that may avert hysterectomy, yet they are resource-intensive [6, 7]. In this context, our preoperative risk score should be interpreted as a tool to support anticipatory planning rather than to mandate invasive intervention. For women identified as higher risk, it may help guide practical preparation, such as scheduling delivery when senior staff are available, ensuring access to balloon tamponade devices and ligation instruments, preparing blood products, and alerting anesthesia, critical care, or interventional teams when appropriate. However, decisions regarding whether and when to perform invasive hemostatic procedures must remain individualized and based on real-time clinical assessment, intraoperative bleeding, response to uterotonics, and operator judgment. Conversely, in low-risk women, it supports avoidance of unnecessary invasive procedures and resource overuse. This aligns with evidence suggesting that earlier, risk-stratified treatment improves outcomes [10].
This study has several strengths and limitations. A key strength is its focus on a clearly defined, clinically relevant outcome of uterus-sparing invasive hemostatic procedures during cesarean delivery, using only preoperative variables routinely available in practice, presented as a simple point-based score. However, the single-center, retrospective design introduces potential bias and limits generalizability, as institutional practices regarding these procedures may vary. In addition, we were unable to fully quantify the impact of operator experience, team composition, and time-of-day or weekend staffing on escalation to invasive procedures, although these institutional factors may influence clinical decision-making. Additionally, the absence of systematically recorded placenta accreta spectrum data, due to its low prevalence and subjective intraoperative diagnosis, precluded its inclusion as a separate predictor, though severe cases were captured under placenta previa and the invasive hemostasis endpoint. The model was validated only internally, and external validation is needed. Future research should prioritize external validation across diverse settings and prospective evaluation of the score’s clinical impact.
Conclusions
Invasive uterus-sparing hemostatic interventions during cesarean delivery were infrequent but associated with more severe hemorrhage and worse short-term outcomes. Using routine preoperative variables, we developed a simple point-based risk score that identifies women at increased risk, demonstrating a clear gradient and concentrating nearly half of all invasive procedures within a small high-risk subgroup. The score is intended to support preparedness and resource allocation, not to mandate intervention or replace clinical judgment. External validation and prospective evaluation are needed before widespread adoption.
Supplementary Information
Supplementary Material 1: Supplemental Figure 1: The AUC for the prediction of invasive hemostatic interventions.
Acknowledgements
The authors would like to express thanks to all the participants.
Abbreviations
- PPH
Postpartum hemorrhage
- BMI
Body mass index
- EFW
Estimated fetal weight
- Hct
Hematocrit
- GDM
Gestational diabetes mellitus
- DM
Diabetes mellitus
- IVF
In-vitro fertilization
- PROM
Premature rupture of membranes
Authors’ contributions
Ying Chen: study design, data analysis, interpretation and manuscript writing; Zishu Zhan: data analysis, interpretation and manuscript writing; Sheng Wan: data analysis, interpretation and manuscript writing; Yan Wang: data acquisition, data interpretation and manuscript revision; Meiqi Chai: data collection and data interpretation; Shengyi Gu: data acquisition and interpretation; Cunjie Lin: study design, data analysis, critical manuscript revision; Xiaolin Hua: study design, data collection and manuscript writing; Jing Peng: study design, data interpretation and manuscript revision.
Funding
This study was supported by Joint Program on Health Science & Technology Innovation of Hainan Province (WSJK2024QN003), Shanghai Key Laboratory of Embryo Original Diseases (Shelab202502), National Natural Science Foundation of China (82401996) and Transformation Project of the Medical-Engineering Interdisciplinary Research Fund of Shanghai Jiao Tong University (YG2026LC05).
Data availability
The datasets utilized and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
This study was approved by the Ethics Committee of Hainan Branch, Shanghai Children’s Medical Center, School of Medicine, Shanghai Jiao Tong University (approval No. SYFYIRB2024005). Informed consent was waived by the same ethics committee due to the retrospective design and the use of anonymized clinical data.
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.
Ying Chen, Zishu Zhan and Sheng Wan contributed equally to this work.
Contributor Information
Cunjie Lin, Email: lincunjie@ruc.edu.cn.
Xiaolin Hua, Email: huaxiaolin_sh@163.com.
Jing Peng, Email: penxiaoduo0725@163.com.
References
- 1.Cresswell JA, Alexander M, Chong MYC, Link HM, Pejchinovska M, Gazeley U, Ahmed SMA, Chou D, Moller AB, Simpson D, Alkema L, Villanueva G, Sguassero Y, Tunçalp Ö, Long Q, Xiao S, Say L. Global and regional causes of maternal deaths 2009-20: a WHO systematic analysis. Lancet Glob Health. 2025;13(4):e626–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.WHO Guidelines Approved by the Guidelines Review Committee. WHO recommendations: Uterotonics for the prevention of postpartum haemorrhage. Geneva: World Health Organization © World Health Organization 2018. 2018. [PubMed]
- 3.Yunas I, Islam MA, Sindhu KN, Devall AJ, Podesek M, Alam SS, Kundu S, Mammoliti KM, Aswat A, Price MJ, Zamora J, Oladapo OT, Gallos I, Coomarasamy A. Causes of and risk factors for postpartum haemorrhage: a systematic review and meta-analysis. Lancet. 2025;405(10488):1468–80. [DOI] [PubMed] [Google Scholar]
- 4.Mammoliti KM, Althabe F, Easter C, Devall A, Martin J, Funmi AL, Yusuf R, Abubakar F, Arigbede LC, Mugambi J, Oyoo P, Sambusa M, Banda A, Samuels F, Willemse S, Khambule SD, Shu’aib HM, Wakili AA, Okore J, Mwampashi A, Singata-Madliki M, Arends E, Muller E, Galadanci H, Qureshi Z, Al-Beity FA, Hofmeyr GJ, Fawcus S, Moran N, Gwako G, Osoti A, Gallos I, Coomarasamy A. When Is Postpartum Haemorrhage Treatment Initiated? A Nested Observational Study Within the E-MOTIVE Trial. BJOG. 2025;132(11):1664–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gallos I, Devall A, Martin J, Middleton L, Beeson L, Galadanci H, Alwy Al-Beity F, Qureshi Z, Hofmeyr GJ, Moran N, Fawcus S, Sheikh L, Gwako G, Osoti A, Aswat A, Mammoliti KM, Sindhu KN, Podesek M, Horne I, Timms R, Yunas I, Okore J, Singata-Madliki M, Arends E, Wakili AA, Mwampashi A, Nausheen S, Muhammad S, Latthe P, Evans C, Akter S, Forbes G, Lissauer D, Meher S, Weeks A, Shennan A, Ammerdorffer A, Williams E, Roberts T, Widmer M, Oladapo OT, Lorencatto F, Bohren MA, Miller S, Althabe F, Gülmezoglu M, Smith JM, Hemming K, Coomarasamy A. Randomized Trial of Early Detection and Treatment of Postpartum Hemorrhage. N Engl J Med. 2023;389(1):11–21. [DOI] [PubMed] [Google Scholar]
- 6.Bouchghoul H, Madar H, Resch B, Pineles BL, Mattuizzi A, Froeliger A, Sentilhes L. Uterine-sparing surgical procedures to control postpartum hemorrhage. Am J Obstet Gynecol. 2024;230(3s):S1066-S75.e4. [DOI] [PubMed]
- 7.Gibier M, Pauly V, Orleans V, Fabre C, Boyer L, Blanc J. Risk Factors for Intrauterine Tamponade Failure in Postpartum Hemorrhage. Obstet Gynecol. 2022;140(3):439–46. [DOI] [PubMed] [Google Scholar]
- 8.Kawakita T, Mokhtari N, Huang JC, Landy HJ. Evaluation of Risk-Assessment Tools for Severe Postpartum Hemorrhage in Women Undergoing Cesarean Delivery. Obstet Gynecol. 2019;134(6):1308–16. [DOI] [PubMed] [Google Scholar]
- 9.Revert M, Rozenberg P, Cottenet J, Quantin C. Intrauterine Balloon Tamponade for Severe Postpartum Hemorrhage. Obstet Gynecol. 2018;131(1):143–9. [DOI] [PubMed] [Google Scholar]
- 10.Scott N, Gallos I, Walsh T, Williams CR, Allotey J, Borovac-Pinheiro A, Devall A, Festin MR, Goudar SS, Haslinger C, Hofmeyr GJ, Price MJ, Qureshi ZP, Sentilhes L, Yunas I, Coomarasamy A, Oladapo OT. Diagnostic criteria for postpartum haemorrhage treatment: a cost-effectiveness study. Lancet Glob Health. 2026;14(2):e213–23. [DOI] [PubMed] [Google Scholar]
- 11.Gestational Hypertension and Preeclampsia. ACOG Practice Bulletin, Number 222. Obstet Gynecol. 2020;135(6):e237–60. [DOI] [PubMed] [Google Scholar]
- 12.Diabetes ADAPPCf. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes-2026. Diabetes Care. 2026;49(Supplement1):S27–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Quiñones-Rivera JN, Bell J, Boelig RC, Azadi S, Anosike O, Gallo DM, Taylor A, Berghella V, Ledford LC, Rincón M, Villegas-Sanchez S, Burwick R, López-Torres L, Tolosa JE. Outcomes after cerclage and preterm prelabor rupture of membranes: data from the international collaborative for cerclage longitudinal evaluation and research (IC-CLEAR). Am J Obstet Gynecol MFM. 2025;7(10):101753. [DOI] [PubMed] [Google Scholar]
- 14.Dang X, Zhang L, Bao Y, Xu J, Du H, Wang S, Liu Y, Deng D, Chen S, Zeng W, Feng L, Liu H. Developing and Validating Nomogram to Predict Severe Postpartum Hemorrhage in Women With Placenta Previa Undergoing Cesarean Delivery: A Multicenter Retrospective Case-Control Study. Front Med (Lausanne). 2021;8:789529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Prevention and Management of Postpartum Haemorrhage. Green-top Guideline 52 Bjog. 2017;124(5):e106–49. [DOI] [PubMed] [Google Scholar]
- 16.Sullivan LM, Massaro JM, D’Agostino RB, Sr. Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Stat Med. 2004;23(10):1631–60. [DOI] [PubMed]
- 17.Liu X, Huang D, Landon MB, Cheng W, Chen Y. Trends in Cesarean Delivery Rate after Cessation of the One-Child Policy in China. Am J Perinatol. 2021;38(S 01):e84–91. [DOI] [PubMed] [Google Scholar]
- 18.Sha X, Hu H, Yang J, Fang D, Li W, Zhang H, Coonrod V, Liu H. Interventions to reduce the cesarean delivery rate in a tertiary hospital in China. J Matern Fetal Neonatal Med. 2022;35(1):30–8. [DOI] [PubMed] [Google Scholar]
- 19.Ageron FX, Ker K, Prowse D, Steyerberg EW, Balogun E, Shakur-Still H, Bello FA, Madar H, Deneux-Tharaux C, Chaudhri R, Olayemi O, Muganyizi P, Sentilhes L, Roberts I. Early assessment of maternal bleeding: development and validation of a prognostic model predicting death and near miss. Lancet Glob Health. 2026;14(3):e378–85. [DOI] [PubMed] [Google Scholar]
- 20.group AADoPADs. Risk factors, prenatal diagnosis, and outcome of posterior placenta accreta spectrum disorders in patients with placenta previa or low-lying placenta: A multicenter study. Acta Obstet Gynecol Scand. 2025;104(7):1328–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Liu CN, Yu FB, Xu YZ, Li JS, Guan ZH, Sun MN, Liu CA, He F, Chen DJ. Prevalence and risk factors of severe postpartum hemorrhage: a retrospective cohort study. BMC Pregnancy Childbirth. 2021;21(1):332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ouh YT, Lee KM, Ahn KH, Hong SC, Oh MJ, Kim HJ, Han SW, Cho GJ. Predicting peripartum blood transfusion: focusing on pre-pregnancy characteristics. BMC Pregnancy Childbirth. 2019;19(1):477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Suarez S, Conde-Agudelo A, Borovac-Pinheiro A, Suarez-Rebling D, Eckardt M, Theron G, Burke TF. Uterine balloon tamponade for the treatment of postpartum hemorrhage: a systematic review and meta-analysis. Am J Obstet Gynecol. 2020;222(4):293.e1-.e52. [DOI] [PubMed] [Google Scholar]
- 24.Li M, Zhang H, Song X, Li L, Meng F, Chu R. Risk factors for emergency cesarean delivery in patients with placenta accreta spectrum: a multicenter retrospective cohort study. Front Med (Lausanne). 2025;12:1539998. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1: Supplemental Figure 1: The AUC for the prediction of invasive hemostatic interventions.
Data Availability Statement
The datasets utilized and/or analyzed during the current study are available from the corresponding author upon reasonable request.