Development and validation of a predictive model for excessive postpartum blood loss: A retrospective, cohort study

Interpretable machine learning predicts postpartum hemorrhage with severe maternal morbidity in a lower-risk laboring obstetric population

BACKGROUND

Early identification of patients at increased risk for postpartum hemorrhage (PPH) associated with severe maternal morbidity (SMM) is critical for preparation and preventative intervention. However, prediction is challenging in patients without obvious risk factors for postpartum hemorrhage with severe maternal morbidity. Current tools for hemorrhage risk assessment use lists of risk factors rather than predictive models.

OBJECTIVE

To develop, validate (internally and externally), and compare a machine learning model for predicting PPH associated with SMM against a standard hemorrhage risk assessment tool in a lower risk laboring obstetric population.

STUDY DESIGN

This retrospective cross-sectional study included clinical data from singleton, term births (>=37 weeks' gestation) at 19 US hospitals (2016-2021) using data from 58,023 births at 11 hospitals to train a generalized additive model (GAM) and 27,743 births at 8 held-out hospitals to externally validate the model. The outcome of interest was PPH with severe maternal morbidity (blood transfusion, hysterectomy, vascular embolization, intrauterine balloon tamponade, uterine artery ligation suture, uterine compression suture, or admission to intensive care). Cesarean birth without a trial of vaginal birth and patients with a history of cesarean were excluded. We compared the model performance to that of the California Maternal Quality Care Collaborative (CMQCC) Obstetric Hemorrhage Risk Factor Assessment Screen.

RESULTS

The GAM predicted PPH with an area under the receiver-operating characteristic curve (AUROC) of 0.67 (95% CI 0.64-0.68) on external validation, significantly outperforming the CMQCC risk screen AUROC of 0.52 (95% CI 0.50-0.53). Additionally, the GAM had better sensitivity of 36.9% (95% CI 33.01-41.02) than the CMQCC screen sensitivity of 20.30% (95% CI 17.40-22.52) at the CMQCC screen positive rate of 16.8%. The GAM identified in-vitro fertilization as a risk factor (adjusted OR 1.5; 95% CI 1.2-1.8) and nulliparous births as the highest PPH risk factor (adjusted OR 1.5; 95% CI 1.4-1.6).

CONCLUSION

Our model identified almost twice as many cases of PPH as the CMQCC rules-based approach for the same screen positive rate and identified in-vitro fertilization and first-time births as risk factors for PPH. Adopting predictive models over traditional screens can enhance PPH prediction.

Risk factors and prediction models for cardiovascular complications of hypertension in older adults with machine learning: A cross-sectional study

Background

Hypertension has emerged as a chronic disease prevalent worldwide that may cause severe cardiovascular complications, particularly in older patients. However, there is a paucity of studies that use risk factors and prediction models for cardiovascular complications associated with hypertension in older adults.

Objectives

To identify the risk factors and develop prediction models for cardiovascular complications among older patients with hypertension.

Methods

A convenience sample of 476 older patients with hypertension was recruited from a university-affiliated hospital in China. Demographic data, clinical physiological indicators, regulatory emotional self-efficacy, medication adherence, and lifestyle information were collected from participants. Binary logistic regression analysis was performed to screen for preliminary risk factors associated with cardiovascular complications. Two machine learning methods, Back-Propagation neural network, and random forest were applied to develop prediction models for cardiovascular complications among the study cohort. The sensitivity, specificity, accuracy, receiver operating characteristic curve, and area under the curve (AUC) values were used to assess the performance of the prediction models.

Results

Binary logistic regression identified nine risk factors for cardiovascular complications among older patients with hypertension. The machine learning models displayed excellent performance in predicting cardiovascular complications, with the random forest model (AUC 0.954) outperforming the Back-Propagation neural network model (AUC 0.811), as confirmed by model comparison analysis. The sensitivity, specificity and accuracy of the Back-Propagation neural network model compared to the random forest model were 74.2% vs. 86.5%, 75.2% vs. 94.3%, and 74.7% vs. 90.4%, respectively.

Conclusion

The machine learning methods employed in this study demonstrated feasibility in predicting cardiovascular complications among older patients with hypertension, with the random forest model based on nine risk factors exhibiting excellent prediction performance. These models could be used to identify high-risk populations and suggest early interventions aimed at preventing cardiovascular complications in such cohorts.

Hemostatic biomarkers associated with postpartum hemorrhage: a systematic review and meta-analysis

Postpartum hemorrhage (PPH) is a leading cause of maternal morbi-mortality. Although obstetric risk factors are well described, the impact of predelivery hematologic and hemostatic biomarkers remains incompletely understood. In this systematic review, we aimed to summarize the available literature on the association between predelivery hemostatic biomarkers and PPH/severe PPH. Searching MEDLINE, EMBASE, and CENTRAL databases from inception to October 2022, we included observational studies on unselected pregnant women without bleeding disorder reporting on PPH and on predelivery hemostatic biomarkers. Two review authors independently performed title, abstract and full-text screening, upon which quantitative syntheses of studies reporting on the same hemostatic biomarker were conducted, calculating the mean difference (MD) between women with PPH/severe PPH and controls. A search on 18 October 2022 yielded 81 articles fitting our inclusion criteria. The heterogeneity between studies was considerable. With regard to PPH, the estimated average MD in the investigated biomarkers (platelets, fibrinogen, hemoglobin, Ddimer, activated partial thromboplastin time, and prothrombin time) were not statistically significant. Women who developed severe PPH had lower predelivery platelets than controls (MD = -26.0 109/L; 95% confidence interval, -35.8 to -16.1), whereas differences in predelivery fibrinogen concentration (MD = -0.31 g/L; 95% confidence interval, -0.75 to 0.13) and levels of factor XIII or hemoglobin were not statistically significant in women with and without severe PPH. Predelivery platelet counts were, on average, lower in women with severe PPH compared with controls, suggesting the potential usefulness of this biomarker for predicting severe PPH. This trial was registered at the International Prospective Register of Systematic Reviews as CRD42022368075.

Family history of postpartum hemorrhage is a risk factor for postpartum hemorrhage after vaginal delivery: results from the French prospective multicenter Haemorrhages and Thromboembolic Venous Disease of the Postpartum cohort study

BACKGROUND

Postpartum hemorrhage is a major component of perinatal morbidity and mortality that affects young women worldwide and is still often unpredictable. Reducing the incidence of postpartum hemorrhage is a major health issue and identifying women at risk for postpartum hemorrhage is a key element in preventing this complication.

OBJECTIVE

This study aimed to estimate postpartum hemorrhage prevalence after vaginal delivery and to identify postpartum hemorrhage risk factors.

STUDY DESIGN

Unselected pregnant women ≥16 years of age admitted to 1 of 6 maternity wards in Brittany (France) for vaginal birth after 15 weeks of gestation were recruited in this prospective, multicenter cohort study between June 1, 2015, and January 31, 2019. Postpartum hemorrhage was defined as blood loss ≥500 mL in the 24 hours following delivery. Independent risk factors for postpartum hemorrhage were determined using logistic regression. Missing data were imputed using the Multivariate Imputation by Chained Equations method.

RESULTS

Among 16,382 included women, the postpartum hemorrhage prevalence was 5.37%. A first-degree family history of postpartum hemorrhage (adjusted odds ratio, 1.63; 95% confidence interval, 1.24-2.14) and a personal transfusion history (adjusted odds ratio, 1.90; 95% confidence interval, 1.23-2.92) were significantly associated with postpartum hemorrhage. The use of oxytocin during labor was also a risk factor for postpartum hemorrhage (adjusted odds ratio, 1.24; 95% confidence interval, 1.06-1.44). Inversely, smoking during pregnancy and intrauterine growth restriction were associated with a reduced risk for postpartum hemorrhage (adjusted odds ratio, 0.76; 95% confidence interval, 0.63-0.91, and 0.34; 95% confidence interval, 0.13-0.87, respectively).

CONCLUSION

In addition to classical risk factors, this study identified a family history of postpartum hemorrhage and personal transfusion history as new characteristics associated with postpartum hemorrhage after vaginal delivery. The association of postpartum hemorrhage with a family history of postpartum hemorrhage suggests a hereditary hemorrhagic phenotype and calls for genetic studies. Identifying women at risk for postpartum hemorrhage is a key element of being prepared for this complication.

Incidence of postpartum hemorrhage based on the improved combined method in evaluating blood loss: A retrospective cohort study

OBJECTIVE

In view of the current clinical inaccuracies and underestimations of postpartum hemorrhage amount, this study aims to investigate the incidence, etiology, clinical characteristics of postpartum hemorrhage in different modes of delivery based on the combination of volumetric method, gravimetric method and area method in evaluating blood loss.

DESIGN

This retrospective cohort study was conducted in Hangzhou Women's Hospital from January 2020 to June 2021, including 725 cases of postpartum hemorrhage among 18,977 parturients. Based on different modes of delivery, the participants were divided into three groups: vaginal delivery, forceps delivery, and cesarean section, for comparison.

METHODS

Using an improved combined assessment method for blood loss, we retrospectively analyzed a cohort of parturients with postpartum hemorrhage who underwent vaginal delivery, forceps delivery, or cesarean section and were hospitalized in Hangzhou Women's Hospital from January 2020 to June 2021.

RESULTS

(1) Among the 18,977 parturients, 725 cases of postpartum hemorrhage occurred, with an incidence rate of 3.8%, and severe postpartum hemorrhage accounted for 0.4% of the cases. (2) The incidence of postpartum hemorrhage was significantly higher in the forceps delivery group than in the vaginal delivery group (χ2 = 19.27, P<0.001), while the incidence of severe postpartum hemorrhage was significantly higher in the cesarean section group than in the vaginal delivery group (χ2 = 8.71, P = 0.003). (3) The causes of postpartum hemorrhage were statistically different among the different delivery modes, with varying underlying factors (P<0.001). (4) Patients with postpartum hemorrhage in different delivery modes showed statistically significant differences in age, body mass index (BMI), birth weight, gestational age, gravidity, parity, the decline of postpartum peripheral blood hemoglobin concentration, and estimated blood loss (P<0.05). (5) The proportion of blood transfusion was significantly higher in the cesarean section group than in the vaginal delivery and forceps delivery groups (χ2 = 231.03, P<0.001).

LIMITATIONS

This study is a single-center retrospective study, which may have led to selection bias in case selection. Additionally, the implementation of the combined three blood loss assessment methods may not have been strictly followed in all cases. Moreover, due to the mixing of bleeding with amniotic and irrigation fluids, the accuracy of evaluation may have been affected, leading to the possibility of inaccuracy of blood loss.

CONCLUSIONS

Forceps delivery and cesarean section increase the risk of postpartum hemorrhage, but forceps delivery does not significantly increase the incidence of severe postpartum hemorrhage. Uterine atony remains the leading cause of postpartum hemorrhage, while birth canal laceration and placental factors are the second most common causes of postpartum hemorrhage in forceps delivery and cesarean section, respectively. In this study, the volumetric method, gravimetric method and area method were combined to quantitatively assess postpartum hemorrhage amount. The combined method has strong clinical practicability and is less affected by subjective factors, although it also has limitations. In the future, we still need to focus on the early prediction and identification of postpartum hemorrhage, and further improve the quantitative assessment of postpartum blood loss.

Prepartum Anemia and Risk of Postpartum Hemorrhage: A Meta-Analysis and Brief Review

Postpartum hemorrhage (PPH) is responsible for 30% to 50% of maternal deaths. There is conflicting evidence if prepartum anemia facilitates PPH. A comprehensive analysis of studies describing their relation is missing. An extensive database search was conducted applying the terms "anemia" OR "hemoglobin" AND "postpartum hemorrhage." We used a random-effects meta-analysis model to estimate an overall odds ratio (OR) for PPH and prepartum anemia, separating studies that were conformant and non-conformant with the World Health Organization (WHO) definitions for anemia. The search yielded 2519 studies, and 46 were appropriate for analysis. The meta-analyses of WHO-conformant (n = 22) and non-conformant (n = 24) studies showed that the risk of PPH was increased when anemia was present. The ORs were 1.45 (CL: 1.23-1.71) for WHO-conformant studies, 2.88 (CL: 1.38-6.02) for studies applying lower thresholds for anemia, and 3.28 (CL: 2.08-5.19) for undefined anemia thresholds. PPH risk appeared to increase with lower anemia thresholds. Prepartum anemia is associated with an increased risk of PPH, an observation that is important regarding improved anemia correction strategies such as iron supplementation.

Prevention of postpartum hemorrhage in vaginal deliveries

Identification of patients at risk for postpartum hemorrhage (PPH) may allow for prompt diagnosis and intervention. Individual risk factors, risk assessment tools and prediction models have been used for determining a patient's risk of PPH. Measures for the prevention of PPH include identification and management of iron deficiency anemia, unit readiness and preparedness through performing regular simulations and having a PPH cart or medication kit readily available, prophylactic uterotonic - carbetocin alone or dual agents such as oxytocin and misoprostol or oxytocin and methylergometrine or antifibrinolytic (oxytocin and tranexamic acid) use in the third stage of labor immediately after fetal head delivery, and controlled cord traction.

Development and Validation of a Predictive Tool for Postpartum Hemorrhage after Vaginal Delivery: A Prospective Cohort Study

Postpartum hemorrhage (PPH) is one of the leading causes of maternal morbidity worldwide. This study aimed to develop and validate a predictive model for PPH after vaginal deliveries, based on routinely available clinical and biological data. The derivation monocentric cohort included pregnant women with vaginal delivery at Brest University Hospital (France) between April 2013 and May 2015. Immediate PPH was defined as a blood loss of ≥500 mL in the first 24 h after delivery and measured with a graduated collector bag. A logistic model, using a combination of multiple imputation and variable selection with bootstrap, was used to construct a predictive model and a score for PPH. An external validation was performed on a prospective cohort of women who delivered between 2015 and 2019 at Brest University Hospital. Among 2742 deliveries, PPH occurred in 141 (5.1%) women. Eight factors were independently associated with PPH: pre-eclampsia (aOR 6.25, 95% CI 2.35−16.65), antepartum bleeding (aOR 2.36, 95% CI 1.43−3.91), multiple pregnancy (aOR 3.24, 95% CI 1.52−6.92), labor duration ≥ 8 h (aOR 1.81, 95% CI 1.20−2.73), macrosomia (aOR 2.33, 95% CI 1.36−4.00), episiotomy (aOR 2.02, 95% CI 1.40−2.93), platelet count < 150 Giga/L (aOR 2.59, 95% CI 1.47−4.55) and aPTT ratio ≥ 1.1 (aOR 2.01, 95% CI 1.25−3.23). The derived predictive score, ranging from 0 to 10 (woman at risk if score ≥ 1), demonstrated a good discriminant power (AUROC 0.69; 95% CI 0.65−0.74) and calibration. The external validation cohort was composed of 3061 vaginal deliveries. The predictive score on this independent cohort showed an acceptable ability to discriminate (AUROC 0.66; 95% CI 0.62−0.70). We derived and validated a robust predictive model identifying women at risk for PPH using in-depth statistical methodology. This score has the potential to improve the care of pregnant women and to take preventive actions on them.

Predicting postpartum haemorrhage: A systematic review of prognostic models

BACKGROUND

Postpartum haemorrhage (PPH) remains a leading cause of maternal mortality and morbidity worldwide, and the rate is increasing. Using a reliable predictive model could identify those at risk, support management and treatment, and improve maternal outcomes.

AIMS

To systematically identify and appraise existing prognostic models for PPH and ascertain suitability for clinical use.

MATERIALS AND METHODS

MEDLINE, CINAHL, Embase, and the Cochrane Library were searched using combinations of terms and synonyms, including 'postpartum haemorrhage', 'prognostic model', and 'risk factors'. Observational or experimental studies describing a prognostic model for risk of PPH, published in English, were included. The Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies checklist informed data extraction and the Prediction Model Risk of Bias Assessment Tool guided analysis.

RESULTS

Sixteen studies met the inclusion criteria after screening 1612 records. All studies were hospital settings from eight different countries. Models were developed for women who experienced vaginal birth (n = 7), caesarean birth (n = 2), any type of birth (n = 2), hypertensive disorders (n = 1) and those with placental abnormalities (n = 4). All studies were at high risk of bias due to use of inappropriate analysis methods or omission of important statistical considerations or suboptimal validation.

CONCLUSIONS

No existing prognostic models for PPH are ready for clinical application. Future research is needed to externally validate existing models and potentially develop a new model that is reliable and applicable to clinical practice.

Prediction of Maternal Hemorrhage Using Machine Learning: Retrospective Cohort Study

Background

Postpartum hemorrhage remains one of the largest causes of maternal morbidity and mortality in the United States.

Objective

The aim of this paper is to use machine learning techniques to identify patients at risk for postpartum hemorrhage at obstetric delivery.

Methods

Women aged 18 to 55 years delivering at a major academic center from July 2013 to October 2018 were included for analysis (N=30,867). A total of 497 variables were collected from the electronic medical record including the following: demographic information; obstetric, medical, surgical, and family history; vital signs; laboratory results; labor medication exposures; and delivery outcomes. Postpartum hemorrhage was defined as a blood loss of ≥1000 mL at the time of delivery, regardless of delivery method, with 2179 (7.1%) positive cases observed. Supervised learning with regression-, tree-, and kernel-based machine learning methods was used to create classification models based upon training (21,606/30,867, 70%) and validation (4630/30,867, 15%) cohorts. Models were tuned using feature selection algorithms and domain knowledge. An independent test cohort (4631/30,867, 15%) determined final performance by assessing for accuracy, area under the receiver operating curve (AUROC), and sensitivity for proper classification of postpartum hemorrhage. Separate models were created using all collected data versus models limited to data available prior to the second stage of labor or at the time of decision to proceed with cesarean delivery. Additional models examined patients by mode of delivery.

Results

Gradient boosted decision trees achieved the best discrimination in the overall model. The model including all data mildly outperformed the second stage model (AUROC 0.979, 95% CI 0.971-0.986 vs AUROC 0.955, 95% CI 0.939-0.970). Optimal model accuracy was 98.1% with a sensitivity of 0.763 for positive prediction of postpartum hemorrhage. The second stage model achieved an accuracy of 98.0% with a sensitivity of 0.737. Other selected algorithms returned models that performed with decreased discrimination. Models stratified by mode of delivery achieved good to excellent discrimination but lacked the sensitivity necessary for clinical applicability.

Conclusions

Machine learning methods can be used to identify women at risk for postpartum hemorrhage who may benefit from individualized preventative measures. Models limited to data available prior to delivery perform nearly as well as those with more complete data sets, supporting their potential utility in the clinical setting. Further work is necessary to create successful models based upon mode of delivery and to validate the findings of this study. An unbiased approach to hemorrhage risk prediction may be superior to human risk assessment and represents an area for future research.

Machine learning-based prediction of postpartum hemorrhage after vaginal delivery: combining bleeding high risk factors and uterine contraction curve

PURPOSE

This work used a machine learning model to improve the accuracy of predicting postpartum hemorrhage in vaginal delivery.

METHODS

Among the 25,098 deliveries in the obstetrics department of the First Hospital of Jinan University recorded from 2016 to 2020, 10,520 were vaginal deliveries with complete study data. Further review selected 850 cases of postpartum hemorrhage (amount of bleeding > 500 mL) and 54 cases of severe postpartum hemorrhage (amount of bleeding > 1000 mL). Indicators of clinical risk factors for postpartum hemorrhage were retrieved from electronic medical records. Features of the uterine contraction curve were extracted 2 h prior to vaginal delivery and modeled using a 49-variable machine learning with 90% of study cases used in the training set and 10% of study cases used in the test set. Accuracy was compared among the assessment table, classical statistical models, and machine learning models used to predict postpartum hemorrhage to assess their clinical use. The assessment table contained 16 high-risk factor scores to predict postpartum hemorrhage. The classical statistical model used was Logistic Regression (LR). The machine learning models were Random Forest (RF), K Nearest Neighbor (KNN), and the one integrated with Lightgbm (LGB) and LR. The effect of model prediction was evaluated by area under the receiver operating characteristic curve (AUC), namely, C-static, calibration curve Brier score, decision curve, F-measure, sensitivity (SE), and specificity (SP).

RESULTS

1: Among the tested tools, the machine learning model LGB + LR has the best performance in predicting postpartum hemorrhage. Its Brier, AUC, and F-measure scores are better than those of other models in each group, and its SE and SP reach 0.694 and 0.800, respectively. The predictive performance of the classical statistical model LR is AUC: 0.729, 95%CI [0.702-0.756]). 2: Verification on the testing set reveals that the features of uterine contraction contribute to the improved accuracy of the model prediction. 3: LGB + LR model suggested that among the 49 indicators for predicting severe postpartum hemorrhage, the importance of the first 10 characteristics in descending order is as follows: hematocrit (%), shock index, frequency of contractions (min-1), white blood cell count, gestational hypertension, neonatal weight (kg), time of second labor (min), mean area of contractions (mmHg s), total amniotic fluid (mL), and body mass index (BMI). The prediction effect is close to that of the model after training with all 49 features. The predictive effect was close to that of the model after training using all 49 features. 4: Contraction frequency and intensity Mean_Area (representing effective contractions) have a high predictive value for severe postpartum hemorrhage. 5: Blood loss amount within 2 h has a high warning effect on postpartum hemorrhage, and the increase in AUC to 0.95 indicates that postpartum bleeding mostly occurs within 2 h after delivery.

CONCLUSION

Machine learning models incorporated with uterine contraction features can further improve the accuracy of postpartum hemorrhage prediction in vaginal delivery and provide a reference for clinicians to intervene early and reduce adverse pregnancy outcomes.

Place of birth and outcomes associated with large volume transfusion: an observational study

BACKGROUND

Guidelines recommend that women at high risk of postpartum haemorrhage deliver at facilities able to handle heavy bleeding. However postpartum haemorrhage is often unexpected. This study aims to compare outcomes and health service use related to transfusion of ≥4 units of red blood cells between women delivering in tertiary and lower level hospitals.

METHODS

The study population was women giving birth in public hospitals in New South Wales, Australia, between July 2006 and December 2010. Data were obtained from linked hospital, birth and blood bank databases. The exposure of interest was transfusion of four or more units of red cells during admission for delivery. Outcomes included maternal morbidity, length of stay, neonatal morbidity and need for other blood products or transfer to higher care. Multivariable regression models were developed to predict need of transfusion of ≥4 units of red cells using variables known early in pregnancy and those known by the birth admission.

RESULTS

Data were available for 231,603 births, of which 4309 involved a blood transfusion, with 1011 (0.4%) receiving 4 or more units. Women giving birth in lower level and/or smaller hospitals were more likely to receive ≥4 units of red cells. Women receiving ≥4 units in tertiary settings were more likely to receive other blood products and have longer hospital stays, but morbidity, readmission and hysterectomy rates were similar. Although 46% of women had no identifiable risk factors early in pregnancy, 20% of transfusions of ≥4 units occurred within this group. By the birth admission 70% of women had at least one risk factor for requiring ≥4 units of red cells.

CONCLUSIONS

Overall outcomes for women receiving ≥4 units of red cells were comparable between tertiary and non-tertiary facilities. This is important given the inability of known risk factors to predict many instances of postpartum haemorrhage.

Incidence and Risk Factors of Postpartum Hemorrhage in China: A Multicenter Retrospective Study

Background: Postpartum hemorrhage (PPH) is a leading cause of maternal morbidity and mortality worldwide but the incidence and its risk factors in China is limited. The objective of this study is to investigate the incidence and the risk factors of PPH in Chinese women.

Methods: A multi-center retrospective study of pregnant women at ≥28 weeks of gestation was conducted. Logistic regression was used to identify potential risk factors of PPH and receiver operating characteristic curve was used to evaluate the predictive performance of the identified risk factors. Subgroup analysis focusing on the number of fetus and the mode of delivery was conducted.

Results: A total of 99,253 pregnant women were enrolled and 804 (0.81%) experienced PPH. The subgroup analysis revealed that the incidence of PPH was 0.75, 2.65, 1.40, and 0.31% in singletons, twin pregnancies, cesarean sections, and vaginal deliveries, respectively. Placenta previa and placenta accreta were the predominant risk factors of PPH in the overall population and all subgroups. A twin pregnancy was a risk factor for PPH regardless of the mode of delivery. Obesity, and multiparity were risk factors for PPH in both singletons and cesarean section cases, but the latter predicted a reduced probability of PPH in vaginal deliveries. Macrosomia was associated with increased risk of PPH in singletons or vaginal deliveries. In women who delivered vaginally, preeclampsia was associated with a higher risk of PPH. The areas under the curve for the overall cohort, singletons, twin pregnancies, cesarean section cases, and vaginal deliveries were 0.832 (95% confidence interval [CI] 0.813–0.851), 0.824 (95% CI 0.803–0.845), 0.686 (95% CI 0.617–0.755), 0.854 (95% CI 0.834–0.874), and 0.690 (95% CI 0.646–0.735), respectively.

Conclusions: The risk factors of PPH varied slightly based on the number of fetuses and the mode of delivery, while placenta previa and placenta accreta were the two major risk factors. A combination of the identified risk factors yielded a satisfactory predictive performance in determining PPH in the overall cohort, singletons pregnancies, and women who delivered by cesarean section, whereas the performance was moderate in twin pregnancies and in women delivering vaginally.

Development and validation of a prediction model for postpartum hemorrhage at a single safety net tertiary care center

BACKGROUND

Postpartum hemorrhage is a leading cause of pregnancy-related morbidity and mortality; however, there is limited ability to identify women at risk of this obstetrical complication.

OBJECTIVE

This study aimed to develop and validate a prediction model for postpartum hemorrhage based on antenatal and intrapartum risk factors.

STUDY DESIGN

This was a retrospective cohort study of women who delivered between April 2016 and March 2019 at a single safety net hospital. The prevalence of postpartum hemorrhage, defined as blood loss of ≥1000 mL at the time of delivery, was determined, and characteristics were compared between women with and without postpartum hemorrhage. Women were randomly assigned to a prediction or a validation cohort. The selection of predictors to be included in the model was based on known antenatal and intrapartum risk factors for postpartum hemorrhage. A multivariable logistic regression with a backward stepwise approach was used to create a prediction model. Area under the receiver operating characteristic curve and 95% bootstrap confidence intervals were calculated. Using the final model, a single threshold for classifying postpartum hemorrhage was chosen, and the resulting sensitivity, specificity, and false-negative and false-positive rates were explored.

RESULTS

The prevalence rates of postpartum hemorrhage in the prediction and validation cohorts were 6.3% (377 of 6000 cases) and 6.4% (241 of 3774 cases), respectively (P=.83). The following predictors were selected for the final model: maternal body mass index (kg/m²), number of fetuses, history of postpartum hemorrhage, admission platelets of <100,000/µL, chorioamnionitis, arrest of descent, placental abruption, and active labor duration. The predictive model had an area under the receiver operating characteristic curve of 0.82 (95% confidence interval, 0.81-0.84). When applied to the validation cohort, the model had an area under the receiver operating characteristic curve of 0.81 (95% confidence interval, 0.78-0.83), a sensitivity of 86.9%, a specificity of 74.2%, a positive predictive value of 18.6%, a negative predictive value of 98.8%, a false-negative rate of 13.1%, and a false-positive rate of 25.9%.

CONCLUSION

The model performed reasonably well in identifying women at risk of postpartum hemorrhage. Further studies are necessary to evaluate the model in clinical practice and its effect on decreasing the prevalence of postpartum hemorrhage and associated maternal morbidity.

Haematological parameters associated with postpartum haemorrhage after vaginal delivery: Results from a French cohort study

BACKGROUND

Immediate postpartum haemorrhage (PPH) is a major, feared and often unpredictable issue. Besides many clinical risk factors, some biological parameters could also be predictive of PPH.

OBJECTIVE

To study simple and easily accessible haematological parameters as potential risk factors for PPH after vaginal delivery.

METHODS

All women who had a vaginal delivery between April 1, 2013 and May 29, 2015 in the maternity ward of Brest University Hospital (France) were included, after oral informed consent obtained. Clinical data were collected by obstetricians or midwives during antenatal care visits, labour and delivery, and recorded by trained research assistants. Haematological variables, including immature platelet fraction, were measured from a blood sample systematically collected at the entrance in the delivery room. PPH, measured with a graduated collector bag, was defined as blood loss of at least 500 ml.

RESULTS

2742 women were included. PPH occurred in 141 (5%) women. Seven clinical factors were independently associated with PPH: pre-eclampsia (OR 5.85, 95%CI 2.02, 16.90), multiple pregnancy (OR 3.28, 95%CI 1.21, 8.91), assisted reproduction (OR 2.75, 95%CI 1.45, 5.20), antepartum bleeding (OR 2.15, 95%CI 1.24,3.73), post-term delivery (OR 1.93, 95%CI 1.17, 3.17), obesity (OR 2.95, 95%CI 1.76, 4.93) and episiotomy (OR 2.51, 95%CI 1.63, 3.74). Three haematological factors were additionally identified as independent risk factors for PPH: platelets < 150 Giga/L (OR 2.98, 95%CI 1.63, 5.46), fibrinogen < 4.5 g/l (OR 1.86, 95%CI 1.21, 2.87) and APTT ratio ≥ 1.1 (OR 2.16, 95%CI 1.31, 3.57). Immature platelet fraction was not associated with PPH.

CONCLUSION

Besides classical clinical risk factors, this study identifies simple haematological parameters as risk factors for PPH.

Prediction of postpartum hemorrhage in pregnant women with immune thrombocytopenia: Development and validation of the MONITOR model in a nationwide multicenter study

Globally, postpartum hemorrhage (PPH) is the leading cause of maternal death. Women with immune thrombocytopenia (ITP) are at increased risk of developing PPH. Early identification of PPH helps to prevent adverse outcomes, but is underused because clinicians do not have a tool to predict PPH for women with ITP. We therefore conducted a nationwide multicenter retrospective study to develop and validate a prediction model of PPH in patients with ITP. We included 432 pregnant women (677 pregnancies) with primary ITP from 18 academic tertiary centers in China from January 2008 to August 2018. A total of 157 (23.2%) pregnancies experienced PPH. The derivation cohort included 450 pregnancies. For the validation cohort, we included 117 pregnancies in the temporal validation cohort and 110 pregnancies in the geographical validation cohort. We assessed 25 clinical parameters as candidate predictors and used multivariable logistic regression to develop our prediction model. The final model included seven variables and was named MONITOR (maternal complication, WHO bleeding score, antepartum platelet transfusion, placental abnormalities, platelet count, previous uterine surgery, and primiparity). We established an easy-to-use risk heatmap and risk score of PPH based on the seven risk factors. We externally validated this model using both a temporal validation cohort and a geographical validation cohort. The MONITOR model had an AUC of 0.868 (95% CI 0.828-0.909) in internal validation, 0.869 (95% CI 0.802-0.937) in the temporal validation, and 0.811 (95% CI 0.713-0.908) in the geographical validation. Calibration plots demonstrated good agreement between MONITOR-predicted probability and actual observation in both internal validation and external validation. Therefore, we developed and validated a very accurate prediction model for PPH. We hope that the model will contribute to more precise clinical care, decreased adverse outcomes, and better health care resource allocation.

Development and validation of a predictive model of exclusive breastfeeding at hospital discharge: Retrospective cohort study

BACKGROUND

The benefits of breastfeeding for both mother and newborn have been widely demonstrated. However, breastfeeding rates at discharge are lower than recommended, so being able to identify women at risk of not breastfeeding at discharge could allow professionals to prioritise care.

OBJECTIVE

To develop and validate a predictive model of exclusive breastfeeding at hospital discharge.

DESIGN

Retrospective cohort study on women who gave birth between 2014 and 2019 in Spain.

DATA SOURCES

The data source was a questionnaire distributed through the Spanish breastfeeding associations. The development of the predictive model was made on a cohort of 3387 women and was validated on a cohort of 1694 women. A multivariate analysis was performed by means of logistic regression, and predictive ability was determined by areas under the ROC curve (AUC).

RESULTS

80.2% (2717) women exclusively breastfed at discharge in the derivation cohort, and 82.1% (1390) in the validation cohort. The predictive factors in the final model were: maternal age at birth; BMI; number of children; previous breastfeeding; birth plan; induced birth; epidural analgesia; type of birth; prematurity; multiple pregnancy; macrosomia; onset of breastfeeding within the first hour; and skin-to-skin contact. The predictive ability (ROC AUC) in the derivation cohort was 0.76 (CI 95%: 0.74-0.78), while in the validation cohort it was 0.74 (CI 95%: 0.71-0.77).

CONCLUSIONS

A predictive model of exclusive maternal breastfeeding at hospital discharge has been developed, based on thirteen variables, with satisfactory predictive ability in both the derivation cohort and the validation cohort according to the Swets' criteria. This model can identify women who are at high risk of not breastfeeding at hospital discharge.

Improvements and limitations in developing multivariate models of hemorrhage and transfusion risk for the obstetric population

BACKGROUND

Maternal hemorrhage protocols involve risk screening. These protocols prepare clinicians for potential hemorrhage and transfusion in individual patients. Patient-specific estimation and stratification of risk may improve maternal outcomes.

STUDY DESIGN AND METHODS

Prediction models for hemorrhage and transfusion were trained and tested in a data set of 74 variables from 63 973 deliveries (97.6% of the source population of 65 560 deliveries included in a perinatal database from an academic urban delivery center) with sufficient data at pertinent time points: antepartum, peripartum, and postpartum. Hemorrhage and transfusion were present in 6% and 1.6% of deliveries, respectively. Model performance was evaluated with the receiver operating characteristic (ROC), precision-recall curves, and the Hosmer-Lemeshow calibration statistic.

RESULTS

For hemorrhage risk prediction, logistic regression model discrimination showed ROCs of 0.633, 0.643, and 0.661 for the antepartum, peripartum, and postpartum models, respectively. These improve upon the California Maternal Quality Care Collaborative (CMQCC) accuracy of 0.613 for hemorrhage. Predictions of transfusion resulted in ROCs of 0.806, 0.822, and 0.854 for the antepartum, peripartum, and postpartum models, respectively. Previously described and new risk factors were identified. Models were not well calibrated with Hosmer-Lemeshow statistic P values between .001 and .6.

CONCLUSIONS

Our models improve on existing risk assessment; however, further enhancement might require the inclusion of more granular, dynamic data. With the goal of increasing translatability, this work was distilled to an online open-source repository, including a form allowing risk factor inputs and outputs of CMQCC risk, alongside our numerical risk estimation and stratification of hemorrhage and transfusion.

Poor uterine contractility and postpartum hemorrhage among low-risk women: A case-control study of a large-scale database from Japan

OBJECTIVE

To examine the association between the risk of postpartum hemorrhage (PPH) and poor uterine contractility, which is suggested by the characteristics of labor.

METHODS

This case-control study used cases recorded in the Japan Perinatal Registry database during the period 2013-2016. After exclusion of women with specified known risk factors for PPH, we enrolled 174 082 primiparas who had a full-term live singleton vaginal birth. Participants were classified into four classes according to the diagnosis of abnormal labor patterns and use of uterotonics. χ² tests were used to compare PPH cases with controls, and odds ratios (OR) were calculated by univariate and multivariate analyses.

RESULTS

Among the enrolled women, 10 508 (6.0%) had PPH. Abnormal labor patterns were significantly associated with an increased risk of PPH. Compared with women without any abnormal labor patterns who had not used uterotonics, women with abnormal labor patterns were at a significantly increased risk for PPH regardless of whether they had used uterotonics (adjusted OR 1.23, 95% confidence interval [CI] 1.10-1.37) or not (adjusted OR 1.30, 95% CI 1.23-1.37).

CONCLUSION

Our study suggests that among low-risk women with PPH, poor uterine contractility in labor could be a significant predisposing risk factor for PPH.

Identifying the risk: a prospective cohort study examining postpartum haemorrhage in a regional Australian health service

Background

In industrialised countries the incidence of postpartum haemorrhage (PPH) is increasing, for which exact etiology is not well understood. Studies have relied upon retrospective data with estimated blood loss as the primary outcome, known to be underestimated by clinicians. This study aimed to explore variables associated with PPH in a cohort of women birthing vaginally in coastal Queensland, Australia, using the gravimetric method to measure blood loss.

Methods

Women were prospectively recruited to participate using an opt-out consent process. Maternal demographics; pregnancy history; model of care; mode of birth; third stage management practices; antenatal, intrapartum and immediate postpartum complications; gravimetric and estimated blood loss; and haematological laboratory data, were collected via a pre-designed data collection instrument. Descriptive statistics were used for demographic, intrapartum and birthing practices. A General Linear Model was used for multivariate analysis to examine relationship between gravimetric blood loss and demographic, birthing practices and intrapartum variables. The primary outcome was a postpartum haemorrhage (blood loss > 500 ml).

Results

522 singleton births were included in the analysis. Maternal mean age was 29 years; 58% were multiparous. Most participants received active (291, 55.7%) or modified active management of third stage (191, 36.6%). Of 451 births with valid gravimetric blood loss recorded, 35% (n = 159) recorded a loss of 500 ml or more and 111 (70%) of these were recorded as PPH. Gravimetric blood loss was strongly correlated with estimated blood loss (r = 0.88; p < 0.001). On average, the estimated blood loss was lower than the gravimetric blood loss, about 78% of the measured value. High neonatal weight, perineal injury, complications during labour, separation of mother and baby, and observation of a gush of blood were associated with PPH. Nulliparity, labour induction and augmentation, syntocinon use were not associated with PPH.

Conclusions

In contrast to previous study findings, nulliparity, labour induction and augmentation were not associated with PPH. Estimation of blood loss was relatively accurate in comparison to gravimetric assessment; raising questions about routine gravimetric assessment of blood loss following uncomplicated births. Further research is required to investigate type and speed of blood loss associated with PPH.