Abstract:
Background: Postpartum hemorrhage is defined as loss of blood of more than 500 milliliters following a vaginal delivery or more than 1000 ml following caesarian section or blood loss that can cause hemodynamic derangement after delivery. The incidence of postpartum hemorrhage has increased in both high and low income countries. Worldwide, postpartum hemorrhage affects about 5% of all deliveries and more than 50% of deaths associated with it are preventable. Maternal mortality ratio an indicator in the sustainable development goal and Ethiopia set strategies to reduce it to less than 70 per 100,000 live births by 2030. Despite the sustained efforts to reduce postpartum hemorrhage, its magnitude is still at highest level and continues to remain the leading cause of maternal mortality. Therefore, for Ethiopia to achieve the SDG target focused on maternal mortality, individualized prediction model for primary postpartum hemorrhage could play additional role but, there are no studies that estimat the collective impact of different factors together on postpartum hemorrhage in Ethiopia as far as my search.
Objective: To develop and validate risk prediction model for primary postpartum hemorrhage using maternal characteristics.
Method: A hospital-based un-matched case–control study design will be conducted from 24th
February 2021 to 30th April 2021. The sample size will be determined by double population proportion formula using r = 2 (ratio of control to case), 80% power and 95% confidence level and total sample size will be 810 mothers (270 cases and 540 controls). Systematic random sampling method will be used to select study units. Data will be coded and entered into Epi data version 3.2 and will be analyzed by STATA version 14 software. For model development simple binary logistic regression will be done to identify the relationship between each predictor and
primary postpartum hemorrhage. Variables with p-value ≤ 0.2 from the univariate analysis will
be entered into a backward stepwise multiple binary logistic regression model, and significant variables with p-value < 0.05 will be retained in the multivariate model.
The model accuracy will be checked by calculating calibration plot, the area under the ROC curve (AUC) (discrimination). To check for model goodness, Hosmer-Lemeshow goodness of fit statistics will be generated. Internal validation of the model will be calculated by bootstrapping method taking 2000 samples with replacement. The results of significant predictors will be reported as coefficients, odds ratios (AORs), with their 95% confidence intervals (CI). Keywords: Primary post partum hemorrhage, risk prediction model, risk score,FHCSH, Ethiopia