Prediction of Postoperative Day 1 Hemoglobin Level after Cesarean Delivery

To create a prediction model for postoperative hemoglobin levels after cesarean delivery, which could reduce routine use of postoperative laboratory test.

This was a secondary analysis of a retrospective cohort study of all women who underwent cesarean delivery (primary or repeat) at or more than 23 weeks' gestation at a single academic center. The cohort was randomly divided into a training cohort to develop a prediction model and a validation cohort to test the model in a 2:1 ratio. Variables with p-value <0.10 were considered for the mixed multivariable linear regression model in a backward stepwise fashion. We obtained the best cut-off point of the predicted hemoglobin level to detect severe anemia (postoperative hemoglobin level less than 7.0 g/dL) in the training cohort. A receiver operating characteristic curve with the area under a curve was created. We calculated the sensitivity and specificity of the model in the validation cohort using the best cut-off point obtained in the training cohort as well as postoperative hemoglobin of 10.0 g/dL.

Of 2,930 women, 1,954 (66.6%) and 976 (33.3%) were randomly allocated to training and validation cohorts. The final model included preoperative hemoglobin level, preoperative platelet level, quantitative blood loss, height, weight, magnesium administration, labor, and general anesthesia. The best cut-off to predict severe anemia was predicted hemoglobin level of 8.57 g/dL in the training cohort. Using this cut-off, the sensitivity and specificity in the validation cohort were 77% (95% confidence interval [CI]: 56-91%) and 87% (95% CI: 85-89%), respectively. The use of postpartum hemorrhage yielded the sensitivity of 58% (95% CI: 37-77%) and specificity 79% (95% CI: 76-81%), respectively.

We developed a validated model to predict the postoperative day 1 hemoglobin levels after cesarean delivery that could assist with identifying women who may not need postoperative laboratory tests.

· Postoperative laboratory tests are routine.. · A prediction model may allow reduce routine tests.. · We developed an accurate mathematical model..