For a study, researchers sought to comprehensively examine all research that created or verified a prediction model for vaginal birth after cesarean (VBAC).
When developing or validating a multivariable VBAC prediction model for women who had a singleton pregnancy and one prior lower-segment cesarean delivery, they considered observational research. 3,758 articles in total were found and evaluated.
The CHARMS (Critical Appraisal and Data Extraction for Systematic Review of Prediction Modelling Studies) checklist-based tool was used to extract data for each of the 57 included studies. The data included participant characteristics, sample size, predictors, timing of application, and performance. Two tools were used to evaluate the risk of bias and the transparency of reporting: PROBAST (Prediction Model Risk of Bias Assessment Tool) and TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis). More than one model was created or verified by several investigations. There were 38 original prediction models, 42 external validations of the 10 models already in use, and 6 revisions already in use. Only 19 (19/38, 50%) of the 38 distinct models were internally verified in the original investigation. In the original investigation, no studies externally evaluated their model. The most frequent predictors were age, prior vaginal births, and prior cesarean deliveries due to labor dystocia. The included studies’ areas under the curve varied from 0.61 to 0.95. In general, models used around delivery fared better than those used earlier in pregnancy. The majority of research (45/57, 79%) showed a high risk of bias, whereas the remaining studies (7/57, 12%) and (5/57, 9%) were ambiguous or low. Adherence to the TRIPOD checklist was 70% on average (range: 32–93%).
There were several models for predicting VBAC success, but many of them lacked outside validation and had a substantial bias risk. Models used around birth fared better than those used earlier in pregnancy, but it’s still unclear how broadly or effectively they may be utilized. To inform clinical usage, high-quality external validation and effect studies are needed.