Assessing the overlap of candidate and included factors among non-validating studies of prognostic models: A review from published literature

In this thesis, I reviewed published prognostic model studies on Adhesive Small Bowel Obstruction (ASBO) and extracted the predictive factors that were retained in each model’s final analysis. ASBO is a common surgical condition that can require either operative or non-operative management. Several prognostic models have been developed to help predict which patients may need surgical or other treatment. However, these models vary widely in the factors they include, and many of them have not been externally validated. This study aims to identify the most common and clinically relevant factors used across these existing models and may serve as a foundation for future work to develop and validate a more consistent prognostic model for ASBO. Methods: A review of published prognostic model studies on ASBO was conducted. Reporting adherence of the included studies was assessed using the TRIPOD statement, and for each included study, all reported variables were extracted and categorized as candidate and final predictors based on their inclusion in the model development process. Extracted predictors were analyzed to evaluate overlap and identify commonly used variables across models. Results: A total of 26 prognostic model studies were included. A wide range of candidate predictors was reported across studies, while relatively few predictors were retained in final models. Overlap among final predictors was very limited, indicating substantial heterogeneity in predictor selection. Reporting adherence according to the TRIPOD statement was variable among studies. Discussion: This review demonstrates considerable heterogeneity and minimal overlap among predictors used in prognostic models for ASBO, limiting their comparability and clinical applicability. Differences in study design, populations, outcomes, and reporting practices may partly explain these findings. Improved standardization of predictor selection and adherence to reporting guidelines are needed to support the development and validation of more consistent and clinically useful prognostic models for ASBO.