Methodological Issues with Head and Neck Cancer Prognostic Risk Prediction Models

Abstract

Objective: Prognostic risk prediction models estimate the probability of developing head and neck cancer (HNC), providing valuable information for managing the disease. While different prognostic HNC risk prediction models have been developed worldwide, a comprehensive evaluation of their methods is lacking. We conducted a scoping review with a critical assessment aiming to identify the methodological strengths and limitations of HNC risk prediction models. Method: We searched Medline, Embase, Scopus, Web of Science, and CAB Abstracts databases and included full-text-available peer-reviewed published papers on developing or validating a prognostic HNC risk prediction model. Study quality was appraised using the PROBAST tool. Results: Nine papers were included. Although all had a high risk of bias, mainly in the analysis domain, only two studies had high concerns about clinical applicability. Conclusion: Currently published studies provide insufficient information on methods, making it difficult to judge the models’ quality and applicability. Future investigations should follow the guidelines in reporting the prediction modelling studies.