Prediction Models for Shared Decision-making by Patients with Localised Renal Masses: A Systematic Review and a...

Counselling a patient with a localised renal mass (LRM) can be complex. While many models can predict various outcomes (nature of the mass, survival, technical points of the intervention), none is widely used. We conducted a systematic review of models that could potentially be used to assist patients' decision-making for LRMs (PROSPERO 1048873). Among 6833 records screened, 284 met the inclusion criteria covering nine broad categories (benign vs malignant, specific pathology, grade, adverse pathology, survival, genomics, renal function, complications, technical factors), of which 101 used existing nomograms (predominantly nephrometry scores). Of the 195 studies that developed an original model, 79 (40.5%) provided a nomogram that could be used for patient decision-making, while 105 (53.8%) provided no validation for the new model. We conducted an online survey that was completed by 31 clinicians and 35 patients/carers and identified six priority areas when considering treatment for a newly diagnosed LRM. Twenty models covered aspects of these priority areas, but no single model adequately covered them all. Further research is required for proper assessment of clinician and patient/carer priorities, but the results from our review and the survey that included multiple stakeholders provide clarity on clinical utility across the heterogeneous field of models available. The design of future models should include clinically relevant outputs, should clearly state the nature of the development cohort and model performance, and should ideally provide a user-friendly version of the model via which patients can obtain relevant information.