LAUSR

Key Points Question Can routinely available data from primary care visits be used to develop and externally validate a prediction model that reliably predicts estimated glomerular filtration rate (eGFR) for upcoming follow-up visits? Findings In this prognostic study involving 4637 adults with type 2 diabetes and chronic kidney disease, a prediction model including 13 routinely collected baseline variables based on data from 3 prospective multinational cohort studies was developed and externally validated. The model was robust, well calibrated, and capable of predicting decreases in eGFR up to 5 years after baseline. Meaning These findings suggest that the prediction model, which is publicly available in a web-based application, may be used to improve prediction of individual eGFR trajectories and kidney function decline.