Adults undergoing any orthopedic surgery, excluding patients receiving chronic renal replacement therapy.
Overall assessment: Moderate quality.

Prediction accuracy: Fair (c-statistic 0.74 in the development cohort and 0.70 in the validation cohort). Substantially over-predicts risk beyond 20%.

Risk to reproducibility: Low. The model was developed and validated in large samples.

Risk of data reporting bias: Moderate. The model was developed and retrospectively validated using routinely-collected data; patients with missing preoperative or postoperative creatinine measurements were excluded initially but when these data were imputed the results did not change appreciably.

Risk to transportability: Moderate. Developed in 2 centers in Scotland using retrospective data from 6220 patients of whom 672 experienced acute kidney injury. Validated using retrospective data from 4395 patients (295 of whom experienced acute kidney injury) at another center in Scotland. Prospective external validation is needed.