Patients undergoing liver resection, excluding those with liver trauma.
Overall assessment: Low quality.
Prediction accuracy: Fair accuracy (c-statistic 0.77) in internal validation.
Risk to reproducibility: Low. The model was developed in a single center using prospectively collected data from 380 patients of whom 58 developed acute kidney injury and internally validated in a randomly selected sample at the same center of 189 patients of whom 28 developed acute kidney injury. The authors tested few variables and used appropriate statistical methods to reduce the risk that the model may perform worse in new patients.
Risk of data reporting bias: Low. The model was developed in a prospective cohort study with systematic monitoring of postoperative renal function.
Risk to transportability: High. May perform adequately in new patients but only data from a single center in Switzerland are available. Prospective validation is needed.
Prediction accuracy: Fair accuracy (c-statistic 0.77) in internal validation.
Risk to reproducibility: Low. The model was developed in a single center using prospectively collected data from 380 patients of whom 58 developed acute kidney injury and internally validated in a randomly selected sample at the same center of 189 patients of whom 28 developed acute kidney injury. The authors tested few variables and used appropriate statistical methods to reduce the risk that the model may perform worse in new patients.
Risk of data reporting bias: Low. The model was developed in a prospective cohort study with systematic monitoring of postoperative renal function.
Risk to transportability: High. May perform adequately in new patients but only data from a single center in Switzerland are available. Prospective validation is needed.