Derivation and Validation of a Simplified Predictive Index for Renal Replacement Therapy After Cardiac Surgery

摘要: ContextA predictive index for renal replacement therapy (RRT; hemodialysis or continuous venovenous hemodiafiltration) after cardiac surgery may improve clinical decision making and research design.ObjectivesTo develop a RRT using preoperative information.Design, Setting, ParticipantsRetrospective cohort of 20 131 patients at 2 hospitals in Ontario, Canada. The derivation consisted 10 751 Toronto General Hospital (1999-2004). validation cohorts 2566 (2004-2005) 6814 Ottawa Heart Institute (1999-2003).Main Outcome MeasurePostoperative RRT.ResultsRRT rates the derivation, validation, were 1.3%, 1.8%, 2.2%, respectively. Multivariable predictors estimated glomerular filtration rate, diabetes mellitus requiring medication, left ventricular ejection fraction, previous surgery, procedure, urgency intra-aortic balloon pump. was scored from 0 to 8 points. An rate less than equal 30 mL/min assigned points; other components 1 point each: 31 60 mL/min, mellitus, fraction 40%, procedure coronary artery bypass grafting, pump, nonelective case. Among 53% with low risk scores (≤1), 0.4%; by comparison, this 10% among 6% high-risk (≥4). had areas under receiver operating characteristic curve 0.81, 0.78, When these stratified based on scores, likelihood ratios more concordant observed rates.ConclusionsRRT is predicted readily available information. A simple information discriminated well between low- cohorts. improved generalizability when used predict RRT.