External validation and comparison of three scores to predict renal replacement therapy after cardiac surgery: a multicenter cohort.

PURPOSE

Cardiac surgery-associated acute kidney injury requiring renal replacement therapy (RRT) is independently associated with mortality. Several risk scores have been developed to predict the need for RRT after cardiac surgery. We have compared and verified the external validity of the three main available scores for RRT prediction after cardiac surgery: the Thakar score, the Mehta tool, and the Simplified Renal Index.

METHODS

The risk scores were calculated in a cohort of 1084 adult patients, 248 of whom required RRT, who underwent open-heart surgery in 24 Spanish hospitals in 2007. The performance of the systems was determined by examining their discrimination (areas under the receiver operating characteristic curves (aROC) and calibration (Lemeshow-Hosmer chi-square goodness-of-fit statistics).

RESULTS

The aROCs in the Thakar score, the Mehta tool, and the Simplified Renal Index were 0.82, 0.76 and 0.79, respectively. The three scoring systems were poorly calibrated and tended to underestimate the actual need for RRT.

CONCLUSIONS

The Thakar score and the Simplified Renal Index discriminated well between low - and high-risk patients in our cohort, and Thakar outperformed the Mehta tool. These best-performing scores may aid in the selection of optimal therapy, facilitate the planning of hospital resource utilization, improve preoperative counseling, select participants for clinical trials of renal-protective therapies and enable an accurate comparison between different institutions or surgeons.