Models predict acute kidney injury after cardiac surgery

Models based on perioperative basic metabolic panel laboratory values performed well in predicting acute kidney injury at 72 hours and 14 days after cardiac surgery, researchers reported in JAMA.

The researchers created four models. The first was based on preoperative serum creatinine, the second on perioperative absolute change in serum creatinine, the third on the first two models combined and the fourth on the third model plus blood urea nitrogen, potassium, bicarbonate, sodium and albumin adjusted from time of surgery to blood draw.

The derivation cohort included 58,526 patients (median age, 66 years; 67% men; 91% white) who underwent CABG, valve surgery or aorta surgery at Cleveland Clinic from 2000 to 2019. The validation cohort included 4,734 patients (median age, 67 years; 71% men; 87% white) who underwent CABG, valve surgery or aorta surgery at three community hospitals.

In the derivation cohort, at 72 hours, 4.6% had moderate to severe acute kidney injury (AKI) and 1.48% had AKI requiring dialysis, while at 14 days, 5.4% had moderate-to-severe AKI and 1.74% had AKI requiring dialysis, SevagDemirjian, MD, a nephrologist at Cleveland Clinic, and colleagues wrote.

The median interval to first metabolic panel after surgery was 10 hours.

AKI after cardiac surgery

Demirjian and colleagues wrote that the models had excellent predictive discrimination in the derivation cohort for moderate to severe AKI at 72 hours (area under the receiver-operating characteristic curve, 0.876; 95% CI, 0.869-0.883), AKI requiring dialysis at 72 hours (AUC, 9.16; 95% CI, 0.907-0.926), moderate to severe AKI at 14 days (AUC, 0.854; 95% CI, 0.85-0.861) and AKI requiring dialysis at 14 days (AUC, 0.9; 95% CI, 0.889-0.909).

In the validation cohort, the models also performed well at predicting moderate to severe AKI at 72 hours (AUC, 0.86; 95% CI, 0.838-0.882), AKI requiring dialysis at 72 hours (AUC, 0.879; 95% CI, 0.84-0.918), moderate to severe AKI at 14 days (AUC, 0.842; 95% CI, 0.82-0.865) and AKI requiring dialysis at 14 days (AUC, 0.873; 95% CI, 0.836-0.91), the researchers wrote.

Demirjian and colleagues wrote all models were well calibrated and had Spiegelhalter z statistic P values greater than .05.

“By using routinely measured, standardized, readily available, objective parameters, the models could be readily implemented elsewhere,” the researchers wrote. “However, these models may require further validation in other populations.”

‘A promising alternative’

In a related editorial, Marlies Ostermann, MD, PhD, consultant in critical care and nephrology at Guy’s and St. Thomas’ Foundation Trust, London, and honorary senior lecturer at King’s College London, and colleagues wrote: “Strategies aimed at optimizing hemodynamics and fluid status and avoiding hyperglycemia and nephrotoxic exposures have been shown to be effective at preventing moderate or severe AKI in high-risk patients after cardiac surgery. Thus, it is widely acknowledged that these patients should be identified as early as possible. A major limitation in clinical practice is the lack of reliable tools to complete this task in a timely and practical manner. Although novel kidney biomarkers are effective and included in some guidelines, these injury biomarkers are not universally available. Demirjian and colleagues offer a promising alternative.”