Acute kidney injury network staging in geriatric postoperative acute kidney injury patients: shortcomings and improvements.

Journal of the American College of Surgeons

PubMedID: 23870218

Chao CT, Lin YF, Tsai HB, Wu VC, Ko WJ. Acute kidney injury network staging in geriatric postoperative acute kidney injury patients: shortcomings and improvements. J Am Coll Surg. 2013;217(2):240-50.
BACKGROUND
The incidence of acute kidney injury (AKI) is rising, particularly among the elderly. However, the optimal risk stratification scheme for these patients is unknown. The Acute Kidney Injury Network (AKIN) classification application in geriatric patients has not been previously confirmed.

STUDY DESIGN
In this multicenter study, elderly patients (>65 years old) who had major surgery and were admitted to ICUs between January 1, 2002 and December 31, 2008 were recruited and grouped according to the AKIN creatinine criteria. The utility of the AKIN criteria for the prediction of in-hospital mortality was determined using Cox proportional hazard regression modeling.

RESULTS
A total of 4,240 eligible patients were identified and separated into "non-AKI" (n = 3,259), AKIN 1 (n = 582), AKIN 2 (n = 78), and AKIN 3 groups (n = 321). Cox proportional hazard regression analysis revealed that the AKIN 3 group has a significantly higher hospital mortality compared with the non-AKI group (hazard ratio [HR] 3.19, 95% CI 2.16 to 4.71; p < 0.001); the AKIN 1 (p = 0.611) and AKIN 2 (p = 0.104) groups have no significant differences compared with the non-AKI group. After excluding patients who received hemodialysis 1 week postoperatively, the AKIN 2 group predicted a significantly higher risk of hospital mortality compared with the non-AKI group (HR 2.31; p = 0.008).

CONCLUSIONS
This is the first study to demonstrate the poor applicability of the AKIN classification in the prediction of in-hospital mortality in geriatric postoperative AKI patients. Consideration of late dialysis status may enhance the discriminative power of AKIN in this specific population.