Abstract
ED Siew, TA Ikizler, ME Matheny, et al. Clinical Journal of the American Society of Nephrology 2012;
Acute kidney injury (AKI) is defined and staged by changes in serum creatinine concentration and requires accurate assessment of baseline kidney function for correct classification. Quick, accurate and automatable methods for estimating baseline creatinine are needed where manual adjudication is impractical.
This study compared baseline creatinine assessed by a panel of clinical nephrologists to surrogate estimation methods in 379 adults admitted to a tertiary referral centre. Patients were included if they had chronic kidney disease (CKD) stage 3/4 or developed AKI during admission, and ≥2 creatinine results 1-730 days before admission were available. Agreement was classified as a <18 μmol/L difference (<10% if >180 μmol/L) and described by the intraclass correlation coefficient (ICC).
Agreement between primary reviewers was 88%, with an ICC of 0.93 (95% confidence interval [CI], 0.91-0.94); all cases of disagreement were resolved by a third reviewer. Mean outpatient creatinine 7-365 days before admission best approximated expert opinion (ICC 0.91, 95% CI 0.88-0.92) and was superior to the most recent outpatient (ICC 0.84, 95% CI 0.80-0.88, P < 0.001) and nadir outpatient creatinine (ICC 0.83, 95% 0.76–0.87, P < 0.001) results, but was not significantly different from the most recent inpatient or outpatient creatinine (ICC 0.88, 95% CI 0.85–0.91) during the same time window. Including results <7 or 365–730 days before admission allowed baseline estimation in more patients, but tended to reduce agreement with expert opinion for all methods.
For the 7–365 day time interval, mean outpatient, most recent outpatient and most recent inpatient/outpatient methods resulted in similar proportions of potential AKI misclassification (>26 μmol/L difference from expert opinion) at 11.1%, 13.7% and 12.1%, respectively, with estimations usually exceeding baseline. Nadir outpatient creatinine resulted in potential AKI misclassification in 20.3% and usually underestimated baseline.
This study highlights the difficulties in defining baseline creatinine and demonstrates that baseline estimation using nadir creatinine shows high potential for AKI misclassification. This study was weighted towards patients with CKD, with no significant difference between methods in the small subset of patients with estimated glomerular filtration rate >60 mL/min/1.73 m2. The applicability of these results to patients with normal baseline renal function is therefore questionable.