Table 2.
Actual AKI misclassification rates using baseline serum creatinine estimation methods
| Estimation Methoda | Overall | eGFR < 60 ml/min per 1.73 m2 (n=1118) | eGFR ≥ 60 ml/min per 1.73 m2 (n=5384) |
|---|---|---|---|
| Single eGFR 75 | 12.3 (11.5–13.2) | 40.5 (37.6–43.5) | 6.5 (5.8–7.2) |
| Basic multiple imputation | 10.0 (9.3–10.8) | 19.1 (16.8–21.5) | 8.1 (7.4–8.9) |
| Basic multiple imputation with SCr | 9.3 (8.6–10.0) | 13.9 (11.9–16.1) | 8.3 (7.6–9.1) |
| Full multiple imputation | 11.3 (10.6–12.1) | 28.3 (25.7–31.0) | 7.8 (7.1–8.6) |
| Full multiple imputation with SCr | 9.0 (8.3–9.7) | 15.3 (13.3–17.6) | 7.7 (7.0–8.4) |
Misclassification rates (%) with 95% Wilson confidence intervals for each method are shown for the overall group. Estimated GFR based on minimum serum creatinine during the first 7 days of hospitalization. All multiple-imputation groups were compared with single eGFR 75 as the reference group. All differences in misclassification were statistically significant (P<0.001) except for the full multiple imputation versus single eGFR 75 in the overall comparison (P=0.06). eGFR, estimated GFR; eGFR 75, eGFR = 75 ml/min per 1.73 m2; SCr, serum creatinine.
Basic multiple imputation = three-variable multiple imputation; basic multiple imputation with SCr = basic three-variable multiple imputation plus minimum inpatient serum creatinine; full multiple imputation = full-variable multiple imputation; full multiple imputation with SCr = full variable multiple imputation plus minimum inpatient serum creatinine.