Long-term survival for ICU patients with acute kidney injury
© Scott et al.; licensee BioMed Central Ltd. 2012
Published: 20 March 2012
A recently published study  validated the criteria used in the Acute Kidney Injury Network (AKIN) definitions  of the three stages of acute kidney injury (AKI) using in-hospital mortality. In the present study, we validate the clinical applicability of the AKIN classifications through long-term survival analysis of AKI patients.
From over 17,000 adult ICU patients in the MIMIC II database [3, 4] (V2.5), we excluded patients having end-stage renal disease and those with insufficient data and determined AKI stages for each patient. Multivariate Cox regression was used to determine hazard ratios (HRs) for 2-year survival, controlling for: age, sex, nonrenal Sequential Organ Failure Assessment (SOFA) score and selected co-morbidities.
Among the final cohort of 14,525 patients, 43% had no AKI and 39%, 14% and 4% developed AKI 1, 2 and 3 respectively. The results of the regression analysis show that AKI 1 (HR 1.12, P < 0.05), AKI 2 (HR 1.10, P = 0.05) and AKI 3 (HR 1.64, P < 0.001) were significantly associated with increased 2-year mortality. In addition, age (HR 1.04, P < 0.001), gender (M) (HR 0.93, P < 0.05), nonrenal SOFA score (HR 1.05, P < 0.001) and all co-morbidities were significant predictors. Adjusted and unadjusted Kaplan-Meier curves for patients with AKI 3 are remarkably different from each other, suggesting that in these most severely ill patients AKI is only one aspect of their illness.
AKI stages 1, 2 and 3 are significant indicators of 2-year mortality. The difference between AKI 1 and 2 is smaller than that between AKI 2 and 3 and it may be prudent to re-examine the criteria used to define AKI to provide better separation among the three classes.
- Mandelbaum T, et al.: Crit Care Med. 2011, 39: 2659-2664.PubMed CentralPubMedGoogle Scholar
- Mehta RL, et al.: Crit Care. 2007, 11: R31. 10.1186/cc5713PubMed CentralView ArticlePubMedGoogle Scholar
- Saeed M, et al.: Crit Care Med. 2011, 39: 952-960. 10.1097/CCM.0b013e31820a92c6PubMed CentralView ArticlePubMedGoogle Scholar
- MIMIC II databases[http://physionet.org/mimic2]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.