Predicting hospital-acquired acute kidney injury - A case-controlled study

Paul E. Drawz, R. Tyler Miller, Ashwini R. Sehgal

Research output: Contribution to journalArticlepeer-review

22 Scopus citations


Acute kidney injury is a major complication of hospitalization, occurring in 5-7 percent of hospitalized patients. The patient characteristics and prognostic variables that help predict acute kidney injury have not been studied in the general hospitalized population. The objectives of this study are to derive and validate a predictive score for hospital-acquired acute kidney injury (HAKI). We conducted a case-controlled study of HAKI involving 180 cases and 360 controls. A multivariate logistic regression model was developed in two-thirds of the subjects and validated in the other third. Upon admission, cases in the developmental sample were older (67 vs. 63 yrs, p =.008) and more likely to have diabetes (51% vs. 35%; p =.003), hypertension (77% vs. 60%, p =.001), heart failure (34% vs. 20%, p =.004), blood urea nitrogen ≥25 mg/dL (38% vs. 20%, p = <.001), creatinine ≥1.1 mg/dL (65% vs. 39%; p <.001), albumin ≤4 g/dL (85% vs. 71%; p =.033), and bicarbonate <24 mEq/L or >30 mEq/L (42% vs. 29%; p =.05) compared to controls. The final risk score included pulse, bicarbonate, creatinine, and specific medications (NSAIDs, ACE inhibitors, ARBs, and/or diuretics). The c-statistic for the risk score in the developmental sample was 0.69. In the validation sample, an increasing number of risk factors was associated with increased risk of HAKI (16% and 62% in the low and high-risk groups, respectively). In conclusion, a simple model based on readily available data stratifies patients according to their risk of developing HAKI and may guide clinical decision making and provide a basis for further research into HAKI.

Original languageEnglish (US)
Pages (from-to)848-855
Number of pages8
JournalRenal Failure
Issue number9
StatePublished - Oct 2008

Bibliographical note

Funding Information:
Paul E. Drawz was supported in part by NIH Training Grant 5T32DK007470-23. R. Tyler Miller was supported by a VA Merit Review and NIH grant RO1 DK59985. Ashwini R. Sehgal was supported by NIH grant DK51472.


  • Acute renal insufficiency
  • Bicarbonates
  • Hospitalization
  • Risk factors


Dive into the research topics of 'Predicting hospital-acquired acute kidney injury - A case-controlled study'. Together they form a unique fingerprint.

Cite this