Retrospective cross-validation of simplified predictive index for renal replacement therapy after cardiac surgery

Interact Cardiovasc Thorac Surg. 2008 Dec;7(6):1101-6. doi: 10.1510/icvts.2008.181438. Epub 2008 Jul 31.

Abstract

Objectives: Acute kidney impairment requiring renal replacement therapy is an infrequent but dangerous complication of cardiac surgery. Its development is associated with high mortality and morbidity. A recently published simple risk stratification engine has been developed and validated in the USA and Canada, but its discriminatory power has never been tested in Europe. We aimed to cross-validate the newly developed risk stratification algorithm in a group of patients operated on in a single centre in Poland.

Methods: From electronic database we selected 1421 patients fulfilling identical inclusion and exclusion criteria as in derivation cohort in Canada. In each patient eligible for analysis we calculated simplified renal index and assessed its predictive power for the need of renal replacement therapy.

Results: After surgery 33 (2.3%) patients developed acute kidney impairment and subsequently underwent renal replacement therapy. The simplified renal index predicted risk of postoperative renal replacement therapy in our group. Patients with low values of simplified renal index (0-1), medium (2-3) and high values (4 and more) were found to have increasingly higher risk for renal replacement therapy of 1.1% (95% CI: 0.5-2.1%), 3.2% (95% CI: 1.9-5%) and 12.5% (95% CI: 5.2-24.1%), respectively. The area under the ROC curve of simplified renal index as predictor of renal replacement therapy in our centre was 0.73 (95% CI: 0.62-0.81) and did not differ significantly from the values obtained in the original paper.

Conclusion: The new risk stratification algorithm is effective in discrimination of patients at high risk for development of acute kidney impairment with the need of renal replacement therapy.

Publication types

  • Validation Study

MeSH terms

  • Acute Disease
  • Aged
  • Algorithms
  • Cardiac Surgical Procedures / adverse effects*
  • Decision Support Techniques
  • Female
  • Health Status Indicators*
  • Humans
  • Kidney Diseases / diagnosis
  • Kidney Diseases / etiology
  • Kidney Diseases / therapy*
  • Male
  • Middle Aged
  • Poland
  • Predictive Value of Tests
  • ROC Curve
  • Renal Replacement Therapy*
  • Reproducibility of Results
  • Retrospective Studies
  • Risk Assessment
  • Sensitivity and Specificity