Predicting Graft Loss Following Acute Kidney Injury in Patients With a Kidney Transplant

• Main Author: Molnar, Amber
• Other Authors: Knoll, Greg
• Language: en
• Published: Université d'Ottawa / University of Ottawa 2016
• Subjects: Acute kidney injury; kidney transplant; Healthcare databases; Risk score
• Online Access: http://hdl.handle.net/10393/34236; http://dx.doi.org/10.20381/ruor-2879

Acute kidney injury (AKI), characterized by an abrupt loss of kidney function with retention of nitrogenous waste products, is common in the months to years following kidney transplantation and is associated with an increased risk of transplant failure (graft loss). Kidney transplant patients who experience graft loss and return to dialysis have an increased mortality risk and a lower quality of life. Research involving kidney transplant patients can prove challenging, as they are relatively small in number. To increase statistical power, researchers may utilize administrative databases. However, these databases are not designed primarily for research, and knowledge of their limitations is needed, as significant bias can occur. When using administrative databases to study AKI in kidney transplantation, the method used to define AKI should be carefully considered. The power of a study may be greatly increased if AKI can be accurately defined using administrative diagnostic codes because data on AKI will be universally available for all patients in the database. However, the methods by which diagnostic codes are assigned to a patient allow for error to be introduced. We confirmed that, when compared to the gold standard definition for AKI of a rise in serum creatinine, the diagnostic code for AKI has low sensitivity but high specificity in the kidney transplant population (the best performing coding algorithm had a sensitivity of 42.9% (95% CI 29.7, 56.8) and specificity of 89.3% (95% CI 86.2, 91.8) (Chapter 3). We therefore determined that for the study outlined in Chapter 4, defining AKI using diagnostic codes would significantly under-­capture AKI and misclassify patients. We decided to define AKI using only serum creatinine criteria even though this would limit our sample size (creatinine data was only available for a subset of patients in the administrative databases). In Chapter 4, we derived an index score to predict the risk of graft loss in kidney transplant patients following an admission to hospital with AKI. The index includes six readily available, objective clinical variables that increased the risk of graft loss: increasing age, increased severity of AKI (as defined by the AKIN staging system), failure to recover from AKI, lower baseline estimated glomerular filtration rate, increased time from kidney transplant to AKI admission, and deceased donor. The derived index requires validation in order to assess its utility in the clinical realm.