Date of Award
Open Access Thesis
Medical Doctor (MD)
Richard Formica, Howard Forman
Renal Ultrasound in the Evaluation of Acute Kidney Injury: Developing a Risk Stratification Framework. Adam Licurse (Yale School of Medicine, New Haven, CT), Michael Kim (Yale College, New Haven, CT), Howard Forman (Yale Department of Radiology, Yale University, School of Medicine, New Haven, CT), Richard Formica and Chirag Parikh (Section of Nephrology, Department of Internal Medicine, Yale University, School of Medicine, New Haven, CT), Danil Makarov (Robert Wood Johnson Clinical Scholars Program, Yale University, School of Medicine, New Haven, CT), James Dziura (Yale Center for Clinical Investigation, Yale University, School of Medicine, New Haven, CT), Cary Gross (Section of General Medicine, Yale University, School of Medicine, New Haven, CT) Background: In adult inpatients with acute kidney injury (AKI), clinicians routinely order a renal ultrasound (RUS). It is unclear how often this test provides clinically useful information. Specific Aims: We aimed to develop a simple decision rule that will identify those patients with AKI who are at low risk of hydronephrosis (HN) on RUS, or HN requiring an intervention (HNRI; defined as nephrostomy tube or stent placement). Using this classification scheme, we also aimed to evaluate the effectiveness of RUS evaluation in each group, in terms of number needed to screen (NNS). Methods: We conducted a cross-sectional study, divided into derivation and validation samples. Our sample consisted of 997 U.S. adult inpatients who were admitted to Yale-New Haven Hospital (YNHH) from January 1, 2005 to May 1, 2009, diagnosed with AKI, and received diagnostic RUS in the evaluation of their elevated Creatinine. Pregnant women, renal transplant recipients, and patients with recently diagnosed HN were excluded. Demographic and clinical characteristics were abstracted from the medical records, including pre-existing comorbidities, inpatient course (e.g. use of pressors) and exposures (e.g. contrast or nephrotoxic medications). A multivariable logistic regression model was used to create risk strata for HN and HNRI; a separate sample was used for validation. We assessed the presence of incidental findings on RUS for each stratum (i.e. other clinically useful findings other than presence of HN). In the validation sample, patients were classified according to their risk of HN; this system was assessed in terms of its sensitivity, specificity, negative predictive value (NPV), NNS, and cost of RUS evaluation per positive study. Results: In the derivation sample of 200 patients, seven factors were found to be associated with HN: history of HN, recurrent urinary tract infections, diagnosis consistent with obstruction, non-black race; and absence of exposure to nephrotoxic medications, congestive heart failure, or pre-renal status prior to AKI. Patients were assigned to the low risk group when 0-2 risk factors were present, medium risk group when 3 factors were present, and high risk group when 4 or more factors were present or when there was a documented history of HN. Among 797 patients in the validation sample (mean age of 65.6 years), 10.6% had HN and 3.3% had HNRI. Of the 223 patients assigned to the low risk group, seven (3.1%) had HN and one (0.4%) had HNRI (223 patients needed to screen to find one HNRI). In this group, there were no incidental findings on RUS unknown to the clinical team. In the higher risk group, 15.7% had HN and 4.7% had HNRI. The NNS to find one case of HN in the low risk group is 32, or $6,371 per positive study (at a cost of $200 per study). For HNRI, the NNS for the low risk group in the same model was 223, at a cost of $44,600 per positive study. Conclusions: In adult inpatients with acute kidney injury, specific factors can identify patients who are unlikely to have hydronephrosis, or hydronephrosis requiring an intervention, on renal ultrasound.
Licurse, Adam, "Renal Ultrasound in the Evaluation of Acute Kidney Injury: Developing a Risk Stratification Framework" (2010). Yale Medicine Thesis Digital Library. 118.