Use of a urine dipstick and brief clinical questionnaire to predict an abnormal serum creatinine in the emergency department.

OBJECTIVES Prior data demonstrated that a urine dipstick used alone was a sensitive predictor of abnormal creatinine, but not sufficiently enough to forego screening of serum creatinine prior to administration of contrast for diagnostic studies. The authors hypothesized that a brief historical questionnaire coupled with a urine dipstick would have high sensitivity for renal dysfunction, potentially eliminating the need for a serum creatinine prior to contrast administration. METHODS This was a prospective study of a convenience sample of patients at two academic tertiary-care emergency departments (EDs) during 2006-2007. Subjects included patients who had both a serum creatinine result reported by the laboratory and a urine dipstick result reported in the medical record. Data included triage vital signs, basic demographic data, 14 medical history items, dipstick urinalysis, and serum creatinine results. The main outcome measure was an abnormal serum creatinine, defined as greater than 1.5 mg/dL. RESULTS Complete data sets were collected on 1,354 patient visits. Of these, there were 161 (12%) with a serum creatinine of >1.5 mg/dL. Logistic regression analysis identified the following independent predictors associated with elevated creatinine: age greater than 60 years, known renal insufficiency, diabetes, hypertension, diuretic use, vomiting, and proteinuria. Nearly all patients with abnormal creatinine (98%) had at least one of these seven predictors. A decision tool combining these predictors would have identified 158 of 161 patients with an abnormal creatinine (sensitivity, 98.1%; 95% confidence interval [CI] = 95.8% to 99.9%) and a specificity of 21.2% (95% CI = 18.8% to 23.2%). CONCLUSIONS The absence of six historical factors and absence of proteinuria can be safely used to identify patients who are unlikely to have an abnormal creatinine.