• Christopher W Jones, Karissa D Culbreath, Abhi Mehrotra, and Peter H Gilligan.
• Department of Emergency Medicine, Christiana Care Health System, Newark, Delaware; Department of Emergency Medicine, University of North Carolina Chapel Hill, Chapel Hill, North Carolina.
• J Emerg Med. 2014 Jan 1;46(1):71-6.

BackgroundThe yield of urine culture testing in the emergency department (ED) is often low, resulting in wasted laboratory and ED resources. Use of a reflex culture cancellation protocol, in which urine cultures are canceled when automated urinalysis results predict that culture yield will be low, may help to conserve these resources.Study ObjectivesTo identify a reflex culture cancellation protocol consisting of urinalysis-based criteria to limit urine culture over-utilization.MethodsWe studied patients aged 5 years and older whose ED evaluation included both an automated urinalysis and urine culture. Logistic regression models incorporating individual urinalysis components were used to predict culture growth. Receiver operating characteristic curves corresponding to each model were constructed, and the area under the curve was used to identify the model that best predicted positive urine culture growth.ResultsThere were 1546 ED patients who met study inclusion criteria. Of these, 314 (20%) had positive urine cultures. Restriction of culture testing to samples with white blood cells > 10 per high-power field, positive nitrites, positive leukocyte esterase, or positive bacteria provided a sensitivity of 96.5% (95% confidence interval [CI] 93.6-98.1%) and specificity of 48.1% (95% CI 45.3-51.0%) for positive urine culture. Implementation of a reflex culture cancellation protocol based on these criteria would have eliminated 604 of 1546 cultures (39%); 11 of 314 positive cultures (3.5%) would have been missed.ConclusionThese results suggest that a substantial reduction in urine culture testing might be achievable by implementing this protocol. Confirmation of these findings in a validation cohort is necessary.Copyright © 2014 Elsevier Inc. All rights reserved.

33 keywords...

hide…