No Hard Cutoffs – Interval Likelihood Ratios for Pediatric UTI

January 26, 2021

Written by Clay Smith

Spoon Feed
For pediatric UTI, different urinalysis (UA) and urine microscopy cutoffs more or less strongly predicted infection. Use of interval likelihood ratios improves accuracy.

Why does this matter?
We often consider the UA and micro exam using arbitrary cutoffs for leukocyte esterase (LE), WBCs, or bacteria. In reality, it’s a continuum. What if we considered varying cutoffs to rule in or out UTI? Consider a patient with trace LE and 5-10 WBCs; the probability they have true infection is only around 10-15%. In the right setting, one might wait for the urine culture (UCx) before starting antibiotics. This could improve antimicrobial stewardship.

Blurring the cutoffs
This was a retrospective, single center study of 2,144 patients under age 2 who had a UA and UCx. Most patients (85%) in this study were black. Prevalence of UTI was 9.2%. Rather than looking at arbitrary cutoffs, they looked at the likelihood ratios for various ranges of leukocyte esterase, hemoglobin, protein, nitrite (this was dichotomous + or -); WBCs, RBCs and bacterial load on microscopy. Key findings are summarized in two tables. Key take points for the UA: 3+ LE or + nitrite were both powerful positive discriminators of a UTI. Negative LE and nitrite were both good negative predictors. For urine micro: 20-50 WBCs or “many” bacteria were powerful positive discriminators of a UTI. On micro, 0-5 WBCs and negative bacteria were both good negative discriminators.

Source
Predicting Urinary Tract Infections With Interval Likelihood Ratios. Pediatrics. 2020 Dec 4;e2020015008. doi: 10.1542/peds.2020-015008. Online ahead of print.