UTI and Faulty Gold Standards

I read with interest the article by Shaikh et al¹ on the association between pyuria and uropathogen type in children being evaluated for urinary tract infection (UTI). Their conclusion differs from our related investigation² in Pediatrics, in which we concluded that the sensitivity of the urinalysis (UA) has been previously underestimated due to a faulty gold standard, namely, urine cultures that are falsely positive because of contamination and asymptomatic bacteruria (AB). In our sample of young infants with bacteremic UTI, a condition that cannot be explained by contamination or AB, 99.5% of subjects had a UA that was positive for either leukocyte esterase or pyuria (≥3 white blood cells per high-power field).² The accompanying commentary to our article concluded that “the absence of pyuria should create great doubt about the presence of a UTI.”³

In this recent investigation, Shaikh et al¹ present data consistent with previous reports that pyuria on the UA is imperfectly sensitive, and provide new information suggesting that the sensitivity might differ by uropathogen. They report that the sensitivity of pyuria of only 90% cannot be explained by AB, because the prevalence of AB is “too low (<1%).” The actual prevalence of AB (detected via suprapubic aspirate) in the study by Wettergren et al,⁴ which the authors cite in making this claim, is 1.4%: 0.9% in girls and 2.5% in boys. Nonetheless, even a prevalence of AB as low as 1% could have a substantial impact on the apparent sensitivity of the UA. In the Shaikh et al¹ study, for example, the estimated prevalence of UTI was ∼5% (1394/26 151). If the population prevalence of AB is 1% in children, then an estimated 1 of 5 positive cultures in their sample will be falsely positive (ie, a positive urine culture with a negative UA). Therefore, if this population were similar to the population of Wettergren et al,⁴ the sensitivity of even a perfect screening test applied to this population would theoretically be ∼80%. As long as urine cultures alone are used as a gold standard to define UTI, we are unlikely to ever see UA sensitivities that approach 100%. The fact that sensitivities in this study appear to differ by pathogen also could be explained by the varying likelihoods of certain organisms to colonize the genitourinary tract and/or contaminate a urine sample.

Shaikh et al¹ also reiterate concerns that a delay in the diagnosis of UTI (as might occur if a physician is misled by a negative UA) may increase the risk of renal scarring. However, this concern is belied by data from their own meta-analysis demonstrating that fever >24 hours before diagnosis of UTI is not associated with renal scarring (odds ratio 1.11, 95% confidence interval 0.72–1.71).⁵

• Copyright © 2017 by the American Academy of Pediatrics