Discordance Between Observed vs Expected Prevalence of UTI in Newborns With Hyperbilirubinemia

Multiple studies in which all jaundiced newborns routinely underwent urine testing found nontrivial prevalences (1.9–7.7%) of urinary tract infections (UTIs); however, because this practice is uncommon in the United States,1 a lack of testing may lead to delayed UTI diagnoses. Alternatively, routine testing may drive UTI misdiagnosis or overdiagnosis. To better understand the clinical significance of the reported UTI prevalence in jaundiced infants, we examined the prevalence of UTI in newborns treated with phototherapy and the incidence of delayed UTI in newborns who did not undergo initial urine testing at the time of phototherapy.This retrospective study used a preexisting database of 1- to 7-day-old newborns born at least 35 weeks’ gestation at 16 Kaiser Permanente Northern California (KPNC) hospitals who received their first home or inpatient phototherapy between January 1, 2013, and December 31, 2018.2 To focus on newborns asymptomatic at the time of phototherapy initiation, we excluded newborns who had temperatures under 36°C or over 38°C within 12 hours before phototherapy initiation, blood cultures collected within 12 hours before urine collection, or urine studies obtained over 12 hours before phototherapy. We quantified urine testing within 12 hours before or after phototherapy initiation, and the prevalence of concurrent UTI (defined as positive urinalysis [UA] with leukocyte esterase, nitrite, or ≥5 white blood cell/high-powered field, plus urine culture with ≥50 000 CFU/mL by catheterization or ≥100 000 colony-forming units [CFU]/mL by bag or clean-catch of a single uropathogen). We also quantified the incidence of delayed UTI (diagnosed between 12 hours and 7 days after phototherapy initiation) in newborns without initial urine testing. This study was approved by the KPNC institutional review board.Of the 15 749 included newborns, the mean (±SD) age at phototherapy initiation was 2.9 (±1.2) days. Eighteen newborns (0.1%) had urine studies within 12 hours of phototherapy initiation, of whom 2 (0.01% of the cohort) met UTI criteria. Of the 15 731 newborns without initial urine studies, 48 (0.3%) had urine studies between 12 hours to 7 days after phototherapy initiation, and 3 met UTI criteria. Figure 1 illustrates the number of included and excluded newborns as well as urine testing results. A lower culture threshold of at least 10 000 CFU/mL by catheterization yielded no additional cases of concurrent or delayed UTI.In newborns 1 to 7 days old who received phototherapy in this large health care system, urine testing was rare, and UTIs were rarely diagnosed (0.01%; 2 cases). Based on the previously reported UTI prevalence of 1.9 to 7.7%,3 if all newborns in our cohort had been tested, we would have expected 299 to 1213 UTIs. Therefore, we quantified the incidence of “delayed UTIs,” ie, missed UTIs in newborns without initial urine testing. We found a low frequency of delayed urine testing (0.3%), suggesting that most newborns did not later present clinically with signs or symptoms concerning UTI, and a low incidence of delayed UTIs (0.02%; 3 cases).This discordance between the frequency of UTI diagnoses, when every newborn undergoes urine testing vs when testing is rare, illustrates fundamental challenges in establishing a rigorous diagnostic “gold standard” for this common infection. Our results suggest that prior UTI prevalence estimates in the jaundiced newborn population are likely inflated by misdiagnosis and/or overdiagnosis–cases due to contamination,4 asymptomatic bacteriuria,4 and/or infections that met laboratory criteria of UTI but would have self-resolved.5 These findings may have implications for reported UTI prevalence in other conditions such as febrile neonates who undergo urine testing routinely.One study limitation is that our cohort ended in 2018, although this should not influence the discordance found between observed and expected UTI prevalence. Although our cohort received phototherapy per the 2004 American Academy of Pediatrics’ thresholds, this was also true for 2 comparison studies.6,7 Second, we focus on newborns 7 days old or younger, a slightly younger age range than in previous studies (≤10–14 days); however, in those studies, the mean ages were all under 7 days,6–10 and in one study (UA+ UTI prevalence 3.9%),8 all UTI cases were under 7 days. In addition, by exclusion criteria, most newborns in the comparison studies and in our cohort likely had breastfeeding jaundice (Table 1).We conclude that at least among well-appearing jaundiced newborns 7 days old or younger, urine testing leads to substantial misdiagnosis or overdiagnosis of UTIs. Future studies should re-assess the UTI prevalence in older infants with prolonged jaundice.