Do Urinalysis Test Characteristics Differ by Method of Urine Collection?

Alan R Schroeder ¹ , Thomas B Newman ¹ , Stacia A Finch ² , Richard C Wasserman ^2,3 and Robert H Pantell ¹ . ¹ Pediatrics, UCSF, San Francisco, CA ; ² Pediatric Research in the Office Setting (PROS), Dept of Practice & Research, Ctr for Child Health Research, AAP, Elk Grove Village, IL and ³ Dept of Peds, Univ of VT, Burlington, VT.

Background: Urinary Tract Infection (UTI) is a common source of fever in young infants. Urinalysis (UA) is often done as a screen for UTI. While there are substantial data on the diagnostic test characteristics of UA in predicting UTI, few studies have compared the performance of UA between urethral catheterization (CATH) and sterile bag methods.

Objective: To compare diagnostic test characteristics of UA between CATH and bag methods.

Design/Methods: We used data from the Pediatric Research in Office Setting (PROS) Febrile Infant Study, an observational study of 3066 infants aged 0-3 months with temperatures 38 C. Diagnostic and management decisions were left to the discretion of individual practitioners. UTI was defined as 100K cfu/mL of a single pathogenic organism for bag and 20K cfu/mL of a single pathogenic organism for CATH. We compared sensitivity and specificity of leukocyte esterase (LE) and nitrites, and likelihood ratios and ROC curves for urine white blood cells (WBC) between CATH and bag methods.

Results: Urine cultures were obtained on presentation in 1605 infants (52%); 1639 (53%) had a UA performed. The UAs were obtained by CATH in 1030 of 1639 infants (63%) . For LE and nitrites, sensitivity and specificty were higher in UAs obtained by CATH (Table 1). For urine WBCs, the area under the ROC curve was higher in UAs obtained by CATH than in those obtained by bag (ROC area = 0.86 vs 0.71, p = .01). These differences persisted even when the threshold colony counts for UTI were modified.

Conclusions: Urinalyses obtained by CATH appear to perform better than those obtained by bag. Possible explanations for this include: (1) the UA actually performs better as a screening test when obtained by CATH, (2) the urine culture definition of UTI is less accurate for bag specimens, and (3) random error. Further studies comparing simultaneous CATH and bag urinalyses would help to address this issue.

TABLE 1: Sensitivity and specificity of LE and nitrites, stratified by UA collection method