Test characteristics of the urine Gram stain in infants

OBJECTIVE: The utility of the Gram stain for the preliminary diagnosis of urinary tract infections (UTI) in infants or= 38.0 degrees C presenting to a pediatric emergency department during 2 consecutive winter seasons. Single pathogen growth of >or= 10(4) cfu/mL from a catheterized specimen and >or= 10(3) cfu/mL from a suprapubic specimen was considered positive. A positive Gram stain was defined as the identification of any organisms. Urinalysis was tested for the presence of nitrites and for leukocyte esterase (LE). Urine microscopy was analyzed for white blood cells per high power field (WBC/hpf). RESULTS: Fourteen of 246 patients were excluded; 11 because no Gram stain was completed. Of the remaining 232 patients, Gram stain had a sensitivity of 85.2% (95% CI 71.9-98.6%), a specificity of 99.0% (95% CI 97.7-100%), a likelihood ratio for a positive test of 87.3 (95% CI 21.8-349.9), and a likelihood ratio for a negative test of 0.15 (95% CI 0.06-0.37). There were 193 specimens for which a Gram stain and a complete UA and microscopy were completed and compared. Urine microscopy had a lower sensitivity and specificity than Gram stain for both >or= 5 WBC/hpf and >or= 10 WBC/hpf. In addition to the Gram stain, a dipstick negative for nitrites and LE had a low negative likelihood ratio (0.16), useful for decreasing the likelihood of a UTI. CONCLUSIONS: The Gram stain has excellent test characteristics for the preliminary diagnosis of a UTI in febrile infants. Patient disposition and therapy will likely change if clinical protocols and guidelines use the Gram stain rather than urine microscopy for preliminary diagnosis of UTI in infants.