Clinical usefulness of urine Gram stain for diagnosing urinary tract infections at the emergency department.

Abstract

Background: Diagnosis of urinary tract infections (UTIs) is a frequent challenge at the emergency department (ED). The clinical usefulness of the urine Gram stain (GS) is uncertain.

Objective: We studied the GS performance to clarify its clinical utility at the ED.

Methods: Urine dipstick (UD), automated urinalysis (UF-1000i), GS and urine culture (UC) were performed in a cohort of consecutive adults presenting at the ED suspected of a UTI. GS performance was assessed and compared to UD and UF-1000i.

Results: A UTI diagnosis was established in 487/1358 (35.9%) episodes. Sensitivity and specificity for 'many' GS leucocytes was 33.7% and 95.4%; for 'many' GS bacteria 51.3% and 91.0%. GS diagnostic performance by ROC analysis was 0.796 for leucocytes and 0.823 for bacteria. GS bacteria performed better than UD nitrite comparable to UF-1000i bacteria. GS leucocytes underperformed compared to UD leucocyte esterase and UF-1000i leucocytes. UC was positive in 455 episodes. GS correctly predicted urine culture of gram-negative rods (PPV 84.6%). Prediction was poor for gram-positive bacteria (PPV 38.4% (cocci), 1.0% (rods)).

Conclusion: With the exception of a moderate prediction of gram-negative bacteria in the UC, urine GS does not improve UTI diagnosis at the ED compared to other urine parameters.