Symptom-based diagnosis of urinary tract infection in women: are we over-prescribing antibiotics?

Abstract

Current clinical guidelines for the management of symptoms suggesting urinary tract infection recommend empiric antibiotic therapy. To determine the diagnostic accuracy of urinary tract symptoms for early identification of urinary tract infection (UTI) in sexually active women when culture results are not available. This was a cross-sectional observational study conducted in a tertiary care hospital between July 2009 and May 2011. Subjects comprised 312 women ≥ 18 year of age who reported to the physician with symptoms suggestive of UTI. A predesigned questionnaire was filled and urine was analysed by microscopic examination and culture. Diagnostic values were calculated against gold standard urine culture results (> 10(2) CFU/ml) and 95% CIs and likelihood ratios are reported. A total of 312 women were enrolled, as culture was contaminated in 36 only 276 women were included in final analysis. Prevalence of UTI was 46.01% amongst symptomatic women. Urgency (p = 0.001), burning sensation during micturition (p = 0.035), dysuria (p = 0.004), frequency of sexual intercourse > 5 per month (p = 0.010) and pyuria (p = 0.000) were significantly associated with culture positivity. Absence of pyuria emerged as best predictor for ruling out UTI even if the woman had symptoms (sensitivity 93.70%, NPV 91.84%, AUC 77.07%, LR- 0.1). The combination of urgency, burning during micturition and pyuria was the best predictor of UTI in our study (sensitivity 85.83%, PPV 71.71%, AUC 78.48%, LR+ 2.97) Symptoms alone have low accuracy when assessed against the reference standard for diagnosing UTI. Empiric treatment of UTI based on symptoms may expose large number of patients to unnecessary antibiotics. Wet mount microscopy for presence of pyuria as a 'near patient test' before starting antibiotics seems a rational approach for management of UTI in symptomatic women.