[A decision analysis model for appropriate laboratory use--should test of urine gram stain be done for selecting antibiotics at the initial visit on women with acute dysuria].

We used a decision analysis model to estimate the clinical and economic implications of testing the gram-stained smears to select the effective antibiotics at the initial visit on women with lower urinary tract infections. A strategy of the initial therapy of antibiotics without routine testing was compared with the initial testing of urine gram stain and selecting antibiotics according to the results. Varying assumptions were used for prevalence of etiologic agents, operating characteristics of gram stain test, treatment efficacy and required days for microbiological tests. We concluded that from the view point of expected direct costs and expected care days, a 90% pretest probability of pathogens to be gram negative rods was the critical point to decide to do or not to do urine gram stain for selection of antibiotics. Sensitivity analysis showed that the gram stain testing strategy was cost effective when the frequency of gram positive coccus infection would be increased or the treatment efficacy after antibiotics therapy would be decreased.