Interpretation and comparison of treatment studies for uncomplicated urinary tract infections in women

Abstract

Recent treatment trials for uncomplicated urinary tract infection were reviewed to determine most commonly used methodologic approaches and to assess whether methodologic problems often compromise internal validity and comparability of these studies. The 62 studies surveyed fulfilled an average of 56% (standard deviation, 19.2%) of 12 standards necessary for accurate interpretation and comparability. Standards most often met were: reporting the incidence of adverse-effects (90%); recording presence or absence of pretreatment symptoms (81%); and describing clinical response to therapy (71%). Standards least often met were: having adequate statistical power to detect a meaningful difference between therapies (21%); double-blinded assigning of treatment regimens (37%); and clearly defining criteria for diagnosing cure and failure (35%). Two recurring characteristics that impaired both interpretation of individual studies and comparisons between studies were failure to separately randomize or stratify patients with risk factors known to adversely affect therapeutic response and inadequate description of outcome measures. Consideration of these factors and use of a proposed system for classifying therapeutic outcome in future treatment trials would improve the basis for clinical decision making.