Abstract
BACKGROUND: Evidence suggests low diagnostic test utilization and high empiric prescribing rates for vaginitis symptoms. This study retrospectively assessed real-world bacterial vaginosis (BV), vulvovaginal candidiasis (VVC), and trichomoniasis (TV) treatment patterns in the United States relative to test results and their timing. METHODS: Patients with vaginitis or related symptom diagnosis were identified between 2018 and 2023 in an ambulatory electronic medical record database linked to prescription and outpatient medical claims databases. Pregnant and nonpregnant subgroups were stratified by test type (nucleic acid amplification test [NAAT] panel [BV, VVC, and TV], direct probe, or traditional methods) up to first diagnosis date +2 days. Treatment claims before (empiric) and up to 7 days after (informed) the test result date were assessed. RESULTS: There were 1465 pregnant and 13,447 nonpregnant patients meeting selection criteria; 31.7%, 26.3%, and 2.0% of pregnant and 26.6%, 16.1%, and 2.0% of nonpregnant cohorts had positive results for BV, VVC, and TV, respectively. Empiric treatment was common across treated pregnant subgroups tested with laboratory NAAT panel (55.0%), direct probe (35.5%), and traditional methods (66.7%), and nonpregnant subgroups (55.7%, 52.2%, 74.2%, respectively). Many empiric treatments were not aligned to positive test results, including coinfections (e.g., 6.5%-8.2% of pregnant and 11.7%-13.0% of nonpregnant patients who tested BV positive by NAAT panel or direct probe were empirically treated with antifungal agents for VVC). CONCLUSIONS: High empiric treatment rates and evidence of inappropriate treatment highlight the potential utility for rapid, accurate tests to diagnose common vaginal infections, to inform clinical decision making and results-guided prescribing at the point of care.