Microscopy-based classification reduces overdiagnosis of acute vulvovaginal candidiasis relative to culture alone in pregnant women

Vulvovaginal candidiasis (VVC) is commonly diagnosed using fungal culture, which cannot distinguish colonisation from infection. This study evaluated discordance between culture and a microscopy-based definition of acute VVC, assessed microscopy performance against culture, and MALDI-ToF species identification. We conducted a cross-sectional secondary analysis of microscopy and culture data from 154 vaginal swab specimens from symptomatic pregnant women in Uganda, recruited between March and June 2023. A five-category microscopy-based definition informed by BASHH 2019 guidelines was applied. We calculated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) using culture as the reference method. Agreement between microscopy phenotype and MALDI-ToF species identification was assessed using Cohen’s kappa. Culture was positive in 103/154 (66.9%) women, whereas microscopy identified 44 (28.6%) as likely Candida albicans infection and 16 (10.4%) as possible non- albicans infection. Based on microscopy, 45.6% of culture-positive cases lacked evidence of active infection (95% CI 35.9–55.7%). Microscopy showed sensitivity of 54.4% (95% CI 44.2–64.2%), specificity of 92.2% (81.1–97.8%), PPV of 93.3% (83.8–98.2%), and NPV of 50.0% (39.5–60.5%). Culture-positive but microscopy-negative cases had significantly lower colony counts ( p = 0.006). All participants were symptomatic, with no difference in symptom burden across microscopy categories ( p = 0.284). Agreement between microscopy phenotype and MALDI-ToF species identification was moderate (κ = 0.454, p < 0.001). Microscopy-based classification reduces apparent overestimation of VVC associated with culture alone while demonstrating high specificity and PPV. Prioritizing microscopy with selective culture may improve diagnostic precision and support antifungal stewardship in low-resource antenatal settings, although this approach requires prospective validation.