Remote BV Management via Metagenomic Vaginal Microbiome Testing and Telemedicine
A fully remote protocol combining shotgun metagenomics and telemedicine for bacterial vaginosis produced self-reported symptom improvement in 84% of patients and significant reduction in BV-associated taxa, but the absence of a control group precludes causal attribution.
| Population | Adults with vaginal symptoms and dysbiotic vaginal microbiome confirmed by shotgun metagenomics, diagnosed with BV via telehealth platform (Nov 2022–Jul 2024); excluded: pregnant, immunocompromised, diabetic, HIV-positive, active cancer, or untreated STI |
|---|---|
| Exposure | Personalized remote protocol: metronidazole or clindamycin selected by metagenomic profile plus adjuvants (boric acid, vaginal estrogen, probiotics, prophylactic antifungals) via Evvy telehealth platform |
| Comparator | None (single-arm study, no control or active comparator) |
| Outcome | Self-reported symptom response; Aggregate symptom score (0–42); Relative abundance of Lactobacillus; Relative abundance of Gardnerella; Relative abundance of Prevotella; BV recurrence rate; Relative abundance of Fannyhessea |
Summary of findings
| Outcome | Effect | 95% CI | Certainty | Clinical relevance | Notes |
|---|---|---|---|---|---|
| Self-reported symptom response | 84% responders (178/212); in the RR/OR reported; in the 95% CI | — | Low | — | 1 studies |
| Aggregate symptom score (0–42) | MD -5.0 (8.9 to 3.9); in the 95% CI reported | — | Low | — | 1 studies |
| Relative abundance of Lactobacillus | mean 26% to 55%, p<0.001; in the 95% CI reported | — | Low | — | 1 studies |
| Relative abundance of Gardnerella | mean 33% to 16%, p<0.001; in the 95% CI reported | — | Low | — | 1 studies |
| Relative abundance of Prevotella | mean 11% to 6%, p<0.001; in the 95% CI reported | — | Low | — | 1 studies |
| BV recurrence rate | 19% recurrence at mean 4.4mo follow-up; in the comparator, in the 95% CI | — | Low | — | 1 studies |
| Relative abundance of Fannyhessea | mean 5% to 2%, p<0.001; in the 95% CI reported | — | Low | — | 1 studies |
Context
BV affects ~30% of individuals with vaginas annually and recurs in >50% within 6 months on standard antibiotics. Limited access to in-person gynecologic care motivates telehealth models. This study evaluates whether self-collected vaginal metagenomics integrated with telemedicine constitutes a clinically valid BV care pathway.
What the study showed
84% of patients (n=178/212) reported significant or moderate symptom improvement. Mean symptom score decreased from 8.9 to 3.9 (absolute reduction 5.0 points; ~56% relative reduction). Mean Lactobacillus relative abundance increased from 26% to 55% (p<0.001); Gardnerella decreased from 33% to 16% (p<0.001); Prevotella from 11% to 6% (p<0.001); Fannyhessea from 5% to 2% (p<0.001). Platform-captured BV recurrence rate was 19% at mean 4.4 months follow-up.
How it was done
Single-arm observational study (retrospective-prospective), n=212 patients. Standardized self-collected vaginal swabs; shotgun metagenomic sequencing (Illumina NovaSeq 600, CLIA/CAP-certified pipeline). Pre- and post-treatment assessment via symptom questionnaires and metagenomic retesting within 1 year. Mean follow-up 4.4 months for symptom response.
Effect magnitude
Absolute reduction of 5.0 points in symptom score (8.9→3.9); absolute increase of ~29 percentage points in Lactobacillus (26%→55%). 95% CIs not reported for most primary outcomes.
Risk of bias
No control group (RoB 2 not formally applied). Primary outcome is self-reported, susceptible to expectation/placebo bias. BV diagnosis based on symptoms plus metagenomics without validated Amsel or Nugent criteria. Single commercial platform (Evvy) limits generalizability. Recurrence captured mainly within-platform; external antibiotic use may be underestimated.
What this study does NOT prove
This study does not prove that shotgun metagenomics is superior to standard clinical diagnosis, nor that the telehealth protocol reduces recurrence compared to conventional treatment. Results are not generalizable to excluded populations (pregnant, immunocompromised, diabetic individuals).
In clinical practice
The remote model demonstrates operational feasibility for BV diagnosis and treatment in populations lacking access to in-person care. Clinicians should interpret the 84% response rate cautiously: without a control arm, part of the improvement may reflect spontaneous resolution or placebo effect. Shotgun metagenomics adds taxonomic granularity, but its cost-effectiveness over standard clinical criteria has not been established in an RCT.
Limitations
No control group (RoB 2 not formally applied). Primary outcome is self-reported, susceptible to expectation/placebo bias. BV diagnosis based on symptoms plus metagenomics without validated Amsel or Nugent criteria. Single commercial platform (Evvy) limits generalizability. Recurrence captured mainly within-platform; external antibiotic use may be underestimated.
What is still missing
An RCT comparing metagenomics-guided protocols versus standard empirical treatment is needed to establish causality and superiority. Recurrence data at 6–12 months with a control group are absent.
Technical appendix
Version history
- 1.0 · 2026-06-21 — Auto-generated under Evidence Standard v1.0