Probiotics as Adjuvant Treatment for Bacterial Vaginosis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Probiotics as Adjuvant Treatment for Bacterial Vaginosis (NCT03894813), a randomized phase 4 study. The primary outcome is Number of Participants With Related Clinical Signs or Symptoms at Baseline.

2. Clinical Question

3. Trial Design

The registry identifies NCT03894813 as a completed Phase 4 randomized trial with a parallel-group design and no masking. Enrollment is listed as 126. The two groups were assigned to probiotics plus vaginal metronidazole or vaginal metronidazole alone. The registry names Peking University Shenzhen Hospital as the sponsor. The recorded study start is 2019-03-26, and the primary completion date is 2020-12-30.

In the combination group, participants received oral Umeta-Miyue probiotics containing Lactobacillus rhamnosus GR-1 and Lactobacillus reuteri RC-14 once daily for 30 days, together with vaginal metronidazole suppositories once daily for 7 days. The comparison group received vaginal metronidazole suppositories once daily for 7 days. Masking is listed as none, so the registry does not describe a blinded comparison.

The registry lists five outcome measures and five analyses, but the posted primary measure is a baseline count rather than a post-treatment recurrence outcome. No secondary results are posted in the registry record. The exact randomization procedure, allocation concealment, eligibility criteria beyond the condition, recruitment setting, and analysis plan are not reported in the registry record. The registry also does not report a sample-size justification or whether any interim analyses were planned or conducted.

The record's stated clinical aim concerns prevention of recurrence, but the posted primary endpoint description concerns signs and symptoms recorded one day before starting treatment. This difference matters when interpreting what the posted comparison can answer: a baseline measure characterizes participants before the treatment period, whereas a recurrence outcome would require an outcome assessed after treatment. Details of such a recurrence analysis are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set of all randomized participants, that is 126, who had baseline samples were tested.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Related Clinical Signs or Symptoms at Baseline, measured over 1 day before starting treatment.

The primary endpoint is the number of participants with related clinical signs or symptoms at baseline, recorded one day before starting treatment. The signs and symptoms named in the registry are homogeneous and thin vaginal discharge, unpleasant odour such as a fishy smell, and vulvar discomfort such as itching or burning. The endpoint is expressed as a participant count, not as a symptom score or time-to-event measure.

A baseline measure can describe the study groups before treatment and provide context for the participants entering the trial. It is not, by itself, a measure of whether symptoms improved, whether bacterial vaginosis recurred, or when recurrence occurred. The endpoint description says that the number of participants with related clinical signs or symptoms was calculated. Both groups have a posted count of 63, with 63 participants in each group.

The endpoint population is described as the full analysis set of all randomized participants, that is 126, who had baseline samples tested. The registry does not explain how signs or symptoms were combined into the endpoint, whether a participant needed one or multiple listed signs to be counted, or how assessments were made beyond the description. Those details are not reported in the registry record. The stated trial aim is recurrence prevention, but the posted primary endpoint is explicitly a baseline count; this distinction limits the clinical question that can be answered from the displayed primary result.

7. Results Posted in the Registry

• Metronidazole Vaginal: 100.0% (63/63) versus 100.0% (63/63) on control; risk difference 0.0 percentage points (95% CI 0.0 to 0.0), risk ratio 1.00 (1.00–1.00); unadjusted two-proportion z-test p = n/a.

8. Statistical Interpretation

9. Statistical Methods Explained

The outcome is a binary participant-level measure: each participant is counted as having or not having the relevant clinical signs or symptoms at baseline. When an outcome has this form, a group count can be compared with the number of participants assessed in that group. The registry posts counts of 63 in each group and group sizes of 63. A risk difference expresses the contrast between the groups' proportions with the event. The posted unadjusted risk difference is 0.0.

What does a risk difference express?

In general, a risk difference is the event proportion in one group minus the event proportion in the comparison group. It is expressed on the probability scale, so a value of zero indicates no observed difference in the proportions for the endpoint being compared. Here, the posted endpoint is the presence of clinical signs or symptoms at baseline. The zero value therefore describes the baseline endpoint; it is not an estimate of a post-treatment recurrence contrast.

What does a confidence interval add?

In general, a confidence interval describes uncertainty around an estimated effect under the assumptions of the method used. It is not a guarantee that the true effect lies within the interval, nor does an interval containing zero prove that the groups are equivalent. The posted interval for the risk difference runs from 0.0 to 0.0. That interval accompanies the displayed recalculation, but the registry does not provide details explaining its construction or how it relates to the sponsor's reported analysis.

How should the reported test be read?

The registry lists a Chi-squared method and labels its analysis as non-inferiority, reporting p <0.01. In general, a Chi-squared test can assess whether observed categorical outcome distributions differ from what would be expected under a specified null hypothesis. The test result depends on the outcome and hypothesis being tested. The registry record does not state the exact tested contrast, the non-inferiority margin, the test direction, or the analysis details needed to reconstruct the sponsor's result. A p-value does not identify the effect size or establish clinical importance on its own.

Non-inferiority is a particular question: whether an intervention is not worse than a comparator by more than a specified margin. In general, the margin must be defined before interpreting a non-inferiority result, and the estimated effect and its uncertainty must be considered against that margin. The registry's p-value is labelled non-inferiority, but the margin and the estimate with its confidence interval are not reported in the registry record. The label alone does not allow a reader to assess the non-inferiority claim.

What do the planning illustrations represent?

In general, a sample-size calculation uses assumptions about the effect to be detected, the significance level, allocation, and statistical method. The registry's planning description says its illustrations use the observed effect, two-sided significance level 0.05, equal allocation, and a normal approximation. For non-time-to-event outcomes, the stated output is participants per arm for the selected power. These are planning illustrations rather than observed trial findings, and they are not the trial's own design justification. Numerical illustration results are not reported in the registry record. They should not be interpreted as observed power.

When a planning exercise is based on an observed effect, it is a retrospective illustration, not proof that the study was adequately powered or that a result is conclusive. In general, observed power calculated from an observed estimate does not replace examination of the confidence interval and the design assumptions. The record does not report the original sample-size rationale, target effect, or power planned for this trial.

Why does the endpoint timing matter?

Statistical methods cannot correct a mismatch between the clinical question and the time at which an outcome was measured. The primary endpoint is recorded one day before treatment. The stated clinical aim concerns recurrence prevention, which would require relevant follow-up after treatment. The registry does not report a post-treatment recurrence endpoint among the posted results. Thus, even a correctly calculated baseline comparison cannot answer the recurrence question.

General statistical interpretation also depends on the analysis population and handling of missing observations. The registry says the baseline endpoint used the full analysis set of all randomized participants, that is 126, who had baseline samples tested. It does not report the missing-data approach, any imputation, or sensitivity analyses. The implications of those omissions cannot be determined from the posted record.

10. Confidence Intervals and Precision

For the baseline count, each group has 63 participants and 63 participants with the listed signs or symptoms. The recalculated risk difference is 0.0, and its confidence interval is 0.0 to 0.0. This is a point estimate and interval for the baseline endpoint as represented in the posted summary statistics, not a precision statement about recurrence after treatment.

In general, a confidence interval conveys the range of effect values compatible with the estimate under a specified statistical method and its assumptions. A narrow interval can indicate greater precision for the quantity being estimated, while a wider interval indicates more uncertainty. An interval from 0.0 to 0.0 has no displayed width, but it should not be read as proof of certainty or as evidence that every clinically relevant outcome is identical between groups. Its meaning is limited to the recalculated baseline contrast and the method used to produce it.

The registry does not provide a sponsor-reported estimate and confidence interval for the labelled non-inferiority analysis. It also does not report the non-inferiority margin. Without those items, the precision and interpretation of the sponsor's non-inferiority analysis cannot be assessed from the registry record. A p-value of p <0.01 does not substitute for an effect estimate and its uncertainty.

Most importantly, precision for the posted baseline endpoint does not supply information about the unreported recurrence outcome. A precise estimate of a different endpoint cannot answer the clinical question of recurrence prevention. No follow-up estimate or confidence interval for recurrence is reported in the registry record.

11. Analyses Reported by the Sponsor

The sponsor's own reported analysis is a Chi-squared analysis labelled non-inferiority, with p <0.01. The registry does not give its estimate, confidence interval, tested outcome details, non-inferiority margin, or other information sufficient to reconstruct its interpretation. The posted recalculation instead describes the baseline count of participants with related signs or symptoms, giving a risk difference of 0.0 and a confidence interval from 0.0 to 0.0.

These results are not interchangeable: the posted recalculation is an unadjusted comparison based on the displayed summary statistics, whereas the registry's reported analysis is identified as a non-inferiority test. The record does not explain whether they concern the same endpoint or how the sponsor's analysis was specified. The registry marks no discrepancy, but that does not add the missing details needed to evaluate the non-inferiority analysis.

The reported p-value should therefore be attributed to the sponsor's analysis and read with the limits of the information in the record. It does not establish a conclusion about recurrence from the baseline comparison. The sponsor's test cannot be fully interpreted without the outcome definition, margin, estimate, and uncertainty, none of which are reported in the registry record.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists five outcome measures and five analyses, while no secondary results are posted. The record does not identify a complete set of hypotheses, explain how outcomes were ranked, or report a multiplicity adjustment. It is therefore not possible to assess from the posted information whether the reported analysis was selected from a larger set of analyses or how multiple comparisons were handled. These details are not reported in the registry record.

In general, multiplicity arises when researchers make several statistical comparisons or evaluate several outcomes. Repeated testing can increase the chance of obtaining a small p-value even when no corresponding effect exists. Methods such as pre-specification, a testing hierarchy, or adjustment can address particular multiplicity concerns, but the appropriate approach depends on the trial's analysis plan. The registry does not report which, if any, approach was used.

The baseline endpoint population is described as the full analysis set of all randomized participants, that is 126, who had baseline samples tested. Each arm lists 63 participants for this endpoint. The record does not report the number randomized to each treatment in a separate flow description, participant exclusions after randomization, or whether all randomized participants contributed to any post-treatment analyses. Since no secondary results are posted, the analysis populations for those outcomes cannot be assessed.

Missing-data handling is not reported in the registry record. There is no description of missing outcome observations, exclusions, imputation, or sensitivity analyses. The fact that the baseline description refers to participants who had baseline samples tested does not establish how missing data were handled at other time points or for other outcomes.

The record also does not report an analysis plan, interim analyses, or the sample-size justification. Those omissions leave uncertainty about pre-specification and the intended analysis population. They should not be filled in by assumptions. The posted baseline comparison can be described as given, but the broader confirmatory analysis strategy and the status of unposted outcomes cannot be determined.

13. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry lists serious adverse events for 63 participants at risk in each group, with 0 affected in each group. Deaths are also listed as 0 in both groups. These are the serious-event and death figures in the posted record; no serious adverse-event terms are listed.

For other adverse events, the registry lists 5 affected participants in the probiotics-and-metronidazole group and 6 in the metronidazole-vaginal group, with 63 participants at risk in each. The registry does not give event descriptions for these other adverse events in the registry record for this analysis. The counts should not be interpreted as a measure of severity or as proof that an event was caused by a treatment.

In general, adverse-event counts describe events recorded during a specified safety observation period and among the participants included in the safety assessment. The registry does not report the observation period, event definitions, attribution rules, or details of collection. Without those details, the counts cannot establish the timing, duration, seriousness of non-serious events, or comparative safety beyond the posted figures. Zero recorded serious events does not prove that no serious event could occur; it states only what the registry reports for this trial.

14. What the Result Does, and Does Not, Mean

The posted data support a narrow descriptive statement: at baseline, the registry lists 63 participants with related clinical signs or symptoms in each treatment group, with 63 participants in each group. The recalculated risk difference for this endpoint is 0.0, with a confidence interval from 0.0 to 0.0. This is a comparison of a baseline count, not a treatment outcome.

The trial's stated aim concerns whether adding oral probiotics to vaginal metronidazole is better than vaginal metronidazole alone in preventing bacterial vaginosis recurrence. The posted primary measure, however, was assessed one day before treatment began. It cannot show whether recurrence was prevented, whether symptoms changed after treatment, or whether the treatments differ on those outcomes. Recurrence results are not reported in the registry record.

The registry's non-inferiority-labelled Chi-squared analysis reports p <0.01. The record does not report the margin, estimate, confidence interval, or other details required to interpret that analysis as evidence about non-inferiority. A p-value alone does not show how large an effect is or whether it is clinically meaningful. Nor does it resolve the difference between the stated clinical aim and the timing of the posted primary endpoint.

The safety figures document no serious adverse events and no deaths in either group in the registry record, alongside other adverse-event counts of 5 and 6. Those numbers do not establish that treatment caused or prevented events, and the record does not provide the event details or observation period needed for a fuller safety comparison.

Accordingly, the result should be understood as a baseline description with limited information about the sponsor's reported test. It does not establish a treatment effect on recurrence or a broad conclusion about comparative safety. Those questions would require interpretable post-treatment outcomes, their analysis details, and appropriate uncertainty estimates.

15. Why This Trial Matters Statistically

This trial illustrates why the definition and timing of an endpoint are central to statistical interpretation. A study can state a clinically important question—in this case, prevention of recurrence—while the posted primary endpoint measures a characteristic before treatment. The baseline measure can describe the groups, but it does not answer whether an outcome occurred after treatment.

It also illustrates the distinction between an estimate and a test label. The registry gives a recalculated risk difference for the baseline endpoint and separately lists a Chi-squared analysis labelled non-inferiority. A non-inferiority interpretation ordinarily depends on an explicit margin and an estimate with uncertainty. Those details are not reported in the registry record, so the label and p-value cannot independently establish what was demonstrated.

Precision must be tied to the quantity being estimated. The interval from 0.0 to 0.0 describes the recalculated baseline comparison as posted; it does not extend to unreported recurrence outcomes. Readers should not mistake a precise description of one endpoint for evidence about another.

The record also shows why transparent reporting of analysis populations, missing-data methods, outcome definitions, and multiplicity plans matters. Without those details, readers cannot fully evaluate how the sponsor's analysis was conducted or how the posted result relates to the broader study aim. These are not merely technical details: they determine what claim a statistical result can support.

The statistical lesson is to match each conclusion to the observed endpoint, its timing, the analysis, and the uncertainty reported. Here, the baseline count and safety figures are interpretable as recorded counts, while recurrence prevention and the meaning of the non-inferiority test remain unresolved by the available registry information.

16. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The most important trial-specific limitation in the posted results is the mismatch between the stated objective and the primary endpoint. The objective concerns preventing recurrence, but the primary endpoint counts signs or symptoms at baseline, one day before treatment. No recurrence result is reported in the registry record, so the primary result cannot directly address the stated objective.

The non-inferiority analysis is difficult to assess because the registry reports only a Chi-squared method, a non-inferiority label, and p <0.01. The outcome tested, non-inferiority margin, estimate, confidence interval, and analysis details are not reported in the registry record. The baseline risk-difference calculation cannot fill these gaps or be assumed to represent the sponsor's analysis.

The endpoint description names several symptoms but does not state how they were combined into a participant-level count, how assessments were conducted, or how symptoms were verified. The record does not report missing-data methods, post-randomization exclusions, or analysis populations for other outcomes. Although five outcome measures and five analyses are listed, no secondary results are posted.

There is no reported sample-size justification, full statistical plan, multiplicity approach, or interim-analysis information. The registry describes planning illustrations based on the observed effect, but numerical illustration results are not reported in the registry record, and those illustrations are not the trial's design rationale.

Safety reporting is also limited: serious-event and death counts are listed, but the observation period and event-collection and attribution details are not reported in the registry record. These limitations do not change the posted counts; they restrict how broadly those counts and analyses can be interpreted.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Probiotics as Adjuvant Treatment for Bacterial Vaginosis — NCT03894813.
• Registry record last updated 2021-07-09; status last verified 2021-06.
• Results first posted 2021-07-09.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

20. Record Summary

NCT03894813 is a completed, randomized, unmasked, parallel-group Phase 4 trial sponsored by Peking University Shenzhen Hospital. It enrolled 126 participants with bacterial vaginosis. The study compared oral probiotics plus vaginal metronidazole with vaginal metronidazole alone. The combination regimen included probiotics taken once daily for 30 days and vaginal metronidazole once daily for 7 days; the comparator regimen used vaginal metronidazole once daily for 7 days.

The posted primary endpoint counts participants with related signs or symptoms at baseline, one day before treatment. The count is 63 in each group, and the recalculated risk difference is 0.0 with a confidence interval from 0.0 to 0.0. The registry separately reports a Chi-squared analysis labelled non-inferiority with p <0.01, but does not report the margin, estimate, or confidence interval for that analysis. No secondary results are posted. Serious adverse events and deaths are listed as 0 in each group; other adverse-event counts are 5 and 6. Recurrence results are not reported in the registry record.