Rwanda Vaginal Microbiota Restoration Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Rwanda Vaginal Microbiota Restoration Study (NCT02459665), a randomized na study. The primary outcome is Bacterial Vaginosis (BV) Incidence by Modified Amsel Criteria.

2. Clinical Question

3. Trial Design

The Rwanda Vaginal Microbiota Restoration Study was a randomized, single-masked, parallel-group pilot clinical trial. It was listed as phase NA and completed. The registry records an enrollment of 68 participants. The study began on 2015-06-05, with primary completion on 2016-02-23. Its stated purpose was preparation for a clinical trial of interventions to maintain normal vaginal microbiota and prevent adverse reproductive health outcomes in Africa. The brief summary describes intermittent use of two vaginal lactobacilli-containing probiotics and oral metronidazole to prevent bacterial vaginosis recurrence.

There were four groups. The negative control group received no intervention after initial treatment for bacterial vaginosis or trichomoniasis. The positive control group received metronidazole pills (500 mg) twice per week for 2 months. One probiotic group used Ecologic Femi+ vaginal capsules daily for 5 days immediately after initial treatment, followed by use three times weekly for two months. The other used a Gynophilus LP vaginal tablet once every 4 days for two months. The study record describes random allocation, single masking, and a parallel model; further details of the randomization sequence or masking procedures are not reported in the registry record.

The primary outcome covers a 2-month intervention period and is bacterial vaginosis incidence by modified Amsel criteria, expressed as events per person-year at risk. The specified analysis population excluded women with Nugent 7-10 at enrollment from the randomized women. Arm-specific outcome counts were 15 for control, 11 for oral metronidazole, 14 for Ecologic Femi+, and 11 for Gynophilus LP. The registry lists one secondary outcome: vaginal microbiota composition by Illumina HiSeq sequencing, specifically Lactobacillus genus concentration. The registry record does not report the trial’s original sample-size rationale, protocol-level analysis plan, or further masking and allocation procedures.

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.

Before assignment: At the end of the screening process, women were either ineligible or diagnosed with bacterial vaginosis (by modified Amsel criteria and/or Nugent score) and/or trichomoniasis (by culture or wet mount) and treated with 500 mg oral metronidazole 2x/day for 7 days. An enrollment visit was scheduled within three days after treatment completion.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Modified ITT population: all randomized women except women with Nugent 7-10 at enrollment.

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 Bacterial Vaginosis (BV) Incidence by Modified Amsel Criteria, measured over 2 months (intervention period).

The primary endpoint is bacterial vaginosis incidence measured using modified Amsel criteria. A participant met the endpoint definition when at least 2 of 3 findings were positive: clue cells, vaginal pH, and the whiff test. Rather than reporting only the proportion of participants with an event, the registry expresses the outcome as events per person-year at risk. This rate formulation uses the amount of time participants were considered at risk as the denominator and is intended to represent how frequently events occurred over exposure time.

The registry specifies a 2-month intervention period and a modified intent-to-treat population. That population includes randomized women except those whose Nugent score was 7-10 at enrollment. The registry does not give additional detail about how individual follow-up time was determined or how repeated events, if any, were handled. Such details matter for interpreting a rate, because its denominator is person-time rather than simply the number of participants.

Incidence rates are useful when exposure time contributes to the interpretation of event occurrence. They do not, by themselves, describe every participant’s experience or the number of people who had at least one event. The reported rate and confidence interval should therefore be read together with the endpoint definition and the specified population. The registry lists vaginal microbiota composition by Illumina HiSeq sequencing as a secondary outcome, but no secondary outcome results are included in the posted outcome details.

7. Results Posted in the Registry

• Oral Metronidazole: 2.04 versus 7.53 (Incidence: events per PY at risk) on control; rate ratio 0.27 (95% CI 0.08–0.96, derived from the posted confidence limits); p ≈ 0.043.
• Ecologic Femi+: 3.36 versus 7.53 (Incidence: events per PY at risk) on control; rate ratio 0.45 (95% CI 0.17–1.19, derived from the posted confidence limits); p ≈ 0.106.
• Gynophilus LP: 3.35 versus 7.53 (Incidence: events per PY at risk) on control; rate ratio 0.44 (95% CI 0.14–1.38, derived from the posted confidence limits); p ≈ 0.160.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is an event rate: events per PY at risk. Person-years at risk represent the accumulated time participants contribute while considered at risk for the event. A rate uses this time as its denominator, so it is different from a simple event proportion. It expresses event frequency relative to exposure time. General statistical interpretation: a rate can help account for different amounts of follow-up, but its meaning depends on how the study defines and measures time at risk. The registry identifies person-years at risk as the denominator but does not report further operational details about follow-up time.

The comparison measure is a rate ratio. It divides the estimated rate in an intervention group by the estimated rate in the control group. A ratio of 1 corresponds to equal rates; a ratio below 1 indicates a lower estimated rate in the intervention group, and a ratio above 1 indicates a higher estimated rate. General statistical interpretation: the ratio describes the relative contrast, while the underlying rates show the event frequency in each group. Both are useful, and neither alone communicates all uncertainty or clinical context.

The posted comparisons are rate ratios with 95% confidence intervals. General statistical interpretation: a confidence interval gives a range of values compatible with the data under the method’s assumptions. It is not a guarantee that the true value lies in the particular interval, and it does not include uncertainty from every possible source, such as measurement or design limitations. The range should be considered alongside the point estimate. Here, the oral metronidazole interval is 0.077 to 0.958; the Ecologic Femi+ interval is 0.168 to 1.189; and the Gynophilus LP interval is 0.144 to 1.377. The probiotic intervals include 1, the value representing equal rates.

The outcome details indicate Poisson-based comparisons. General statistical explanation: a Poisson model is commonly used for counts of events, often with exposure time incorporated so that rates can be compared. Such methods rely on assumptions about the event-count process. Over-dispersion means that event counts vary more than the model expects; if present and not handled, uncertainty estimates can be too narrow. The registry record does not report a diagnostic for over-dispersion, an adjustment for it, or further details of the model. These are general considerations, not claims that over-dispersion occurred in this trial.

The registry also posts unadjusted p-values: 0.043 for oral metronidazole, 0.106 for Ecologic Femi+, and 0.16 for Gynophilus LP. General statistical explanation: a p-value summarizes how unusual data at least as extreme as those observed would be under a specified null hypothesis and model. It does not measure the size or importance of an effect, nor the probability that the null hypothesis is true. Because these values are unadjusted, they do not account for multiplicity. The registry record does not report a prespecified significance threshold or a multiplicity adjustment for the primary comparisons.

The primary population is described as modified intent-to-treat: all randomized women except those with Nugent 7-10 at enrollment. This definition is important because the estimates apply to that population, not automatically to every randomized participant. The arm-specific primary outcome counts are 15, 11, 14, and 11. The registry does not provide a participant flow account explaining exclusions or the difference between these counts and the overall enrollment.

Planning calculations use a stated effect fraction and are illustrations based on the observed effect, not the trial’s original design assumptions. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect. General statistical explanation: sample-size planning balances the effect one aims to detect, variability, significance level, and desired power. The trial’s own original sample-size justification and planning assumptions are not reported in the registry record. The planning figures should not be interpreted as observed power or as evidence that the trial was adequately or inadequately designed.

The registry labels the study randomized, single-masked, and parallel-group, but does not report the randomization method, concealment procedures, identity of the masked parties, or masking success. It lists one secondary outcome but does not include secondary outcome results in the registry entries. The detailed statistical analysis plan, handling of missing observations, and interim analyses are not reported in the registry record.

10. Confidence Intervals and Precision

The primary rate estimates have broad 95% confidence intervals. For control, the incidence was 7.53 events per PY at risk, with a confidence interval from 4.28 to 13.26. For oral metronidazole, the rate was 2.04, with an interval from 0.66 to 6.31. Ecologic Femi+ had a rate of 3.36, with an interval from 1.51 to 7.48, and Gynophilus LP had a rate of 3.35, with an interval from 1.26 to 8.92. The width of these ranges indicates uncertainty about the underlying rates, not only uncertainty about the comparisons.

The rate-ratio intervals also allow a wide range of relative effects. Oral metronidazole’s interval is 0.077 to 0.958; Ecologic Femi+’s is 0.168 to 1.189; and Gynophilus LP’s is 0.144 to 1.377. The probiotic intervals include 1, so the posted estimates are compatible with equal rates as well as with rate ratios below 1. The oral metronidazole interval does not include 1 under the reported method, but it remains wide and its upper endpoint is close to 1. The interval is not a measure of certainty about the study’s conduct, endpoint validity, or generalizability.

Small outcome-specific arm counts—15, 11, 14, and 11—are relevant to precision. The registry does not explain the differences between these counts and the overall enrollment of 68. General statistical explanation: estimates based on limited data often have more uncertainty, especially when events are uncommon or exposure time is limited. The actual event counts and person-time by arm are not reported in the record’s posted outcome details, so the intervals are especially important when reading the point estimates. They show that the apparent differences should not be treated as precise measurements.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are per arm for these comparisons; they are not total event counts. They are based on the observed effect and illustrate how the required number of participants per arm changes when the target effect is set to 100%, 75%, or 50% of that observed effect. For oral metronidazole, planning for 100% of the observed effect gives 18 participants per arm for 80% power and 24 participants per arm for 90% power; planning for 75% gives 24 and 32; planning for 50% gives 43 and 58. For Ecologic Femi+, the corresponding pairs are 32 and 42, 49 and 65, and 96 and 129. For Gynophilus LP, they are 31 and 42, 49 and 65, and 96 and 128.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. In plain terms, the 80% power column gives the participants per arm estimated to yield an 80% chance of detecting the specified effect under those assumptions if that effect were the true effect. The 90% column gives the corresponding estimate for a 90% chance. These are planning probabilities, not probabilities that a result is true.

For every comparison, planning for a smaller fraction of the observed effect requires more participants per arm. These figures are illustrations calculated from the observed estimates; they are not observed power and do not state the original trial’s intended enrollment target. The trial’s own sample-size justification and design assumptions are not reported in the registry record.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, λ₀ and λ₁ are event rates per patient-year, and T is follow-up per participant in years. This assumes Poisson counts and equal follow-up.

Worked calculation for this trial

1. Control rate λ₀ = 7.5300; treatment rate λ₁ = 2.0400 events per patient-year; follow-up T = 0.17 years.
2. n = (1.96 + 0.842)² × (1/7.5300 + 1/2.0400) / (0.17 × [ln(0.271)]²) = 17.2, rounded up to 18 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
lambda0 <- 7.53                      # control rate, events per patient-year
lambda1 <- 2.04                      # treatment rate
tt      <- 0.166666666666667                       # follow-up per participant, years
n_per_arm <- function(power, frac = 1) {
  l1 <- lambda0 * (lambda1 / lambda0)^frac   # smaller effect = fraction of the observed log rate ratio
  ceiling(round((z_a + qnorm(power))^2 * (1 / lambda0 + 1 / l1) / (tt * log(l1 / lambda0)^2), 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Simulation 1: Poisson counts. Simulation 2: over-dispersed counts (variance = 1.5 x mean), analysed with the matching variance.
set.seed(2459665)
n    <- n_per_arm(0.80)
reps <- 20000
m0 <- lambda0 * n * tt; m1 <- lambda1 * n * tt
x0 <- rpois(reps, m0); x1 <- rpois(reps, m1)
z  <- log(x1 / x0) / sqrt(1 / x0 + 1 / x1)
cat(sprintf("Simulated power, Poisson counts, n = %s per arm: %.3f\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE)))
phi <- 1.5
y0 <- rnbinom(reps, mu = m0, size = m0 / (phi - 1)); y1 <- rnbinom(reps, mu = m1, size = m1 / (phi - 1))
z2 <- log(y1 / y0) / sqrt(phi * (1 / y0 + 1 / y1))
cat(sprintf("Simulated power, over-dispersed counts (phi = 1.5), same n: %.3f\n", mean(abs(z2) > z_a, na.rm = TRUE)))

Output from running the code

100% of observed effect: n per arm = 18 (80% power), 24 (90% power)
 75% of observed effect: n per arm = 24 (80% power), 32 (90% power)
 50% of observed effect: n per arm = 43 (80% power), 58 (90% power)
Simulated power, Poisson counts, n = 18 per arm: 0.887
Simulated power, over-dispersed counts (phi = 1.5), same n: 0.697

The first simulation draws Poisson counts and should land close to the target power. The second allows extra variability (variance 1.5 times the mean, a purely illustrative value), which lowers the power at the same sample size, so rate trials usually need more participants than the Poisson formula suggests.

13. Multiplicity, Missing Data and Analysis Populations

The posted primary results include three intervention-versus-control comparisons: oral metronidazole, Ecologic Femi+, and Gynophilus LP. The p-values are reported as unadjusted: 0.043, 0.106, and 0.16, respectively. The registry record does not report whether one comparison was designated as the primary contrast for inference, whether a family of comparisons was prespecified, or whether any correction for multiple testing was planned or used. Consequently, the p-values should be read as unadjusted results, without assuming they control the chance of a false-positive finding across the comparisons.

The registry describes a modified intent-to-treat population that excludes randomized women with Nugent 7-10 at enrollment. It does not report how many participants met that exclusion, how missing outcome observations were handled, or whether any further participants were omitted from each analysis. The primary outcome counts by arm are 15, 11, 14, and 11, while overall enrollment is 68; the reason for this difference is not reported in the registry record. No missing-data method, sensitivity analysis, or participant flow details are available in the registry entries.

The registry lists one secondary outcome, vaginal microbiota composition by Illumina HiSeq sequencing, focused on Lactobacillus genus concentration. The number of outcome measures is listed as 6, but detailed results for the secondary outcome are not included in the registry entries. The registry record does not report a hierarchy of outcomes, a statistical analysis plan, or whether analyses were exploratory or confirmatory. It also does not report interim analyses or stopping rules. Those items cannot be inferred from the posted estimates or from the study’s pilot description.

These omissions affect how confidently the nominal comparisons can be interpreted. They do not negate the posted estimates, but they limit assessment of analysis choices, missingness, and the risk that selective reporting or multiple testing influenced the reported findings.

14. 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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: typhoid fever (5.9%), malaria during pregnancy (5.9%).

The posted safety figures list serious adverse events for 17 participants at risk in each of the four groups. The number with a serious adverse event was 0 in control, 1 with oral metronidazole, 1 with Ecologic Femi+, and 0 with Gynophilus LP. Deaths were 0 in every group. These are participant counts, not incidence rates, and they should not be combined with the primary outcome rate estimates.

The record lists Typhoid fever and Malaria during pregnancy among the top serious adverse event terms, each at 5.9%. It does not give group-specific attribution for these terms in the registry entries, so they cannot be assigned to a particular study group from this information. The non-serious adverse event counts were 2 in control, 2 with oral metronidazole, 2 with Ecologic Femi+, and 3 with Gynophilus LP, among 17 participants at risk in each group.

These counts are descriptive. With the small numbers shown and no further information about event timing, severity details, relatedness, or statistical comparisons, they do not establish whether groups differed in safety. Adverse event reporting and the primary efficacy outcome are distinct; neither should be used as a substitute for the other.

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

The registry’s primary outcome estimates describe bacterial vaginosis incidence by modified Amsel criteria during a 2-month intervention period, among randomized women except those with Nugent 7-10 at enrollment. The control rate was 7.53 events per PY at risk. The active-group rates were 2.04 for oral metronidazole, 3.36 for Ecologic Femi+, and 3.35 for Gynophilus LP. Each active-group rate ratio was below 1 in the posted comparison with control, but the confidence intervals are broad. The interval for each probiotic comparison includes 1; the oral metronidazole interval does not.

This pattern is a description of the posted estimates, not a conclusion that any strategy works or does not work. The comparisons come from a pilot study with small outcome-specific arm counts, and the uncertainty intervals allow materially different underlying relative rates. The unadjusted p-values do not address multiplicity, and the registry record does not report enough detail to evaluate the full analysis plan, missing-data handling, or all aspects of the outcome analysis.

The endpoint is defined by a particular set of clinical criteria and expressed as events per person-year at risk. It is not identical to a measure of the proportion of participants with at least one recurrence, and it does not directly describe outcomes beyond the 2-month intervention period. The record does not report the actual person-time by arm or further details of event ascertainment. General statistical interpretation: estimates tied to a specific population, outcome definition, and follow-up period should not automatically be extended to different settings or periods.

The trial’s stated purpose was to prepare for a clinical trial of interventions to maintain normal vaginal microbiota and prevent adverse reproductive health outcomes in Africa. The registry lists one secondary microbiota outcome but does not include its result among the registry entries. Therefore, the primary incidence estimates alone do not establish effects on microbiota composition or broader reproductive health outcomes. The record supports a cautious reading of exploratory, uncertain estimates rather than a definitive comparative conclusion.

16. Why This Trial Matters Statistically

This study illustrates several statistical issues that matter in small randomized trials. First, the primary endpoint is a rate, not a simple proportion. Its denominator is person-years at risk, so a reader needs to distinguish event frequency over exposure time from the number of participants who experienced an event. The record states the rate unit and modified intent-to-treat population, but does not provide further detail about how person-time was accumulated.

Second, rate ratios and absolute rates answer related but different questions. The rate ratios compare each intervention group with control, while the arm-specific rates describe event frequency in each group. Reading both helps prevent a relative comparison from obscuring the scale and uncertainty of the underlying rates. In this study, the confidence intervals are wide, underscoring that the point estimates are not precise.

Third, randomized assignment does not eliminate uncertainty in a pilot study with small outcome-specific groups. Broad intervals can arise even when the design is randomized, and a p-value alone does not indicate precision or importance. The unadjusted p-values also raise the need to understand how multiple comparisons were planned; that information is not reported in the registry record.

Finally, the sample-size planning illustrations show how planning depends on the target effect. Smaller effects require more participants per arm under the stated assumptions. Those calculations use observed estimates and are not a substitute for a prospectively justified design. Taken together, the study is statistically useful as an example of how endpoint definition, person-time, interval width, multiplicity, and planning assumptions shape interpretation, while leaving important design and analysis details unreported.

17. 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.
• Rate comparisons use a Poisson approximation unless the sponsor posted confidence limits; real event counts are often over-dispersed.

Several trial-specific limitations constrain interpretation. The registry identifies the trial as a pilot, with an overall enrollment of 68 and primary outcome counts of 15, 11, 14, and 11 across the four groups. The registry does not explain why the outcome counts differ from overall enrollment. It also does not report a participant flow account, the number excluded for Nugent 7-10 at enrollment, or the handling of missing outcome observations.

The primary outcome uses modified Amsel criteria and a person-year denominator, but the record does not report additional details about ascertainment, follow-up time, or treatment of repeated events. The follow-up window is the 2-month intervention period, so the posted results do not describe longer-term recurrence. The study’s single-masking label is not accompanied by details about who was masked or whether masking was maintained.

The posted record does not include the original sample-size justification, the full statistical analysis plan, multiplicity procedures, interim analysis details, or sensitivity analyses. It gives unadjusted p-values for three comparisons. The confidence intervals are broad, and the small outcome-specific arm counts limit precision. The registry lists one secondary microbiota outcome but does not include its results in the registry entries.

These limitations do not change the numerical estimates, but they restrict what can be concluded from them. In particular, the record does not enable a full assessment of missingness, model assumptions, analysis selection, or the relationship between the primary outcome and broader reproductive health outcomes.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Rwanda Vaginal Microbiota Restoration Study — NCT02459665.
• Registry record last updated 2019-07-16; status last verified 2019-06.
• Results first posted 2019-07-16.
• 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.

21. Record Summary

NCT02459665, the Rwanda Vaginal Microbiota Restoration Study, was a completed randomized, single-masked, parallel-group pilot study listed as phase NA. The registry records 68 participants enrolled. The four groups received no intervention after initial treatment, oral metronidazole, Ecologic Femi+, or Gynophilus LP.

The primary outcome was bacterial vaginosis incidence by modified Amsel criteria during the 2-month intervention period, expressed as events per PY at risk. The modified intent-to-treat population excluded women with Nugent 7-10 at enrollment. Posted rates were 7.53 for control, 2.04 for oral metronidazole, 3.36 for Ecologic Femi+, and 3.35 for Gynophilus LP. Confidence intervals were broad, and outcome-specific arm counts were small.

The registry lists vaginal microbiota composition by Illumina HiSeq sequencing as a secondary outcome. It does not report the original sample-size justification, detailed analysis plan, or missing-data procedures.