Evaluation of Many Men, Many Voices, An STD/HIV Prevention Intervention for Black MSM: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of Many Men, Many Voices, An STD/HIV Prevention Intervention for Black MSM (NCT00137631), a randomized phase 1, phase 2 study. The primary outcome is Any Unprotected Anal Intercourse (UAI) With Casual Partners.

2. Clinical Question

The registry describes the program evaluation as examining whether the intervention reduces HIV sex-risk behaviors and increases HIV testing. The primary outcome considered in this analysis is narrower: the number of episodes of unprotected anal intercourse with casual partners. It does not, on its own, represent every behavior or objective named in the broader evaluation purpose. HIV testing is part of that stated purpose, but no HIV-testing result is included in the posted primary outcome details.

The comparison is between immediate participation in the intervention and delayed participation after a wait-list period. The principal statistical question is therefore about the difference in the recorded behavioral measure between those randomized groups at the stated follow-up. The registry record does not report enough information to establish how the primary outcome relates to the other outcome measures or whether it was the only outcome designated as primary in the original evaluation plan.

3. Trial Design

The registry identifies this as a randomized, parallel-group evaluation with no masking listed. One group received the six-session Many Men, Many Voices intervention immediately after baseline assessment and randomization. The wait-list comparison group was scheduled to receive it after a six-month delay. The record lists PHASE1 and PHASE2, describes the intervention as behavioral, and lists the Centers for Disease Control and Prevention as sponsor.

Enrollment was 341. The study began in 2005-08, had primary completion in 2007-08, and is marked completed. The setting named in the official title is the New York metropolitan area; the title also identifies People of Color in Crisis in Brooklyn, NY. The registry’s brief summary characterizes the effort as a program evaluation intended to support assessment of an existing intervention and provide feedback to the implementing organization, rather than to conduct research.

Random assignment establishes the intended basis for comparing the two groups, but the registry record does not report the allocation ratio, details of sequence generation, concealment procedures, or other implementation information. No masking is listed; the record does not describe whether outcome assessment or analysis was masked. The original protocol’s sample-size justification and design assumptions are not reported in the registry record. Nor are interim analyses or stopping rules reported in the registry record.

The primary outcome results list 127 participants in the intervention arm and 133 in the wait-list arm. Those are the group counts attached to the posted continuous outcome, not a statement about the number randomized or the number with complete data at every assessment. The registry record does not explain the relationship between those endpoint counts and total enrollment, or give a participant flow account that would clarify follow-up and analysis inclusion.

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:

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 Any Unprotected Anal Intercourse (UAI) With Casual Partners, measured over 6 months.

The primary endpoint is titled Any Unprotected Anal Intercourse (UAI) With Casual Partners. It is a continuous measure summarized as a mean number of episodes, with standard deviations also reported. The registry describes the measure as sexual activities with casual male partners in the past three months, specifically any unprotected insertive or receptive anal sex. The recorded timeframe for the outcome is six months.

This measure captures the frequency of a particular self-reported sexual behavior rather than whether a participant had any episode at all, whether an infection occurred, or whether HIV testing took place. A count can distinguish a report of no episodes from reports of one or more episodes, while the mean describes the average episode count in each group. The registry does not report the questionnaire wording, how responses were collected, or how the six-month assessment and past-three-month recall period were operationalized.

The endpoint is relevant to the stated evaluation purpose because the registry names reduction in HIV sex-risk behaviors as one aim. It remains a behavioral measure, not a direct measure of HIV or sexually transmitted disease incidence. The registry record does not report a rationale for selecting this specific endpoint, its measurement properties, or its status relative to the other outcome measures beyond identifying it as the primary outcome in the posted results.

7. Results Posted in the Registry

• Wait List Comparison: mean 6.20 versus 2.00 on control; difference 4.20 (95% CI 0.22 to 8.18), standardized effect d = 0.25; z-test p = 0.039.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a count of episodes, but the registry also describes it as a continuous measure and posts group means and standard deviations. The simple re-calculation uses these arm-level summaries to compare the mean episode count in the wait-list group with the mean in the intervention group. It is unadjusted: it does not account for participant-level covariates or apply the sponsor’s regression model. The resulting mean difference is 4.2 episodes, with a p-value of 0.039. This calculation is useful as a transparent comparison of the posted summary statistics, but it does not duplicate the sponsor’s analysis.

What does the negative binomial model do?

The sponsor reports a negative binomial regression with a rate ratio as its effect measure. General explanation: negative binomial regression is a model commonly used for count outcomes, especially when the observed counts vary more than would be expected under a simple Poisson model. This extra variability is often called over-dispersion. A model that accommodates over-dispersion can provide a more suitable way to represent variable count data than a model that assumes the Poisson mean and variance are equal. That general description does not establish the exact specification used in this trial. The registry record does not report covariates, offsets, link function details, or diagnostic checks for the sponsor’s model.

A rate ratio expresses a relative comparison on a ratio scale, whereas a mean difference expresses an absolute contrast in the units of the outcome. The reported rate ratio is 0.58, with a 95% confidence interval from 0.33 to 1.01. Because the rate ratio and mean difference are different estimands and come from different methods, their values should not be expected to match numerically. The registry’s discrepancy flag is important: the sponsor’s own analysis takes precedence over the simple re-calculation.

What does the p-value say?

General explanation: a p-value measures how incompatible data are with a specified null model, under the assumptions of the test. It is not the probability that the null hypothesis is true, nor does it measure the magnitude or practical importance of an effect. The recalculated p-value of 0.039 pertains to the unadjusted comparison based on posted means and standard deviations. The sponsor’s p-value is reported as p < 0.05 for the negative binomial regression. The record does not provide enough information to reconstruct the sponsor’s test or explain the relation between that p-value and the interval that extends to 1.01.

General explanation: an interval estimate combines an effect estimate with information about statistical uncertainty under a model. A wider interval indicates less precision than a narrower interval on the same scale. Intervals do not correct for possible measurement problems, missing outcomes, selection, or model misspecification. Their interpretation depends on the model and the data contributing to it.

Why distinguish the two analyses?

The direct calculation uses group-level summary statistics, while the sponsor’s analysis uses negative binomial regression. This difference can affect the estimated contrast and its uncertainty. The count distribution may be skewed, and the arm standard deviations differ considerably, but the registry does not report distributional plots or individual observations that would allow a fuller assessment. The recalculated mean difference therefore serves as a descriptive, unadjusted check rather than an alternative authoritative analysis.

The registry lists five outcome measures and five analyses, while the posted results identify no secondary outcomes. The record does not provide a full statistical analysis plan, explain the relationship between the listed measures and analyses, or state whether adjustments for multiple comparisons were used. The exact analysis population for the sponsor’s model and the handling of incomplete outcome data are also not reported in the registry record. These omissions limit how far the posted statistical results can be evaluated independently.

10. Confidence Intervals and Precision

The unadjusted mean comparison is estimated as a difference of 4.2 episodes between the wait-list and intervention group means. The recalculated confidence interval is broad, indicating that the estimate is not highly precise. The posted summary statistics therefore support describing the observed contrast, but they leave substantial uncertainty about its size. A point estimate alone would give an incomplete account of that uncertainty.

The sponsor’s regression result has a 95% confidence interval from 0.33 to 1.01 around the rate ratio of 0.58. General explanation: a confidence interval is a range of values compatible with the data and the assumptions of the statistical method at the stated confidence level. It is not a guarantee that the true effect lies in that range, and it does not account for every source of uncertainty. In this case, the interval approaches and extends to 1.01, so its upper end is close to the ratio value of 1 that represents no relative difference. The registry also reports p < 0.05 for this analysis. The two summaries are not readily reconciled under conventional inference, and the registry does not explain the discrepancy.

The interval for the unadjusted mean difference and the interval for the rate ratio describe different scales and different analyses; they should not be compared as if they were interchangeable. The sponsor’s reported analysis takes precedence because the registry flags a discrepancy. Neither estimate should be interpreted as exact. In particular, the wide variation in the posted episode counts and the absence of reported details about missing observations, model specification, and measurement procedures constrain interpretation of precision.

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 participants per arm, not total events. They use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at 100%, 75%, or 50% of the observed effect. The original trial’s own design assumptions and sample-size justification are not reported in the registry record.

For the 100% observed-effect planning scenario, the estimates are 249 participants per arm for 80% power and 333 participants per arm for 90% power. For 75% of the observed effect, the corresponding estimates are 442 and 591 participants per arm. For 50% of the observed effect, they are 993 and 1,330 participants per arm. The 80% power column represents the planned group size for an 80% chance of detecting the specified effect under the stated assumptions, if that effect is truly present. The 90% column describes the larger group size associated with a 90% chance under those same assumptions.

Smaller effects require more participants to detect under these assumptions. The listed numbers are planning illustrations based on the observed effect, not observed power and not a claim about the trial’s original design. The curve presents the planning relationship visually; its meaning is governed by the same assumptions and should not be read as a separate analysis of observed trial performance.

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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 4.200; pooled standard deviation σ = 16.702.
2. n = 2 × (1.96 + 0.842)² × 16.702² / 4.200² = 248.2, rounded up to 249 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
delta <- 4.2                      # observed difference in means
sigma <- 16.7018589690865                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(137631)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 249 (80% power), 333 (90% power)
 75% of observed effect: n per arm = 442 (80% power), 591 (90% power)
 50% of observed effect: n per arm = 993 (80% power), 1,330 (90% power)
power.t.test: 249.2 
Simulated power at n = 249 per arm: 0.803 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The sponsor reports negative binomial regression, with a rate ratio of 0.58, a 95% confidence interval from 0.33 to 1.01, and p < 0.05. The separate re-calculation is a simple unadjusted comparison of posted group means and gives a mean difference of 4.2 episodes and p = 0.039. The estimate and p-value from the re-calculation are not a reproduction of the sponsor’s model-based result.

The registry flags a discrepancy, so the sponsor’s own analysis takes precedence. The two analyses use different statistical approaches and express the comparison on different scales. The registry does not report sufficient model details to determine exactly why the inferential summaries differ, and the apparent tension between the sponsor’s reported interval and p-value is not explained in the record. Accordingly, the sponsor’s regression result should be quoted as reported, while the unadjusted calculation should be labeled as a recalculation from summary statistics rather than treated as a correction or replacement.

The registry does not report the analysis code, model covariates, missing-data procedure, or full analysis plan. Those absent details prevent a more complete reconstruction of the sponsor’s result.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists five outcome measures and five analyses, but no secondary outcomes are posted. It identifies the unprotected anal intercourse measure as the primary outcome for the results. The record does not describe whether the other measures were prespecified as secondary, how the five measures relate to the five analyses, or whether there was a hierarchy among outcomes.

Multiplicity refers generally to the increased chance of a false-positive finding when many hypotheses are tested. General explanation: statistical plans may address multiplicity by defining a primary endpoint, ordering tests, or adjusting significance thresholds. The registry record does not report whether any such approach was used. Since the full analysis plan and the intended family of tests are not reported, it is not possible to determine whether the posted p-values were adjusted or to assess the overall multiplicity strategy.

The primary outcome lists 127 participants in the intervention arm and 133 in the wait-list arm. The registry does not explain why those numbers contributed to the posted endpoint, whether all randomized participants were included, or how missing outcome data were handled. It does not report whether missing values were imputed, whether analyses were restricted to participants with observed outcomes, or whether sensitivity analyses were performed. The analysis populations for the reported regression and recalculation are not fully described in the registry record.

General explanation: missing data can affect an estimate when the participants with observed outcomes differ systematically from those without them. The direction and size of any effect depend on the pattern and reasons for missingness. Without participant flow information and a stated missing-data method, the impact cannot be assessed from the posted results alone. The trial’s allocation method is listed as randomized, but the record does not give enough detail to evaluate implementation, adherence, or deviations from assigned timing.

15. Safety Results

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

Other (non-serious) adverse events

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

The serious adverse event entries list 164 participants at risk in the intervention group and 174 at risk in the wait-list group. Zero serious adverse events are recorded in each group. The entries also list 164 and 174 participants at risk, respectively, for other adverse events, with zero other adverse events recorded in either group.

These are the registry’s posted counts for the stated risk groups. They should not be interpreted as evidence that no unrecorded event occurred, nor do the entries establish whether all participants were monitored for the same duration. The registry does not report the method of adverse-event collection, definitions, monitoring period, or attribution procedures. The death fields are not populated with counts, so death information is not reported in the registry record; a blank field is not the same as an explicit count of zero deaths.

The safety entries are separate from the behavioral primary endpoint and do not answer whether the intervention changed the episode count. They describe the adverse-event counts recorded in the registry, with the limitations that follow from the missing monitoring details.

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

The posted primary outcome summaries show a lower mean number of reported episodes in the intervention group than in the wait-list group at the six-month outcome timeframe: 2.0 compared with 6.2. The simple unadjusted mean difference is 4.2 episodes, with a recalculated p-value of 0.039. The count distributions are variable, particularly in the wait-list group, and the re-calculated interval is broad. These results describe the posted outcome data and do not establish a precise magnitude for the group contrast.

The sponsor’s negative binomial analysis reports a rate ratio of 0.58, a 95% confidence interval from 0.33 to 1.01, and p < 0.05. Because the registry flags a discrepancy, the sponsor’s own analysis takes precedence. The interval and p-value are not readily aligned under conventional interpretation, and the registry does not explain why. This inconsistency should remain visible rather than being resolved by choosing the more favorable statistic.

The outcome is a reported behavioral count, not a direct measure of HIV infection, sexually transmitted disease, or HIV testing. The broader evaluation purpose includes HIV testing, but no result for that outcome is posted in the primary details. The results therefore do not establish effects on those outcomes, nor do they address every aspect of the intervention’s intended purpose.

Randomization and a wait-list comparison provide a framework for comparing groups, but the registry record omits information needed to evaluate follow-up, missing data, outcome measurement, and model specification in depth. The findings should be read as registry-posted statistical results for this specific measure and assessment, not as a general conclusion about all participants or settings.

17. Why This Trial Matters Statistically

This trial illustrates why an endpoint’s scale and statistical model matter. The primary outcome is an episode count summarized with means and standard deviations, while the sponsor reports a negative binomial regression and a rate ratio. A direct comparison of arm means and a count-regression estimate answer related but not identical statistical questions. Their estimates and uncertainty measures should not be treated as interchangeable.

The registry also demonstrates the value of comparing reported results with a transparent calculation from summary statistics. That check yields a simple unadjusted mean difference and p-value, but it does not replace a sponsor analysis. When the registry flags a discrepancy, the sponsor’s own analysis takes precedence; retaining the distinction between analyses avoids presenting a re-calculation as the original model result.

Precision is another central issue. The group means differ, yet the posted variability is substantial and the recalculated confidence interval is broad. The sponsor’s reported interval approaches the ratio null value even as its p-value is listed as below 0.05. A reader needs both the effect estimate and its uncertainty, as well as a clear account of the statistical method, to interpret such results.

Finally, the registry record leaves key questions about missing data, analysis populations, multiplicity, and protocol-level design assumptions unanswered. Statistical interpretation is not only about a p-value; it also depends on what was measured, who contributed data, how the model was specified, and how outcomes were selected and reported.

18. 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.

Several trial-specific details needed for a fuller appraisal are not reported in the registry record. The participant eligibility criteria beyond the broad population description, recruitment process, allocation ratio, randomization procedures, and participant flow are not described in the registry record of the posted record. The primary outcome counts do not account for all enrollment by themselves, and the reason for the difference is not reported in the registry record.

The endpoint is described as a report of sexual activities with casual male partners in the past three months at the six-month timeframe. The questionnaire, data-collection mode, and treatment of ambiguous responses are not reported in the registry record. Because the outcome is a self-reported behavioral measure, interpretation depends on measurement procedures that are not documented in the posted details.

The registry names negative binomial regression but does not report its full specification, covariates, diagnostics, or missing-data strategy. The analysis populations are not explained. The record also does not report an original sample-size rationale, interim analysis plan, stopping rules, or a multiplicity strategy. These gaps limit independent evaluation of how the sponsor’s inferential result was produced.

The sponsor’s confidence interval and p-value are reported in a way that appears difficult to reconcile under conventional interpretation. The registry does not explain that discrepancy. The separate unadjusted calculation cannot resolve it, because it is based on summary statistics and uses a different analysis. The safety record lists zero serious and other adverse events in the specified risk groups, but the monitoring methods and death counts are not reported in the registry record. These limitations call for a bounded reading of the posted results rather than broader conclusions.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Evaluation of Many Men, Many Voices, An STD/HIV Prevention Intervention for Black MSM — NCT00137631.
• Registry record last updated 2010-06-24; status last verified 2010-04.
• Results first posted 2010-05-14.
• 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.

22. Record Summary

NCT00137631 is a completed randomized, parallel-group evaluation of Many Men, Many Voices, a behavioral six-session HIV/STD prevention intervention. The registry lists PHASE1 and PHASE2, no masking, enrollment of 341, and the Centers for Disease Control and Prevention as sponsor. The study began in 2005-08 and had primary completion in 2007-08.

The primary outcome is the number of episodes of unprotected anal intercourse with casual male partners, assessed at six months with a past-three-month recall period. Posted means are 2.0 for the intervention group and 6.2 for the wait-list group. The sponsor reports negative binomial regression with a rate ratio of 0.58, a 95% confidence interval from 0.33 to 1.01, and p < 0.05. The registry flags a discrepancy; the sponsor’s own analysis takes precedence over the unadjusted recalculation.

No secondary outcomes are posted. Serious and other adverse event counts are zero in both listed groups; death information is not reported in the registry record.