Evaluation of the Safer Bars Prevention Program for Sexual Assault: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of the Safer Bars Prevention Program for Sexual Assault (NCT04141839), a randomized na study. The primary outcome is Number of Participants Who Reported Always Intervening Behavior at 6 Months.

2. Clinical Question

The substantive question is about reported action, rather than only awareness, attitudes, or intention. The outcome description refers to survey items about actions taken after observing sexual aggression. Such reports can help characterize whether participants say they intervene in the setting targeted by the training. They do not, by themselves, establish how often an intervention occurred in real time or whether the behavior was independently observed.

The evaluation is described as randomized and cluster-randomized, with bars as the units assigned to training or delayed training. That design makes the comparison relevant to an intervention delivered at a venue level: staff at the same bar may share training, working conditions, and opportunities to intervene. The registry record does not report an analysis of the primary outcome, so it does not establish whether reported behavior differed between conditions.

3. Trial Design

NCT04141839 is a completed, randomized evaluation of the Safer Bars Prevention Program for Sexual Assault. Its official title describes a cluster-randomized effectiveness evaluation of alcohol-related sexual violence prevention through bar-staff bystander training. The record lists a crossover model and single masking. It does not identify who was masked, or describe further crossover procedures.

The intervention was behavioral training delivered at bars and attended by their liquor-serving staff. Bars in the intervention arm were assessed directly before and after training, and at 3 and 6 months after training. The control arm received a delayed intervention: baseline and 3-month assessment preceded Safer Bars training. The study goal was to evaluate program implementation at three sites. The registry lists enrollment as 681 and the status as completed. Its recorded start date is 2020-10-16 and its primary completion date is 2023-11-01.

Cluster randomization means that assignment occurs at a group level rather than by independently assigning each person. In this evaluation, the arm descriptions identify bars as randomized. Because staff within a bar share a setting, their outcome reports may be more alike than reports from staff at different bars. An analysis that treats every person as fully independent can therefore understate uncertainty. The registry record flags this concern and indicates that a simple individual-level recalculation would be misleading.

The record does not report the number of bars, the allocation procedure beyond randomization, the masking role, or the statistical model used for the sponsor’s analysis. It also does not describe how the crossover designation was implemented. These details matter when assessing the comparison, but their absence from the registry record should not be filled in by assumption.

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

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 Who Reported Always Intervening Behavior at 6 Months, measured over at baseline and up to 6 months.

The registered primary endpoint is the number of participants who reported always intervening behavior at 6 months. It is based on self-reported survey items about actual behavior, described as actions taken in response to observing sexual aggression in the bar environment. The record gives the timeframe as baseline and up to 6 months, while the primary outcome title specifies the 6-month report.

This endpoint is relevant to the program’s stated aim because the training concerns bystander action in bars. Asking about reported action focuses on behavior rather than knowledge or intention alone. The registered description says the survey items will be analyzed to identify significant differences after the intervention. However, no result or analysis for this endpoint is reported in the registry record. The number of participants who reported always intervening, the comparison between arms, and the precision of any estimated difference are therefore not available.

The record classifies the outcome as binary and describes the reported measure as a participant count. In general, a participant-level binary outcome distinguishes whether a person meets the specified response criterion; a group-level count is the number meeting it. The count alone does not convey the uncertainty in a comparison. Interpretation also depends on how responses were handled, which participants contributed data, and how clustering and repeated assessments were accounted for. Those analysis details are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies a randomized, cluster-randomized evaluation and explicitly cautions that a simple individual-level recalculation would be misleading. That caution is central to interpreting the statistical methods. The statistical ideas below are general explanations; the registry record does not report the model or procedures used for the sponsor’s analysis.

Why does cluster assignment affect analysis?

In a cluster-randomized study, groups rather than individual participants are assigned to conditions. Here, the arm descriptions say bars were randomized. People working at the same bar may share exposure to training and a work environment, so their answers may be correlated. General statistical methods account for that dependence rather than treating every response as an independent piece of information. If clustering is ignored, uncertainty can be understated. The extent of within-bar correlation, number of bars, and method used to account for clustering are not reported in the registry record.

Cluster-aware approaches can include models with cluster-level effects or methods that adjust standard errors for clustering. Which approach is appropriate depends on the design, outcome, and analysis objective. The record does not identify the sponsor’s method, so it is not possible to describe its assumptions or assess whether those assumptions were met. The registry’s warning is a reason not to reconstruct a standard individual-level comparison from aggregate participant counts.

What does a binary outcome mean in this setting?

The primary outcome is titled as a count of participants who reported always intervening at 6 months. The registry classifies the result type as binary and describes the endpoint as a count of participants. In general, the underlying response can be treated as whether each participant meets the reported criterion; the group count then reflects how many meet it. A comparison may be expressed using a difference in proportions, a ratio, or a model-based contrast, but the registry record does not report which measure was used. No such measure should be inferred from the endpoint title.

For a binary outcome, a confidence interval communicates the range of effect values reasonably compatible with the data and analysis assumptions. It reflects uncertainty, not the probability that a particular effect is true. A p-value, when used, assesses compatibility of the observed data with a specified null model; it does not measure the size or practical importance of an effect. Neither an interval nor a p-value for the primary outcome is reported in the registry record.

How do repeated assessments affect interpretation?

The intervention arm had assessment directly before and after training, then at 3 and 6 months after training. The delayed-intervention control arm had baseline and 3-month assessment before it received training. Repeated assessments can describe the timing and persistence of reported behavior, but analyses must define which observations contribute to the primary contrast. In general, repeated observations from the same person are also correlated. A suitable model or other method must account for that dependence, as well as dependence among people at the same bar. The registry record does not state how either source of correlation was handled.

The crossover designation indicates a crossover model in the registry, but the arm descriptions do not provide the details needed to reconstruct sequence, period, or carryover handling. In general, a crossover analysis may need to distinguish treatment effects from changes related to time or order. Whether such considerations apply to the sponsor’s primary analysis, and how they were addressed, are not reported in the registry record.

What can the planning assumptions tell readers?

The record describes sample-size illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. It says that the effect to detect is a stated fraction of the observed effect, and that non-time-to-event illustrations are expressed as participants per arm. These are planning assumptions, not evidence that the trial was designed using those calculations and not evidence of observed power. The record does not report a sample-size justification for the trial. The observed primary effect and numerical illustration results are not reported in the registry record, so the assumptions alone cannot show the precision achieved or the ability of this trial to detect a particular effect.

In general, a planning calculation estimates the sample size needed under specified assumptions about effect, variability, significance level, and allocation. Changing an assumed effect changes the resulting planning requirement. A calculation based on an observed effect is retrospective and should not be presented as the probability that the completed trial found a true effect. Cluster randomization also requires attention to the number and size of clusters and their correlation; those inputs are not described in the registry record’s stated assumptions.

What is reported about the actual analysis?

The registry lists zero analyses. It does not report an effect estimate, confidence interval, p-value, model specification, analysis population, or missing-data method for the primary endpoint. The record’s statement that only the sponsor’s own analysis is appropriate for this cluster-randomized design does not supply that analysis’s results. Without those results, there is no basis in the registry record for estimating the intervention effect or its precision.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Control Staff", "Intervention Staff")
n      <- c(310, 321)
events <- c(207, 239)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Control Staff  310  207  66.8  61.2  72.0  61.4  71.8  61.3  71.8
ARM  Intervention Staff  321  239  74.5  69.3  79.1  69.4  78.9  69.4  78.9

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 2 outcome measures and identifies the participant count reporting always intervening behavior at 6 months as the primary outcome. It does not give the details of the other outcome measure in the registry record recorded for this summary. The record lists zero analyses, so it does not report statistical tests or estimates for the primary or other outcomes.

Multiplicity arises when an analysis examines several outcomes, time points, subgroups, or contrasts. In general, conducting many tests can increase the chance of at least one apparently unusual result even when there is no underlying effect. A prespecified hierarchy or adjustment can address some multiplicity concerns, but the registry record does not report whether any such plan was used. The primary designation alone does not establish how other outcomes or repeated assessments were handled.

The outcome description includes baseline and follow-up through 6 months, and the arms have different assessment schedules around delayed training. The registry record does not say which observations or contrasts were included in the primary analysis. It also does not report how missing survey responses were addressed, whether participants were analyzed according to their assigned condition, or how the analysis population was defined. These omissions prevent assessment of attrition, the handling of incomplete data, and the relationship between the registered outcome and any analysis set.

These are reporting gaps in the registry record, not evidence that the trial lacked a statistical plan or that any particular procedure was omitted. The analysis plan, multiplicity procedures, missing-data methods, and analysis populations are not reported in the registry record.

12. 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.

The registry lists serious adverse event counts of 0 in the Control group and 0 in the Intervention group. The corresponding serious-event at-risk counts are 335 and 346. The record also lists other adverse event counts of 0 in each group, with at-risk counts of 335 and 346, respectively, and deaths of 0 in each group. No serious adverse event terms are listed.

These entries describe what is recorded in the registry safety fields. A reported count of 0 means that no event was entered in that category for the listed group in this record. It does not, by itself, establish that no event occurred outside the reporting process, that all participants were observed for the same duration, or that the intervention has no safety concerns. The registry record does not describe the event collection procedures, definitions, follow-up completeness, or whether the at-risk counts represent the population used for the primary endpoint.

Safety counts should also be distinguished from the primary behavioral outcome. The serious and other adverse event fields do not provide evidence about reported intervening behavior, and the primary outcome description does not substitute for safety monitoring. The record’s group-specific at-risk values should be read as reported; they do not establish the endpoint analysis denominators.

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

The record establishes that Safer Bars was evaluated in a completed randomized study with a cluster-randomized design, and it identifies self-reported always intervening behavior at 6 months as the primary endpoint. It also describes the intervention as training for liquor-serving staff at bars and a delayed-training control condition. Those details clarify what question the study set out to address.

The record does not report the primary outcome findings. There is no reported group comparison, effect estimate, confidence interval, or p-value. Therefore, it cannot establish from the registry record whether reported intervening behavior was higher, lower, or similar between the study conditions. It also cannot show the size or precision of a possible difference. The absence of a reported result should not be interpreted as evidence that the conditions were equivalent, or as evidence that the intervention did or did not change behavior.

The endpoint is self-reported action after observing sexual aggression in a bar environment. Even if a result were available, it would describe responses captured by those survey items, not necessarily independently observed behavior or every aspect of sexual violence prevention. The record does not report an outcome for sexual assaults prevented, nor does it establish effects beyond the evaluated settings and participants.

Cluster assignment and repeated assessment require methods that respect dependence among people at the same bar and observations over time. The registry record warns against simple individual-level recalculation, while listing no analyses. This leaves no appropriate numerical estimate to interpret. The planning assumptions described in the record are illustrations, not findings about treatment effects or observed power. The meaningful conclusion is limited: the research question and design are registered, but the outcome evidence needed to answer that question is not reported in the registry record.

14. Why This Trial Matters Statistically

This trial raises an important statistical issue for studies of interventions delivered in shared environments: the unit assigned to a condition may differ from the person whose behavior is measured. Safer Bars was assigned at the bar level, while the primary endpoint is a count of participants reporting a behavior. The analysis must connect those levels appropriately. Treating every participant as statistically independent can overstate the information in the data when responses within bars are correlated.

The registry record explicitly flags that concern, which is useful for interpreting why a simple recalculation from individual-level summaries would not be suitable. It also makes clear why a cluster-aware sponsor analysis would be needed to assess the primary comparison. However, the record lists no analyses and gives no primary outcome result. Clear design documentation cannot replace the estimate and uncertainty needed to evaluate the observed evidence.

The endpoint also illustrates the distinction between a behavior that matters to a prevention program and the way it is measured. Self-reported action is directly relevant to the stated program aim, but its interpretation depends on question wording, response handling, follow-up, and whether reported action corresponds to behavior in practice. The record describes the survey basis but does not report these additional measurement or analysis details.

Finally, the study has a delayed-intervention comparison and repeated assessments, making the timing of the contrast important. A transparent report would explain which assessments informed the primary comparison and how clustering, repeated measures, missing responses, and multiplicity were handled. Those details and the sponsor’s outcome analysis are not reported in the registry record. The trial therefore serves as a reminder that design details and outcome definitions are necessary context, while an interpretable result also requires a reported analysis and its precision.

15. 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 main limitation for interpreting this trial from the registry record is the absence of reported primary outcome results. Although the primary endpoint is identified, the record provides no group counts for that endpoint, comparison estimate, confidence interval, or test. It also lists zero analyses. The direction, size, and precision of any difference therefore cannot be assessed from the registry record.

The design is described as cluster-randomized, and the record cautions that an individual-level recalculation would be misleading. Yet the number of bars, details of assignment, and sponsor’s analytical method are not reported. The record also lists a crossover model without enough detail to assess how its periods or any potential carryover were handled. These limits prevent an independent assessment of the statistical analysis.

The endpoint is self-reported and concerns actions taken after observing sexual aggression in a bar. The record does not describe survey wording, validation, response rates, missing responses, or whether reported actions were externally verified. It does not report the analysis population or whether the primary endpoint data were available at each assessment. Enrollment of 681 should not be treated as the number contributing to the primary endpoint.

The record notes 2 outcome measures but does not report a multiplicity procedure or detailed analysis plan. Its sample-size assumptions are planning illustrations, not the trial’s design justification or observed power. The stated safety counts are not a substitute for primary-outcome findings and do not describe safety ascertainment procedures. These limits concern what is reported in the registry record; they do not establish that additional methods or results were absent from other records.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Evaluation of the Safer Bars Prevention Program for Sexual Assault — NCT04141839.
• Registry record last updated 2025-04-27; status last verified 2025-04.
• Results first posted 2025-04-27.
• 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.

19. Record Summary

NCT04141839 is a completed randomized, cluster-randomized evaluation of the Safer Bars behavioral prevention program. The study goal was to evaluate implementation at three sites. Bars in the intervention arm received staff training, with assessments before and after training and at 3 and 6 months afterward. The delayed-intervention control arm had baseline and 3-month assessment before training.

The primary endpoint is the number of participants reporting always intervening behavior at 6 months, based on self-reported survey items about actions taken after observing sexual aggression in a bar environment. Enrollment is recorded as 681. The registry lists 2 outcome measures and zero analyses. It does not report a primary outcome estimate, uncertainty interval, or statistical test.

The record reports zero serious adverse events, zero other adverse events, and zero deaths in both listed groups. It cautions that a simple individual-level recalculation would be misleading because of the cluster-randomized design. The sponsor’s analysis results are not reported in the registry record.