Reducing HIV Risk Among Adolescents: Evaluating Project HEART: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Reducing HIV Risk Among Adolescents: Evaluating Project HEART (NCT02579135), a randomized na study. The primary outcome is 7-item Self-report of Sexual Communication Self-efficacy.

2. Clinical Question

3. Trial Design

The registry describes a randomized, parallel-group study with no masking. Participants were allocated to Project HEART or Project Growing Minds, a control website matched for attention. The study is marked completed, with a reported enrollment of 222. The registry lists the study start as 2015-09-07 and primary completion as 2016-05-04. The listed phase is NA, consistent with the behavioral educational interventions described in the record rather than a drug-phase designation.

Project HEART has five modules focused on safer-sex motivation, knowledge, attitudes and norms, self-efficacy, and communication skills. The control website has five modules focused on growth mindsets and self-regulation. Both programs are described as taking approximately 30-45 minutes to complete. This active comparison can help distinguish the specific content of Project HEART from the experience of using an educational website, although the registry record does not report measures of participants' engagement or adherence to the modules.

The primary outcome is a continuous score: a mean across seven self-report items on communication self-efficacy. Its stated assessment schedule includes an immediate post-test at intervention completion and a 3-month follow-up. The outcome summaries report n=107 for Project HEART and n=114 for the control, and give a mean and standard deviation for each group. Those summaries do not identify whether they refer to the immediate assessment, the follow-up, or another way of summarizing both assessments.

The record lists three outcome measures and two analyses, but only the primary outcome and one comparison are described in the available results. It does not report a detailed protocol, the original design assumptions, or the method used to handle repeated assessments. Those details cannot be inferred from the parallel-group label or from the posted outcome statistics.

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

Analysis population for the primary outcome, as defined in the registry: Data on acceptability only analyzed among girls who completed the Project HEART intervention

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 7-item Self-report of Sexual Communication Self-efficacy, measured over Immediate post-test at completion of intervention and 3-month follow-up.

The primary endpoint is the 7-item Self-report of Sexual Communication Self-efficacy. It uses the validated Self-Efficacy for HIV Prevention Scale to assess participants' confidence in communicating about sexual topics. Example content asks how sure a participant is that she could talk with a partner about safer sex. Responses range from 1, meaning couldn't do it, to 4, meaning very sure. The seven responses are averaged, with higher scores indicating greater reported confidence. The registry reports alpha = 0.82 for the measure.

A self-efficacy measure is used because confidence in carrying out a behavior is a distinct construct from knowledge or attitudes. In general, measuring it can help researchers assess whether an educational intervention is associated with a change in participants' perceived ability to communicate, rather than assuming that exposure to educational content changes that confidence. The score remains a self-report: it records participants' reported confidence, not observed communication behavior or sexual health outcomes.

The endpoint is listed at immediate post-test and 3-month follow-up. However, the posted arm means and comparison do not identify the timepoint represented, and separate results for the two assessments are not reported in the registry record. The interpretation therefore has to stay with the reported self-efficacy score and cannot establish from these summaries whether any difference persisted across the follow-up period.

7. Results Posted in the Registry

• Project HEART: mean 3.62 versus 3.51 on control; difference 0.11 (95% CI -0.01 to 0.23), standardized effect d = 0.24; z-test p = 0.079.

8. Statistical Interpretation

9. Statistical Methods Explained

What does the mean difference represent?

For a continuous outcome, a group mean describes the average observed score among participants included in that group summary. The posted means are 3.62 for Project HEART and 3.51 for Project Growing Minds. The unadjusted recalculation subtracts the control mean from the Project HEART mean, yielding 0.11 units on the scale. A positive value means the Project HEART mean is higher in this comparison. It does not mean every participant scored higher, and it does not describe a change from baseline because baseline outcome values are not reported in the registry record.

The arm standard deviations, 0.47 and 0.46, describe the spread of individual scores around each arm's mean. Standard deviation is different from standard error: the former describes variation among participants, while the latter describes the uncertainty of an estimated mean. The recalculation uses the posted group summaries to estimate an unadjusted difference and its uncertainty. It is not the same as a regression model that may adjust for other variables or use a different outcome definition.

What does a confidence interval tell us?

General explanation: a confidence interval gives a range of values compatible with the data under the assumptions of the method used. Its width conveys precision: a narrower interval indicates more precise estimation, while a wider interval leaves more uncertainty about the size and direction of the difference. The recalculated interval for this comparison spans -0.013 to 0.233. Because it crosses zero, it includes no difference as one of the values compatible with the data. That does not prove there is no difference, and the interval does not assign probabilities to individual possible values.

General explanation: a p-value is calculated under a specified null model. It indicates how unusual data at least as extreme as the observed data would be if that model were correct. The recalculated unadjusted p-value is 0.079. A p-value does not express the probability that the intervention is effective, the probability that a finding is due to chance, or the practical importance of an estimate. Interpretation should consider the estimate, its interval, the measurement scale, and the analysis design together.

Why can regression and a simple recalculation differ?

General explanation: an unadjusted comparison uses the group summaries without controlling for other variables. Regression can estimate a group contrast while accounting for additional variables or features of the data, depending on how the model is specified. Regression results can therefore differ from a direct comparison of means. Here, the registry reports a linear regression beta of 0.05 with p = .32, while the unadjusted recalculation reports a mean difference of 0.11 with p = 0.079. The registry does not describe the regression specification or identify enough details to reconcile the two analyses. No adjustment set, repeated-measures method, or outcome timepoint should be assumed.

Why can over-dispersion matter?

General explanation: over-dispersion occurs when observed variability is greater than the statistical model expects. If a model assumes less variation than the data contain, its uncertainty estimates can be too small and its inferential results overly confident. This concept is especially relevant to outcomes such as counts or proportions modeled with particular distributions. The primary outcome in this trial is a continuous mean score, and the registry does not report an over-dispersion assessment or a model feature that would establish it as a concern. It would be inappropriate to claim over-dispersion occurred or to apply a correction based on the record.

What is the role of a standardized effect?

General explanation: a standardized effect expresses a difference relative to variability, which can help compare results measured on different scales. The recalculation includes a standardized effect estimate, but its detailed value is not needed to interpret the primary result and is not repeated. Standardized effects also do not determine practical importance on their own. For this outcome, the original scale and its response anchors are essential context: the measure averages seven responses from 1 to 4, with higher scores indicating greater reported communication confidence.

What can be inferred about the design?

The randomized parallel design indicates that participants were assigned to one of two groups, and the record states there was no masking. Random allocation is intended, in general, to reduce systematic differences between groups at assignment. It does not guarantee that all groups are identical, that participants completed the programs, or that data are complete. The registry record does not report details about the randomization procedure, allocation concealment, adherence, or the handling of missing observations. Those aspects therefore cannot be assessed from the posted information.

The outcome schedule names an immediate post-test and a 3-month follow-up, but the posted comparison does not identify its timepoint or explain whether repeated observations were modeled jointly. In general, repeated measurements from the same participant are related rather than independent. Appropriate analysis may need to account for that relationship, but the record does not say whether or how it did so. The regression label alone is not enough to determine the model used.

10. Confidence Intervals and Precision

The recalculated mean difference is 0.11 units, with a confidence interval from -0.013 to 0.233. The interval is relatively close to zero at its lower end but extends to positive differences at its upper end; most importantly, it crosses zero. This means the recalculated data do not pin down the direction of the population difference with high precision. The interval is compatible with a very small negative difference, no difference, or a positive difference within its limits, under the method's assumptions.

Precision concerns how narrowly an estimate is determined by the data, not whether the estimate is important. The group standard deviations are 0.47 for Project HEART and 0.46 for Project Growing Minds, indicating individual scores vary around their group averages. The posted group counts are 107 and 114. General explanation: estimates usually become more precise when there is more information and less variability, but precision also depends on the analysis and the completeness and structure of the data.

The registry's linear regression result is beta = 0.05 with p = .32, but no confidence interval is posted for that estimate. Its precision therefore cannot be directly judged from an interval in the registry record. The difference between this regression result and the recalculated unadjusted estimate may relate to analysis choices or data definitions, but the necessary details are not reported. The timepoint represented by either posted comparison is also not identified, limiting interpretation of precision over time.

The interval quantifies uncertainty under the recalculation's assumptions; it does not account for unreported sources of uncertainty such as missing data handling, measurement limitations, or departures from the analysis assumptions. Its statistical range should therefore be read alongside the record's incomplete description of the analysis.

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. The primary outcome is continuous, so the table's figures refer to participants per arm. The registry's general planning note specifies a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The assumed effect to detect is the stated fraction of the observed effect. These are planning assumptions for an illustration based on the observed effect; the trial's original design assumptions are not reported in the registry record.

For planning at 100% of the observed effect, the figures are 281 participants per arm for 80% power and 376 participants per arm for 90% power. At 75% of the observed effect, they are 499 participants per arm for 80% power and 668 participants per arm for 90% power. At 50% of the observed effect, they are 1,122 participants per arm for 80% power and 1,502 participants per arm for 90% power. The rows illustrate that planning for a smaller effect requires more participants under the same stated assumptions.

In plain language, the 80% power column gives the approximate number per arm that the planning calculation associates with an 80% chance of detecting the specified effect under its assumptions. The 90% column expresses the corresponding planning target for a higher chance of detection. Neither column describes the probability that the observed result is true, and neither is a measure of what happened in the completed trial. These are planning illustrations, not observed power.

The values are sensitive to the assumed effect. Because the observed effect is used as the reference for each fraction, the figures should not be treated as a design recommendation without an independent rationale for the effect that matters. The registry does not report the trial's own sample-size justification or original design assumptions.

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 = 0.110; pooled standard deviation σ = 0.465.
2. n = 2 × (1.96 + 0.842)² × 0.465² / 0.110² = 280.4, rounded up to 281 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 <- 0.11                      # observed difference in means
sigma <- 0.464867045361374                     # 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(2579135)
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 = 281 (80% power), 376 (90% power)
 75% of observed effect: n per arm = 499 (80% power), 668 (90% power)
 50% of observed effect: n per arm = 1,122 (80% power), 1,502 (90% power)
power.t.test: 281.3 
Simulated power at n = 281 per arm: 0.795 (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 registry reports a linear regression analysis with beta = 0.05 and p = .32. It does not provide a confidence interval, model details, covariates, or a clear indication of which assessment timepoint the result represents. This is the sponsor's posted analysis as recorded in the registry.

The separate unadjusted recalculation from the posted group summaries gives a mean difference of 0.11 and p = 0.079, with a confidence interval from -0.013 to 0.233. That recalculation compares the listed means directly; it is not a reproduction of the regression model. Differences in estimates or p-values can arise when analyses use different models, timepoints, covariates, or data handling. The registry does not report enough information to identify which, if any, of these explains the contrast in results.

These results should therefore be reported as distinct analyses rather than combined into a single estimate. The recalculation provides a transparent unadjusted comparison of the posted summaries, while the regression result is the analysis reported in the registry. The record does not establish that both analyses use the same population, endpoint definition, or assessment. Neither result alone supplies the missing model details or resolves the uncertainty about the timepoint represented.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists three outcome measures and two analyses, but the results information describes one primary outcome comparison and does not post secondary outcome results. It does not report a multiplicity plan, such as whether there were designated primary comparisons, an adjustment across outcomes, or a hierarchy for interpreting multiple tests. When many outcomes or comparisons are examined, the chance of obtaining at least one small p-value can increase. General explanation: multiplicity procedures are used in some analyses to manage that issue, but no such procedure can be attributed to this study from the record.

The primary endpoint schedule includes an immediate post-test and a 3-month follow-up. The posted arm summaries and comparison do not specify which assessment they represent, and the record does not say whether the two timepoints were analyzed separately or jointly. It also does not report whether a baseline measure was collected or used in the analysis. The regression label does not resolve these questions.

Missing-data methods and the analysis population are not reported in the registry record. It is therefore not possible to determine whether the posted analyses included all randomized participants, only participants with observed outcome data, or another defined group. The registry does not explain how missing outcome values were handled, whether any participants were excluded from analyses, or whether an intention-to-treat approach was used. The outcome counts are 107 for Project HEART and 114 for Project Growing Minds, but the record does not explain their relationship to the enrollment figure of 222 or the safety counts.

These omissions matter because choices about outcomes, timepoints, missingness, and analysis populations can influence estimates and their uncertainty. The available record supports describing the posted score summaries and regression result, but not reconstructing the full analysis plan or judging whether all planned comparisons were reported. The original statistical analysis plan and protocol details are not reported in the registry record.

15. 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 reports zero serious adverse events and zero deaths in both groups. For serious adverse events, the listed at-risk counts are 107 for Project HEART and 115 for Project Growing Minds; the other adverse-event counts also list 107 and 115, respectively. No serious adverse event terms are listed. These figures describe what is recorded in this registry entry, not a complete account of every possible safety outcome.

The serious-event counts are distinct from the primary outcome group counts, which list 107 for Project HEART and 114 for Project Growing Minds. The registry does not explain why the control-group counts differ across these entries. The adverse-event record also does not provide descriptions of how events were collected, the observation period for safety reporting, or whether non-serious events were assessed. No explanation should be inferred from the counts alone.

For a behavioral website intervention, zero recorded serious events does not establish that no adverse experiences occurred or that the intervention is risk-free. It means only that the registry lists no serious adverse events or deaths for the reported groups. The safety information should be read in the context of its limited detail and alongside the fact that the study evaluated educational websites, not a medical treatment.

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

The posted summaries show average self-reported communication self-efficacy scores of 3.62 in Project HEART and 3.51 in Project Growing Minds. The unadjusted recalculation estimates a difference of 0.11 units in favor of the higher Project HEART mean. The recalculated interval crosses zero and the unadjusted p-value is 0.079. Taken together, these statistics indicate that the estimated difference is imprecise and that the data do not sharply distinguish a zero difference from differences in either direction within the interval.

The registry's own linear regression result is beta = 0.05 with p = .32. Since the model details and timepoint are not reported, the regression result cannot be fully interpreted or reconciled with the unadjusted calculation. Both findings concern the listed self-efficacy outcome; neither establishes an effect on actual communication, sexual behavior, HIV, other sexually transmitted diseases, or unplanned pregnancy.

The measure captures reported confidence in communicating about sexual topics, not directly observed skills or behavior. A higher score indicates greater reported confidence according to the scale definition. The registry states that assessments were planned immediately after intervention completion and at 3-month follow-up, but does not identify which timepoint is represented in the posted summaries. The results therefore cannot show from this record whether a difference persisted, changed, or was limited to one assessment.

The study's stated ultimate goal is to help girls reduce risk of HIV and other sexually transmitted diseases and avoid unplanned pregnancies. Those goals are not demonstrated by the primary outcome statistics alone. The registry record does not report the necessary clinical or behavioral results to determine whether those outcomes changed. The safest statistical interpretation is limited to the reported confidence score, the unadjusted comparison, and the separate regression result, with uncertainty about analysis details and timing.

These results are not a clinical recommendation. They describe the statistical evidence recorded for a randomized comparison of two educational websites. The record's omissions mean that claims about broader effectiveness, persistence, or outcomes beyond the self-report measure would go beyond the posted evidence.

17. Why This Trial Matters Statistically

This trial illustrates how a behavioral intervention can be evaluated against an active educational comparison rather than against no program. Project Growing Minds was attention-matched to Project HEART and had a similar stated completion time. That design makes the comparison relevant to the difference between the two websites' content, although the registry record does not report engagement or adherence measures that would clarify participants' actual exposure.

It also highlights why an endpoint needs careful interpretation. The primary measure is a mean of seven self-reported items about communication confidence. That construct can be important to an intervention's theory, but it is not equivalent to observed communication behavior or to the health outcomes named as ultimate goals. Statistical interpretation should stay anchored to what the scale directly measures.

The posted means are close, and the recalculated estimate has an interval crossing zero. This demonstrates the value of reporting an estimate and its uncertainty rather than relying on a single p-value. At the same time, the registry's regression result differs from the unadjusted recalculation, and the record does not give enough model information to understand why. Reporting the timepoint, analysis population, missing-data handling, and model specification would make the results more interpretable.

Finally, the sample-size illustration makes clear that planning depends on the effect researchers intend to detect: smaller assumed effects require larger planned groups under the stated assumptions. The record does not report the original design justification, so the illustration should not be mistaken for the trial's own planning rationale. Together, these points show why transparent methods and appropriately bounded conclusions are central to interpreting clinical and behavioral research.

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.

The principal reporting limitation is that the registry does not identify which planned assessment is represented by the posted means and comparison. Although the outcome timeframe includes an immediate post-test and 3-month follow-up, separate results are not reported for these assessments. The record therefore cannot establish the timing or persistence of the reported difference.

The registry provides little detail about the reported regression. Its covariates, model specification, analysis population, and confidence interval are not reported. The unadjusted recalculation and sponsor's regression result differ, but the record does not permit a specific explanation. A full protocol and statistical analysis plan are not reported in the registry record, including the original sample-size justification and any multiplicity strategy.

Missing-data handling and exclusions are also not reported. The outcome summaries list 107 participants in Project HEART and 114 in Project Growing Minds, while enrollment is listed as 222 and safety at-risk counts differ from the control outcome count. The record does not explain these differences, so no reason for them should be assumed. The randomization procedure and allocation concealment are not described, and the no-masking entry does not specify who knew allocation.

The primary endpoint is self-reported confidence, which is not a direct measure of communication behavior or health outcomes. The registry does not report adherence or engagement results, nor does it post secondary outcome results despite listing three outcome measures. Finally, serious adverse-event reporting is sparse: zero events and zero deaths are listed, but collection procedures and the safety observation period are not described. These limitations constrain interpretation to the specific posted results and prevent broader conclusions about intervention effects.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Reducing HIV Risk Among Adolescents: Evaluating Project HEART — NCT02579135.
• Registry record last updated 2018-10-16; status last verified 2018-09.
• Results first posted 2018-10-16.
• 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.

22. Record Summary

The registry record describes NCT02579135, a completed randomized, parallel-group, unmasked study of Project HEART and the attention-matched Project Growing Minds website. Enrollment is listed as 222. The primary outcome is a seven-item self-report of sexual communication self-efficacy, assessed immediately after intervention completion and at 3-month follow-up.

The posted arm means are 3.62 for Project HEART and 3.51 for the control, with standard deviations of 0.47 and 0.46. The unadjusted recalculation gives a mean difference of 0.11, a confidence interval from -0.013 to 0.233, and p = 0.079. The registry's linear regression result is beta = 0.05 with p = .32; model details and the represented timepoint are not reported.

The registry lists zero serious adverse events and zero deaths in both groups. It does not report the original sample-size justification, missing-data approach, analysis populations, multiplicity plan, or separate results by assessment timepoint. The evidence described is specific to the self-reported confidence measure and does not by itself establish effects on behavior or health outcomes.