Promoting Benzodiazepine Cessation Through an Electronically-delivered Patient Self-management Intervention: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Promoting Benzodiazepine Cessation Through an Electronically-delivered Patient Self-management Intervention (NCT04572750), a randomized na study. The primary outcome is 25% Reduction in Benzodiazepine Use.

2. Clinical Question

This is a question about a defined change in medication dose, rather than complete cessation alone: a participant meets the endpoint after a reduction of at least one quarter, whether or not benzodiazepine use has stopped. The endpoint is reported as a count of participants, and the comparison expresses the arm contrast as a risk difference. The record also lists anxiety, sleep quality, and overall health and quality of life as secondary outcomes. It does not report the measures or results for those outcomes in the registry record for this analysis.

The study is described as randomized, with a parallel-arm design and single masking. The registry record does not identify who was masked. It reports an enrollment of 170 and an outcome population that excludes nine patients described as ineligible. The central statistical question is therefore how much the posted arm-specific outcome counts support a difference at the stated follow-up, and how much uncertainty remains around that estimate. The registry record does not report enough detail to assess all aspects of eligibility, implementation, or outcome ascertainment.

3. Trial Design

The registry identifies this as a randomized, parallel-group study with single masking. Participants were assigned to EMPOWER-ED, an electronic self-help app, or care as usual. The app was available on each participant’s preferred platform and focused on reducing benzodiazepine use. The masking designation does not say which people were masked, and that detail is not reported in the registry record.

The listed enrollment is 170. The registry record gives an outcome population description stating that nine patients were excluded because they were not eligible. The primary outcome arm summaries include 82 participants for EMPOWER-ED and 79 for control. The record does not explain the relationship between enrollment, eligibility exclusions, and the arm-specific outcome counts beyond the stated exclusion. It also does not report how participants were recruited, the eligibility criteria in full, or the reasons for any other differences in the reported populations.

The primary endpoint is assessed at 6 months and counts participants whose benzodiazepine dose fell by at least one quarter. The study is marked completed, with a start date of 2022-06-01 and primary completion date of 2024-07-03. Phase is listed as NA. The sponsor is the VA Office of Research and Development.

Three secondary outcomes are named: anxiety, sleep quality, and overall health and quality of life. The registry lists five outcome measures, but detailed results for those secondary outcomes are not reported in the record details used. The registry record also does not report the original sample-size justification, planned analysis model, allocation concealment procedures, masking details, or whether interim analyses were planned. These gaps limit what can be concluded about the conduct and prespecified statistical approach beyond the design characteristics explicitly recorded.

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: We excluded nine patients because they were not eligible.

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.

Population: Data not available for 9 participants due to ineligibility

6. Primary Endpoint

The registry lists the primary outcome as 25% Reduction in Benzodiazepine Use, measured over 6 months.

The primary endpoint counts participants who reduced their benzodiazepine dose by one quarter or more by 6 months. It captures a thresholded change in dose: participants either meet the stated reduction criterion or do not. The registry describes the measure as a count of participants, not as a continuous measure of dose change or a time-to-event outcome.

A threshold can make an outcome easier to interpret as a participant-level milestone, but it also compresses information. A participant with a reduction just under the threshold is classified differently from someone who reaches it, while the size of reductions beyond the threshold is not distinguished in the posted endpoint. The registry record does not report whether dose reductions were measured through prescriptions, participant report, clinical records, or another method. It also does not give additional details about how dose changes were handled when participants used more than one benzodiazepine.

The measure is relevant to the study’s aim of promoting reduced benzodiazepine use, but it does not by itself establish complete cessation. Meeting the endpoint means a reduction of at least one quarter, not necessarily stopping medication. Conversely, participants who do not meet this threshold may still have changed their use by a smaller amount. The registry record specifies the 6-month timeframe but does not report other assessment timepoints or the handling of missing endpoint information.

For the arm comparison, the endpoint is expressed as the proportion of participants meeting the criterion and as a risk difference. This permits a direct contrast between the randomized groups on the stated outcome. Interpretation still depends on the eligibility and analysis populations, outcome ascertainment, and missing-data handling; those details are not reported in the registry record.

7. Results Posted in the Registry

• EMPOWER-ED: 17.1% (14/82) versus 7.6% (6/79) on control; risk difference 9.5 percentage points (95% CI -0.5 to 19.5), risk ratio 2.25 (0.91–5.56); unadjusted two-proportion z-test p = 0.068.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary at the participant level: each person either meets the reduction threshold or does not. Although the registry reports counts, comparing counts alone would not account for different group sizes. The posted comparison therefore includes arm-specific proportions and a risk difference. The risk difference subtracts the control proportion from the intervention proportion; its sign indicates the direction of the contrast. The reported estimate is 0.095.

What does the risk difference describe?

The risk difference is an absolute contrast on the proportion scale. It answers how far apart the two arm-level outcome proportions are, rather than how many times larger one is than the other. Its confidence interval runs from -0.005 to 0.195. Since this interval crosses the no-difference value for a difference, the compatible values include a small contrast favoring control as well as contrasts favoring EMPOWER-ED. The interval also indicates that the point estimate should not be treated as an exact description of the underlying effect.

The registry reports a relative risk of 2.248 and an odds ratio of 2.505 as additional comparisons. A relative risk compares proportions as a ratio. An odds ratio compares the odds of the endpoint between arms, rather than the proportions directly. Both measures are relative scales, so their null value differs from the null value for a risk difference. The reported intervals for these measures extend from 0.909 to 5.557 for the relative risk and from 0.911 to 6.89 for the odds ratio. Because both intervals include their ratio-scale null value, they do not rule out no relative contrast at the interval’s stated confidence level. The confidence level itself is not reported in the registry record.

What does the p-value add?

The unadjusted p-value is 0.068. In general, a p-value measures how unusual results at least as far from the null as the observed result would be if the null model and the test assumptions held. It is not a measure of the probability that the null hypothesis is true, and it does not indicate the size or practical importance of a difference. The registry record does not report the full test specification or the trial’s analysis threshold. The p-value is therefore best read together with the effect estimate, interval, endpoint definition, and analysis-population details.

The posted comparison is labelled unadjusted. That indicates the reported p-value is not presented as an adjusted result, but the registry entries do not specify which covariates, if any, were considered in other analyses. The record lists no separate analyses in its analysis count, and it does not report whether a prespecified model was used. No adjusted estimates or p-values are available in the record details used.

Why do the outcome counts matter?

The arm summaries report 14 participants meeting the endpoint among 82 in EMPOWER-ED and 6 among 79 in control. These are the outcome counts and denominators used for the posted comparison. The primary outcome description says nine patients were excluded because they were not eligible. The registry record does not describe the allocation or outcome status of those excluded patients, or whether the analysis followed an intention-to-treat principle. Those details are important when assessing whether the reported comparison preserves the benefits of randomization.

Since the endpoint is binary, the analysis treats participants who reach the reduction threshold alike, regardless of how far beyond the threshold their doses changed. This is appropriate for answering the registry’s stated threshold question, but it discards information about the magnitude of dose change. It also does not tell whether participants completely ceased use. These are limits of the endpoint’s definition rather than errors in the reported comparison.

What is not described?

Details of the trial’s own statistical plan are not reported in the registry record. This includes the original sample-size justification, planned primary analysis, handling of missing data, adjustment strategy, interim analyses, and the intended confidence level. The separate planning illustration is based on the observed effect and does not reveal the assumptions used to design the completed trial. The registry record also does not describe how the primary outcome was ascertained or whether assessors were masked. Without those details, the numerical estimates can be explained, but the full statistical conduct cannot be independently reconstructed from the registry record.

10. Confidence Intervals and Precision

Precision concerns how tightly an estimate is determined by the observed data. The risk-difference estimate is 0.095, and its confidence interval extends from -0.005 to 0.195. The interval includes the no-difference value and reaches both a small contrast favoring control and a larger contrast favoring EMPOWER-ED. That range signals uncertainty about the exact size and direction of the underlying difference.

A confidence interval is a range generated by a statistical procedure that, under its assumptions, is intended to capture the parameter in repeated use. It is not a range containing a stated probability that the true effect lies inside it after the data have been observed. The confidence level and interval method are not reported in the registry record, so the interval should be interpreted as posted rather than reconstructed.

The ratio estimates also have broad intervals: the relative risk is 2.248, with an interval from 0.909 to 5.557, and the odds ratio is 2.505, with an interval from 0.911 to 6.89. Each includes its ratio-scale null value, so the posted estimates do not precisely distinguish a relative contrast from no relative contrast. Relative measures can appear numerically large while still being uncertain, particularly when outcome counts are limited; the registry record reports the counts but does not provide a precision target or an original design justification.

Precision is not the same as importance or validity. A narrow interval would not address possible bias in eligibility, outcome measurement, or missing-data handling, while a wide interval does not prove there is no meaningful difference. Here, the interval leaves room for a range of possible contrasts. The registry record does not report enough methodological detail to determine whether the interval reflects the planned analysis or which assumptions were used to calculate it.

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 sample-size figures are planning illustrations based on the observed effect, not the trial’s original design assumptions. The assumptions listed for this illustration are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The target effect is set at the stated fraction of the observed effect. These calculations are not observed power.

The figures are participants per arm, not total participants. For planning at 100% of the observed effect, the illustration gives 188 participants per arm for 80% power and 251 participants per arm for 90% power. At 75% of the observed effect, it gives 307 participants per arm for 80% power and 410 participants per arm for 90% power. At 50% of the observed effect, it gives 626 participants per arm for 80% power and 838 participants per arm for 90% power.

In plain terms, the 80% power column gives the planned number per arm that would yield an 80% chance of detecting the specified effect under the calculation’s assumptions, if that effect were the true one. The 90% column uses a higher target chance of detecting that effect and therefore requires more participants per arm in these illustrations. The values are hypothetical planning requirements, not a description of the trial’s observed enrollment or its probability of finding a result.

Smaller effects are harder to distinguish from random variation, so planning for a smaller fraction of the observed effect requires more participants per arm in this illustration. The original study’s sample-size assumptions, target effect, and power calculation are not reported in the registry record. The planning curve represents this same general relationship; it should be read as a design illustration rather than as additional trial outcome evidence.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.076; treatment proportion p₁ = 0.171; average p̄ = 0.123.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.4650 = 0.9114.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.4602 = 0.3873.
4. n = (0.9114 + 0.3873)² / (0.0948)² = 187.8, rounded up to 188 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
p0 <- 0.0759493670886076                          # control proportion
p1 <- 0.170731707317073                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(4572750)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 188 (80% power), 251 (90% power)
 75% of observed effect: n per arm = 307 (80% power), 410 (90% power)
 50% of observed effect: n per arm = 626 (80% power), 838 (90% power)
power.prop.test: 187.8 
Simulated power at n = 188 per arm: 0.811 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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. Multiplicity, Missing Data and Analysis Populations

The registry lists three secondary outcomes: anxiety, sleep quality, and overall health and quality of life. It also lists five outcome measures, but the registry entries available for this analysis do not include results for those secondary outcomes. The registry record does not report whether additional outcomes or subgroup analyses were planned, which outcomes were designated primary at the protocol stage, or whether the primary outcome definition changed over time.

When several outcomes or analyses are examined, the chance of at least one apparently notable result can increase. Statistical approaches to multiplicity can address this, but the appropriate approach depends on the prespecified analysis plan and the scientific questions. Whether a multiplicity adjustment was planned or used is not reported in the registry record. The unadjusted p-value of 0.068 should therefore not be taken as evidence that any multiplicity procedure was or was not applied elsewhere.

Missing-data handling is also not reported. The primary outcome description says nine patients were excluded because they were not eligible, but it does not say how missing 6-month outcomes were handled, whether any outcomes were missing, or whether imputation or sensitivity analyses were used. The registry record does not identify the primary analysis population as intention-to-treat, per-protocol, or another defined set. As a result, the reported arm denominators cannot by themselves establish how all randomized participants were accounted for.

The registry record does not report a detailed statistical analysis plan, interim analyses, stopping rules, or procedures for addressing multiple comparisons. These omissions do not change the posted counts or estimates, but they limit assessment of whether the analysis followed a prespecified approach. They also constrain interpretation of the p-value and any possible selective reporting. No claims about the presence or absence of interim monitoring or multiplicity adjustments can be made from the record details available.

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.

The adverse-event record reports serious adverse events affecting 0 participants in EMPOWER-ED and 0 in control, with 85 at risk in each arm for this measure. It also reports other adverse events affecting 0 participants in EMPOWER-ED and 0 in control, again with 85 at risk in each arm. The record lists 0 deaths in EMPOWER-ED and 2 in control.

These are counts in the registry record, not estimates of comparative risk adjusted for follow-up or other factors. The record does not report event timing, causes of death, whether deaths were considered related to study participation, or additional clinical details. Zero reported affected participants does not establish that no events occurred outside the recorded definitions or reporting period.

The serious-event term list is empty. The record therefore provides no serious-event terms to describe. These sparse safety counts should be interpreted as registry reporting, not as proof of equivalence between groups or a causal explanation for the deaths. Further details about ascertainment and the safety analysis are not reported in the registry record.

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

The posted primary result is a comparison of the proportion of participants meeting a specific dose-reduction threshold at 6 months. The point estimate is higher in the EMPOWER-ED arm, and the risk-difference estimate is 0.095. Its confidence interval, from -0.005 to 0.195, includes the possibility of a small contrast in the opposite direction as well as a larger contrast favoring EMPOWER-ED. The unadjusted p-value is 0.068. Together, these figures describe an estimate with meaningful uncertainty, not a definitive answer about the underlying difference.

The endpoint does not measure complete discontinuation. It counts a participant as meeting the outcome after a reduction of one quarter or more, and it does not distinguish the amount of reduction beyond that threshold. The findings also do not tell whether any dose change was maintained after 6 months. The registry lists anxiety, sleep quality, and overall health and quality of life as secondary outcomes, but results for those outcomes are not included in the available the registry entries.

Random assignment is intended to support a fair comparison between arms, but interpretation also depends on how eligibility, outcome measurement, and missing data were handled. The record says nine patients were excluded as ineligible and gives outcome arm sizes, but it does not report a full account of the analysis population or the missing-data approach. It also does not identify who was masked in this single-masked study.

These results should not be taken as proof that the app caused a difference, nor as proof that it had no effect. The interval and p-value do not settle the size or direction of the underlying contrast. The registry record also does not support conclusions about individual care, side effects beyond the recorded counts, or the broader experience of people not represented in this trial. The statistical result is specific to the recorded endpoint, population, follow-up, and comparison.

16. Why This Trial Matters Statistically

This trial illustrates why a point estimate should be read together with its uncertainty. The estimated risk difference is 0.095, but its interval extends from -0.005 to 0.195. The estimate points toward a higher proportion meeting the reduction threshold in the app arm, while the interval does not pin down the direction or magnitude with precision. Reporting the interval alongside the point estimate makes that uncertainty visible.

The endpoint also shows how statistical conclusions depend on outcome definition. A reduction of at least one quarter is an interpretable threshold, but it is not the same as complete cessation, and it does not preserve the full magnitude of dose changes. The analysis answers the registry’s stated threshold question, not every question about medication use, sustained reduction, or well-being.

The sample-size illustration highlights the relationship between effect size, desired power, and the number of participants needed under specified assumptions. It is based on the observed effect and cannot be mistaken for the original design calculation or for evidence about the trial’s observed power. The original sample-size justification is not reported in the registry record.

Finally, the record demonstrates the importance of complete reporting. Details about the analysis population, missing data, masking, outcome ascertainment, and multiplicity are not reported. Without them, readers can interpret the posted estimates but cannot fully evaluate how sensitive they might be to design and analysis choices. Statistical interpretation is strongest when the numerical result and the methods needed to understand it are both available.

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.

The registry record does not identify who was masked in the single-masked design. This matters because the intervention is an app and the comparator is usual care, but the record does not describe whether participants, outcome assessors, or other study personnel were masked. The details of usual care and the intervention’s implementation, including engagement with the app, are also not reported in the registry entries.

The primary outcome description states that nine patients were excluded because they were not eligible. It does not provide the eligibility criteria in full or explain how those exclusions relate to randomization and the analysis population. The arm summaries report the participants included for the outcome, but the registry record does not describe the handling of missing 6-month information or whether the analysis retained participants in their assigned groups.

Outcome ascertainment is not described. The record specifies a reduction of one quarter or more but does not state how dose changes were measured or verified. The binary threshold also does not distinguish smaller changes from no change, or larger reductions from complete cessation. The posted primary outcome therefore cannot answer questions about the size or durability of dose changes beyond the stated threshold.

The full statistical analysis plan, original sample-size justification, multiplicity strategy, interim-analysis plan, and confidence-interval method are not reported in the registry record. The secondary outcomes are named, but their results are not available in the registry record used. These omissions limit assessment of prespecification, missing-data sensitivity, and the broader balance of outcomes. The safety record gives counts but not event timing, causes, or relatedness, so it cannot explain the reported deaths or establish comparative safety.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Promoting Benzodiazepine Cessation Through an Electronically-delivered Patient Self-management Intervention — NCT04572750.
• Registry record last updated 2026-01-23; status last verified 2026-01.
• Results first posted 2025-05-23.
• 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

NCT04572750 is a completed randomized, single-masked, parallel-group study of EMPOWER-ED, an electronic self-help app focused on reducing benzodiazepine use, compared with care as usual. The listed enrollment is 170. The primary outcome is the count of participants who reduced their benzodiazepine dose by one quarter or more at 6 months.

The registry reports 14 outcome participants in the EMPOWER-ED arm and 6 in control meeting the threshold. The risk-difference estimate is 0.095, with a confidence interval from -0.005 to 0.195; the unadjusted p-value is 0.068. The outcome description states that nine patients were excluded because they were not eligible.

Anxiety, sleep quality, and overall health and quality of life are listed as secondary outcomes. The record also reports adverse-event counts, including 0 serious adverse events affecting participants in either arm and 2 deaths in control. Important details about the original statistical plan, missing data, outcome ascertainment, and analysis populations are not reported in the registry record.