Medication Development in Alcoholism: Suvorexant Versus Placebo: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Medication Development in Alcoholism: Suvorexant Versus Placebo (NCT04229095), a randomized phase 2 study. The primary outcome is Visual Analogue Scale (VAS) of Craving Severity: 2 Arms.

2. Clinical Question

3. Trial Design

This was a randomized, quadruple-masked, parallel phase 2 study sponsored by The Scripps Research Institute. The registry lists Alcohol Use Disorder (AUD) as the condition and gives enrollment as 26. The study status is completed. The recorded start was 2021-11-17 and primary completion was 2022-11-08. These dates describe the registry record; they do not by themselves establish when every analysis or report was completed.

The two arms received a single dose on an inpatient clinical research unit: suvorexant 20 mg or placebo. The primary endpoint was measured during a laboratory cue reactivity session, while a secondary outcome assessed standard drinks per day up to one week following single-dose administration. The study summary also refers to drinking under naturalistic conditions as an intended research question. The registry's outcome descriptions, rather than the broad hypothesis wording, define what is available for interpreting the posted results.

The summary says participants were stratified by sex and by Pittsburgh Sleep Quality Index total score, using the stated categories of > 5 versus <5. It explains that sleep disturbance was considered because suvorexant may be indicated for insomnia. The registry does not give further details about how those strata were incorporated into allocation or analysis. The randomization method, allocation concealment, protocol deviations, and details of masking are not reported in the registry record.

The primary arm summaries include the number analyzed for that outcome, and the registry gives separate arm summaries for the secondary standard-drinks outcome. It does not state whether these populations were defined by intention to treat, per protocol, or another rule. The trial's original sample-size justification and design assumptions are not reported in the registry record. Planning illustrations based on the observed comparison are a separate exercise and should not be mistaken for the trial's original design rationale.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Thirty-two subjects did not meet admission criteria and 12 subjects declined participation.

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 Visual Analogue Scale (VAS) of Craving Severity: 2 Arms, measured over 1 hour during cue reactivity session.

The primary endpoint is a Visual Analogue Scale (VAS) measure of craving severity during the cue reactivity session. It is defined as the VAS response to alcohol cues minus the VAS response to water cues. The registry describes the scale as running from 0 to 20, with higher scores indicating greater craving strength. The reported timeframe is 1 hour during the session.

A cue-based measure can capture a participant's immediate subjective response when exposed to alcohol-related cues in a controlled laboratory setting. Subtracting the response to water cues is intended to express craving specifically associated with alcohol cues relative to a comparison stimulus. This is a general explanation of the measure's logic, not a claim about the validity of this particular implementation.

The endpoint is a continuous score, and the registry reports group means with standard errors. The between-group mean difference therefore describes the average separation in the posted scores, while its confidence interval conveys uncertainty about that separation. The endpoint measures reported craving in the laboratory; it is not itself a measure of drinking behavior, longer-term recovery, or clinical benefit. The registry does not report a threshold for a meaningful change in this score, so the clinical importance of a particular difference is not specified in the registry record.

7. Results Posted in the Registry

• Belsomra,(Suvorexant): mean 2.38 versus 1.44 on control; difference 0.94 (95% CI -1.20 to 3.08), standardized effect d = 0.33; z-test p = 0.390.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Number of Standard Drinks Per Day: 2 Arms

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Belsomra,(Suvorexant): mean 3.59 versus 3.46 on control; difference 0.13 (95% CI -2.30 to 2.56), standardized effect d = 0.04; z-test p = 0.916.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Belsomra,(Suvorexant) and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists two secondary outcomes concerning standard drinks per day: a two-arm comparison and a combined-arms conditional model. The posted detailed arm summaries and unadjusted comparison concern the two-arm outcome, measured up to one week following the single dose. The registry does not report an analysis for the combined-arms conditional model, so the overall secondary pattern cannot be fully characterized from the posted results.

The available two-arm summary does not show a clear separation between the groups. This is broadly consistent with the primary result in the limited sense that neither posted comparison establishes a clear between-group difference. The outcomes address different questions, however: the primary endpoint measures laboratory cue-related craving, while the secondary drinking endpoint concerns drinking behavior after dosing. Similarity in the broad statistical pattern does not make the measures interchangeable or establish that one predicts the other.

Secondary outcomes are often interpreted with greater caution when several outcomes or analyses are considered. As a general statistical principle, examining multiple results can increase the chance that an apparently noteworthy result occurs by chance. The registry does not report a multiplicity adjustment, a hierarchy among secondary outcomes, or an analysis plan governing how these results should be interpreted. The secondary results therefore carry less weight as confirmatory evidence than a clearly prespecified primary analysis would.

The registry's reported analyses list is empty, and no detailed result is posted for the conditional model. The absence of that result should not be filled in by inference from the two-arm summary. These secondary outcomes can be treated as descriptive context for the primary finding, but the record does not support a complete evaluation of every secondary question listed.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Number of Standard Drinks Per Day: 2 Arms
• Number of Standard Drinks Per Day: Combined Arms Conditional Model

9. Statistical Interpretation

10. Statistical Methods Explained

The posted primary comparison uses continuous-outcome summaries: a mean for each arm and a mean difference between arms. The values are unadjusted recalculations from the arm-level summary statistics. They are useful for describing the observed separation and its uncertainty, but they should not be mistaken for a sponsor-reported adjusted analysis. The registry's reported analyses list is empty.

What does a mean difference represent?

A mean is the arithmetic average of the observed scores in a group. A mean difference compares those averages, with the direction determined by the order of subtraction. In the primary comparison, the stated difference is suvorexant minus placebo. A positive estimate means the posted suvorexant mean is higher; for this endpoint, higher scores indicate greater craving strength. This direction is a description of the scale and estimate, not a conclusion about causation or clinical importance.

The primary endpoint is a VAS score formed from alcohol-cue ratings minus water-cue ratings. The registry reports means of 2.38 and 1.44, and an unadjusted mean difference of 0.94. As a general statistical principle, a difference in observed means is an estimate subject to sampling variation. It does not describe every participant's score, and it does not show whether a difference would persist in another sample.

What does the standard error tell the reader?

The registry reports standard errors alongside the group means. A standard error describes the uncertainty of an estimated mean, rather than the spread of individual observations. In general, it reflects how much an estimate would vary across repeated samples under a statistical model. It should not be interpreted as a range containing most participant scores. The registry does not post the underlying participant-level data, so the distribution of individual scores cannot be assessed from these summaries alone.

For continuous outcomes, the comparison also uses a pooled standard deviation in its recalculation. A standard deviation describes variation among observations; pooling combines information about variation across the two groups under the calculation's assumptions. The comparison record indicates that this standard deviation was not approximated from the standard errors. Such a pooled summary supports a standardized effect description, but it does not replace the scale-specific mean difference or its confidence interval.

How should the confidence interval and p-value be read?

A confidence interval gives a range of estimates compatible with the observed data and the method used to construct it. In general, a wider interval indicates less precision than a narrower interval. The primary interval runs from -1.202 to 3.082. It spans both negative and positive differences, so the posted data are compatible with more than one direction of the group contrast. The confidence level and the exact interval procedure are not reported in the registry record.

The unadjusted p-value is 0.39. A p-value measures how unusual data at least as discrepant as those observed would be under a specified null model and its assumptions. It is not the probability that the null hypothesis is true, and it is not a measure of the size or clinical importance of an effect. Here, the value should be read alongside the estimate and interval, not in isolation. The registry does not report whether a different model or covariate adjustment was used for an official analysis.

The comparison is explicitly unadjusted. In general, an unadjusted comparison evaluates the group summaries without accounting for other variables in a regression model. The study summary describes stratification by sex and sleep-quality score, but the posted comparison does not document how those factors were incorporated. It would be inappropriate to infer an adjusted result from the description of stratification alone.

Why can over-dispersion matter?

For count or rate outcomes, over-dispersion means that observed variability exceeds what a simple statistical model assumes. A model that understates variability can make uncertainty look smaller than it is. This is a general statistical consideration, not a reported feature of this trial. The primary outcome is continuous, and the registry does not identify an over-dispersion assessment for the secondary drinking outcome. No model specification for the listed combined-arms conditional model is reported in the registry record.

The registry's planning illustration uses a normal approximation, equal allocation between arms, and a two-sided significance level of 0.05. Those are assumptions for that illustration only. The original design assumptions are not reported in the registry record. The planning calculation is based on a fraction of the observed effect; it does not show the probability that this completed trial would detect an effect. The resulting participant counts are planning illustrations, not observed power.

Finally, a sound interpretation also depends on how outcomes were handled, which participants entered each analysis, and whether analyses were prespecified. Missing-data procedures, analysis-population definitions, and the primary statistical analysis plan are not reported in the registry record. Without those details, the posted summary statistics can be described and their uncertainty explained, but aspects of the analysis process cannot be independently assessed.

11. Confidence Intervals and Precision

The primary mean difference is 0.94, with a confidence interval from -1.202 to 3.082. The interval is wide enough to include a negative difference as well as a positive one. In practical terms, the observed estimate is not precise enough to pin down the direction or magnitude of a possible group difference. The reported means and their standard errors provide context, but the interval is the more direct summary of uncertainty about the contrast.

Confidence intervals are generally interpreted as ranges generated by a specified procedure that, under its assumptions, would cover the target parameter at the procedure's stated rate over repeated samples. The confidence level and interval method are not reported in the registry record. The interval should therefore be quoted as posted without assigning it an unreported confidence level. Its inclusion of zero is consistent with no difference under the comparison's scale, but it does not prove equality between the arms.

The uncertainty matters because the primary endpoint group summaries come from limited outcome groups. The registry does not report a target precision, clinically important difference, or original power calculation against which to judge the interval. As a result, it is not possible to tell from the registry record whether the trial was designed to estimate a narrow range for craving or to test a particular minimum effect.

The estimate is also unadjusted and based on summary statistics. No reported adjusted interval or model-based result is available in the registry record. Therefore, the posted interval describes the recalculated unadjusted comparison, not necessarily every analysis that may have been undertaken. The conclusion should remain focused on the available estimate: the observed group means differ, but uncertainty around the difference is substantial.

12. 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 participants and not total events. They illustrate the group size that a hypothetical study might require under the listed assumptions when targeting the stated fraction of the observed effect. The rows describe planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. The columns describe participants per arm for 80% power and participants per arm for 90% power.

Power is the probability, under a specified model and assumed effect, that a study will reject its null hypothesis at the chosen significance level. In plain language, the 80% column represents the planning target for a design intended to detect the assumed effect with 80% power; the 90% column uses a higher power target. The registry's stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect.

The displayed requirements rise when the assumed target effect is smaller: planning for 75% or 50% of the observed effect calls for more participants per arm than planning for the full observed effect. This is a general feature of power planning, because smaller effects are harder to distinguish from sampling variation. The illustration is based on the observed estimate, which may itself be uncertain, and is not the original sample-size justification. The trial's original design assumptions are not reported in the registry record. These figures are planning illustrations, not observed power.

13. 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 = 0.940; pooled standard deviation σ = 2.873.
2. n = 2 × (1.96 + 0.842)² × 2.873² / 0.940² = 146.7, rounded up to 147 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 <- 0.94                      # observed difference in means
sigma <- 2.87344740686166                     # 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(4229095)
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 = 147 (80% power), 197 (90% power)
 75% of observed effect: n per arm = 261 (80% power), 350 (90% power)
 50% of observed effect: n per arm = 587 (80% power), 786 (90% power)
power.t.test: 147.7 
Simulated power at n = 147 per arm: 0.799 (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.

14. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary craving endpoint and lists secondary outcomes related to standard drinks per day. It does not report a procedure for controlling multiplicity, such as an adjustment across outcomes or a testing hierarchy. When more than one outcome or analysis is examined, the chance of obtaining an apparently unusual result by chance can increase. This is a general statistical concern; it does not establish that any particular posted result is spurious.

The primary comparison available in the record is an unadjusted recalculation from summary statistics. The registry's reported analyses list is empty, and the statistical analysis plan is not reported in the registry record. It is therefore unknown whether the posted comparisons match a prespecified analysis, whether other models were used, or whether the primary hypothesis was tested with adjustment for the described stratification factors.

Missing-data handling is also not reported. The record does not state whether observations were missing, how missing values were handled, or whether any sensitivity analyses assessed their effect. It would be inappropriate to assume complete follow-up or to infer a particular imputation method. The analysis-population definition is likewise not reported: the record does not identify the primary analysis as intention to treat, per protocol, or another population.

The registry gives outcome-specific group counts, but those counts alone do not explain participant flow, withdrawals, exclusions, or the relationship between enrollment and each analysis population. No details about protocol deviations or adherence are reported. These omissions limit assessment of possible bias and reproducibility. They do not change the arithmetic of the posted summaries, but they narrow what can be concluded about how the underlying data were selected and analyzed.

A full evaluation would require the prespecified analysis plan, details of missingness and participant flow, and the intended handling of multiple endpoints. Those details are not reported in the registry record, so conclusions should remain anchored to the posted unadjusted summaries and their uncertainty.

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.

Most frequent non-serious adverse event terms

For serious adverse events, the registry records 0 affected participants in the suvorexant arm and 0 in the placebo arm. The at-risk counts for these summaries were 14 and 12, respectively. The registry also records 0 deaths in each arm. These are counts recorded during the trial; they do not establish that serious events are impossible or that the study can rule out uncommon harms.

The registry lists 8 participants with other adverse events in the suvorexant arm and 7 in the placebo arm. It does not provide serious adverse-event terms, and the serious-event terms list is empty. The record therefore offers limited detail about the types, timing, severity, and attribution of events. The adverse-event summaries should be read as descriptive counts for this study, not as a comprehensive comparison of safety or as a basis for medical advice.

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

The posted primary results describe an observed difference in mean craving scores between the suvorexant and placebo arms. The estimated difference is positive, but the confidence interval extends from a negative value to a positive value, and the unadjusted p-value is 0.39. The data therefore leave substantial uncertainty about the direction and size of a possible difference. The result does not establish that suvorexant reduced craving, increased craving, or had no effect.

Nor does a non-significant comparison establish equivalence. Equivalence requires a defined margin and an analysis designed to assess whether the plausible effect lies within that margin. No equivalence margin or equivalence objective is reported in the registry record. A failure to demonstrate a difference and evidence that two treatments are sufficiently similar are distinct statistical conclusions.

The primary measure captures a reported response to alcohol cues relative to water cues in a laboratory session. It does not directly measure drinking, longer-term outcomes, or changes in alcohol use over an extended period. The secondary drinking outcome offers a separate short-term measure, but the registry does not report an analysis for the listed combined-arms conditional model. The available outcomes should not be combined into a broader claim about clinical benefit.

The study summary states hypotheses about craving after laboratory alcohol exposure and drinking under naturalistic conditions. Hypotheses describe what a study set out to examine, not what its results demonstrate. The posted summaries are limited by imprecision and by the absence of reported analysis-plan, missing-data, and analysis-population details. They also do not define a clinically meaningful threshold for the VAS difference.

Accordingly, the registry supports a careful descriptive account of the measured scores, the unadjusted comparisons, and the recorded adverse-event counts. It does not support a definitive claim that the intervention works or does not work, a claim of equivalence, or a general conclusion about safety. Those claims would require evidence and reporting beyond the information available in the registry record.

17. Why This Trial Matters Statistically

This trial is statistically instructive because its primary point estimate and its uncertainty need to be read together. A single estimated mean difference can appear to suggest a direction, yet a broad confidence interval may allow materially different interpretations. The interval is not a decorative addition to the result; it shows why a point estimate alone is an incomplete account of the evidence.

The study also illustrates the distinction between a registered research question and a reported analysis. The summary describes hypotheses about craving and drinking, and the outcomes include both a primary craving score and secondary drinking measures. However, the reported analyses list is empty, and the posted comparisons are unadjusted recalculations from group summaries. Clear identification of the prespecified model, analysis population, missing-data approach, and multiplicity procedure would make it easier to assess how the results answer the original questions.

Planning calculations based on an observed effect can help illustrate how assumptions about effect size and desired power influence a future study's scale. They are not a substitute for the original design rationale and cannot show the power of the completed trial. In this case, the registry explicitly identifies the figures as planning illustrations rather than the trial's own design calculation.

Finally, the record shows why descriptive adverse-event counts should be distinguished from comprehensive safety evidence. Zero recorded serious events in these groups is a report about this study's observations, not proof that serious events cannot occur. In combination, these points make the trial useful for learning how estimates, precision, design reporting, and evidential limits shape interpretation.

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 available primary comparison is unadjusted and based on summary statistics. The registry describes stratification by sex and sleep-quality score but does not explain how stratification was used in analysis. No adjusted estimate is reported in the registry record. The result therefore cannot be treated as a documented covariate-adjusted comparison.

The primary endpoint is a laboratory VAS score, and its clinical interpretation is limited by the absence of a reported threshold for a meaningful difference. The secondary drinking measure covers up to one week following the single dose. The record does not provide longer-term outcome results, and the listed combined-arms conditional model has no reported analysis. Generalizing beyond the measured timeframe and settings would not be supported by these results.

The record does not report detailed eligibility criteria, participant flow, adherence, withdrawals, or reasons for missing outcome data. It also does not define the analysis populations or describe missing-data methods. Those gaps prevent assessment of whether the participants represented in the outcome summaries align with the randomized population and whether missingness could affect the estimates.

The original sample-size justification, primary analysis plan, confidence-interval level, and multiplicity procedure are not reported in the registry record. The planning illustrations are based on the observed effect and stated assumptions, not on the trial's original design. Together with the limited group sizes and broad primary interval, these reporting gaps call for cautious interpretation rather than strong conclusions about the intervention or the precision of its effects.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Medication Development in Alcoholism: Suvorexant Versus Placebo — NCT04229095.
• Registry record last updated 2023-04-26; status last verified 2023-04.
• Results first posted 2023-04-26.
• 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

NCT04229095 is a completed randomized, quadruple-masked, parallel phase 2 trial comparing a single 20 mg dose of suvorexant with a single dose of placebo on an inpatient clinical research unit. The registry lists Alcohol Use Disorder (AUD), enrollment of 26, and The Scripps Research Institute as sponsor.

The primary outcome is a VAS measure of alcohol-cue craving relative to water-cue response during a cue reactivity session. Its unadjusted mean difference is 0.94, with a confidence interval from -1.202 to 3.082 and an unadjusted p-value of 0.39. The interval indicates substantial uncertainty. The registry also posts a secondary standard-drinks comparison and adverse-event counts, but does not report the listed conditional-model analysis or a detailed statistical analysis plan.