Examination of the Effectiveness of Suvorexant in Improving Daytime Sleep in Shift Workers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Examination of the Effectiveness of Suvorexant in Improving Daytime Sleep in Shift Workers (NCT02491788), a randomized phase 4 study. The primary outcome is Change in Average Total Sleep Time.

2. Clinical Question

The trial’s stated purpose was to test whether suvorexant before daytime sleep would improve objective and subjective measures of daytime sleep. The listed primary endpoint focuses on one objective measure: change in average total sleep time. The registry describes the sleep opportunity as daytime sleep following night-shift work, making the result specific to that setting and measurement.

The reported arm summaries give the mean change for each group, while the registry’s reported analysis gives a mixed-model mean difference and a p-value. The result therefore concerns the posted endpoint and analysis, rather than every possible aspect of sleep. Although the study summary also mentions sleep efficiency, wake after sleep onset, and subjective sleep quality, secondary posted outcomes are listed as none. Their results are not reported in the registry record.

The registry does not report participant characteristics in the primary outcome entry or explain whether the analysis population included all randomized participants. The findings should therefore be read as a record of the stated comparison, with those details left unresolved.

3. Trial Design

This was a randomized, triple-masked, parallel-group phase 4 study. It compared a drug arm receiving suvorexant with a placebo arm. The drug-arm description specifies 10 mg of suvorexant 30 minutes before the daytime sleep opportunity; the placebo arm received a placebo pill at the same stated interval. The registry lists enrollment as 19 and reports arm sizes of 8 for Drug and 11 for Placebo.

The study was conducted by the VA Palo Alto Health Care System. Its status is completed. The registry lists a start date of 2016-02-01 and a primary completion date of 2019-08-01. These dates describe the trial record; they do not establish when results became available or when an analysis was conducted.

The primary outcome is change in average total sleep time, measured in hours per sleep opportunity. The registry says daytime sleep would be examined from baseline to after 3 weeks. The endpoint description refers to the average amount of sleep during daytime sleep episodes following night-shift work, compared with baseline.

Randomization means assignment to treatment groups was intended to be by chance, while parallel-group design means the groups received their assigned interventions rather than crossing over within the described design. Triple masking indicates masking at three levels, but which people or roles were masked is not reported in the registry record. Details of sequence generation, allocation concealment, adherence, protocol deviations, and the exact procedures used to maintain masking are also not reported in the registry record.

The registry identifies one primary outcome measure and one reported analysis. It does not report the planned statistical analysis in enough detail to establish whether the posted model followed a prespecified protocol or whether other analyses were conducted.

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 Change in Average Total Sleep Time, measured over Daytime sleep will be examined from baseline to after 3 weeks.

The primary endpoint is change in average total sleep time during daytime sleep episodes after night-shift work, relative to baseline. It is a continuous measure expressed in hours per sleep opportunity. In practical terms, it captures the amount of sleep recorded during the daytime sleep opportunity, averaged as described in the registry.

Change from baseline is used to describe how a measurement at follow-up differs from the participant’s baseline measurement. Comparing changes between groups can account for the fact that people may begin the study with different sleep patterns. The registry reports arm-level mean changes and a mixed-model comparison, but does not provide details about the exact model specification or how baseline values entered the model.

Total sleep time is an objective measure of sleep duration. The study summary also identifies sleep efficiency, wake after sleep onset, and subjective sleep quality as outcomes of interest, but the registry lists no secondary posted outcomes. Thus, the reported primary result addresses sleep duration; it cannot by itself describe sleep continuity, perceived quality, or other dimensions of daytime sleep.

The endpoint is specific to the daytime sleep opportunity after night work and to the assessment described as occurring from baseline to after 3 weeks. The registry does not report the measurement device or procedure, the number of sleep opportunities contributing to each average, or how incomplete sleep measurements were handled. Those details matter for understanding measurement precision and the interpretation of an average, but are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What do the arm summaries represent?

The endpoint is continuous: it is measured on a scale of hours per sleep opportunity. For a continuous outcome, a mean describes the group’s average value, while a standard error describes uncertainty in that estimated average. The reported Drug mean change is 1.83, with standard error 0.62; the Placebo mean change is -0.33, with standard error 0.49. These values summarize changes from baseline in the respective groups. They do not show how values varied among individual participants.

A standard error should not be mistaken for a standard deviation. A standard deviation describes variability among observations, while a standard error concerns the precision of an estimated mean. The registry identifies the reported dispersion as standard error. It does not report standard deviations or participant-level observations for this endpoint.

What does the mixed-model result say?

The sponsor reports a Mixed Models Analysis with a mean difference in final values of 2.16 and p < 0.05, classified as superiority. In general, a mixed model can represent relationships among repeated observations and account for dependencies among measurements, depending on its specification. The registry does not report the model’s terms, covariance structure, treatment of baseline, or other implementation details. It is therefore not possible to infer those choices from the method label alone.

The registry identifies the comparison groups as Drug and Placebo. It does not give a confidence interval, a confidence level, or further details on the test used to obtain the p-value. A p-value summarizes how unusual the observed data, or more extreme data, would be under a specified null model and its assumptions. It is not an estimate of the probability that the result will replicate, and it does not describe the range of plausible effect sizes.

Why does a confidence interval matter?

In general, a confidence interval combines an estimate with a measure of its uncertainty. A narrow interval indicates greater precision under the method’s assumptions; a wide interval indicates less precision. The interval can also help readers assess whether effects of different magnitudes remain compatible with the data. Because no interval is reported for the mixed-model mean difference, the record does not show that range of uncertainty for the sponsor’s comparison.

What is and is not recalculated?

A recalculated comparison would be a simple, unadjusted analysis derived from posted group summaries. Such a calculation may be useful as a check, but it is not the same as the sponsor’s fitted model. In this case, a two-arm recalculation is not available: the registry record indicates that values were missing or not estimable for a continuous endpoint comparison. No recalculated effect or p-value should therefore be inferred from the arm means and standard errors.

When a recalculation is available, it should be treated as an independent approximation based on summary statistics, not as a replacement for the trial’s analysis. Here there is no such estimate to set alongside the sponsor’s result. The sponsor’s analysis remains the only posted inferential comparison, and the registry does not provide enough model detail for an independent reproduction.

How should planning illustrations be understood?

In general, a sample-size illustration asks how many participants per arm might be needed under specified assumptions to detect an effect at a chosen significance level and power. Such an illustration is conditional on its assumptions, including the target effect and the method used to model uncertainty. It is not evidence that a completed trial had that level of power, and it cannot establish the probability that the study detected an effect.

The record describes planning illustrations as based on the observed effect, with two-sided significance level 0.05, equal allocation, and a normal approximation; the effect to detect is expressed as a fraction of the observed effect. These are planning assumptions, not a report of the trial’s original design rationale. The protocol’s sample-size justification is not reported in the registry record. The illustrations should be understood as hypothetical planning quantities, not as observed power or as a validation of the trial’s design.

What statistical details remain unknown?

The registry reports a mixed-model method but does not report the statistical analysis plan, the model’s full specification, the confidence interval, or the handling of missing measurements. It also does not report whether any interim analyses were planned or conducted, or whether the analysis population was defined as all randomized participants, participants who received treatment, or another group. These omissions limit reproduction and appraisal of the result. General statistical principles can explain what the reported measures mean, but they cannot fill in those trial-specific details.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Drug", "Placebo")
n   <- c(8, 11)
m   <- c(1.83, -0.33)
sp  <- c(0.62, 0.49)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Drug  8  1.830  1.754  0.364  3.296
CARM  Placebo  11  -0.330  1.625  -1.422  0.762

11. Analyses Reported by the Sponsor

The registry reports the sponsor’s mixed-model comparison as a mean difference in final values of 2.16, with p < 0.05, and classifies the result as superiority. It does not report a confidence interval for that estimate. The arm summaries separately give mean changes from baseline and their standard errors, but those summaries are not a substitute for the fitted mixed-model comparison.

A simple recalculation from posted group summaries is not available because the required values are reported as missing or not estimable for a two-arm continuous-endpoint comparison. There is therefore no recalculated estimate against which to check the sponsor’s result. The registry record does not support claims that the model result was reproduced, contradicted, or confirmed by an unadjusted recalculation.

The sponsor’s own analysis is the only reported inferential analysis for this endpoint. The record does not provide sufficient model details to reproduce it independently. The reported p-value indicates the test result under that analysis, but the absence of a confidence interval and fuller model specification limits assessment of precision and reproducibility.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure and one reported analysis, and it lists no secondary posted outcomes. The study summary mentions sleep efficiency, wake after sleep onset, and subjective sleep quality in addition to total sleep time, but their results are not reported in the registry record. The record also does not explain whether these measures were formally designated secondary endpoints in the protocol or whether they were analyzed.

Multiplicity arises when several outcomes, comparisons, time points, or analyses are tested. In general, conducting many tests can increase the chance of obtaining at least one apparently notable result by chance. Statistical plans may address this through prespecified outcome hierarchies or adjustment procedures. Whether any such approach was used is not reported in the registry record. With no posted secondary results and no stated multiplicity plan, the primary result cannot be interpreted in the context of a documented family of analyses.

The registry does not report missing-data counts, reasons for missingness, imputation methods, or sensitivity analyses. It also does not report which participants were included in the primary analysis or how treatment discontinuation and protocol deviations were handled. These details are important because analyses based on different participant sets or missing-data assumptions can produce different estimates.

The arm sizes are reported as 8 for Drug and 11 for Placebo, and the registry enrollment is 19. The available record does not establish whether every enrolled or randomized participant contributed usable endpoint data. No assumption about complete follow-up or analysis population should be made from the group summaries alone.

In general, missingness is not only a matter of how many observations are absent; why they are absent can also matter. If missing measurements relate to sleep outcomes or treatment experience, an analysis that excludes them may not represent the originally assigned groups. The record does not provide the information needed to assess that possibility or to determine the robustness of the reported result to alternative missing-data assumptions.

13. 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

The registry reports no serious adverse events in either arm and no deaths. For other adverse events, it reports 0 in the Drug arm and 2 in the Placebo arm. The corresponding at-risk counts are 8 for Drug and 11 for Placebo. These are the posted event counts, not estimates of comparative risk.

Serious adverse events and other adverse events are presented as separate categories in the record. No serious-event terms are listed. The record does not provide event descriptions for the other adverse events, their timing, severity, duration, relationship to treatment, or whether any participant experienced more than one event.

Counts alone do not establish that one treatment is safer than the other. The registry does not report a formal statistical comparison of adverse events, and the available counts are limited to the listed groups and categories. When event numbers are small, uncertainty around comparative estimates can be substantial; a count should not be read as a precise measure of risk.

These safety figures are descriptive registry results. They do not establish that no unrecorded events occurred, nor do they settle the safety profile beyond the trial and its reported follow-up. The record does not provide enough detail to assess completeness of adverse-event ascertainment.

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

The registry records a randomized, triple-masked comparison of suvorexant and placebo for daytime sleep in shift workers. For the primary endpoint, it reports mean changes in total sleep time for both groups and a sponsor-reported mixed-model mean difference of 2.16 with p < 0.05. The sponsor classifies the comparison as superiority. These statements describe what the posted results report; they do not establish a treatment recommendation or determine what an individual should expect.

The inferential result has important limits. The registry does not provide a confidence interval for the mixed-model estimate, so the precision and range of effect sizes compatible with the analysis are not shown. The model is not described in enough detail for independent reproduction. A recalculated two-arm comparison is unavailable from the posted summaries, so there is no independent unadjusted estimate to compare with the sponsor’s analysis.

The primary endpoint addresses average total sleep time during daytime sleep opportunities after night-shift work, from baseline to after 3 weeks. It does not directly answer questions about sleep efficiency, wake after sleep onset, subjective sleep quality, longer-term outcomes, or other populations. Although the study summary mentions some of these sleep measures, no secondary outcomes are posted in the record.

The record also leaves analysis-population and missing-data details unresolved. It does not report whether all randomized participants contributed to the primary analysis, how missing sleep measurements were handled, or how adherence and deviations from the intended intervention were addressed. Those uncertainties affect how fully the reported comparison can be interpreted.

In short, the posted result is a statistically reported superiority finding for the specified primary endpoint, with limited information about its precision and analysis details. It should be read as a finding within this trial record, not as a complete account of all outcomes or as proof of effects beyond the study setting.

15. Why This Trial Matters Statistically

This trial is a useful example of why a reported p-value is only one part of statistical interpretation. The registry gives a mixed-model estimate and p-value, but no confidence interval. Readers can identify the reported direction and significance classification while still lacking a direct view of the estimate’s precision. Statistical significance alone does not show whether an estimate is precise or how a range of plausible effects should be understood.

The record also illustrates the difference between arm summaries and a model-based comparison. The group means describe changes from baseline, with standard errors for those means. The sponsor’s analysis is described as a mean difference in final values from a mixed model. Without the model specification, readers cannot fully reconcile the two presentations or reproduce the sponsor’s analysis. Summary statistics and model results answer related but not necessarily identical questions.

Another important issue is the boundary between a trial’s stated scientific question and its posted outcome record. The study summary refers to several dimensions of sleep, but the registry posts a primary endpoint and no secondary outcomes. Interpreting one endpoint should not be treated as evidence about unposted measures.

Finally, the missing details are themselves relevant statistical information. The record does not report a confidence interval, the analysis population, missing-data methods, or a full statistical plan. These omissions constrain independent review and make it harder to judge how robust the result is to analytic choices. Transparent reporting of estimates, uncertainty, and analysis decisions helps readers distinguish a reported finding from the broader questions it cannot resolve.

16. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several trial-specific limitations of interpretation arise from what the registry does not report. The primary outcome description identifies total sleep time and the assessment period, but the measurement procedure and instrument are not reported. The record does not state how many sleep opportunities contributed to each average or how measurements were aggregated for each participant. Those details would help readers assess the outcome’s measurement properties.

The sponsor reports a mixed-model comparison but does not provide the model specification, confidence interval, or further details about the test. The registry does not explain how baseline measurements were handled in that model or whether additional covariates were included. The reported arm summaries are changes from baseline, while the analysis is described as a mean difference in final values; the relation between these summaries and the modeled estimate is not detailed.

The registry does not report the analysis population, adherence, treatment discontinuation, protocol deviations, missing outcome data, or methods for handling missingness. It also does not report a sample-size justification or a prespecified statistical analysis plan. Planning illustrations based on an observed effect are not evidence of the original design rationale or of observed power.

Masking is described as triple, but the identities of the masked roles and the procedures used to maintain masking are not reported. Participant characteristics and the details needed to assess how broadly the results might apply are also not reported in the registry record. The completed status and reported dates do not fill these gaps.

Finally, the study summary mentions sleep outcomes beyond the primary endpoint, but no secondary outcomes are posted. The absence of those results means the record does not support conclusions about the broader set of sleep measures named in the summary.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Examination of the Effectiveness of Suvorexant in Improving Daytime Sleep in Shift Workers — NCT02491788.
• Registry record last updated 2020-07-01; status last verified 2020-06.
• Results first posted 2020-06-16.
• 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.

20. Record Summary

NCT02491788 is a completed randomized, triple-masked, parallel-group phase 4 trial sponsored by the VA Palo Alto Health Care System. The registry lists enrollment as 19 and compares suvorexant with placebo in shift workers with sleep disorder related to shift work.

The primary endpoint is change in average total sleep time during daytime sleep episodes after night-shift work, from baseline to after 3 weeks. The arm summaries report mean changes of 1.83 hours per sleep opportunity for Drug and -0.33 for Placebo. The sponsor reports a mixed-model mean difference in final values of 2.16 with p < 0.05, classified as superiority; no confidence interval is reported.

A recalculated two-arm comparison is unavailable from the posted summaries. The registry does not report the analysis population, missing-data methods, full model details, or secondary outcome results. No serious adverse events or deaths are reported; other adverse-event counts are 0 for Drug and 2 for Placebo.