Trial of Adherence App for Buprenorphine Treatment (TAAB) Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Trial of Adherence App for Buprenorphine Treatment (TAAB) Study (NCT03779997), a randomized na study. The primary outcome is Percentage Opioid Negative Urine Tests.

2. Clinical Question

The clinical question is about a behavioral strategy intended to support medication-taking, not a comparison of buprenorphine with another medication. The intervention added video-based confirmation to office-based treatment; the comparison arm continued usual care. Because the study was randomized, its intended comparison is between participants assigned to these two approaches. However, the posted primary outcome is recorded as counts of units, rather than as arm-level percentages with denominators. That limits what can be concluded from the posted primary results alone about the percentage of tests that were negative.

The registry describes the study's purpose as pilot testing the application. It does not report details about recruitment, eligibility, the settings involved, or how the application was used beyond daily video confirmation. Those omissions matter when considering which patients or treatment settings the findings might represent. The registry also does not report a primary outcome estimate with a confidence interval that can be interpreted directly as a difference in percentages. The reported model result should therefore be read alongside the outcome definition and the form in which the arm values are recorded.

3. Trial Design

TAAB is a completed, randomized, parallel-group study with no masking. The registry records the study as phase NA and lists the University of Washington as sponsor. Its stated purpose was to pilot test a smartphone application for video-based directly observed therapy among participants receiving buprenorphine treatment for opioid use disorder in office-based settings.

The two assigned approaches were the Video-based DOT Application and Treatment as Usual (TAU). The application arm involved daily videos confirming medication ingestion. The brief description says the comparison arm continued with care as usual. More detailed descriptions of either arm are not reported in the registry record. The registry lists 39 participants in each arm and an enrollment of 78. These are the recorded group sizes; the registry does not provide further information about the analysis population or participant flow.

The recorded study start date is 2019-02-15, and primary completion is recorded as 2020-05-15. The study status is completed. The primary outcome timeframe runs from baseline to 12 weeks post-randomization. The secondary outcome concerns treatment engagement at week 12 post-randomization.

Random allocation is relevant because, in general, randomization is intended to make assigned groups comparable with respect to factors that could affect outcomes. It does not ensure that the groups will be identical, particularly in a pilot study, and it does not eliminate uncertainty in estimates. With no masking, participants and study personnel were not masked to assignment according to the registry entry. How that might have affected outcome measurement or conduct is not described in the registry record, so its practical impact cannot be determined from the posted information.

Details of sequence generation, allocation concealment, eligibility criteria, recruitment, protocol deviations, and the planned sample-size justification are not reported in the registry record. The registry also does not report whether any interim analyses were planned or conducted. These gaps limit assessment of the design beyond the randomized, unmasked, parallel-group structure and the intervention descriptions that are 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

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 Percentage Opioid Negative Urine Tests, measured over Baseline to 12 weeks post-randomization.

The primary endpoint is titled Percentage Opioid Negative Urine Tests. It concerns the percentage of weekly study urine drug tests that were negative for opioids between baseline and 12 weeks post-randomization. The registry states that a missing test was assumed to be positive. This rule treats an unavailable test as not negative for the purpose of defining the endpoint, rather than omitting it from the outcome under that definition.

Urine testing is used in this study as an objective measure related to opioid use during follow-up. In general, repeated measurements across a follow-up period can describe a pattern over time rather than relying on a single measurement. The registry's outcome definition focuses on the proportion of weekly tests that were negative, not solely on a participant-level status at the final visit. That makes the way repeated tests and missing tests are handled important to interpretation.

The registry posts arm values of 234 for the Video-based DOT Application and 299 for Treatment as Usual, with 39 participants listed in each arm. These values are recorded as counts of units. The denominator for the stated percentage is not reported with those arm values, and no dispersion measure is posted. It is therefore not possible to derive an arm-level percentage or an unadjusted percentage comparison from the reported values without information that is not in the registry record.

The endpoint is relevant to the stated study question because it records whether weekly tests were negative over the follow-up period. Still, the posted count values and the model-based result are not interchangeable with a directly reported percentage for each arm. The registry does not report further detail on test collection, missed visits, or the number of tests expected per participant.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a log-linear generalized estimating equation (GEE) regression for the primary outcome. GEE methods are generally used to estimate relationships when observations may be correlated within a participant, as can occur when outcomes are recorded repeatedly over time. The primary endpoint concerns weekly tests, so accounting for within-participant dependence can be important. The registry does not give enough model detail to establish exactly how repeated tests, time, or missing observations entered this analysis.

What does the reported risk ratio convey?

A risk ratio compares the outcome level in one group with the outcome level in another group, according to the model's definition and group ordering. The registry gives a risk ratio of 0.78 for the two listed groups, but the available outcome reporting raises an interpretive issue: the endpoint is titled as a percentage, while the arm values are count values without denominators. The posted result should therefore be described as the registry's model-based risk ratio, not converted into a percentage-point difference or an absolute number of negative tests attributable to assignment.

For a ratio measure, the no-effect value is the value at which the compared groups have the same outcome level. The 95% confidence interval runs from 0.60 to 1.02, including that no-effect value. General explanation: a confidence interval gives a range of estimates that remain reasonably compatible with the observed data under the statistical model and its assumptions. It is not a range containing a fixed proportion of individual participant outcomes, nor does it represent the probability that the true effect lies in the interval.

How should the p-value be read?

The primary p-value is 0.07. In general, a p-value describes how unusual data at least as incompatible with a specified null hypothesis would be under that hypothesis and the analysis assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of an effect. It should be interpreted together with the effect estimate, its confidence interval, the outcome definition, and the study's design. The registry does not report a prespecified decision threshold or a complete statistical analysis plan.

For the secondary engagement outcome, the registry reports Poisson regression with robust standard errors and a risk ratio. In general, Poisson regression models count-like outcomes, while robust standard errors can make the estimated uncertainty less dependent on the usual variance assumption of the model. The registry identifies the reported parameter as a risk ratio and gives a 95% confidence interval and p-value. It does not provide enough detail to assess all model specifications or assumptions. A model-based result and an unadjusted recalculation need not match exactly because they can use different effect measures, assumptions, and adjustment strategies.

Why do denominators and repeated observations matter?

The primary endpoint is expressed as a percentage of weekly tests, but the registry posts counts of units without the corresponding denominators. A percentage requires a numerator and a denominator. Without the denominator, the posted arm values cannot be translated into percentages or compared as percentages. The presence of repeated tests also means that observations from the same participant may not be independent. General explanation: treating correlated measurements as if they were independent can misstate uncertainty, which is why methods such as GEE are often used for repeated observations. The registry's reported GEE method is consistent with addressing correlated observations in general, but the exact implementation is not reported.

The missing-test rule is specified for the primary endpoint: missing tests were assumed positive. General explanation: assumptions about missing outcomes can affect an estimate because the unobserved values are not directly known. In this case the rule defines how missing tests contribute to the negative-test outcome, but other details—such as the amount of missingness and whether additional sensitivity analyses were conducted—are not reported in the registry record.

What do the planning illustrations mean?

The posted sample-size planning illustrations use the observed effect for the secondary engagement outcome, rather than the trial's own design. The record specifies a two-sided significance level of 0.05, equal allocation, and a normal approximation. It states that the effect to detect is a stated fraction of the observed effect, and that the listed values are participants per arm. For planning for 100% of the observed effect, the illustrations give 175 participants per arm for 80% power and 234 participants per arm for 90% power. For planning for 75% of the observed effect, they give 298 participants per arm for 80% power and 398 participants per arm for 90% power. For planning for 50% of the observed effect, they give 636 participants per arm for 80% power and 851 participants per arm for 90% power.

These are planning illustrations, not observed power and not evidence that the trial was designed around those assumptions. They depend on the observed secondary effect, which is itself uncertain. The registry does not report that these calculations formed the study's original design or sample-size justification. No comparable planning illustration is posted for the primary count outcome; the registry notes that values are missing or not estimable for a two-arm comparison of that endpoint.

10. Analyses Reported by the Sponsor

For the primary endpoint, the registry reports the sponsor's log-linear GEE analysis: risk ratio 0.78, 95% confidence interval 0.60 to 1.02, and p-value 0.07. A simple unadjusted recalculation is not available for this count outcome because the record says values are missing or not estimable for a two-arm comparison. The arm values are reported as counts, without the denominators needed to recreate the stated percentage comparison. There is therefore no directly comparable unadjusted primary estimate against which to check the model result.

For treatment engagement, the sponsor reports Poisson regression with robust standard errors, with risk ratio 0.84, 95% confidence interval 0.65 to 1.10, and p-value 0.20. The unadjusted recalculation uses a risk difference rather than the sponsor's risk ratio and reports a p-value of 0.187. These results are not identical measures of effect, so numerical agreement should not be expected. The recalculation is a simple unadjusted comparison based on posted summary statistics, whereas the sponsor's model is the registry's reported analysis.

The registry does not flag a discrepancy. The sponsor's analysis remains the relevant reported analysis; the unadjusted recalculation is supplementary and should not be treated as a replacement for it. Neither approach removes uncertainty or resolves limitations in the outcome reporting.

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary outcome and a secondary outcome, but it also lists 10 outcome measures and 9 analyses. It does not report how those measures were ranked, which analyses were prespecified, or whether a multiplicity adjustment was planned. Multiplicity refers generally to the increased chance of chance findings when many outcomes or comparisons are examined. Without a reported strategy for handling multiple analyses, secondary p-values should be interpreted cautiously and should not be treated as if they were the sole confirmatory test.

The registry's missing-data rule for the primary endpoint is explicit: a missing urine test was assumed to be positive. That rule affects the endpoint definition by treating unavailable tests as not negative. The registry does not report the number of missing tests, the reasons tests were missing, or sensitivity analyses using different assumptions. It also does not explain whether missingness was handled in any additional way in the GEE model. For the secondary outcome, a missing-data approach is not reported in the registry record.

The analysis populations are not reported in the registry record. In particular, the record does not state whether the analyses included all randomized participants, only participants with observed outcome data, or another defined population. Participant flow, adherence to assigned intervention, withdrawals, and exclusions from the analyses are also not reported in the information available for these results. Without those details, it is not possible to assess whether the reported comparisons preserve the randomized groups or how missing outcome data affected the analyses.

Interim analyses and any statistical stopping rules are not reported in the registry record. The full statistical analysis plan, including whether the models or contrasts were specified before examining outcomes, is not reported either. These omissions do not show that such procedures were absent; they mean the registry record does not establish what was planned or performed. Interpretation should therefore stay close to the posted methods and acknowledge the uncertainties in multiplicity, missingness, and analysis populations.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: in-patient hospitalization (10.3%).

The registry lists serious adverse events for the Video-based DOT Application arm as 9 among 39 at risk, and for Treatment as Usual as 3 among 39 at risk. Other adverse events are listed as 3 among 39 at risk in the application arm and 1 among 39 at risk in the usual-care arm. The record lists 1 death in the application arm and 0 in the usual-care arm. In-patient hospitalization is the top serious adverse event term listed, at 10.3%.

These figures are descriptive safety information from the registry. They do not, by themselves, show that an event was caused by the assigned approach. The registry record does not report the relationship of the listed events or death to the intervention, nor does it provide enough detail to interpret event timing, severity beyond the serious classification, or clinical circumstances. The term listed as the top serious event should not be assumed to account for all serious events.

The counts also need to be read in the context of a pilot study, whose stated purpose was to test the application. The registry does not report a formal comparative safety analysis or a statistical uncertainty measure for these figures. Differences in the reported counts should therefore be described, not converted into claims about comparative safety or causation.

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

The registry record gives evidence about a randomized pilot comparison of video-based directly observed therapy with treatment as usual in office-based buprenorphine treatment. The posted primary model estimate is a risk ratio of 0.78, with a 95% confidence interval from 0.60 to 1.02 and p-value 0.07. The interval includes the ratio's no-effect value, and the result does not provide a precise distinction between a difference and no difference under the model. This is not evidence that the application works, and it is not evidence that it does not work.

The main interpretive constraint is the mismatch between the primary endpoint's description as a percentage and the posted arm values as counts of units. Denominators are not reported with those counts, and the registry does not give a direct percentage comparison. The model-based result can be reported as recorded, but it cannot be translated from the available information into an absolute change in the percentage of negative tests. The registry's rule that missing tests were assumed positive is relevant, but the extent of missing data and the effect of that rule are not reported.

The secondary outcome is treatment engagement at week 12. It gives context about a different dimension of the intervention's stated purpose, but its analysis is also uncertain and should carry less weight than a prespecified primary result. The registry does not report a multiplicity plan, a full analysis population, or enough details to assess the complete analysis strategy.

More broadly, the results apply to the study as recorded, not automatically to every person receiving buprenorphine or every office-based setting. Recruitment details, eligibility criteria, intervention fidelity, and usual-care details are not reported in the registry record. The study is described as a pilot, and the sample-size planning illustrations are based on an observed secondary effect rather than the study's original design. The findings are therefore best understood as preliminary registry results with material uncertainty, rather than as a definitive estimate of comparative benefit or harm.

14. Why This Trial Matters Statistically

This trial raises statistical issues common in behavioral intervention studies with repeated clinical measurements. Its primary outcome is described as a percentage across weekly urine tests, while the registry posts counts of units and does not report the denominators. That distinction matters: a count alone does not specify a percentage, and an analysis of repeated tests needs to account for observations that may be correlated within participants. The registry reports GEE regression, a method generally used to handle correlated observations, but the missing denominator information limits independent interpretation of the posted primary result.

The record also illustrates why outcome definitions and missing-data rules should be read together. The registry states that missing urine tests were assumed positive. That is a clear rule for defining the endpoint, but the amount and pattern of missingness are not reported. Without those details, readers cannot judge how influential the rule may have been or compare it with alternative assumptions.

Another statistical lesson is the distinction between a model-based result and an unadjusted recalculation. For the secondary outcome, the registry reports a model-based risk ratio, while an unadjusted recalculation uses a risk difference. These measures describe the group comparison on different scales. Their results should not be treated as competing estimates of exactly the same quantity, and a recalculation from summary statistics does not replace the sponsor's reported analysis.

Finally, the study is explicitly described as a pilot, and the planning illustrations are derived from the observed secondary effect rather than the trial's own design. Such illustrations can help communicate how future planning changes under different assumed effects, but they are not evidence of observed power or of the original sample-size rationale. The absence of a reported multiplicity plan and analysis population also demonstrates why statistical interpretation depends on transparent reporting beyond a p-value alone.

15. Limitations

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

Several limitations specific to this trial affect interpretation. The primary endpoint is named as a percentage of weekly urine tests, but the posted arm values are counts of units and no denominators are given. The registry therefore does not support reconstruction of the arm-level percentages or a simple unadjusted comparison of those percentages. The model estimate is available, but the relationship between that estimate and the posted count values is not fully explained by the registry entry.

The study is described as a pilot, and 39 participants are listed in each arm. The registry does not report the original sample-size justification or a target effect used to design the trial. The later sample-size illustrations are based on the observed secondary outcome and should not be read as the trial's design assumptions. The precision of the primary estimate is limited by its confidence interval, which includes the no-effect value for a ratio.

Important design and analysis details are not reported in the registry record, including eligibility criteria, recruitment and participant flow, allocation concealment, analysis populations, and the complete statistical analysis plan. Further detail on usual care, intervention delivery, adherence to assignment, and test collection is also not reported. The registry gives a missing-test rule for the primary outcome but not the extent or pattern of missing tests, or sensitivity analyses assessing that rule.

The record lists multiple outcome measures and analyses but does not report how multiplicity was handled. It also does not report interim analyses. These omissions make it difficult to determine which analyses were confirmatory and to assess the potential influence of missing data or selective emphasis. The unmasked design is recorded, but its consequences for conduct or outcome measurement cannot be assessed from the registry entries. These limitations do not demonstrate that the study procedures were inadequate; they describe what the registry record does not establish.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Trial of Adherence App for Buprenorphine Treatment (TAAB) Study — NCT03779997.
• Registry record last updated 2021-10-06; status last verified 2021-06.
• Results first posted 2021-07-01.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

TAAB is a completed, randomized, parallel-group, unmasked pilot study sponsored by the University of Washington. It compared a video-based DOT smartphone application with Treatment as Usual among participants receiving buprenorphine treatment for opioid use disorder in office-based settings. The registry lists 39 participants in each arm and enrollment of 78.

The primary endpoint was the percentage of weekly study urine tests negative for opioids from baseline to 12 weeks post-randomization, with missing tests assumed positive. The registry reports a log-linear GEE risk ratio of 0.78, a 95% confidence interval of 0.60 to 1.02, and p-value 0.07. Arm-level values are posted as counts without denominators, limiting direct interpretation as percentages.

The secondary outcome was treatment engagement at week 12. The registry reports a Poisson regression with robust standard errors. Interpretation of both outcomes is limited by uncertainty, incomplete reporting of analysis details, and the absence of a reported multiplicity plan. The study's planning illustrations are based on the observed secondary effect and are not evidence of the original design or observed power.