Three Strategies for Implementing Motivational Interviewing on Medical Inpatient Units: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Three Strategies for Implementing Motivational Interviewing on Medical Inpatient Units (NCT01825057), a randomized na study. The primary outcome is Percentage of Motivation Interviewing Sessions Audio Recorded.

2. Clinical Question

The trial addresses an implementation question: whether different approaches to introducing motivational interviewing are associated with different levels of recorded provider practice in a hospital service. It is not a direct comparison of patient health outcomes or of motivational interviewing against no motivational interviewing. The primary endpoint is a provider-level percentage, calculated from recorded sessions for each provider and then averaged by group.

The record describes a cluster randomized controlled trial of three strategies. It reports analyses between coded groups, but does not identify in the registry record which coded group corresponds to each named strategy. Consequently, the reported differences and p-values cannot be confidently attributed to a particular pair of named approaches from the record alone.

3. Trial Design

This was a randomized, parallel-group trial described as cluster randomized. It enrolled 1211 participants and is marked completed. The listed study dates are 2013-02-21 for the start and 2018-11-19 for primary completion. The registry lists the phase as not applicable. The sponsor is the National Institute on Drug Abuse (NIDA).

The three strategies were See One, Do One, and Order One. See One consisted of motivational interviewing workshop training; the trainer encouraged providers to screen patients for substance misuse and apply motivational interviewing when indicated. Do One included workshop training followed by live bedside supervision from trained consultation-liaison clinicians twice before the trial and once midstream. Providers could also request additional live supervision during the trial. The registry description ends before fully specifying that additional supervision. Order One allowed providers either to administer motivational interviewing themselves or to request delivery by a trained consultation-liaison clinician. Only this group could request motivational interviewing through a separate electronic health record order.

The record identifies single masking, but does not report who was masked. It does not state the precise unit or procedure used to form the randomization clusters. The primary follow-up timeframe was an expected average of 24 weeks per provider, to see 40 study-enrolled patients per provider. These are described as expectations in the outcome timeframe, not as observed follow-up for every provider.

The listed primary outcome was the percentage of motivational interviewing sessions audio recorded. Secondary outcomes included adherence and competence scores from the Independent Tape Rater Scale and whether providers were adequately performing motivational interviewing. The specific measurement details for those secondary outcomes are not reported in the registry record.

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: The study's primary outcome and secondary outcomes are only collected and reported at the provider level - not at the patient level.

Patient Study Flow

Provider Study Flow

Analysis population for the primary outcome, as defined in the registry: A percentage of MI sessions audio recorded was calculated for each provider, and a mean percentage was calculated by group.

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: Total provider and patient participant characteristics

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Motivation Interviewing Sessions Audio Recorded, measured over Provider participants will be followed for the duration of the trial period, an expected average of 24 weeks to see 40 study-enrolled patients per provider..

The primary endpoint measures how often providers audio recorded motivational interviewing sessions for their 40 consecutively enrolled study-eligible patients. A percentage was calculated for each provider, then a mean percentage was calculated for each group. Thus, the endpoint summarizes recording behavior across providers rather than patient outcomes, the number of patients treated, or a direct rating of counseling quality.

Recording is relevant to an implementation study because it provides an observable indicator that a session took place in a form that could be reviewed. It may also support assessment of how providers deliver the approach. But recording is not identical to delivering motivational interviewing competently: a session may occur without being recorded, and a recorded session does not by itself establish adherence or competence. The registry lists separate secondary outcomes for fundamental and advanced adherence and competence, as well as adequate performance.

The reported endpoint is a continuous percentage summarized by a group mean. The record does not report the group means themselves, their standard deviations, or confidence intervals for the primary outcome. The comparison summaries therefore need to be interpreted as reported differences and p-values, not as a complete account of each group’s outcome distribution or precision.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a parallel randomized trial with cluster randomization. In a parallel design, study groups are followed as distinct strategies rather than participants crossing through each strategy. Cluster randomization assigns groups of individuals together rather than assigning each individual independently. This matters statistically because outcomes from members of the same cluster can be correlated. If that dependence is ignored, standard errors can be too small and p-values can appear more decisive than warranted. The registry specifically says a simple individual-level re-analysis would be misleading, and only the sponsor’s own analysis is shown.

What does the endpoint summarize?

The primary outcome is a percentage calculated for each provider and then averaged within each group. It is therefore a continuous measure, with the provider as the unit for the percentage calculation described in the record. The endpoint concerns audio recording of sessions for consecutively enrolled eligible patients. It is not described as a proportion of all enrolled patients, nor as a patient-level clinical endpoint. The registry does not report the number of providers contributing to each group or the distribution of provider-level percentages.

A group mean can be useful as a summary of the average provider-level percentage, but it does not show whether providers’ values were tightly grouped or varied substantially. The record identifies standard deviation as the dispersion measure for the primary outcome but does not include the group-specific standard deviations in the posted results. It also does not give confidence intervals for the reported contrasts.

Why might a rank-based test be used?

The reported method is Kruskal-Wallis. As a general statistical concept, this is a rank-based method for comparing distributions across groups. It can be useful when a comparison is not well represented by assumptions of a conventional parametric test, such as a normally distributed outcome with comparable variance. Rank-based methods use the ordering of observations rather than relying directly on their original numerical distances. Their interpretation is not automatically a test of means; the exact interpretation depends on the design and analysis implementation.

The registry reports both mean and median differences alongside Kruskal-Wallis analyses. It does not explain how those difference summaries were calculated, how the test was implemented for the cluster-randomized design, or whether the reported p-values account for clustering. Those implementation details are important for judging the reported inferential results and are not reported in the registry record. The numerical difference and the test result should therefore be treated as distinct pieces of information rather than assuming the p-value tests precisely the displayed mean or median difference.

What does a p-value convey?

In general, a p-value is calculated under a null model and analysis assumptions; it describes how unusual results at least as incompatible with that model would be if the model were true. A smaller p-value indicates greater incompatibility with that null model, but does not quantify the probability that a strategy is effective, the size of a meaningful effect, or the probability that the finding will replicate. The reported p-values are 0.001, < 0.001, and 0.498. The record does not report confidence intervals, which would help describe the range of effect sizes compatible with the data under the analysis.

The registry labels the analyses as superiority analyses. That label does not identify the direction of the observed difference, and the group codes are not linked to named strategies in the record. The p-values alone cannot identify which strategy had a higher mean recording percentage or establish a clinical or implementation benefit.

How should cluster structure affect interpretation?

When assignment occurs by cluster, analyses generally need to reflect both the allocation unit and the correlation among observations within clusters. The effective information may be less than would be implied by the same number of independent individuals. A valid approach can use cluster-level summaries or a model that accounts for clustering, but the registry does not state which precise method the sponsor used beyond naming Kruskal-Wallis. It does not report cluster counts, cluster sizes, or the intracluster correlation. No independent-participant re-analysis is appropriate to substitute for the sponsor’s reported analysis in this setting.

Single masking is recorded, but the masked party is not specified. The randomization procedure, allocation concealment details, analysis population, missing-data methods, and treatment of providers or clusters with incomplete endpoint data are also not reported in the registry record. These details would help assess potential bias and the relationship between the randomized groups and the analyzed results.

What does the planning illustration represent?

Any sample-size figures associated with this analysis are planning illustrations based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The stated effect is a fraction of the observed effect. Such illustrations describe hypothetical planning under those assumptions; they are not the trial’s original design justification and are not observed power. The registry does not report the trial’s sample-size justification or the precise planning values for these illustrations.

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("See One", "Do One", "Order One")
n   <- c(13, 12, 13)
m   <- c(0.8, 3, 20.5)
sp  <- c(2.3, 5.4, 17.9)
sd  <- sp      # posted standard deviations
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  See One  13  0.800  2.300  -0.590  2.190
CARM  Do One  12  3.000  5.400  -0.431  6.431
CARM  Order One  13  20.500  17.900  9.683  31.317

11. Analyses Reported by the Sponsor

The registry presents the sponsor’s own analyses and explicitly excludes a simple individual-level re-analysis because the trial was cluster randomized. That distinction is important: a recalculation from aggregate summary statistics that treats individuals as independent would not reflect the randomized structure and could give misleading uncertainty estimates. No recalculated comparison is reported alongside the sponsor’s results.

The sponsor analyses are identified as Kruskal-Wallis comparisons and report differences of 17.5, 19.7, and 2.178 for the coded group pairs, with p-values of 0.001, < 0.001, and 0.498. Confidence intervals are not reported in the registry record. Nor does the record explain the test implementation or map the coded groups to the named strategies. The figures can therefore be quoted as the sponsor’s reported results, but their direction and strategy-specific meaning remain unclear from the record.

Because no independent recalculation is shown and the design is clustered, there is no competing recalculated estimate to weigh against the sponsor’s analysis. The sponsor’s analysis is the appropriate reported basis for describing the numerical findings, while the missing methodological detail limits assessment of their precision and interpretation.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists five secondary outcomes: fundamental adherence, fundamental competence, advanced adherence, advanced competence, and adequate performance on motivational interviewing. Along with the primary outcome, it lists six outcome measures and 15 analyses. The reported results concern the primary recording outcome; detailed results for the secondary outcomes are not reported in the registry record.

Testing multiple outcomes or contrasts can increase the chance that at least one result appears statistically significant by chance. As a general statistical principle, investigators may define a primary comparison, use a hierarchical testing plan, or adjust significance thresholds to address multiplicity. The registry record does not report whether such a plan or adjustment was used. It also does not identify a single primary contrast among the three strategies or explain how the pairwise analyses relate to a broader test of group differences. The p-values should therefore be read as reported, without assuming a multiplicity procedure.

The record does not report interim analyses or stopping rules. It does not provide a sample-size justification or state whether the planned enrollment was based on a specified detectable difference. The listed enrollment is 1211, but the number analyzed for the primary endpoint and the analysis population are not reported in the registry record.

Missing-data handling is also not reported. The endpoint depends on provider-level recording information, but the record does not explain how unavailable recordings, incomplete provider follow-up, or other missing endpoint values were handled. It does not state whether all randomized units were included in the analysis or whether exclusions occurred. Without these details, the relationship between enrollment, endpoint availability, and the reported comparisons cannot be fully assessed.

These gaps do not establish that multiplicity, missingness, or interim monitoring affected the findings. They mean that the registry record does not offer enough information to evaluate those aspects of the statistical plan.

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.

The registry reports zero serious adverse events, zero other adverse events, and zero deaths in each of the three named groups. The numbers at risk for serious and other adverse events are 349 for See One, 391 for Do One, and 471 for Order One. These are the reported safety counts; they should not be interpreted as proof that no adverse events occurred outside the registry’s reporting scope or observation procedures.

The study compares implementation strategies for motivational interviewing rather than reporting a medication or other clinical treatment comparison. The registry does not list serious adverse event terms. It does not explain how adverse events were collected, the precise observation period for safety, or whether the at-risk counts represent all enrolled participants or a defined safety population. The primary outcome timeframe describes expected provider follow-up, but the safety reporting details are not separately specified.

As a general statistical point, a zero count is a description of the events recorded in the stated group, not evidence that the underlying event probability is exactly zero. Without further information about ascertainment and follow-up, the figures are best read as registry-reported event counts rather than as a comprehensive safety assessment.

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

The findings concern implementation behavior: the percentage of motivational interviewing sessions providers audio recorded for eligible patients. The record reports differences between coded groups and corresponding p-values, but it does not map those codes to See One, Do One, and Order One. It also does not report the direction of the differences or confidence intervals. Readers therefore cannot use these summaries alone to identify which named strategy had the higher recording percentage or how precisely the contrasts were estimated.

The endpoint is not a direct measure of patient substance use, health, satisfaction, or clinical response. Nor does recording itself demonstrate that motivational interviewing was delivered with fidelity or competence. The registry lists separate adherence, competence, and adequate-performance outcomes, but the results for those measures are not reported in the information available in the registry record.

As a general statistical distinction, a statistically small p-value does not establish that a difference is practically important. Conversely, a p-value of 0.498 does not prove equivalence or show that two strategies are identical. Establishing equivalence would require a prespecified margin and an analysis designed to assess it; such details are not reported in the registry record.

The trial is described as cluster randomized, and the registry cautions against simple individual-level re-analysis. The reported estimates should be interpreted in light of the sponsor’s analysis, while recognizing that the exact treatment of clustering is not explained in the record. Missing confidence intervals, group-level summaries, and analysis-population details further limit assessment of uncertainty.

Overall, the posted results describe statistical comparisons for a process measure. They do not by themselves establish that a strategy improves the quality of motivational interviewing, changes patient outcomes, or should be selected in practice. Those conclusions would require fuller outcome reporting and a clearer account of the analysis.

15. Why This Trial Matters Statistically

This trial is statistically interesting because its question is about implementing a behavioral practice within a working hospital service, where strategies may differ in how they fit existing workflows. The approaches range from workshop training, to supervised bedside practice, to an option to request specialist delivery through an electronic health record order. Comparing implementation strategies requires outcomes that describe adoption or delivery, and this trial’s primary endpoint focuses on session recording.

The design also highlights why the unit of randomization matters. The trial is described as cluster randomized, so analyses need to respect the grouped allocation rather than treating every participant as independent. The registry explicitly flags this concern and does not present an individual-level re-analysis. That is an important reminder that a large enrollment count alone does not determine the precision of a cluster-randomized comparison; the clustering structure and analysis approach matter too.

The registry’s reporting also illustrates the importance of connecting statistical contrasts to recognizable groups. Differences and p-values are reported for coded groups, but the codes are not mapped to the named strategies in the record. Without that link, the numerical findings cannot be translated clearly into a strategy-by-strategy conclusion. Confidence intervals and group-level outcome summaries would further help readers understand uncertainty and practical magnitude.

Finally, the distinction between implementation process measures and clinical outcomes is central. Recording can be an observable marker of activity, but it is not a substitute for assessing counseling quality or patient outcomes. The listed secondary outcomes point toward adherence and competence, yet their results are not reported in the registry record. The trial’s statistical contribution therefore depends not only on its comparisons, but also on the clarity and completeness of the outcome and analysis reporting.

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 limitations are specific to the registry record. Although the study is described as cluster randomized, the record does not specify the precise cluster unit, the number of clusters, cluster sizes, or the method used to account for within-cluster correlation. It warns that an individual-level re-analysis would be misleading, but the details of the sponsor’s clustering approach are not reported.

The coded comparison groups are not linked to the named strategies, and the direction of the reported differences is absent. The registry also does not report group-specific primary outcome means or standard deviations, or confidence intervals for the contrasts. These omissions prevent a clear strategy-specific reading and limit assessment of the range of plausible effects.

The description of Do One is incomplete: it states that additional live supervision could be requested during the trial, but the registry text ends before fully describing that provision. The single-masking designation does not identify who was masked. Randomization details and allocation concealment are not reported in the registry record.

The analysis population, missing-data procedures, and handling of incomplete recordings are unspecified. The registry does not report the sample-size justification, a multiplicity plan, or interim analyses. It lists secondary outcomes and analyses but does not provide their results in the record.

These reporting gaps do not establish that the trial was conducted improperly. They limit how fully the reported results can be independently assessed and how confidently the contrasts can be interpreted as differences between the three implementation strategies.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Three Strategies for Implementing Motivational Interviewing on Medical Inpatient Units — NCT01825057.
• Registry record last updated 2021-03-25; status last verified 2021-03.
• Results first posted 2021-03-25.
• 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

NCT01825057 is a completed, randomized, parallel, cluster-randomized trial of three strategies for implementing motivational interviewing on medical inpatient units. It enrolled 1211 participants and was sponsored by the National Institute on Drug Abuse (NIDA). The primary outcome was the mean percentage of motivational interviewing sessions audio recorded for eligible patients, calculated first for each provider and then by group.

The registry reports three Kruskal-Wallis comparisons between coded groups, with differences of 17.5, 19.7, and 2.178 and p-values of 0.001, < 0.001, and 0.498. The coded groups are not mapped to named strategies in the registry record, and confidence intervals are not reported. The record also reports zero serious adverse events, zero other adverse events, and zero deaths in each group.

The registry cautions that simple individual-level re-analysis would be misleading because of the clustered design. Further details about clustering, missing data, multiplicity, and the analysis populations are not reported in the registry record.