Evaluation of a Mental Health Physician Support Program in Nova Scotia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of a Mental Health Physician Support Program in Nova Scotia (NCT01975948), a randomized na study. The primary outcome is Depression Severity (Change in Patient Health Questionnaire-9 (PHQ-9) Score From Baseline.

2. Clinical Question

The study’s central patient-level question was whether the pattern of depression symptom scores across follow-up differed between patients associated with physicians who completed the program training and patients receiving treatment as usual. The PHQ-9 was assessed at baseline and at 1, 2, 3, and 6 months. The registry describes the comparison as a between-group difference in mean scores during follow-up, with the intervention effect represented by an intervention-by-time interaction.

The registry’s brief summary sets out a broader evaluation than this primary patient endpoint alone. It names reducing stigma among medical personnel, increasing physicians’ and staff members’ comfort in providing care to people living with mental illness, and improving client well-being and mental health as study aims. The listed secondary outcome concerns between-group changes in occupational functioning from baseline to 6 months. Those aims and the secondary outcome do not establish that corresponding results were reported.

The record does not report the estimated group-by-time effect, its uncertainty, or a test result. Consequently, the clinical question can be described from the planned outcome and analysis, but the posted results do not permit a conclusion about the observed direction, size, or precision of any difference.

3. Trial Design

The registry describes a randomized, double-masked, parallel study of the Mental Health Practice Support Program and Treatment as Usual. The recorded enrollment is 285, and the status is Completed. The study start is listed as 2013-11 and primary completion as 2015-12. The sponsor is Nova Scotia Health Authority, and the recorded phase is NA.

The arm descriptions distinguish physicians from patients. Physicians in the program arm were described as undertaking Adult Mental Health Practice Support Program training; physicians in the comparator arm administered treatment as usual for depression. Patients were associated with a physician who completed program training or received treatment as usual for depression, according to their respective arm descriptions. These descriptions indicate that the evaluation included physician and patient roles, but details of the allocation procedure, the unit of randomization, and the implementation of masking are not reported in the registry record.

The primary outcome was assessed repeatedly rather than at a single follow-up visit. The stated time frame includes baseline and follow-up at 1, 2, 3, and 6 months. The planned analysis accounted for physicians clustered within practices, patients clustered within their corresponding physicians, and repeated PHQ-9 ratings clustered within patients. This structure is relevant because measurements from people in the same practice or physician group, and repeated observations from the same patient, may be related.

The record lists 10 outcome measures and 14 analyses. It identifies occupational functioning from baseline to 6 months as a secondary outcome. It does not report the allocation ratio, recruitment and eligibility details, the planned sample-size justification, or a detailed description of usual care. Enrollment should therefore be distinguished from the primary outcome analysis population, which the record states included participants with at least one follow-up data (n=116).

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: 285 participants were enrolled in the study: Physicians (n=111);Patients (n=129); Medical office Staff (MOAs) (n=45) MOA Group: MOAs excluded from outcome analysis due to inconsistent particpation and small sample size

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All participants with at least one follow-up data (n=116) were included in the analysis

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: It should be noted that baseline measures for the physician group are based on pre and post training completers (n=73). Patient data is based on those who completed at least one data time point (n=116).

6. Primary Endpoint

The registry lists the primary outcome as Depression Severity (Change in Patient Health Questionnaire-9 (PHQ-9) Score From Baseline, measured over Baseline, 1, 2, 3, and 6 months.

The primary endpoint concerns change in depression severity measured with the Patient Health Questionnaire-9, or PHQ-9. The measure covers nine symptom-based Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria for major depressive disorder. Its scores range from 0-27, with higher scores indicating more severe depression severity. The outcome is described in units on a scale, and the summary measure is the mean with standard deviation.

Repeated assessment can capture how scores develop over time, rather than restricting the comparison to a single follow-up point. The registry specifies measurements at baseline and at 1, 2, 3, and 6 months. Its planned primary comparison is not simply a contrast at one visit: it is a comparison of between-group mean differences during follow-up, expressed through a group-by-time interaction. In general, a group-by-time approach asks whether the average outcome trajectory changes differently across groups over the observed occasions.

The PHQ-9 is used as a symptom-severity measure aligned with the recorded condition of Major Depressive Disorder. Its score scale gives the outcome a defined direction: higher values correspond to greater reported symptom severity. The primary endpoint is not a direct measure of stigma, physician comfort, or occupational functioning, even though those topics appear among the study aims or other outcomes.

The registry record does not report the observed mean scores, changes, group contrast, interaction estimate, confidence interval, or test result. It therefore describes what was measured and how the comparison was planned, but not the magnitude or precision of the primary outcome finding.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The planned primary analysis is a multi-level mixed model for repeated PHQ-9 ratings. In general, a mixed model can represent both the average pattern of an outcome and dependencies among observations that share a grouping structure. This matters when repeated measurements from the same person are related, or when people cared for by the same physician or practice may have more similar outcomes than unrelated observations.

Why account for clustering?

The registry describes three nested sources of clustering: physicians within practices, patients within their corresponding physicians, and longitudinal PHQ-9 ratings within patients. General statistical reasoning says that observations in a cluster are not necessarily independent. For example, measurements from the same patient are repeated observations on one person, not separate people. Ignoring such dependence can make uncertainty estimates inappropriate because the analysis would treat related measurements as if they carried wholly independent information.

A multi-level model can represent these levels of dependence while estimating an overall group pattern. The registry does not report the model’s exact specification, the covariance structure, random-effects structure, estimation details, or diagnostic checks. Those details should not be inferred from the general label “multi-level mixed model.”

What does the group-by-time interaction ask?

The registry says that the four follow-up time points were represented by indicator variables and that the intervention effect was measured as an intervention-by-time interaction. In general, indicator variables allow a model to represent distinct time occasions without requiring the outcome to change in a straight-line pattern across time. An interaction allows the difference between groups to vary across those occasions. The time-by-group interaction was assessed using a likelihood ratio test.

In general, a likelihood ratio test compares how well a specified model describes the data with how well a related model describes the data under a restriction. Here the registry identifies the test as the assessment of the time-by-group interaction. It does not report the compared model specifications, the test statistic, degrees of freedom, or result. The test result therefore cannot be reconstructed or interpreted from the record.

What does a mean and standard deviation describe?

The primary outcome is a continuous score summarized using a mean and standard deviation. In general, the mean describes the average score in a group, while the standard deviation describes the spread of individual scores around that average. A mean difference focuses on a contrast in group averages; the spread is not itself a measure of uncertainty in that contrast. Uncertainty about an estimated mean difference is usually conveyed by an interval or a test, but neither is reported for the primary comparison in the registry record.

The PHQ-9 values are bounded by the scale range stated in the record. A statistical model’s fit and interpretation depend on the data and assumptions, and bounded scores may not behave like an unrestricted continuous variable in every setting. This general point does not establish that the chosen analysis was unsuitable. The record does not report residual assessments, distributional checks, alternative analyses, or sensitivity analyses, so their adequacy cannot be evaluated.

What can the analysis population tell us?

The primary outcome population is described as all participants with at least one follow-up data (n=116). In general, an analysis requiring follow-up measurements may have less information than an analysis that also uses baseline data from participants who did not return for follow-up. Mixed models can, under particular assumptions and specifications, use observed repeated measurements without requiring every participant to have a complete sequence. That general capability does not reveal how missing observations were handled in this trial. The registry does not report the missing-data assumptions, imputation procedures, or sensitivity analyses.

The registry also does not give the allocation ratio, model coefficients, standard errors, confidence intervals, or primary test result. No numerical effect or precision estimate can be derived from the description of the model alone. The outcome results and model details would be needed to assess the estimated difference and the assumptions supporting its uncertainty.

How should planning illustrations be read?

In general, sample-size calculations describe the number of participants per arm that would be required under stated design assumptions to detect a specified effect with a chosen power and significance level. Such calculations are planning illustrations, not observed trial results and not a statement of the probability that the trial’s conclusion is correct. The registry description specifies a two-sided significance level 0.05, equal allocation, a normal approximation, and an effect expressed as a stated fraction of the observed effect for its planning illustration. It does not report a trial-specific design justification, and no planning sample sizes or observed primary effect are reported in the registry record. Therefore, no numerical planning requirement can be discussed from this record.

10. Multiplicity, Missing Data and Analysis Populations

The record lists 10 outcome measures and 14 analyses, while the primary outcome itself is measured at baseline and at 1, 2, 3, and 6 months. It describes a group-by-time interaction test for the repeated primary outcome. The registry record does not report how multiplicity across outcomes, time points, or analyses was addressed. It does not describe a hierarchy of hypotheses, an adjustment procedure, or whether any analyses were designated exploratory. Without that information, the extent to which the reported analyses were protected against multiple testing cannot be assessed.

The primary outcome population includes participants with at least one follow-up data (n=116). The record does not provide a flow of participants from enrollment to analysis, reasons for missing follow-up, the amount of missingness at each measurement occasion, or whether missingness differed by group. It also does not report a missing-data strategy, assumptions about why measurements were missing, imputation, or sensitivity analyses. General statistical methods for incomplete repeated measures rely on assumptions; the record does not identify which assumptions were used or evaluated.

The analysis population is described for the primary endpoint, but the registry does not say whether this was an intention-to-treat population, a per-protocol population, or another prespecified set. It does not report whether participants were analyzed according to their randomized assignment or how departures from assignment were handled. The precise allocation unit and the allocation procedure are not reported in the registry record either.

These omissions limit interpretation of both the analysis population and the robustness of any estimated effect. They do not by themselves establish that the analysis was biased or inappropriate. They mean that the posted record does not contain enough information to evaluate the handling of multiplicity, missing data, or analysis-population definitions.

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

The registry describes a randomized comparison of a physician support program with treatment as usual and identifies repeated PHQ-9 score change as the primary patient outcome. It also states that the planned analysis used a multi-level mixed model and assessed a group-by-time interaction. These details establish the intended statistical question and the broad structure of the analysis.

They do not establish what the analysis found. The primary estimate, uncertainty interval, and test result are not reported in the registry record. The direction, size, and precision of the difference between PHQ-9 trajectories therefore cannot be determined. The record’s outcome description is not a substitute for an observed result, and enrollment or analysis-population counts do not reveal the outcome effect.

Even if an estimated difference were available, statistical evidence would not automatically determine whether the difference was important to patients or clinicians. In general, interpretation would require considering the magnitude and uncertainty of the estimate, the scale’s meaning, and a relevant standard for importance. The registry does not state a clinically important change threshold, and it does not report an estimated change to evaluate.

The stated aims also include reducing stigma, increasing physician and staff comfort in providing care, and improving client well-being and mental health. The primary endpoint is PHQ-9 depression severity, not a direct measure of every aim. The record lists occupational functioning as a secondary outcome, but does not report its result. The listed aims should therefore not be treated as findings.

The adverse-event fields likewise require restraint. Although the patient-group entries show zero at risk and zero affected for serious and other events, the record does not explain those entries; deaths are recorded as null. These registry values do not support a broad conclusion about safety or the absence of events.

In short, the posted information supports a description of the design and planned primary analysis, but not a conclusion about whether the groups differed in depression severity or by how much. It also cannot establish effects on stigma, staff comfort, occupational functioning, or broader well-being.

12. Why This Trial Matters Statistically

This trial is statistically notable in the way its planned analysis recognizes the structure of the observations. The primary endpoint was measured repeatedly, and the registry describes patients nested within physicians and physicians nested within practices. It also describes repeated ratings clustered within patients. These features make the independence assumptions of a simple comparison of unrelated observations an important consideration. The planned multi-level mixed model reflects those dependencies in its broad structure.

The group-by-time question is also distinct from asking whether the groups differ at a single visit. In general, a longitudinal interaction tests whether group patterns vary over time. That framing can be useful when an intervention is intended to influence the course of an outcome, but interpreting it requires the estimated pattern, uncertainty, and model details. The registry does not report those results, so the statistical question is clearer than the statistical answer.

The gap between recorded enrollment of 285 and the stated primary outcome analysis population of participants with at least one follow-up data (n=116) makes transparent reporting of participant flow and missing-data handling especially relevant. The record does not explain this difference or describe the missing-data strategy. Those omissions prevent a reader from evaluating how much information contributed to the primary analysis or how dependent the analysis might be on assumptions about unavailable measurements.

The record also lists multiple outcomes and analyses without describing multiplicity handling. This matters statistically because repeated testing can affect how individual results should be interpreted. The record does not provide enough detail to determine whether analyses were adjusted, prioritized, or exploratory.

Thus, the trial record illustrates why understanding a result requires more than knowing the outcome name and model type. The analysis structure, population, missingness, multiplicity, effect estimate, and precision all contribute to interpretation. Here, the design and model are described at a useful broad level, while the results and several key analytic details are not reported.

13. 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.

A central limitation of the registry information is that the primary outcome results are not reported. There is no estimated change or between-group contrast, no confidence interval, and no likelihood ratio test result. As a result, neither the observed effect nor its precision can be assessed. The record identifies the planned statistical approach but does not provide enough numerical information to evaluate its findings.

The primary outcome analysis population is stated as participants with at least one follow-up data (n=116), while enrollment is 285. The record does not describe participant flow, reasons for the difference, follow-up availability by group, or the number of observations at each time point. This prevents assessment of how missing follow-up may affect interpretation. The missing-data method and assumptions are also not reported in the registry record.

Important design and model details are absent. The registry does not report the allocation procedure, allocation unit, allocation ratio, exact masking procedures, full eligibility criteria, or a detailed account of the care delivered in either arm. It also does not report the exact model specification, covariance assumptions, diagnostic checks, or sensitivity analyses. These gaps limit evaluation of implementation and analytical robustness without showing that any particular design or method was deficient.

The record lists 10 outcome measures and 14 analyses but does not describe multiplicity handling or distinguish confirmatory from exploratory analyses. It gives no trial-specific sample-size justification. Although the registry describes assumptions for a planning illustration, the design rationale and numerical planning results are not reported.

Finally, the brief summary includes aims beyond the primary PHQ-9 outcome, while corresponding outcome estimates are not reported. The adverse-event fields are not explained, and deaths are null. These limitations mean the registry record offers a partial account of the trial, rather than a complete basis for assessing efficacy, clinical importance, or safety.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: Evaluation of a Mental Health Physician Support Program in Nova Scotia — NCT01975948.
• Registry record last updated 2021-10-14; status last verified 2021-09.
• Results first posted 2017-06-26.
• 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.

17. Record Summary

NCT01975948 is a completed, randomized, double-masked, parallel evaluation sponsored by Nova Scotia Health Authority. The registry records enrollment of 285, a start of 2013-11, and primary completion of 2015-12. The condition is Major Depressive Disorder. The intervention was the Mental Health Practice Support Program, compared with Treatment as Usual.

The primary outcome was change in PHQ-9 depression severity from baseline, with measurements at baseline and at 1, 2, 3, and 6 months. The planned analysis used a multi-level mixed model and assessed a group-by-time interaction. The primary outcome analysis population is reported as participants with at least one follow-up data (n=116).

The registry does not report the primary outcome estimate, confidence interval, or test result. It also does not report detailed participant flow, missing-data methods, multiplicity handling, or a trial-specific sample-size justification. The broader aims include stigma, staff comfort, client well-being, and mental health; occupational functioning is listed as a secondary outcome. Results for those aims are not reported in the registry record.