Treatment of Schizophrenia and Comorbid Cannabis Use Disorder: Comparing Clozapine to Treatment-as-Usual: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of Schizophrenia and Comorbid Cannabis Use Disorder: Comparing Clozapine to Treatment-as-Usual (NCT00498550), a randomized phase 4 study. The primary outcome is Average Over Time of Intensity of Cannabis Use (Used to Evaluate Treatment Efficacy).

2. Clinical Question

3. Trial Design

The study was registered as a Phase 4, randomized, single-masked, parallel-group trial. Its title identifies the comparison as clozapine versus treatment as usual for people with schizophrenia and co-occurring cannabis use disorder. The intervention descriptions specify clozapine in one group and treatment as usual with any antipsychotic other than clozapine in the other. The registry does not report which people were masked or how masking was maintained.

The registry records enrollment of 31. The posted arm records list 15 for clozapine and 16 for treatment as usual. These are the group counts attached to the primary outcome; the record does not explain any differences between enrollment and the outcome-specific counts. The trial status is completed. Its recorded start is 2000-10, and primary completion is recorded as 2009-03.

The primary outcome is the average over time of cannabis-use intensity, defined as the number of joints smoked during the prior week. Weekly use was assessed retrospectively with the Timeline Followback Scale. The analysis period was Week 1 to week 12. The registry says mixed models were used to estimate efficacy from partial data contributed while participants adhered to assigned treatment, under a missing-at-random assumption.

The registry record does not report the randomization procedure, allocation concealment, detailed masking procedures, treatment duration beyond the stated outcome timeframe, or the specific treatment-as-usual medicines. It also does not report a sample-size justification, a detailed statistical analysis plan, interim analyses, or a fuller account of the analysis population. Those omissions limit how fully the trial procedures can be independently assessed.

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 Average Over Time of Intensity of Cannabis Use (Used to Evaluate Treatment Efficacy), measured over Week 1 to week 12.

The primary endpoint is the average intensity of cannabis use over time, measured in joints per week. For each week, the number of joints smoked during the prior week was assessed retrospectively using the Timeline Followback Scale. The reported timeframe runs from Week 1 to week 12.

This endpoint expresses use as a count of joints rather than as a yes-or-no measure of any cannabis use. That makes it capable of representing differences in reported intensity among people who use cannabis, while also allowing no use to be represented as a count. The registry’s purpose was to compare the average intensity between the clozapine and treatment-as-usual groups.

The registry describes a mixed-model estimate based on partial data available while participants remained adherent to assigned treatment. Its stated target is an explanatory estimand: population quantities that would have occurred had all participants remained on assigned treatment throughout the study. The model relies on a missing-at-random assumption. This is an assumption of the analysis, not a guarantee that missing observations resemble observed ones.

The outcome is therefore not simply a count taken at one visit. It summarizes repeated, retrospective weekly reports through a model. The registry record does not report further details about the scale’s administration, validation in this trial, or how the model handled the count distribution. Those details would help readers judge measurement and modeling choices.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the mixed model estimate?

A mixed model is a statistical approach for outcomes measured repeatedly within the same person. Repeated measurements from one participant tend to be related, so a model can account for that within-person structure rather than treating each weekly report as an independent observation. The registry says the model used partial data from each subject while adherent to assigned treatment. It does not report the specific model formula, covariance structure, or distributional specification.

The target is described as explanatory: the population quantities that would have occurred had all participants remained on assigned treatment throughout the study. This is a modeled target, not a claim that every participant actually remained on treatment. It differs from merely averaging whatever observations happen to be available, because the model is intended to use the available repeated information to estimate the stated population quantity.

What does missing at random mean?

In general, a missing-at-random assumption means that, after accounting for information included in the analysis, whether an outcome is missing does not depend on its unobserved value. This assumption can make analysis of incomplete repeated measurements possible, but it cannot be confirmed just by fitting a model. If missingness depends on unobserved cannabis use in ways not captured by the analysis, estimates could differ from those produced under the assumption.

The registry record names the assumption but does not report how much outcome data were missing, when observations were missing, or whether sensitivity analyses examined departures from the assumption. It also does not report a detailed analysis-population definition. As a result, readers cannot determine from the registry record how strongly the analysis depends on the assumption in practice.

How should the group estimates be read?

The arm estimates are 0.02 joints per week for clozapine and 4.56 joints per week for treatment as usual. Each is paired with a 95% confidence interval: -2.1 to 2.5 for clozapine and -0.2 to 9.3 for treatment as usual. These intervals express uncertainty in the estimated arm-level means under the fitted model. They do not describe the range of individual participants’ outcomes.

The sponsor reports the between-group difference as -4.54 and gives p = .088. A difference estimate summarizes the direction and size of the modeled contrast on the joints-per-week scale. The p-value measures how incompatible the observed result is with a specified null hypothesis under the analysis assumptions; it is not the probability that the null hypothesis is true. The registry does not report a confidence interval for this contrast, which prevents direct assessment of the range of contrast values compatible with the analysis at a stated confidence level.

Why does the outcome’s count form matter?

General statistical considerations apply because the endpoint is a count. Count data may be asymmetric and may have variability that changes with the mean. Some count distributions also exhibit over-dispersion, meaning that the observed variability is greater than expected under a simpler count model. If over-dispersion is present and not accommodated, standard errors and confidence intervals can be poorly calibrated. This is a general concern, not a claim that over-dispersion occurred in this trial.

The record identifies the endpoint as a count and the approach as mixed models, but does not report the model’s distribution, link function, handling of excess zeros, or assessment of over-dispersion. It is therefore not possible to determine from the registry record whether these particular features were evaluated or how they affected inference.

What can the p-value and confidence intervals establish?

A p-value is a measure of how unusual data at least as incompatible with the null hypothesis would be, if the null and analysis assumptions held. It does not measure the clinical importance of a result. Here, the reported p-value is .088, and the registry does not provide a confidence interval for the contrast. The separate arm-level confidence intervals should not be used to reconstruct the contrast interval.

A confidence interval conveys uncertainty about an estimated parameter under a statistical model and its assumptions. Wider intervals indicate less precision than narrower intervals on the same scale. The arm-level intervals span both negative and positive values, but the target is an estimated mean count and the record does not explain the negative lower bounds. Those bounds arise in the posted model-based intervals and should not be mistaken for literal negative numbers of joints consumed by an individual.

The registry does not report a detailed statistical plan, model diagnostics, adjustments for baseline measures, or interim analyses. It also does not state whether the reported p-value was adjusted for multiple testing. No inference about those procedures can be made from the record. The registry lists one primary outcome measure and one analysis, but that alone does not establish that no other analyses were conducted.

10. Analyses Reported by the Sponsor

The sponsor’s own reported analysis is a mixed models analysis of the difference in treatment means. It reports a contrast of -4.54 and p = .088. The registry does not report a confidence interval for this contrast. The arm-specific estimates and confidence intervals are posted separately, but they do not supply the missing contrast interval.

The registry record does not identify a discrepancy between the sponsor’s reported analysis and a recalculation, and it does not provide a separate recalculated contrast to compare with the sponsor’s result. Accordingly, the sponsor’s reported estimate and p-value are the available between-group analysis. The posted arm-level estimates can be read as model-based group summaries, not as a replacement for the sponsor’s contrast test.

The p-value is above the conventional 0.05 reference level, but that does not establish equivalence or absence of an effect. The missing contrast confidence interval also means the precision of the reported difference cannot be fully evaluated from the registry record. Interpretation should remain bounded by the model assumptions and incomplete reporting.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure and one reported analysis. It does not report a detailed statistical analysis plan, whether other outcomes or analyses were considered, or whether any multiplicity adjustment was used. The absence of a posted secondary outcome analysis does not prove that no additional analyses were conducted. Without the prespecified analysis plan, it is not possible to determine how the primary result relates to all analyses that may have been undertaken.

The primary analysis used mixed models and, according to the registry, incorporated partial data while participants were adherent to assigned treatment. The target was defined in terms of what would have occurred had all participants remained on assigned treatment. This is an explanatory estimand, and the analysis depended on a missing-at-random assumption. The registry record does not report the extent or pattern of missing outcome observations, the number of observations contributing at each time point, or sensitivity analyses for alternative missing-data assumptions.

The record also does not provide a detailed definition of the analysis population, the handling of treatment discontinuation, or whether participants were analyzed according to randomized assignment regardless of adherence. The statement that available data while adherent informed the model does not answer all of those questions. These details matter because missingness and adherence can affect which observations contribute to the estimate and how the estimate should be interpreted.

No interim analyses or sample-size justification are reported in the registry record. The record gives no basis for inferring whether the study was planned around a particular effect size or whether the analysis decisions were specified before outcomes were examined. These reporting gaps do not invalidate the posted estimate, but they limit assessment of its robustness and the risk of selective analysis.

12. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: vomiting (6.7%), suicide attempt (6.7%), depression (6.7%), psychosis (6.7%), psychiatric disorder - other: accidental overdose (6.2%).

The registry reports serious adverse events for 4 participants at risk in the clozapine group and 4 participants at risk in the treatment-as-usual group. For other adverse events, the reported figures are 15 affected among 15 at risk for clozapine and 15 affected among 16 at risk for treatment as usual. The record does not report deaths.

The listed serious-event terms include vomiting, suicide attempt, depression, and psychosis, each reported as 6.7%, along with accidental overdose under psychiatric disorder and non-cardiac chest pain, each reported as 6.2%. These are registry-listed terms and percentages. The record does not clarify whether events could be counted under more than one term, how event severity was classified beyond the serious-event listing, or how the percentages were calculated.

These figures should be read as descriptive safety reporting, not as proof that a study medicine caused an event. The record does not provide enough detail about event timing, attribution, exposure duration, or comparative statistical testing to support a causal comparison. The absence of reported deaths is not evidence of zero risk beyond the record’s reporting.

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

The posted result is a model-based comparison of average cannabis-use intensity, expressed in joints per week, between clozapine and treatment as usual from Week 1 to week 12. The sponsor reports a difference in treatment means of -4.54, with p = .088. The arm-specific estimates are 0.02 for clozapine and 4.56 for treatment as usual. These values summarize the trial’s reported analysis; they are not an individual-level prediction.

The p-value does not meet the conventional 0.05 reference threshold, but this does not show that the treatments are equivalent or that the true difference is zero. Conversely, the numerical difference alone does not establish a reliable treatment effect. The registry does not report a confidence interval for the between-group contrast, so readers lack a direct interval estimate of its precision.

The results also depend on the specified mixed-model approach and its missing-at-random assumption. The registry does not provide the details needed to examine model diagnostics, missing-data patterns, or sensitivity to other assumptions. The posted arm-level intervals are broad and include values on both sides of zero; they represent uncertainty under the model rather than individual variability.

This is a completed Phase 4 randomized trial with a recorded enrollment of 31 and a single primary endpoint. Its record does not provide a posted secondary outcome analysis or enough detail to assess all aspects of design and analysis. The findings should therefore be interpreted as limited evidence from this trial’s recorded primary analysis, not as a standalone determination of efficacy, safety, or clinical practice.

14. Why This Trial Matters Statistically

This trial illustrates several statistical challenges that commonly arise in clinical research. The outcome was measured repeatedly and retrospectively, as weekly counts of joints smoked. Repeated observations within a participant are related, while a count outcome may not follow the assumptions of methods designed for continuous, symmetric data. A mixed model can use repeated observations, but the model’s specification and assumptions matter. The registry does not report enough detail to assess those choices fully.

The analysis also highlights the distinction between an estimated group contrast and the uncertainty around it. The sponsor reports a difference and p-value, but not a confidence interval for the contrast. Without that interval, readers cannot directly see the range of contrast values compatible with the fitted analysis at a stated confidence level. Separate confidence intervals for the two groups do not replace that information.

Missing data and adherence are central to the stated estimand. The model used partial data while participants were adherent and relied on a missing-at-random assumption, while targeting quantities under continued assigned treatment. This makes clear why an analysis description should identify both the target question and the assumptions linking incomplete observations to that target. The registry does not report the missingness pattern or sensitivity analyses, leaving uncertainty about the practical impact of the assumption.

Finally, the modest recorded enrollment and the absence of a reported sample-size justification limit what can be inferred about precision and planning. The reported p-value should not be converted into a claim about the probability that a treatment effect exists. Statistical interpretation is strongest when estimates, uncertainty intervals, assumptions, and design details are considered together.

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.

The registry record leaves several trial-specific details unclear. It identifies the design as randomized, single-masked, and parallel, but does not describe sequence generation, allocation concealment, who was masked, or how masking was maintained. It also does not report detailed eligibility criteria or baseline characteristics. These omissions make it difficult to assess selection processes and the comparability of participants beyond the randomization label.

The treatment-as-usual group could receive any antipsychotic other than clozapine, but the record does not identify which medicines participants received or how those choices were made. It does not report the clozapine dose, treatment adjustments, adherence information, or additional care. The intervention contrast is therefore not fully characterized by the registry description.

The outcome was a retrospective weekly count assessed with the Timeline Followback Scale. The record does not give further information on administration, completeness, or measurement quality in this study. It also does not report the mixed model’s distributional assumptions, covariance structure, diagnostics, or handling of the count outcome. The negative lower bounds in the posted arm-level confidence intervals are model-based bounds, not possible individual consumption values; the record does not explain how they were generated.

The sponsor reports a between-group estimate and p-value without a confidence interval for the contrast. The registry does not report the missing-data pattern, sensitivity analyses, detailed analysis population, sample-size justification, interim analyses, or a complete statistical plan. It also does not report deaths. These limitations constrain interpretation of the primary result and safety reporting. They do not establish that any unreported procedure or event did or did not occur.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Treatment of Schizophrenia and Comorbid Cannabis Use Disorder: Comparing Clozapine to Treatment-as-Usual — NCT00498550.
• Registry record last updated 2019-03-13; status last verified 2019-02.
• Results first posted 2012-01-19.
• 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

The registry describes a completed, randomized, single-masked, parallel-group Phase 4 trial comparing clozapine with treatment as usual using another antipsychotic. Recorded enrollment was 31. The primary outcome was average cannabis-use intensity in joints per week from Week 1 to week 12, measured retrospectively with the Timeline Followback Scale.

The posted arm estimates were 0.02 for clozapine and 4.56 for treatment as usual. The sponsor’s mixed models analysis reported a difference in treatment means of -4.54 and p = .088, without a confidence interval for the contrast. The analysis used partial data while participants were adherent and relied on a missing-at-random assumption. Important design, analysis, and missing-data details are not reported in the registry record, so the result should be interpreted with those uncertainties in view.