This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
CAMP was a randomized, parallel, single-masked phase 4 treatment trial in patients with schizophrenia or schizoaffective disorder. The trial enrolled 215 participants and compared a stay strategy with a switch strategy, using change in non-HDL cholesterol from baseline to 24 weeks as the primary efficacy endpoint.
| Feature | CAMP |
|---|---|
| Official title | Comparison of Antipsychotics for Metabolic Problems in Schizophrenia or Schizoaffective Disorder |
| Phase | Phase 4 |
| Status | Completed |
| Population | Schizophrenia; Schizoaffective Disorder |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Primary purpose | Treatment |
| Enrollment | 215 |
| Start date | 2007-01 |
| Primary completion date | 2009-10 |
| Lead sponsor | National Institute of Mental Health (NIMH) |
| Sponsor type | NIH |
| ClinicalTrials.gov | NCT00423878 |
2. Clinical Question
The statistical question was whether assigning patients to a switch strategy rather than a stay strategy produced a different change in non-HDL cholesterol over 24 weeks.
Population
Patients with schizophrenia or schizoaffective disorder enrolled in the phase 4 CAMP treatment trial.
Intervention strategy
Assignment to the Switch Group.
Comparator strategy
Assignment to the Stay Group.
Primary question
Does change in non-HDL cholesterol from baseline to 24 weeks differ between the stay and switch treatment groups?
3. Trial Design
Switch Group
- Assigned to the switch treatment strategy.
- Serious adverse events: 18/107.
Stay Group
- Assigned to the stay treatment strategy.
- Serious adverse events: 9/106.
The intervention list in the registry contains four antipsychotic drugs: Risperidone, Olanzapine, Quetiapine, and Aripiprazole. The primary endpoint, however, is defined at the treatment-strategy level as a comparison between patients assigned to stay and patients assigned to switch.
4. Endpoints
| Endpoint | Registry definition | Time frame | Registered analysis |
|---|---|---|---|
| Change in Non-HDL Cholesterol Level for Patients Assigned to Stay and Patients Assigned to Switch Over 24 Weeks | Change in non-HDL cholesterol measured at baseline and every 4 weeks for 24 weeks. The efficacy analysis corresponded to a comparison of change in non-HDL cholesterol from baseline to 24 weeks between treatment groups (stay versus switch). | 24 weeks | Repeated measurements mixed effects linear models were fit for the primary analysis. |
Why repeated measurements matter
The primary endpoint was not limited to a single measurement taken at the end of follow-up. Non-HDL cholesterol was measured at baseline and every 4 weeks for 24 weeks. That creates a longitudinal dataset in which each participant can contribute multiple observations.
A repeated-measures model is therefore well suited to the structure described in the registry. Rather than treating each measurement as independent, the model can account for the fact that measurements from the same participant are correlated.
5. Statistical Methodology
The registry specifies repeated measurements mixed effects linear models for the primary analysis. The stated efficacy comparison is the change in non-HDL cholesterol from baseline to 24 weeks between the stay and switch treatment groups.
Here, Yij represents a participant's non-HDL cholesterol measurement at a particular time point. A mixed-effects framework distinguishes population-level treatment and time effects from participant-level variation.
Why a mixed-effects model fits this endpoint
Each participant has repeated non-HDL cholesterol measurements over time. Those observations are naturally related because measurements from the same person tend to be more alike than measurements from unrelated participants. A mixed-effects model provides a way to represent that within-participant structure while estimating the treatment-group comparison.
Baseline and follow-up measurements
The endpoint definition explicitly includes measurement at baseline and every 4 weeks through 24 weeks. The primary efficacy comparison is framed as the difference in change from baseline to 24 weeks between the stay and switch groups.
Conceptually, for participant i, a simple change score can be written as:
The treatment comparison then concerns the difference in average change between the randomized strategies. The registry specifies a repeated-measures mixed-effects model rather than a simple comparison of change scores.
Longitudinal information versus a single endpoint
A major statistical advantage of repeated measurements is that the analysis can use information from the trajectory of non-HDL cholesterol rather than discarding the intermediate observations. This is especially relevant when measurements are obtained at several scheduled time points.
6. Results
The ClinicalTrials.gov record indicates that results have been posted, but its statistical-analysis section contains no posted formal statistical analyses for the primary endpoint. The record does report serious adverse events by treatment group.
Primary endpoint: non-HDL cholesterol
| Element | Registry information |
|---|---|
| Endpoint | Change in Non-HDL Cholesterol Level for Patients Assigned to Stay and Patients Assigned to Switch Over 24 Weeks |
| Measurement schedule | Baseline and every 4 weeks for 24 weeks |
| Comparison | Change in non-HDL cholesterol from baseline to 24 weeks between treatment groups (stay versus switch) |
| Specified statistical model | Repeated measurements mixed effects linear models |
| Effect estimate | Not reported in the posted statistical analyses |
| Confidence interval | Not reported in the posted statistical analyses |
| P-value | Not reported in the posted statistical analyses |
The clinically relevant quantity for this endpoint is the difference in change in non-HDL cholesterol between the switch and stay strategies over 24 weeks. A negative or positive estimated difference would describe the direction and magnitude of the between-group change, depending on how the contrast is defined.
The estimate would not mean that every participant experienced that amount of change. It would represent a group-level treatment contrast from the specified longitudinal analysis.
A confidence interval would describe the statistical precision of that estimated contrast. A narrow interval would indicate greater precision, while a wide interval would indicate greater uncertainty. The interval would not describe the range of individual patients' cholesterol changes.
A p-value would address the evidence against the relevant null hypothesis under the specified statistical model. It would not measure the size or clinical importance of the treatment effect.
Because the primary endpoint uses repeated measurements, interpretation also depends on the assumptions of the mixed-effects model and on how within-participant correlation and incomplete longitudinal observations are handled.
7. Safety Results
The registry reports serious adverse events by treatment group. These counts are presented as affected participants divided by participants at risk.
| Safety measure | Switch Group | Stay Group |
|---|---|---|
| Serious adverse events | 18/107 | 9/106 |
Reported serious adverse events
Switch Group vs Stay Group
These figures describe the number of participants affected among those at risk in each group. They do not by themselves establish a causal difference in safety between the strategies. A formal comparison would require the prespecified safety analysis, including its definition of the event, analysis population, follow-up, and statistical method.
8. Statistical Methods Explained
Why was a mixed-effects model used?
The primary endpoint contains repeated non-HDL cholesterol measurements from the same participants. A mixed-effects model is designed for this type of longitudinal structure because it can represent both systematic treatment/time effects and participant-level variability.
Why not simply compare the 24-week measurements?
A comparison using only the 24-week measurement would discard the intermediate observations collected every 4 weeks. The registry instead specifies a repeated-measures analysis, allowing the longitudinal nature of the endpoint to be incorporated into the primary model.
What does "change from baseline" mean?
Change from baseline compares a participant's value after treatment with that participant's starting value. For the CAMP primary endpoint, the registry specifically defines the efficacy comparison as the difference in change from baseline to 24 weeks between the stay and switch groups.
Why is within-participant correlation important?
Measurements from the same participant are not independent observations. A person's non-HDL cholesterol at one time point is related to that person's measurement at another time point. Treating all repeated observations as unrelated would misrepresent the structure of the data and can lead to incorrect uncertainty estimates.
What would a treatment-effect estimate mean?
The treatment-effect estimate would quantify the difference in modeled change between the two assigned strategies. Its interpretation is a comparison of groups, not a statement that every patient should experience the same amount of change.
What would a confidence interval add?
A confidence interval would place statistical uncertainty around the estimated between-group contrast. It is important because the point estimate alone does not show how precisely the trial estimated the treatment difference.
What would a p-value add?
A p-value would quantify the compatibility of the observed data with a specified null hypothesis under the statistical model. It is not a measure of effect size, clinical importance, or the probability that one treatment strategy is beneficial.
9. Understanding the Longitudinal Analysis
The primary endpoint is unusually well suited to thinking about the difference between a cross-sectional comparison and a longitudinal comparison.
Cross-sectional question
How different are the groups at one selected time point?
Longitudinal question
How do measurements evolve over time, and how does the change differ between groups?
CAMP endpoint
The registry frames the primary comparison around change from baseline to 24 weeks.
CAMP model
The registry specifies repeated measurements mixed effects linear models.
This distinction matters because repeated measurements contain information about the path taken between baseline and the final assessment. A longitudinal model can therefore provide a framework for using the scheduled measurements collectively rather than treating each time point as an unrelated experiment.
10. Interpreting Randomization
Randomization is the central design feature that supports the comparison between the stay and switch strategies. In a randomized trial, treatment assignment is determined independently of the participant's subsequent outcome, subject to the trial's allocation process.
This creates the foundation for a causal treatment comparison. The statistical model then addresses how the repeated outcome measurements are analyzed; it does not replace the role of randomization.
The validity of the causal comparison depends on the integrity of the randomized design and on appropriate handling of follow-up measurements and analysis populations.
The parallel design means the two assigned strategies are evaluated as separate treatment groups rather than as a sequence in which every participant receives both strategies.
11. Masking and Interpretation
The registry describes CAMP as single-masked. Masking can reduce the potential for knowledge of treatment assignment to influence aspects of trial conduct or outcome assessment.
The registry also states an important limitation: open-label treatment is a limitation of this study, particularly for outcomes not measured in the laboratory but instead subject to clinical judgment.
12. Statistical Interpretation of the Primary Endpoint
The primary endpoint asks whether the amount of change in non-HDL cholesterol from baseline to 24 weeks differs between patients assigned to stay and patients assigned to switch.
The repeated-measures mixed-effects model provides a statistical framework for analyzing the repeated non-HDL cholesterol measurements collected at baseline and every 4 weeks through 24 weeks. This accounts for the longitudinal structure rather than treating all observations as unrelated.
The primary endpoint does not by itself summarize all clinical outcomes associated with antipsychotic treatment. It is a specific metabolic laboratory endpoint. Safety outcomes, clinical outcomes, and other measures require their own interpretation.
13. Limitations
- Open-label treatment: the registry identifies open-label treatment as a limitation, particularly for outcomes not measured in the laboratory but subject to clinical judgment.
- No formal effect estimate in the posted statistical analyses: the registry identifies the primary endpoint and its mixed-effects model but does not provide a treatment-effect estimate, confidence interval, or p-value in the posted statistical-analysis information represented here.
- Longitudinal modeling assumptions: mixed-effects analyses require assumptions about the mean structure, participant-level random effects, residual variability, and within-participant dependence.
- Repeated observations: measurements from the same participant are correlated, making the specification of the longitudinal covariance structure an important modeling consideration.
- Safety interpretation: serious adverse-event counts are reported by group, but counts alone do not provide a complete comparative safety analysis.
- Endpoint scope: the primary endpoint is change in a laboratory measure, so it should not be interpreted as a comprehensive measure of overall clinical benefit or harm.
14. Why This Trial Matters Statistically
CAMP is a useful teaching example because its primary endpoint illustrates a common problem in clinical research: a clinically meaningful outcome may be measured repeatedly over time rather than at only one final visit.
| Concept | How it appears in CAMP |
|---|---|
| Randomization | Participants were randomized to treatment strategies. |
| Parallel-group design | The stay and switch strategies form separate treatment groups. |
| Longitudinal measurement | Non-HDL cholesterol was measured at baseline and every 4 weeks for 24 weeks. |
| Change from baseline | The primary efficacy comparison concerns change from baseline to 24 weeks. |
| Mixed-effects modeling | Repeated measurements mixed effects linear models were specified for the primary analysis. |
| Within-participant correlation | Repeated measurements from the same participant require a longitudinal statistical framework. |
| Laboratory endpoint | Non-HDL cholesterol is measured rather than judged clinically. |
| Safety comparison | Serious adverse events are reported by randomized treatment strategy. |
15. A Practical Framework for Reading the CAMP Analysis
A useful way to read the primary analysis is to separate the question into four layers.
Layer 1 · Design
Randomization determines the two treatment strategies being compared and provides the foundation for the between-group inference.
Layer 2 · Outcome
Non-HDL cholesterol is repeatedly measured from baseline through 24 weeks.
Layer 3 · Model
A repeated-measures mixed-effects linear model is specified to analyze the longitudinal measurements.
Layer 4 · Interpretation
The treatment contrast concerns the difference in change between the stay and switch strategies, with statistical uncertainty ideally expressed by an effect estimate and confidence interval.
This layered approach prevents a common statistical mistake: jumping directly from a treatment label to a conclusion without first identifying the estimand, the measurement schedule, the randomized comparison, and the model used to estimate the treatment effect.
16. Serious Adverse Events: Statistical Context
The reported serious adverse-event counts are 18/107 for the Switch Group and 9/106 for the Stay Group.
| Group | Affected | At risk | What the reported figure represents |
|---|---|---|---|
| Switch Group | 18 | 107 | Participants with serious adverse events divided by participants at risk |
| Stay Group | 9 | 106 | Participants with serious adverse events divided by participants at risk |
The denominators differ from the overall enrollment figure of 215 because the safety information is presented specifically as affected participants divided by participants at risk. This is an important reminder that a trial's enrollment population and the population contributing to a particular safety analysis need not be identical.
A formal safety comparison could use an appropriate measure of association or difference depending on the prespecified analysis. The choice would depend on whether the analysis concerns participants with at least one event, event counts, time to first event, or another safety estimand.
17. Trial Timeline
Trial start
The CAMP trial began in January 2007.
Primary completion
The registry lists October 2009 as the primary completion date.
Trial status
The ClinicalTrials.gov record lists the study status as completed.
18. Four Antipsychotic Interventions in the Record
| Intervention | Registry classification |
|---|---|
| Risperidone | Drug |
| Olanzapine | Drug |
| Quetiapine | Drug |
| Aripiprazole | Drug |
The presence of four antipsychotic interventions in the registry should not be confused with the number of randomized comparison groups. The study is described as having 2 arms, and the primary endpoint explicitly compares patients assigned to the stay and switch treatment groups.
19. What a Complete Primary Analysis Would Usually Report
For a longitudinal continuous endpoint such as the one specified in CAMP, a statistically complete presentation would ordinarily identify the estimated treatment contrast, its uncertainty, and the inferential test associated with the prespecified model.
| Component | Purpose |
|---|---|
| Treatment contrast | Quantifies the estimated difference in change between the switch and stay strategies. |
| Confidence interval | Shows the statistical precision of the estimated treatment contrast. |
| P-value | Assesses compatibility with the specified null hypothesis under the model. |
| Model specification | Defines how repeated measurements, treatment, time, and participant-level variability are represented. |
| Analysis population | Defines which participants contribute to the efficacy analysis. |
| Missing-data approach | Describes how incomplete longitudinal measurements are handled within the analysis. |
The CAMP registry record identifies the primary endpoint and the repeated-measures mixed-effects model, but the posted statistical-analysis information represented here does not contain the corresponding numerical treatment estimate, confidence interval, or p-value.
20. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary analysis is a longitudinal comparison of change in non-HDL cholesterol between randomized stay and switch groups, using repeated measurements mixed effects linear models.
Clinical interpretation
The primary endpoint concerns a metabolic laboratory measure over 24 weeks. Its interpretation should remain distinct from serious adverse events and from outcomes that require clinical judgment.
The distinction is important because statistical significance and clinical importance are different concepts. Even when a treatment contrast is statistically distinguishable from a null value, its clinical relevance depends on the magnitude of the change, the clinical context, and the consequences of treatment.
21. What CAMP Teaches About Repeated-Measures Clinical Trials
The design illustrates several general principles that recur throughout clinical biostatistics.
| Question | Statistical issue |
|---|---|
| What is being compared? | Stay versus switch treatment strategies. |
| What is measured? | Non-HDL cholesterol. |
| When is it measured? | At baseline and every 4 weeks for 24 weeks. |
| What is the primary contrast? | Difference in change from baseline to 24 weeks between treatment groups. |
| Why use a longitudinal model? | Each participant contributes repeated measurements over time. |
| What model is specified? | Repeated measurements mixed effects linear models. |
| What safety information is reported? | Serious adverse events by treatment group. |
For students and analysts, the central lesson is that the statistical model should follow the structure of the endpoint. Repeated measurements call for methods that recognize repeated observations, rather than treating every measurement as an independent observation.
22. Limitations of the Available Statistical Reporting
The ClinicalTrials.gov record contains the primary endpoint definition and the stated primary model, but the statistical-analysis information represented here does not provide the numerical treatment effect for the primary endpoint.
That distinction is important for interpretation. Knowing the endpoint and model tells the reader what was intended to be estimated and how it was modeled; it does not provide the magnitude or precision of the observed treatment contrast.
23. Record Summary
CAMP was a completed randomized, parallel, single-masked phase 4 treatment trial enrolling 215 participants with schizophrenia or schizoaffective disorder. The trial compared stay and switch treatment strategies and included risperidone, olanzapine, quetiapine, and aripiprazole among its listed drug interventions.
The primary endpoint was change in non-HDL cholesterol from baseline to 24 weeks, with measurements collected at baseline and every 4 weeks. The registry specifies repeated measurements mixed effects linear models for the primary analysis. This makes CAMP a useful example of how randomized treatment comparisons can be combined with longitudinal statistical modeling.
The registry reports serious adverse events of 18/107 in the Switch Group and 9/106 in the Stay Group. It also identifies open-label treatment as a limitation, particularly for outcomes not measured in the laboratory but subject to clinical judgment.
From a biostatistical perspective, the most important feature of CAMP is the alignment between the repeated-measurement endpoint and the mixed-effects analysis. The treatment comparison is not simply a comparison of two isolated measurements; it is framed around change over time within a randomized parallel-group trial.
Explore more Clinical Biostats trial analyses
Clinical trial results provide practical examples of how study design, endpoints, longitudinal measurements, and statistical models work together in applied research.
24. Sources
- ClinicalTrials.gov: CAMP — NCT00423878.