This page separates registry-reported trial information from statistical interpretation. The ClinicalTrials.gov record provides the official trial registry record and identifies the primary endpoint as change in angina symptom score between groups at 12 weeks.
1. Trial at a Glance
ORBITA-2 was a randomized, parallel-group, quadruple-masked, placebo-controlled trial of percutaneous coronary intervention for the relief of stable angina. The study enrolled 439 participants and had two arms. Its registered primary endpoint was change in angina symptom score between groups at 12 weeks.
| Feature | ORBITA-2 |
|---|---|
| Study acronym | ORBITA-2 |
| Clinical question | Relief of stable angina |
| Intervention | Percutaneous coronary intervention |
| Phase | NA |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 439 |
| Number of arms | 2 |
| Lead sponsor | Imperial College London |
| Sponsor type | Other |
| Status | Completed |
| Start date | 2018-11-12 |
| Primary completion date | 2023-09-06 |
| ClinicalTrials.gov | NCT03742050 |
2. Clinical Question
The registered clinical question concerns whether percutaneous coronary intervention affects angina symptoms in people with stable angina. The primary comparison is expressed directly as a between-group change in angina symptom score over a 12-week time frame.
Population
The registered condition is stable angina.
Intervention
Percutaneous coronary intervention, recorded as a procedure.
Comparator
The study is described as placebo-controlled, with two randomized parallel arms.
Primary question
How does change in angina symptom score differ between the randomized groups at 12 weeks?
3. Trial Design
The parallel randomized structure is statistically important because the principal comparison is between outcomes observed in groups established by random allocation. Randomization provides the framework for attributing between-group differences to the intervention, subject to adherence, missing observations, measurement, and other features of the analysis.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Primary endpoint | Change in angina symptom score between groups | 12 weeks |
The endpoint is a change score, meaning that the statistical comparison concerns the difference between groups in how the angina symptom score changes over the specified 12-week period. The registry identifies the between-group comparison rather than reporting a numerical result in its posted statistical analyses.
5. Statistical Structure of the Primary Endpoint
A change-from-baseline endpoint creates several related quantities that should not be confused. For each participant, the change can be represented conceptually as:
The primary treatment comparison then concerns the distribution or mean change in the two randomized groups. The clinically relevant statistical quantity is the between-group difference in change, rather than a change observed in only one arm.
For a continuous symptom score, a common analysis would compare the mean change between randomized groups. An analysis of covariance (ANCOVA) can also be used, with the 12-week score as the outcome and baseline score included as a covariate. The choice between a simple change-score analysis and an ANCOVA approach depends on the prespecified statistical analysis plan and the properties of the symptom measure.
Why compare changes between groups?
Patients can experience changes in symptoms over time even without an active treatment effect. A randomized between-group comparison separates general time-related change from the difference associated with assignment.
Why baseline adjustment can help
Baseline symptom severity can explain variation in follow-up scores. ANCOVA uses baseline information directly and can improve precision when the baseline measure is strongly related to the follow-up outcome.
The central estimand should therefore be stated clearly. If the analysis is based on mean change, it is the difference in mean change between randomized groups. If ANCOVA is prespecified instead, the corresponding adjusted group contrast at 12 weeks is the principal estimate. These quantities can be related but are not automatically identical.
6. Planned Analysis
Primary endpoint
Between-group comparison · 12 weeks
The registry identifies change in angina symptom score between groups as the primary endpoint at 12 weeks. No formal statistical analyses are posted in the ClinicalTrials.gov record.
How an endpoint of this type is typically analysed
For a continuous symptom score measured at baseline and again at 12 weeks, the principal analysis would ordinarily estimate a between-group treatment contrast. A straightforward approach is a comparison of mean changes:
The estimate is interpreted on the original symptom-score scale. Its sign depends on how improvement is defined by the scoring system; the registry record identifies the endpoint but does not provide the numerical score definition in the information reported here.
Another common approach is ANCOVA, modelling the 12-week symptom score while adjusting for baseline symptom score and treatment group. This can provide a more statistically efficient comparison when baseline and follow-up scores are correlated.
The statistical analysis should also specify the analysis population, handling of missing 12-week assessments, treatment adherence rules, any prespecified covariates, and the confidence interval and hypothesis-testing framework. Those details determine exactly what the estimated treatment contrast represents.
7. Statistical Methods
Change-score analysis
When an outcome is recorded at baseline and follow-up, one natural approach is to calculate an individual change score and compare those changes between randomized groups. If improvement corresponds to a decrease in the symptom score, a negative between-group difference would indicate greater average reduction in the intervention group. If improvement corresponds to an increase, the direction would reverse.
The direction cannot be interpreted correctly without knowing how the particular symptom score is scaled. The registry's endpoint wording identifies change in angina symptom score but does not provide the score's numerical orientation in the information reported here.
ANCOVA
ANCOVA is a common alternative for continuous outcomes measured at baseline and follow-up. Instead of first subtracting baseline from follow-up, the model predicts the follow-up score using treatment group and baseline score.
Here, β1 represents the adjusted treatment-group contrast at 12 weeks. The equation is a general illustration of an appropriate model for this type of endpoint, not a claim about the specific model used in the registry.
Confidence intervals
A confidence interval around the treatment contrast quantifies statistical uncertainty under the specified model and sampling framework. A narrow interval indicates greater precision, while a wide interval indicates less precision. The interval describes uncertainty in the estimated population contrast; it does not describe the range of symptom changes experienced by individual participants.
Hypothesis testing
A conventional two-group hypothesis test would evaluate a null hypothesis that the treatment contrast is zero. The p-value describes the compatibility of the observed data with that null hypothesis under the assumed model. It is not a measure of the magnitude of the treatment effect.
For this type of endpoint, the most informative reporting would pair the estimated between-group difference in symptom-score change with a confidence interval and p-value. The effect estimate communicates magnitude, the confidence interval communicates precision, and the p-value addresses evidence against a specified null hypothesis.
A small p-value would not by itself indicate a large or clinically important improvement, just as a confidence interval that includes zero would not establish that the two groups are clinically identical.
8. Statistical Methods Explained
Why analyse change in symptom score rather than only the 12-week score?
A change endpoint accounts directly for the participant's starting symptom level. Comparing change between randomized groups asks whether the amount of improvement or worsening differs between groups over the 12-week period.
Why might ANCOVA be used?
ANCOVA can account for baseline symptom score while comparing 12-week outcomes. If baseline score predicts follow-up score, incorporating it into the model can reduce residual variability and improve precision. It also avoids treating baseline measurement simply as a quantity that must be subtracted.
What is the treatment effect for a continuous endpoint?
For a mean-change analysis, the treatment effect is the difference between the two group means of individual change scores. The value remains on the symptom-score scale, making it directly interpretable in terms of the instrument used to measure symptoms.
Why is the confidence interval important?
The confidence interval shows how precisely the treatment contrast has been estimated. An estimate near zero with a very narrow interval tells a different statistical story from an estimate near zero with a very wide interval, even if both have similar point estimates.
Why does the p-value not measure effect size?
The p-value depends on both the estimated effect and its statistical uncertainty. Sample size and variability can therefore affect the p-value substantially. The magnitude of the effect should be read from the treatment estimate itself, with its confidence interval providing the associated precision.
Why does randomization matter?
Randomization creates the basis for a causal comparison because treatment assignment is determined independently of participants' baseline characteristics, subject to the usual limitations of randomization and subsequent trial conduct. The resulting analysis should preserve that randomized comparison rather than selectively comparing participants according to post-randomization characteristics.
9. Missing Data and Follow-Up Considerations
A 12-week change endpoint requires information from the relevant baseline and 12-week assessments. Participants without an evaluable follow-up measurement create a missing-data problem that must be addressed in the statistical analysis plan.
Complete-case analysis
A complete-case analysis uses participants with the required measurements. It is simple but can introduce bias if missingness is related to prognosis or treatment response.
Model-based approaches
Mixed-effects or likelihood-based methods can use available repeated measurements under explicit assumptions about the missing-data mechanism.
Multiple imputation
Multiple imputation can represent uncertainty about missing values by creating several completed datasets and combining the resulting estimates.
Sensitivity analysis
Alternative assumptions about missing outcomes can test how dependent the treatment conclusion is on the primary missing-data assumption.
For a patient-reported symptom endpoint, missingness can be particularly important because failure to provide a follow-up symptom assessment may itself be related to symptom burden, treatment experience, or other participant characteristics. The credibility of the primary analysis therefore depends not only on the statistical test but also on the prespecified strategy for incomplete observations.
10. Randomization and Causal Interpretation
The randomized design gives the primary between-group comparison a clear causal framework. Before treatment assignment, the groups are intended to be exchangeable in expectation. Consequently, a difference in the primary endpoint can be interpreted as a treatment-assignment effect under the assumptions required for the analysis.
| Feature | Statistical implication |
|---|---|
| Randomized allocation | Supports causal comparison between assigned treatment groups. |
| Parallel design | Each participant contributes to one randomized treatment pathway rather than crossing between treatment conditions. |
| Quadruple masking | Reduces the potential influence of treatment knowledge on trial conduct and symptom assessment. |
| Placebo-controlled structure | Provides a comparator for interpreting changes in angina symptoms over the study period. |
| Two arms | Creates a direct randomized comparison between the intervention and comparator groups. |
The causal interpretation applies to the randomized treatment comparison represented by the study. It does not automatically establish that every individual participant would experience the same change, nor does it eliminate uncertainty arising from measurement error, missing data, nonadherence, or limitations in the enrolled population.
11. Masking and Measurement of Symptoms
The quadruple-masked design is particularly relevant to an endpoint involving angina symptoms. Subjective outcomes can be influenced by expectations, interactions with study personnel, and knowledge of treatment assignment. Masking is therefore part of the statistical credibility of the endpoint, not merely a procedural detail.
A between-group difference in a symptom score reflects both the underlying clinical experience and the properties of the measurement process. Randomization addresses treatment-group comparability, while masking helps limit expectation-related influences on how symptoms are reported or assessed.
For a symptom-based endpoint, the statistical analysis should therefore be considered together with the design features that determine how the measurement was obtained. A mathematically precise estimate cannot compensate for systematic differences in outcome measurement between randomized groups.
12. Primary Endpoint: What the Estimate Would Mean
Because the registry identifies the primary endpoint as a between-group change in angina symptom score, the central estimand is a contrast between randomized groups over the 12-week period.
Conceptual treatment contrast
Mean change in the intervention group minus mean change in the comparator group
If the score is oriented so that lower values represent fewer symptoms, a negative treatment contrast would indicate greater symptom reduction in the intervention group. If higher values represent improvement, the interpretation would be reversed. The direction of benefit therefore depends on the scoring convention of the angina symptom instrument.
A between-group mean change is not the same as the change experienced by every participant. Some individuals can improve more than the group average, some less, and some can move in the opposite direction. The group contrast describes an average treatment difference under the chosen analysis.
The confidence interval would show the statistical precision of the estimated between-group difference. It would not show the range of individual symptom responses or the probability that a particular participant benefits.
The p-value would quantify evidence against a prespecified null hypothesis under the statistical model. It would not quantify clinical importance, the percentage of participants who improved, or the magnitude of benefit.
13. Multiplicity and Endpoint Hierarchy
The registry identifies one registered primary endpoint: change in angina symptom score between groups at 12 weeks. Because only this primary endpoint is identified in the information reported here, there is no basis for constructing a broader confirmatory endpoint hierarchy.
| Analysis component | Role | Statistical consideration |
|---|---|---|
| Change in angina symptom score | Registered primary endpoint | Between-group comparison at 12 weeks |
| Additional endpoints | Not identified in the registered primary-endpoint information | No additional multiplicity structure can be established from the reported registry information |
Multiplicity becomes important whenever several hypotheses are tested and interpreted as confirmatory evidence. If additional outcomes or subgroup comparisons are examined, their role should be distinguished from the registered primary endpoint and any prespecified multiplicity-control strategy.
14. Interim Analysis, Non-Inferiority, and Other Design Features
The registry information reported here identifies a randomized parallel design with quadruple masking and a single registered primary endpoint. It does not provide a non-inferiority margin, crossover design, factorial structure, Bayesian analysis, interim-analysis specification, or multiplicity procedure.
Non-inferiority margin
No non-inferiority margin is identified in the reported registry information. The primary endpoint is therefore presented as a between-group change rather than as a non-inferiority comparison.
Crossover
The registered design model is parallel. No crossover structure is identified.
Factorial design
The registered design model specifies two parallel arms rather than a factorial design.
Bayesian analysis
No Bayesian statistical method is identified in the reported registry information.
15. Safety and Secondary Endpoints
The ClinicalTrials.gov information reported here identifies the primary endpoint as change in angina symptom score between groups at 12 weeks. It does not provide posted statistical analyses or numerical serious-adverse-event results by arm in the information available for this analysis.
Safety and secondary efficacy outcomes should ordinarily be evaluated separately from the primary symptom endpoint. Safety summaries commonly describe the number and proportion of participants experiencing adverse events, while secondary continuous outcomes may require their own treatment contrasts and confidence intervals. The exact estimand and analysis population should follow the prespecified statistical analysis plan.
16. Trial Timeline
Study start
The registered study start date was November 12, 2018.
439 participants
The registry reports total enrollment of 439 participants in two arms.
12-week assessment
The registered primary endpoint is change in angina symptom score between groups at 12 weeks.
Primary completion
The registered primary completion date was September 6, 2023.
17. How to Read a Future Result for ORBITA-2
If a formal numerical analysis is subsequently reported, the first quantity to examine should be the estimated between-group treatment contrast for change in angina symptom score. The statistical interpretation should then proceed in a defined sequence.
| Question | What to examine |
|---|---|
| What is the estimated effect? | The between-group difference in symptom-score change, or the prespecified adjusted treatment contrast. |
| How precise is it? | The corresponding confidence interval. |
| What is the evidence against the null? | The prespecified hypothesis test and p-value. |
| What population was analysed? | The prespecified efficacy analysis population and any exclusions from the randomized cohort. |
| How were missing observations handled? | The primary missing-data strategy and relevant sensitivity analyses. |
| Does the scale have a clear clinical interpretation? | The direction and magnitude of change should be interpreted using the symptom instrument's scoring convention. |
This sequence prevents a common statistical mistake: treating the p-value as the result itself. The estimated treatment contrast is the primary measure of magnitude, while the confidence interval and hypothesis test provide complementary information about uncertainty and compatibility with the null hypothesis.
18. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record contains no posted statistical analyses for the registered primary endpoint.
- Limited endpoint detail: the registry wording identifies change in angina symptom score between groups at 12 weeks but does not provide the numerical scoring properties needed to determine the direction of clinical improvement from a raw score change alone.
- Analysis specification: the exact primary statistical model, covariates, hypothesis-testing procedure, and confidence-interval construction are not identified in the registry information reported here.
- Missing data: interpretation of a 12-week change endpoint depends on how participants without evaluable follow-up measurements are handled.
- Measurement considerations: symptom scores are measurements rather than direct physiological quantities, making masking and consistent assessment important components of the design.
- Generalizability: the study population is defined by the registered condition and eligibility criteria; conclusions about populations outside the enrolled study population require additional evidence.
- Clinical versus statistical significance: even a statistically precise between-group difference should be interpreted in the context of the symptom scale and the magnitude of change that is clinically meaningful.
19. Why This Trial Matters Statistically
ORBITA-2 provides a useful teaching example because the central outcome is not a binary event or a time-to-event endpoint. Instead, the registered primary endpoint is a change in a symptom score compared between randomized groups. That shifts the statistical focus toward continuous-outcome estimation, baseline adjustment, measurement, missing data, and interpretation of absolute differences.
| Concept | How it appears in ORBITA-2 |
|---|---|
| Randomization | Participants were randomly allocated to two parallel arms. |
| Placebo control | The trial is described as placebo-controlled. |
| Quadruple masking | The registry identifies quadruple masking. |
| Continuous/change endpoint | The registered primary endpoint is change in angina symptom score between groups. |
| Between-group treatment effect | The primary question concerns the difference in symptom-score change between randomized groups. |
| Baseline adjustment | ANCOVA is a common statistical framework for a baseline-and-follow-up continuous outcome. |
| Confidence intervals | Precision should accompany any estimated between-group treatment contrast. |
| Missing-data analysis | Participants without complete 12-week information require a prespecified analytical strategy. |
| Measurement validity | Because the primary outcome is a symptom score, masking and consistent outcome assessment are important to interpretation. |
The trial is therefore useful for understanding an important distinction in clinical-trial statistics: the statistical design determines how an effect is estimated, while the measurement scale determines what that effect means clinically. A sophisticated analysis cannot compensate for an unclear endpoint definition, and a well-defined endpoint still requires an analysis that respects randomization and uncertainty.
20. Statistical Methods Explained in Practice
Change versus endpoint
A change analysis asks how much each participant moved from baseline to 12 weeks. An endpoint analysis asks where participants are at 12 weeks. ANCOVA can incorporate the baseline measurement while modelling the follow-up endpoint.
Unadjusted versus adjusted
An unadjusted mean-change comparison uses the observed changes directly. An adjusted model incorporates baseline information or other prespecified covariates. These analyses answer closely related questions but can produce different estimates.
Statistical versus clinical importance
A confidence interval and p-value describe statistical evidence. Clinical importance requires interpretation on the symptom scale and consideration of the magnitude of the observed difference.
Individual versus average response
A mean treatment effect describes an average contrast between randomized groups. It does not imply that every participant experienced the same direction or magnitude of symptom change.
21. Sources
- ClinicalTrials.gov: ORBITA-2, NCT03742050.