This page separates reported trial results from statistical interpretation. The numerical results presented here are restricted to the ClinicalTrials.gov record and its posted statistical analyses.
1. Trial at a Glance
COMBI-AD was a randomized, parallel, double-blind phase 3 trial in melanoma. The trial enrolled 870 participants and compared dabrafenib with trametinib against dabrafenib and trametinib placebos, with relapse-free survival as a primary time-to-event endpoint.
| Feature | COMBI-AD |
|---|---|
| Trial name | COMBI-AD |
| NCT identifier | NCT01682083 |
| Phase | Phase 3 |
| Condition | Melanoma |
| Population | High-risk BRAF V600 mutation-positive melanoma |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 870 |
| Interventions | Dabrafenib; trametinib; placebos |
| Lead sponsor | Novartis Pharmaceuticals |
| Sponsor type | Industry |
| Trial status | Completed |
| Start | 08-Jan-2013 |
| Primary completion | 30-Jun-2017 |
2. Clinical Question
The central statistical question was whether adjuvant dabrafenib with trametinib changed relapse-related time-to-event outcomes compared with dabrafenib and trametinib placebos in participants with high-risk BRAF V600 mutation-positive melanoma.
Population
Participants with high-risk BRAF V600 mutation-positive melanoma enrolled in the phase 3 COMBI-AD trial.
Intervention
Dabrafenib with trametinib.
Comparator
Dabrafenib and trametinib placebos.
Primary question
Does combination therapy alter relapse-free survival relative to the placebo comparison?
3. Trial Design
Dabrafenib + Trametinib
- Dabrafenib
- Trametinib
- Randomized treatment assignment
- Primary efficacy analysis in the ITT population
Dabrafenib + Trametinib Placebos
- Dabrafenib placebo
- Trametinib placebo
- Randomized treatment assignment
- Primary efficacy analysis in the ITT population
The design has two important features for statistical interpretation. First, randomization establishes the treatment groups as the basis for the efficacy comparison. Second, double masking reduces the opportunity for knowledge of treatment assignment to influence trial conduct or outcome assessment.
4. Randomization, Stratification, and Analysis Population
The posted statistical analyses identify the Intent-to-Treat (ITT) population as the analysis population for the primary and secondary efficacy comparisons. The secondary analyses of overall survival, distant metastasis-free survival, and freedom from relapse were adjusted for the randomized strata of disease stage and BRAF mutation status.
| Analysis feature | Registry-supported description |
|---|---|
| Primary efficacy population | Intent-to-Treat (ITT) Population |
| Primary comparison | Dabrafenib and Trametinib Combination Therapy vs Dabrafenib and Trametinib Placebos |
| Primary test | Log-rank test |
| Primary effect measure | Hazard ratio |
| Primary hypothesis | Superiority |
| Randomized strata identified in posted secondary analyses | Disease stage and BRAF mutation status |
| Stratified estimator identified | Pike estimator |
The ITT principle is important because it preserves the comparison created by randomization. A participant remains part of the analysis group to which that participant was randomized, rather than being reassigned according to treatment actually received.
5. Primary Endpoints
| Endpoint | Registered time frame | Type | Definition |
|---|---|---|---|
| Percentage of Participants With Relapse-free Survival (RFS) Events | Approximately 4 years | Time-to-event | Relapse-Free Survival was defined as the time from randomization to disease recurrence or death from any cause. RFS events included loco-regional recurrence, distant metastases, new primary melanoma, and death without prior documented recurrence; such deaths were counted as events and not censored. |
| Relapse-free Survival (RFS) | Approximately 4 years | Time-to-event | Relapse-Free Survival was defined as the time from randomization to disease recurrence or death from any cause. RFS events included loco-regional recurrence, distant metastases, new primary melanoma, and death without prior documented recurrence; such deaths were counted as events and not censored. |
6. Statistical Methodology
Log-rank testing
The posted primary analysis used a log-rank test to compare RFS between the randomized treatment groups. The log-rank test is designed for time-to-event data and evaluates whether the event experience differs between groups over follow-up while accounting for the timing of events and censoring.
The registry's primary analysis notes specify the null hypothesis as a hazard ratio of 1 and the alternative as a hazard ratio different from 1. The posted hypothesis type is superiority.
Hazard ratio
The primary effect measure was a hazard ratio. The registry reports that the hazard ratio was obtained from the stratified Pike estimator. A hazard ratio below 1 indicates a lower estimated hazard with dabrafenib plus trametinib compared with placebo.
A hazard ratio is a relative time-to-event measure. It is not an absolute probability, an absolute risk difference, or a statement that every individual participant experiences the same proportional reduction.
Stratified analysis
The posted secondary analyses state that the hazard ratio and p-value from the stratified log-rank test were adjusted for randomized strata of disease stage and BRAF mutation status. Stratification allows the comparison to account for these prespecified randomized strata rather than treating the entire study population as if those factors had not entered the randomization scheme.
Intention-to-treat analysis
All four posted statistical analyses identify the ITT population. For a randomized trial, analyzing participants according to their randomized assignment maintains the treatment comparison generated by randomization and avoids redefining the efficacy population based on events that occur after assignment.
Censoring in RFS
RFS is a time-to-event endpoint, so participants who have not experienced an RFS event by the analysis cut-off contribute follow-up information until the point at which they are censored. The registry specifically states that death without prior documented recurrence counts as an RFS event rather than being censored.
7. Primary Results: Relapse-Free Survival
The posted formal statistical analysis for RFS compared dabrafenib and trametinib combination therapy with dabrafenib and trametinib placebos in the ITT population over the registered time frame of approximately 4 years.
Hazard ratio for relapse-free survival
95% CI: 0.39–0.58 · P < 0.0001
Two-sided confidence interval · Superiority hypothesis
| Primary analysis feature | Reported result |
|---|---|
| Outcome | Relapse-free Survival (RFS) |
| Time frame | Approximately 4 years |
| Analysis population | Intent-to-Treat (ITT) Population |
| Comparison | Dabrafenib and Trametinib Combination Therapy vs Dabrafenib and Trametinib Placebos |
| Method | Log Rank |
| Effect measure | Hazard Ratio, log |
| Estimate | 0.47 |
| 95% CI | 0.39–0.58 |
| P-value | < 0.0001 |
| Hypothesis | Superiority |
| Estimator | Stratified Pike estimator |
An HR of 0.47 means that the estimated hazard of an RFS event was approximately 53% lower with dabrafenib plus trametinib than with the placebo comparison, under the time-to-event analysis reported by the registry.
The HR does not mean that 53% of participants avoided relapse, that each participant had exactly a 53% lower probability of relapse, or that the absolute difference in relapse-free survival was 53 percentage points. It is a relative measure of the event hazard.
The 95% CI of 0.39–0.58 describes the statistical uncertainty around the estimated hazard ratio. It indicates that the reported estimate is not being presented as an exact population value; the interval quantifies uncertainty under the stated statistical framework.
The P < 0.0001 result addresses evidence against the null hypothesis of a hazard ratio of 1. It does not measure the magnitude or clinical importance of the treatment effect. Effect size is conveyed by the HR and its confidence interval.
Because this is a time-to-event analysis, interpretation also depends on censoring and on the relationship between hazards over time. A single hazard ratio summarizes a relative event-hazard comparison rather than providing the complete shape of the underlying RFS curves.
The other registered primary RFS measure
The registry also lists Percentage of Participants With Relapse-free Survival (RFS) Events as a primary endpoint with a time frame of approximately 4 years. The posted statistical analyses in the ClinicalTrials.gov record do not provide a separate estimate, confidence interval, or p-value for that endpoint. The formal posted primary analysis is the RFS time-to-event comparison summarized above.
8. Secondary Endpoint Results
Three secondary time-to-event analyses are included in the ClinicalTrials.gov record: overall survival, distant metastasis-free survival, and freedom from relapse. Each was analyzed in the ITT population using a log-rank framework, with hazard ratios obtained using the Pike estimator.
Overall Survival
Hazard ratio for overall survival
95% CI: 0.62–1.01 · P = 0.063
Approximately 10-year time frame · Two-sided CI
| Feature | Overall Survival |
|---|---|
| Analysis population | Intent-to-Treat (ITT) Population |
| Comparison | Dabrafenib and Trametinib Combination Therapy vs Dabrafenib and Trametinib Placebos |
| Method | Log Rank |
| Effect measure | Hazard Ratio, log |
| Estimate | 0.80 |
| 95% CI | 0.62–1.01 |
| P-value | 0.063 |
| Time frame | Approximately 10 years |
| Adjustment | Stratified analysis for disease stage and BRAF mutation status |
An HR of 0.80 corresponds to an estimated 20% lower hazard of death with the combination therapy relative to placebo under the reported analysis. The estimate is therefore below 1, but the confidence interval extends from 0.62 to 1.01.
The confidence interval is important because it includes 1.00. The reported P = 0.063 likewise does not provide conventional evidence against a null hazard ratio of 1 at a two-sided 0.05 threshold. The p-value should not be interpreted as a 6.3% probability that the treatment effect is real, nor does it quantify the size of the observed effect.
The result should be read as an estimated relative difference in the hazard of death with uncertainty, not as proof that there is no treatment effect. The interval includes both a potentially meaningful reduction in hazard and values close to no difference under the reported model.
The analysis was stratified for disease stage and BRAF mutation status, which means those randomized strata form part of the statistical comparison rather than being ignored in the reported hazard-ratio analysis.
Distant Metastasis-Free Survival
Hazard ratio for distant metastasis-free survival
95% CI: 0.40–0.65 · P < .001
Approximately 4-year time frame · Two-sided CI
| Feature | Distant Metastasis-Free Survival |
|---|---|
| Analysis population | Intent-to-Treat (ITT) Population |
| Comparison | Dabrafenib and Trametinib Combination Therapy vs Dabrafenib and Trametinib Placebos |
| Method | Log Rank |
| Effect measure | Hazard Ratio, log |
| Estimate | 0.51 |
| 95% CI | 0.40–0.65 |
| P-value | < .001 |
| Time frame | Approximately 4 years |
| Adjustment | Stratified analysis for disease stage and BRAF mutation status |
An HR of 0.51 corresponds to an approximately 49% lower estimated hazard of a distant metastasis-free survival event with combination therapy compared with placebo under the reported analysis.
The 95% CI of 0.40–0.65 provides the uncertainty range around that relative estimate. Because the interval remains below 1, the estimate is consistently in the direction of a lower event hazard across the interval.
The P < .001 result is evidence against a null hazard ratio of 1 under the stated test. It is not an effect-size measure. The HR and confidence interval are the appropriate quantities for describing magnitude and precision.
As with the primary RFS analysis, this is a time-to-event result. The hazard ratio should not be translated directly into a fixed percentage of participants who will or will not experience distant metastasis.
Freedom From Relapse
Hazard ratio for freedom from relapse
95% CI: 0.39–0.57 · P < .001
Approximately 4-year time frame · Two-sided CI
| Feature | Freedom From Relapse |
|---|---|
| Analysis population | Intent-to-Treat (ITT) Population |
| Comparison | Dabrafenib and Trametinib Combination Therapy vs Dabrafenib and Trametinib Placebos |
| Method | Log Rank |
| Effect measure | Hazard Ratio |
| Estimate | 0.47 |
| 95% CI | 0.39–0.57 |
| P-value | < .001 |
| Time frame | Approximately 4 years |
| Adjustment | Stratified analysis for disease stage and BRAF mutation status |
An HR of 0.47 corresponds to an approximately 53% lower estimated hazard of a freedom-from-relapse event with combination therapy relative to placebo under the reported statistical framework.
The 95% CI of 0.39–0.57 indicates the precision of the estimated relative effect. It is not a prediction interval for individual patients and does not mean that individual treatment effects must fall inside that range.
The P < .001 result addresses the null hypothesis of a hazard ratio of 1. It should not be interpreted as measuring how large or clinically important the treatment effect is.
The close similarity between the reported RFS HR of 0.47 and FFR HR of 0.47 should be treated as a description of the registry-reported estimates, not as evidence that the endpoints are statistically interchangeable. Endpoint definitions and censoring rules determine what each analysis actually measures.
9. Comparing the Reported Time-to-Event Effects
| Endpoint | Time frame | HR | 95% CI | P-value | Analysis |
|---|---|---|---|---|---|
| Relapse-free Survival (primary) | Approximately 4 years | 0.47 | 0.39–0.58 | < 0.0001 | Log-rank; stratified Pike estimator |
| Overall Survival | Approximately 10 years | 0.80 | 0.62–1.01 | 0.063 | Stratified log-rank; Pike estimator |
| Distant Metastasis-Free Survival | Approximately 4 years | 0.51 | 0.40–0.65 | < .001 | Stratified log-rank; Pike estimator |
| Freedom From Relapse | Approximately 4 years | 0.47 | 0.39–0.57 | < .001 | Stratified log-rank; Pike estimator |
The table illustrates why a trial should not be reduced to one p-value. The endpoints address different clinical events and have different registered time frames. The primary RFS analysis reports an HR of 0.47, while the secondary OS analysis reports an HR of 0.80 with a 95% CI extending to 1.01. These are not contradictory simply because their estimates differ: relapse-related events and death are different endpoints, and the ClinicalTrials.gov record does not provide enough additional information to attribute the difference to a particular mechanism.
10. Safety: Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by treatment arm as the number affected among the number at risk.
| Study arm | Serious adverse events | At risk |
|---|---|---|
| Dabrafenib + Trametinib Combination Therapy | 155 | 435 |
| Dabrafenib and Trametinib Placebos | 44 | 432 |
The denominators differ slightly between the two reported groups: 435 participants were at risk in the combination-therapy group and 432 in the placebo group. Because the ClinicalTrials.gov record provides affected and at-risk counts rather than a formal comparative safety analysis, these figures should be treated as descriptive safety results rather than as a reported hypothesis test.
11. Statistical Methods Explained
Why was a log-rank test used?
RFS, OS, DMFS, and FFR are time-to-event outcomes. A simple comparison of proportions at one arbitrary time point would discard information about when events occurred and how long participants were followed. The log-rank test instead compares the observed event experience between groups across the follow-up period while accommodating censored observations.
What does an RFS hazard ratio of 0.47 mean?
Under the reported analysis, the estimated hazard of an RFS event with dabrafenib plus trametinib was approximately 47% of the estimated hazard with placebo. Equivalently, this corresponds to an approximately 53% lower estimated hazard. It does not mean that 53% of participants avoided relapse or that the absolute probability of relapse was reduced by 53 percentage points.
Why is the confidence interval important?
The point estimate is only one part of a statistical result. The 95% CI of 0.39–0.58 for the primary RFS analysis describes the uncertainty surrounding the estimated hazard ratio. A narrower interval would indicate greater precision; a wider interval would indicate more uncertainty. The confidence interval is not a range in which the true treatment effect for every individual participant must lie.
What does the p-value tell us?
The primary P < 0.0001 result evaluates evidence against the specified null hypothesis of a hazard ratio equal to 1. A small p-value does not tell us that the treatment effect is large, nor does it give the probability that the null hypothesis is true. The magnitude of the observed effect is communicated by the hazard ratio, while the confidence interval communicates precision.
Why was an ITT population used?
The posted analyses use the Intent-to-Treat population. Keeping participants in their randomized groups preserves the comparison created by randomization. This is especially important when interpreting an efficacy analysis because treatment discontinuation, later treatment exposure, or other events after randomization can otherwise change the composition of the groups being compared.
What does stratified analysis add?
The posted secondary analyses state that the hazard ratio and p-value were adjusted for the randomized strata of disease stage and BRAF mutation status. Stratification allows the time-to-event comparison to account for those randomized strata while retaining the overall randomized comparison.
Why does censoring matter?
Time-to-event methods distinguish an observed event from the point at which follow-up ends without an event. In the RFS definition posted on ClinicalTrials.gov for COMBI-AD, participants without an event at the 30-Jun-2017 analysis cut-off were censored at the date of their last relevant assessment. Death without prior documented recurrence, however, was explicitly counted as an RFS event rather than censored.
12. Understanding the Primary RFS Result
The primary RFS HR of 0.47 is the central relative-effect estimate in the registry-reported statistical record. A value below 1 favors the combination-therapy group under the analysis convention used by the registry.
The 95% CI of 0.39–0.58 shows the uncertainty around that estimate. The interval remains below 1, so every value within the reported interval corresponds to a lower estimated hazard than the null value of 1.
The reported P < 0.0001 indicates strong evidence against the null hypothesis specified for the primary analysis. It should not be confused with the magnitude of effect or with the probability that the treatment is beneficial for an individual participant.
The ClinicalTrials.gov record does not include median RFS, Kaplan-Meier survival probabilities, event counts for the primary comparison, or subgroup-specific RFS estimates. Those quantities are therefore not presented here.
13. Interpreting the Overall Survival Analysis
The OS result illustrates why statistical interpretation should consider both the point estimate and its uncertainty. The reported HR is 0.80, with a 95% CI of 0.62–1.01 and P = 0.063.
Point estimate
HR 0.80 corresponds to an approximately 20% lower estimated hazard of death with combination therapy under the reported model.
Confidence interval
The 95% CI extends from 0.62 to 1.01, so it includes the null hazard ratio of 1.
P-value
P = 0.063 evaluates evidence against the null hypothesis; it is not a probability that the treatment effect is absent.
Stratification
The reported analysis adjusts for disease stage and BRAF mutation status through stratified analysis.
The OS result should therefore not be reduced to either "the treatment has no effect" or "the treatment reduces mortality by 20%." The statistically precise statement is that the reported point estimate is 0.80, with the registry-reported confidence interval and p-value providing the corresponding uncertainty and evidence against the null.
14. Primary vs Secondary Time-to-Event Endpoints
| Role | Endpoint | What it measures | Reported estimate |
|---|---|---|---|
| Primary | Relapse-free Survival | Time from randomization to disease recurrence or death from any cause | HR 0.47 |
| Primary | Percentage of Participants With RFS Events | Participants experiencing the registered RFS event definition | No separate formal estimate reported in the posted statistical analyses |
| Secondary | Overall Survival | Time-to-event endpoint for overall survival | HR 0.80 |
| Secondary | Distant Metastasis-Free Survival | Time-to-event endpoint concerning distant metastasis-free survival | HR 0.51 |
| Secondary | Freedom From Relapse | Time-to-event endpoint concerning freedom from relapse | HR 0.47 |
These endpoints should not be treated as duplicate measurements. RFS, DMFS, FFR, and OS represent different event definitions and therefore answer different questions about the clinical course. The fact that several hazard ratios are below 1 indicates consistency in the direction of the reported estimates, but it does not make the endpoints statistically identical.
15. Multiplicity and Hypothesis Testing
The ClinicalTrials.gov record identifies two registered primary endpoints and four posted statistical analyses, with one primary-endpoint analysis having a reported estimate, confidence interval, and p-value. The primary analyses use a superiority framework.
| Feature | What the ClinicalTrials.gov record establishes |
|---|---|
| Registered primary endpoints | 2 |
| Primary endpoint type | Time-to-event |
| Primary formal analysis reported | Relapse-free Survival |
| Primary hypothesis type | Superiority |
| CI direction | Two-sided |
| Statistical analyses posted | 4 |
| Primary analyses with estimate + CI | 1 |
The ClinicalTrials.gov record does not specify an alpha-allocation scheme, hierarchical testing procedure, interim-analysis boundary, or multiplicity-adjustment procedure. Those design features therefore should not be inferred from the reported p-values.
16. What the Hazard Ratio Does — and Does Not — Mean
An HR of 0.47 means the estimated event hazard in the combination-therapy group is 47% of that in the placebo group under the reported analysis. This is equivalent to an approximately 53% lower estimated hazard.
The HR does not state how many participants experience an event, how many are spared an event, or the difference in cumulative event probabilities at a particular time point.
A population-level hazard ratio does not imply that every participant experiences the same proportional change in event risk. Individual outcomes can differ substantially from the population-level estimate.
The hazard ratio is a time-to-event summary derived from the statistical analysis. Its interpretation depends on the event definition, censoring rules, follow-up, and assumptions underlying the analysis. A single HR does not reproduce the complete event-time distributions.
17. Important Limitations and Interpretation Issues
- Endpoint-specific interpretation: RFS, OS, DMFS, and FFR describe different time-to-event outcomes. Their hazard ratios should not be treated as interchangeable.
- Different time frames: the primary RFS and secondary DMFS and FFR analyses use an approximately 4-year time frame, whereas the registry-reported OS analysis uses an approximately 10-year time frame.
- Missing absolute survival estimates: the ClinicalTrials.gov record does not provide median survival, Kaplan-Meier survival probabilities, or cumulative event probabilities for the reported analyses.
- Limited primary endpoint reporting: the registry lists a percentage-of-participants RFS-event endpoint as primary, but the statistical analyses posted on ClinicalTrials.gov do not provide a separate estimate, confidence interval, or p-value for it.
- Hazard-ratio assumptions: a single hazard ratio is a model-based summary of a time-to-event comparison. It should not automatically be interpreted as a constant relative risk at every point in time.
- Subgroup information: the ClinicalTrials.gov record identifies stratification by disease stage and BRAF mutation status but do not provide subgroup-specific treatment-effect estimates.
- Multiplicity: the ClinicalTrials.gov record does not specify an alpha-spending or multiplicity-adjustment strategy, so none should be inferred.
- Safety comparison: serious adverse-event counts are reported by arm, but the ClinicalTrials.gov record does not provide a formal comparative confidence interval or p-value for safety.
- Registry-level evidence: this page is restricted to the ClinicalTrials.gov record and linked publication records rather than reconstructing additional results from outside sources.
18. Why This Trial Matters Statistically
COMBI-AD is a useful teaching example because the ClinicalTrials.gov record combines randomized treatment assignment, double masking, an ITT efficacy population, time-to-event endpoints, log-rank testing, hazard ratios, confidence intervals, and stratified analysis.
| Statistical concept | How it appears in COMBI-AD |
|---|---|
| Randomization | Participants were randomized between two parallel study arms. |
| Blinding | The trial was double masked. |
| ITT analysis | The primary and secondary statistical analyses use the ITT population. |
| Time-to-event analysis | RFS, OS, DMFS, and FFR were analyzed as time-to-event outcomes. |
| Log-rank test | The reported statistical analyses use the log-rank method. |
| Hazard ratio | The treatment effect is reported as a hazard ratio. |
| Confidence interval | Each registry-reported formal effect estimate has a two-sided 95% confidence interval. |
| Stratified analysis | Secondary analyses were adjusted for disease stage and BRAF mutation status. |
| Pike estimator | The registry states that the hazard ratio was obtained from the stratified Pike estimator. |
| Censoring | The RFS definition explicitly distinguishes events from censoring at the analysis cut-off. |
| Superiority testing | The posted hypothesis type is superiority. |
19. Statistical Methods Explained in Context
Why use a hazard ratio?
Because the endpoints are time-to-event outcomes, the analysis needs a measure that incorporates both whether an event occurred and when it occurred. The hazard ratio provides a relative comparison of event hazards between randomized groups.
Why report a confidence interval?
The point estimate alone does not communicate precision. The 95% CI shows how uncertain the estimated treatment effect is under the statistical framework.
Why use an ITT analysis?
The ITT population maintains the randomized comparison. It prevents post-randomization treatment experience from redefining the principal efficacy groups.
Why stratify the analysis?
The registry-reported secondary analyses adjust for disease stage and BRAF mutation status, the randomized strata identified in the registry record.
20. A Statistical Reading of the Entire Evidence Set
Viewed together, the ClinicalTrials.gov record shows a consistent direction across the reported relapse-related endpoints. RFS has an HR of 0.47, DMFS has an HR of 0.51, and FFR has an HR of 0.47. Each corresponding 95% confidence interval lies below 1, and each registry-reported p-value is less than 0.001 or, for the primary RFS analysis, less than 0.0001.
The OS estimate is different: HR 0.80 with a 95% CI of 0.62–1.01 and P = 0.063. This illustrates an important statistical distinction. A trial can have several event-related endpoints whose estimates are below 1 without those estimates being identical, because each endpoint defines the event differently and may have a different time frame and event process.
The appropriate interpretation is therefore not to collapse all four hazard ratios into one number. Instead, each endpoint should be interpreted using its own definition, analysis population, time frame, estimate, confidence interval, and p-value.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: COMBI-AD — NCT01682083.
- Linked publication record: PubMed PMID 40250457.
- Linked publication record: PubMed PMID 38899716.
- Linked publication record: PubMed PMID 36653848.
- Linked publication record: PubMed PMID 34225229.
- Linked publication record: PubMed PMID 32877599.
Continue learning through Clinical Biostats
Explore the statistical concepts behind randomized trials, survival analysis, confidence intervals, hazard ratios, and related clinical-trial methods.
24. Record Summary
COMBI-AD provides a clear example of randomized clinical-trial survival analysis. The trial was a phase 3, randomized, parallel, double-blind study with 870 participants and two treatment arms. Its primary RFS analysis used a log-rank test in the ITT population and reported a hazard ratio of 0.47 with a 95% CI of 0.39–0.58 and P < 0.0001.
The registry-reported secondary analyses extend the statistical picture to overall survival, distant metastasis-free survival, and freedom from relapse. Their reported hazard ratios were 0.80, 0.51, and 0.47, respectively, with the OS confidence interval extending to 1.01 and the DMFS and FFR confidence intervals remaining below 1. The analyses used stratified methods incorporating disease stage and BRAF mutation status.
The most useful statistical reading therefore combines the hazard ratio, confidence interval, p-value, endpoint definition, analysis population, time frame, and stratification approach. A p-value alone cannot describe treatment-effect magnitude, and a hazard ratio alone cannot describe the complete time-to-event experience.