This page separates reported trial results from statistical interpretation. Trial-specific numbers and reported analyses are restricted to the ClinicalTrials.gov data posted on ClinicalTrials.gov for CheckMate-026. General explanations of statistical methods are educational descriptions of those methods.
1. Trial at a Glance
CheckMate-026 was an open-label, randomized, parallel, phase 3 trial evaluating nivolumab versus investigator's choice chemotherapy as first-line therapy for participants with stage IV or recurrent non-small cell lung cancer. The registry reports 541 enrolled participants, two treatment arms, one registered primary endpoint, six posted outcome measures, and five posted statistical analyses.
| Feature | CheckMate-026 |
|---|---|
| Trial name | CheckMate-026 |
| Phase | Phase 3 |
| Status | COMPLETED |
| Population | Stage IV or recurrent non-small cell lung cancer |
| Design | Randomized, parallel, open-label |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 541 |
| Lead sponsor | Bristol-Myers Squibb |
| Start | March 27, 2014 |
| Primary completion | July 1, 2016 |
| ClinicalTrials.gov | NCT02041533 |
2. Clinical Question
The central statistical question was whether first-line nivolumab differed from investigator's choice chemotherapy with respect to progression-free survival in participants with PD-L1 expression ≥ 5%. The registered primary hypothesis type was superiority.
Population
Participants with stage IV or recurrent non-small cell lung cancer, with the primary efficacy analysis restricted to all randomized participants with PD-L1 expression ≥ 5%.
Intervention
Nivolumab. The registry also identifies nivolumab as a biological intervention.
Comparator
Investigator's choice of chemotherapy, with gemcitabine, cisplatin, carboplatin, paclitaxel, and pemetrexed listed among the trial interventions.
Primary question
Among randomized participants with PD-L1 expression ≥ 5%, how does nivolumab compare with investigator's choice chemotherapy for progression-free survival?
3. Trial Design
Nivolumab
- Nivolumab was the biological intervention.
- The primary comparison was nivolumab versus investigator's choice chemotherapy.
- Serious adverse events affected 195 of 267 participants at risk in the reported safety data.
Investigator's Choice of Chemotherapy
- Investigator's choice chemotherapy was the comparator.
- Listed chemotherapy interventions included gemcitabine, cisplatin, carboplatin, paclitaxel, and pemetrexed.
- Serious adverse events affected 203 of 263 participants at risk in the reported safety data.
4. Trial Timeline
Trial start
The registry lists March 27, 2014 as the study start date.
Primary completion
The registry lists July 1, 2016 as the primary completion date.
Primary PFS assessment window
The registered primary endpoint was assessed up to August 2016, with the registry time frame described as approximately 28 months.
Reported overall-survival analysis window
The posted secondary overall-survival analyses use a time frame from randomization to death, up to approximately 89 months.
5. Primary Endpoint
| Endpoint | Registry definition / time frame | Primary statistical analysis |
|---|---|---|
| Progression-Free Survival in Participants With PD-L1 Expression ≥ 5% | From date of randomization until date of documented tumor progression (assessed up to August 2016, approximately 28 months)). Progression-Free Survival (PFS) was defined as the time between the date of randomization and the first date of documented tumor progression, as determined by the Independent Radiology Review Committee (IRRC) per Response Evaluation Criteria in Solid Tumors (RECIST) v1.1, or death due to any cause, whichever occurs first. | Stratified log-rank test with stratified Cox proportional-hazards model |
The primary endpoint is a time-to-event endpoint. Its analysis therefore focuses on the timing of progression or death rather than simply classifying each participant as having experienced an event or not.
6. Primary Result: Progression-Free Survival in Participants With PD-L1 Expression ≥ 5%
The registry reports a formal superiority analysis among all randomized participants with PD-L1 expression ≥ 5%, comparing nivolumab with investigator's choice of chemotherapy.
Hazard ratio for progression or death
95% CI: 0.91–1.45 · P = 0.2511
Analysis: stratified log-rank test and stratified Cox proportional-hazards model.
| Primary endpoint | Nivolumab vs chemotherapy |
|---|---|
| Analysis population | All randomized participants with PD-L1 expression ≥ 5% |
| Endpoint | Progression-Free Survival |
| Effect measure | Hazard ratio |
| Hazard ratio | 1.15 |
| 95% CI | 0.91–1.45 |
| P-value | 0.2511 |
| Hypothesis | Superiority |
| Stratification | Histology (squamous vs. non-squamous) as entered into the IVRS |
The reported hazard ratio of 1.15 is the estimated hazard of progression or death for nivolumab relative to investigator's choice chemotherapy under the stratified Cox model. Because the estimate is above 1, the estimated instantaneous event rate is higher for nivolumab in this analysis; equivalently, 1.15 represents an estimated 15% higher hazard relative to chemotherapy under the model.
The hazard ratio does not mean that 15% more participants necessarily progressed or died. A hazard ratio is a relative time-to-event measure, not an absolute event-rate difference, a risk ratio, or a percentage of patients who benefited or failed treatment.
The 95% confidence interval of 0.91–1.45 describes statistical uncertainty around the estimated hazard ratio under the analysis model. It includes 1, so the data are compatible with a range of relative hazard values on both sides of the no-difference value.
The P-value of 0.2511 measures the compatibility of the observed test statistic with the specified null hypothesis under the statistical testing framework. It is not a measure of the magnitude or clinical importance of the hazard ratio, and it does not describe the probability that either treatment is effective.
The analysis was stratified by histology, and the reported model was a stratified Cox proportional-hazards model. As with other Cox-model analyses, interpretation of a single hazard ratio requires attention to the proportional-hazards assumption and to censoring and follow-up patterns.
7. Secondary Efficacy Results
The registry contains four additional time-to-event analyses and one binary response analysis. These analyses use the same broad randomized treatment comparison but differ in endpoint, analysis population, and statistical method.
Progression-Free Survival in All Randomized Participants
Hazard ratio
95% CI: 0.95–1.43
Stratified Cox proportional-hazards model.
The hazard ratio of 1.17 represents the estimated hazard of progression or death for nivolumab relative to investigator's choice chemotherapy among all randomized participants. Under the model, the estimate corresponds to an approximately 17% higher estimated hazard in the nivolumab group relative to chemotherapy.
The 95% CI of 0.95–1.43 spans 1.00, indicating uncertainty around the direction and magnitude of the relative treatment effect. The interval is not a range of individual patient outcomes.
No P-value is posted on ClinicalTrials.gov for this secondary analysis in the ClinicalTrials.gov record. The absence of a posted P-value means that the page does not assign a formal hypothesis-test result beyond the reported hazard ratio and confidence interval.
Overall Survival in Participants With PD-L1 Expression ≥ 5%
Hazard ratio
95% CI: 0.79–1.20
Time frame: from date of randomization to date of death, up to approximately 89 months.
The hazard ratio of 0.97 estimates the relative instantaneous hazard of death for nivolumab versus investigator's choice chemotherapy among randomized participants with PD-L1 expression ≥ 5%. An HR below 1 corresponds to a lower estimated hazard under the fitted model, but an estimate close to 1 indicates little separation on this relative scale.
The 95% CI of 0.79–1.20 includes 1.00. It therefore allows for both a lower and a higher hazard of death for nivolumab relative to chemotherapy within the uncertainty represented by the interval.
No P-value is reported for this analysis in the ClinicalTrials.gov record. The hazard ratio should not be interpreted as a percentage reduction in mortality or as an absolute survival difference.
Overall Survival in All Randomized Participants
Hazard ratio
95% CI: 0.86–1.24
Time frame: from date of randomization to date of death, up to approximately 89 months.
The hazard ratio of 1.03 estimates the hazard of death for nivolumab relative to investigator's choice chemotherapy in all randomized participants. The point estimate is close to 1, meaning the estimated relative hazards are close on this model-based scale.
The 95% CI of 0.86–1.24 includes 1.00. It represents uncertainty around the estimated hazard ratio and should not be read as the range of survival experiences for individual participants.
No P-value is posted for this secondary analysis in the ClinicalTrials.gov record. A hazard ratio by itself does not establish a clinically meaningful absolute difference in survival.
Objective Response Rate in Participants With PD-L1 Expression ≥ 5%
Odds ratio for objective response
95% CI: 0.46–1.06 · Two-sided
Analysis: Cochran-Mantel-Haenszel test, stratified by histology.
| Secondary endpoint | Effect measure | Estimate | 95% CI | P-value | Method |
|---|---|---|---|---|---|
| Progression-Free Survival in All Randomized Participants | Hazard ratio | 1.17 | 0.95–1.43 | Not posted | Stratified Cox proportional-hazards model |
| Overall Survival in Participants With PD-L1 Expression ≥ 5% | Hazard ratio | 0.97 | 0.79–1.20 | Not posted | Stratified Cox proportional-hazards model |
| Overall Survival in All Randomized Participants | Hazard ratio | 1.03 | 0.86–1.24 | Not posted | Stratified Cox proportional-hazards model |
| Objective Response Rate in Participants With PD-L1 Expression ≥ 5% | Odds ratio | 0.70 | 0.46–1.06 | Not posted | Cochran-Mantel-Haenszel test |
The odds ratio of 0.70 compares the odds of objective response for nivolumab with the odds of objective response for investigator's choice chemotherapy, using the reported Mantel-Haenszel stratified analysis. An odds ratio below 1 does not directly mean that the probability of response was 30% lower; odds and probabilities are different quantities.
The 95% CI of 0.46–1.06 spans 1.00, so the interval includes both lower and higher odds of response for nivolumab relative to chemotherapy.
The P-value is not posted on ClinicalTrials.gov for this posted secondary analysis. The reported two-sided confidence interval should therefore be presented as the available measure of statistical uncertainty rather than supplemented with an inferred P-value.
8. Statistical Results Summary
| Endpoint | Population | Effect | 95% CI | P-value | Method |
|---|---|---|---|---|---|
| Primary PFS, PD-L1 ≥ 5% | All randomized participants with PD-L1 expression ≥ 5% | HR 1.15 | 0.91–1.45 | 0.2511 | Stratified log-rank; stratified Cox model |
| PFS, all randomized | All randomized participants | HR 1.17 | 0.95–1.43 | Not posted | Stratified Cox model |
| OS, PD-L1 ≥ 5% | All randomized participants with PD-L1 expression ≥ 5% | HR 0.97 | 0.79–1.20 | Not posted | Stratified Cox model |
| OS, all randomized | All randomized participants | HR 1.03 | 0.86–1.24 | Not posted | Stratified Cox model |
| ORR, PD-L1 ≥ 5% | All randomized participants with PD-L1 expression ≥ 5% | OR 0.70 | 0.46–1.06 | Not posted | Cochran-Mantel-Haenszel test |
This table illustrates an important statistical distinction: the primary endpoint has a formally posted hypothesis test with P = 0.2511, whereas the ClinicalTrials.gov record for the four secondary analyses provide effect estimates and confidence intervals but no corresponding P-values.
9. Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by treatment group as affected participants divided by participants at risk. These figures should be kept separate from efficacy analyses because safety populations and treatment exposure can be defined differently from randomized efficacy populations.
| Reported group | Participants affected | Participants at risk | Reported measure |
|---|---|---|---|
| Nivolumab | 195 | 267 | 195/267 |
| Investigator Choice of Chemotherapy | 203 | 263 | 203/263 |
| Post Chemotherapy Optional Nivolumab | 110 | 159 | 110/159 |
10. Statistical Methodology
Kaplan-Meier estimation
Kaplan-Meier estimation is the standard descriptive framework for a time-to-event endpoint such as progression-free survival or overall survival. It estimates the probability of remaining event-free over time while allowing participants to be right-censored when their event status is not observed through the end of available follow-up.
Here, di is the number of events at time ti, while ni is the number at risk immediately before that time.
The ClinicalTrials.gov record does not provide the participant-level event and censoring times needed to reproduce the Kaplan-Meier curve. The method is therefore useful for understanding the registered endpoint, but a reconstructed curve should not be presented from the summary statistics alone.
Stratified log-rank test
The primary PFS analysis used a log-rank test stratified by histology, specifically squamous versus non-squamous, as entered into the IVRS. Stratification allows the time-to-event comparison to account for the prespecified stratification factor rather than treating all participants as belonging to one unstratified risk set.
Cox proportional-hazards model
The primary analysis also used a stratified Cox proportional-hazards model. The model expresses the treatment comparison through a hazard ratio, with the reported hazard ratio defined as nivolumab relative to investigator's choice chemotherapy.
HR = 1 corresponds to equal estimated hazards under the model. HR > 1 corresponds to a higher estimated instantaneous event rate for nivolumab relative to the comparator.
Cochran-Mantel-Haenszel analysis
The objective response rate analysis used the Cochran-Mantel-Haenszel test, with the analysis stratified by histology. The resulting effect measure was a strata-adjusted odds ratio for nivolumab over investigator's choice of chemotherapy.
Odds ratio
For a binary endpoint, an odds ratio compares the odds of an event between two groups. It is not identical to a risk ratio. The distinction becomes particularly important when the outcome is not rare, because an odds ratio can differ substantially from the corresponding ratio of probabilities.
Confidence intervals
A two-sided 95% confidence interval describes uncertainty around an estimated effect under the statistical model and sampling framework. For ratios such as hazard ratios and odds ratios, the no-difference value is 1.00. An interval that crosses 1.00 therefore includes both directions of relative effect.
11. Statistical Methods Explained
Why was a stratified log-rank test used?
The registered primary analysis used a log-rank test stratified by histology. A stratified log-rank procedure compares time-to-event experience between randomized groups while accounting for the specified strata. In this trial, the reported strata were squamous versus non-squamous histology as entered into the IVRS.
What does a hazard ratio of 1.15 mean?
A hazard ratio of 1.15 means that the fitted model estimates the instantaneous hazard of progression or death to be 1.15 times as high for nivolumab as for investigator's choice chemotherapy. The simple interpretation of the point estimate is an approximately 15% higher estimated hazard, but this is not equivalent to saying that 15% more participants experienced progression or death.
Why is the confidence interval important?
The point estimate is only one summary of the treatment comparison. The 95% CI of 0.91–1.45 for the primary PFS analysis shows the uncertainty around that estimate. Because the interval extends below and above 1.00, the ClinicalTrials.gov record does not identify a narrow range entirely on one side of the no-difference value.
Why does the P-value not measure effect size?
The primary P-value of 0.2511 is a hypothesis-testing quantity. It describes how compatible the observed test statistic is with the specified null hypothesis under the test's assumptions. It does not tell us how large the hazard ratio is, how clinically important the effect is, or the probability that one treatment is superior.
Why is the odds ratio different from a risk ratio?
The OR of 0.70 for objective response compares odds rather than probabilities. Odds are defined as the probability of an event divided by the probability of no event. Consequently, an OR of 0.70 cannot be translated directly into a 30% reduction in the probability of response without additional information.
Why does stratification matter?
Stratification allows the analysis to account for a factor used to structure the randomized comparison. In the primary PFS analysis, histology was the reported stratification factor. In the ORR analysis, the Cochran-Mantel-Haenszel approach likewise adjusted for histology.
12. Primary Analysis Population and Stratification
| Feature | Registry-supported detail |
|---|---|
| Primary analysis population | All randomized participants with PD-L1 expression ≥ 5% |
| Primary endpoint | Progression-Free Survival |
| Primary comparison | Nivolumab vs Investigator Choice of Chemotherapy |
| Primary test | Log Rank |
| Primary model | Stratified Cox proportional hazard model |
| Primary stratification | Histology (squamous vs. non-squamous) as entered into the IVRS |
| Hypothesis type | Superiority |
The analysis population is important because the primary result does not describe all 541 enrolled participants without qualification. It specifically refers to all randomized participants with PD-L1 expression ≥ 5%. The secondary all-randomized analyses address a broader randomized population.
13. Understanding the Primary Endpoint
Progression-free survival combines two possible events into a single time-to-event endpoint: documented tumor progression or death from any cause, whichever occurs first. This definition makes PFS different from a simple tumor-response proportion because it incorporates when the event occurs.
What is measured?
Time from randomization to the first qualifying progression event or death, according to the registered definition.
Why time matters
Two participants can both experience progression while having different progression-free times. Time-to-event methods retain this temporal information.
What is censoring?
Censoring occurs when a participant's event time is not observed within the relevant follow-up period. Survival methods use the available follow-up without treating every censored participant as an event.
What does HR summarize?
The hazard ratio summarizes the relative instantaneous event rate under the fitted Cox model rather than giving an absolute difference in event probability.
14. Interpreting the Primary PFS Result
The primary hazard ratio was 1.15. Because the comparison was defined as nivolumab to investigator's choice chemotherapy, the point estimate is above 1. Under the model, this corresponds to a higher estimated instantaneous hazard of progression or death for nivolumab.
The 95% CI was 0.91–1.45. This interval is considerably more informative than the point estimate alone because it displays the statistical uncertainty surrounding the estimated relative hazard.
The reported two-sided P-value was 0.2511. The registered hypothesis type was superiority. The P-value is evidence about the specified hypothesis test, not a direct measure of treatment effect size or patient-level benefit.
The Cox hazard ratio is a model-based summary. It should not automatically be translated into an absolute probability difference, median survival difference, or percentage of patients who benefited.
15. Secondary Time-to-Event Analyses
The secondary analyses show why the analysis population and endpoint should always accompany a hazard ratio.
| Endpoint | Population | HR | 95% CI | Interpretive point |
|---|---|---|---|---|
| PFS | All randomized participants | 1.17 | 0.95–1.43 | The broader population produces an estimate above 1, with the confidence interval crossing 1. |
| OS | PD-L1 expression ≥ 5% | 0.97 | 0.79–1.20 | The point estimate is close to 1 and the interval crosses 1. |
| OS | All randomized participants | 1.03 | 0.86–1.24 | The point estimate is close to 1 and the interval crosses 1. |
These estimates should not be collapsed into a single overall treatment effect. Each corresponds to a different endpoint or analysis population. In particular, the two OS analyses differ in whether the population is restricted to participants with PD-L1 expression ≥ 5% or includes all randomized participants.
16. Objective Response Rate and the Cochran-Mantel-Haenszel Method
Objective Response Rate in participants with PD-L1 expression ≥ 5% was analyzed as a binary endpoint using the Cochran-Mantel-Haenszel test. The analysis was stratified by histology, and the reported effect measure was a strata-adjusted odds ratio.
The important idea is that the overall treatment comparison is constructed while accounting for the specified stratification variable rather than simply pooling all observations without regard to strata.
The reported OR of 0.70 has a 95% CI of 0.46–1.06. The interval includes 1.00, so the ClinicalTrials.gov record does not support a narrow one-direction interpretation of the treatment effect based on the interval alone.
17. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary PFS analysis estimated an HR of 1.15 with a two-sided 95% CI of 0.91–1.45 and P = 0.2511. The analysis used a stratified log-rank test and stratified Cox model.
Clinical interpretation
The ClinicalTrials.gov record alone do not provide enough information to quantify median PFS, absolute survival probabilities, response percentages, or patient-level benefit. Clinical interpretation therefore should not be expanded beyond the reported statistical estimates.
This distinction is important. Statistical analysis describes the evidence produced by the prespecified comparison. Clinical interpretation requires additional context, including absolute event probabilities, treatment exposure, safety, patient characteristics, and the clinical consequences of the measured endpoint.
18. Important Limitations and Interpretation Issues
- Primary endpoint population: the primary PFS analysis is restricted to all randomized participants with PD-L1 expression ≥ 5%, so the primary result should not be presented as if it automatically described every enrolled participant.
- Limited summary information: the ClinicalTrials.gov record contains effect estimates and confidence intervals but not the underlying participant-level event and censoring data.
- No absolute PFS estimates: median PFS and time-specific PFS probabilities are not included in the ClinicalTrials.gov record.
- No baseline table: the ClinicalTrials.gov record does not provide the baseline demographic or disease-characteristic distribution by randomized arm.
- Hazard-ratio assumptions: Cox-model hazard ratios are model-based summaries and should not automatically be interpreted as constant relative risks at every point in time.
- Confidence intervals: the intervals should be interpreted as measures of statistical uncertainty around the estimated effect, not as ranges containing individual patient outcomes.
- Secondary P-values: the ClinicalTrials.gov record does not report P-values for the four secondary statistical analyses shown on this page.
- Safety population: serious adverse-event data are reported using affected/at-risk counts and include a separate Post Chemotherapy Optional Nivolumab group; these should not be treated as equivalent to the randomized efficacy populations.
- Missing-data methodology: the ClinicalTrials.gov record does not describe a specific missing-data or imputation procedure, so no such procedure is attributed to the trial here.
- Other design features: the ClinicalTrials.gov record does not provide information supporting conclusions about crossover, factorial design, non-inferiority margins, Bayesian methods, or interim-analysis rules.
19. Why This Trial Matters Statistically
CheckMate-026 is a useful statistical teaching case because the registry results connect several core methods in clinical-trial analysis: randomized comparison, time-to-event endpoints, stratification, log-rank testing, Cox proportional-hazards modeling, hazard ratios, confidence intervals, and categorical-data analysis using the Cochran-Mantel-Haenszel framework.
| Concept | How it appears in CheckMate-026 |
|---|---|
| Randomization | The trial uses randomized allocation in a parallel phase 3 design. |
| Time-to-event endpoint | The registered primary endpoint is progression-free survival. |
| Stratified analysis | The primary PFS analysis was stratified by histology. |
| Log-rank test | The primary PFS comparison used a stratified log-rank test. |
| Cox model | A stratified Cox proportional-hazards model was used for the primary PFS analysis and reported secondary time-to-event analyses. |
| Hazard ratio | PFS and OS results are expressed as hazard ratios comparing nivolumab with investigator's choice chemotherapy. |
| Confidence interval | 95% two-sided confidence intervals accompany the reported effect estimates. |
| Categorical analysis | Objective response rate was analyzed using a Cochran-Mantel-Haenszel approach. |
| Odds ratio | The ORR analysis reports a strata-adjusted odds ratio. |
| Analysis populations | The primary PFS population is PD-L1 expression ≥ 5%; other analyses use all randomized participants. |
20. What the Hazard Ratio Does — and Does Not — Mean
The primary PFS hazard ratio of 1.15 means that, under the stratified Cox model and with nivolumab as the numerator of the comparison, the estimated instantaneous hazard of progression or death was 1.15 times that associated with investigator's choice chemotherapy.
It does not mean that 15% more participants progressed, that 15% more participants died, or that every individual patient had a 15% higher probability of progression or death.
The 95% CI of 0.91–1.45 expresses uncertainty around the primary hazard-ratio estimate. It includes the no-difference value of 1.00, so the point estimate should not be interpreted independently of the interval.
The P-value of 0.2511 addresses the statistical hypothesis test associated with the primary analysis. It does not measure effect magnitude. Effect magnitude is conveyed by the hazard ratio, while precision is conveyed by the confidence interval.
21. Results by Statistical Question
| Question | Reported result | How to read it |
|---|---|---|
| Does nivolumab have a lower estimated PFS hazard in the PD-L1 ≥ 5% primary population? | HR 1.15; 95% CI 0.91–1.45; P = 0.2511 | The point estimate is above 1, while the confidence interval includes 1. |
| What is the PFS hazard ratio in all randomized participants? | HR 1.17; 95% CI 0.95–1.43 | The point estimate is above 1; the interval includes 1. |
| What is the OS hazard ratio in the PD-L1 ≥ 5% population? | HR 0.97; 95% CI 0.79–1.20 | The point estimate is close to 1; the interval includes 1. |
| What is the OS hazard ratio in all randomized participants? | HR 1.03; 95% CI 0.86–1.24 | The point estimate is close to 1; the interval includes 1. |
| What is the ORR odds ratio in the PD-L1 ≥ 5% population? | OR 0.70; 95% CI 0.46–1.06 | The point estimate is below 1; the interval includes 1. |
22. Data Boundaries for This Analysis
The ClinicalTrials.gov record contains enough information to describe the registered design, primary endpoint, five posted statistical analyses, and reported serious adverse-event counts. They do not contain several types of information commonly included in a full clinical-trial publication.
Not included in the ClinicalTrials.gov record
Median survival estimates, time-specific survival probabilities, detailed baseline characteristics, subgroup hazard ratios, or forest-plot results.
Not reconstructed
No Kaplan-Meier curves, event counts by randomized efficacy arm, or derived response percentages are created from incomplete summary information.
Not inferred
No randomization ratio, treatment-arm sample size, crossover rate, interim boundary, multiplicity adjustment, or missing-data method is inferred when it is absent from the ClinicalTrials.gov record.
Why this matters
Keeping these boundaries explicit prevents secondary information from being presented as though it were part of the registry evidence.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
25. Sources
- ClinicalTrials.gov: NCT02041533 — CheckMate-026. Trial design, registered endpoint definitions, posted statistical analyses, and safety counts summarized on this page are restricted to the ClinicalTrials.gov record.
- Linked publication: PubMed record for PMID 38383458.
- Linked publication: PubMed record for PMID 28636851.
Continue with the statistical methods
Explore the survival-analysis, categorical-data, confidence-interval, and clinical-trial methods connected to the CheckMate-026 analysis.
26. Record Summary
CheckMate-026 provides a compact example of several fundamental clinical-trial statistical methods. The trial was a randomized, parallel, open-label phase 3 study with 541 enrolled participants and a registered primary time-to-event endpoint of progression-free survival in participants with PD-L1 expression ≥ 5%. The primary analysis used a histology-stratified log-rank test and a stratified Cox proportional-hazards model, producing an HR of 1.15 with a two-sided 95% CI of 0.91–1.45 and P = 0.2511.
The posted secondary analyses demonstrate the importance of keeping the endpoint and analysis population attached to every effect estimate. PFS in all randomized participants produced an HR of 1.17, OS produced HRs of 0.97 in the PD-L1 ≥ 5% population and 1.03 in all randomized participants, and objective response rate produced an odds ratio of 0.70 using a histology-stratified Cochran-Mantel-Haenszel analysis. The reported serious adverse-event data likewise require attention to the population being described, particularly because a separate Post Chemotherapy Optional Nivolumab group is reported.