This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are limited to the information contained in the ClinicalTrials.gov record.
1. Trial at a Glance
REACH was a randomized, double-blind, phase 3 parallel trial evaluating ramucirumab plus best supportive care against placebo plus best supportive care as second-line treatment in participants with hepatocellular carcinoma after first-line therapy with sorafenib.
| Feature | REACH |
|---|---|
| Phase | Phase 3 |
| Condition | Hepatocellular carcinoma |
| Clinical question | Ramucirumab plus best supportive care versus placebo plus best supportive care as second-line treatment after first-line therapy with sorafenib |
| Design | Randomized, double-blind, parallel |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 565.0 |
| Primary endpoint | Overall Survival (OS) |
| Primary endpoint type | Time-to-event |
| Statistical methods reported | Log-rank test; Cochran-Mantel-Haenszel test |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
| Trial status | Completed |
| ClinicalTrials.gov | NCT01140347 |
2. Clinical Question
The central statistical question was whether participants randomized to ramucirumab plus best supportive care experienced a different overall survival distribution from participants randomized to placebo plus best supportive care, with superiority as the registered hypothesis type.
Population
Participants with hepatocellular carcinoma who had received first-line therapy with sorafenib.
Intervention
Ramucirumab DP (IMC-1121B) plus best supportive care.
Comparator
Placebo plus best supportive care.
Primary question
Does ramucirumab plus best supportive care improve overall survival relative to placebo plus best supportive care?
3. Trial Design
Ramucirumab + best supportive care
- Ramucirumab DP (IMC-1121B)
- Best supportive care
- Serious adverse events: 124/277 affected participants
Placebo + best supportive care
- Placebo
- Best supportive care
- Serious adverse events: 92/276 affected participants
The registry identifies the design as randomized, parallel, and double-blind. The trial had two arms and a total enrollment of 565.0 participants.
4. Endpoints
| Endpoint | Registry definition / time frame | Type |
|---|---|---|
| Overall Survival (OS) | OS was defined as the time from the date of randomization to the date of death from any cause. Participants who were alive at the end of the follow-up period or were lost to follow-up were censored on the last date the participant was known to be alive. Time frame: Randomization to death from any cause (up to 37 months). | Primary; time-to-event |
| Progression-Free Survival (PFS) | Randomization to PD (up to 36 months). | Secondary; time-to-event |
| Time to Radiographic Progression (TTP) | Randomization to PD (up to 36 months). | Secondary; time-to-event |
| Objective Response Rate (ORR) | Percentage of participants with complete response (CR) or partial response (PR), from baseline to the date of first evidence of confirmed CR or PR (up to 37 months). | Secondary; binary |
The registry data report 1 registered primary endpoint and 11 posted outcome measures. Four statistical analyses are posted in the ClinicalTrials.gov record: one primary-endpoint analysis and three secondary time-to-event/categorical analyses.
5. Statistical Analysis Plan in the Registry Data
The ClinicalTrials.gov record identifies two main statistical-method families: log-rank testing for time-to-event endpoints and the Cochran-Mantel-Haenszel test for the binary objective response endpoint. The time-to-event hazard ratios were estimated using stratified Cox proportional hazards regression.
| Endpoint | Analysis method | Effect measure | Analysis population |
|---|---|---|---|
| Overall Survival | Log-rank test; stratified Cox proportional hazards regression | Hazard ratio | ITT: all randomized participants |
| Progression-Free Survival | Log-rank test; stratified Cox proportional hazards regression | Hazard ratio | ITT: all randomized participants |
| Time to Radiographic Progression | Log-rank test; stratified Cox proportional hazards regression | Hazard ratio | ITT: all randomized participants |
| Objective Response Rate | Cochran-Mantel-Haenszel test | Percentage of participants with CR or PR | ITT: all randomized participants |
The analysis notes specify that the Cox model used the Interactive Web Response System (IWRS) stratification factors, namely geographical regions and etiology of liver disease. The ORR analysis used a Cochran-Mantel-Haenszel adjustment for geographic region and etiology of liver disease.
6. Primary Endpoint: Overall Survival
The registered primary endpoint was overall survival, defined as time from randomization to death from any cause. The analysis used the ITT population, meaning all randomized participants were included according to randomized treatment assignment. The registry reports 65 participants censored in the ramucirumab plus BSC group and 58 in the placebo plus BSC group.
Hazard ratio for overall survival
95% CI: 0.717–1.046 · P = 0.1391
Two-sided confidence interval · Superiority hypothesis
| Primary endpoint | Ramucirumab + BSC vs Placebo + BSC |
|---|---|
| Analysis method | Log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.866 |
| 95% CI | 0.717–1.046 |
| P-value | 0.1391 |
| Analysis population | Intent-to-treat population: all randomized participants |
| Censored participants | Ramucirumab + BSC: 65; Placebo + BSC: 58 |
| Model | Stratified Cox proportional hazards regression |
| Stratification factors | Geographical regions and etiology of liver disease |
The estimated hazard ratio of 0.866 means that the fitted model estimated the instantaneous rate of death in the ramucirumab-plus-BSC group at about 86.6% of the corresponding rate in the placebo-plus-BSC group over the analyzed follow-up. Expressed as a relative model-based quantity, 1 − 0.866 = 0.134, so the estimate corresponds to approximately a 13.4% lower estimated hazard.
That interpretation does not mean that 13.4% of participants benefited, that individual patients had exactly a 13.4% reduction in risk, or that the absolute probability of death was reduced by 13.4 percentage points. A hazard ratio is a relative time-to-event measure.
The 95% confidence interval of 0.717–1.046 describes uncertainty around the estimated hazard ratio under the specified statistical model and sampling framework. Because the interval includes 1, the interval is compatible with both a lower and a higher hazard under the model.
The P-value of 0.1391 is evidence against the null hypothesis within the specified testing framework; it is not a measure of the magnitude or clinical importance of the treatment effect. A P-value also does not give the probability that the treatment has no effect.
Interpretation of a Cox hazard ratio additionally depends on the proportional-hazards framework. If relative hazards change materially over time, one summary hazard ratio may not fully describe the treatment-effect pattern. The ClinicalTrials.gov record does not report a proportional-hazards diagnostic or an alternative time-varying analysis.
7. Secondary Endpoint: Progression-Free Survival
Progression-Free Survival was defined in the registry as time from randomization to PD, with a time frame of up to 36 months. The analysis used the ITT population and a log-rank test. The registry reports 43 censored participants in the ramucirumab-plus-BSC group and 19 in the placebo-plus-BSC group.
Hazard ratio for progression-free survival
95% CI: 0.522–0.750 · P < 0.0001
Two-sided confidence interval · Superiority hypothesis
| Secondary endpoint | Ramucirumab + BSC vs Placebo + BSC |
|---|---|
| Analysis method | Log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.625 |
| 95% CI | 0.522–0.750 |
| P-value | <0.0001 |
| Analysis population | Intent-to-treat population: all randomized participants |
| Censored participants | Ramucirumab + BSC: 43; Placebo + BSC: 19 |
| Model | Stratified Cox proportional hazards regression |
| Stratification factors | Geographical regions and etiology of liver disease |
The estimated PFS hazard ratio of 0.625 corresponds to an estimated instantaneous event rate about 62.5% as large in the ramucirumab-plus-BSC group as in the placebo-plus-BSC group. Equivalently, 1 − 0.625 = 0.375, so the estimated relative hazard is approximately 37.5% lower under the fitted model.
The confidence interval of 0.522–0.750 quantifies uncertainty around that estimate. It remains below 1 throughout the reported interval, but it should not be interpreted as the range of individual patient effects.
The reported P < 0.0001 indicates strong evidence against the null hypothesis within the specified statistical test. It does not quantify the size of the treatment effect, and it does not mean there is a probability of less than 0.0001 that the null hypothesis is true.
Because PFS is a time-to-event endpoint, censoring contributes to the analysis. The number of censored participants differs between the two randomized groups, so the interpretation depends on the censoring framework underlying the analysis. The ClinicalTrials.gov record does not provide a detailed censoring-mechanism assessment.
8. Secondary Endpoint: Time to Radiographic Progression
Time to Radiographic Progression was evaluated from randomization to PD, with a time frame of up to 36 months. The analysis used the ITT population and a log-rank test, with the hazard ratio estimated from a stratified Cox proportional hazards model.
Hazard ratio for time to radiographic progression
95% CI: 0.487–0.722 · P < 0.0001
Two-sided confidence interval · Superiority hypothesis
| Secondary endpoint | Ramucirumab + BSC vs Placebo + BSC |
|---|---|
| Analysis method | Log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.593 |
| 95% CI | 0.487–0.722 |
| P-value | <0.0001 |
| Analysis population | Intent-to-treat population: all randomized participants |
| Censored participants | Ramucirumab + BSC: 86; Placebo + BSC: 52 |
| Model | Stratified Cox proportional hazards regression |
| Stratification factors | Geographical regions and etiology of liver disease |
The TTP hazard ratio of 0.593 corresponds to an estimated instantaneous rate of radiographic progression about 59.3% as large in the ramucirumab-plus-BSC group as in the placebo-plus-BSC group. The corresponding relative reduction in estimated hazard is approximately 40.7%.
The 95% confidence interval, 0.487–0.722, remains below 1. This provides information about the precision and direction of the estimated relative effect, but it does not establish that every participant experiences the same relative reduction.
The reported P < 0.0001 concerns the statistical evidence against the null hypothesis under the specified test. It should not be read as a probability that the treatment effect is real, nor as a measure of effect magnitude.
The TTP result is related to PFS but is not identical conceptually: PFS is the registered secondary endpoint described as randomization to PD, whereas TTP is specifically labeled Time to Radiographic Progression. Both are analyzed as time-to-event outcomes in the ClinicalTrials.gov record.
9. Secondary Endpoint: Objective Response Rate
Objective Response Rate was defined as the percentage of participants with complete response or partial response, measured from baseline to the date of first evidence of confirmed CR or PR, up to 37 months. The analysis used the ITT population and a Cochran-Mantel-Haenszel test adjusted for geographic region and etiology of liver disease.
Objective response rate comparison
Cochran-Mantel-Haenszel test · Two-sided · Superiority hypothesis
Adjustment: geographic region and etiology of liver disease
| Feature | ORR analysis |
|---|---|
| Outcome | Percentage of participants with complete response or partial response |
| Time frame | Baseline to the date of first evidence of confirmed CR or PR (up to 37 months) |
| Endpoint type | Binary |
| Analysis population | Intent-to-treat population: all randomized participants |
| Statistical method | Cochran-Mantel-Haenszel test |
| Adjustment factors | Geographic region and etiology of liver disease |
| P-value | <0.0001 |
The ORR analysis asks whether the proportion of randomized participants achieving a confirmed complete or partial response differs between treatment groups after adjustment for geographic region and etiology of liver disease.
The reported P < 0.0001 indicates strong evidence against the null hypothesis under the Cochran-Mantel-Haenszel testing framework. The ClinicalTrials.gov record does not provide the arm-specific response percentages, a confidence interval for the response-rate difference or ratio, or the corresponding response counts. Those quantities therefore are not reconstructed here.
This distinction is important: a statistically significant comparison does not by itself quantify the magnitude of the difference. For a binary endpoint, an effect size such as an absolute risk difference, risk ratio, odds ratio, or number needed to treat would require the underlying arm-specific response data.
10. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized treatment arm using affected participants divided by participants at risk.
| Safety measure | Ramucirumab + BSC | Placebo + BSC |
|---|---|---|
| Serious adverse events | 124/277 | 92/276 |
The ClinicalTrials.gov record does not provide a formal statistical comparison of serious adverse events, confidence intervals, or a P-value. The displayed affected/at-risk fractions should therefore be treated as descriptive safety information rather than as a formal between-group hypothesis test.
11. Statistical Methods Explained
Why was a log-rank test used for overall survival?
Overall survival is a time-to-event endpoint because both the timing of death and censoring matter. A log-rank test compares the survival experience of randomized groups across follow-up rather than reducing the outcome to a single binary status at one arbitrary time point. In the registry-reported REACH data, the OS analysis used a log-rank test and a stratified Cox model for the hazard ratio.
What does an OS hazard ratio of 0.866 mean?
A hazard ratio of 0.866 means that the fitted model estimated the instantaneous rate of death in the ramucirumab-plus-BSC group to be 0.866 times the corresponding rate in the placebo-plus-BSC group. It does not mean that 86.6% of patients survived, nor that absolute mortality was reduced by 13.4 percentage points.
Why was the Cox model stratified?
The registry analysis notes specify that the hazard ratio and its 95% confidence interval were estimated using a stratified Cox proportional hazards regression model based on the IWRS stratification factors: geographical regions and etiology of liver disease. Stratification allows these prespecified factors to contribute to the comparison without requiring a single common baseline hazard across their strata.
What does the confidence interval add beyond the hazard ratio?
The point estimate summarizes the fitted relative effect, whereas the confidence interval describes its statistical uncertainty. For OS, the reported interval is 0.717–1.046 around the estimate 0.866. For PFS, the interval is 0.522–0.750 around 0.625. The width of an interval is therefore part of the evidence about precision.
Why is the P-value not the effect size?
A P-value summarizes how incompatible the observed data are with a specified null hypothesis under the statistical model. It does not quantify the size of the treatment effect. For example, the PFS analysis reports P < 0.0001 alongside an HR of 0.625; both pieces of information are needed because they answer different questions.
Why use a Cochran-Mantel-Haenszel test for ORR?
ORR is a binary endpoint: each participant either has a confirmed complete or partial response under the registered definition or does not. The Cochran-Mantel-Haenszel procedure provides a way to compare treatment groups while adjusting for the reported stratification factors. In REACH, the registry-reported analysis note specifies adjustment for geographic region and etiology of liver disease.
Why is the ITT population important?
The efficacy analyses were conducted in the ITT population, defined in the ClinicalTrials.gov record as all randomized participants. An ITT analysis retains participants according to randomized assignment rather than selectively excluding participants based on what happened after randomization. This preserves the treatment comparison established by randomization, subject to the assumptions required by the endpoint analysis.
12. Kaplan-Meier Estimation and Censoring
The ClinicalTrials.gov record classifies OS, PFS, and TTP as time-to-event endpoints and report censored participants for each analysis. Kaplan-Meier estimation is the standard descriptive framework for displaying a time-to-event distribution when some participants have not experienced the event by their last observed follow-up.
Here, di represents the number of events at event time ti, while ni is the number at risk immediately before that time.
Censoring means that the exact event time is not observed for a participant during the period contributing information to the analysis. For OS, the registry explicitly states that participants alive at the end of follow-up or lost to follow-up were censored on the last date they were known to be alive.
| Endpoint | Ramucirumab + BSC censored | Placebo + BSC censored |
|---|---|---|
| Overall Survival | 65 | 58 |
| Progression-Free Survival | 43 | 19 |
| Time to Radiographic Progression | 86 | 52 |
The censoring counts should not be interpreted as treatment failures. They represent participants whose endpoint event was not observed within the relevant observation framework.
13. Stratified Analysis
Stratification appears repeatedly in the registry-reported REACH analyses. The OS, PFS, and TTP hazard ratios were estimated with a stratified Cox proportional hazards model using geographical regions and etiology of liver disease as the IWRS stratification factors. The ORR analysis used a Cochran-Mantel-Haenszel adjustment for the same two factors.
| Endpoint | Stratification / adjustment | Purpose in the reported analysis |
|---|---|---|
| Overall Survival | Stratified Cox model using geographical regions and etiology of liver disease | Estimate the hazard ratio while accounting for the reported IWRS stratification factors |
| Progression-Free Survival | Stratified Cox model using geographical regions and etiology of liver disease | Estimate the hazard ratio while accounting for the reported IWRS stratification factors |
| Time to Radiographic Progression | Stratified Cox model using geographical regions and etiology of liver disease | Estimate the hazard ratio while accounting for the reported IWRS stratification factors |
| Objective Response Rate | Cochran-Mantel-Haenszel adjustment for geographic region and etiology of liver disease | Compare the binary response outcome after adjustment for the reported factors |
Stratification does not turn observational subgroup comparisons into separate randomized trials. Its role here is to incorporate the reported design factors into the primary treatment comparisons.
14. Confidence Intervals and Statistical Precision
The three reported time-to-event hazard ratios have different levels of precision, reflected in the widths of their 95% confidence intervals.
| Endpoint | HR | 95% CI | P-value |
|---|---|---|---|
| Overall Survival | 0.866 | 0.717–1.046 | 0.1391 |
| Progression-Free Survival | 0.625 | 0.522–0.750 | <0.0001 |
| Time to Radiographic Progression | 0.593 | 0.487–0.722 | <0.0001 |
The OS interval crosses 1, while the PFS and TTP intervals remain below 1. This describes the statistical evidence reported for each analysis; it does not mean that PFS and TTP are intrinsically more clinically important than OS. They are different endpoints measuring different aspects of the disease course.
A 95% confidence interval is a property of the statistical procedure used to estimate the parameter. It should not be described as a statement that there is a 95% probability that the true hazard ratio lies inside this particular interval. Under repeated sampling and the assumptions of the procedure, intervals constructed in the specified way would contain the corresponding true parameter at the stated long-run rate.
15. Multiplicity and Multiple Endpoints
The ClinicalTrials.gov record identifies one registered primary endpoint and three secondary statistical analyses. The primary endpoint is OS; PFS, TTP, and ORR are reported as secondary endpoints.
| Endpoint | Role | Formal statistical analysis posted? |
|---|---|---|
| Overall Survival | Primary | Yes |
| Progression-Free Survival | Secondary | Yes |
| Time to Radiographic Progression | Secondary | Yes |
| Objective Response Rate | Secondary | Yes |
Multiple endpoints raise an important statistical question: how should evidence be interpreted when several outcomes are tested? The ClinicalTrials.gov record identifies the hypothesis type as superiority and provide the individual P-values, but they do not provide a multiplicity-adjustment strategy, alpha-allocation scheme, or hierarchical testing sequence. Consequently, the page reports the posted P-values without inventing a familywise-error procedure that is not contained in the ClinicalTrials.gov record.
16. Missing Data and Imputation
The registry data provide censoring counts for the time-to-event analyses but do not specify a general missing-data or imputation strategy. For OS, the registered definition explicitly states how participants who were alive at the end of follow-up or lost to follow-up were censored.
| Issue | What the ClinicalTrials.gov record establishes |
|---|---|
| OS censoring | Participants alive at the end of follow-up or lost to follow-up were censored on the last date known to be alive. |
| OS censored participants | Ramucirumab + BSC: 65; Placebo + BSC: 58. |
| PFS censored participants | Ramucirumab + BSC: 43; Placebo + BSC: 19. |
| TTP censored participants | Ramucirumab + BSC: 86; Placebo + BSC: 52. |
| Formal imputation method | Not reported in the ClinicalTrials.gov record. |
For time-to-event outcomes, censoring is not the same as imputing an unobserved event time. The analysis uses the information available up to the censoring time under the assumptions of the survival-analysis framework.
17. Non-Inferiority, Crossover, Factorial, and Bayesian Methods
Superiority
The registered hypothesis type is superiority. The ClinicalTrials.gov record does not report a non-inferiority margin.
Crossover
The ClinicalTrials.gov record does not report a crossover procedure.
Factorial design
The design model is parallel with 2 arms. The ClinicalTrials.gov record does not describe a factorial design.
Bayesian analysis
The posted statistical methods are log-rank and Cochran-Mantel-Haenszel; the ClinicalTrials.gov record does not report a Bayesian method.
These distinctions matter because the statistical interpretation of a trial depends on the question it was designed to answer. A superiority hazard-ratio analysis is not interpreted using a non-inferiority margin, and a randomized parallel comparison is not interpreted as though it were a factorial experiment.
18. What the Hazard Ratio Does — and Does Not — Mean
The OS estimate of 0.866 means that the fitted stratified Cox model estimated the instantaneous rate of death in the ramucirumab-plus-BSC group to be approximately 86.6% of the corresponding rate in the placebo-plus-BSC group. The simple transformation 1 − 0.866 gives an estimated relative hazard reduction of approximately 13.4%.
It does not mean that 13.4% of participants avoided death, that survival increased by 13.4 percentage points, or that each individual participant experienced the same relative reduction.
The PFS estimate of 0.625 corresponds to an estimated relative hazard reduction of approximately 37.5% under the fitted model. The estimate concerns the instantaneous event rate represented by the model; it is not an absolute difference in the probability of progression.
The TTP estimate of 0.593 corresponds to an estimated relative reduction of approximately 40.7% in the modeled hazard of the event. Again, this is not equivalent to saying that 40.7% of participants avoided progression.
19. Primary Endpoint Interpretation in Context
The primary OS analysis reports an HR of 0.866 with a 95% CI of 0.717–1.046 and P = 0.1391. The point estimate is below 1, but the confidence interval includes 1. The appropriate statistical description is therefore to report the estimate, its uncertainty, and the prespecified superiority-testing result rather than reducing the analysis to the point estimate alone.
| Component | What it tells us | What it does not tell us |
|---|---|---|
| HR = 0.866 | Estimated relative instantaneous hazard under the Cox model | An absolute survival difference or individual patient benefit |
| 95% CI = 0.717–1.046 | Uncertainty around the estimated hazard ratio | The range of effects for individual patients |
| P = 0.1391 | Evidence against the specified null under the reported test | Probability that the treatment works or magnitude of effect |
| ITT population | Analysis according to randomized assignment | That all participants necessarily remained on assigned treatment |
This separation between estimate, uncertainty, and hypothesis testing is central to sound interpretation of randomized clinical-trial results.
20. Secondary Endpoint Interpretation in Context
The secondary time-to-event analyses show smaller estimated hazard ratios than the primary OS analysis: 0.625 for PFS and 0.593 for TTP. These are distinct endpoints, however, and the estimates should not be treated as interchangeable measures of one underlying quantity.
| Endpoint | HR | Relative hazard interpretation | 95% CI |
|---|---|---|---|
| Overall Survival | 0.866 | Approximately 13.4% lower estimated hazard | 0.717–1.046 |
| Progression-Free Survival | 0.625 | Approximately 37.5% lower estimated hazard | 0.522–0.750 |
| Time to Radiographic Progression | 0.593 | Approximately 40.7% lower estimated hazard | 0.487–0.722 |
The numerical transformations above are simple interpretations of the reported hazard ratios. They do not create new trial estimates and should not be confused with absolute treatment effects.
21. Important Limitations and Interpretation Issues
- Primary endpoint uncertainty: the OS estimate is 0.866 with a 95% CI of 0.717–1.046, so the reported interval includes 1.
- No median survival values reported: the ClinicalTrials.gov record does not report median OS, median PFS, or median TTP, so those quantities are not presented.
- No arm-specific ORR estimates reported: the ORR analysis reports P < 0.0001 but does not provide the response percentages or counts in the ClinicalTrials.gov record.
- Multiplicity information is incomplete: the registry identifies primary and secondary endpoints but does not provide a multiplicity-adjustment strategy in the ClinicalTrials.gov recordset.
- Cox-model assumptions: the hazard ratios rely on a proportional-hazards model. The ClinicalTrials.gov record does not report a formal proportional-hazards diagnostic.
- Censoring: time-to-event results depend on how censored observations contribute information and on assumptions about censoring.
- Safety interpretation: serious adverse events are reported descriptively as affected/at-risk counts, without a formal statistical comparison in the ClinicalTrials.gov record.
- Limited endpoint detail: the ClinicalTrials.gov record does not provide median survival, subgroup analyses, baseline characteristics, forest plots, or long-term follow-up estimates.
- Generalizability: the trial population is specifically described as participants with hepatocellular carcinoma after first-line therapy with sorafenib; the ClinicalTrials.gov record does not establish applicability to populations outside those trial characteristics.
22. Why This Trial Matters Statistically
REACH is a useful statistical teaching case because its registry results connect randomized trial design with several core methods for time-to-event and categorical outcomes.
| Concept | How it appears in REACH |
|---|---|
| Randomization | The study is registered as randomized with 565.0 participants and 2 parallel treatment arms. |
| Double blinding | The registered masking is double. |
| Intention-to-treat analysis | All reported efficacy analyses use the ITT population: all randomized participants. |
| Time-to-event endpoints | OS, PFS, and TTP are analyzed as time-to-event outcomes. |
| Log-rank testing | Used for OS, PFS, and TTP comparisons. |
| Hazard ratio | Reported as the effect measure for the three time-to-event analyses. |
| Stratified Cox regression | Used to estimate hazard ratios and 95% confidence intervals. |
| Stratification | Geographical regions and etiology of liver disease were used in the reported Cox analyses. |
| Cochran-Mantel-Haenszel testing | Used for the binary ORR analysis with adjustment for geographic region and etiology of liver disease. |
| Confidence intervals | Reported for the three hazard-ratio analyses. |
| Censoring | Censored participant counts are reported for OS, PFS, and TTP. |
| Superiority testing | The registered hypothesis type is superiority. |
23. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
24. Related Statistical Calculators
Explore calculators aligned with the main statistical methods used in REACH:
25. Sources
- ClinicalTrials.gov: REACH, NCT01140347.
- PubMed: PMID 35247922.
- PubMed: PMID 34795131.
- PubMed: PMID 33531690.
- PubMed: PMID 33188713.
- PubMed: PMID 32817068.
Continue with Clinical Biostats statistical methods
Use the related tutorials and calculators to explore the survival-analysis and categorical-data methods represented in the REACH trial.
26. Record Summary
REACH provides a clear example of a randomized, double-blind phase 3 clinical trial in which the primary endpoint was a time-to-event outcome analyzed using a log-rank test and a stratified Cox proportional hazards model. The reported OS hazard ratio was 0.866 with a 95% CI of 0.717–1.046 and P = 0.1391. Secondary analyses reported hazard ratios of 0.625 for PFS and 0.593 for TTP, each with a P-value of <0.0001. ORR was analyzed using a Cochran-Mantel-Haenszel test adjusted for geographic region and etiology of liver disease, with P < 0.0001.
The most useful statistical interpretation keeps these components separate: the effect estimate describes the relative treatment effect under the model; the confidence interval describes statistical uncertainty; the P-value addresses evidence against a specified null hypothesis; and the analysis population, stratification, censoring, and endpoint definition determine how the result should be understood.