This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
PATENT-1 was a randomized, parallel-group, triple-masked phase 3 trial evaluating oral riociguat in patients with pulmonary arterial hypertension. The registry reports 445 enrolled participants, three arms, and a primary endpoint based on change in 6-minute walking distance from baseline to week 12.
| Feature | PATENT-1 |
|---|---|
| Trial name | PATENT-1 |
| ClinicalTrials.gov identifier | NCT00810693 |
| Phase | Phase 3 |
| Condition | Pulmonary Hypertension |
| Therapeutic area | Pulmonology |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 445 |
| Arms | 3 |
| Interventions | Riociguat (Adempas, BAY63-2521) (drug); Riociguat (Adempas, BAY63-2521) (drug); Placebo (drug) |
| Start | 2008-12-17 |
| Primary completion | 2012-05-14 |
| Status | Completed |
| Lead sponsor | Bayer |
2. Clinical Question
The central statistical question was whether participants assigned to riociguat up to 2.5 mg, as analyzed in the registry's reported comparison, experienced a greater change in functional exercise capacity from baseline to week 12 than participants assigned to placebo.
Population
Patients with pulmonary arterial hypertension, as represented by the trial's registered condition of pulmonary hypertension.
Intervention
Riociguat (Adempas, BAY63-2521), including the reported analysis comparing the riociguat group up to 2.5 mg with placebo.
Comparator
Placebo.
Primary question
What is the difference between riociguat and placebo in 6-minute walking distance change from baseline to week 12?
3. Trial Design
Riociguat
- Riociguat (Adempas, BAY63-2521)
- Reported primary and secondary statistical comparisons use the riociguat group up to 2.5 mg versus placebo.
Riociguat
- Riociguat (Adempas, BAY63-2521)
- This is one of the three registered trial arms.
Placebo
- Placebo.
- Placebo is the comparator in the reported statistical analyses.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame | Analysis type |
|---|---|---|---|
| 6 Minutes Walking Distance (6MWD) - Change From Baseline to Week 12 | 6-minute walking distance (6MWD) is a measure for the objective evaluation of a patient's functional exercise capacity. | Baseline and week 12 | Continuous; meters |
The primary endpoint is a change-from-baseline continuous outcome. That matters statistically because the analysis can use both the baseline measurement and the week-12 outcome rather than treating the week-12 measurement as an isolated value.
The registry's analysis notes state that the ANCOVA used baseline 6MWD as a covariate and treatment group, region, and treatment naive/add-on therapy as main effects.
5. Analysis Population and Covariate Adjustment
The reported primary analysis used the Intent to Treat (ITT) population. The registry defines a randomized participant as valid for ITT analyses if at least one dose of study medication was administered.
| Feature | Registry-supported detail |
|---|---|
| Primary analysis population | Intent to Treat (ITT) |
| ITT definition | A randomized participant was valid for ITT analyses if at least one dose of study medication was administered. |
| Primary outcome | 6MWD change from baseline to week 12 |
| Primary model | ANCOVA |
| Baseline covariate | Baseline 6MWD |
| Main effects | Treatment group, region, and treatment naive/add-on therapy |
| Hypothesis type | Superiority |
Covariate adjustment is important here because baseline exercise capacity can be related to the amount of change observed at week 12. Incorporating baseline 6MWD into ANCOVA can therefore improve the precision of the treatment comparison and account for baseline differences in the outcome measure.
The presence of region and treatment naive/add-on therapy as main effects also means the reported comparison was not simply an unadjusted difference between two raw sets of week-12 values. The treatment contrast was estimated within a model that incorporated the prespecified baseline and design-related factors reported by the registry.
6. Statistical Methodology
ANCOVA
ANCOVA, or analysis of covariance, was the principal reported method for the primary 6MWD endpoint. The effect measure was a mean difference in final values, with the registry reporting the estimate as 35.78 meters and a two-sided 95% confidence interval.
For a change-from-baseline outcome, ANCOVA can be viewed as a regression model that estimates the treatment contrast while adjusting for baseline 6MWD and other included factors. The resulting treatment estimate is therefore model-adjusted rather than merely the arithmetic difference between two unadjusted group means.
Wilcoxon / Mann-Whitney analysis
The registry also reports a Wilcoxon (Mann-Whitney) analysis for the primary endpoint, with P < 0.0001. The analysis notes indicate that the comparison was stratified by region and therapy naive/add-on therapy and that the same missing-value framework was incorporated for the primary efficacy parameter.
This provides a nonparametric perspective. Unlike ANCOVA, the Wilcoxon/Mann-Whitney procedure does not require the same normal-error framework as a conventional linear model. It is useful as a complementary analysis when the distribution of the endpoint or model residuals is a concern.
Shapiro-Wilk assessment
The registry reports a Shapiro-Wilk test for normality of the ANCOVA residuals, with P = 0.0001 for the primary endpoint. This is a diagnostic result rather than a treatment-effect estimate.
A normality test should not be confused with a test of whether treatment works. It asks whether a particular distributional assumption is consistent with the observed residuals. A small P-value can indicate departure from normality, but the practical effect of that departure on an ANCOVA treatment estimate depends on the magnitude and nature of the departure, sample size, balance, variance structure, and robustness of the estimator.
Stratified log-rank analysis
The secondary endpoint for percentage of participants with clinical worsening was analyzed using a log-rank method. The registry notes that the test was stratified by region and therapy naive/add-on therapy and that the risk-difference estimate was based on a Mantel-Haenszel estimate stratified by those factors.
Missing-data procedures
The registry provides endpoint-specific imputation rules. For the primary 6MWD analysis, participants who withdrew or died before week 12 were assigned a worst value of 0 m in case of death or clinical worsening without a termination visit, and the last observed value otherwise.
For several secondary continuous endpoints, missing week-12 values were imputed using the last available post-baseline observation. Other endpoints used specific worst-value rules for death or clinical worsening. These choices are part of the estimand-defining analysis strategy: they affect which values enter the comparison and therefore affect the resulting treatment estimate or test statistic.
7. Primary Result: 6-Minute Walking Distance
The primary endpoint was analyzed in the ITT population as change in 6MWD from baseline to week 12. The registry reports a superiority comparison of riociguat up to 2.5 mg versus placebo using ANCOVA.
Adjusted mean difference in 6MWD
95% CI: 20.06–51.51 · P < 0.0001
Two-sided 95% confidence interval; mean difference in final values.
| Primary analysis feature | Reported value |
|---|---|
| Outcome | 6MWD change from baseline to week 12 |
| Comparison | Riociguat up to 2.5 mg_IDT vs Placebo |
| Population | Intent to Treat |
| Method | ANCOVA |
| Effect measure | Mean Difference (Final Values) |
| Estimate | 35.78 meters |
| 95% CI | 20.06 to 51.51 meters |
| P-value | <0.0001 |
| Hypothesis | Superiority |
The estimated mean difference of 35.78 meters means that, under the reported ANCOVA model, the riociguat group had a model-adjusted week-12 6MWD outcome that was 35.78 meters higher than the placebo group in the analyzed comparison.
It does not mean that every participant increased walking distance by 35.78 meters. It is a group-level estimated treatment contrast. It also does not describe the response of an individual patient, the proportion of participants who benefited, or the probability that a particular participant would experience a 35.78-meter improvement.
The 95% confidence interval of 20.06 to 51.51 meters describes statistical uncertainty around the estimated mean difference under the analysis framework. Its width indicates that the estimate is not known with arbitrary precision. The interval is entirely above zero, which is consistent with a positive treatment difference under this model.
The P-value <0.0001 measures how incompatible the observed data are with the null hypothesis represented by the statistical test. It is not a measure of effect size. A P-value does not say that there is a <0.01% probability that the null hypothesis is true, nor does it tell us whether the estimated difference is large or small in clinical terms.
The interpretation also depends on the analysis population, the ANCOVA specification, the baseline covariate, the stratification-related factors, and the prespecified missing-value rules. Those elements are part of the statistical definition of the reported result rather than optional adjustments made after seeing the estimate.
Supporting primary analyses
| Method | Reported result | Purpose |
|---|---|---|
| Wilcoxon (Mann-Whitney) | P < 0.0001 | Nonparametric analysis of the primary endpoint |
| Shapiro-Wilk | P = 0.0001 | Test for normality of ANCOVA residuals |
The consistency of the primary treatment comparison across the reported ANCOVA and Wilcoxon analyses is informative, but the two tests answer their questions through different statistical frameworks. The Shapiro-Wilk result should be interpreted separately because it evaluates a model diagnostic rather than the treatment contrast itself.
8. Secondary Endpoint Results
The registry reports statistical analyses for several secondary outcomes. Most continuous endpoints were evaluated with ANCOVA and/or Wilcoxon/Mann-Whitney methods, while clinical worsening was analyzed with a stratified log-rank procedure and reported as a risk difference.
Pulmonary Vascular Resistance
Mean difference in PVR change
95% CI: −281.37 to −170.08 · P < 0.0001
ANCOVA; two-sided 95% confidence interval.
The reported mean difference is negative because the endpoint is change in pulmonary vascular resistance. Under the stated comparison, the model-adjusted difference in final values was −225.72 dyn*s*cm−5. The confidence interval remains below zero, and the registry reports P < 0.0001.
A separate Wilcoxon/Mann-Whitney analysis also reported P < 0.0001. Its analysis notes state that missing week-12 values were imputed using the last available post-baseline observation and that the test was stratified by region and therapy naive/add-on therapy. A Shapiro-Wilk assessment of ANCOVA residual normality reported P = 0.0001.
N-terminal Prohormone of Brain Natriuretic Peptide
Mean difference in NT-proBNP change
95% CI: −781.52 to −82.10 · P = 0.0157
ANCOVA; two-sided 95% confidence interval.
The ANCOVA estimate is a mean difference of −431.81 pg/mL for change from baseline to week 12, with a 95% CI of −781.52 to −82.10 and P = 0.0157.
The registry also reports a Wilcoxon/Mann-Whitney analysis with P < 0.0001. For that analysis, missing week-12 values were imputed using the last available post-baseline observation, and the test was stratified by region and therapy naive/add-on therapy. The Shapiro-Wilk assessment of ANCOVA residuals reported P = 0.0001.
Clinical Worsening
Risk difference for clinical worsening
95% CI: −9.85 to −0.55 · P = 0.0046
Stratified log-rank analysis; risk difference based on a Mantel-Haenszel estimate.
The registered endpoint was Percentage of Participants With Clinical Worsening, assessed at week 12. The analysis used a log-rank method, and the registry states that the test was for the difference in occurrence of "Any event." It was stratified by region and therapy naive/add-on therapy.
The reported risk difference of −5.20 means that the estimated percentage of participants experiencing the event was 5.20 percentage points lower in the riociguat comparison group than in placebo under the reported analysis. The confidence interval extends from −9.85 to −0.55 percentage points.
EQ-5D Utility Score
| Analysis | Estimate | 95% CI | P-value |
|---|---|---|---|
| ANCOVA | Mean difference 0.06 | 0.01 to 0.11 | 0.0197 |
| Wilcoxon / Mann-Whitney | Not reported as an estimate | — | 0.0663 |
| Shapiro-Wilk | Diagnostic test | — | 0.0001 |
The ANCOVA estimated a mean difference of 0.06 in EQ-5D Utility Score change from baseline to week 12, with a 95% CI of 0.01 to 0.11 and P = 0.0197. The registry also reports a Wilcoxon/Mann-Whitney P-value of 0.0663.
For the nonparametric analysis, missing baseline values were imputed using the last available observation prior to start of study treatment. Missing values for participants who withdrew or died before 12 weeks were assigned a worst value of −0.594 in case of death or clinical worsening without a termination visit, and the last observed value otherwise. The test was stratified by region and therapy naive/add-on therapy.
Living With Pulmonary Hypertension Questionnaire
Mean difference in LPH Questionnaire change
95% CI: −9.79 to −2.54 · P = 0.0009
ANCOVA; two-sided 95% confidence interval.
The ANCOVA estimate for change from baseline to week 12 was a mean difference of −6.17, with a 95% CI of −9.79 to −2.54 and P = 0.0009.
The Wilcoxon/Mann-Whitney analysis reported P = 0.0019. Its missing-data rule assigned a worst value of 105 in case of death or clinical worsening without a termination visit and used the last observed value otherwise. The test was stratified by region and therapy naive/add-on therapy. The Shapiro-Wilk diagnostic for ANCOVA residuals reported P = 0.0001.
WHO Functional Class
| Endpoint | Method | P-value | Key analysis detail |
|---|---|---|---|
| WHO Functional Class - Change From Baseline to Week 12 | Wilcoxon (Mann-Whitney) | 0.0033 | Stratified by region and therapy naive/add-on therapy |
The registry reports P = 0.0033 for the Wilcoxon/Mann-Whitney analysis. Missing values for participants who withdrew or died before 12 weeks were imputed with a worst value of IV in case of clinical worsening without a termination visit or measurement at that termination visit, and with a worst value of V in case of death; the last observed value was used otherwise.
The endpoint is described as binary in the normalized registry data, while the reported analysis is Wilcoxon/Mann-Whitney. The safest interpretation is therefore to report the registry's analysis exactly as posted rather than substitute a different binary-outcome model.
Borg CR 10 Scale
| Endpoint | Method | P-value | Missing-data approach |
|---|---|---|---|
| Borg CR 10 Scale - Change From Baseline to Week 12 | Wilcoxon (Mann-Whitney) | 0.0022 | Worst value of 10 for death or clinical worsening without termination visit; last observed value otherwise |
The reported P-value is 0.0022. The test was stratified by region and therapy naive/add-on therapy. The registry's imputation rule assigns a worst value of 10 in case of death or clinical worsening without a termination visit and otherwise uses the last observed value.
9. Secondary Results: Statistical Summary
| Secondary endpoint | Method | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Pulmonary Vascular Resistance - Change From Baseline to Week 12 | ANCOVA | Mean difference | −225.72 | −281.37 to −170.08 | <0.0001 |
| NT-proBNP - Change From Baseline to Week 12 | ANCOVA | Mean difference | −431.81 | −781.52 to −82.10 | 0.0157 |
| Percentage of Participants With Clinical Worsening | Log-rank | Risk difference | −5.20 | −9.85 to −0.55 | 0.0046 |
| EQ-5D Utility Score - Change From Baseline to Week 12 | ANCOVA | Mean difference | 0.06 | 0.01 to 0.11 | 0.0197 |
| Living With Pulmonary Hypertension Questionnaire - Change From Baseline to Week 12 | ANCOVA | Mean difference | −6.17 | −9.79 to −2.54 | 0.0009 |
| WHO Functional Class - Change From Baseline to Week 12 | Wilcoxon / Mann-Whitney | Not reported | Not reported | Not reported | 0.0033 |
| Borg CR 10 Scale - Change From Baseline to Week 12 | Wilcoxon / Mann-Whitney | Not reported | Not reported | Not reported | 0.0022 |
10. What the Primary Mean Difference Means
The primary estimate of 35.78 meters is a between-group mean difference from the reported ANCOVA analysis. Because the analysis is based on a continuous outcome, the estimate is expressed in the same unit as 6MWD: meters.
The estimate does not imply that every participant had a 35.78-meter treatment effect. Individual treatment responses can vary substantially around a group-level mean. It also does not provide the percentage of participants who improved, the percentage who achieved a particular threshold, or the probability of benefit for an individual patient.
The 95% CI of 20.06 to 51.51 meters quantifies uncertainty around the estimated mean treatment difference under the reported statistical model. A narrower interval would indicate greater precision; the width of this interval reflects the uncertainty that remains after the data and model have been applied.
The reported P < 0.0001 is evidence against the null hypothesis used for the superiority test. It does not measure the magnitude of the 35.78-meter difference and does not tell us whether the difference is clinically important. Effect size, uncertainty, endpoint meaning, and patient-level variability must be considered separately.
11. Statistical Methods Explained
Why was ANCOVA used for 6MWD?
6MWD is a continuous outcome, and the registered primary endpoint is its change from baseline to week 12. ANCOVA is well suited to this setting because it can compare treatment groups while incorporating baseline 6MWD as a covariate. In PATENT-1, the registry additionally reports treatment group, region, and treatment naive/add-on therapy as main effects.
What does a mean difference of 35.78 meters mean?
It is the estimated difference between the two analyzed groups in the final-value treatment contrast under the reported ANCOVA model. A positive value favors the riociguat comparison in terms of higher 6MWD. The number is a group-level estimate, not a prediction for an individual participant.
Why report a confidence interval as well as a P-value?
The two quantities answer different questions. The P-value addresses compatibility with the null hypothesis used by the test, while the confidence interval communicates the statistical precision of the effect estimate. Here, the confidence interval provides a range from 20.06 to 51.51 meters for the model-based mean difference, while P < 0.0001 summarizes evidence against the null hypothesis.
Why was a Wilcoxon/Mann-Whitney analysis also reported?
The Wilcoxon/Mann-Whitney procedure provides a nonparametric comparison that does not rely on the same distributional assumptions as ANCOVA. PATENT-1 reports this analysis for the primary endpoint as well as several secondary endpoints. For the primary endpoint, the reported P-value was <0.0001.
What does the Shapiro-Wilk test tell us?
The Shapiro-Wilk result reported for the primary endpoint was P = 0.0001 and was described as a test of normality of ANCOVA residuals. It is therefore a diagnostic of the model's residual distribution, not evidence for or against the treatment effect itself. A small P-value does not automatically invalidate ANCOVA; the practical importance of non-normal residuals depends on the full analysis context.
Why does missing-data handling matter?
For the primary 6MWD analysis, the registry specifies a worst value of 0 m for death or clinical worsening without a termination visit and the last observed value otherwise. Those rules determine which values enter the week-12 comparison. Different assumptions about what missing observations would have been can change an estimated treatment effect, so the imputation strategy is part of the interpretation of the result.
Why was the clinical-worsening endpoint analyzed differently?
Clinical worsening was treated as a time-to-event endpoint and analyzed using a log-rank method. The reported effect measure was a risk difference, with a Mantel-Haenszel estimate stratified by region and therapy naive/add-on therapy. This illustrates an important principle in trial statistics: different endpoint structures require different estimators and tests.
12. Missing Data and Imputation
Missing-data handling is one of the most consequential methodological details in the PATENT-1 registry analyses because several outcomes use explicit endpoint-specific rules.
| Endpoint / analysis | Reported missing-data approach |
|---|---|
| Primary 6MWD | For participants who withdrew/died before 12 weeks: worst value of 0 m for death or clinical worsening without termination visit; last observed value otherwise. |
| PVR | Missing week-12 values imputed using the last available post-baseline observation for the reported Wilcoxon analysis. |
| NT-proBNP | Missing week-12 values imputed using the last available post-baseline observation for the reported Wilcoxon analysis. |
| WHO Functional Class | Worst value IV for clinical worsening without termination visit or measurement at termination visit; worst value V for death; last observed value otherwise. |
| Borg CR 10 Scale | Worst value of 10 for death or clinical worsening without termination visit; last observed value otherwise. |
| EQ-5D | Missing baseline values imputed using the last available observation prior to start of study treatment; death/clinical worsening before 12 weeks assigned a worst value of −0.594, with last observed value otherwise. |
| LPH Questionnaire | Missing baseline values imputed using the last available observation prior to start of study treatment; death/clinical worsening before 12 weeks assigned a worst value of 105, with last observed value otherwise. |
These approaches are not interchangeable. Some analyses use a last-observation rule, while others deliberately assign a worst value to death or clinical worsening. The statistical consequence is that missingness is not treated merely as an administrative nuisance; the registry's rules encode assumptions about the meaning of an unavailable week-12 measurement.
13. Stratified Analysis
Several reported analyses were stratified by region and therapy naive/add-on therapy. For the primary endpoint, the ANCOVA analysis notes identify treatment group, region, and treatment naive/add-on therapy as main effects. The Wilcoxon analysis was also described as stratified by these factors.
For clinical worsening, the log-rank test was stratified by the same factors, and the risk-difference estimate was based on a Mantel-Haenszel estimate stratified by region and therapy naive/add-on therapy.
Stratification can improve comparability and precision when the stratification variables are related to the endpoint. It also preserves the distinction between a randomized treatment comparison and a simple unadjusted comparison of observed outcomes.
14. Clinical Worsening as a Time-to-Event Endpoint
The registry identifies percentage of participants with clinical worsening as a time-to-event endpoint with assessment at week 12. The analysis uses a log-rank method rather than treating the outcome solely as a simple binary proportion.
This distinction matters because a time-to-event framework can incorporate information about when an event occurs, while a simple week-12 binary analysis would only classify participants according to whether an event had occurred by the specified time.
The reported risk difference was −5.20, with a two-sided 95% CI of −9.85 to −0.55 and P = 0.0046. The registry notes that the test was for the difference in occurrence of "Any event" and that the estimate was based on a stratified Mantel-Haenszel approach.
A risk difference is an absolute difference in event probability or estimated event occurrence between groups, expressed here in percentage-point units. A value of −5.20 indicates a lower estimated occurrence in the riociguat comparison group under the reported analysis. It is fundamentally different from a hazard ratio: it does not describe an instantaneous event-rate ratio.
15. Secondary Endpoint Interpretation
The secondary analyses illustrate why a trial should not be reduced to a single P-value. PATENT-1 reports several dimensions of outcome assessment: functional exercise capacity, pulmonary vascular resistance, a cardiac biomarker, clinical worsening, health utility, disease-specific quality of life, functional class, and perceived exertion.
Functional capacity
6MWD directly evaluates functional exercise capacity and was the registered primary endpoint.
Hemodynamic-related measure
PVR was evaluated as a continuous secondary endpoint and had a reported ANCOVA mean difference of −225.72 dyn*s*cm−5.
Biomarker
NT-proBNP was analyzed as change from baseline to week 12, with a reported ANCOVA mean difference of −431.81 pg/mL.
Patient-reported outcomes
EQ-5D and the Living With Pulmonary Hypertension Questionnaire provide patient-centered measures that were analyzed alongside physiologic and functional outcomes.
The different endpoint types also explain the different statistical methods. Continuous outcomes can be analyzed with ANCOVA, nonparametric methods can provide complementary analyses, and time-to-event outcomes can use survival-analysis methods. The choice of method should follow the endpoint structure rather than being driven by which test produces the smallest P-value.
16. Safety Results
The registry provides serious adverse events by arm as affected participants over participants at risk. These data should be reported as counts rather than converted into additional percentages because the ClinicalTrials.gov record specifies the affected/at-risk format.
| Arm | Serious adverse events | Interpretation of the ClinicalTrials.gov record |
|---|---|---|
| Riociguat (Adempas, BAY63-2521) up to 2. | 29 / 254 | 29 affected participants among 254 at risk |
| Riociguat (Adempas, BAY63-2521) up to 1. | 11 / 63 | 11 affected participants among 63 at risk |
| Placebo | 23 / 126 | 23 affected participants among 126 at risk |
These serious-adverse-event counts should be interpreted separately from the efficacy analyses. A serious adverse event is a safety outcome, while the primary 6MWD analysis is an efficacy outcome. The ClinicalTrials.gov record does not provide enough additional context to construct a broader composite benefit-risk estimate from these counts alone.
17. Randomization and Blinding
PATENT-1 is registered as randomized, with a parallel design and triple masking. Randomization is statistically important because it is the design mechanism that creates the basis for a causal comparison between treatment assignments, subject to the usual assumptions of randomized trials.
Triple masking reduces the opportunity for knowledge of treatment assignment to influence participant behavior, investigator decisions, outcome assessment, or other aspects of trial conduct. The exact identities of the masked parties are not specified in the ClinicalTrials.gov record, so this page does not infer them.
Randomization
Creates the treatment groups to which participants are assigned for the trial comparison.
Triple masking
Limits knowledge of assignment among the masked parties, reducing opportunities for differential behavior or assessment.
18. Why This Trial Matters Statistically
PATENT-1 is a useful statistical teaching case because its registry results combine several important methods within a relatively focused primary endpoint framework.
| Concept | How it appears in PATENT-1 |
|---|---|
| Randomization | Randomized parallel-group phase 3 design. |
| Blinding | Triple masking. |
| ANCOVA | Primary 6MWD analysis and several secondary continuous endpoints. |
| Covariate adjustment | Baseline 6MWD included as a covariate for the primary ANCOVA. |
| Stratified analysis | Region and therapy naive/add-on therapy incorporated into reported analyses. |
| Wilcoxon / Mann-Whitney | Nonparametric analyses of the primary endpoint and multiple secondary outcomes. |
| Log-rank test | Clinical-worsening analysis. |
| Risk difference | Reported for percentage of participants with clinical worsening. |
| Confidence interval | Reported for the primary mean difference and several secondary estimates. |
| Missing-data methods | Endpoint-specific imputation and worst-value rules. |
| ITT analysis | Primary and secondary efficacy analyses identify the ITT population. |
| Model diagnostics | Shapiro-Wilk assessment of ANCOVA residual normality. |
The trial therefore illustrates a broader principle in clinical biostatistics: the statistical analysis is a chain of linked decisions. Randomization establishes the comparison; masking helps protect trial conduct; endpoint definition determines what is being estimated; the analysis population defines who contributes to the estimate; the statistical model determines how the comparison is calculated; missing-data rules determine how unavailable observations are handled; and the confidence interval and P-value describe different aspects of uncertainty and evidence.
19. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The reported primary ANCOVA estimated a mean difference of 35.78 meters, with a two-sided 95% CI of 20.06 to 51.51 and P < 0.0001. The registry also reports a Wilcoxon/Mann-Whitney P-value of <0.0001.
Clinical interpretation
The primary endpoint measures functional exercise capacity using 6-minute walking distance. The statistical result quantifies a group-level difference in this endpoint; deciding how clinically meaningful that difference is requires clinical context beyond the P-value.
The same distinction applies to the secondary endpoints. A mean difference, risk difference, or P-value is a statistical summary of an endpoint. It does not by itself determine whether an individual participant will experience benefit or whether an observed difference is meaningful for every patient.
20. Important Limitations and Interpretation Issues
- Group-level effects: the reported mean differences describe average treatment contrasts and do not characterize the full distribution of individual participant responses.
- Model dependence: the primary estimate comes from ANCOVA and therefore depends on the specified covariate and main-effect structure.
- Missing-data assumptions: several endpoints use explicit imputation or worst-value rules. The resulting estimates depend in part on those assumptions.
- Multiple reported analyses: the registry contains 18 statistical analyses across primary and secondary endpoints. Individual P-values should therefore not automatically be interpreted as if each were an isolated confirmatory test without considering the overall analysis structure.
- Nonparametric versus parametric analyses: a Wilcoxon result and an ANCOVA result are not interchangeable statistics. They use different assumptions and summarize the data through different procedures.
- Diagnostic testing: Shapiro-Wilk P-values concern residual normality, not treatment efficacy. They should not be ranked alongside treatment P-values as though they answered the same question.
- Clinical worsening: the registry uses a time-to-event method and reports a risk difference. A risk difference should not be interpreted as a hazard ratio.
- Safety denominators: serious adverse events are reported as affected participants divided by participants at risk and should be interpreted using those denominators.
- Analysis-population wording: the registry's ITT definition requires that a randomized participant have received at least one dose of study medication. This should be preserved rather than silently replaced with a generic definition of ITT.
21. A Practical Reading of the Primary Analysis
A useful way to read the primary result is to separate five questions.
| Question | PATENT-1 answer from the ClinicalTrials.gov record |
|---|---|
| What was measured? | Change in 6-minute walking distance from baseline to week 12. |
| Who was compared? | Riociguat up to 2.5 mg_IDT versus placebo in the ITT analysis. |
| How was it analyzed? | ANCOVA, with baseline 6MWD as a covariate and treatment group, region, and treatment naive/add-on therapy as main effects. |
| What was estimated? | A mean difference in final values of 35.78 meters. |
| How precise was it? | Two-sided 95% CI 20.06 to 51.51 meters; P < 0.0001. |
This structure is more informative than simply stating that the result was "statistically significant." It identifies the estimand, analysis population, model, effect measure, magnitude, uncertainty, and hypothesis-test result separately.
22. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
23. Related Statistical Calculators
Use these calculator pathways to explore the statistical methods represented in PATENT-1:
24. Sources
- ClinicalTrials.gov: PATENT-1, NCT00810693.
- PubMed record: PMID 23883383.
- PubMed record: PMID 23883378.
- PubMed record: PMID 26135803.
- PubMed record: PMID 27326239.
- PubMed record: PMID 27263466.
The numerical trial results presented on this page are restricted to the ClinicalTrials.gov record. The linked PubMed records are provided as source records associated with the trial.
Continue through the Clinical Biostats statistical pathway
Connect PATENT-1's endpoints and methods to deeper tutorials and statistical calculators for clinical-trial analysis.
25. Record Summary
PATENT-1 provides a compact but statistically rich example of a randomized phase 3 clinical trial. Its registered primary endpoint was change in 6-minute walking distance from baseline to week 12, analyzed in the ITT population using ANCOVA with baseline 6MWD as a covariate and treatment group, region, and treatment naive/add-on therapy as main effects. The reported mean difference was 35.78 meters, with a two-sided 95% CI of 20.06 to 51.51 and P < 0.0001.
The registry also provides complementary nonparametric analyses, a Shapiro-Wilk residual diagnostic, stratified analysis, a log-rank analysis of clinical worsening, and endpoint-specific missing-data rules. Secondary outcomes include PVR, NT-proBNP, clinical worsening, EQ-5D, the Living With Pulmonary Hypertension Questionnaire, WHO Functional Class, and the Borg CR 10 Scale. Together, these analyses demonstrate why clinical-trial interpretation requires more than reading a single P-value: the endpoint definition, analysis population, effect measure, confidence interval, model, missing-data strategy, and clinical meaning all contribute to the statistical story.