This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical results on this page are restricted to the trial data reported for FREEDOM-M.
1. Trial at a Glance
FREEDOM-M was a randomized, double-blind, parallel phase 3 trial evaluating oral treprostinil (UT-15C) sustained release tablets versus placebo as monotherapy for pulmonary arterial hypertension. The primary endpoint was the placebo-corrected change in six-minute walk distance from Baseline to Week 12.
| Feature | FREEDOM-M |
|---|---|
| Trial name | FREEDOM-M |
| NCT identifier | NCT00325403 |
| Phase | Phase 3 |
| Status | COMPLETED |
| Condition | Pulmonary Hypertension |
| Brief title | FREEDOM - M: Oral Treprostinil as Monotherapy for the Treatment of Pulmonary Arterial Hypertension (PAH) |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 349 |
| Interventions | Oral treprostinil (UT-15C) Sustained Release Tablets; Placebo |
| Hypothesis type | Superiority |
| Results posted | Yes |
2. Clinical Question
The central statistical question was whether oral treprostinil produced a difference in the change in six-minute walk distance from Baseline to Week 12 compared with placebo in patients with pulmonary arterial hypertension.
Population
Participants enrolled in the FREEDOM-M phase 3 trial for pulmonary arterial hypertension.
Intervention
Oral treprostinil (UT-15C) sustained release tablets.
Comparator
Placebo.
Primary question
Does oral treprostinil produce a superior placebo-corrected change in six-minute walk distance from Baseline to Week 12?
3. Trial Design
UT-15C / Oral Treprostinil
- Oral treprostinil (UT-15C)
- Sustained release tablets
- Included in the primary comparison against placebo
Placebo
- Placebo
- Included in the primary comparison against oral treprostinil
4. Randomization and Analysis Population
The primary analysis was conducted in a modified intention-to-treat (mITT) group of n=228. The registry defines this group as subjects with access to 0.25 mg tablets at randomization. The analysis notes state that all alpha was spent on this subgroup.
| Population / feature | Registry-supported description |
|---|---|
| Overall enrollment | 349 participants |
| Primary efficacy population | Modified intention-to-treat group |
| mITT size | n=228 |
| mITT definition | Subjects with access to 0.25 mg tablets at randomization |
| Alpha allocation | All alpha was spent on this subgroup |
| Analysis comparison | Placebo vs UT-15C (Oral Treprostinil) |
5. Primary Endpoint
| Endpoint | Definition / assessment | Analysis |
|---|---|---|
| Six Minute Walk Distance (6MWD) | Placebo corrected change in six minute walk distance (6MWD) from Baseline to Week 12, which correlates with the historical clinical standard for assessing patient functional status in the treatment of PAH and is considered an objective measure of patient functional status by the American Thoracic Society (ATS). | ANCOVA |
The registered time frame was Baseline and Week 12. The endpoint was continuous and measured in meters. The statistical analysis used the Hodges-Lehmann (H-L) estimate as the reported effect measure, with a two-sided 95% confidence interval.
6. Primary Result: Six Minute Walk Distance
Placebo-corrected change in 6MWD
95% CI: 4–41 meters · P = 0.0125
ANCOVA · modified intention-to-treat population, n=228 · superiority hypothesis
| Primary endpoint | Time frame | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Six Minute Walk Distance (6MWD) | Baseline and Week 12 | Hodges-Lehmann (H-L) | 23 meters | 4 to 41 meters | 0.0125 |
Method: ANCOVA
The registry reports a Hodges-Lehmann estimate of 23 with a two-sided 95% confidence interval of 4 to 41 and a P-value of 0.0125.
The reported estimate of 23 meters represents the placebo-corrected treatment difference in the change in six-minute walk distance from Baseline to Week 12, as estimated using the registry-reported analysis framework. Because the estimate is positive, the comparison favors the oral-treprostinil group in terms of the direction of change in 6MWD.
The 95% confidence interval of 4 to 41 meters expresses statistical uncertainty around the estimated treatment difference. It does not mean that individual patients experienced changes only within this interval, nor does it describe the distribution of individual treatment responses.
The P-value of 0.0125 quantifies the statistical evidence against the null hypothesis under the specified testing framework. It is not a measure of how large or clinically important the treatment effect is. Effect magnitude and uncertainty are better communicated by the estimate and its confidence interval.
The analysis is also population-specific: the primary result was obtained in the mITT group of n=228, rather than simply in all 349 enrolled participants. The registry also states that all alpha was spent on this subgroup.
Sample-size and power context
The registry analysis notes state that, using an allocation ratio of 2:1 between oral treprostinil and placebo, a fixed sample size of approximately 195 subjects with access to 0.25 mg tablets at randomization would provide at least 90% power at a significance level of 0.01 using a two-sided hypothesis to detect a between-treatment difference in the change from Baseline to Week 12 in distance traversed during the six-minute walk, under the assumptions specified in the registry analysis text.
This provides important context for interpreting the primary result: the trial's statistical planning was tied specifically to the subgroup with access to 0.25 mg tablets, and the reported primary analysis likewise used the mITT population of 228.
7. Secondary Endpoint Results
The registry also reports several secondary analyses. These use different endpoints and, in some cases, different statistical methods. They should therefore not be treated as interchangeable with the prespecified primary analysis.
6MWD at Week 11
Placebo-corrected change in 6MWD
95% CI: −2 to 33 meters · P = 0.0653
Baseline and Week 11 · ANCOVA · Hodges-Lehmann estimate
The estimated treatment difference at Week 11 was 13 meters, with a two-sided 95% confidence interval from −2 to 33 meters. The interval includes zero, so the registry-reported estimate is less precise than the Week 12 primary result with respect to excluding no treatment difference.
The P-value of 0.0653 is not an effect-size measure. It should be read together with the estimate and confidence interval rather than used as a substitute for them.
6MWD at Week 8
Placebo-corrected change in 6MWD
95% CI: 1–33 meters · P = 0.0307
Baseline and Week 8 · ANCOVA · Hodges-Lehmann estimate
The Week 8 estimate was 17 meters, with a 95% confidence interval from 1 to 33 meters. The positive estimate and interval provide a treatment-group difference in the same direction as the Week 12 primary analysis, while the interval indicates the statistical precision of this earlier assessment.
Because this is a secondary time point rather than the registered primary endpoint, it should not replace the Week 12 analysis when describing the primary trial question.
6MWD at Week 4
Placebo-corrected change in 6MWD
95% CI: 0–24 meters · P = 0.0518
Baseline and Week 4 · ANCOVA · Hodges-Lehmann estimate
The Week 4 estimate was 12 meters, with a two-sided 95% confidence interval from 0 to 24 meters. The estimate is positive, but the interval reaches zero. The P-value of 0.0518 describes the statistical evidence at this time point; it does not determine the magnitude of the estimated difference.
World Health Organization Functional Classification for PAH
| Endpoint | Time frame | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| World Health Organization Functional Classification for PAH | Baseline and Week 12 | Wilcoxon rank sum test | 0 | 0 to 0 | 0.7380 |
The analysis population was subjects with a WHO functional classification assessment at Week 12. The registry reports a Hodges-Lehmann estimate of 0 and a two-sided 95% confidence interval of 0 to 0. The registry notes that the estimated parameter and confidence interval were calculated.
Borg Dyspnea Score
| Endpoint | Time frame | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Borg Dyspnea Score | Baseline and Week 12 | Wilcoxon rank-sum test | 0 | −1 to 0 | 0.4887 |
The endpoint was measured in units on a scale. The nonparametric Wilcoxon rank-sum approach produced a Hodges-Lehmann estimate of 0, with a two-sided 95% confidence interval from −1 to 0 and a P-value of 0.4887.
Dyspnea-Fatigue Index
| Endpoint | Time frame | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Dyspnea-Fatigue Index | Baseline and Week 12 | Wilcoxon sum-rank test | 0 | 0 to 1 | 0.6116 |
Two subjects, one in the placebo arm and one in the oral-treprostinil arm, from the primary analysis population of n=228 did not have a Baseline dyspnea-fatigue index score and were not included in this analysis. The reported Hodges-Lehmann estimate was 0, with a two-sided 95% confidence interval of 0 to 1.
Clinical Worsening Assessment
| Endpoint | Time frame | Method | 95% CI | P-value |
|---|---|---|---|---|
| Clinical Worsening Assessment | Baseline and Week 12 | Fisher Exact | 95% | 1.000 |
The Clinical Worsening Assessment was analyzed as a binary endpoint using Fisher Exact. The registry reports a 1.000 P-value and a 95% confidence interval level, but does not provide an effect estimate in the posted statistical analysis.
8. Post-Hoc Analyses
The registry also reports post-hoc analyses of six-minute walk distance. These analyses can help describe patterns within particular baseline categories or across the entire study population, but they are distinct from the registered primary endpoint.
| Post-hoc analysis | Time frame | Population / subgroup | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| 6MWD by baseline WHO Functional Classification III or IV | Baseline and Week 12 | Baseline WHO Functional Classification III or IV | 26 meters | 1–49 meters | 0.0326 |
| 6MWD by baseline WHO Functional Classification: I or II | Baseline and Week 12 | Baseline WHO Functional Classification I or II | 16.0 meters | −15 to 47 meters | 0.2275 |
| 6MWD by PAH Etiology: Idiopathic or Heritable PAH | Baseline and Week 12 | Idiopathic or Heritable PAH | 32 meters | 10–55 meters | 0.0024 |
| 6MWD for the Entire Study Population | Baseline and Week 12 | All subjects enrolled, regardless of tablet strength availability at randomization | 25.5 meters | 10–41 meters | 0.0001 |
| 6MWD for the Entire Study Population | Baseline and Week 11 | All subjects enrolled, regardless of tablet strength availability at randomization | 17 meters | 3–33 meters | 0.0025 |
| 6MWD for the Entire Study Population | Baseline and Week 8 | All subjects enrolled, regardless of tablet strength availability at randomization | 20 meters | 7–34 meters | 0.0008 |
| 6MWD for the Entire Study Population | Baseline and Week 4 | All subjects enrolled, regardless of tablet strength availability at randomization | 14 meters | 3.9–25 meters | 0.0025 |
All of these post-hoc 6MWD analyses used ANCOVA and reported a Hodges-Lehmann estimate with a two-sided 95% confidence interval.
The subgroup estimates are descriptive extensions of the primary analysis rather than replacements for it. For example, the estimated difference was 26 meters for participants with baseline WHO Functional Classification III or IV and 16.0 meters for those with baseline WHO Functional Classification I or II. Their confidence intervals differ substantially in width, illustrating how subgroup analysis can reduce statistical precision.
The two subgroup P-values also differ, but that alone does not establish that the treatment effect differs between the two functional-classification groups. Demonstrating treatment-effect heterogeneity would ordinarily require an appropriate interaction analysis rather than comparing P-values from separate subgroups.
The entire-study-population analyses are also analytically distinct from the primary mITT analysis because the registry explicitly states that they included all enrolled subjects regardless of tablet strength availability at randomization.
9. Statistical Methodology
| Method | Role in FREEDOM-M | Endpoint(s) |
|---|---|---|
| ANCOVA | Primary linear-model comparison of treatment groups | Primary and post-hoc 6MWD analyses |
| Wilcoxon / Mann-Whitney approach | Nonparametric comparison between treatment groups | WHO Functional Classification, Borg Dyspnea Score, Dyspnea-Fatigue Index |
| Fisher exact test | Exact categorical comparison | Clinical Worsening Assessment |
| Hodges-Lehmann estimate | Reported effect measure for the continuous/nonparametric analyses where provided | 6MWD and several functional measures |
ANCOVA for the primary endpoint
The registry reports ANCOVA for the primary 6MWD analysis. In a clinical trial, ANCOVA is commonly used to compare treatment groups on a continuous outcome while accounting for baseline measurement. For a change-from-baseline endpoint, the underlying goal is to estimate the treatment-group difference while improving statistical efficiency relative to an analysis that ignores baseline information.
The exact model specification beyond the registry's reported ANCOVA method is not provided in the ClinicalTrials.gov record, so this page does not add unreported covariates or model terms.
Nonparametric analyses
The WHO Functional Classification, Borg Dyspnea Score, and Dyspnea-Fatigue Index analyses used Wilcoxon-based methods. These methods compare the distributions or ranks of observations rather than relying on the same normal-error assumptions as a conventional parametric mean comparison.
Fisher exact test
Clinical Worsening Assessment was analyzed using Fisher Exact. This is appropriate for categorical comparisons when an exact test of the treatment-group association is desired, particularly when cell counts may be small.
Hodges-Lehmann estimates
The registry reports the Hodges-Lehmann (H-L) estimate for the primary 6MWD analysis and several other analyses. In a two-group setting, the Hodges-Lehmann approach provides a robust location estimate based on pairwise treatment-group comparisons. Its interpretation should follow the endpoint and analysis framework reported by the registry rather than being treated automatically as a conventional arithmetic mean difference.
10. Statistical Methods Explained
Why was ANCOVA used for 6MWD?
6MWD is a continuous outcome, making a linear-model framework such as ANCOVA a natural choice. ANCOVA can incorporate baseline information when estimating the treatment difference at the follow-up assessment. The ClinicalTrials.gov record identifies ANCOVA as the method used, but do not provide the complete model specification, so additional covariates should not be assumed.
What does a 23-meter estimate mean?
The primary Hodges-Lehmann estimate was 23 meters. In the context of this trial, that is the reported placebo-corrected treatment difference in change in 6MWD from Baseline to Week 12. It is not the same thing as saying that every participant walked 23 meters farther, nor does it describe the response of any particular patient.
What does the 95% confidence interval of 4 to 41 mean?
The interval quantifies statistical uncertainty around the estimated treatment difference. The lower and upper limits are 4 and 41 meters. A confidence interval is about the uncertainty of the estimated treatment effect under the statistical framework; it is not a prediction interval for individual patients.
Why doesn't the P-value measure treatment effect size?
The primary P-value was 0.0125. A P-value measures the strength of evidence against a null hypothesis under a specified testing framework. It depends on both the magnitude of an observed difference and the amount of information available. The estimate and confidence interval are therefore necessary to understand the size and precision of the treatment effect.
Why are the Week 8, Week 11, and Week 12 results not interchangeable?
They answer related but distinct time-point questions. The primary endpoint was defined at Week 12, while Week 8 and Week 11 were secondary endpoints. The estimates were 17, 13, and 23 meters, respectively. Even though they concern the same underlying measure, their inferential roles in the trial are different.
Why does the analysis population matter?
The primary result was based on the mITT group of n=228, defined by access to 0.25 mg tablets at randomization. Post-hoc analyses of the entire study population used all enrolled subjects regardless of tablet strength availability. Comparing these results therefore involves more than comparing two estimates: the populations being analyzed are different.
Why should subgroup P-values be interpreted cautiously?
The post-hoc analyses by baseline WHO Functional Classification illustrate the issue. A subgroup estimate can be statistically significant in one category and not in another without demonstrating that the underlying treatment effects truly differ. A formal interaction test is generally needed to evaluate treatment-effect heterogeneity.
11. Secondary and Post-Hoc Results in Context
| Analysis | Estimate | 95% CI | P-value | Interpretive role |
|---|---|---|---|---|
| Primary 6MWD, Week 12, mITT | 23 meters | 4–41 meters | 0.0125 | Registered primary comparison |
| 6MWD, Week 11, mITT | 13 meters | −2–33 meters | 0.0653 | Secondary time point |
| 6MWD, Week 8, mITT | 17 meters | 1–33 meters | 0.0307 | Secondary time point |
| 6MWD, Week 4, mITT | 12 meters | 0–24 meters | 0.0518 | Secondary time point |
| 6MWD, WHO III or IV | 26 meters | 1–49 meters | 0.0326 | Post-hoc subgroup |
| 6MWD, WHO I or II | 16.0 meters | −15–47 meters | 0.2275 | Post-hoc subgroup |
| 6MWD, idiopathic or heritable PAH | 32 meters | 10–55 meters | 0.0024 | Post-hoc subgroup |
| 6MWD, entire study population, Week 12 | 25.5 meters | 10–41 meters | 0.0001 | Post-hoc full population |
| 6MWD, entire study population, Week 11 | 17 meters | 3–33 meters | 0.0025 | Post-hoc full population |
| 6MWD, entire study population, Week 8 | 20 meters | 7–34 meters | 0.0008 | Post-hoc full population |
| 6MWD, entire study population, Week 4 | 14 meters | 3.9–25 meters | 0.0025 | Post-hoc full population |
The pattern across the reported 6MWD analyses is statistically informative because the estimates are not identical at each time point or population definition. That is expected: treatment effects are estimated with sampling variability, and the analyzed populations differ. The correct comparison is therefore not to look for a single universal number, but to identify which estimate answers the prespecified primary question.
12. Safety
The ClinicalTrials.gov record reports serious adverse events by arm as affected participants divided by participants at risk for the reported safety category.
| Arm | Serious adverse events affected / at risk |
|---|---|
| Placebo | 15/77 |
| UT-15C (Oral Treprostinil) | 27/151 |
The reported serious-adverse-event counts are 15/77 for placebo and 27/151 for UT-15C. These denominators are the affected/at-risk figures reported in the ClinicalTrials.gov record and should not be replaced with the overall enrollment of 349 or the primary mITT size of 228.
The ClinicalTrials.gov record does not provide a formal between-arm hypothesis test for these serious-adverse-event figures, so this page does not calculate one or infer a safety comparison beyond the reported counts.
13. Multiplicity and Alpha
The primary analysis notes state that all alpha was spent on the mITT subgroup. They also describe a sample-size framework using a significance level of 0.01 with a two-sided hypothesis.
This matters because a clinical trial can contain many endpoints and repeated time points, while its confirmatory statistical framework may allocate the type I error budget to a particular primary comparison. The fact that later secondary and post-hoc analyses have P-values does not automatically give them the same confirmatory status as the registered primary endpoint.
14. Missing Data and Measurement Considerations
The ClinicalTrials.gov record identifies one explicit missing-baseline issue: two subjects, one in the placebo arm and one in the oral-treprostinil arm, from the primary analysis population of n=228 did not have a Baseline dyspnea-fatigue index score and were not included in that analysis.
The ClinicalTrials.gov record do not specify an imputation method for that missing Dyspnea-Fatigue Index information, nor do they provide a broader missing-data strategy for the primary 6MWD analysis. This page therefore does not infer an imputation procedure.
15. Blinding and Randomization
FREEDOM-M is identified in the registry as randomized, parallel, and double-masked. Randomization provides the basis for comparing treatment groups without deliberately assigning treatment according to baseline prognosis, while double masking is intended to reduce knowledge of treatment assignment during the trial.
The analysis notes specify a 2:1 allocation ratio between oral treprostinil and placebo. The ClinicalTrials.gov record does not identify the complete randomization stratification scheme, so no additional stratification factors are presented here.
16. What the Different Endpoints Measure
6MWD
A continuous measure reported in meters. The primary endpoint was the placebo-corrected change from Baseline to Week 12.
WHO Functional Classification
A functional classification endpoint analyzed with a Wilcoxon rank sum test in the posted analysis.
Borg Dyspnea Score
A scale-based endpoint analyzed using a Wilcoxon rank-sum approach.
Clinical Worsening
A binary endpoint analyzed using Fisher Exact.
These endpoints should not be interpreted as interchangeable. The six-minute walk distance provides a continuous measure of walking performance, while the other endpoints capture different dimensions of functional status, symptoms, or clinical worsening. A treatment effect can therefore differ in magnitude and statistical evidence across endpoints without the results being mathematically inconsistent.
17. Reading the Primary Result Correctly
The primary estimated placebo-corrected difference was 23 meters. This is the central estimate of the treatment-group difference reported by the registry.
The two-sided 95% CI was 4 to 41 meters. The width of the interval communicates the uncertainty surrounding the estimated difference.
The primary P-value was 0.0125. It addresses the statistical testing question; it does not quantify clinical importance or the probability that the treatment is effective.
The primary analysis used the modified intention-to-treat population of n=228. This is essential context for the reported estimate and confidence interval.
18. Primary vs Secondary vs Post-Hoc Analysis
| Analysis class | Example in FREEDOM-M | Interpretive status |
|---|---|---|
| Primary | 6MWD, Baseline and Week 12 | Registered primary efficacy question |
| Secondary | 6MWD at Weeks 4, 8, and 11 | Additional prespecified outcome analyses |
| Secondary | WHO Functional Classification, Borg Dyspnea Score, Dyspnea-Fatigue Index, Clinical Worsening Assessment | Additional outcome analyses |
| Post-hoc | 6MWD by baseline WHO Functional Classification and PAH etiology | Exploratory subgroup analyses |
| Post-hoc | 6MWD for the entire study population | Exploratory analysis using all enrolled subjects |
This hierarchy is important because statistical significance is not the only dimension of interpretation. The prespecified primary endpoint has a different evidentiary role from a post-hoc subgroup analysis, even if the latter has a smaller P-value.
19. Why This Trial Matters Statistically
FREEDOM-M is a useful teaching case because the posted analyses bring several core clinical-trial concepts together in a single randomized phase 3 study: a continuous primary endpoint, ANCOVA, a modified intention-to-treat population, a two-sided confidence interval, a prespecified superiority hypothesis, nonparametric secondary analyses, an exact categorical test, and post-hoc population and subgroup analyses.
| Concept | How it appears in FREEDOM-M |
|---|---|
| Randomization | Randomized parallel-group design |
| Blinding | Double masking |
| Allocation ratio | 2:1 oral treprostinil to placebo in the primary analysis notes |
| Modified ITT | Primary analysis in the mITT group, n=228 |
| ANCOVA | Primary and post-hoc 6MWD analyses |
| Hodges-Lehmann estimate | Reported effect measure for the primary 6MWD analysis and several other analyses |
| Confidence interval | Two-sided 95% CI for the primary treatment estimate |
| P-value | 0.0125 for the primary Week 12 analysis |
| Wilcoxon test | WHO Functional Classification, Borg Dyspnea Score, Dyspnea-Fatigue Index |
| Fisher exact test | Clinical Worsening Assessment |
| Multiplicity / alpha | Analysis notes state that all alpha was spent on the mITT subgroup |
| Post-hoc analysis | WHO functional-classification subgroups, PAH etiology, and entire study population |
| Missing data | Two subjects lacked Baseline Dyspnea-Fatigue Index scores |
20. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary analysis estimated a 23-meter placebo-corrected difference in change in 6MWD from Baseline to Week 12, with a two-sided 95% CI of 4 to 41 meters and a P-value of 0.0125 in the mITT population of n=228.
Clinical interpretation
The primary endpoint was six-minute walk distance, an objective measure of patient functional status described in the registry as correlating with the historical clinical standard for assessing functional status in PAH. Clinical meaning should consider the magnitude of the estimated difference alongside uncertainty and the other reported endpoints.
The distinction matters. Statistical evidence tells us how compatible the observed data are with a specified null hypothesis under the analysis framework. Clinical interpretation asks what the size and nature of the observed difference mean in the context of patient function. Neither question is fully answered by the P-value alone.
21. Important Limitations and Interpretation Issues
- mITT population: the primary analysis used n=228 subjects with access to 0.25 mg tablets at randomization rather than all 349 enrolled participants.
- Different analysis populations: the post-hoc entire-study-population analyses included all subjects enrolled regardless of tablet strength availability at randomization, making them different from the primary analysis population.
- Post-hoc subgroup analyses: subgroup results should be interpreted as exploratory and should not be used to infer treatment-effect heterogeneity simply by comparing separate P-values.
- Secondary endpoints: Week 4, Week 8, and Week 11 6MWD analyses answer different time-point questions from the registered Week 12 primary endpoint.
- Multiplicity: the analysis notes state that all alpha was spent on the mITT subgroup. Secondary and post-hoc P-values should not automatically be given the same confirmatory interpretation as the primary analysis.
- Missing data: the ClinicalTrials.gov record identifies two missing Baseline Dyspnea-Fatigue Index scores but do not specify a broader imputation strategy.
- Effect measure: the registry reports a Hodges-Lehmann estimate. It should not automatically be described as an arithmetic mean difference without additional information.
- Safety denominators: the registry-reported serious-adverse-event figures use affected/at-risk denominators of 15/77 and 27/151; these should not be replaced by the overall enrollment or mITT denominator.
- Incomplete model specification: the registry identifies ANCOVA but the ClinicalTrials.gov record does not provide the complete covariate structure of the model.
22. A Practical Statistical Reading of FREEDOM-M
Randomized, double-blind, parallel phase 3 trial
The trial enrolled 349 participants and compared oral treprostinil with placebo.
Modified intention-to-treat group
The primary analysis used n=228 subjects with access to 0.25 mg tablets at randomization, with all alpha spent on this subgroup.
6MWD at Week 12
The registered endpoint was the placebo-corrected change in six-minute walk distance from Baseline to Week 12.
ANCOVA
The treatment comparison produced a Hodges-Lehmann estimate of 23 meters, with a two-sided 95% CI of 4 to 41 meters and P=0.0125.
Secondary and post-hoc analyses
Additional 6MWD time points, functional measures, clinical worsening, subgroup analyses, and entire-study-population analyses were also reported.
23. What the Primary Estimate Does — and Does Not — Mean
The 23-meter estimate summarizes the reported placebo-corrected difference in change in 6MWD from Baseline to Week 12 under the primary analysis framework.
It does not mean that every participant receiving oral treprostinil increased their walking distance by exactly 23 meters. It is a population-level treatment comparison, not an individual prediction.
The 4 to 41 meter interval shows the uncertainty around the estimated treatment difference. The interval is more informative about precision than the point estimate alone.
The 0.0125 P-value provides the statistical-test result under the specified superiority framework. It should not be interpreted as the probability that the treatment effect is exactly 23 meters or as a measure of clinical importance.
24. Related Tutorials
Learn more about the methods used in this trial:
25. Related Calculators
26. Sources
- ClinicalTrials.gov: FREEDOM-M, NCT00325403.
- PubMed: PMID 23307827.
Continue through the Clinical Biostats statistical pathway
Explore the statistical methods behind randomized trials, continuous outcomes, confidence intervals, nonparametric tests, and categorical comparisons.
27. Record Summary
FREEDOM-M provides a compact example of how a randomized phase 3 trial can combine several statistical approaches around a clinically meaningful functional endpoint. The registered primary endpoint was the placebo-corrected change in six-minute walk distance from Baseline to Week 12, analyzed by ANCOVA in a modified intention-to-treat population of n=228. The reported Hodges-Lehmann estimate was 23 meters, with a two-sided 95% CI of 4 to 41 meters and a P-value of 0.0125.
The broader statistical record includes secondary 6MWD assessments at Weeks 4, 8, and 11, nonparametric analyses of functional and symptom measures, a Fisher exact analysis of clinical worsening, and post-hoc analyses by baseline functional classification, PAH etiology, and the entire study population. Reading these results correctly requires keeping the endpoint hierarchy, analysis population, confidence intervals, P-values, and post-hoc status distinct rather than reducing the trial to a single numerical result.