This page separates reported trial results from statistical interpretation. Numerical results are restricted to the information reported in the ClinicalTrials.gov record. The registry provides the official trial record.
1. Trial at a Glance
STRIVE was a randomized, parallel, quadruple-masked phase 3 treatment trial evaluating erenumab (AMG 334) for migraine prevention. The registry reports 955 enrolled participants and three arms, with efficacy comparisons between placebo and each of the two erenumab doses.
| Feature | STRIVE |
|---|---|
| Trial name | STRIVE |
| ClinicalTrials.gov identifier | NCT02456740 |
| Brief title | Study to Evaluate the Efficacy and Safety of Erenumab (AMG 334) in Migraine Prevention |
| Therapeutic area | Neurology |
| Condition | Migraine |
| Phase | Phase 3 |
| Status | Completed |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 955 |
| Interventions | Erenumab; placebo |
| Lead sponsor | Amgen |
| Sponsor type | Industry |
| Results posted | Yes |
| Outcome measures posted | 5 |
| Statistical analyses posted | 10 |
2. Clinical Question
The central statistical question was whether treatment with erenumab produced a greater improvement than placebo in migraine-related outcomes during the double-blind treatment period.
Population
Participants enrolled in the phase 3 STRIVE trial for the prevention of migraine.
Intervention
Erenumab (AMG 334), evaluated at 70 mg QM and 140 mg QM in the reported efficacy comparisons.
Comparator
Placebo.
Primary question
Does erenumab reduce mean monthly migraine days from baseline during the last 3 months of the 24-week double-blind treatment phase compared with placebo?
3. Trial Design
Placebo
- Placebo was the comparator for both primary efficacy analyses.
- Double-blind treatment phase safety data included 7 serious adverse events among 319 participants at risk.
Erenumab
- Erenumab 70 mg QM was compared with placebo for the primary and secondary efficacy analyses.
- Erenumab 140 mg QM was compared with placebo for the primary and secondary efficacy analyses.
- Double-blind treatment phase safety data included 8 serious adverse events among 314 participants at risk for Erenumab 70 mg and 8 among 319 participants at risk for the other erenumab arm identified in the registry data as Erenumab 1.
Parallel-group design with three arms.
The registry identifies the study as quadruple-masked.
The trial evaluates efficacy and safety in migraine prevention.
The ClinicalTrials.gov record identifies these as the study start and primary completion dates.
4. Endpoints
The registry identifies one primary endpoint and reports five outcome measures overall. The primary endpoint was assessed using the 4-week baseline phase and the last 3 months of the 24-week double-blind treatment phase.
| Role | Endpoint | Time frame | Type |
|---|---|---|---|
| Primary | Change From Baseline in Mean Monthly Migraine Days to the Last 3 Months of the Double-blind Treatment Period | 4-week baseline phase and the last 3 months (months 4, 5, and 6) of the 24-week double-blind treatment phase | Continuous |
| Secondary | Percentage of Participants With at Least a 50% Reduction From Baseline in Monthly Migraine Days in the Last 3 Months of the Double-blind Treatment Phase | 4-week baseline phase and the last 3 months (months 4, 5, and 6) of the 24-week double-blind treatment phase | Binary |
| Secondary | Change From Baseline in Monthly Acute Migraine-specific Medication Treatment Days to the Last 3 Months of the Double-blind Treatment Period | 4-week baseline phase and the last 3 months (months 4, 5, and 6) of the 24-week double-blind treatment phase | Continuous |
| Secondary | Change From Baseline in Mean Monthly Average Physical Impairment Domain Score Measured by MPFID in the Last 3 Months of the Double-blind Treatment Phase | 4-week baseline phase and the last 3 months (months 4, 5, and 6) of the 24-week double-blind treatment phase | Continuous |
| Secondary | Change From Baseline in Mean Monthly Average Impact on Everyday Activities Score Measured by MPFID in the Last 3 Months of the Double-blind Treatment Phase | 4-week baseline phase and the last 3 months (months 4, 5, and 6) of the 24-week double-blind treatment phase | Continuous |
Primary endpoint definition
The registry defines a migraine day as any calendar day in which the participant experienced a qualified migraine headache, including onset, continuation, or recurrence. A qualified migraine headache was defined either as a migraine with or without aura. The change from baseline in monthly migraine days was calculated as the average number of migraine days per month during the last 3 months, corresponding to months 4, 5, and 6 of the double-blind treatment phase.
5. Statistical Analysis Populations
The statistical analyses posted on ClinicalTrials.gov identify the efficacy analysis set as the population used for the primary and secondary efficacy comparisons. The registry describes this set as including participants who received at least 1 dose of study drug and had at least 1 change from baseline measurement in the relevant monthly outcome.
| Population | Role in reported analyses |
|---|---|
| Efficacy analysis set | Used for the primary and secondary efficacy analyses reported in the ClinicalTrials.gov record. |
| Safety population | Serious adverse events are reported by arm using affected participants and participants at risk for the double-blind and active treatment phases. |
6. Statistical Methodology
Generalized Linear Mixed Model
The primary endpoint and several continuous secondary endpoints were analyzed using a generalized linear mixed model. The model incorporated treatment, visit, treatment-by-visit interaction, stratification factors, and baseline value as covariates.
The reported stratification factors were region and prior/current treatment with migraine prophylactic medication.
The longitudinal structure is important because the endpoint is measured over visits rather than represented only by a single unadjusted post-treatment value. The treatment-by-visit interaction permits the treatment difference to vary across visits, while the baseline value adjusts the comparison for the corresponding starting measurement.
Cochran-Mantel-Haenszel test
The binary secondary endpoint measuring the percentage of participants with at least a 50% reduction from baseline in monthly migraine days was analyzed using a Cochran-Mantel-Haenszel test. The analysis was stratified by the randomization stratification factors: region and prior/current treatment with migraine prophylactic medication.
The reported odds ratios are stratified treatment-effect measures. They are not risk ratios, probability differences, or the percentage-point difference in response rates.
Least-squares mean difference
The continuous endpoint analyses report a LS Mean Difference. This is the model-adjusted difference between treatment groups after accounting for the covariates and longitudinal structure specified in the generalized linear mixed model.
Covariate adjustment
Baseline value was included as a covariate for the reported continuous analyses. Covariate adjustment can improve precision when the covariate is prognostic and is prespecified as part of the analysis model. It does not turn an observational comparison into a randomized trial; here, randomization supplies the underlying treatment-comparison framework.
Stratified analysis
Both major analysis families used stratification information. The mixed-effects models included region and prior/current treatment with migraine prophylactic medication, while the Cochran-Mantel-Haenszel analyses were explicitly stratified by these randomization factors.
7. Primary Results: Monthly Migraine Days
The registry reports two formal primary endpoint analyses, comparing placebo with each erenumab dose. Both analyses used a generalized linear mixed model and report least-squares mean differences with two-sided 95% confidence intervals.
Erenumab 70 mg QM vs Placebo
LS Mean Difference in monthly migraine days
95% CI: -1.88 to -0.92 · P < 0.001
Outcome unit: migraine days / month
The reported LS mean difference of -1.40 means that the model-adjusted mean change in monthly migraine days favored erenumab 70 mg by 1.40 migraine days per month relative to placebo, under the specified analysis model and assessment period.
The negative sign identifies the direction of the treatment contrast: fewer migraine days in the erenumab comparison relative to placebo. It does not mean that every participant experienced exactly 1.40 fewer migraine days, nor does it describe the response of an individual patient.
The 95% confidence interval extends from -1.88 to -0.92. This interval describes uncertainty around the estimated treatment difference under the model and sampling framework. It is not an interval containing 95% of individual treatment effects.
The reported P < 0.001 addresses the statistical evidence against the relevant null hypothesis under the prespecified testing framework. A p-value does not measure the size or clinical importance of the treatment effect; the estimate and confidence interval provide the information about magnitude and precision.
The analysis was conducted in the efficacy analysis set and incorporated treatment, visit, treatment-by-visit interaction, region, prior/current treatment with migraine prophylactic medication, and baseline value. Interpretation therefore depends on that model specification and the handling of repeated observations.
Erenumab 140 mg QM vs Placebo
LS Mean Difference in monthly migraine days
95% CI: -2.33 to -1.37 · P < 0.001
Outcome unit: migraine days / month
The reported LS mean difference of -1.85 means that the model-adjusted mean change in monthly migraine days favored erenumab 140 mg by 1.85 migraine days per month relative to placebo, under the specified analysis model and assessment period.
Again, this is a population-level adjusted comparison, not a statement that every participant experienced the same reduction. It also should not be interpreted as a percentage reduction unless a corresponding baseline and response-scale calculation is explicitly reported.
The 95% confidence interval is -2.33 to -1.37. Its relatively narrow span around the point estimate indicates that the model-based estimate is more precisely localized within that interval than an estimate with a much wider interval would be.
The reported P < 0.001 indicates strong statistical evidence under the specified hypothesis test, but it is not a measure of effect size. The effect size is represented by the LS mean difference, while the confidence interval describes its statistical precision.
The analysis used the same generalized linear mixed-model framework, including treatment, visit, treatment-by-visit interaction, region, prior/current treatment with migraine prophylactic medication, and baseline value. As with any model-based longitudinal analysis, interpretation is conditional on the adequacy of that model and the handling of incomplete observations.
Primary endpoint comparison
| Comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Erenumab 70 mg QM vs placebo | LS Mean Difference | -1.40 | -1.88 to -0.92 | < 0.001 |
| Erenumab 140 mg QM vs placebo | LS Mean Difference | -1.85 | -2.33 to -1.37 | < 0.001 |
8. Secondary Results: 50% Reduction in Monthly Migraine Days
The first reported binary secondary endpoint was the percentage of participants with at least a 50% reduction from baseline in monthly migraine days during the last 3 months of the double-blind treatment phase. The registry reports odds ratios from Cochran-Mantel-Haenszel analyses.
Erenumab 70 mg QM vs Placebo
Odds ratio for at least a 50% reduction
95% CI: 1.52 to 2.98 · P < 0.001
An odds ratio of 2.13 means that the estimated odds of achieving at least a 50% reduction were 2.13 times the corresponding odds in the placebo group under the stratified analysis. Odds are not probabilities, so this should not be read as "2.13 times as many participants responded."
Erenumab 140 mg QM vs Placebo
Odds ratio for at least a 50% reduction
95% CI: 2.01 to 3.94 · P < 0.001
An odds ratio of 2.81 indicates higher estimated odds of achieving the specified response in the erenumab 140 mg group relative to placebo. The 95% confidence interval of 2.01 to 3.94 quantifies uncertainty around the estimated odds ratio.
| Comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Erenumab 70 mg QM vs placebo | Odds Ratio | 2.13 | 1.52 to 2.98 | < 0.001 |
| Erenumab 140 mg QM vs placebo | Odds Ratio | 2.81 | 2.01 to 3.94 | < 0.001 |
9. Secondary Results: Acute Migraine-Specific Medication Treatment Days
The registry reports change from baseline in monthly acute migraine-specific medication treatment days during the last 3 months of the double-blind treatment period. Both erenumab doses were compared with placebo using generalized linear mixed models.
| Comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Erenumab 70 mg QM vs placebo | LS Mean Difference | -0.94 | -1.23 to -0.64 | < 0.001 |
| Erenumab 140 mg QM vs placebo | LS Mean Difference | -1.42 | -1.71 to -1.12 | < 0.001 |
The negative LS mean differences indicate fewer monthly acute migraine-specific medication treatment days in the erenumab groups relative to placebo under the specified model. The estimated difference was -0.94 days/month for 70 mg and -1.42 days/month for 140 mg.
The confidence intervals provide uncertainty around each model-adjusted estimate. The reported p-values provide evidence for the corresponding statistical tests but do not quantify the magnitude of the treatment effect.
10. Secondary Results: MPFID Physical Impairment
The registry reports a secondary endpoint measuring change from baseline in the mean monthly average Physical Impairment Domain Score measured by MPFID during the last 3 months of the double-blind treatment phase.
| Comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Erenumab 140 mg QM vs placebo | LS Mean Difference | -2.43 | -3.51 to -1.35 | < 0.001 |
| Erenumab 70 mg QM vs placebo | LS Mean Difference | -1.86 | -2.95 to -0.77 | < 0.001 |
The reported negative estimates represent lower model-adjusted MPFID average physical impairment scores in the erenumab groups relative to placebo. The outcome unit is reported as units on a scale; the ClinicalTrials.gov record does not provide a separate minimally important difference or other threshold for translating these estimates into a categorical clinical-response claim.
11. Secondary Results: MPFID Impact on Everyday Activities
The second MPFID endpoint measured change from baseline in the mean monthly average Impact on Everyday Activities Score during the last 3 months of the double-blind treatment phase.
| Comparison | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Erenumab 140 mg QM vs placebo | LS Mean Difference | -2.57 | -3.62 to -1.51 | < 0.001 |
| Erenumab 70 mg QM vs placebo | LS Mean Difference | -2.22 | -3.28 to -1.16 | < 0.001 |
As with the physical-impairment endpoint, the negative estimates represent lower model-adjusted scores in the erenumab groups relative to placebo. The reported confidence intervals describe uncertainty around the corresponding LS mean differences.
12. Multiplicity and Hierarchical Testing
Multiplicity is a central feature of the STRIVE statistical analysis because the trial evaluated the same primary endpoint against placebo at two erenumab doses and then proceeded through multiple secondary endpoints.
| Testing feature | Registry-supported description |
|---|---|
| Primary endpoint | The primary endpoint was tested independently for each erenumab dose. |
| 70 mg primary alpha | Alpha level of 0.04. |
| 140 mg primary alpha | Alpha level of 0.01. |
| Overall type I error | The registry wording states that these levels were used to maintain the type 1 error rate at an alpha level of 0.05. |
| First-tier secondary testing | If the primary endpoint was statistically significant for the relevant erenumab group, the first two secondary endpoints were tested using the Hochberg method. |
| Secondary alpha levels | The first-tier secondary endpoints were tested at an alpha level of 0.04 for the 70 mg group and 0.01 for the 140 mg group, according to the registry analyses. |
| MPFID hierarchy | The analyses posted on ClinicalTrials.gov describe additional Hochberg-based testing rules for the remaining MPFID secondary endpoints depending on significance of earlier endpoints. |
Why the hierarchy matters
If several hypotheses are tested without accounting for their multiplicity, the probability of obtaining at least one apparently positive result by chance can exceed the nominal type I error rate. STRIVE used an explicit testing hierarchy rather than treating every reported p-value as an independent confirmatory result.
The practical consequence is important: a p-value should be interpreted in the context of which hypothesis was being tested, at what stage of the hierarchy, and at which alpha level. A value below 0.05 is not, by itself, enough to reconstruct the trial's formal multiplicity decision process.
13. Missing Data and Multiple Imputation
The registry analysis metadata identify multiple imputation / missing data as a concept associated with the binary 50%-reduction endpoint. The continuous endpoint analyses were conducted in efficacy analysis sets defined by receipt of at least 1 dose and availability of at least 1 change-from-baseline measurement.
Why missing data matter
When participants do not contribute an observed endpoint, the statistical analysis must determine how that missing information enters the treatment comparison.
Why imputation is consequential
Multiple imputation replaces missing values with multiple plausible datasets, combines the resulting estimates, and carries uncertainty from the imputation process into the final inference.
The ClinicalTrials.gov record does not provide the full imputation model, assumptions, number of imputations, or sensitivity-analysis specifications. Those details should therefore not be inferred from the presence of the "multiple imputation / missing data" analysis concept alone.
14. Statistical Methods Explained
Why was a mixed-effects model used for monthly migraine days?
The primary outcome is longitudinal: the analysis concerns the last 3 months of a 24-week double-blind treatment phase and includes visit as a model component. A mixed-effects framework can account for repeated measurements and model treatment, time, treatment-by-time interaction, baseline value, and stratification factors simultaneously.
What does an LS mean difference of -1.40 mean?
It represents the model-adjusted difference between the erenumab 70 mg and placebo groups in change from baseline in mean monthly migraine days. A negative value indicates fewer migraine days in the erenumab group relative to placebo. It is not a percentage reduction and is not an individual-level prediction.
Why was baseline value included as a covariate?
Baseline migraine burden is directly relevant to the subsequent outcome. Including baseline value in the model allows the treatment comparison to be adjusted for the starting measurement rather than relying only on an unadjusted comparison of follow-up values.
Why use the Cochran-Mantel-Haenszel test?
The 50%-reduction endpoint is binary, while the trial also used randomization stratification factors. A Cochran-Mantel-Haenszel analysis provides a way to compare treatment groups across strata while producing a stratified association measure such as an odds ratio.
What does an odds ratio of 2.81 mean?
An odds ratio of 2.81 means that the estimated odds of achieving at least a 50% reduction in monthly migraine days were 2.81 times the corresponding odds under placebo in the specified stratified comparison. Odds are different from probabilities, so the estimate cannot be directly translated into a percentage of responders without the underlying response probabilities.
Why does multiplicity change how p-values are read?
STRIVE did not treat all hypotheses as independent tests at the same alpha level. The primary endpoint had separate alpha allocations for the two erenumab doses, and secondary testing followed a gate-keeping and Hochberg framework. Therefore, the formal evidentiary meaning of each p-value depends on its place in that hierarchy.
What does a 95% confidence interval tell us?
For example, the 70 mg primary LS mean difference was -1.40 with a 95% CI of -1.88 to -0.92. The interval describes statistical uncertainty around the model-based estimate under the analysis framework. It does not mean that 95% of individual participants had treatment effects inside that range.
15. Interpreting the Two Primary Treatment Effects
| Feature | Erenumab 70 mg QM | Erenumab 140 mg QM |
|---|---|---|
| LS Mean Difference | -1.40 | -1.85 |
| 95% CI | -1.88 to -0.92 | -2.33 to -1.37 |
| P-value | < 0.001 | < 0.001 |
| Primary alpha level | 0.04 | 0.01 |
| Analysis | Generalized Linear Mixed Model | Generalized Linear Mixed Model |
The point estimates differ by 0.45 migraine days per month, with the 140 mg comparison showing the more negative estimate. That arithmetic difference is a simple descriptive comparison of the two reported point estimates; it is not a formal statistical test comparing 70 mg directly with 140 mg. The ClinicalTrials.gov record does not report a direct dose-versus-dose inferential comparison.
The placebo comparisons do not, by themselves, establish that 140 mg is statistically superior to 70 mg. That would require a prespecified and appropriately analyzed direct comparison between the two erenumab doses. Similarly, the p-values do not measure the probability that either dose "works" for an individual participant.
16. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm and treatment phase. These are counts of affected participants among those at risk; they are not accompanied in the ClinicalTrials.gov record by formal between-arm hypothesis tests.
Double-blind Treatment Phase
| Arm | Serious adverse events | Participants at risk |
|---|---|---|
| Placebo | 7 | 319 |
| Erenumab 70 mg | 8 | 314 |
| Erenumab 1 | 8 | 319 |
Active Treatment Phase
| Arm | Serious adverse events | Participants at risk |
|---|---|---|
| Erenumab 70 mg | 14 | 421 |
| Erenumab 140 mg | 14 | 424 |
The ClinicalTrials.gov record does not provide formal confidence intervals or p-values for these serious-adverse-event comparisons. For a binary safety endpoint, a risk difference, risk ratio, odds ratio, or exact/stratified test could be considered depending on the prespecified statistical analysis plan, but those analyses should not be reconstructed from the reported counts and presented as if they were registry-reported inferential results.
17. Statistical Interpretation of the Secondary Endpoints
Binary response
The 50%-reduction endpoint converts a continuous clinical outcome into a responder/non-responder classification. Its odds ratio answers a different question from the LS mean difference in monthly migraine days.
Medication-use days
The acute migraine-specific medication endpoint is continuous and longitudinal, so its LS mean difference comes from the same mixed-model family as the primary endpoint.
Physical impairment
The MPFID physical impairment endpoint quantifies a change on a scale. Its negative LS mean difference indicates lower adjusted scores relative to placebo, but clinical importance requires an appropriate interpretation of that scale.
Everyday activities
The MPFID everyday-activities endpoint similarly uses a model-adjusted change score and should be interpreted independently from the migraine-day endpoint.
18. Confidence Intervals, Effect Size, and P-values
These quantities should not be treated as interchangeable. For example, both primary comparisons have P < 0.001, but their point estimates and confidence intervals differ. The p-value alone cannot tell a reader whether the effect is -1.40 or -1.85 migraine days per month, nor can it communicate the width of the uncertainty interval.
Conversely, a confidence interval is not a measure of clinical importance. Clinical interpretation requires understanding the endpoint's scale and context, while statistical precision is represented by the interval around the estimate.
19. Longitudinal Structure of the Analysis
The primary and several secondary continuous outcomes were evaluated over repeated visits. The generalized linear mixed model included both visit and a treatment-by-visit interaction.
Visit effect
Including visit allows the modeled outcome to vary over time rather than forcing every observation into a single post-treatment average.
Treatment-by-visit interaction
This term permits the treatment difference to change across visits, rather than assuming one identical treatment difference at every measurement time.
The primary estimand reported by the registry, however, is specifically the change from baseline in mean monthly migraine days to the last 3 months of the double-blind treatment period. The model's longitudinal structure therefore supports the estimation of the endpoint while accounting for its repeated-measure nature.
20. What the Odds Ratio Does — and Does Not — Mean
An odds ratio of 2.13 for erenumab 70 mg versus placebo means the estimated odds of achieving at least a 50% reduction in monthly migraine days were 2.13 times the placebo odds under the specified stratified analysis.
It does not mean that 2.13 times as many participants responded, that the probability of response increased by 113%, or that an individual participant was 2.13 times as likely to benefit in a literal individual-level sense.
An odds ratio of 2.81 means that the estimated odds of the specified response were 2.81 times those under placebo in the reported 140 mg comparison. The 95% CI of 2.01 to 3.94 represents uncertainty around that estimated odds ratio.
Odds ratios are mathematically related to probabilities but are not themselves probability differences. Without the underlying response proportions, the odds ratios cannot be converted into an absolute percentage-point treatment effect from the ClinicalTrials.gov record alone.
21. What the Mixed-Model Estimate Does — and Does Not — Mean
The LS mean difference of -1.85 for erenumab 140 mg represents a model-adjusted treatment contrast for change from baseline in mean monthly migraine days over the specified analysis period.
It does not mean that each participant experienced 1.85 fewer migraine days per month, that the treatment reduced migraine days by a fixed amount for every person, or that the estimate represents an unadjusted arithmetic difference between two simple sample means.
The 95% CI of -2.33 to -1.37 gives the uncertainty interval around the model-based treatment estimate. A narrower interval generally represents greater statistical precision than a wider interval, although precision should always be considered together with the endpoint scale and model assumptions.
22. Trial Timeline
Study start
The ClinicalTrials.gov record identifies 2015-07-17 as the study start date.
Primary completion
The ClinicalTrials.gov record identifies 2016-09-05 as the primary completion date.
Primary efficacy assessment window
The primary endpoint used the last 3 months, months 4, 5, and 6, following a 4-week baseline phase.
Formal statistical analyses available
The ClinicalTrials.gov record reports 5 outcome measures and 10 statistical analyses, including 2 primary-endpoint analyses with estimates and confidence intervals.
23. Important Limitations and Interpretation Issues
- Analysis population: the primary and secondary efficacy analyses were conducted in the specified efficacy analysis set rather than simply describing every enrolled participant.
- Model dependence: the continuous endpoint estimates depend on the generalized linear mixed model and its treatment, visit, treatment-by-visit, stratification, and baseline covariates.
- Missing data: multiple imputation / missing data is identified as an analysis concept for the binary 50%-reduction endpoint, but the ClinicalTrials.gov record does not provide the complete imputation specification.
- Multiplicity: the two primary dose comparisons used different alpha levels, and the secondary endpoints were subject to a hierarchical Hochberg strategy. Individual p-values therefore require their testing context.
- No direct dose comparison: the reported 70 mg and 140 mg placebo comparisons do not constitute a formal statistical comparison between the two erenumab doses.
- Odds ratio interpretation: the binary secondary endpoints are reported as odds ratios rather than risk ratios or absolute risk differences.
- Safety inference: serious adverse-event counts are reported by arm, but the ClinicalTrials.gov record does not include formal comparative confidence intervals or p-values for these safety outcomes.
- Registry completeness: several potentially useful descriptive elements, such as full baseline characteristics, detailed missing-data assumptions, and a complete statistical analysis plan, are not contained in the ClinicalTrials.gov record and therefore are not reconstructed here.
- Endpoint scale: the ClinicalTrials.gov record provides estimates for MPFID outcomes in scale units but do not provide a minimally important difference with which to translate those estimates into a categorical clinical-importance statement.
24. Why This Trial Matters Statistically
STRIVE is a useful teaching case because its registry-posted analyses combine several core methods encountered in modern randomized clinical trials: randomized parallel-group design, quadruple masking, longitudinal mixed-effects modeling, covariate adjustment, stratified categorical analysis, odds ratios, missing-data methodology, and explicit multiplicity control.
| Concept | How it appears in STRIVE |
|---|---|
| Randomization | The study uses randomized allocation in a parallel-group phase 3 design. |
| Blinding | The registry identifies the study as quadruple-masked. |
| Mixed-effects model | Used for the primary continuous endpoint and several continuous secondary endpoints. |
| Covariate adjustment | Models included baseline value and the specified stratification factors. |
| Longitudinal analysis | Visit and treatment-by-visit interaction were included in the generalized linear mixed models. |
| Cochran-Mantel-Haenszel test | Used for the binary 50%-reduction endpoint with stratification. |
| Odds ratio | Used as the effect measure for the binary responder endpoint. |
| Confidence interval | 95% two-sided intervals accompany the reported primary and secondary effect estimates. |
| P-values | Reported for primary and secondary comparisons, with interpretation governed by the multiplicity framework. |
| Multiple imputation / missing data | Identified in the registry analysis metadata for the binary responder analysis. |
| Multiplicity | Primary alpha levels differed by erenumab dose, with subsequent Hochberg-based secondary testing. |
| Stratified analysis | Region and prior/current treatment with migraine prophylactic medication were used as stratification factors. |
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Statistical Calculators
27. Sources
- ClinicalTrials.gov: NCT02456740 — STRIVE.
- PubMed: PMID 35978286.
- PubMed: PMID 34928306.
- PubMed: PMID 34301173.
- PubMed: PMID 33939497.
- PubMed: PMID 32851644.
Continue through the Clinical Biostats statistical pathway
Explore tutorials and statistical calculators related to randomized trials, longitudinal models, categorical analysis, confidence intervals, odds ratios, and hypothesis testing.
28. Record Summary
STRIVE provides a compact example of how a randomized phase 3 trial can combine different statistical methods for different endpoint types. The primary continuous endpoint used a generalized linear mixed model incorporating treatment, visit, treatment-by-visit interaction, region, prior/current treatment with migraine prophylactic medication, and baseline value. The binary 50%-reduction endpoint used a stratified Cochran-Mantel-Haenszel analysis with odds ratios. Multiple imputation / missing data was identified as an analysis concept, and the testing hierarchy explicitly addressed multiplicity across the two erenumab dose comparisons and subsequent secondary endpoints.
The reported primary LS mean differences were -1.40 for erenumab 70 mg QM versus placebo and -1.85 for erenumab 140 mg QM versus placebo, both with two-sided 95% confidence intervals excluding zero and P < 0.001. The corresponding binary responder analyses reported odds ratios of 2.13 and 2.81, respectively, again with two-sided 95% confidence intervals and P < 0.001. These estimates should be interpreted in the context of the efficacy analysis set, the specified models and strata, and the prespecified multiplicity framework.