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
ACHIEVE I was a randomized, double-blind, parallel-group phase 3 trial evaluating the efficacy, safety, and tolerability of oral ubrogepant in the acute treatment of migraine with or without aura. The registry reports 1,672 enrolled participants across three arms.
| Feature | ACHIEVE I |
|---|---|
| Phase | Phase 3 |
| Condition | Migraine, with or without aura |
| Design | Randomized, double-blind, parallel-group |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Primary endpoints | Two binary efficacy endpoints assessed at 2 hours after the initial dose |
| Primary analysis | Logistic regression |
| Effect measure | Odds ratio |
| Hypothesis type | Superiority |
| Status | Completed |
| Start | 2016-07-22 |
| Primary completion | 2017-12-13 |
| Lead sponsor | Allergan |
2. Clinical Question
The central question was whether oral ubrogepant improved acute migraine outcomes compared with placebo, with the primary efficacy assessment focused on pain freedom and absence of the participant-identified most bothersome migraine-associated symptom at 2 hours after the initial dose.
Population
Participants with migraine, with or without aura.
Intervention
Oral ubrogepant, with analyses reported separately for 50 mg and 100 mg.
Comparator
Placebo-matching ubrogepant.
Primary question
Does ubrogepant increase the odds of achieving pain freedom and absence of the most bothersome migraine-associated symptom at 2 hours?
3. Trial Design
Ubrogepant 50 mg
- Oral ubrogepant
- Compared with placebo in the posted statistical analyses
- Primary analyses reported odds ratios for both primary endpoints
Ubrogepant 100 mg
- Oral ubrogepant
- Compared with placebo in the posted statistical analyses
- Primary analyses reported odds ratios for both primary endpoints
4. Endpoints
Primary endpoint 1: Pain freedom
Percentage of Participants With Pain Freedom at 2 Hours After Initial Dose. The registry defines pain freedom as a reduction in headache severity from moderate/severe at baseline to no pain at 2 hours after the initial dose. Participants used an electronic diary to rate headache severity from no pain to severe pain. The number analyzed was the number of participants with a non-missing postdose pain-severity assessment at or before 2 hours after the initial dose.
| Element | Registry definition |
|---|---|
| Time frame | Baseline (Predose) to 2 hours after initial dose |
| Endpoint type | Binary |
| Outcome unit | Percentage of participants |
| Statistical method | Logistic regression |
| Effect measure | Odds ratio |
| Hypothesis | Superiority |
Primary endpoint 2: Most bothersome migraine-associated symptom
Percentage of Participants With Absence of the Most Bothersome Migraine-Associated Symptom Identified at Baseline at 2 Hours After Initial Dose. The most bothersome symptom was the symptom among photophobia, phonophobia, or nausea that was present at baseline and identified by the participant as most bothersome. Participants recorded presence or absence of migraine-associated symptoms using an electronic diary. The number analyzed was the number of participants with a non-missing postdose assessment of the most bothersome migraine-associated symptom.
| Element | Registry definition |
|---|---|
| Time frame | Baseline (Predose) to 2 hours after initial dose |
| Endpoint type | Binary |
| Outcome unit | Percentage of participants |
| Statistical method | Logistic regression |
| Effect measure | Odds ratio |
| Hypothesis | Superiority |
5. Statistical Methodology
Modified intention-to-treat analysis
The primary pain-freedom analyses used a modified intention-to-treat population. This population included all randomized participants who received at least 1 dose of study drug, recorded a baseline migraine headache-severity measurement, and had at least 1 postdose migraine headache-severity or migraine-associated symptom measurement at or before the 2-hour timepoint, using last observation carried forward.
The corresponding most-bothersome-symptom analyses used an mITT population with the same general structure: randomized participants who received at least 1 dose, had a baseline migraine headache-severity measurement, and had at least 1 relevant postdose measurement at or before 2 hours, with LOCF.
Logistic regression
All posted primary analyses used logistic regression because the endpoints were binary: a participant either achieved the specified outcome by the assessment time or did not. The model estimates the relationship between treatment assignment and the odds of the binary outcome while incorporating prespecified explanatory variables.
For these analyses, the treatment coefficient is translated into an odds ratio. The registry identifies treatment group and additional explanatory variables in the fitted models.
Model covariates
For the pain-freedom analyses, the explanatory variables were treatment group, historical triptan response, medication use for migraine prevention, and baseline headache severity. For the most-bothersome-symptom analyses, the model also included the underlying symptom relevant to the endpoint.
Odds ratio as the effect measure
The registry reports odds ratios rather than risk ratios or absolute risk differences. An odds ratio compares the odds of the binary outcome between treatment groups after accounting for the variables included in the logistic model.
An OR greater than 1 indicates higher estimated odds of the specified outcome in the ubrogepant group relative to placebo. It does not directly state how many additional participants experienced the outcome.
Confidence intervals and p-values
The primary analyses report two-sided 95% confidence intervals for the odds ratios. These intervals describe statistical uncertainty around the estimated treatment effect under the analysis model. The p-value addresses the statistical evidence against the null hypothesis represented by an odds ratio of 1; it is not a measure of the magnitude or clinical importance of the effect.
Missing-data handling
The posted primary analyses identify last observation carried forward as part of the analysis population definition. Several analyses also explicitly identify multiple imputation / missing data as an analysis concept. For sustained outcomes, the registry specifies determinable cases rather than LOCF in the posted population definition.
6. Primary Results: Pain Freedom at 2 Hours
Ubrogepant 50 mg versus placebo
Odds ratio for pain freedom
95% CI: 1.25–2.66 · P = 0.0023
Superiority analysis using logistic regression in the mITT population.
The model estimated 1.83 times the odds of pain freedom at 2 hours for ubrogepant 50 mg relative to placebo, after accounting for the explanatory variables specified in the registry analysis: treatment group, historical triptan response, medication use for migraine prevention, and baseline headache severity.
The OR of 1.83 means that the estimated odds of achieving pain freedom at 2 hours were 83% higher with ubrogepant 50 mg than with placebo under the fitted logistic model. This is an odds comparison, not an 83-percentage-point increase in the proportion of participants with pain freedom.
The 95% CI of 1.25–2.66 describes uncertainty around the estimated odds ratio. Because the interval is entirely above 1, the estimated treatment effect is directionally positive throughout the reported interval. The interval does not tell us the range of individual patient responses.
The P = 0.0023 quantifies evidence against the null value of an odds ratio of 1 under the stated model and hypothesis. It does not measure effect size, probability of benefit, or clinical importance.
The analysis is specific to the mITT population and the 2-hour endpoint definition. The LOCF component and the requirement for postdose measurements are important when interpreting the result, because the analyzed population is not simply every randomized participant.
Ubrogepant 100 mg versus placebo
Odds ratio for pain freedom
95% CI: 1.41–2.95 · P = 0.0003
Superiority analysis using logistic regression in the mITT population.
The model estimated 2.04 times the odds of pain freedom at 2 hours for ubrogepant 100 mg relative to placebo, using the same principal model structure and explanatory variables.
The OR of 2.04 corresponds to approximately twice the estimated odds of pain freedom with ubrogepant 100 mg compared with placebo under the fitted model. It should not be read as saying that twice as many participants achieved pain freedom, because odds and probabilities are different quantities.
The 95% CI of 1.41–2.95 indicates the precision of the estimated odds ratio under the model. The entire interval lies above 1, so the reported estimate and its confidence interval are consistent with higher odds of pain freedom for the ubrogepant 100 mg comparison.
The P = 0.0003 is evidence against the null odds ratio of 1 under the stated superiority test. A smaller p-value does not mean that the treatment effect is larger than the 50 mg comparison; effect magnitude is described by the odds ratio and its interval, not by the p-value alone.
The analysis remains conditional on the mITT definition, the 2-hour assessment window, and the logistic-regression model. The result is therefore a model-based comparison of a specific binary endpoint rather than a universal measure of migraine benefit.
7. Primary Results: Absence of the Most Bothersome Migraine-Associated Symptom
Ubrogepant 50 mg versus placebo
Odds ratio for absence of the most bothersome symptom
95% CI: 1.27–2.28 · P = 0.0023
Superiority analysis using logistic regression in the mITT population.
For this endpoint, the registry analysis adjusted for treatment group, historical triptan response, migraine prevention medication, baseline headache severity, and the underlying symptom.
The OR of 1.70 indicates 70% higher estimated odds of absence of the participant-identified most bothersome migraine-associated symptom at 2 hours with ubrogepant 50 mg than with placebo, under the fitted model. This does not mean that 70% of participants became symptom-free or that the absolute difference between groups was 70 percentage points.
The 95% CI of 1.27–2.28 provides a measure of precision for the odds-ratio estimate. Its lower and upper limits both exceed 1, consistent with the reported superiority result.
The P = 0.0023 describes the evidence against an odds ratio of 1 under the specified statistical model. It does not measure the size of the treatment effect or the probability that the treatment is effective.
The endpoint is participant-specific because the most bothersome symptom was identified at baseline from photophobia, phonophobia, or nausea. Interpretation therefore depends on the endpoint's baseline symptom definition as well as the 2-hour assessment.
Ubrogepant 100 mg versus placebo
Odds ratio for absence of the most bothersome symptom
95% CI: 1.22–2.17 · P = 0.0023
Superiority analysis using logistic regression in the mITT population.
The 100 mg analysis used the same principal mITT structure and included treatment group, historical triptan response, migraine prevention medication, baseline headache severity, and underlying symptom as explanatory variables.
The OR of 1.63 corresponds to 63% higher estimated odds of absence of the most bothersome symptom at 2 hours with ubrogepant 100 mg compared with placebo under the fitted model. It is not a relative increase in the percentage of participants and does not provide an absolute risk difference.
The 95% CI of 1.22–2.17 indicates the uncertainty around the estimated odds ratio. Because the interval remains above 1, the reported estimate is consistent with higher odds of the specified outcome in the ubrogepant 100 mg group.
The P = 0.0023 addresses the null hypothesis of an odds ratio of 1 within the reported superiority analysis. It should not be interpreted as an effect-size metric.
As with the pain-freedom endpoint, the result is specific to the mITT analysis population, the baseline identification of the most bothersome symptom, and the 2-hour assessment framework.
8. Secondary Endpoint Results
Pain relief at 2 hours
Pain relief at 2 hours was analyzed as a binary endpoint using logistic regression in the mITT population.
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 1.69 | 1.28–2.23 | 0.0023 |
| Ubrogepant 100 mg vs placebo | 1.69 | 1.28–2.21 | 0.0023 |
The registry defines the endpoint as the percentage of participants with pain relief at 2 hours after the initial dose, measured from baseline to 2 hours. Both comparisons used treatment group, historical triptan response, medication use for migraine prevention, and baseline headache severity as explanatory variables.
Sustained pain relief from 2 to 24 hours
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 2.25 | 1.65–3.07 | 0.0023 |
| Ubrogepant 100 mg vs placebo | 2.39 | 1.77–3.24 | 0.0023 |
The endpoint was the percentage of participants with sustained pain relief from 2 to 24 hours after the initial dose. The posted analysis used logistic regression and a determinable-case mITT population. The registry identifies treatment group, historical triptan response, medication use for migraine prevention, and baseline headache severity as explanatory variables.
Sustained pain freedom from 2 to 24 hours
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 1.57 | 1.01–2.44 | 0.0577 |
| Ubrogepant 100 mg vs placebo | 1.95 | 1.28–2.97 | 0.0037 |
The registry defines sustained pain freedom as the percentage of participants with sustained pain freedom from 2 to 24 hours after the initial dose. Both analyses used logistic regression in the determinable-case mITT population.
Absence of photophobia at 2 hours
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 1.63 | 1.22–2.19 | 0.0577 |
| Ubrogepant 100 mg vs placebo | 1.81 | 1.36–2.42 | 0.0037 |
The endpoint was the percentage of participants with absence of photophobia at 2 hours after the initial dose. The logistic models included treatment group, historical triptan response, migraine prevention medication, baseline headache severity, and photophobia.
Absence of phonophobia at 2 hours
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 1.56 | 1.16–2.09 | 0.0577 |
| Ubrogepant 100 mg vs placebo | 1.47 | 1.10–1.95 | 0.0577 |
The endpoint was the percentage of participants with absence of phonophobia at 2 hours after the initial dose. The logistic models included treatment group, historical triptan response, migraine prevention medication, baseline headache severity, and phonophobia.
Absence of nausea at 2 hours
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Ubrogepant 50 mg vs placebo | 1.31 | 0.96–1.79 | 0.0962 |
| Ubrogepant 100 mg vs placebo | 1.35 | 1.00–1.83 | 0.0962 |
The endpoint was the percentage of participants with absence of nausea at 2 hours after the initial dose. The logistic models included treatment group, historical triptan response, migraine prevention medication, baseline headache severity, and nausea.
9. Results Across the Binary Endpoints
The reported results show a consistent statistical structure: the primary and secondary outcomes were binary, and the registry used logistic regression with odds ratios and two-sided 95% confidence intervals. This makes the results directly comparable at the level of statistical estimands, while the clinical meaning differs according to the specific endpoint.
The bars above are a visual representation of the reported odds-ratio estimates, not probabilities or percentages. An odds ratio of 2.04 does not mean that 204% of participants achieved the endpoint, nor does it establish a 104-percentage-point difference.
10. Safety
The registry reports serious adverse events by arm using affected participants over participants at risk. The available arm-specific safety counts are:
| Arm | Serious adverse events | Affected / at risk |
|---|---|---|
| Placebo | Serious adverse events | 0 / 485 |
| Ubrogepant 50 mg | Serious adverse events | 3 / 466 |
| Ubrogepant 100 mg | Serious adverse events | 2 / 485 |
These counts describe serious adverse events rather than the binary efficacy endpoints. They should therefore be interpreted separately from the odds ratios for pain freedom and migraine-associated symptoms.
Placebo
0 affected participants among 485 at risk were reported for serious adverse events.
Ubrogepant 50 mg
3 affected participants among 466 at risk were reported for serious adverse events.
Ubrogepant 100 mg
2 affected participants among 485 at risk were reported for serious adverse events.
Interpretive distinction
Serious adverse-event counts are safety observations and are not directly comparable with the efficacy odds ratios.
11. Statistical Methods Explained
Why was logistic regression used?
The primary endpoints were binary outcomes measured at a defined time point: pain freedom at 2 hours and absence of the most bothersome migraine-associated symptom at 2 hours. Logistic regression is designed for this outcome structure and produces an odds ratio as a treatment-effect measure.
What does an odds ratio of 1.83 mean?
An odds ratio of 1.83 means that the estimated odds of the specified outcome are 1.83 times as high in the ubrogepant 50 mg group as in the placebo group under the fitted model. It does not mean that the probability is 1.83 times as high in every setting, because the relationship between odds and probability depends on the underlying event probability.
Why is the odds ratio not a percentage-point difference?
Probability and odds use different scales. If an outcome occurs in 60% of participants, its probability is 0.60 while its odds are 0.60/0.40 = 1.50. An odds ratio compares those odds between groups. An absolute risk difference would instead subtract the two probabilities.
What does the 95% confidence interval add?
The point estimate alone gives one estimated treatment effect. The confidence interval adds information about precision. For example, the pain-freedom estimate for 50 mg is 1.83 with a 95% CI of 1.25–2.66. The interval communicates that the estimate is uncertain rather than treating 1.83 as an exact population value.
Why doesn't the p-value measure the treatment effect?
The p-value describes the compatibility of the observed data with a specified null hypothesis under the statistical model. A very small p-value can occur with a modest effect in a large dataset, while a larger p-value can occur with a potentially meaningful estimate when uncertainty is substantial. The odds ratio and its confidence interval describe the estimated effect and its precision.
What does the mITT population change?
The mITT definition means that the primary analysis does not simply include every randomized participant. Participants had to receive at least 1 dose, have the required baseline assessment, and have relevant postdose information at or before the 2-hour timepoint. This makes the exact analysis population part of the interpretation of the reported odds ratios.
What does LOCF mean here?
Last observation carried forward means that an earlier available observation can be carried forward to represent the endpoint when specified by the analysis. It is a missing-data strategy with assumptions about how an earlier observation relates to the later endpoint. Consequently, the use of LOCF is part of the statistical context of the primary analyses rather than a neutral bookkeeping step.
12. Interpreting the Primary Results Together
| Primary endpoint | 50 mg vs placebo | 100 mg vs placebo |
|---|---|---|
| Pain freedom at 2 hours | OR 1.83; 95% CI 1.25–2.66; P = 0.0023 | OR 2.04; 95% CI 1.41–2.95; P = 0.0003 |
| Absence of most bothersome symptom at 2 hours | OR 1.70; 95% CI 1.27–2.28; P = 0.0023 | OR 1.63; 95% CI 1.22–2.17; P = 0.0023 |
All four primary comparisons have odds-ratio estimates above 1. The confidence intervals for all four comparisons also lie above 1. The two primary endpoints therefore point in the same direction in the reported analyses: higher estimated odds of the prespecified favorable binary outcome for both ubrogepant doses compared with placebo.
Importantly, the four estimates should not be collapsed into a single treatment-effect number. Pain freedom and absence of the most bothersome symptom are distinct endpoints, and 50 mg and 100 mg are separate randomized treatment comparisons with their own model estimates and confidence intervals.
13. Missing Data and Analysis Population
Missing-data handling is particularly important for this trial because the primary outcomes depended on electronic-diary measurements around a specific 2-hour timepoint. The registry explicitly describes an mITT population requiring relevant postdose information and identifies LOCF in the primary analysis definitions.
14. Multiplicity and Multiple Comparisons
The registry reports two primary endpoints and two active-dose comparisons for each primary endpoint. The posted analyses therefore contain multiple statistical comparisons within the efficacy program.
The reported p-values should be read together with the trial's designation of the endpoints as primary or secondary and with the fact that the registry identifies the hypothesis type as superiority. A p-value is not, by itself, a complete description of a multiple-testing strategy.
15. What the Odds Ratio Does — and Does Not — Mean
An odds ratio above 1 indicates higher estimated odds of the specified binary outcome in the ubrogepant group relative to placebo. For example, an OR of 2.04 for pain freedom with 100 mg means the estimated odds were approximately twice as high under the fitted logistic model.
It does not mean that 204% of participants achieved pain freedom, that the probability doubled for every participant, or that every individual experienced the same treatment effect.
The 95% CI describes uncertainty around the estimated odds ratio. For the 100 mg pain-freedom comparison, the interval of 1.41–2.95 communicates considerably more information than the point estimate of 2.04 alone.
The registry describes the outcomes as percentages of participants, but the posted statistical analyses report odds ratios rather than absolute percentage-point differences. An odds ratio describes a relative comparison on the odds scale; an absolute risk difference would require the corresponding group-specific event proportions.
16. Endpoint-Specific Interpretation
Pain freedom
This endpoint requires movement from moderate/severe headache at baseline to no pain at 2 hours. It is therefore a stringent binary definition that incorporates both baseline severity and the postdose assessment.
Most bothersome symptom
This endpoint is individualized at baseline because the participant identifies which of photophobia, phonophobia, or nausea is most bothersome.
Sustained pain outcomes
The 2-to-24-hour endpoints address persistence of pain relief or pain freedom rather than only the single 2-hour assessment.
Specific symptoms
Photophobia, phonophobia, and nausea were also evaluated individually at 2 hours, allowing the symptom-specific analyses to be distinguished from the participant-selected most bothersome symptom endpoint.
17. Limitations
- mITT analysis population: primary efficacy analyses were based on a modified intention-to-treat definition requiring treatment exposure and relevant baseline and postdose measurements.
- LOCF: last observation carried forward is incorporated into the primary analysis definitions and can affect how incomplete longitudinal observations contribute to the binary endpoint.
- Odds-ratio interpretation: odds ratios are not risk ratios or absolute risk differences, so they should not be translated directly into percentage-point changes.
- Multiple comparisons: the trial contains two primary endpoints and separate comparisons for two ubrogepant doses, making the broader multiplicity framework important to interpretation.
- Time-specific outcomes: the primary efficacy conclusions concern the prespecified 2-hour assessments and should not automatically be generalized to later timepoints.
- Secondary endpoints: secondary outcomes provide additional evidence about pain and migraine-associated symptoms but have a different inferential role from the registered primary endpoints.
- Safety denominator: the serious-adverse-event counts are reported using arm-specific affected/at-risk denominators and should be interpreted separately from the efficacy models.
18. Why This Trial Matters Statistically
ACHIEVE I is a useful teaching case because it illustrates how a modern randomized trial can analyze several clinically related binary outcomes while maintaining a consistent modeling framework.
| Concept | How it appears in ACHIEVE I |
|---|---|
| Randomization | Randomized, parallel-group phase 3 design with three arms |
| Blinding | Double-blind design |
| Binary endpoints | Pain freedom, pain relief, sustained outcomes, and symptom absence |
| Logistic regression | Posted statistical method for the reported efficacy analyses |
| Odds ratio | Primary effect measure for the binary comparisons |
| Confidence interval | Two-sided 95% intervals reported for the odds ratios |
| Intention-to-treat concepts | Modified intention-to-treat population used for efficacy analyses |
| Missing data | LOCF and multiple imputation / missing-data concepts identified in the analyses |
| Superiority testing | Primary and secondary efficacy analyses identified as superiority hypotheses |
| Safety analysis | Serious adverse events reported by randomized treatment arm |
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Calculators
21. Sources
- ClinicalTrials.gov: NCT02828020 — ACHIEVE I.
- Linked publication: PubMed record — PMID 39982105.
- Linked publication: PubMed record — PMID 36125279.
- Linked publication: PubMed record — PMID 35468729.
- Linked publication: PubMed record — PMID 34874514.
- Linked publication: PubMed record — PMID 33874756.
Continue with the Clinical Biostats methods library
Explore tutorials and statistical calculators related to randomized trials, binary endpoints, regression models, effect measures, and statistical inference.
22. Record Summary
ACHIEVE I provides a clear example of statistical analysis for binary clinical-trial endpoints. The trial used a randomized, double-blind, parallel-group phase 3 design with 1,672 participants and evaluated oral ubrogepant in the acute treatment of migraine with or without aura. Its primary endpoints were pain freedom and absence of the most bothersome migraine-associated symptom at 2 hours, both analyzed using logistic regression with odds ratios and two-sided 95% confidence intervals.
The primary analyses reported odds ratios of 1.83 and 2.04 for pain freedom with ubrogepant 50 mg and 100 mg, respectively, and odds ratios of 1.70 and 1.63 for absence of the most bothersome migraine-associated symptom. The secondary analyses extended the same modeling framework to pain relief, sustained outcomes, and individual migraine-associated symptoms.
The statistical interpretation is strongest when the odds ratios are read together with their confidence intervals, p-values, analysis populations, endpoint definitions, and missing-data framework. The registry's reported serious-adverse-event counts provide a separate safety perspective and should not be conflated with the efficacy effect measures.