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
SAMURAI was a randomized, parallel, quadruple-masked phase 3 trial evaluating two lasmiditan doses against placebo in participants with acute migraine. The registry reports two primary binary endpoints measured 2 hours after dosing: headache pain freedom and most bothersome symptom freedom.
| Feature | SAMURAI |
|---|---|
| Phase | Phase 3 |
| Condition | Acute Migraine |
| Design | Randomized, parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 2231 |
| Interventions | Lasmiditan 100 mg; lasmiditan 200 mg; placebo matching lasmiditan doses |
| Primary endpoints | Percentage of participants who are headache pain free; percentage who are most bothersome symptom free |
| Hypothesis type | Superiority |
| Primary endpoint type | Binary |
| Statistical methods posted | Logistic regression and linear regression |
| Primary effect measure | Odds ratio |
| ClinicalTrials.gov | NCT02439320 |
| Lead sponsor | Eli Lilly and Company |
| Sponsor type | Industry |
2. Clinical Question
The central question was whether lasmiditan improved acute migraine outcomes compared with placebo when assessed 2 hours after dosing. The registry evaluated two primary binary outcomes: complete absence of headache pain and absence of the participant's most bothersome symptom.
Population
Participants with acute migraine enrolled in the phase 3 SAMURAI trial.
Intervention
Lasmiditan 100 mg or lasmiditan 200 mg.
Comparator
Placebo matching the lasmiditan doses.
Primary question
Do either lasmiditan dose groups have higher odds of being headache pain free or MBS free 2 hours after dosing than placebo?
3. Trial Design
Lasmiditan 100 mg
- Active lasmiditan treatment
- Compared directly with placebo in the posted analyses
- Primary outcomes assessed at 2 hours post dose
Lasmiditan 200 mg
- Active lasmiditan treatment
- Compared directly with placebo in the posted analyses
- Primary outcomes assessed at 2 hours post dose
Placebo
- Placebo matching lasmiditan doses
- Reference group for the posted comparisons
- Used in the randomized parallel comparison
The study began in April 2015 and reached primary completion in July 2016. The registry classifies the study as completed.
4. Endpoints
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Percentage of Participants Who Are Headache Pain Free | The percentage of participants defined as mild, moderate, or severe headache pain becoming none. | 2 hours post dose | Binary |
| Percentage of Participants Who Are Most Bothersome Symptom (MBS) Free | The percentage of participants defined as the associated symptom present and identified as MBS (nausea, photophobia, or phonophobia) prior to dosing being absent. | 2 hours post dose | Binary |
Primary endpoint structure
Both primary endpoints are binary: each participant either meets the endpoint definition at the specified time point or does not. This structure naturally supports a comparison of the odds of success between each active treatment and placebo.
The registry reports two primary endpoints and four posted primary-analyses comparisons: each primary endpoint was analyzed for lasmiditan 100 mg versus placebo and for lasmiditan 200 mg versus placebo.
5. Statistical Methodology
Logistic regression for binary outcomes
The primary analyses were posted as logistic regression analyses. Logistic regression models the probability of a binary outcome as a function of treatment and any other model terms included in the analysis. The resulting treatment contrast was expressed as an odds ratio.
For a treatment indicator, exponentiating the treatment coefficient gives an odds ratio: OR = eβ1.
An odds ratio above 1 means that the treatment group has higher estimated odds of meeting the binary endpoint than the comparator. An odds ratio below 1 means lower estimated odds. An odds ratio of 1 corresponds to equal odds.
Confidence intervals
Each posted primary analysis reports a two-sided 95% confidence interval. The confidence interval provides a measure of precision around the estimated odds ratio. For example, an odds ratio of 2.2 with a 95% CI of 1.6–3.0 is more informative than the point estimate alone because it describes the range of values compatible with the statistical estimation framework used for the analysis.
P-values
The primary analyses use superiority hypotheses. The reported p-values quantify evidence against the corresponding null hypothesis within the statistical testing framework. They do not measure the size of the treatment effect and do not indicate the probability that the null hypothesis is true.
Posted linear regression analyses
The registry also identifies linear regression among the posted methods. Two secondary analyses for the percentage of participants who were nausea free are labeled linear regression in the registry while reporting odds ratios as their effect measure. That combination should be read as the registry's recorded analysis metadata rather than as evidence that an odds ratio is the conventional direct output of an ordinary linear regression model.
Analysis populations
The primary headache pain-free and MBS-free analyses used all randomized participants who received at least one dose of study drug and had evaluable data for the respective endpoint. This distinction matters: the analysis population is narrower than simply "all enrolled participants," and the exact evaluability requirement differs between the two primary endpoints.
| Endpoint | Analysis population |
|---|---|
| Headache pain free | All randomized participants who received at least one dose of study drug and had evaluable headache pain free data. |
| MBS free | All randomized participants who received at least one dose of study drug and had evaluable MBS data. |
6. Primary Results: Headache Pain Free
The primary endpoint was the percentage of participants whose headache pain became none 2 hours after dosing. The registry posted separate superiority comparisons for each lasmiditan dose versus placebo.
Lasmiditan 100 mg vs placebo
Odds ratio for headache pain freedom
95% CI: 1.6–3.0 · P < 0.001
Logistic regression; two-sided 95% confidence interval.
An odds ratio of 2.2 means that the estimated odds of being headache pain free at 2 hours were 2.2 times the odds in the placebo comparison group under the posted logistic regression analysis.
What the estimate means: The estimated odds of headache pain freedom were higher with lasmiditan 100 mg than with placebo, with an odds ratio of 2.2.
What it does not mean: An OR of 2.2 does not mean that 2.2 times as many participants became pain free, nor does it mean that the absolute probability of success increased by 120 percentage points. Odds and probabilities are related but are not interchangeable.
Precision: The 95% CI of 1.6–3.0 indicates that the estimated odds ratio is subject to statistical uncertainty, while the entire interval remains above 1.
P-value: The p-value of <0.001 indicates strong evidence against the corresponding superiority null hypothesis under the reported analysis. It does not quantify the magnitude or clinical importance of the effect.
Analysis population: The estimate applies to randomized participants who received at least one dose and had evaluable headache pain-free data, not automatically to every enrolled participant.
Lasmiditan 200 mg vs placebo
Odds ratio for headache pain freedom
95% CI: 2.0–3.6 · P < 0.001
Logistic regression; two-sided 95% confidence interval.
The 200 mg comparison produced an estimated odds ratio of 2.6. In odds terms, the estimated odds of being headache pain free at 2 hours were 2.6 times those in the placebo group under the posted analysis.
What the estimate means: The estimated odds of headache pain freedom were higher in the 200 mg lasmiditan group than in the placebo group.
What it does not mean: The odds ratio should not be translated directly into a percentage-point increase in the endpoint rate without knowing the underlying baseline probability. It also does not mean that every participant experienced the same treatment effect.
Precision: The 95% CI of 2.0–3.6 indicates uncertainty around the point estimate while remaining above the null value of 1.
P-value: The p-value of <0.001 provides evidence against the superiority null hypothesis under the reported model. It is not a measure of effect size.
Interpretive caution: The endpoint is measured at a single prespecified time point, 2 hours post dose. The result therefore describes that assessment rather than the entire subsequent migraine course.
7. Primary Results: Most Bothersome Symptom Free
The second primary endpoint assessed whether the associated symptom identified as the most bothersome symptom before dosing was absent at 2 hours. The registry defines the possible MBS categories as nausea, photophobia, or phonophobia.
Lasmiditan 100 mg vs placebo
Odds ratio for MBS freedom
95% CI: 1.3–2.2 · P < 0.001
Logistic regression; two-sided 95% confidence interval.
The estimated odds of being MBS free at 2 hours were 1.7 times the placebo odds in the posted 100 mg comparison.
What the estimate means: The treatment comparison corresponds to an odds ratio of 1.7 for absence of the participant's identified most bothersome symptom at 2 hours.
What it does not mean: The estimate does not indicate that 70% more participants became MBS free. Odds ratios are nonlinear transformations of probabilities and depend on the underlying event rate.
Precision: The 95% CI of 1.3–2.2 describes uncertainty around the estimated odds ratio and does not represent the range of individual participant responses.
P-value: The p-value of <0.001 is evidence against the superiority null hypothesis within the reported statistical framework, but it does not indicate the clinical magnitude of benefit.
Lasmiditan 200 mg vs placebo
Odds ratio for MBS freedom
95% CI: 1.3–2.1 · P < 0.001
Logistic regression; two-sided 95% confidence interval.
The 200 mg comparison produced an estimated odds ratio of 1.6 for MBS freedom at 2 hours.
What the estimate means: The estimated odds of being free of the identified most bothersome symptom at 2 hours were 1.6 times the odds in the placebo group.
What it does not mean: It is not a relative risk of 1.6 and should not be described as a 60% increase in the probability of being MBS free.
Precision: The 95% CI of 1.3–2.1 provides the statistical uncertainty around the estimated odds ratio.
P-value: The p-value of <0.001 provides evidence against the corresponding superiority null hypothesis. It does not establish that the magnitude of effect is clinically important by itself.
8. Primary Results in One Table
| Primary endpoint | Comparison | Method | OR | 95% CI | P-value |
|---|---|---|---|---|---|
| Headache pain free, 2 hours | 100 mg vs placebo | Logistic regression | 2.2 | 1.6–3.0 | <0.001 |
| Headache pain free, 2 hours | 200 mg vs placebo | Logistic regression | 2.6 | 2.0–3.6 | <0.001 |
| MBS free, 2 hours | 100 mg vs placebo | Logistic regression | 1.7 | 1.3–2.2 | <0.001 |
| MBS free, 2 hours | 200 mg vs placebo | Logistic regression | 1.6 | 1.3–2.1 | <0.001 |
9. Secondary Efficacy Results
The registry also reports secondary binary outcomes covering headache relief, headache recurrence, rescue medication use, nausea freedom, phonophobia freedom, and photophobia freedom.
Headache relief after first dose
| Comparison | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| 100 mg lasmiditan vs placebo | 2.4 | 1.8–3.1 | <0.001 |
| 200 mg lasmiditan vs placebo | 2.5 | 1.9–3.3 | <0.001 |
For both comparisons, the endpoint was assessed 2 hours after dosing. The analysis population was all randomized participants who received at least one dose and had evaluable headache relief data.
Headache recurrence
| Comparison | Time frame | Odds ratio | 95% CI | P-value |
|---|---|---|---|---|
| 100 mg lasmiditan vs placebo | From 2 hours post dose up to 48 hours | 1.7 | 1.1–2.8 | =0.029 |
| 200 mg lasmiditan vs placebo | From 2 hours post dose up to 48 hours | 2.1 | 1.3–3.4 | =0.002 |
These estimates concern recurrence during a longer observation window than the 2-hour primary endpoints. The corresponding analysis population consisted of randomized participants who received a dose and had postdose headache severity or symptom assessments.
The recurrence analysis illustrates why the direction of an odds ratio must be interpreted using the endpoint definition. An odds ratio above 1 means higher estimated odds of the coded recurrence outcome in the lasmiditan group relative to placebo. It should not automatically be described as a favorable treatment effect simply because it is above 1.
This is a useful distinction between statistical direction and clinical desirability: the meaning of an odds ratio depends on whether the outcome itself is desirable or undesirable.
Rescue medication use
| Comparison | Time frame | Odds ratio | 95% CI | P-value |
|---|---|---|---|---|
| 100 mg vs placebo | 2 hours post dose | 0.4 | 0.3–0.5 | <0.001 |
| 200 mg vs placebo | 2 hours post dose | 0.3 | 0.2–0.4 | <0.001 |
| 100 mg vs placebo | Anytime between 2–24 hours post dose | 0.7 | 0.5–1.1 | 0.120 |
| 200 mg vs placebo | Anytime between 2–24 hours post dose | 0.6 | 0.4–1.0 | 0.035 |
The direction is different from the recurrence endpoint because the outcome is rescue medication use. An odds ratio below 1 represents lower estimated odds of rescue medication use in the lasmiditan group relative to placebo.
Nausea freedom
| Comparison | Time frame | Posted method | Odds ratio | 95% CI | P-value |
|---|---|---|---|---|---|
| 100 mg vs placebo | 2 hours post dose | Linear regression | 1.1 | 0.9–1.4 | 0.386 |
| 200 mg vs placebo | 2 hours post dose | Linear regression | 1.1 | 0.9–1.4 | 0.470 |
The registry reports odds ratios of 1.1 for both comparisons, with confidence intervals spanning 1. The recorded p-values are 0.386 and 0.470.
Phonophobia freedom
| Comparison | Time frame | Odds ratio | 95% CI | P-value |
|---|---|---|---|---|
| 100 mg lasmiditan | 2 hours post dose | 1.4 | 1.1–1.8 | 0.006 |
| 200 mg lasmiditan | 2 hours post dose | 1.3 | 1.0–1.6 | 0.037 |
The posted analysis population was all randomized participants who used at least 1 dose of study drug and had any post-dose headache severity or symptom assessments.
Photophobia freedom
| Comparison | Time frame | Odds ratio | 95% CI | P-value |
|---|---|---|---|---|
| 100 mg lasmiditan | 2 hours post dose | 1.9 | 1.5–2.4 | <0.001 |
| 200 mg lasmiditan | 2 hours post dose | 1.7 | 1.4–02.2 | <0.001 |
The registry reports the upper confidence limit for the 200 mg comparison as 02.2; it is reproduced exactly as recorded.
10. Secondary Results: Statistical Perspective
Endpoint direction matters
An OR above 1 can represent either a favorable or unfavorable result depending on whether the endpoint is desirable, such as symptom freedom, or undesirable, such as recurrence.
Magnitude matters
A p-value alone does not tell the reader whether an observed association is large or small. The odds ratio and its confidence interval provide the effect estimate and its precision.
Time frame matters
Two-hour endpoints and endpoints measured from 2 to 48 hours answer different questions and should not be treated as interchangeable.
Analysis population matters
The registry uses different evaluability definitions for different outcomes. The population attached to an estimate is part of the result.
The secondary analyses also illustrate why a long list of p-values requires careful interpretation. The registry reports multiple secondary endpoints and comparisons, but the registry-reported statistical record does not describe a multiplicity-adjustment strategy. The nominal p-values should therefore be read as the reported values for their individual analyses rather than automatically treated as a single familywise-error-controlled collection of confirmatory tests.
11. Safety
The registry reports serious adverse events by treatment arm. These are presented as affected participants over participants at risk.
| Arm | Serious adverse events | At risk |
|---|---|---|
| 100 mg Lasmiditan | 0 | 630 |
| 200 mg Lasmiditan | 2 | 609 |
| Placebo | 1 | 617 |
The serious-adverse-event counts are small relative to the numbers at risk. This makes the safety comparison particularly sensitive to the fact that a small number of events can produce substantial relative changes in an estimated rate. The ClinicalTrials.gov record does not include a formal statistical comparison of serious adverse events.
12. Statistical Methods Explained
Why was logistic regression appropriate for the primary endpoints?
The primary endpoints are binary: participants either became headache pain free or did not, and either became MBS free or did not. Logistic regression is designed for this outcome structure and expresses treatment differences naturally through odds ratios.
What does an odds ratio of 2.2 mean?
An odds ratio of 2.2 means that the estimated odds of the endpoint were 2.2 times the odds in the comparator group. Odds are defined as probability divided by one minus probability. Because odds are not probabilities, an OR of 2.2 should not be described as a 120-percentage-point increase in the response rate.
Why can an odds ratio be below 1 for rescue medication?
Because the endpoint is the occurrence of rescue medication use. An OR of 0.4 means lower estimated odds of that event in the treatment group relative to placebo. The direction of the statistical estimate must always be interpreted in relation to the endpoint definition.
What does the 95% confidence interval tell us?
The confidence interval describes uncertainty around the estimated treatment effect under the statistical model and sampling framework. A narrower interval indicates greater precision than a wider interval. It does not describe the range of individual treatment effects across participants.
Why does the p-value not measure treatment effect size?
A p-value measures the degree of incompatibility between the observed data and a specified null hypothesis under the analysis framework. It depends on both the effect and the amount of information in the analysis. Effect size is conveyed by the odds ratio, while the confidence interval communicates its precision.
Why should the two-hour and 48-hour endpoints be kept separate?
The primary endpoints are fixed at 2 hours after dosing, whereas headache recurrence is assessed from 2 hours post dose up to 48 hours. They represent different clinical time windows and different questions. Combining them into a single interpretation would obscure that distinction.
What is unusual about the posted nausea-free analyses?
The registry records linear regression as the method for the two nausea-free analyses while also recording odds ratios as the effect measure. For a binary endpoint, logistic regression is the conventional regression framework for an odds ratio, so the registry's method label and effect-measure field should be read as separate pieces of the posted analysis metadata.
13. Confidence Intervals, Odds Ratios, and Clinical Meaning
Because the odds ratio compares odds rather than probabilities, the same odds ratio can correspond to different absolute probability differences depending on the comparator event rate.
This distinction is central to interpreting SAMURAI. For example, the headache pain-free ORs of 2.2 and 2.6 establish the relative comparison on the odds scale, but they do not provide the absolute percentage of participants who were pain free in either group. The ClinicalTrials.gov record reports the effect estimates, confidence intervals, and p-values, but not the corresponding group-specific endpoint percentages.
For binary endpoints, read the result in this order: endpoint definition → comparison direction → odds ratio → confidence interval → p-value → analysis population → time frame. This prevents a statistically correct number from being given an incorrect clinical meaning.
14. Superiority Testing
All posted analyses are classified as superiority hypotheses. The null hypothesis for an odds-ratio comparison is conventionally centered at 1: an odds ratio of 1 corresponds to equal odds between the two comparison groups.
| Odds ratio | Statistical interpretation |
|---|---|
| OR = 1 | Equal odds under the modeled treatment comparison. |
| OR > 1 | Higher estimated odds of the endpoint in the treatment group. |
| OR < 1 | Lower estimated odds of the endpoint in the treatment group. |
For the two primary endpoints, all four posted odds-ratio confidence intervals lie above 1, and all four reported p-values are <0.001. This describes the statistical evidence reported for those four comparisons. It does not by itself determine the clinical importance of the observed differences.
15. Multiple Comparisons and Interpretation
The trial contains more than one efficacy endpoint and more than one treatment-versus-placebo comparison. The registry reports two primary endpoints, with four corresponding primary analyses.
| Primary endpoint | 100 mg vs placebo | 200 mg vs placebo |
|---|---|---|
| Headache pain free | OR 2.2; 95% CI 1.6–3.0; P < 0.001 | OR 2.6; 95% CI 2.0–3.6; P < 0.001 |
| MBS free | OR 1.7; 95% CI 1.3–2.2; P < 0.001 | OR 1.6; 95% CI 1.3–2.1; P < 0.001 |
Multiple testing matters because each statistical test creates an opportunity for a false-positive finding. The ClinicalTrials.gov record identifies the endpoints, methods, effect estimates, confidence intervals, and p-values, but does not provide a multiplicity-adjustment procedure for these analyses. The safest interpretation is therefore to distinguish the reported nominal results from any broader claim about familywise error control.
16. Analysis Population and Missing Data
The primary headache pain-free analysis included randomized participants who received at least one dose and had evaluable headache pain-free data. The MBS analysis used the analogous population with evaluable MBS data.
Randomization is retained
Participants entered the analysis according to their randomized treatment assignment, subject to the posted requirement that they received at least one dose and had evaluable endpoint data.
Evaluability matters
The analysis does not simply equal the full enrollment count of 2231. Participants must satisfy the endpoint-specific analysis-population definition.
The ClinicalTrials.gov record does not describe a specific missing-data imputation method for the primary endpoints. Consequently, the posted odds ratios should be interpreted in the context of the stated evaluable analysis populations rather than attributed to a particular imputation strategy.
17. Blinding and Randomization as Statistical Protection
The trial used randomized allocation and quadruple masking. Randomization is fundamental because it creates the framework for comparing outcomes between treatment groups without relying solely on measured baseline characteristics to explain differences between groups.
Masking addresses a different source of bias. By keeping multiple groups involved in the trial unaware of treatment assignment, it can reduce differential expectations, assessment behavior, and other influences that might otherwise affect measured outcomes. For subjective or participant-reported migraine outcomes, the protection provided by masking is particularly relevant to interpretation of the observed binary endpoint comparisons.
Randomization primarily protects the validity of the treatment comparison by balancing measured and unmeasured factors in expectation. Masking primarily reduces the possibility that knowledge of treatment assignment influences participant behavior, outcome assessment, or other trial processes. They solve different statistical and operational problems.
18. Statistical Interpretation of the Primary Findings
Headache pain freedom
The posted odds ratios were 2.2 for 100 mg and 2.6 for 200 mg versus placebo, with both 95% confidence intervals entirely above 1 and both p-values <0.001.
MBS freedom
The posted odds ratios were 1.7 for 100 mg and 1.6 for 200 mg versus placebo, again with confidence intervals entirely above 1 and p-values <0.001.
Effect measure
The primary effect measure is the odds ratio. It describes relative odds, not absolute probability or percentage-point difference.
Time point
Both primary outcomes were measured at 2 hours post dose. The estimates therefore describe the prespecified two-hour assessment.
The pattern across the four primary comparisons is internally consistent in direction: each reported odds ratio is above 1, and every corresponding 95% confidence interval excludes 1. That statistical pattern should be kept separate from questions about absolute response rates, patient-level benefit, durability, or comparative clinical importance, none of which is established by the odds ratios alone.
19. Limitations
- Odds ratios are not risk ratios: the magnitude of an odds ratio depends on the baseline event probability and should not be translated directly into a probability ratio.
- Absolute response rates are important: the posted ClinicalTrials.gov record provide odds ratios and confidence intervals but do not provide the corresponding arm-specific percentages for the primary endpoints.
- Evaluable populations: the primary analyses require at least one dose and evaluable endpoint data, so enrollment of 2231 should not be treated as the denominator for every analysis.
- Multiple analyses: two primary endpoints and multiple secondary endpoints create a multiplicity issue. The ClinicalTrials.gov record does not specify an adjustment strategy.
- Secondary endpoints: the secondary results cover several different clinical concepts and time frames. They should not be collapsed into a single efficacy measure.
- Linear-regression metadata: the registry labels two nausea-free analyses as linear regression while recording odds ratios as their effect measure; that pairing should not be interpreted as a conventional direct linear-regression odds-ratio model without additional statistical documentation.
- Safety event counts: the serious adverse-event counts are small, limiting what can be learned from simple descriptive counts alone.
- No proportional-hazards issue for the primary analyses: the primary endpoints are binary at a fixed time point, so the central modeling issue is logistic rather than Cox proportional-hazards modeling.
20. Why This Trial Matters Statistically
SAMURAI is a useful statistical teaching example because the primary analysis is straightforward in outcome structure but rich in interpretation. The trial combines randomized treatment allocation, quadruple masking, binary efficacy endpoints, logistic regression, odds ratios, confidence intervals, superiority testing, multiple treatment comparisons, and several secondary endpoints with different time frames.
| Concept | How it appears in SAMURAI |
|---|---|
| Randomization | Participants were randomly allocated in a parallel three-arm phase 3 trial. |
| Blinding | The study used quadruple masking. |
| Binary endpoint | Both primary outcomes classify participants according to whether a specified migraine outcome is achieved. |
| Logistic regression | Posted for all four primary endpoint comparisons. |
| Odds ratio | The primary effect measure for the posted analyses. |
| Confidence interval | All primary analyses report two-sided 95% confidence intervals. |
| P-value | All four primary comparisons report P < 0.001. |
| Superiority | Primary and secondary posted analyses are classified as superiority hypotheses. |
| Multiplicity | Two primary endpoints and numerous secondary comparisons require careful interpretation of nominal p-values. |
| Analysis populations | Endpoint-specific evaluability criteria define the populations used for the primary analyses. |
21. A Worked Example: Reading OR 2.6 Correctly
Consider the posted headache pain-free comparison of 200 mg lasmiditan versus placebo.
Endpoint: headache pain free at 2 hours post dose.
The first step is to identify the outcome: headache pain freedom is a desirable binary event. The second step is to identify the comparison: 200 mg lasmiditan is compared with placebo. The third step is to interpret the effect measure: the estimated odds of headache pain freedom are 2.6 times the placebo odds.
The confidence interval then adds precision. The interval from 2.0 to 3.6 indicates uncertainty around the estimated odds ratio and remains above the null value of 1. Finally, the p-value of <0.001 indicates strong statistical evidence against the null hypothesis of no difference in odds under the posted analysis.
What cannot be obtained from the OR alone is the absolute percentage of participants who became headache pain free. That requires the underlying response probabilities or event counts. This is one reason odds ratios should be interpreted together with absolute risks whenever those quantities are available.
22. Results Summary
| Outcome | 100 mg vs placebo | 200 mg vs placebo |
|---|---|---|
| Headache pain free, 2 hours | OR 2.2; 95% CI 1.6–3.0; P < 0.001 | OR 2.6; 95% CI 2.0–3.6; P < 0.001 |
| MBS free, 2 hours | OR 1.7; 95% CI 1.3–2.2; P < 0.001 | OR 1.6; 95% CI 1.3–2.1; P < 0.001 |
| Headache relief, 2 hours | OR 2.4; 95% CI 1.8–3.1; P < 0.001 | OR 2.5; 95% CI 1.9–3.3; P < 0.001 |
| Headache recurrence, 2–48 hours | OR 1.7; 95% CI 1.1–2.8; P = 0.029 | OR 2.1; 95% CI 1.3–3.4; P = 0.002 |
| Rescue medication, 2 hours | OR 0.4; 95% CI 0.3–0.5; P < 0.001 | OR 0.3; 95% CI 0.2–0.4; P < 0.001 |
| Rescue medication, 2–24 hours | OR 0.7; 95% CI 0.5–1.1; P = 0.120 | OR 0.6; 95% CI 0.4–1.0; P = 0.035 |
| Nausea free, 2 hours | OR 1.1; 95% CI 0.9–1.4; P = 0.386 | OR 1.1; 95% CI 0.9–1.4; P = 0.470 |
| Phonophobia free, 2 hours | OR 1.4; 95% CI 1.1–1.8; P = 0.006 | OR 1.3; 95% CI 1.0–1.6; P = 0.037 |
| Photophobia free, 2 hours | OR 1.9; 95% CI 1.5–2.4; P < 0.001 | OR 1.7; 95% CI 1.4–02.2; P < 0.001 |
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
25. Sources
- ClinicalTrials.gov record: NCT02439320.
- PubMed record: PMID 36125279.
- PubMed record: PMID 35779194.
- PubMed record: PMID 35471625.
- PubMed record: PMID 35350937.
- PubMed record: PMID 34644189.
Continue through the Clinical Biostats statistical library
Connect the trial's endpoints and methods to deeper biostatistics tutorials and statistical calculators.
26. Record Summary
SAMURAI provides a clear example of statistical analysis for randomized clinical-trial endpoints that are binary at a prespecified time point. The primary analyses used logistic regression and reported odds ratios with two-sided 95% confidence intervals and superiority p-values. For headache pain freedom, the reported odds ratios were 2.2 for 100 mg lasmiditan versus placebo and 2.6 for 200 mg versus placebo. For MBS freedom, the corresponding odds ratios were 1.7 and 1.6. All four primary p-values were <0.001.
The most important statistical lesson is that the odds ratio must be interpreted together with the endpoint definition, comparator, confidence interval, p-value, analysis population, and time frame. Secondary endpoints further demonstrate how the same statistical measure can have different clinical meaning depending on whether the outcome represents symptom freedom, recurrence, or rescue medication use. The trial therefore provides a useful framework for understanding logistic regression, odds ratios, confidence intervals, superiority testing, randomization, masking, and the interpretation of multiple binary endpoints.