Ticagrelor Therapy for RefrACTORy Migraine Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Ticagrelor Therapy for RefrACTORy Migraine Study (NCT02518464), a na phase 4 study. The primary outcome is Total Number of Responders.

2. Clinical Question

This is a within-participant question rather than a randomized comparison between treatment groups. It asks whether headache frequency changed from baseline during the period when participants took ticagrelor. The registry describes the study as a prospective, open-label, single-arm pilot. Thus, participants and study personnel were not masked to the intervention, and the record does not describe an untreated or active-comparator group.

The population description is focused on migraine and right-to-left shunt, but the registry does not report detailed eligibility criteria or the number of participants with episodic versus chronic migraine. It also does not report participant-level baseline headache frequency. The clinical question is therefore clear at the level of the intended endpoint, but the record does not provide the outcome result needed to answer how many participants met that endpoint.

3. Trial Design

TRACTOR was a prospective, open-label, single-group pilot study listed as phase 4. Columbia University is named as the sponsor. The registry lists the study as completed, with a planned enrollment of 40. The recorded start was 2015-10 and primary completion was 2017-10.

All participants were assigned to the ticagrelor intervention; the registry identifies no allocation between multiple arms. The intervention description specifies a 180 mg loading dose given in the office at enrollment, followed by ticagrelor 90 mg twice per day for 28 days, with the doses taken in the morning and evening as close to 12 hours apart as possible. The design is described as single-group, and masking is listed as none.

The central comparison was between headache frequency on therapy and each participant's documented baseline. Daily headache logs were part of the described study procedures. Participants whose response met the stated more-than-50% reduction criterion could elect to continue therapy for an additional two months, described as 56 days, while continuing daily headache logs. This optional continuation does not replace the primary endpoint timeframe, which is listed as one month from baseline.

A single-group pilot can describe changes observed during the study period, but it does not make a randomized treatment-versus-control comparison. Because masking was absent and the baseline served as the comparison, the design is vulnerable to influences that a concurrent control group or masking might help address. The registry does not report the protocol's sample-size justification, a detailed statistical analysis plan, interim analyses, or how the optional continuation was handled in the primary analysis.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Total Number of Responders, measured over 1 month from baseline.

The primary endpoint is the total number of participants classified as responders, with the unit recorded as participants. A responder was defined as someone with more than 50% reduction in monthly headache days during the month of therapy compared with that participant's own baseline. The record classifies the measure as a participant count and gives its timeframe as one month from baseline. A reduction below 50% was defined as non-response; the wording does not specify how a value exactly equal to 50% would be classified.

A threshold-based endpoint makes a continuous change in headache frequency easier to summarize as a count of participants who crossed a prespecified boundary. It can be useful for communicating how many participants met a defined response criterion. However, it does not preserve the full information in the underlying change: participants just on either side of the threshold may have similar changes but receive different classifications, while participants within the responder group may have substantially different reductions.

The record does not report the number of responders, the underlying headache-day changes, or the variability of those changes. Therefore, the endpoint definition is available, but the posted record does not establish the observed response rate or show how precisely it was estimated.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Statistical methods should be interpreted in light of what the registry actually records. The listed design is prospective, open-label, and single-group, and the primary outcome is a count of participants meeting a binary responder definition. The registry lists zero analyses and does not report a detailed analysis plan. The explanations below describe general statistical concepts relevant to this design; they are not a claim that the trial used methods that are not documented.

What does a within-participant comparison estimate?

In a within-participant comparison, each participant's outcome during a later period is compared with that same participant's earlier measurement. This can account for stable characteristics of that participant because those characteristics are shared across the two measurements. Here, the registry describes comparing headache frequency during therapy with documented baseline frequency.

That pairing does not control for influences that change over time. Symptoms can vary, measurement practices can change, and other events can coincide with the treatment period. A concurrent control group can help distinguish changes occurring during treatment from changes that would have occurred over the same period without the intervention. Since this study is listed as single-group, the registry record does not provide that between-group contrast.

What does the responder threshold do?

A binary responder outcome reduces each participant's change to one of two categories according to the stated cutoff. This can make the endpoint straightforward to count and interpret against the protocol's definition. It also discards detail. The amount of change within each category is not represented by the responder count, and small differences around the cutoff can change classification. A fuller description would ordinarily include the underlying change in headache days as well as the threshold-based count, but those results are not reported in the registry record.

The record defines response as more than 50% reduction and non-response as less than 50% reduction. The wording does not state how exactly 50% is classified. That boundary detail matters for reproducibility when an outcome falls exactly at the threshold. The record also does not describe how monthly headache days were calculated from the daily logs or how incomplete logs were handled.

Why does precision matter?

An estimate describes the observed result, while a confidence interval communicates its statistical uncertainty under the assumptions of the method used. A narrower interval generally indicates greater precision; a wider interval indicates less precision. An interval does not remove design limitations, and it does not measure every source of uncertainty. For this study, the registry does not report a responder estimate or confidence interval, so the precision of the primary outcome cannot be evaluated from the posted record.

For a binary endpoint, uncertainty depends in part on how many participants are observed and how the outcome is distributed. The planned enrollment was 40, but the registry's enrollment figure should not be treated as an observed primary-outcome denominator without a reported analysis population and outcome data. The record does not describe the number of participants with evaluable logs, exclusions from analysis, or the handling of withdrawals.

What can open-label conduct affect?

In an open-label study, participants and study personnel know which intervention is being given. For an outcome involving headache days recorded in logs, awareness can potentially influence symptom reporting, expectations, or study procedures. This is a general methodological concern, not a finding that such influence occurred in this study. The registry lists masking as none and does not report additional safeguards or an assessment of reporting effects.

Baseline comparisons also invite attention to natural variability and regression to the mean. In general, if a person enters a study during a period when symptoms are unusually frequent, later measurements may be closer to that person's usual level even without an intervention effect. The registry does not report participant-level symptom trajectories or information that would establish whether this occurred.

How should planning calculations be read?

Sample-size calculations are planning tools that describe the design conditions under which a study might detect a specified effect with a chosen statistical procedure. They are not outcome results and do not establish that the study had a particular probability of finding an effect after the data were collected. The registry's planning-illustration description specifies a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect set as a stated fraction of an observed effect. It also says that the illustration is computed from the observed effect and is not the trial's own design.

Those assumptions do not match the listed single-group study design, which has no two-arm allocation. The illustration must therefore be kept distinct from the trial's actual design and from its observed result. Participants-per-arm figures in a planning illustration are hypothetical design quantities, not observed participants in this trial. They are not “observed power.” The registry does not report the trial's own sample-size justification or enough primary outcome results to assess the illustration's underlying observed effect.

Why can a control group change interpretation?

A randomized control comparison aims to create groups that, on average, differ mainly in the assigned intervention, making causal interpretation more credible when the design and analysis are sound. This trial has no randomized comparison according to the registry record. Its baseline comparison is still informative as a description of change over time, but it answers a narrower question than a controlled comparison. Statistical adjustment cannot automatically replace the missing counterfactual comparison, particularly when the registry does not report participant-level covariates or an adjustment plan.

What is not documented?

The registry does not report a primary-outcome analysis, an analysis population, a confidence interval, a statistical test, missing-data methods, interim analyses, or detailed model choices. It also does not report how the optional continuation for participants meeting the response criterion entered any analysis. These omissions limit reproducibility and interpretation. They should be treated as gaps in the registry record, not as evidence that the methods were or were not used in another study document.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Ticagrelor 90 mg Twice Per Day")
n      <- c(40)
events <- c(17)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Ticagrelor 90 mg Twice Per Day  40  17  42.5  27.0  59.1  28.5  57.8  28.5  57.9

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure, the total number of responders, and zero analyses. It does not report secondary outcome measures or a multiplicity strategy. Without a broader set of prespecified tests and a statistical analysis plan, it is not possible to determine whether multiplicity was a concern in the full study conduct or how it may have been addressed. The record does not report interim analyses.

Missing outcome data can affect both the responder count and its interpretation. Participants with incomplete daily logs may have less certain estimates of headache frequency, and the way such records are handled can influence who is classified as a responder. The registry does not report how missing logs, withdrawals, or other unavailable measurements were treated. It also does not report the number of participants with evaluable primary outcomes.

The registry's planned enrollment is 40, and the safety entries use 40 participants at risk. These details do not establish the analysis population for the primary endpoint. It is not reported whether the primary analysis was based on all enrolled participants, participants who completed the month of therapy, or another population. The record also does not explain how the optional continuation for participants meeting the response criterion was handled statistically.

These are reporting gaps, not proof that the underlying study lacked procedures. Without the methods and analysis details in the registry record, readers cannot assess the effect of missing data, identify the exact population represented by any primary result, or evaluate whether outcome definitions and analysis decisions were applied consistently.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry records zero serious adverse events among 40 participants at risk in the ticagrelor group. It also records zero deaths. Separately, it lists 10 participants affected by other, non-serious adverse events among 40 at risk. The record provides no serious adverse event terms and does not list the terms for the other events.

These are counts as entered in the registry, not a comparison against an untreated or active-control group. They do not by themselves establish whether an event was caused by ticagrelor, and the record does not provide details about event severity beyond the serious versus other distinction, timing, duration, or clinical course. The absence of recorded serious events in this group should be read as a description of the posted safety data, not as proof that serious events cannot occur.

The safety counts also do not answer the primary efficacy question. The responder count is not reported in the registry record, and no statistical analysis linking adverse events with headache outcomes is reported. The available safety entries are therefore limited to the counts and denominators recorded for the study group.

13. What the Result Does, and Does Not, Mean

The registry record establishes that TRACTOR was designed as a prospective, open-label, single-group pilot enrolling 40 participants, with ticagrelor given over a 28-day treatment period. Its primary endpoint was the number of participants whose monthly headache days fell by more than 50% from their own baseline during the month of therapy. Those details explain what the study set out to measure.

The record does not report the number of participants who met the responder definition, the number who did not, or an estimate of change in headache days. It also does not report a confidence interval or primary statistical analysis. Consequently, the posted information does not show the observed primary result or how precise an estimate would be. It cannot establish how frequently the defined response occurred in this study.

Even if a within-participant change were reported, a single-group open-label design would describe change over time rather than provide a randomized treatment-versus-control estimate. Other influences on symptom frequency or recording could contribute to an observed change. That is a general limitation of the design, not a claim about what happened to these participants.

The responder threshold is only one way to describe an outcome. It does not show the full range of headache-day changes or distinguish participants just above the threshold from those with larger reductions. The registry does not report those underlying data. Nor does it report how missing logs or the optional extension for some participants were incorporated into analysis.

The safety entries report no serious adverse events and no deaths, with 10 participants affected by other adverse events. These are descriptive registry counts in the study group and are not a comparative safety analysis. Overall, the record documents the study question and basic design, but leaves the main efficacy result and important analysis details unreported. The available information should not be treated as proof for or against a treatment effect.

14. Why This Trial Matters Statistically

This study raises several statistical issues that are important when reading small, single-group clinical studies. The primary endpoint converts headache-day change into a responder count using a more-than-50% threshold. This makes the outcome easy to state, but a threshold can conceal the size and distribution of changes. A useful interpretation would ordinarily consider both the count crossing the threshold and the underlying continuous measurements; the latter results are not reported in the registry record.

The baseline comparison also illustrates the difference between describing change and estimating a comparative effect. Repeated measurements in the same participants can show whether outcomes changed during a study period, but without a control group they cannot isolate the intervention as the reason for that change. The open-label design adds a potential source of influence on self-recorded symptoms, although the record does not establish that such influence occurred.

Reporting the analysis population, missing-data procedures, and uncertainty around an endpoint is central to judging how much information a pilot contributes. Those details, along with the primary responder result, are not reported in the registry record. Their absence prevents readers from assessing the precision and robustness of the planned endpoint.

The sample-size planning description is also a reminder that design illustrations and trial design are different things. The described illustration assumes a two-arm allocation, whereas the study itself is listed as single-group. Planning quantities based on an observed effect do not turn into an observed efficacy result, and they are not evidence of observed power. Careful separation of design, analysis, and outcome reporting is essential to avoid overstating what a pilot record can establish.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The most direct limitation for interpretation is that the registry record does not report the primary endpoint result. It gives the responder definition and timeframe, but not the count of responders or non-responders, the observed change in headache days, or a measure of statistical uncertainty. The listed count of zero analyses reinforces that no primary analysis is available in the record.

The single-group, open-label design has no concurrent comparator and no masking. The baseline is each participant's own reference, so the comparison can describe change but cannot account for all time-varying influences. Headache frequency was to be recorded in daily logs, but the registry does not explain the method for deriving monthly headache days or handling incomplete logs.

The record does not report detailed eligibility criteria, participant characteristics, baseline outcome distributions, or the number with evaluable primary data. It also does not report a statistical analysis plan, sample-size justification, missing-data approach, or analysis population. The classification of an outcome exactly at the 50% boundary is not specified in the endpoint wording.

Some participants meeting the response threshold could elect to continue therapy for 56 days, but the registry does not report how that optional period was incorporated into analyses. Safety reporting is also limited: the counts are available, but event terms and further details are not. These limitations describe what is absent from the registry record; they do not establish what may have been documented in other study materials.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Ticagrelor Therapy for RefrACTORy Migraine Study — NCT02518464.
• Registry record last updated 2019-06-04; status last verified 2019-06.
• Results first posted 2019-06-04.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

TRACTOR is recorded as a completed phase 4, prospective, open-label, single-group pilot sponsored by Columbia University. The registry lists planned enrollment of 40, a start in 2015-10, and primary completion in 2017-10. Participants received ticagrelor after an office loading dose, followed by 90 mg twice per day for 28 days.

The primary endpoint was the number of participants with more than 50% reduction in monthly headache days versus their own baseline during one month from baseline. The registry does not report the responder result or a primary analysis. Safety entries record zero serious adverse events, 10 participants affected by other adverse events, and zero deaths among 40 at risk.

The single-group design provides no randomized control comparison. Important details about the analysis population, missing data, and statistical methods are not reported in the registry record.