Pilot Study Evaluating the Use of Nasal Carbon Dioxide for the Treatment of Cluster Headache: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pilot Study Evaluating the Use of Nasal Carbon Dioxide for the Treatment of Cluster Headache (NCT02381795), a na phase 2 study. The primary outcome is Headache Pain Intensity.

2. Clinical Question

The clinical question is exploratory: it concerns the recorded pain change after an attack was treated, rather than a comparison between assigned treatment groups. The registry identifies the primary measure as the greatest change in pain intensity at any time within 30 minutes of administration, relative to the pretreatment measurement. Pain intensity was recorded on a scale ranging from no pain to severe pain, with mild and moderate categories between those endpoints.

The registry record does not report the numerical outcome estimate, the individual observations, or an uncertainty interval. As a result, it is not possible to describe the size or precision of the observed change from the record. The design also cannot separate a treatment-associated change from other influences on pain reporting or the natural course of an attack. The study can therefore be described in terms of its question, methods and recorded participation, but its primary clinical result cannot be quantified from the registry information.

3. Trial Design

This was a Phase 2, investigator-sponsored pilot study with an open-label, single-group design. The registry lists an enrollment of 10 and describes allocation as not applicable. There was no masking and no control arm. The study status is listed as terminated. The record gives a start of 2015-07 and primary completion in 2017-04. The sponsor is listed as Cady, Roger, M.D.

Participants were screened for eligibility. The time from screening to treatment could range from 0 days to a maximum of 12 weeks. Someone already experiencing a cluster headache episode at screening could enter treatment after meeting eligibility criteria and could choose to treat the episode in the clinic. Someone without an active episode could remain in screening for up to 12 weeks while awaiting an episode, then enter the treatment period. The record does not report how many participants followed either path.

The intervention was a hand-held nasal carbon dioxide dispenser calibrated to deliver 0.5 standard liters per minute. The instructions described 10 seconds of use in each nostril, with doses separated by 3-5 minutes and up to 6 administrations for one attack. Participants were instructed to treat only one attack in a 24-hour period and could treat up to three attacks during the treatment phase. The arm description also states that 0.17 liters of carbon dioxide would be delivered through two 10 second administrations in each nostril, up to 6 times, with a stated total of 1.0 L for an attack and 3.0 L over the treatment phase. One hour after the first dose, participants could choose to use investigator-approved rescue medication.

These design features make the trial an uncontrolled pilot, not a randomized comparison. The registry record does not report a protocol rationale for the enrollment target, nor does it report a detailed statistical analysis plan.

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:

Analysis population for the primary outcome, as defined in the registry: Enrolled subjects reporting at least one headache during study period. One enrolled subject did not experience any headaches during the study period and thus was not included in the analysis.

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 Headache Pain Intensity, measured over Immediately preceeding treatment to 30 minutes post-treatment.

The primary outcome was headache pain intensity, measured from immediately before treatment to 30 minutes after treatment. The registry defines the outcome as the greatest change in pain intensity at any time within that 30-minute period, relative to the pretreatment measurement. Pain was rated on a scale with 0 for no pain, 1 for mild pain, 2 for moderate pain and 3 for severe pain.

In general, a repeated pain rating can describe whether a reported symptom changes over a specified observation period. Selecting the greatest change within that period focuses on the most pronounced recorded movement from the starting measurement, rather than requiring the change to be present at one prespecified instant. This approach is sensitive to the timing of observations and to fluctuations in reported pain. The registry does not describe the measurement schedule within the 30-minute interval, so the timing and density of those observations are not reported in the registry record.

The record calls the reported values “Full Range” values but clarifies that they were calculated change values and do not represent the full range of the pain scale. No numerical value for the primary outcome is reported in the registry record. The endpoint population was enrolled subjects reporting at least one headache during the study period; one enrolled subject without a headache was excluded from that analysis. The record does not report an uncertainty interval or a detailed method for calculating the endpoint beyond its description as the greatest change.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The design is a single-group study: there is one intervention group and no concurrent comparison group. In a randomized trial, random assignment is used to create groups that can be compared while reducing systematic differences between them. That framework is absent. The allocation field is listed as not applicable, and the record identifies no control arm. Consequently, the primary measure is a change observed after treatment within the study group, not a between-group treatment contrast.

What an open-label design means

Open label means masking was not used. In general, participants and study personnel who know which intervention is being used may influence how symptoms are reported, assessed or managed. This does not show that such influence occurred in this study; it describes a possible source of bias that is relevant when interpreting subjective outcomes such as reported pain. The registry does not report procedures intended to reduce or evaluate that possibility.

How to read a before-and-after pain measure

The primary outcome compares pain intensity immediately before treatment with the greatest change recorded within 30 minutes after administration. A within-person comparison can describe the trajectory of a symptom during a defined period. It does not, on its own, reveal what would have happened over the same period without the intervention. In general, a control group helps estimate that unobserved counterfactual by providing a comparison under similar study conditions. No such group was included in this design.

The outcome uses a four-category numerical scale from 0 to 3, with verbal descriptions attached to each value. A scale of this kind represents ordered categories: higher values indicate more severe pain, but the distance between adjacent categories is not necessarily equal in the underlying experience. Analysts may choose different summaries for such data, but the registry record does not report the specific statistical method used for the primary outcome. It reports a mean parameter and full-range dispersion, while clarifying that the full-range values are calculated change values rather than the full range of the scale. It does not give the actual mean, dispersion values or an uncertainty interval.

Why observation timing matters

In general, taking the greatest change over a time window makes the endpoint dependent on the measurements available within that window. If observations are made at different times or with different frequency, the most extreme recorded change can differ even when the underlying symptom pattern is similar. The registry gives the overall period as 30 minutes but does not report the schedule of measurements within it. It also does not report how missing pain ratings, if any, were handled.

Repeated attacks and participant-level dependence

Participants could treat up to three attacks during the treatment phase. When multiple observations come from one person, they can be correlated because they share characteristics of that participant. General statistical methods for repeated observations account for that dependence rather than treating every attack as fully independent. The record does not report how many attacks contributed to the endpoint or whether the primary analysis used one observation per participant, multiple attacks, or a method that accounted for repeated observations. No particular approach can be assumed.

Uncertainty and confidence intervals

In general, a confidence interval expresses uncertainty around an estimated quantity under the assumptions of the statistical method. A narrower interval indicates greater precision, while a wider interval indicates less precision. An interval does not eliminate bias from an uncontrolled design or show that the assumptions are correct. The registry record does not report a confidence interval for the primary outcome, so its precision cannot be evaluated from the posted information. A p-value is also not reported for that outcome. The absence of those quantities means the primary estimate's statistical uncertainty is not described in the record.

Planning illustrations are not trial findings

The sample-size planning assumptions describe a hypothetical two-arm comparison, with equal allocation, a two-sided significance level of 0.05 and a normal approximation. They are separate from the single-group design used in this pilot. The assumptions also describe detecting a stated fraction of an observed effect, but the numeric planning outputs and effect fractions are not reported in the registry record. The illustrations are not an account of the original trial's design justification; that justification is not reported in the registry record.

In general, power is the probability that a specified statistical procedure will detect an effect of a chosen size under stated assumptions. A calculation based on an observed effect after a study is not “observed power,” and it does not replace an uncertainty interval or repair the lack of a control group. Sample-size illustrations are planning aids for a possible design, not observed participant counts or evidence of efficacy. No inference about the power of this trial should be drawn from the planning assumptions.

Over-dispersion and count outcomes

In general, over-dispersion occurs when the variability in count data is greater than a model assumes. It can make standard errors too small if the model does not account for the extra variability. This concept may matter in studies that analyze counts, such as numbers of attacks or events, but the registry does not report a count-based primary analysis for this trial. Since no detailed analysis method is listed, there is no basis for saying that over-dispersion was assessed or that a particular model was used.

The registry lists one primary outcome measure and zero analysis entries. It does not report an analysis plan, interim analysis procedures, test statistics or model specification. The statistical interpretation must therefore remain limited to the outcome definition and design features that are recorded.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure, headache pain intensity, and zero analysis entries. It does not report a statistical analysis plan describing the tests, models or decision rules intended for that outcome. It also does not report how any secondary or exploratory outcomes were designated or analyzed. A multiplicity adjustment, or a reason for not applying one, is not reported in the registry record.

In general, multiplicity arises when several outcomes, time points, subgroups or comparisons are examined. Repeated testing can increase the chance of a false-positive finding if all tests are treated as independent opportunities for significance. Prespecifying a primary outcome and analysis can help clarify which result is confirmatory and how additional analyses should be interpreted. Here, the primary outcome is identified, but the record does not describe a testing hierarchy, adjustment procedure, or whether additional analyses were planned.

The record identifies the endpoint population as enrolled subjects reporting at least one headache during the study period. One enrolled subject did not experience a headache during the study period and was not included in the analysis. It does not report an intention-to-treat analysis, a per-protocol analysis, or another named analysis population. How missing pain ratings or incomplete treatment observations were handled is not reported in the registry record. Nor does the record describe imputation, sensitivity analyses, or reasons for missing measurements.

These omissions leave uncertainty about the planned statistical framework and the treatment of incomplete observations. They do not establish that missing data or multiplicity affected the results; the registry does not report enough information to determine that. The pilot's primary outcome population can be described as stated, but broader analysis decisions should not be inferred.

11. 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 lists the Nasal Carbon Dioxide group with 10 participants at risk for serious adverse events and 0 affected. For other adverse events, 10 were at risk and 5 were affected. These are participant counts as recorded, not an estimate of the probability that the intervention caused an event. The record lists no serious adverse event terms.

The serious-event count should be read as no affected participants recorded among the 10 at risk in this group, not as proof that serious harm is impossible or could not occur in other settings. A zero count in a small study provides limited information about uncommon events. The other-adverse-event count indicates that the registry recorded affected participants, but it does not identify the event types, severity, timing, recurrence or relationship to the intervention. Those details are not reported in the registry record.

There is no comparator group, so these counts cannot be used to compare safety against another treatment or no treatment. The registry also does not report participant-level exposure duration or a detailed adverse-event analysis. The counts are therefore descriptive and should be interpreted within the limits of this small, uncontrolled pilot.

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

The record establishes that an open-label pilot study of nasal carbon dioxide for episodic cluster headache was conducted with a single group and an enrollment of 10. It describes a primary pain-intensity measure based on change from immediately before treatment to within 30 minutes afterward. The endpoint population excluded one enrolled subject who did not experience a headache during the study period. The registry does not report a numerical estimate for the primary outcome, so the magnitude, direction and precision of the pain change cannot be stated.

The absence of a control group means that a before-and-after observation would not by itself show what caused a change. In general, symptoms can change over time for reasons unrelated to the study intervention, and participants' awareness of treatment can influence subjective reports. These are limitations of interpretation, not claims that any specific influence occurred among these participants. A randomized comparator would address a different question by enabling a contrast between groups; this pilot did not make that comparison.

The registry record does not report a statistical analysis plan, a numerical uncertainty estimate, or how repeated attacks and missing measurements were handled. The safety information records no serious adverse events among 10 at risk and other adverse events in 5 of 10, without event details. Those observations describe what the registry reports, but cannot establish comparative safety or causation.

The record therefore supports a description of the study design, treatment instructions, endpoint definition and safety counts. It does not support a numerical conclusion about the primary pain outcome or a claim that nasal carbon dioxide caused a particular change. It also does not establish how results would compare with another treatment, placebo or the natural course of an attack. The planning assumptions for a hypothetical two-arm design are not a substitute for trial findings, and are not evidence of observed power. The study's status as a pilot is consistent with an exploratory role, but its rationale and intended statistical purpose beyond the registry description are not reported in the registry record.

13. Why This Trial Matters Statistically

This trial illustrates why study design and result reporting both matter in biostatistical interpretation. The registry identifies a concrete clinical question and describes a measurable symptom outcome, but it does not report the numerical primary result. Without that estimate, there is no basis to discuss the observed size or precision of the pain change. A well-defined endpoint is useful, but it does not replace the data needed to interpret it.

The single-group, open-label structure also makes the estimand—the quantity the study can describe—different from a randomized treatment effect. The within-person endpoint concerns change after treatment during a short observation period. It is not a comparison of outcomes under nasal carbon dioxide versus an alternative. That distinction is central to reading uncontrolled studies and avoiding causal conclusions that the design cannot support.

The outcome definition presents additional statistical questions. It selects the greatest change within a 30-minute window, and participants could treat repeated attacks. In general, extrema can depend on measurement timing, and repeated observations from one participant are often correlated. The record does not report measurement schedules or the method used to handle repeat attacks. Those details would help readers understand how observations were summarized and how uncertainty was assessed.

The contrast between this single-group pilot and the two-arm planning assumptions is also instructive. Planning scenarios can help describe requirements for a future comparison under chosen assumptions, but they do not retroactively explain the trial's enrollment target or demonstrate that its design was adequately powered. A planning calculation based on an observed effect should not be confused with an observed effect estimate or observed power.

Finally, the safety counts are informative as descriptive registry entries but have limited scope: the serious-event count is zero among those at risk, while other adverse events were recorded for some participants. Event details and a comparator are absent. Reading these findings in their proper statistical context separates recorded details from unanswered questions.

14. 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 trial's single-group design provides no concurrent control for interpreting change over time. Its open-label conduct also means masking was not used. Because the primary outcome is a subjective pain rating, general concerns about awareness and reporting are relevant, although the registry does not establish that they altered the observations. The design cannot support a direct comparison with another treatment or placebo.

The registry does not report a numerical primary outcome, uncertainty interval, p-value, or detailed analysis method. It therefore does not permit assessment of the result's size or precision. The endpoint is the greatest change within 30 minutes, but the observation schedule inside that interval is not reported. The record also does not explain how repeated attacks from one participant were incorporated, despite allowing treatment of up to three attacks.

The primary outcome population excluded one enrolled subject who had no headache during the study period. The record does not report how missing pain ratings, incomplete follow-up, or other protocol deviations were addressed. It does not report a statistical analysis plan, multiplicity approach, interim analyses, or sample-size justification. These details are not reported in the registry record and should not be inferred.

Enrollment was 10, and the status is listed as terminated. The registry record does not state why the study was terminated, so no reason can be assigned. The safety record has no serious adverse events among the 10 participants at risk, but the size and uncontrolled design limit what those counts can establish. Other adverse events are reported for 5 participants, without terms or further descriptive detail. Overall, the absence of outcome and analysis detail constrains interpretation beyond the limitations inherent in the design.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Pilot Study Evaluating the Use of Nasal Carbon Dioxide for the Treatment of Cluster Headache — NCT02381795.
• Registry record last updated 2018-08-14; status last verified 2018-07.
• Results first posted 2018-05-01.
• 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.

18. Record Summary

The registry describes a Phase 2, open-label, single-group pilot study of nasal carbon dioxide for episodic cluster headache. Enrollment was 10; the study status is listed as terminated. The primary outcome was the greatest change in headache pain intensity within 30 minutes after treatment compared with immediately before treatment. One enrolled subject reported no headache during the study period and was excluded from that outcome population.

The registry record does not report a numerical primary outcome estimate or a detailed statistical analysis plan. No serious adverse events were recorded among 10 participants at risk; other adverse events affected 5 of 10. The design has no control group, so the record does not provide a randomized treatment comparison.