This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Antiepileptic Efficacy Study of GWP42003-P in Children and Young Adults With Dravet Syndrome (GWPCARE1) (NCT02091375), a randomized phase 3 study. The primary outcome is Percentage Change From Baseline In Convulsive Seizure Frequency During The Treatment Period.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT02091375 |
| Phase | Phase 3 |
| Condition | Epilepsy, Dravet Syndrome |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Enrollment | 120 |
| Arms in the primary outcome | 2 |
| Lead sponsor | Jazz Pharmaceuticals |
| Results posted | Yes |
2. Clinical Question
Population
Children and young adults with Dravet syndrome. The registry record identifies epilepsy and Dravet syndrome as the conditions studied; further eligibility details are not reported in the registry record.
Intervention
GWP42003-P at 20 mg/kg/day, administered orally. Participants titrated to the dose over 11 days and remained at that dose during the 12-week maintenance period.
Comparator
Oral placebo, volume-matched to the active dose level. The placebo schedule also involved titration over 11 days to maintain masking.
Primary question
How did the median percentage change from baseline in convulsive seizure frequency during treatment compare between the GWP42003-P group and the placebo group?
The trial’s broader stated aim was to investigate the potential antiepileptic effects of cannabidiol in this population. Its primary outcome focused on convulsive seizures, while secondary outcomes included responder thresholds, non-convulsive seizure frequency, caregiver impressions, rescue medication use, epilepsy-related hospitalization, and sleep-related measures.
The primary outcome describes change from each participant’s baseline, then compares the groups. Negative percentage changes indicate a reduction from baseline under the registry’s definition. The comparison is therefore about the recorded seizure-frequency measure during the treatment period; it does not by itself answer questions about every symptom, longer-term outcomes, or individual experience.
3. Trial Design
NCT02091375 is recorded as a Phase 3, randomized, quadruple-masked, parallel-group study. The registry describes a two-part study: dose-ranging safety and pharmacokinetics followed by efficacy and safety evaluation of cannabidiol (GWP42003-P). The posted intervention arms for the efficacy comparison are GWP42003-P 20 mg/kg/day and placebo.
The registry reports enrollment of 120. The primary outcome results list 61 participants in the GWP42003-P group and 59 in the placebo group. These are the outcome-specific group counts; the record defines the primary efficacy population as participants in the Part B safety analysis set who had post-baseline efficacy data. The efficacy analyses followed randomized treatment assignment.
Participants or caregivers recorded convulsive seizures in an interactive voice response system diary. The active treatment was taken orally, with half administered in the morning and half in the evening. Participants titrated over 11 days and then remained at 20 mg/kg/day for the 12-week maintenance period; placebo was volume-matched and followed a titration schedule to preserve masking.
The primary outcome timeframe was baseline to end of treatment at Day 99 or early termination. The registry lists a start date of 2015-03-30, primary completion date of 2015-11-26, and status of completed. Further operational details, including the full eligibility criteria and the specific procedures used to maintain masking, are not reported in the registry record.
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
| Stage | GWP42003-P 20 mg/kg/Day Dose | Placebo |
|---|---|---|
| STARTED | 61 | 59 |
| Safety Analysis Set | 61 | 59 |
| Intention to Treat (ITT) Analysis Set | 61 | 59 |
| COMPLETED | 52 | 56 |
| NOT COMPLETED | 9 | 3 |
Reasons for not completing this period:
| Reason | GWP42003-P 20 mg/kg/Day Dose | Placebo |
|---|---|---|
| Adverse Event | 8 | 1 |
| Lost to Follow-up | 0 | 1 |
| Withdrawal by Subject | 0 | 1 |
| Withdrawn by the investigator | 1 | 0 |
Analysis population for the primary outcome, as defined in the registry: ITT analysis set: included all participants in the Part B safety analysis set who had post-baseline efficacy data. Participants were analyzed according to the treatment group to which they were randomized. The ITT analysis set was the primary analysis set for all efficacy outcome measures.
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.
Population: Safety Analysis Set: included all participants randomized to treatment who received at least 1 dose of IMP. Participants were analyzed according to the actual treatment they received.
| Characteristic | Unit | GWP42003-P 20 mg/kg/Day Dose | Placebo | Total |
|---|---|---|---|---|
| Participants analysed | 61 | 59 | 120 | |
| Age, Continuous | Years | 9.736 (SD 4.7309) | 9.779 (SD 4.8505) | 9.757 (SD 4.7699) |
| Sex: Female, Male: Female | Participants | 26 | 32 | 58 |
| Sex: Female, Male: Male | Participants | 35 | 27 | 62 |
6. Primary Endpoint
The registry lists the primary outcome as Percentage Change From Baseline In Convulsive Seizure Frequency During The Treatment Period, measured over Baseline to End of Treatment (EOT) (Day 99) or Early Termination (ET).
| Element | Registry entry |
|---|---|
| Result type | Median |
| Measure of spread | Inter-Quartile Range |
| Unit | percent change |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint was percentage change from baseline in convulsive seizure frequency during treatment. Convulsive seizures were defined in the registry as atonic, clonic, tonic, or tonic-clonic seizures recorded by the participant or caregiver in an interactive voice response diary.
Frequency was expressed as a 28-day average, using the number of recorded seizures and the number of reported diary days. Baseline included available data before Day 1; the treatment-period measure ran to end of treatment at Day 99 or early termination. Percentage change was calculated relative to baseline. Under this definition, negative values indicate a reduction from baseline.
Using a change measure makes each participant’s treatment-period frequency interpretable relative to their own baseline, while the group comparison evaluates how those changes differed between randomized arms. A median is useful when the distribution of individual changes is skewed or contains extreme values, because it describes the middle of the observed distribution rather than relying on the arithmetic average. The registry reports medians and inter-quartile range as the endpoint’s specified summary, but numerical inter-quartile ranges are not reported in the registry record.
This endpoint captures recorded convulsive seizure frequency over the specified treatment window. It does not alone describe seizure severity, non-convulsive seizures, daily functioning, or outcomes after the treatment period.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The primary outcome estimates were negative in both groups: the median percentage change was -38.94 for GWP42003-P and -13.29 for placebo. Negative values represent a reduction from baseline according to the endpoint definition. The posted bounds for the group estimates were -69.53 to -4.83 and -52.53 to 20.2, respectively; the registry does not state the confidence level for these arm-specific bounds.
The sponsor’s reported comparison used a Wilcoxon rank-sum test and expressed the result as a median difference in final values. The reported median difference was -22.79, with a 95% confidence interval from -41.06 to -5.43 and a p-value of 0.012. The interval lies on the negative side of zero, and the reported p-value is below the 0.05 significance level used in the registry’s planning assumptions. These findings indicate evidence of a between-group difference in the primary measure under the reported analysis. The result does not establish whether the size of that difference is important to patients or applies beyond the trial population and timeframe.
A confidence interval describes the uncertainty around an estimate under the statistical method used. Its width matters as well as its position: an interval that excludes zero is compatible with a nonzero group difference, but it does not eliminate uncertainty about the magnitude. The interval for the sponsor’s primary comparison extends across a range of possible differences, so the point estimate should not be read as the only plausible value.
The posted arm-level medians and the sponsor’s comparison are related but are not interchangeable. The arm summaries describe each group’s median percentage change, whereas the sponsor’s test compares the distribution of final values between groups and reports a median difference. A difference between group summaries should not be reconstructed by subtracting the medians; the relevant inferential result is the sponsor’s stated analysis.
The primary result is distinct from the secondary outcomes. Those outcomes address different features of seizure response, care, and sleep, and their posted comparisons do not create a single combined measure of benefit. The registry does not report enough information to assess all secondary outcomes comparatively. Interpretation should therefore remain anchored to the prespecified primary measure as described in the record, without treating the secondary results as independent confirmation of the primary finding.
9. Statistical Methods Explained
The primary outcome is a continuous change measure summarized by its median. A median is the central observation after values are ordered; it is less influenced by unusually large or small observations than a mean. The registry’s endpoint specification also names the inter-quartile range, which describes the middle half of the distribution. Numerical inter-quartile ranges are not reported in the registry record, so the spread of individual changes cannot be assessed from that summary.
What does the Wilcoxon rank-sum test assess?
The sponsor used a Wilcoxon rank-sum test for the primary comparison. In general, this nonparametric test compares the ranks of observations across independent groups. It can be useful when a normal-distribution model for the outcome is not appropriate. It assesses whether the groups’ outcome distributions differ; it is not simply a test of the arithmetic means, and its interpretation as a difference in medians requires conditions about the shapes of the group distributions. The registry reports a median difference in final values alongside the test, but further details of the test implementation are not reported in the registry record.
The sponsor reported a median difference of -22.79 and a 95% confidence interval from -41.06 to -5.43. In general, a confidence interval gives a range of values reasonably compatible with the data and the method, subject to the method’s assumptions. It is not a range containing a known fixed proportion of individual participant outcomes. Because this reported interval does not include zero, it is consistent with a difference between groups under the analysis used. The p-value of 0.012 is evidence against the null comparison under that test; it does not measure the probability that the null hypothesis is true or quantify clinical importance.
How should binary outcomes be read?
Some secondary outcomes are counts of participants meeting a condition, such as a responder threshold or use of rescue medication. A risk difference expresses the contrast in event proportions between groups. Its confidence interval describes uncertainty in that contrast; when the interval includes zero, the data remain compatible with no difference as well as with differences in either direction. The recalculated binary comparisons use posted counts and are simple, unadjusted re-calculations. They are not replacements for the sponsor’s own analyses.
For the responder outcome, the sponsor reported a Cochran-Mantel-Haenszel odds ratio of 2.00, with a 95% confidence interval from 0.93 to 4.30 and a p-value of 0.078. An odds ratio compares the odds of meeting the defined outcome between groups; it is not the same quantity as a risk difference or a risk ratio. The reported interval includes the null odds ratio of 1, so it does not provide clear evidence against no association at the stated interval level. The registry gives no sponsor-reported comparison for rescue medication use.
Over-dispersion is a general concern when count outcomes vary more than a simple statistical model expects. In such settings, a model that assumes less variability can produce overly narrow intervals or overly small p-values. The primary endpoint is analyzed as a change measure with a rank-based test, and the posted binary summaries do not report an over-dispersed count model. The record does not report a model for seizure counts that would allow further assessment of this issue.
What do the planning illustrations mean?
The sample-size illustrations are planning calculations based on the observed effects, not the trial’s own design justification. They assume a two-sided significance level of 0.05, equal allocation, and a normal approximation; the effect to detect is a stated fraction of the observed effect. The values are participants per arm for 80% power or 90% power. They are hypothetical planning illustrations, not observed power and not evidence that the completed trial was designed around those assumptions. The trial’s sample-size justification is not reported in the registry record.
More generally, power is a design-stage probability of detecting a specified effect under stated assumptions if that effect is present. It is not a measure of whether a completed trial’s result is reliable. Recalculating power from an observed effect is especially dependent on the observed estimate and should not be used to interpret the primary p-value. The planning calculations do not resolve uncertainty from missing data, multiplicity, or the choice of analysis model.
The primary population was the ITT analysis set as defined in the registry: participants in the Part B safety analysis set with post-baseline efficacy data, analyzed according to randomized group. This definition is not equivalent to including every randomized participant regardless of outcome availability. The amount and handling of missing primary outcome data are not reported in the registry record.
10. Analyses Reported by the Sponsor
| Method | Estimate | 95% CI | P-value |
|---|---|---|---|
| Wilcoxon rank-sum test | Median Difference (Final Values) -22.79 | -41.06 to -5.43 | 0.012 |
The sponsor’s primary analysis is the principal inferential result in the registry record. It used a Wilcoxon rank-sum test and reported a median difference in final values of -22.79, with a 95% confidence interval from -41.06 to -5.43 and a p-value of 0.012. The result is consistent with a difference between randomized groups in the primary endpoint under the reported method.
The recalculated binary summaries are simple, unadjusted analyses based on posted group counts. They answer different questions from the sponsor’s primary rank-based analysis and should not be treated as substitutes for it. For the responder outcome, the sponsor also reported a Cochran-Mantel-Haenszel analysis: odds ratio 2.00, 95% confidence interval 0.93 to 4.30, and p-value 0.078. The sponsor did not report a comparison for rescue medication use in the registry record.
The record does not flag a discrepancy between the posted primary result and a recalculation. The sponsor’s analysis remains the relevant reported test, while recalculations offer an unadjusted check of selected binary data rather than a reanalysis of the trial’s full dataset. Participant-level data and additional analytic details are not reported in the registry record.
11. Multiplicity, Missing Data and Analysis Populations
The registry identifies an ITT analysis set for efficacy outcomes. It includes participants in the Part B safety analysis set who had post-baseline efficacy data, analyzed according to randomized treatment assignment. This definition preserves randomized assignment for those included, but it does not by itself establish that every randomized participant contributed to the primary analysis.
The record does not report the number of participants with missing primary endpoint data, the reasons for missingness, or the method used to handle missing observations. It also does not describe sensitivity analyses for missing data. These omissions matter because the primary measure depends on seizure diary records: differences in diary completeness could affect which observations contribute to the endpoint. No assumption about the direction or size of any missing-data effect can be made from the registry record.
The registry lists multiple secondary outcomes and several responder thresholds. Testing many outcomes can increase the likelihood of nominally significant findings by chance. The multiplicity strategy, including whether the primary and secondary tests were placed in a hierarchy or adjusted for multiple comparisons, is not reported in the registry record. The primary p-value is therefore interpreted as the result of its reported test, not as evidence that every secondary comparison was controlled within a family-wise testing plan.
Interim analyses, stopping rules, and the statistical analysis plan are not reported in the registry record. The absence of these details does not show that such procedures were or were not used. It limits assessment of how the reported tests relate to the trial’s full prespecified analysis strategy.
The record specifies a primary analysis population and a primary endpoint, but does not give further detail on analysis-set exclusions, diary-data rules, or sensitivity checks. Such information would help assess how robust the result is to alternative assumptions about incomplete records and participant inclusion.
12. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| GWP42003-P 20 mg/kg/Day Dose | 10 | 61 | 16.4% |
| Placebo | 3 | 59 | 5.1% |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| GWP42003-P 20 mg/kg/Day Dose | 46 of 61 |
| Placebo | 28 of 59 |
Most frequent serious adverse event terms
| Event term | Organ system | GWP42003-P 20 mg/kg/Day Dose | Placebo |
|---|---|---|---|
| Status epilepticus | Nervous system disorders | 3 of 61 | 3 of 59 |
| Convulsion | Nervous system disorders | 2 of 61 | 1 of 59 |
| Somnolence | Nervous system disorders | 3 of 61 | 0 of 59 |
| Respiratory failure | Respiratory, thoracic and mediastinal disorders | 1 of 61 | 1 of 59 |
| Abdominal distension | Gastrointestinal disorders | 1 of 61 | 0 of 59 |
| Abdominal pain | Gastrointestinal disorders | 1 of 61 | 0 of 59 |
| Gastrointestinal haemorrhage | Gastrointestinal disorders | 1 of 61 | 0 of 59 |
| Asthenia | General disorders | 1 of 61 | 0 of 59 |
| Fatigue | General disorders | 1 of 61 | 0 of 59 |
| Lower respiratory tract infection | Infections and infestations | 1 of 61 | 0 of 59 |
Most frequent non-serious adverse event terms
| Event term | Organ system | GWP42003-P 20 mg/kg/Day Dose | Placebo |
|---|---|---|---|
| Diarrhoea | Gastrointestinal disorders | 19 of 61 | 6 of 59 |
| Somnolence | Nervous system disorders | 19 of 61 | 6 of 59 |
| Decreased appetite | Metabolism and nutrition disorders | 16 of 61 | 3 of 59 |
| Pyrexia | General disorders | 9 of 61 | 5 of 59 |
| Fatigue | General disorders | 11 of 61 | 2 of 59 |
| Vomiting | Gastrointestinal disorders | 9 of 61 | 3 of 59 |
| Upper respiratory tract infection | Infections and infestations | 7 of 61 | 5 of 59 |
| Lethargy | Nervous system disorders | 7 of 61 | 3 of 59 |
| Convulsion | Nervous system disorders | 5 of 61 | 2 of 59 |
| Nasopharyngitis | Infections and infestations | 3 of 61 | 3 of 59 |
Serious events with the highest arm-level rate: status epilepticus (5.1%), somnolence (4.9%), convulsion (3.3%), respiratory failure (1.7%), abdominal distension (1.6%).
The registry reports serious adverse events for 10 participants in the GWP42003-P group and 3 participants in the placebo group. It also reports other adverse events for 46 participants and 28 participants, respectively. These are participant counts as posted; they do not describe event severity beyond the serious-event classification, duration, or whether events were judged related to treatment.
The record lists serious-event terms including status epilepticus, somnolence, convulsion, respiratory failure, abdominal distension, and abdominal pain, with percentages attached to the terms. The list does not specify group attribution for those percentages, so they should not be assigned to either treatment arm. Death counts are not reported in the registry record.
Adverse-event counts require context. They do not, on their own, establish causation or permit a full comparison of safety profiles. The registry record does not provide the event-level timing, recurrence, exposure-adjusted rates, or detailed attribution needed for those assessments. These figures describe the posted safety record and should not be interpreted as a complete account of risk.
13. What the Result Does, and Does Not, Mean
The primary result is evidence of a between-group difference in the recorded percentage change in convulsive seizure frequency during the trial’s treatment period, under the sponsor’s reported rank-based analysis. The group medians were both below baseline, with the GWP42003-P median more negative than the placebo median. The reported confidence interval for the comparison excludes zero, while still spanning a range of possible differences.
This statistical result does not by itself establish the clinical importance of the difference. The registry record does not provide a threshold for a meaningful change in this endpoint, nor does it show how the measured difference relates to participants’ daily functioning or longer-term outcomes. The result also concerns the specified population, dose, comparator, and treatment timeframe; it should not be generalized automatically to other populations, doses, or durations.
The endpoint is based on participant or caregiver diary records and a 28-day average. It reflects convulsive seizure frequency as defined in the record, not all types of seizures or every aspect of Dravet syndrome. Secondary outcomes cover additional domains but have limited comparative reporting, and their interpretation is complicated by the unreported multiplicity strategy.
Neither a p-value nor a confidence interval states the probability that a treatment is beneficial for an individual. Statistical inference describes group-level data under assumptions; it does not predict an individual outcome. The registry results are evidence about the trial’s measured outcomes, not medical advice or a complete judgment about treatment decisions.
14. Why This Trial Matters Statistically
This trial illustrates why the choice of summary measure and test matters in clinical research. The primary endpoint is a percentage change from baseline, summarized by medians, and the sponsor used a rank-sum test rather than presenting a mean-based comparison. That approach is relevant when individual changes may not be well described by a symmetric distribution, although the registry does not provide the full distribution or numerical inter-quartile ranges needed to inspect that possibility.
It also illustrates the difference between an endpoint-level summary and a between-group analysis. The arm medians describe observed changes within each group; the sponsor’s median-difference analysis addresses the comparison between groups. These summaries answer related but distinct questions, and the inferential conclusion should rely on the stated comparison rather than a newly calculated subtraction of group medians.
The record also shows why statistical significance and precision should be considered together. The primary comparison’s reported interval lies away from zero, but its range still reflects uncertainty about the magnitude. Selected secondary comparisons have less decisive intervals, and the registry does not report a multiplicity strategy. A statistically notable primary result therefore should not be extended automatically to all measured outcomes.
Finally, planning calculations based on an observed effect are not the same as a trial’s original sample-size justification or observed power. Clear separation of design assumptions, sponsor analyses, and post hoc re-calculations helps readers understand what the registry can establish and what remains uncertain.
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 registry record does not report the primary endpoint’s numerical inter-quartile ranges, the confidence level for the arm-specific bounds, or the full distribution of participant-level changes. Those omissions limit assessment of variability and of how representative the medians are of individual responses.
The primary analysis population required post-baseline efficacy data. The registry does not report missing-data counts, diary completeness, reasons for missingness, or sensitivity analyses. Because seizure frequency was diary-recorded, these details would be useful for assessing whether incomplete records could affect the group comparison.
Although the study is described as a two-part investigation, the available efficacy summaries do not provide full details of the transition between parts or the analytic handling of all participants. The registry’s account of arm descriptions is incomplete, and further protocol and analysis-plan details are not reported in the registry record. The sample-size justification, interim analyses, and multiplicity procedures are also not reported.
Secondary outcome reporting is limited for comparative interpretation. The record includes several types of outcome, but comparative estimates are unavailable for many of them. The results therefore cannot establish a coherent pattern across all secondary measures or show whether secondary analyses were confirmatory.
Safety information is presented as participant counts and a list of serious-event terms. Event timing, duration, attribution, group-specific term percentages, and deaths are not reported in the registry record. These limitations preclude a detailed assessment of comparative safety from the posted summaries alone.
16. Related Tutorials
17. Related Statistical Calculators
18. Sources
- ClinicalTrials.gov: Antiepileptic Efficacy Study of GWP42003-P in Children and Young Adults With Dravet Syndrome (GWPCARE1) — NCT02091375.
- Registry record last updated 2022-09-28; status last verified 2022-09.
- Results first posted 2018-07-20.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
19. Record Summary
NCT02091375 is a completed Phase 3 randomized, quadruple-masked, parallel-group study of GWP42003-P 20 mg/kg/day versus placebo in children and young adults with Dravet syndrome. The primary endpoint was percentage change from baseline in convulsive seizure frequency through end of treatment at Day 99 or early termination. The sponsor reported a Wilcoxon rank-sum comparison with a median difference of -22.79 and a 95% confidence interval from -41.06 to -5.43. The registry also lists secondary outcomes and adverse-event counts, but comparative reporting and analysis-plan details are limited. Missing-data handling and multiplicity procedures are not reported in the registry record.