A Study to Look at the Efficacy and Safety of Keppra® Extended Release Formulation - XR: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Look at the Efficacy and Safety of Keppra® Extended Release Formulation - XR (NCT00368069), a randomized phase 3 study. The primary outcome is Partial Onset Seizure (POS) Frequency Per Week - Intention-To-Treat (ITT) Population.

2. Clinical Question

The study record identifies epilepsy as the condition and specifies refractory epilepsy with partial onset seizures in its official title. The primary outcome counts partial onset seizures and standardizes the count to a one-week period. This makes seizure frequency the central measure of the efficacy comparison, rather than a measure of whether any seizure occurred or the time until a seizure.

The registry also lists outcomes covering all seizure types, response thresholds, and categorized reductions in weekly partial onset seizure frequency. Those outcomes provide related perspectives on seizure activity, but the primary question is the between-group comparison in weekly partial onset seizure frequency. The posted primary analysis includes 75 participants in the Keppra® group and 78 in the placebo group. These are the analysis counts shown for that outcome, not a statement that every enrolled participant contributed to every analysis.

The registry record does not report a more detailed clinical rationale for choosing this endpoint or the study’s original assumptions for how large a difference would be important to participants. The result therefore describes a statistical comparison of seizure frequency; its clinical importance cannot be established from the outcome estimate alone.

3. Trial Design

This was a phase 3, randomized, quadruple-masked, parallel-group study comparing Keppra® extended release formulation with placebo. The registry describes the Keppra® XR as once-daily add-on therapy and identifies the study population as subjects from 12 to 70 years with refractory epilepsy and partial onset seizures. The sponsor was UCB Pharma, and the record lists the study as completed.

The registry gives the start as 2006-08 and the primary completion as 2007-05. Enrollment was 158. The primary outcome analysis lists 75 participants for Keppra® and 78 for placebo. These outcome-specific counts are the denominators attached to the posted primary endpoint summaries; the record does not explain the difference between those analysis counts and total enrollment.

Random allocation and a parallel-group structure mean that the treatment groups were followed as separate groups rather than participants crossing between study treatments. Quadruple masking is the registry’s masking classification. The record does not further describe which roles were masked or how masking was maintained. The study is described as double-blind in its official title, while the structured record classifies masking as quadruple.

The primary outcome covers a 12-week treatment period and measures partial onset seizure frequency per week. Its stated population is intention-to-treat, qualified by a requirement for non-missing baseline and treatment-period information. Accordingly, the population label should be read together with that analysis eligibility description, rather than as proof that every randomized participant was included in the endpoint calculation.

The registry record does not report the original sample-size justification, the planned effect size, interim analyses, or detailed operational information about randomization and masking. Those design features cannot be reconstructed from the posted outcome results. The displayed sample-size planning values are illustrations based on the observed effect, not evidence of the trial’s original design assumptions.

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.

Before assignment: Baseline and Participant Flow data consists of the Intent-to-Treat (ITT) analysis group. The ITT group consists of all randomized subjects.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intention-to-treat (ITT) (Analyses were performed on subjects from the ITT with non-missing information during baseline and treatment 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 Partial Onset Seizure (POS) Frequency Per Week - Intention-To-Treat (ITT) Population, measured over Treatment period (12 weeks).

The primary endpoint is partial onset seizure frequency per week during the 12-week treatment period. The registry describes the measure as the number of partial onset seizures over treatment, standardized to a one-week period. It is a continuous outcome, and the reported least-squares means are based on log-transformed data. The registry identifies the dispersion accompanying those means as a standard error.

Seizure frequency is used to compare the burden of recurring seizures between the randomized groups over the treatment period. A weekly rate provides a common time scale for participants whose observation periods may differ, while focusing the endpoint on partial onset seizures matches the population specified in the trial title. This is a general explanation of the measurement approach, not a claim that the registry reports a particular clinical threshold for meaningful change.

Log transformation is used in analyses when the distribution of a continuous measurement is skewed or when variation tends to depend on the measurement’s scale. In general, analyzing transformed values can make model assumptions more suitable, but it also means that the reported values and treatment contrast need careful interpretation: the posted least-squares means are not described as raw arithmetic counts of seizures. The registry additionally reports an ANCOVA estimate and a transformed percentage reduction over placebo.

The outcome is restricted to participants with non-missing information at baseline and during treatment, according to the registry’s population description. The record does not provide further details about how missing seizure data were handled, how baseline seizure frequency was measured, or how the transformation and model were specified.

7. Results Posted in the Registry

• Keppra®: mean 0.91 versus 1.07 on control; difference -0.15 (95% CI -0.30 to -0.01), standardized effect d = -0.34; z-test p = 0.037.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

All (Type I+II+III) Seizures Frequency Per Week

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Keppra®: mean 0.93 versus 1.09 on control; difference -0.16 (95% CI -0.30 to -0.01), standardized effect d = -0.34; z-test p = 0.033.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Keppra® than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry lists four secondary outcomes, including all-seizure frequency, partial onset seizure frequency, a 50% response measure, and a categorized response measure over the 12-week treatment period. Numerical results are posted for two secondary outcomes. The broad pattern of the posted all-seizure frequency comparison is consistent in direction with the primary frequency result. The secondary response outcomes are listed, but their numerical results are not reported in the registry record.

These secondary measures address related but distinct questions. A frequency outcome describes average seizure activity, while response categories describe how participants are distributed across levels of change. Agreement in direction across outcomes can provide context, but it does not make the outcomes interchangeable or establish that every participant experienced a similar result. Nor does a related secondary endpoint independently confirm the primary finding.

Multiplicity matters because testing several outcomes or contrasts creates more opportunities for apparently notable findings to arise by chance. The registry reports multiple outcome measures and analyses but does not report whether the secondary analyses were adjusted for multiplicity or how the outcomes were prioritized in the original analysis plan. Without that information, unadjusted secondary results carry less weight than the primary analysis and should be treated as exploratory or supportive context.

The posted secondary findings should therefore be read alongside the primary endpoint rather than as a set of independent conclusions. The registry record does not report numerical findings for every listed secondary outcome, so the complete pattern across all planned or registered secondary measures cannot be assessed from the available record.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• POS Seizure Frequency Per Week Over Baseline and Treatment Period
• All (Type I+II+III) Seizures Frequency Per Week
• 50% Response in Weekly POS Frequency
• Response in Weekly POS Frequency (Categorized Into 6 Categories According to Reduction) Over the Treatment Period of 12 Weeks

9. Statistical Interpretation

10. Statistical Methods Explained

The registry reports least-squares means for a continuous outcome analyzed on a log-transformed scale. Least-squares means are model-based estimates of group means, typically adjusted for other terms in the analysis model. The record identifies ANCOVA as the sponsor’s method for the primary comparison but does not list the model covariates or other specification details. Therefore, the results can be described as ANCOVA-adjusted without inferring exactly which variables were included.

Why analyze seizure frequency on a log scale?

As a general statistical explanation, counts and rates can be unevenly distributed, with a concentration of lower values and a smaller number of higher values. A logarithmic transformation can reduce skew and can make the residual variation more compatible with the assumptions of a linear model. It also changes how differences are expressed: contrasts on the transformed scale are not directly the same as differences between raw counts. The registry explicitly labels the primary data as log-transformed and separately reports a transformed percent reduction over placebo.

The record describes the endpoint unit as seizures per week, but the least-squares means are reported for log-transformed data. They should not be treated as untransformed participant-level seizure counts. A reader should distinguish the measure’s clinical definition, weekly partial onset seizure frequency, from the statistical scale used for the reported group summaries.

What does ANCOVA contribute?

ANCOVA is a general modeling approach that compares groups while accounting for specified covariates. In randomized studies, adjustment can improve precision when relevant baseline or design variables are included. The registry reports ANCOVA and labels its primary contrast mean difference (net). Because the covariates and model details are not reported in the registry record, the adjustment cannot be independently reconstructed from the posted summaries. The adjusted sponsor result should therefore be distinguished from the simple unadjusted recalculation.

The posted adjusted estimate is 0.155, with a 95% confidence interval from 0.009 to 0.301 and p = 0.038. The registry also gives an estimate of 14.4 for percent reduction over placebo after transformation of ANCOVA results on log data. These are two ways the sponsor presents its analysis, rather than two unrelated primary tests. The registry does not give a p-value for the transformed percent-reduction presentation.

How should the unadjusted recalculation be understood?

The recalculated comparison uses posted group summaries to form a simple, unadjusted comparison. Its estimate is -0.155 and its p-value is 0.037, with the sign reflecting Keppra® minus placebo. This calculation does not reproduce the sponsor’s covariate-adjusted ANCOVA; it is a check based on the summary statistics. Such recalculations can help readers understand direction and approximate scale, but they cannot replace an analysis using participant-level data and the prespecified model. The registry does not mark a discrepancy, and the sponsor’s own adjusted analysis remains the main reported result.

The arm-level standard errors are 0.053 for Keppra® and 0.052 for placebo. A standard error describes the estimated uncertainty of a group summary, not the variability among individual participants. The registry’s reported comparison includes a confidence interval, which expresses uncertainty about the group contrast. The standard errors should not be mistaken for standard deviations of participant-level seizure frequencies.

Why can over-dispersion matter?

In general, count data may show greater variability than a simple count model assumes. This is called over-dispersion. If a model underestimates variability, standard errors and confidence intervals can be too small, making a result appear more precise than warranted. The endpoint is analyzed as log-transformed continuous data, and the registry does not report a count-model specification or an over-dispersion assessment. It is therefore not possible to infer from the record whether over-dispersion was evaluated or how it might have affected the analysis.

More broadly, the methods record is limited. It does not report the full statistical analysis plan, the model’s covariates, procedures for missing observations, interim analyses, or multiplicity adjustments. The population description says that analyses used participants with non-missing information during baseline and treatment. That qualification matters: an intention-to-treat label alone does not show how participants with incomplete endpoint information were handled.

The result is best read as a reported ANCOVA comparison of transformed seizure-frequency data, accompanied by a simple unadjusted recalculation for context. Neither the p-value nor the confidence interval answers every question about robustness, clinical relevance, or missing data. Those questions require design and analysis details that are not reported in the registry record.

11. Confidence Intervals and Precision

The primary ANCOVA estimate is 0.155, and its 95% confidence interval ranges from 0.009 to 0.301. The interval describes the uncertainty around the model-based mean difference (net) under the analysis assumptions. It does not mean that 95% of individual participants had outcomes within those limits, nor does it give the probability that the true effect lies in the interval.

The interval is on the sponsor’s reported contrast scale. Its lower limit is close to zero, while the upper limit allows a larger difference. This indicates that the result is more precise about the direction and approximate scale than it is about a single exact effect size. The confidence interval for the transformed percent reduction over placebo is 0.9 to 26.0, which likewise spans a range of possible effect magnitudes on that transformed presentation.

The p-value of 0.038 is related to the confidence interval but answers a different question: under the model and a null hypothesis of no difference, it measures how incompatible the observed result is with that hypothesis. It is not a measure of the size, importance, or probability of the effect. A value close to a conventional threshold does not remove the uncertainty represented by the interval.

The analysis counts were 75 for Keppra® and 78 for placebo. Precision depends on the amount and variability of outcome information, as well as the analysis model. The registry gives standard errors for the arm-level means but does not report all model details needed to reproduce the sponsor’s interval. The recalculated unadjusted comparison is close to the reported estimate, but it is not a substitute for the adjusted analysis.

These intervals quantify sampling uncertainty conditional on the analysis; they do not account for every limitation, such as missing-data assumptions, unreported modeling choices, or multiplicity. They also do not identify a threshold for clinical importance, which is not reported in the registry record.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning values are participants per arm, not total events. They illustrate the group sizes that would be needed under the stated assumptions to detect a chosen fraction of the observed effect. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The target effect is the stated fraction of the observed effect. These are planning illustrations based on the observed effect, not the trial’s original sample-size justification.

The 80% power column describes the number of participants per arm associated with an 80% chance of detecting the specified effect under those assumptions, if that effect were the true effect. The 90% power column expresses the corresponding planning target for a 90% chance. These are design calculations, not statements about how likely the trial’s observed result is to be correct.

Each row uses a different fraction of the observed effect. Planning for 100% of the observed effect gives smaller required group sizes than planning for 75% or 50% of that effect. Smaller effects are harder to distinguish from random variation, so they require more participants under the same assumptions. This relationship is a general feature of statistical planning.

The planning figures are not observed power. Observed power calculated from the effect seen in a completed study does not add useful information beyond the estimate and its uncertainty. The registry record does not report the trial’s original design assumptions, target effect, or planned power calculation, so these illustrations should not be read as a reconstruction of the study design.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.155; pooled standard deviation σ = 0.459.
2. n = 2 × (1.96 + 0.842)² × 0.459² / -0.155² = 137.7, rounded up to 138 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.155                      # observed difference in means
sigma <- 0.4591250965746                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(368069)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 138 (80% power), 185 (90% power)
 75% of observed effect: n per arm = 245 (80% power), 328 (90% power)
 50% of observed effect: n per arm = 551 (80% power), 738 (90% power)
power.t.test: 138.7 
Simulated power at n = 138 per arm: 0.798 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The sponsor’s primary analysis is an ANCOVA comparison reported as a mean difference (net) of 0.155, with a 95% confidence interval from 0.009 to 0.301 and p = 0.038. The accompanying transformed presentation reports a percent reduction over placebo of 14.4, with a 95% confidence interval from 0.9 to 26.0. The registry does not provide a p-value for that transformed presentation.

The simple recalculation from the posted group summaries gives a mean difference of -0.155 and p = 0.037. The sign is reversed because the recalculation expresses the contrast as Keppra® minus placebo, while the sponsor’s net contrast is reported in the opposite direction. The estimates are close in magnitude and their statistical pattern is similar, but the recalculation is unadjusted and does not reproduce the ANCOVA model.

The registry record does not flag a discrepancy between the sponsor result and the recalculation. The sponsor’s own adjusted analysis therefore remains the main result to interpret. The recalculation is useful as a consistency check on the direction and approximate scale, but it cannot establish the sponsor model’s details or independently verify an analysis based on participant-level data.

15. Multiplicity, Missing Data and Analysis Populations

The registry reports six outcome measures and eight analyses, and lists four secondary outcomes. Multiple outcomes and analyses raise the question of multiplicity: when several statistical tests are considered, the chance of at least one apparently positive result can increase. The record does not report whether any adjustment was made, how the primary outcome was protected from other comparisons, or whether a hierarchy of outcomes was prespecified. The p-values for secondary analyses should therefore not be interpreted as if they were necessarily isolated tests.

The primary population is labeled intention-to-treat, with the qualification that analyses were performed on participants with non-missing information during baseline and treatment. This description does not explain how missing observations were handled, whether participants were excluded for other reasons, or whether any imputation or sensitivity analyses were used. The number and reasons for missing primary outcome data are not reported in the registry record.

The total enrollment was 158, while the primary outcome summaries list 75 participants for Keppra® and 78 for placebo. The record does not explain the relation between these numbers or provide a participant flow account for the endpoint. The outcome counts should thus be understood as the groups contributing to the posted primary summaries, not as a complete description of all randomized participants’ follow-up.

The record also does not report the original sample-size calculation, the statistical analysis plan, interim analyses, or detailed model specifications. These omissions limit assessment of whether the analyses followed prespecified choices and how robust the results are to alternative assumptions. They do not negate the reported ANCOVA result, but they restrict the conclusions that can be drawn about the analysis process and the role of secondary findings.

16. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: concussion (1.3%), epilepsy (1.3%), ischaemic stroke (1.3%), simple partial seizures (1.3%), respiratory failure (1.3%).

For serious adverse events, the registry reports 6 affected participants among 77 at risk in the Keppra® group and 2 among 79 at risk in the placebo group. These are participant counts with serious adverse events, not necessarily counts of distinct event episodes. The record lists several serious-event terms, each reported at 1.3%: concussion, epilepsy, ischaemic stroke, simple partial seizures, respiratory failure, and rash.

For other adverse events, the record reports 23 affected participants among 77 at risk in the Keppra® group and 21 among 79 at risk in the placebo group. These summaries describe the posted safety counts; the registry record does not provide enough information to determine the seriousness criteria, event timing, relationship to study treatment, or whether participants could have more than one event.

Death counts are not reported in the registry record. The adverse-event summaries should be read descriptively and in light of the differing at-risk counts. They do not establish causation, comparative risk, or the full safety profile. No medical advice or treatment conclusion follows from these counts alone.

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

The posted primary result is a between-group comparison in weekly partial onset seizure frequency during a 12-week treatment period. The sponsor’s ANCOVA analysis reports a mean difference (net) of 0.155, with a 95% confidence interval from 0.009 to 0.301 and p = 0.038. The associated transformed estimate is a 14.4 percent reduction over placebo, with a 95% confidence interval from 0.9 to 26.0. These estimates describe the analyzed group comparison, not the experience of every participant.

The result does not show that all participants had fewer seizures, that a particular person would respond, or that the estimated difference is important to patients. The registry does not report a threshold defining a clinically meaningful change. A statistically distinguishable group result and a clinically meaningful benefit are different questions, and the posted summaries do not settle both.

The finding applies to the enrolled population and the treatment period studied. It does not establish what happens over a longer period, in people outside the registry’s specified population, or under treatment conditions not described in the record. The registry reports the study as randomized and masked, but it does not give the full methods needed to assess all possible sources of bias or the robustness of the analysis.

The outcome was analyzed using log-transformed data and an ANCOVA comparison. The arm summaries and adjusted contrast should be interpreted on their stated scales; they are not raw counts of seizures for an individual. The confidence interval shows uncertainty about the estimated group difference, while unreported details about missing data and multiplicity add limits that are not captured by the interval.

The secondary findings and safety counts provide context but do not turn the primary result into a complete account of treatment effects. Several registered outcomes lack posted numerical results, and the registry does not report multiplicity procedures or complete safety details. The record therefore supports a focused statistical description of the posted comparison, not a broad claim about all benefits, risks, or individual outcomes.

18. Why This Trial Matters Statistically

This study illustrates why statistical interpretation must keep the endpoint scale, analysis method, and estimate direction together. The primary outcome is weekly seizure frequency, but the means are reported for log-transformed data and the sponsor’s adjusted contrast is reported as a net difference. A simple recalculation uses the opposite contrast direction. Recognizing that sign convention prevents a reader from mistaking a presentation difference for a disagreement about which group has the lower displayed mean.

The record also illustrates the distinction between an adjusted primary analysis and a recalculation from summary statistics. The two estimates are close, but only the sponsor’s ANCOVA is the reported adjusted analysis. A summary-based check can clarify the broad pattern, while the model details and participant-level data remain necessary for reproducing the sponsor’s analysis.

Precision matters alongside statistical significance. The confidence interval places the reported effect within a range, rather than identifying one exact value. A p-value alone would not communicate that uncertainty or the possible clinical importance of the result. The registry does not report a meaningful-change threshold, so the statistical evidence cannot by itself answer that clinical question.

Finally, the trial record shows why missing-data methods, multiplicity, and sample-size assumptions are important parts of evidence interpretation. Several outcomes are registered, but adjustment for multiple testing and original planning assumptions are not reported. Planning illustrations based on an observed effect cannot fill that gap. Together, these distinctions help separate what the posted analysis estimates from what remains uncertain.

19. 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 gives a limited account of the primary model. It identifies ANCOVA and log-transformed data but does not report the covariates, model diagnostics, transformation details, or full analysis plan. As a result, the adjusted estimate cannot be independently reconstructed from the registry summaries.

The intention-to-treat description is qualified by requiring non-missing baseline and treatment-period information. The record does not explain how missing data were handled, how many participants had incomplete information, or whether sensitivity analyses were conducted. The difference between total enrollment and the primary outcome analysis counts is also not explained.

Multiplicity procedures are not reported despite multiple outcomes and analyses. Four secondary outcomes are listed, while numerical summaries are posted for only two. The record does not explain why results for the remaining outcomes are absent, whether they were analyzed, or how secondary analyses were prioritized.

The registry does not report the original sample-size justification, target effect, planned power, interim analyses, or detailed randomization and masking procedures. The study duration for the primary endpoint is specified as 12 weeks, but longer-term outcomes are not established by these results. The safety record gives affected-participant counts and selected serious-event terms, but does not provide deaths, event attribution, timing, or complete detail about event handling.

These gaps limit assessment of reproducibility, robustness, and generalizability. They do not change the values reported for the primary analysis, but they constrain how much can be inferred beyond that analysis and the population and treatment period specified in the registry.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study to Look at the Efficacy and Safety of Keppra® Extended Release Formulation - XR — NCT00368069.
• Registry record last updated 2020-07-15; status last verified 2020-06.
• Results first posted 2009-07-28.
• 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.

23. Record Summary

The registry describes a completed phase 3 randomized, quadruple-masked, parallel-group study sponsored by UCB Pharma. It compared once-daily Keppra® extended release formulation as add-on therapy with placebo in people with refractory epilepsy and partial onset seizures. Enrollment was 158, and the primary endpoint covered a 12-week treatment period.

The primary outcome was weekly partial onset seizure frequency analyzed on log-transformed data. The sponsor’s ANCOVA result was a mean difference (net) of 0.155, with a 95% confidence interval from 0.009 to 0.301 and p = 0.038. The transformed percent reduction over placebo was 14.4, with a 95% confidence interval from 0.9 to 26.0. The registry record does not report the original sample-size assumptions, multiplicity procedures, or detailed missing-data methods.