Safety and Efficacy of Dexlansoprazole Modified Release Formulation to Treat Gastroesophageal Reflux Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Safety and Efficacy of Dexlansoprazole Modified Release Formulation to Treat Gastroesophageal Reflux Disease (NCT00321984), a randomized phase 3 study. The primary outcome is Percentage of Days With Neither Daytime Nor Nighttime Heartburn During Treatment as Assessed by Daily Electronic Diary-Median.

2. Clinical Question

The trial focused on a diary-based symptom outcome in people with symptomatic, nonerosive GERD. It was registered as a Phase 3, randomized, triple-masked, parallel-group study. The registry record gives an enrollment of 947 and lists the trial as completed. Its stated purpose was to assess both efficacy and safety; the primary endpoint addressed symptom relief, while adverse-event information was also recorded.

The main statistical question is about the distribution of heartburn-free days during treatment, rather than a single symptom assessment. The measure combines daytime and nighttime diary results: a day counts as heartburn-free when neither daytime nor nighttime heartburn is recorded. The registry record does not report a clinical threshold for a meaningful change in this percentage, so statistical evidence of a group difference cannot by itself establish how important that difference would be to participants.

3. Trial Design

This was a Phase 3 randomized, triple-masked, parallel-group clinical trial sponsored by Takeda. The registry record lists gastroesophageal reflux disease as the condition and describes the enrolled population as subjects with symptomatic, nonerosive GERD. The planned comparison involved placebo QD and dexlansoprazole MR at 30 mg QD and 60 mg QD. Enrollment was 947. The record lists the study as completed, with a start in 2006-06 and primary completion in 2006-12.

Randomized allocation means assignment to study groups was determined through randomization, rather than chosen by participants or investigators. Triple masking indicates that masking was recorded at that level, but the registry record does not identify which specific parties were masked or describe how masking was maintained. Parallel-group design means the groups were followed as separate treatment groups rather than participants crossing between assigned treatments. Further details of the randomization procedure, allocation concealment, adherence, and treatment discontinuation are not reported in the registry record.

The primary analysis population was an intent-to-treat (ITT) population defined in the registry as randomized subjects who received at least 1 dose of study drug and completed at least 1 heartburn diary entry during treatment. The record also says that all ITT populations excluded subjects with confirmed Barrett's esophagus and/or definite, but the remainder of that exclusion description is not reported in the registry record. Thus, the registry's use of ITT does not mean that every randomized subject was necessarily included in the posted primary analysis.

The primary outcome was assessed over 4 weeks using daily electronic diary data. The registry record lists 4 outcome measures and 6 analyses, but it does not report the full protocol, a sample-size justification, planned interim analyses, or a complete statistical analysis plan. Those omissions limit what can be inferred about prespecified decisions and how all recorded analyses relate to one another.

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: Subjects with endoscopically documented normal esophageal mucosa were enrolled in Dexlansoprazole Modified Release (MR) or Placebo once daily (QD) treatment group; subjects were instructed that lifestyle or behavioral modifications designed to treat their symptoms of gastroesophageal reflux disease (GERD) should not be altered throughout the study.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis was conducted on an intent-to-treat (ITT) population that included all randomized subjects who received at least 1 dose of study drug and completed at least 1 diary entry for heartburn during treatment. All ITT populations excluded subjects with confirmed Barrett's esophagus and/or definite dysplastic changes.

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 Percentage of Days With Neither Daytime Nor Nighttime Heartburn During Treatment as Assessed by Daily Electronic Diary-Median, measured over 4 weeks.

The primary endpoint was the median percentage of treatment days with neither daytime nor nighttime heartburn, assessed using a daily electronic diary over 4 weeks. The percentage was calculated from the days with no heartburn out of the total days for which either a daytime or nighttime result was marked. The measure therefore describes the portion of recorded treatment days that were free of both types of symptom, rather than the number of episodes or symptom intensity.

A diary-based endpoint can capture symptoms repeatedly during treatment and can distinguish symptom-free days from days with reported heartburn. Requiring a marked result for daytime or nighttime means that a day without the relevant diary information does not automatically count as a heartburn-free day. The registry record does not report further rules for incomplete diaries or how unmarked days were handled beyond the stated denominator definition.

The median is a measure of the distribution's center: it identifies the middle value when observations are ordered. The inter-quartile range (IQR) describes the spread of the central portion of the observed values. These summaries are useful when the distribution may be skewed or include values clustered near the boundaries of a percentage measure. They do not, on their own, convey every feature of the distribution or the experience of an individual participant. The registry record does not report a threshold for a clinically important difference on this endpoint, so the size of a numerical contrast must be considered alongside that missing context.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is summarized with a median and an inter-quartile range. A median is the central observation in an ordered distribution, while the IQR describes the range between the first and third quartiles. Together, these statistics convey a typical value and the spread of the central part of the observations. They are not a complete description of the distribution: values in the tails, patterns of clustering, and the shape of the distribution are not fully represented. For a percentage endpoint bounded by its scale, a median-based summary can be informative when values are not symmetrically distributed.

What does the Wilcoxon comparison assess?

The sponsor-reported primary comparisons used the Wilcoxon (Mann-Whitney) method. This rank-based test compares the ordering of observations across groups. It does not directly test a difference between the two reported medians in the same way that a test based on a mean difference would. A shift in ranks can arise from differences in location, spread, or distributional shape. Interpreting the test solely as a test of medians requires additional assumptions about how the distributions differ; the registry record does not report such assumptions.

The reported p-value measures how incompatible the observed test result is with the null hypothesis used by the test, under its statistical assumptions. A small p-value is evidence against that null, but it is not an estimate of effect size, a probability that the null is true, or evidence about clinical importance. The registry reports p < 0.001 for each active-dose comparison with placebo and p = 0.729 for the comparison between active doses. No primary effect estimate or confidence interval is reported alongside those tests.

How should a confidence interval be read?

A confidence interval describes the uncertainty around an estimated effect under a statistical procedure and its assumptions. The posted secondary recalculations report confidence intervals for mean differences in the nighttime-heartburn-free percentage. These intervals are unadjusted recalculations based on posted summary statistics, rather than sponsor-reported analyses. They quantify uncertainty for those secondary mean contrasts, not for the primary median endpoint. A confidence interval is not a guarantee that the true effect lies within its limits, and its interpretation depends on the method used and the validity of its assumptions.

For the secondary mean outcome, the posted recalculations use the difference between group means. This estimand differs from a comparison of medians: a mean summarizes the arithmetic average and can be more affected by extreme observations. A standard deviation describes variability among observations, whereas the standard error describes uncertainty in an estimated mean. The secondary record reports group means and standard deviations, and the recalculations use those summaries to estimate mean differences and confidence intervals. The reported unadjusted p-values are attached to these recalculated comparisons; the registry lists no sponsor-reported analyses for this secondary mean outcome.

Why does over-dispersion matter?

Over-dispersion is a general statistical issue in which observed variability is greater than a model assumes. If an analysis assumes less variability than is present, standard errors can be too small and confidence intervals too narrow, making statistical evidence appear stronger than it is. The posted secondary outcome is a bounded percentage with substantial variability, and the group standard deviations are reported. The registry record does not report a model-based over-dispersion assessment or an adjustment of that kind, so no such adjustment should be presumed. The secondary recalculations are based on summary statistics and their stated normal-approximation planning illustration; that illustration is not a model fitted to the trial's participant-level outcomes.

What do the planning illustrations mean?

Planning figures based on observed effects are illustrative calculations for a hypothetical future study, not observed power for the completed trial. Their assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect set to a stated fraction of the observed effect. For the 30 mg comparison, planning for 100% of the observed effect gives 42 participants per arm for 80% power and 56 participants per arm for 90% power; planning for 75% gives 74 and 99, and planning for 50% gives 167 and 223. For the 60 mg comparison, the corresponding planning figures are 55 and 74 for 100%, 98 and 131 for 75%, and 220 and 294 for 50% of the observed effect. These are participants per arm, not a claim about the trial's actual design assumptions or its achieved power.

Because the planning calculations use effects observed in this study, they can be optimistic if those estimates overstate the effect that would be expected in a future study. The registry record does not report the original sample-size justification, so the planning illustration cannot be used to reconstruct why the trial enrolled its stated population. Nor does a nominal power calculation establish that an observed result is reliable or clinically important.

10. Analyses Reported by the Sponsor

The sponsor's reported primary analysis is the Wilcoxon (Mann-Whitney) comparison of the diary-based primary endpoint. It reports p < 0.001 for placebo versus dexlansoprazole MR 30 mg QD, p < 0.001 for placebo versus 60 mg QD, and p = 0.729 for the comparison between the active-dose groups. These are the registry's reported analyses and should be distinguished from any separate recalculation using summary statistics.

No recalculated primary pairwise effect estimates or confidence intervals are available because the relevant two-group values are missing or not estimable for this median endpoint. The posted primary results therefore support interpretation through the group medians, IQRs, and sponsor-reported rank tests, rather than a separately recalculated median difference. The sponsor's own analysis takes precedence over recalculations when the analyses differ; the registry record does not flag a discrepancy.

The mean nighttime-heartburn-free endpoint has unadjusted recalculated comparisons based on posted group summary statistics, but the registry lists no sponsor-reported analyses for that outcome. Those calculations are not a reproduction or verification of the sponsor's primary analysis: they concern a different endpoint and a different summary measure. They should be read as supplementary, unadjusted analyses, with their assumptions and the absence of reported multiplicity handling in mind.

11. Multiplicity, Missing Data and Analysis Populations

The registry reports three pairwise primary comparisons: placebo with each dexlansoprazole MR dose and a comparison between the two active-dose groups. It does not report whether the statistical plan adjusted the significance threshold for these comparisons or for the broader set of outcome measures and analyses. When multiple hypotheses are tested, each additional test can increase the chance of at least one result appearing statistically unusual by chance. Adjustments or a prespecified testing hierarchy can address this concern, but no such procedure is described in the registry record. The reported p-values should therefore be understood as the sponsor's reported values, without assuming they were or were not multiplicity-adjusted.

Missing-data procedures are also not reported in enough detail to assess their influence. The primary ITT definition required at least 1 dose and at least 1 diary entry, so it did not include randomized subjects who failed either condition. The registry describes the endpoint denominator as days with either a daytime or nighttime result marked, but does not report further details about diary completeness, handling of missing days, imputation, or sensitivity analyses. The full exclusion wording for subjects with Barrett's esophagus and/or definite is incomplete in the record. These gaps make it difficult to evaluate the robustness of the analysis population and diary-derived summaries.

The record does not report the protocol's sample-size justification, planned interim analyses, or a complete statistical analysis plan. It is therefore not possible to assess whether the reported tests and populations match all prespecified decisions, whether analyses were selected after observing the data, or how unreported analyses were treated. The planning illustrations based on observed effects are not a substitute for the original design justification.

These omissions do not demonstrate that the analyses were inappropriate. They limit independent assessment of multiplicity, missing data, prespecification, and precision. In particular, the primary p-values are not accompanied by effect confidence intervals, and the registry does not report clinically important thresholds for the endpoint. Those limitations constrain how strongly statistical evidence can be translated into claims about the size or importance of a difference.

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

The posted primary results describe differences in diary-recorded heartburn-free days between groups during the 4-week treatment period. The active-dose groups had higher reported median percentages of days without either daytime or nighttime heartburn than placebo, and the sponsor's rank-based comparisons with placebo had p < 0.001. The registry also reports no statistically significant difference between the two active-dose groups in the Wilcoxon comparison, with p = 0.729. These statements describe the posted data and analyses; they do not establish that one dose is equivalent to the other.

A statistically small p-value does not say how much a participant's symptoms changed, how durable any difference would be, or whether the difference would be important to patients. The registry record does not define a clinically meaningful threshold for the primary percentage. The group medians and IQRs describe the distributions, but there is no reported primary effect estimate with a confidence interval to quantify the uncertainty around a contrast between groups.

The results apply to the registry's analyzed population and the recorded diary-based endpoint. The ITT population required receipt of at least 1 dose and at least 1 diary entry, and excluded subjects with specified esophageal conditions, although the complete exclusion criterion is not reported. The results do not automatically generalize to populations, settings, treatment durations, or outcomes not described in this record. The registry does not report longer-term symptom outcomes or a full account of adherence and missing diary data.

Safety information is also incomplete for drawing broad conclusions. The record lists adverse-event counts by group and selected serious-event terms, but does not provide the at-risk denominators for those counts or deaths. Counts without the corresponding denominator and fuller event ascertainment details do not allow reliable comparisons of event rates. Taken together, the posted record supports a statistical description of the specified symptom endpoint and sponsor-reported comparisons, but not medical advice or a definitive conclusion about overall benefit, risk, or clinical importance.

13. Why This Trial Matters Statistically

This trial illustrates why a clinical result needs more than a p-value. The primary endpoint is a repeated diary-based percentage summarized by medians and IQRs, while the sponsor's inferential method is rank-based. Understanding that pairing helps distinguish a distributional comparison from a direct estimate of a median difference. The very small reported p-values for the placebo comparisons indicate statistical evidence under the reported tests, but without a primary effect estimate and confidence interval the magnitude and precision of each contrast remain less clear.

The comparison of the two active-dose groups also demonstrates an important interpretive distinction: failure to find a statistically significant difference is not proof that the doses are equivalent. Equivalence requires an explicit margin and a design and analysis suited to that question. The registry does not report those elements. A reader should therefore interpret the non-significant comparison as an absence of strong evidence of a difference under the reported test, rather than as evidence that the dose outcomes are interchangeable.

The trial also provides a useful example of endpoint specificity. The primary measure required the absence of both daytime and nighttime heartburn, while the secondary mean measure concerns nighttime heartburn alone. These outcomes can be directionally consistent without being identical in meaning or analysis. Their overlap also matters when considering multiplicity: related outcomes and multiple comparisons should not automatically be treated as independent confirmations.

Finally, the registry's incomplete reporting of missing-data procedures, multiplicity, and the statistical analysis plan shows why transparency affects interpretation. The posted summaries allow readers to describe the observed results, but some questions about prespecification, precision, and robustness cannot be resolved from this record. Statistical significance is one part of evidence, not a substitute for those details or for a judgment about clinical importance.

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.

Several trial-specific reporting gaps limit interpretation beyond the posted summaries. The registry record does not provide primary pairwise effect estimates or confidence intervals, so the precision of the primary contrasts cannot be evaluated directly. The medians and IQRs describe each group but are not themselves estimates of the between-group effect. Nor does the record state a threshold for a clinically important change in the primary outcome.

The analysis population is described as ITT but excludes subjects who did not receive at least 1 dose or complete at least 1 diary entry. The record does not report the number excluded for these reasons or detail the handling of missing diary days. Its wording about exclusions for Barrett's esophagus and/or definite is incomplete. Without these details, the impact of exclusions and incomplete reporting on the analyzed population cannot be assessed.

Further limitations include the absence of a reported multiplicity procedure, full statistical analysis plan, original sample-size justification, and planned interim-analysis information. The registry does not explain which parties were masked or how masking was maintained. It also does not report detailed adherence or discontinuation information. These omissions prevent a full assessment of prespecification and potential sources of bias; they do not establish that any particular problem occurred.

For safety, the posted event counts lack corresponding at-risk denominators, and deaths are not reported in the record. Selected serious-event terms are listed, but that is not a complete account of comparative safety. The registry also does not provide further numerical findings for the listed secondary median outcomes. These limits mean the record supports a focused interpretation of the posted symptom summaries and tests, rather than a comprehensive appraisal of all outcomes or trial conduct.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Safety and Efficacy of Dexlansoprazole Modified Release Formulation to Treat Gastroesophageal Reflux Disease — NCT00321984.
• Registry record last updated 2011-04-28; status last verified 2011-04.
• Results first posted 2009-03-23.
• 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

NCT00321984 is a completed Phase 3 randomized, triple-masked, parallel-group study of dexlansoprazole MR 30 mg QD and 60 mg QD versus placebo in subjects with symptomatic, nonerosive GERD. Enrollment was 947. The study started in 2006-06 and reached primary completion in 2006-12.

The primary outcome was the median percentage of treatment days without either daytime or nighttime heartburn over 4 weeks, based on daily electronic diary entries. The registry reports higher group medians for both active-dose groups than for placebo, with Wilcoxon comparisons against placebo at p < 0.001; the comparison between active-dose groups had p = 0.729. The primary record does not report pairwise effect estimates or confidence intervals.

Secondary mean nighttime-heartburn-free results have unadjusted recalculations, while the registry lists no sponsor-reported analyses for that measure. Important details about multiplicity, missing-data handling, the full statistical plan, and safety denominators are not reported in the registry record.