Milnacipran (Savella) in Irritable Bowel Syndrome (IBS): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Milnacipran (Savella) in Irritable Bowel Syndrome (IBS) (NCT01471379), a randomized phase 2 study. The primary outcome is Number of Participants With Pain Response.

2. Clinical Question

3. Trial Design

The registry describes a randomized, double-blind, parallel Phase 2 study of milnacipran in IBS, with placebo included as a comparator. Its official title describes an efficacy assessment, while the brief summary says the investigators planned to compare pain and other patient-reported outcomes, including quality of life, psychological distress, and related psychosocial measures. The study status is TERMINATED.

Three arms are described. Group A was to begin with milnacipran 50mg BID in Phase I and increase to 100mg BID during Phase II. Group B was to remain on milnacipran 50mg BID for the stated 12 weeks. Group C was to receive placebo BID in Phase I and 50mg BID in Phase II. The registry does not explain the rationale for these dosing sequences or the details of the phase transition.

The planned eligibility description is women meeting Rome III diagnostic criteria for IBS. The registry records enrollment as 2, but the arm descriptions refer to n=20 in Phase I for Group A and Group C. Posted outcome counts are sparse: Group A has a reported arm size of 1.0, while Groups B and C have reported arm sizes of 0.0. The record does not reconcile these entries or establish how enrollment relates to the posted arm counts.

The recorded study start is 2012-04 and primary completion is 2013-02. Although the primary outcome timeframe is labeled Twelve Weeks, its description says the subject was terminated at the 10-week visit. The registry does not report a detailed account of the termination or the resulting analysis population.

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 undergo screening labs and questionnaires to make sure subjects are healthy and don't have any underlying conditions. Subjects who had clinically significant labs or Hospital Anxiety and depression scale(HADS)score more than 17 were excluded.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

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

Population: The study was terminated early resulting in a small number of subjects enrolled leaving the ability to analyze only certain categories.

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Pain Response, measured over Twelve Weeks.

The primary endpoint counts participants whose pain VAS score meets the registry's response definition. VAS scores range from 0-100 mm, with 0 meaning no pain and 100 the worst pain. Scores were recorded before the study, at 6 weeks of treatment, and at the final study visit, described as 10 weeks. A positive response is defined as a >30% decrease from baseline to the final visit. The outcome is therefore a binary classification: each assessed participant either meets that threshold or does not.

A threshold-based response can be easier to interpret than a change in average score alone because it expresses the outcome as the count of participants who meet a stated criterion. It also discards detail: participants just above or below the threshold are classified differently, even if their underlying changes are similar. That is a general feature of binary endpoints, not a finding specific to this trial.

The endpoint description has a timing inconsistency that matters for interpretation. The listed timeframe is Twelve Weeks, but the description states that the subject was terminated at the 10-week visit, where the final score was obtained. The record does not explain whether this timing applies to every participant or how the planned timeframe was adjusted. The posted primary result is 0.0 for Group A with a reported arm size of 1.0. No primary outcome value is recorded for Groups B or C, so the posted data do not support a between-arm estimate.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Only a limited set of statistical information is recorded. The study is identified as randomized, double-blind, and parallel, and the primary endpoint is a binary count of participants meeting a pain-response threshold. The registry lists no analyses. It does not report a detailed statistical analysis plan, a method for handling missing outcomes, or an approach to adjustment for multiple outcomes. Those omissions make it important to distinguish the recorded result from any broader inference.

What does a binary endpoint represent?

For this endpoint, each participant's baseline and final VAS scores are used to classify pain response according to the stated >30% decrease criterion. The result is a count of responders rather than a continuous summary of all score changes. A count can be summarized as a proportion when the relevant denominator is known and comparable. In this record, outcome values and arm sizes are absent for two arms, so a complete arm-by-arm summary cannot be made. This is a general explanation of binary outcomes, not an additional trial result.

Binary outcomes can be useful when a threshold has a clear clinical interpretation, but they depend on the chosen threshold and assessment time. Small differences around the cutoff may change a participant's classification, while the size of the underlying score change is not reflected in the binary count. The registry gives the threshold and the VAS scale, but does not report a justification for choosing this response definition.

Why is a treatment comparison unavailable?

A comparative analysis requires outcome information from the arms being compared, as well as denominators appropriate to the analysis. Group A has a posted response count of 0.0 and an arm size of 1.0; Group B and Group C have arm sizes of 0.0 and no posted primary outcome values. The registry marks the binary two-arm comparison as not estimable because values are missing. Without the required arm-level observations, a treatment difference or uncertainty interval cannot be calculated from the registry entries.

In general, a confidence interval describes the range of effect estimates compatible with the data and the statistical model, subject to its assumptions. It communicates uncertainty as well as the estimated effect. No such interval is reported for this endpoint. With the comparator result absent, there is no interval to interpret, and the one recorded arm count does not substitute for a between-arm analysis.

What can planning illustrations tell readers?

The registry describes planning illustrations based on an observed effect, a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says that the effect to detect is a stated fraction of the observed effect, and that non-time-to-event outputs are expressed as participants per arm. No planning values are included in the registry record for this interpretation. Such sample-size illustrations are planning scenarios, not observed power and not evidence that a trial did or did not have adequate power. When an observed effect is used as the basis for a planning scenario, the estimate itself may be highly uncertain, especially when the underlying outcome record is sparse.

General statistical context: power is a design property describing the probability that a specified analysis will detect a defined effect under assumptions. It is not a measure of how likely a result is true after observing data. An observed-power calculation based on the effect estimate does not repair missing outcomes or create a valid treatment comparison. The registry's planning illustrations should therefore be kept separate from interpretation of the posted primary outcome.

Why do allocation and masking not resolve the data gaps?

Randomization is generally intended to make treatment groups comparable with respect to both measured and unmeasured characteristics, on average, when allocation is properly implemented. Double masking is generally intended to reduce the influence of treatment knowledge on participants' reports and study assessments. These design features do not ensure that outcomes are observed, that all randomized participants are represented in a particular analysis, or that the results are precise. The registry does not report allocation procedures, masking details, or an analysis population definition.

The design is labeled parallel, but Group C is described as moving from placebo to milnacipran during Phase II. The record does not explain the design implications of this sequence or specify how the primary comparison would be defined over time. No assumptions about the analysis should be inferred from the arm descriptions alone.

How should over-dispersion be understood?

General statistical context: over-dispersion occurs when variability in count data is greater than assumed by a model. If present, it can make standard errors too small and confidence intervals too narrow unless the analysis accounts for it. This primary endpoint is recorded as a participant count with a binary response definition, and the registry does not report an over-dispersion assessment or a model-based analysis. There is no basis in the record to say that over-dispersion was observed or that a particular correction was used.

Overall, the registry allows the endpoint definition and a limited arm-specific count to be described. It does not provide enough outcome information to evaluate a comparative statistical effect, its precision, or the analysis assumptions that would support such an estimate.

10. Exact Estimates for Each Arm

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

R code and output

arm    <- c("Group A (50mg - 100mg)")
n      <- c(1)
events <- c(0)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Group A (50mg - 100mg)  1  0  0.0  0.0  97.5  0.0  79.3  1.3  84.2

11. Multiplicity, Missing Data and Analysis Populations

The registry names four secondary outcomes: IBS-QOL, subject self-reported adequate relief of pain, the Treatment Efficacy Questionnaire, and dose-related incremental benefit in pain reduction based on VAS. It also records five outcome measures overall. No results for these secondary outcomes are included in the posted information. The analysis record lists no analyses, and a plan for controlling or interpreting multiplicity is not reported in the registry record.

Multiplicity is a general statistical concern when several outcomes or comparisons are examined: the chance of at least one apparently notable finding can increase as the number of statistical tests grows. Methods such as prespecifying a primary outcome or adjusting for multiple comparisons can address different aspects of this issue, but the registry does not state whether such methods were planned or used. The primary outcome can be identified from the record, but its analysis strategy is not detailed.

The record also does not report how missing outcome data were handled, whether outcomes were collected after termination, or which analysis population was intended. Those details matter because an analysis limited to participants with recorded outcomes may differ from one that follows all randomized participants, particularly if missingness is related to participant characteristics or outcomes. General statistical methods for missing data rely on assumptions; no specific assumption or method is reported for this study.

The posted primary data are incomplete across arms, and the registry itself indicates that a binary comparison is not estimable. Without an analysis population, missing-data strategy, and comparable arm outcomes, the treatment contrast and its uncertainty cannot be assessed. The record does not report whether the secondary measures were analyzed or how any multiple-outcome findings would have been interpreted.

12. 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 1% reporting threshold.

Most frequent non-serious adverse event terms

The serious adverse event entries show 0 affected among 1 at risk in Group A and 0 affected among 1 at risk in Group B. Group C has 0 at risk and 0 affected. These are registry counts, not comparative safety estimates. In particular, the Group C entry has no participants at risk recorded, so it does not provide an observed placebo safety denominator.

Zero serious adverse events recorded in an arm with a nonzero at-risk count means none were recorded among the listed at-risk participants. It does not establish that the risk is absent or precisely estimated. The recorded counts are very limited, and no comparative analysis is reported. Deaths are not reported in the registry record.

For other adverse events, Group A has 1 affected among 1 at risk. Group B and Group C each have 0 at risk and 0 affected. The registry does not provide further event descriptions in the serious-event terms list. These entries should be read as sparse reporting, not as a basis for comparing adverse-event rates across regimens.

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

The posted primary outcome supports a narrow descriptive statement: Group A has a recorded pain-response count of 0.0, with a reported arm size of 1.0. It does not support an estimate of how the milnacipran regimen compares with the placebo sequence or the other milnacipran regimen. Group B and Group C have no posted primary outcome values, and the registry indicates that a binary two-arm comparison is not estimable.

This absence of a comparative estimate is not evidence that the treatments are equivalent, nor does it establish that one regimen is better or worse. The record lacks the arm-level outcome data required to assess a difference and its precision. There is no confidence interval or statistical test result to use in judging the range of effects compatible with the data.

The timing also constrains interpretation. The endpoint is labeled Twelve Weeks, while the description identifies the final visit as 10 weeks because the subject was terminated at that visit. The registry does not clarify whether this was the only available final assessment or how it relates to the intended timeframe. The result should therefore be understood in the context of the recorded assessment description rather than as a fully documented 12-week comparison.

The record's TERMINATED status, enrollment entry of 2, and sparse arm outcome entries do not establish why the study ended or what conclusions investigators reached. The registry does not report those details, a full analysis plan, or results for the named secondary outcomes. Statistical interpretation should remain limited to what is recorded: a trial design and endpoint definition, one arm-specific primary count, and an unavailable comparative analysis.

In general, absence of an estimable comparison is different from a well-estimated null effect. A null or small estimate with a sufficiently informative confidence interval can constrain plausible effects; missing comparator outcomes cannot do that. This is a general distinction and does not add a result to this trial. The record provides too little comparative information to characterize treatment effects or their precision.

14. Why This Trial Matters Statistically

This trial illustrates why a registry's outcome definition and its actual posted results need to be considered together. The primary endpoint is described clearly enough to identify its construct: a binary pain-response count based on a specified decrease in VAS score from baseline to the final visit. But the posted result is available for only one arm, and the record states that the binary comparison is not estimable. A precise endpoint definition cannot compensate for missing comparative observations.

The entries also show how design descriptions and analysis data answer different questions. Randomization, double masking, and a placebo comparator describe planned design features. They do not demonstrate that the intended comparison was completed or that results are available for every arm. The record's enrollment and arm entries, together with its terminated status, make it especially important not to treat the planned design as a completed comparative analysis.

The outcome timing is another statistical issue rather than a minor label. A Twelve Weeks timeframe appears alongside a description of a final visit at 10 weeks. Endpoint timing determines which observations are being compared, so the registry's unexplained timing difference limits interpretation of the recorded count.

Finally, the trial's listed secondary outcomes and planning assumptions do not fill the primary data gap. Without posted secondary results, a declared multiplicity strategy, or a documented missing-data approach, readers cannot assess whether other analyses were performed or how robust any findings might be. The record is therefore a useful example of the distinction between a planned research question and an estimable result: the former is described, while the latter is not available as a between-arm estimate.

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 is limited by sparse and internally unexplained enrollment and arm reporting. Enrollment is recorded as 2, the arm descriptions mention n=20 for Group A and Group C during Phase I, and the posted arm sizes are 1.0, 0.0, and 0.0. The record does not explain how these entries relate to one another, so they cannot be used to reconstruct the number of participants in the primary analysis or the intended sample.

The primary result is incomplete across arms, and the registry identifies the binary comparison as not estimable. The endpoint timeframe is listed as Twelve Weeks, while the outcome description says the subject was terminated at the 10-week visit. The record does not clarify the timing for each arm or describe additional follow-up.

The study status is TERMINATED, but the reason for termination is not reported in the registry record. The record also does not report a detailed statistical analysis plan, a missing-data method, the analysis population, or a multiplicity approach. Four secondary outcomes are named, but results for them are not included in the information available. These gaps prevent evaluation of secondary evidence, sensitivity to missingness, and the precision of a comparative primary estimate.

The safety entries are also sparse, including arms with 0 at risk recorded, and deaths are not reported in the registry record. Finally, the registry's description of Group C changes from placebo in Phase I to milnacipran in Phase II, but does not explain how that sequence was to be handled in the primary comparison. These limitations constrain interpretation of both efficacy and safety reporting.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Milnacipran (Savella) in Irritable Bowel Syndrome (IBS) — NCT01471379.
• Registry record last updated 2017-04-13; status last verified 2017-03.
• Results first posted 2013-12-24.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry describes a randomized, double-blind, parallel Phase 2 study of milnacipran strategies and placebo for IBS. The planned population is women meeting Rome III criteria. The study is recorded as TERMINATED, with enrollment of 2, a start date of 2012-04, and primary completion in 2013-02.

The primary endpoint counts participants with a >30% decrease in VAS pain score from baseline to the final visit. The timeframe is listed as Twelve Weeks, while the outcome description identifies a 10-week final visit. Group A has a posted primary value of 0.0 and arm size of 1.0; Groups B and C have arm sizes of 0.0 and no primary outcome values. A two-arm binary comparison is identified as not estimable. Four secondary outcomes are listed, but their results and a detailed analysis plan are not reported in the registry record.