Tragus Stimulation to Prevent Atrial Fibrillation After Cardiac Surgery: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Tragus Stimulation to Prevent Atrial Fibrillation After Cardiac Surgery (NCT03392649), a randomized na study. The primary outcome is Time to the First Episode of Atrial Fibrillation (AF) >30 Seconds.

2. Clinical Question

The clinical question concerns the timing of a first qualifying episode, rather than whether any atrial fibrillation occurred in the full randomized groups. That distinction matters: the primary endpoint description defines its population as patients who had atrial fibrillation, and the registry record does not report how many patients in each group met that condition. The outcome was assessed through patient-chart review and inpatient telemetry, from the end of surgery to the end of hospitalization, up to 1 month.

The registry summary describes a rationale involving autonomic stimulation and inflammation, but the posted primary result addresses a narrower statistical question: whether the observed time distributions differed between groups. It does not, by itself, answer every question about occurrence, total burden, or consequences of atrial fibrillation. Those are separate outcomes, and the registry lists several secondary measures in addition to the primary endpoint.

3. Trial Design

TraP-AF was a randomized, parallel-group study sponsored by the University of Chicago. The registry lists single masking; it does not report who was masked. The study status is completed, and its phase is listed as NA. The registry record gives an enrollment of 81. The posted primary outcome results include 40 participants in the stimulation group and 41 in the sham group.

After cardiac surgery, the stimulation group received continuous Parasym stimulation at the tragus for 48 hours. The sham group had the clip placed on the tragus, but the device was not turned on. In both groups, the clip was moved to the other ear every 4 hours. The registry describes the eligible surgery as coronary artery bypass graft and/or heart valve replacement surgery.

The primary outcome was time from the end of surgery to the first atrial fibrillation episode lasting longer than 30 seconds, assessed from chart review and inpatient telemetry. Its stated timeframe was from the end of surgery to the end of hospitalization, up to 1 month. The record also lists secondary outcomes involving atrial fibrillation burden, medications and support used during atrial fibrillation, stroke or transient ischemic attack, and all-cause mortality.

The study began on 2018-03-09, and primary completion was recorded as 2021-07-30. The registry record does not report the planned sample-size justification, detailed eligibility criteria, or the full analysis populations. Those omissions limit what can be inferred about how the observed enrollment related to the original design or how participants were handled in each analysis.

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

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 Time to the First Episode of Atrial Fibrillation (AF) >30 Seconds, measured over From end of surgery to end of hospitalization, up to 1 month..

The primary endpoint measures elapsed time from the end of surgery to a patient's first atrial fibrillation episode lasting longer than 30 seconds. It is a time-to-event measure, but the registry reports it as a median with an inter-quartile range, rather than as a hazard ratio or a survival probability. The endpoint was assessed using chart review and inpatient telemetry during hospitalization, up to 1 month.

Time to first event can capture when an event occurs, not only whether it occurs. For this outcome, however, the stated endpoint description applies to patients who had atrial fibrillation. The posted medians therefore describe the timing among that described group; they should not be read as the proportion of all randomized patients who developed atrial fibrillation. The registry record does not report the number of patients with atrial fibrillation in each group or clarify whether the analysis included only those patients or used another handling of participants without an event.

The reported median was 53.1 hours in the stimulation group, with an inter-quartile range from 39.8 to 77.3. In the sham group, the median was 52.2 hours, with an inter-quartile range from 28.7 to 62.3. These summaries describe the observed distributions; the overlap in the ranges is not itself a formal test of a group difference. The reported Wilcoxon (Mann-Whitney) comparison, with p = 0.60, is the registry's inferential test for the primary comparison.

In general, a median identifies the central point of an observed distribution, while an inter-quartile range describes its middle spread. Neither summary alone conveys the full distribution or establishes a causal effect.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the primary test compare?

The registry reports a Wilcoxon (Mann-Whitney) test for the primary comparison. In general, this rank-based test compares the ordering of observations across groups. It does not require the data to follow a normal distribution, and it is not simply a test of whether two reported medians are equal. Its interpretation concerns the relative distributions, subject to the study's analysis and assumptions. The record reports p = 0.60 but does not provide a test statistic, a confidence interval for the group contrast, or further details about handling of observations without an event.

The median and inter-quartile range are descriptive summaries. The median is the central value of a distribution; the inter-quartile range spans its middle portion. A broad or overlapping inter-quartile range can indicate spread, but overlap is not a substitute for a formal comparison. Conversely, a p-value alone does not show the range of group differences consistent with the data. A confidence interval for a contrast would help express that uncertainty, but one is not reported for this primary result.

How should the binary outcomes be read?

For binary outcomes, a risk difference expresses the difference between the observed proportions in the groups. The recalculated unadjusted risk difference is positive for each of the three posted medication or support outcomes, meaning the observed proportion is higher in the stimulation group for those outcomes. The corresponding confidence intervals extend across zero, so the recalculated summaries are compatible with no difference as well as with differences in either direction. They are not adjusted for other factors.

A confidence interval is a range of values that, under the method's assumptions, is compatible with the data and the chosen confidence level. It is not a guarantee that the true value lies in that particular interval. Wider intervals indicate less precision. In this study, the posted binary outcomes involve relatively few events, so uncertainty is substantial. General statistical reasoning: sparse binary data can make estimates unstable, and a small change in observed counts may materially affect a comparison.

The registry lists Fisher Exact tests for the binary secondary comparisons. In general, Fisher's exact test evaluates evidence of association in a two-by-two table and is often used when counts are small. The sponsor-reported p-values are 0.09 for rate-control medication, 0.16 for antiarrhythmic medication, and 0.67 for inotropes. Recalculated unadjusted p-values are 0.214, 0.132, and 0.379, respectively. These are not identical because the registry's reported tests and the simple recalculations are not necessarily based on the same inferential method or conventions. The recalculated values are explicitly unadjusted and do not replace the sponsor's reported analysis.

Why does over-dispersion matter?

General explanation: over-dispersion occurs when variation in observed data exceeds what a statistical model assumes. It can make model-based standard errors too small and confidence intervals too narrow if it is ignored. The primary endpoint is reported with medians and an inter-quartile range, and the registry does not report a model-based estimate for it. The record does not report a dispersion assessment or a model for the secondary outcomes, so no conclusion about over-dispersion in this trial can be drawn.

What do the planning illustrations mean?

The sample-size illustrations are planning calculations based on observed effects, not evidence of the trial's actual power and not a retrospective measure of whether the trial succeeded. The stated assumptions are a two-sided significance level of 0.05, equal allocation, and a normal approximation. For the binary outcomes, the figures are participants per arm under scenarios targeting 100%, 75%, or 50% of the observed effect. For example, the rate-control medication illustration gives 206 participants per arm for 80% power when planning for 100% of the observed effect, and 358 participants per arm when planning for 75% of it. These are hypothetical future-study planning values, not observed enrollment requirements or a justification reported for TraP-AF.

Observed effects from a small study can be imprecise, and using them directly for planning may give an unstable target. A smaller target effect generally requires a larger study to detect under otherwise similar assumptions. The illustrations do not account for every possible design feature, such as loss to follow-up, multiplicity, or a different analysis plan. The registry record does not report the study's own sample-size assumptions, so the planning calculations should not be mistaken for the original design rationale.

10. Analyses Reported by the Sponsor

For the primary endpoint, the registry reports the sponsor's Wilcoxon (Mann-Whitney) comparison: p = 0.60. The posted group summaries are medians of 53.1 hours for stimulation and 52.2 hours for sham, with their respective inter-quartile ranges. No recalculated primary contrast or confidence interval is reported in the registry record, so there is no separate recalculated primary estimate to compare with the sponsor's test.

For the binary secondary outcomes, the registry reports Fisher Exact tests with p-values of 0.09 for rate-control medication, 0.16 for antiarrhythmic medication, and 0.67 for inotropes. Simple unadjusted recalculations from the posted counts produce p-values of 0.214, 0.132, and 0.379, respectively, alongside risk-difference intervals. Different testing procedures and conventions can yield different p-values, particularly when counts are sparse. The recalculations are not adjusted analyses and should be distinguished from the registry-reported tests.

The registry does not identify why the numerical p-values differ or provide enough detail to reproduce every aspect of the sponsor's analysis. The recalculated comparisons are therefore best read as supplementary, unadjusted descriptions of the posted counts, not as corrections to the sponsor's results. Neither set of secondary p-values supplies a clear account of multiplicity or resolves the uncertainty in the observed effects.

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary endpoint and multiple secondary outcomes, but a plan for handling multiplicity is not reported in the registry record. Multiplicity arises when several outcomes or comparisons are tested: even if no real effects exist, repeated testing can increase the chance that one result appears unusual by chance. General statistical practice addresses this through a prespecified hierarchy, adjusted significance thresholds, or careful emphasis on estimation and uncertainty. The record does not say whether any such strategy was planned or used.

Missing-data procedures and analysis populations are also not reported in the registry record. The primary outcome is described among patients who had atrial fibrillation, but the record does not give the number meeting that condition in each group or explain how patients without an episode were treated in the time analysis. It also does not describe exclusions, protocol deviations, or whether analyses followed randomized assignment. These details are important for understanding what population the primary comparison represents.

The outcome was assessed by chart review and inpatient telemetry through the end of hospitalization, up to 1 month. The registry record does not report how incomplete telemetry, early discharge, or unavailable chart information were handled. Nor does it report a sensitivity analysis for missing or uncertain event times. Absence of such information in the record should not be interpreted as proof that data were missing or that a particular method was used.

These reporting gaps affect both interpretation and reproducibility. The reported p-values and group summaries can be described as posted, but readers cannot reconstruct all analysis decisions or assess the consequences of unreported missingness and multiplicity. A more detailed protocol and analysis plan would be needed to evaluate those issues fully.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: stroke or transient ischemic attack (tia) (2.5%).

The registry lists serious adverse events for 1 participant in the stimulation group and 1 participant in the sham group. It identifies stroke or transient ischemic attack as a serious adverse event term at 2.5%; the record does not assign that percentage to a specific group in the information reported. Other adverse events are listed for 0 participants in the stimulation group and 1 participant in the sham group. The registry also lists 3 deaths in the stimulation group and 0 in the sham group.

These figures are descriptive counts, not a comparison adjusted for chance or participant characteristics. The record does not report whether events were judged related to the device or procedure, their timing, or the causes of death. A difference in observed counts cannot establish that the intervention caused an event, and the absence of a listed event in a group does not establish absence of risk. With the group sizes in this study, safety comparisons are especially uncertain. The serious-event and mortality figures should therefore be read as registry-reported observations, not as proof of harm or safety.

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

The primary result describes a comparison of time to a first atrial fibrillation episode lasting longer than 30 seconds among patients who had atrial fibrillation. The group medians were 53.1 hours for stimulation and 52.2 hours for sham, and the reported test gave p = 0.60. These data do not show a clear statistical separation in the primary time comparison. They also do not prove that the groups are equivalent or rule out effects that the study could not estimate precisely.

The endpoint is conditional on atrial fibrillation, as worded in the registry, and is not a direct estimate of the probability of developing atrial fibrillation in all randomized participants. The registry record does not provide the group counts for occurrence, so the primary summaries cannot answer that separate question. Nor do they provide a complete picture of episode burden, later recurrence, or outcomes after hospitalization.

General statistical interpretation: a non-small p-value is not evidence that there is no effect. It indicates that the observed result is not unusual enough under the test's null model to provide strong evidence against that model. Precision is better judged using an interval for the group contrast, which is not reported for the primary endpoint. For the binary secondary outcomes, recalculated intervals include no difference and are wide enough to reflect considerable uncertainty.

The results also do not establish that tragus stimulation prevents atrial fibrillation, shortens episodes, or changes other clinical outcomes. Those claims would require appropriate outcome estimates, analysis details, and precision. Conversely, the posted findings do not establish that stimulation has no possible effect. The study's observed data should be interpreted within its randomized design, its endpoint definitions, the limited enrollment, and the reporting gaps described in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why endpoint definitions are central to interpreting a clinical comparison. A time-to-first-event outcome among participants who experienced atrial fibrillation answers a different question from the occurrence of atrial fibrillation among everyone randomized. The distinction can be easy to miss when the clinical motivation concerns prevention, but the primary measure and population determine what the estimate can support.

It also illustrates the value of separating a descriptive summary from an inferential result. Group medians and inter-quartile ranges describe observed timing and spread; the rank-based test assesses a distributional comparison. Neither the medians alone nor the p-value alone conveys a complete account of effect size and precision. A confidence interval for the primary contrast would have helped readers judge the range of differences consistent with the observed data.

The secondary outcomes demonstrate another recurring issue in biostatistics: many outcomes can produce a complicated pattern of estimates and p-values. Recalculated binary comparisons are unadjusted and imprecise, while the registry does not report a multiplicity strategy. Treating isolated secondary results as confirmatory would therefore be unwarranted. The planning illustrations also show why planning targets are distinct from observed power: estimates based on observed effects are hypothetical design scenarios, not a measure of what the completed trial was capable of establishing.

Finally, a registry record can contain useful group summaries while omitting important details about analysis populations, missing data, and prespecified methods. Recognizing those limits is part of statistical interpretation. It helps separate what the posted results directly describe from 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 primary endpoint's stated population is patients who had atrial fibrillation, but the registry record does not report the number of such patients in each group. This limits interpretation of the time summaries and prevents them from being read as group-wide occurrence rates. The record also does not report a confidence interval for the primary group contrast, leaving its precision less fully characterized than the descriptive group ranges.

The primary outcome was assessed during hospitalization, up to 1 month. The registry does not report follow-up beyond that period for this outcome, or explain how incomplete telemetry, missing chart information, or patients without a qualifying episode were treated. The analysis population and detailed handling of censoring are not reported in the registry record.

Enrollment was 81, with posted primary groups of 40 and 41. The relatively limited number of participants and the sparse counts in several secondary outcomes contribute to imprecision. Results for some listed secondary outcomes are not reported in the registry record. A multiplicity strategy, sample-size justification, and detailed statistical analysis plan are also not reported, so it is not possible to assess the relationship between the completed study and its planned design in detail.

The registry identifies single masking but does not say who was masked. It also does not report whether outcome assessment was independent of treatment assignment or whether clinical decisions about medication and support were standardized. These limitations do not invalidate the posted summaries, but they restrict assessment of bias, reproducibility, and the strength of conclusions that can be drawn from them.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Tragus Stimulation to Prevent Atrial Fibrillation After Cardiac Surgery — NCT03392649.
• Registry record last updated 2022-08-04; status last verified 2022-07.
• Results first posted 2022-08-04.
• 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

TraP-AF was a randomized, parallel-group study of Parasym tragus stimulation versus a sham procedure after cardiac surgery. The registry lists single masking and an enrollment of 81. Stimulation was continuous for 48 hours; in the sham group the device was not turned on. In both groups, the clip was moved to the other ear every 4 hours.

The primary endpoint was time to a first atrial fibrillation episode lasting longer than 30 seconds among patients who had atrial fibrillation. The median was 53.1 hours in the stimulation group and 52.2 hours in the sham group; the registry-reported Wilcoxon (Mann-Whitney) test gave p = 0.60. This comparison does not show a clear statistical separation, but the record does not report a confidence interval for the primary contrast or the number of patients with atrial fibrillation by group.

The registry lists secondary outcomes involving atrial fibrillation burden, treatment and support, stroke or transient ischemic attack, and mortality. It does not report results for all listed secondary outcomes or key details about multiplicity, missing data, and analysis populations. The posted results are informative as descriptive observations but leave important uncertainty about precision and interpretation.