A Study to Assess the Effectiveness and Side Effects of GSK2798745 in Participants With Chronic Cough: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Assess the Effectiveness and Side Effects of GSK2798745 in Participants With Chronic Cough (NCT03372603), a randomized phase 1, phase 2 study. The primary outcome is Total Cough Counts During Day Time Hours Following 7-days of Dosing.

2. Clinical Question

This is a question about a short treatment period and a monitored cough-count outcome, not about long-term symptoms or outcomes beyond the registry’s specified measurement window. The record describes GSK2798745 as a TRPV4 channel blocker being investigated for chronic cough. That description supplies the study context, but the posted endpoint comparison is the statistical evidence available for this question.

The crossover arrangement means participants were assigned to receive both treatments, in different orders. It allows the treatment comparison to use measurements from the same participants across treatment periods. The record does not report how many participants were assigned to each sequence or give detailed information about participants’ clinical characteristics.

3. Trial Design

The registry identifies this as a randomized, double-blind, placebo-controlled crossover study in participants with chronic cough. It lists Phase 1 and Phase 2. The study began on 2018-04-05, had a primary completion date of 2018-10-08, and is recorded as terminated. The reason for termination is not reported in the registry record.

Each participant was assigned to a sequence of two treatment periods: placebo followed by GSK2798745, or GSK2798745 followed by placebo. The periods were separated by a washout period of 14 to 21 days. Each treatment period lasted 7 days. The GSK2798745 regimen was 4.8 mg on Day 1 followed by 2.4 mg once daily for 6 days; the placebo regimen was matching tablets once daily for 7 days. The registry record describes oral administration.

The enrollment recorded for the study is 17. The study summary states that a maximum of 48 subjects would be enrolled, but the registry record does not report a sample-size justification or say how the enrollment target was determined. It also does not report counts by treatment sequence. The primary efficacy analysis included participants who took at least 1 dose and had data at the specified time points; participants were analyzed according to the treatment they actually received.

The title describes the study as double-blind and sponsor open. The record does not explain which sponsor personnel had access to treatment assignments or how masking was maintained for participants and investigators. It describes the primary analysis as a Bayesian mixed model that included treatment, period, and baseline terms. Further details of the model specification, including any assessment of carryover, are not reported in the registry record.

Secondary outcomes included adverse events, serious adverse events, and laboratory measures. The study status and posted outcomes should be read together: the registry records a terminated study and reports the primary endpoint result, but it does not give a reason for termination or a detailed account of the decision to end the study.

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: A total of 34 participants were screened of which 17 failed screening and 17 participants were included in the study.

Treatment Period 1 (Up to 7 Days)

Wash-out Period (14-21 Days)

Reasons for not completing this period:

Treatment Period 2 (Up to 7 Days)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All Subjects Population. Only those participants with data available at the specified data points were analyzed.

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 Total Cough Counts During Day Time Hours Following 7-days of Dosing, measured over Up to 10 hours post-dose on Day 7 of each treatment period.

The primary endpoint was total cough counts during daytime hours after 7 days of dosing. Coughs were monitored with the VitaloJAK cough monitor. On Day 7 of each treatment period, the monitoring window began immediately after dosing and extended up to 10 hours. The endpoint therefore measures cough frequency during a defined observation period, rather than relying solely on a participant’s retrospective symptom report.

The endpoint description says cough counts were log-transformed before analysis. It reports a posterior median and 95% credible interval from a Bayesian mixed model. The posted placebo estimate is 180.6, with an interval from 137.9 to 235.4; the GSK2798745 estimate is 241.1, with an interval from 181.4 to 320.6. The reported treatment comparison is a ratio of 1.336, with a 90% confidence interval from 0.965 to 1.847.

General statistical context: a cough monitor provides a count-based measure over a specified period, making it possible to compare observed cough frequency between treatment periods. The log transformation is a modeling approach often used when an analysis is framed in relative terms; the resulting ratio communicates a relative comparison rather than an absolute difference in cough counts. The recorded outcome is relevant to the study question because it directly quantifies coughing during the stated daytime monitoring window. It does not, by itself, describe cough at other times or establish how participants felt about their symptoms.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the crossover model contribute?

The study used a crossover design: participants received placebo and GSK2798745 in separate periods, with treatment order randomized. General statistical context: a crossover comparison can account for the fact that each participant contributes measurements under both treatments. This can reduce the influence of stable differences between participants on the treatment comparison. It does not remove uncertainty, and its validity depends on appropriate handling of period effects and on the relevance of assumptions about effects carried between periods.

The registry says the Bayesian mixed model adjusted for treatment and period, as well as subject-level and period-adjusted baselines. The record does not say whether carryover was assessed or included in the model. Nor does it provide a complete model formula, details of the prior beyond calling it non-informative, or diagnostic information. Those details matter when judging how well the analysis reflects the crossover structure, so they cannot be inferred from the brief model description.

Why transform cough counts?

The primary endpoint is a count, and the record says total cough counts were log-transformed before analysis. General statistical context: count measurements can be unevenly distributed, and a logarithmic transformation can make a relative treatment comparison more suitable for modeling. A ratio is a natural way to express a relative comparison. It is different from an absolute difference: the ratio describes how one treatment’s count relates proportionally to the other, whereas a difference would describe the gap in counts.

The registry reports a treatment ratio of 1.336 and a 90% confidence interval from 0.965 to 1.847. The primary endpoint description also says the model reports posterior medians and 95% credible intervals. The registry’s wording thus refers to Bayesian summaries for the endpoint and a confidence interval for the posted treatment ratio. The exact derivation of that ratio and its interval is not described in greater detail. No p-value is posted for the primary comparison.

How should interval estimates be read?

General statistical context: a confidence interval describes the uncertainty associated with an estimate under a specified statistical procedure. A Bayesian credible interval instead summarizes the posterior distribution under the model and prior. Neither interval is a measure of clinical importance by itself. Their width and position relative to a no-change value help show how precisely the data locate the effect and what directions remain plausible under the stated analysis.

The ratio interval extends across the no-change value. This means the reported analysis does not isolate a single direction for the relative cough-count comparison. The point estimate is one value within that uncertainty range, not a definitive result independent of the interval. The arm-level intervals likewise reflect substantial uncertainty around the reported medians. Since the record labels the endpoint summaries as posterior medians and credible intervals, those summaries should not be interpreted as if they were simple unadjusted means or as a substitute for the model-based comparison.

What do baseline adjustments mean?

The model description defines subject-level baseline as the mean of the two period-specific baselines. It defines period-adjusted baseline as the difference between a period-specific baseline and that subject-level baseline for each period. General statistical context: including baseline measures can account for differences in cough counts before treatment, while separating subject-level and period-specific information can distinguish a participant’s general starting level from variation between periods. The registry does not give the baseline values or explain the estimated contribution of each baseline term.

The model also adjusted for period. General statistical context: outcomes can differ across periods for reasons unrelated to treatment, so period adjustment is relevant in crossover studies. The record does not describe any sequence effect, treatment-by-period interaction, or sensitivity analysis. Their absence from the registry description does not prove that they were not considered; it means the registry record does not report them.

What does the analysis population represent?

The primary endpoint description identifies an All Subjects Population of randomized participants who took at least 1 dose of study treatment. It also says that only participants with data available at the specified time points were analyzed, and that participants were analyzed according to treatment actually received. These details define the analysis population more narrowly than enrollment alone. The posted arm summaries give 17 for placebo and 15 for GSK2798745. The registry does not provide a detailed flow of participants through randomization, treatment periods, monitoring, and analysis.

General statistical context: excluding participants without endpoint data can affect an estimate if missingness is related to treatment or outcome. The record does not report how missing cough-monitor data were handled, whether any values were imputed, or whether a sensitivity analysis examined missingness. It also does not report a separate intention-to-treat analysis. These omissions limit assessment of whether the available measurements represent all randomized participants in a way that preserves the benefits of randomization.

How do the secondary re-calculations differ?

For selected continuous laboratory outcomes, unadjusted re-calculations compare posted group summary statistics. General statistical context: such calculations are not the same as a trial’s adjusted analysis, and they do not account automatically for the crossover model or other covariates. The registry does not report sponsor analyses for those selected secondary outcomes. The re-calculations should therefore be read as descriptive comparisons of the posted summaries, not as replacements for a prespecified analysis.

The posted planning illustrations associated with some re-calculations are not evidence of observed power and do not reconstruct the trial’s original design assumptions. They illustrate hypothetical planning based on the observed effects and stated assumptions. The registry record does not report the original sample-size justification, the intended power, or any interim-analysis plan. More generally, estimates from a small study may be imprecise; planning values based on such estimates should not be mistaken for reliable forecasts of the effect in a future study.

10. Analyses Reported by the Sponsor

The registry reports a sponsor analysis for the primary endpoint: a Bayesian mixed-model comparison with a ratio of 1.336 and a 90% confidence interval from 0.965 to 1.847. The endpoint description states that the model adjusted for subject-level and period-adjusted baselines, treatment, and period, and that a non-informative prior was used. No primary p-value is reported.

A separate unadjusted two-arm re-calculation is not available for this endpoint. The posted summary identifies the endpoint as another type of outcome for which values were missing or not estimable for a two-arm comparison. That absence is important: a simple re-calculation from arm summaries would not reproduce the sponsor’s crossover model in any case. The arm-level medians do not contain enough information to reconstruct the model’s baseline and period adjustments.

There is no reported discrepancy between a re-calculation and the sponsor result. The sponsor’s model-based comparison is therefore the relevant posted analysis for the primary endpoint, rather than an independent re-calculation. Its estimate and interval should be interpreted together: the point estimate is above the no-change ratio, while the interval includes values on either side of no change. The registry record does not provide further model output, such as a posterior probability for a treatment effect or additional sensitivity analyses.

11. Multiplicity, Missing Data and Analysis Populations

The registry record identifies one primary endpoint: total daytime cough counts after 7 days of dosing. It also lists multiple secondary outcomes, including adverse events, serious adverse events, and laboratory measures. The record does not report a multiplicity strategy for interpreting the primary and secondary outcomes, such as an adjustment procedure or a hierarchy of hypotheses. Without that information, secondary findings should be treated as exploratory rather than as independent confirmatory evidence.

The record reports 19 outcome measures and 18 posted secondary outcomes. It does not explain why one registered outcome measure is not among the posted secondary outcomes, or whether it was not collected, not analyzed, or not reported. Those possibilities cannot be distinguished from the registry information. The record also does not report a prespecified statistical analysis plan in enough detail to establish which outcomes or comparisons were planned before data were examined.

For the primary endpoint, the registry states that only participants with data at the specified data points were analyzed. The analysis population included randomized participants who took at least 1 dose, and participants were analyzed according to treatment actually received. The reported arm summaries include 17 participants for placebo and 15 for GSK2798745. The registry does not provide reasons for missing endpoint data, details of participant withdrawals, or a missing-data method. It does not report imputation or sensitivity analyses for missing cough-monitor measurements.

General statistical context: missing outcome data can affect interpretation when the likelihood of having a measurement is related to treatment or the outcome itself. Complete-case summaries can be informative, but their reliability depends on how missingness arose and on the analysis assumptions. Without participant flow and missing-data details, it is not possible to assess how robust the primary estimate is to missing measurements.

The registry describes a Bayesian mixed model and a non-informative prior, but does not report interim analyses, stopping criteria, or detailed analysis-plan procedures. It records the study as terminated, but the reason and timing of the termination decision are not reported in the registry record. These gaps constrain assessment of whether outcome selection, multiplicity, or stopping decisions affected the posted results; they do not establish that any such influence occurred.

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 non-serious adverse event terms

The posted serious adverse event figures are 0 among 17 participants at risk for placebo and 0 among 16 participants at risk for GSK2798745. The record also reports 0 deaths in each treatment group. For other adverse events, it records 9 affected participants among 17 at risk for placebo and 11 among 16 at risk for GSK2798745. No top serious adverse-event terms are listed.

These are counts of participants, not estimates of comparative risk adjusted for the crossover design. A participant may have contributed safety information in both treatment periods, and the registry does not describe how repeat events or exposure time were handled in these summaries. The different at-risk counts also mean the figures should be read as the registry reports them, rather than compared as though the groups had identical denominators.

General statistical context: observing no serious adverse events in a small study does not rule out uncommon harms or establish that the treatments have the same safety profile. Likewise, the counts of other adverse events are descriptive and do not by themselves show whether a difference is due to treatment. The record does not provide event-level timing, severity, relatedness, or a detailed safety analysis. These figures describe the posted observations; they are not a complete characterization of safety.

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

The primary result is an estimate of relative daytime cough counts during a defined monitoring window after 7 days of treatment. The reported ratio is 1.336, with a 90% confidence interval from 0.965 to 1.847. The point estimate is above the no-change ratio, but the interval extends to values on both sides of no change. The estimate therefore does not precisely establish the direction or size of the relative comparison.

The finding does not demonstrate that GSK2798745 reduces cough counts, and it does not establish that the treatment increases them. It also does not show that GSK2798745 and placebo are equivalent: equivalence requires an appropriate margin and analysis, neither of which is reported in the registry record. A confidence interval that includes no change is not proof of no effect. It indicates that the available data do not isolate a clear treatment difference under the reported analysis.

The outcome concerns monitored daytime coughing for up to 10 hours after dosing on Day 7 of each treatment period. It is not a direct measure of long-term cough burden, quality of life, or effects outside the monitoring window. The study’s short dosing periods and recorded enrollment of 17 constrain what can be inferred about longer-term use or other populations. The primary efficacy summaries include fewer participants in the GSK2798745 arm than in the placebo arm, and the record does not explain all missing data or provide a detailed participant flow.

The record also reports safety counts and selected laboratory outcomes, but those results do not establish broad safety or resolve the primary efficacy uncertainty. The study is listed as terminated, with no reason reported. That status is relevant context but does not, on its own, identify why the study ended or determine how the treatment comparison should be interpreted.

Overall, the posted data describe a short-term, model-adjusted crossover comparison with substantial uncertainty. They support a cautious statistical reading of the observed measurements, not a clinical recommendation or a definitive conclusion about treatment benefit or harm.

14. Why This Trial Matters Statistically

This trial illustrates why the analysis structure matters as much as the arm summaries. Participants were randomized to treatment sequences and received both placebo and GSK2798745 in separate periods. A model that accounts for period and baseline terms addresses features of this design that a simple comparison of arm medians cannot reproduce. The posted primary result is therefore the Bayesian mixed-model ratio, not a stand-alone contrast of the two displayed medians.

It also illustrates how uncertainty can change the reading of a point estimate. The ratio of 1.336 is above the no-change ratio, yet its 90% confidence interval extends from 0.965 to 1.847. The direction suggested by the point estimate is not established by the interval. Reporting both the estimate and its uncertainty prevents a single value from being mistaken for a precise conclusion.

The endpoint itself is a count obtained with a cough monitor over a defined period. That offers an objective measure of cough frequency for the study window, while also making the result specific to the measurement method, timing, and treatment duration. The log transformation and ratio-based comparison are appropriate concepts to explain, but the registry’s brief description leaves important details of the model and missing-data handling unavailable.

Finally, the study record exposes limits that are common in small registry results: limited precision, incomplete reporting of analysis procedures, and a status of terminated without a reported reason. Statistical interpretation requires keeping those limits in view. A result can be informative about what was measured while remaining insufficient to settle broader questions about efficacy, safety, or outcomes over longer periods.

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 recorded enrollment is 17, and the primary endpoint summaries include 17 participants for placebo and 15 for GSK2798745. These figures indicate that the available endpoint data differ by treatment, but the registry does not provide a complete participant-flow account or explain the missing measurements. The uncertainty around the primary ratio is substantial, which limits the precision of the comparison.

The registry describes the study as a crossover but does not report counts by sequence. It does not provide enough detail to assess balance between treatment orders, the handling of possible carryover, or whether treatment effects varied by period. The stated model adjusts for period and baseline terms, but the full model specification and diagnostic checks are not reported. The exact masking procedures are also unclear despite the title’s description as double-blind and sponsor open.

The primary outcome is limited to total daytime cough counts recorded by a monitor for up to 10 hours after dosing on Day 7 of each treatment period. The endpoint description does not report how the monitor data were processed when recording was incomplete, nor does it describe an analysis of other symptom or patient-reported outcomes. The study’s stated treatment periods are short, so the posted endpoint does not characterize longer-term outcomes.

The record lists the study as terminated without reporting why. It does not report the original sample-size justification, interim analyses, stopping rules, or a detailed multiplicity plan. It also does not give a detailed missing-data strategy or sensitivity analyses. The registry lists 19 outcome measures and reports 18 secondary outcomes as posted, without explaining the difference. These omissions constrain evaluation of the analysis and reporting context, but they do not establish that any particular bias occurred.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study to Assess the Effectiveness and Side Effects of GSK2798745 in Participants With Chronic Cough — NCT03372603.
• Registry record last updated 2019-10-02; status last verified 2019-08.
• Results first posted 2019-10-02.
• 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

NCT03372603 is a randomized, double-blind, placebo-controlled crossover study of GSK2798745 in participants with chronic cough. It is listed as Phase 1 and Phase 2, with enrollment of 17 and status recorded as terminated. The primary endpoint was total daytime cough counts after 7 days of dosing, monitored for up to 10 hours after the Day 7 dose.

The registry reports a Bayesian mixed-model treatment ratio of 1.336, with a 90% confidence interval from 0.965 to 1.847. The interval spans values on both sides of the no-change ratio, leaving substantial uncertainty about the direction of the relative comparison. The record does not report the reason for termination, the sample-size justification, detailed missing-data handling, or a multiplicity strategy.