Study for the Treatment of Ocular Chronic Graft-Versus-Host Disease (GVHD) With Amniotic Fluid Eye Drops (AFED): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study for the Treatment of Ocular Chronic Graft-Versus-Host Disease (GVHD) With Amniotic Fluid Eye Drops (AFED) (NCT03298815), a randomized phase 1, phase 2 study. The primary outcome is Overall Response Rate.

2. Clinical Question

The study concerns ocular chronic graft-versus-host disease and compares AFED with saline solution in a randomized, double-blinded, placebo-controlled design. The primary endpoint is an overall response rate measured in eyes at 30 days. A responder is defined by a one point improvement in the Dry Eye Workshop (DEWS) grading scale without worsening in the National Institutes of Health (NIH) Consensus Conference eye score.

This endpoint combines a change in dry-eye grading with a condition that the eye score must not worsen. It therefore asks about a particular joint pattern of changes, rather than improvement on either measure alone. The posted primary results give counts of 3.0 for AFED and 6.0 for saline solution, with 12.0 eyes listed in each group. The record reports a McNemar test p-value of 0.375 for the comparison.

The registry record describes the study as Phase 1 and Phase 2, randomized, double-blinded, and parallel-group. It lists one eye per participant in each arm description. Further details about how eyes or participants were assigned to the analysis groups, and how the analysis handled the unit of assignment, are not reported in the registry record. That leaves important design and analysis details unresolved when interpreting a comparison expressed in eyes.

3. Trial Design

The registry record describes a randomized, double-blinded, placebo-controlled, parallel-group study. The listed phases are Phase 1 and Phase 2. The condition is ocular graft-versus-host disease, and the interventions are Amniotic Fluid Eye Drops (AFED) and saline solution. The study status is completed. Enrollment is listed as 15, with a start date of 2019-12-20 and primary completion date of 2022-12-16.

The primary endpoint is an eye-level count of responders at 30 days. The registry lists 12.0 eyes for each treatment group in the primary results. The arm descriptions say “All Participants, One Eye,” but do not provide further details about which eye was selected or how participants and eyes relate to the two groups. The safety results list 15 participants at risk in each group, rather than 12.0 eyes. Why these denominators differ is not reported in the registry record.

The trial record lists 13 outcome measures and 17 analyses. It reports 12 secondary outcomes. These include changes in Functional Assessment of Cancer Therapy: General (FACT-G) at 30, 60, and 100 days; changes in the NIH Consensus Criteria ocular score at those same time points; and changes in the DEWS 2007 dry-eye severity grading, with the listed timeframes including 30 days and 60 days. The registry entries do not give results for those secondary outcomes.

Further information about sequence generation, allocation concealment, the masking procedure, treatment schedule, eligibility criteria, participant flow, and the analysis population is not reported in the registry record. The registry record also does not report its own sample-size justification or a detailed statistical analysis plan. The design labels establish broad features of the trial, but do not resolve those operational and analytic details.

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 Overall Response Rate, measured over 30 days.

The primary endpoint is the overall response rate at 30 days, counted in eyes. Response is defined using two measures: the DEWS grading scale and the NIH Consensus Conference eye score. An eye qualifies as a responder if its DEWS grading scale improves by one point and its eye score does not worsen.

This composite definition is intended to distinguish improvement in dry-eye severity from a change that occurs alongside worsening in the other eye assessment. Combining criteria can make the clinical meaning of a response more specific, but it also means the endpoint does not show the separate contribution of each component. The posted responder count alone does not reveal how many eyes improved on DEWS, how many had an unchanged eye score, or whether any individual component changed without meeting the full definition.

The registry reports 3.0 responders among 12.0 eyes in the AFED group and 6.0 among 12.0 eyes in the saline group. It reports the endpoint at 30 days only. The registry record does not report a confidence interval for the treatment comparison or additional details about how the count was analyzed. The stated endpoint is a count-based outcome, so a mean difference or standard deviation is not the natural summary of the posted results.

As a general statistical point, a clearly defined composite endpoint can provide a single outcome for a study, but its interpretation depends on the components and the rule joining them. The response definition is available in the registry record; component-level results and further details on endpoint assessment are not reported there.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The posted primary outcome is a count of eyes meeting a binary response definition. Each eye either meets the composite criteria or does not. The endpoint is not a continuous measurement, and the registry reports counts rather than means or measures of spread. A comparison of such outcomes requires methods appropriate to the data structure and to whether observations are independent or paired.

What does the reported test assess?

The registry reports a McNemar test with p = 0.375. As a general statistical explanation, McNemar’s test is designed for paired binary outcomes. It uses the discordant pairs—pairs in which the two observations have different outcomes—to assess whether the pattern of discordance is balanced. The total counts in each group do not by themselves reveal the number or direction of discordant pairs, so they are not enough to reconstruct the test result.

The registry describes a randomized parallel-group study and says each arm involved one eye per participant. It does not explain whether the analysis treated observations as paired, how pairs were identified, or what data were entered into the test. Consequently, the choice and implementation of the reported test cannot be fully evaluated from the registry record. The label “parallel” and the use of McNemar’s test raise a question about the analysis structure, but the record does not provide enough detail to resolve it.

How should the p-value be read?

In general, a p-value is calculated under a null hypothesis and a specified statistical model. It describes how unusual data at least as incompatible with that null model would be if the model and its assumptions were correct. It is not a measure of the probability that the null hypothesis is true, and it is not a measure of the probability that a result occurred by chance in an everyday sense. The reported p-value of 0.375 does not demonstrate equivalence or establish that a difference is absent.

A p-value also does not describe the precision of the estimated effect. Confidence intervals are commonly used to show a range of effect estimates compatible with the data and model. No confidence interval for the primary comparison is reported in the registry record. The posted counts and p-value therefore do not provide the same information as an effect estimate accompanied by an interval.

What can be inferred from the counts?

The registry lists 3.0 responders among 12.0 eyes for AFED and 6.0 among 12.0 eyes for saline solution. These are the reported observed counts. No new comparative calculation is presented because the posted count outcome is not estimable as a two-arm comparison in the available analysis summary. In general, a calculation from marginal counts alone may not reproduce a paired-data analysis: the joint pattern of outcomes is needed to identify discordant pairs.

The registry lists 15 enrolled participants and 15 participants at risk in each safety group, while the primary results list 12.0 eyes in each group. A denominator is the number of observations represented in a particular analysis. Different analyses may use different populations, but the reasons for these particular denominators are not reported in the registry record. Without participant flow or analysis-population details, the relationship among enrollment, safety denominators, and primary endpoint denominators remains unclear.

What do the planning assumptions mean?

The stated sample-size assumptions describe a planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation, and a normal approximation. The illustrated effect is defined as a stated fraction of the observed effect. These are planning assumptions, not evidence about the trial’s original design process, and the registry record does not report its own sample-size justification. A planning illustration is not observed power: it cannot establish the probability that this completed trial detected a true effect, nor does it show that the original design was adequate.

In general, an illustration based on an observed effect should be interpreted cautiously. The observed effect itself is uncertain, particularly in a small study, and the illustration does not account for every feature of the actual design unless those features are explicitly included. No additional sample-size results or design calculations are stated in the available the registry entries.

Why do analysis details matter?

Statistical methods depend on the unit of analysis, the allocation structure, and the assumptions about dependence among observations. If two eyes from one participant are analyzed, their outcomes may be related; if one eye per participant is assigned to each group, the analysis structure may differ. The arm descriptions and trial design labels do not provide enough detail to determine which situation applies. The analysis population, missing-data approach, and handling of any protocol deviations are also not reported.

The registry lists 13 outcome measures, 12 posted secondary outcomes, and 17 analyses. In general, examining multiple outcomes or analyses can increase the chance that some results appear notable even when no single effect is clear. Whether any multiplicity adjustment, hierarchy, or prespecified testing sequence was used is not reported in the registry record. The number of listed outcomes alone does not establish how the analyses were planned or interpreted.

10. Analyses Reported by the Sponsor

The registry’s reported primary analysis is a McNemar test with p = 0.375. The posted primary counts are 3.0 responders among 12.0 eyes for AFED and 6.0 responders among 12.0 eyes for saline solution. No recalculated two-arm effect estimate is available for this count outcome in the posted analysis summary: the comparison is not estimable from the reported summary statistics.

There is no reported discrepancy between a sponsor analysis and a recalculation. The registry’s McNemar result is therefore the available inferential result, but its interpretation remains limited by the lack of a confidence interval and by missing details about pairing and the analysis unit. A McNemar analysis relies on paired binary data, while the record describes a randomized parallel-group study. The record does not explain the relationship between that design description and the test’s data structure.

The reported p-value should not be treated as a complete account of the result. It does not quantify the precision of a comparative effect, and it does not prove equivalence or absence of a difference. The sponsor’s reported test is the result to cite, while the counts, endpoint definition, and unresolved analysis details provide essential context for reading it.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 outcome measures, 12 posted secondary outcomes, and 17 analyses. The secondary outcomes include FACT-G changes at 30, 60, and 100 days; NIH Consensus Criteria ocular score changes at 30, 60, and 100 days; and DEWS 2007 dry-eye severity grading changes at the listed 30-day and 60-day timeframes. The available record does not include results for those secondary outcomes.

When a study examines multiple outcomes, time points, or analyses, the chance of obtaining at least one apparently notable result can increase. As a general statistical approach, investigators may address this through a prespecified hierarchy, adjustment, or other strategy. Whether the study used any such approach, and which outcomes were designated for formal testing, is not reported in the registry record. The outcome and analysis counts alone do not establish that a multiplicity problem occurred.

Missing-data methods and the primary analysis population are also not reported in the registry record. There is no account in the registry entries of how missing endpoint measurements, withdrawals, or protocol deviations were handled. The primary results list 12.0 eyes per group, while the record lists 15 enrolled participants and 15 participants at risk per safety group. The reasons for the different denominators are not reported, so it is not possible to determine from these figures whether the primary analysis included all eligible observations or how any excluded observations were treated.

The registry record does not report an intention-to-treat population, a per-protocol population, or a prespecified approach to analysis of missing outcomes. It also does not report an interim analysis plan or the statistical analysis plan. These omissions limit assessment of whether the posted analyses followed a prespecified strategy and whether the endpoint counts represent the randomized groups as intended.

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 safety results list 15 participants at risk in each group. Serious adverse events affected 0 participants in the AFED group and 0 in the saline solution group; deaths were also 0 in each group. Other adverse events affected 3 participants in the AFED group and 1 in the saline solution group. No serious adverse event terms are listed.

These figures describe the counts recorded in the registry. They do not provide detail about the nature, timing, duration, severity, or attribution of the other adverse events. The registry record also does not report the period over which these events were collected or further information about safety monitoring. The absence of listed serious adverse events in these groups is not proof that no such events could occur, especially when the observed group counts are limited.

Safety denominators differ from the primary endpoint denominators: the safety groups list 15 participants at risk each, while the primary results list 12.0 eyes in each group. The record does not explain the difference. These safety counts should not be combined with the eye-level primary endpoint counts or interpreted as if they use the same analysis population. They are descriptive registry figures, not a complete assessment of comparative safety.

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

The primary endpoint results describe the number of eyes meeting the stated response definition at 30 days: 3.0 in the AFED group and 6.0 in the saline solution group, with 12.0 eyes listed in each group. The registry reports a McNemar test p-value of 0.375. These registry entries establish what the posted primary results contain; they do not by themselves establish a treatment effect, its precision, or the reason for the difference in counts.

The response definition requires a one point improvement in DEWS grading without worsening in the NIH Consensus Conference eye score. The observed counts therefore concern eyes meeting both parts of that composite rule. They do not show the separate distributions of the two components. The registry record does not report a confidence interval or an effect estimate for the primary comparison, so the range of plausible comparative effects cannot be assessed from the posted primary summary.

A non-small p-value is not evidence that treatments are equivalent. Establishing equivalence or non-inferiority would require an appropriate design, a prespecified margin, and analysis details; those are not reported in the registry record. Likewise, the reported p-value does not tell readers the probability that either intervention is effective or ineffective. It is a result of a particular test under its assumptions, and the record does not fully explain how the test relates to the stated parallel-group design.

The enrollment count is 15, whereas the primary results list 12.0 eyes in each arm and the safety results list 15 participants at risk per group. The record does not explain how these counts correspond. Without that information, participant flow, missing observations, and the population represented by the endpoint cannot be reconstructed. The listed secondary outcomes also have no results in the registry entries, so they cannot be used to support or qualify the primary finding.

Accordingly, the result is best understood as a limited report of observed responder counts and a test result, accompanied by important uncertainty about precision, denominators, and analysis structure. It does not establish that one intervention works or does not work, nor does it settle whether the groups differ in the underlying response rate.

14. Why This Trial Matters Statistically

This study illustrates why the statistical unit and the clinical unit need to be clear. The primary result is counted in eyes, while enrollment and safety information are expressed in participants. The arm descriptions refer to one eye per participant, but the record does not explain how the eye-level observations relate to the parallel-group allocation or why the primary and safety denominators differ. Such details matter because statistical methods depend on whether observations are independent or paired.

The endpoint also shows how a composite response rule shapes interpretation. A responder must improve by one point on the DEWS grading scale without worsening on the NIH eye score. A count based on that joint rule cannot reveal how each component behaved on its own. Clear component-level reporting would help readers understand what contributes to the endpoint, but component results are not reported in the registry entries.

The p-value is presented without a confidence interval or comparative effect estimate. This makes the result harder to assess in terms of precision and plausible effect sizes. The registry lists many outcomes and analyses, while the multiplicity strategy and analysis plan are not reported. Together, these features demonstrate why inferential results need context: endpoint definitions, analysis populations, denominators, test assumptions, and prespecified analysis choices all affect interpretation.

The record’s sample-size assumptions are described as a planning illustration based on the observed effect. This is distinct from a reported rationale for the trial’s original design and from observed power. The registry does not report its own sample-size justification. The distinction is important because a retrospective planning illustration cannot replace a prospective design rationale or resolve the uncertainty in a result.

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 result lacks a reported confidence interval and a comparative effect estimate. The registry gives responder counts and a McNemar test p-value, but no interval with which to assess precision. The p-value alone cannot show the magnitude or range of plausible effects.

The analysis structure is not fully described. McNemar’s test is generally used for paired binary observations, while the registry identifies a randomized parallel-group design. The arm descriptions refer to one eye per participant, but do not explain whether observations were paired or how the reported test was implemented. It is not possible to resolve this issue from the registry record, and the ambiguity should not be taken as proof of an error.

Denominators are also unclear. The record lists 15 enrolled participants and 15 participants at risk for safety outcomes in each group, but the primary results list 12.0 eyes per group. It does not explain how these counts relate to each other, which participants or eyes were included in the primary analysis, or whether missing observations account for the difference.

The registry record does not report the original sample-size justification, a detailed statistical analysis plan, a multiplicity strategy, the analysis populations, or methods for handling missing data. It does not provide primary endpoint component results or the secondary outcome results in the registry entries. Further details about allocation, eye selection, and outcome assessment are also not reported. These limitations constrain independent evaluation of the reported test and the scope of conclusions that can be drawn from the posted results.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study for the Treatment of Ocular Chronic Graft-Versus-Host Disease (GVHD) With Amniotic Fluid Eye Drops (AFED) — NCT03298815.
• Registry record last updated 2025-04-09; status last verified 2025-04.
• Results first posted 2025-04-09.
• 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 record describes a completed, randomized, double-blinded, placebo-controlled, parallel-group Phase 1 and Phase 2 study of AFED compared with saline solution for ocular chronic graft-versus-host disease. Enrollment is listed as 15. The primary endpoint is the number of eyes meeting a composite response definition at 30 days: a one point DEWS improvement without worsening in the NIH Consensus Conference eye score.

The primary results list 3.0 responders among 12.0 eyes in the AFED group and 6.0 among 12.0 eyes in the saline group. The reported McNemar test p-value is 0.375. No confidence interval is reported. Safety results list 0 serious adverse events and 0 deaths in each group, with other adverse events affecting 3 participants in the AFED group and 1 in the saline group.

The record does not explain the difference between primary and safety denominators, the relationship between the parallel-group design and the reported test, or the analysis populations. It also does not report a detailed statistical plan, multiplicity strategy, or missing-data methods. These gaps limit interpretation of the primary comparison.