Visual Restoration for Hemianopia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Visual Restoration for Hemianopia (NCT03350919), a randomized na study. The primary outcome is 24-2 Humphrey PMD Change From Baseline to 6-month Post-training Timepoint.

2. Clinical Question

The research objective was to assess whether visual training could reduce the size of a visual field deficit and improve visual functions in adults with brain damage. The posted primary measure is PMD from standard Humphrey automated perimetry using a 24-2 testing pattern. It captures a change in a perimetric measure over the training interval; the registry record does not define a threshold for a clinically meaningful change.

The central comparison of interest is training in the blind field versus training in the intact field. The registry record presents mean changes for the two groups separately for the left and right eyes, but does not report a direct two-arm estimate or statistical test. The posted numbers therefore describe the outcome within each group and eye; they do not by themselves quantify the difference attributable to the training location.

3. Trial Design

The registry describes a completed, randomized, quadruple-masked, parallel-group trial sponsored by the University of Rochester. The listed enrollment is 48. The trial began on 2018-03-15, and the primary completion date is 2019-12-03. The phase is listed as NA.

Participants were assigned to software-based training in either the blind visual field or the intact visual field. Both are active interventions, so the comparison concerns where training was conducted rather than training versus no training. The registry does not report further details of the randomization procedure, the allocation ratio, or what the quadruple-masking label means in practice. Details of who was masked and how masking was maintained are not reported in the registry record.

The primary outcome was assessed after a six-month training interval. It is a continuous measure, reported as a mean change with a 95% confidence interval. The primary population description says that subjects who completed assigned training were included, and that each eye was analyzed independently. The posted outcome groups list 23 observations for each training-by-eye result. Those eye-specific counts should not be interpreted as an independent count of randomized participants: the record states that each eye was analyzed independently, but does not explain how the eye-level observations relate to participant-level enrollment.

The safety summary uses different group counts: 25 participants at risk in the blind-field group and 23 in the intact-field group. The registry record does not explain the difference between these safety counts, the listed enrollment, and the primary outcome counts. It also does not report a detailed participant flow or the number assigned to each arm at randomization.

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: No subjects were excluded from the study between recruitment and randomization to assigned groups.

Overall Study

Analysis population for the primary outcome, as defined in the registry: All subjects that completed the assigned training were included in analysis. Each eye was analyzed independently for the primary outcome measure.

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 24-2 Humphrey PMD Change From Baseline to 6-month Post-training Timepoint, measured over 6 months.

The primary endpoint is change in perimetric mean deviation (PMD) measured with standard Humphrey automated perimetry using a 24-2 testing pattern. It compares the baseline measurement with the post-training measurement after a six-month training interval. The measure is reported in decibels, with the group results expressed as mean changes and 95% confidence intervals.

Perimetry assesses visual field sensitivity across tested locations. A summary measure such as PMD condenses information from the field test into a single value, making it possible to describe average change over time. In a training study, a change from baseline is relevant because the research question concerns whether visual function differs after a period of training. The endpoint does not, on its own, describe the location or pattern of any change across the visual field.

The registry record does not state a clinically meaningful change threshold or explain how a particular size of PMD change should be interpreted for an individual. It also does not provide a direct between-group estimate for the endpoint. As a result, the posted means describe change within each group and eye, but do not establish the size or precision of a contrast between training in the blind field and training in the intact field.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous: PMD change is summarized by a mean. A mean gives the arithmetic average of the recorded changes in a group. It is useful for describing the central level of change, but it does not show the distribution of individual responses. The posted summaries do not include a standard deviation or another measure of participant-level spread.

What does a 95% confidence interval express?

In general, a confidence interval communicates uncertainty around an estimated quantity. A narrower interval indicates greater precision for that estimate, while a wider interval indicates less precision. A 95% confidence interval is not a guarantee that a particular interval contains the true value, and it does not mean that 95% of individual observations lie inside its limits. The intervals in the registry are attached to the eye-specific mean changes, not to a comparison between the two training groups.

For the left eye, the blind-field mean change is 0.84 decibels, with limits 0.22 and 1.47; the intact-field mean change is 0.1 decibels, with limits -0.52 and 0.72. For the right eye, the corresponding blind-field result is 0.58 decibels, with limits 0.07 and 1.08; the intact-field result is 0.12 decibels, with limits -0.38 and 0.62. The registry labels these limits as 95% confidence intervals. No variance estimates or detailed method for calculating the intervals are reported in the registry record.

Why is a direct group contrast needed?

The question raised by a randomized comparison is whether the outcome differs between the assigned groups. That requires estimating the between-group contrast, along with uncertainty for that contrast. Separate group means are not the contrast itself. Nor does the fact that one group’s interval includes zero and another group’s interval does not establish a statistically significant difference between them. The registry record says values for a two-arm comparison of this continuous endpoint are missing or not estimable, and reports no analysis results. A direct estimate and its uncertainty therefore cannot be interpreted from the posted results.

What does analyzing each eye independently imply?

The endpoint population description says each eye was analyzed independently. This indicates that the reported summaries are eye-specific rather than a single combined result for each participant. The record does not explain the statistical handling of the two eyes belonging to the same person, or whether dependence between eyes was incorporated in any model. No analyses are posted, so the analysis details needed to assess that issue are not available. The eye-level sample counts should not be treated as evidence that left and right eyes represent different participants.

What can be said about statistical significance?

The registry reports no p-values and no test for the difference between the training groups. The confidence intervals for the means can be read as precision summaries for those means, but they cannot supply the missing comparative test. Even a test of change from baseline within each arm would address a different question from the randomized between-arm comparison. The registry record does not report such within-arm tests either.

How should planning calculations be understood?

In general, sample-size planning illustrations ask how many participants per arm might be needed under specified assumptions to detect a chosen effect with a selected significance level and power. They are planning exercises, not evidence about the power achieved in a completed trial. Any such illustrations based on an observed effect should not be called observed power. The record describes a planning approach, but the registry entries do not include numerical planning results. The trial’s own sample-size justification is not reported in the registry record.

More generally, the reliability of a mean and its interval depends on the analysis population, measurement variability, and assumptions used for estimation. Those details matter particularly when observations may be related, as eye measurements from one participant can be. The available record gives the endpoint summaries and states that eyes were analyzed independently, but does not provide enough analysis detail to assess those assumptions fully.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure and no secondary outcomes. It also lists zero analyses. There is therefore no posted family of outcome tests from which to assess how multiplicity was handled. A prespecified multiplicity strategy, any hierarchy of outcomes, and any adjustment for multiple comparisons are not reported in the registry record. The absence of posted secondary outcomes does not establish what may have been measured during the trial; it describes what is recorded as an outcome in this record.

The primary population is described as subjects who completed their assigned training, with each eye analyzed independently. The record does not report whether this was the only analysis population, how many participants were randomized to each arm, or how many did not complete training. It does not state whether an intention-to-treat analysis was conducted or whether any alternative population was used.

Methods for missing data are not reported in the registry record. In particular, the record does not say how missing baseline or follow-up PMD measurements were handled, whether participants with incomplete measurements were excluded, or whether any imputation method was used. The posted outcome counts are 23 per training-by-eye result, while the safety counts are 25 and 23; the record does not explain these differences.

Without analysis-population details and missing-data methods, it is difficult to determine how the posted means relate to all randomized participants. No statistical analysis plan, interim analysis details, or multiplicity procedure is reported in the registry record. These are limits on interpretation, not evidence that a particular method was or was not used outside what the record reports.

11. 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 5% reporting threshold.

The serious adverse event summary reports zero affected participants among 25 at risk in the blind-field training group and zero among 23 at risk in the intact-field training group. The death count is zero in both groups. The summary also reports zero participants with other adverse events among the same respective at-risk counts.

These figures are counts within the groups and risk denominators recorded in the registry. They do not mean that the trial established the absence of any possible harm. The record does not provide event follow-up details, event definitions, or a fuller account of how adverse events were collected. No serious adverse event terms are listed.

The safety denominators differ from the 23 observations reported for each eye-specific efficacy result. The registry does not explain the difference, so the safety and efficacy counts should be read as their own posted summaries rather than treated as interchangeable populations. A zero count is a statement about events recorded in this summary, not proof that the underlying risk is zero.

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

The posted results show positive mean PMD changes in both eyes among participants assigned to blind-field training, with 95% confidence intervals above zero for both eye-specific means. In the intact-field training group, the mean changes are smaller in the posted summaries, and both confidence intervals include zero. These are descriptive patterns in the recorded within-group changes.

They do not establish that training in the blind field produced a greater change than training in the intact field. The randomized question concerns a difference between groups, but the registry does not report a direct estimate or test for that contrast. Differences between separate means cannot be interpreted as a trial-level treatment effect without the comparative analysis and its uncertainty. Nor can the separate confidence intervals be used as a substitute for it.

The results also do not show how much change an individual participant experienced. Means can conceal variation in the direction and size of individual outcomes. No individual-level distribution is included in the registry summaries, and the record does not state a threshold for a clinically meaningful PMD change. The statistical summaries therefore cannot be translated into a claim about the importance of the change for an individual.

The recorded outcome population consists of subjects who completed assigned training, and each eye was analyzed independently. Because the record does not report how missing outcomes or participants who did not complete training were handled, the summaries should not be assumed to represent all randomized participants. The posted results also do not provide secondary outcome analyses.

In short, the record supplies eye-specific mean changes and confidence intervals, but not the comparative estimate needed to evaluate the principal between-group question. The safety summary records no serious adverse events, other adverse events, or deaths in the stated groups, while leaving event-collection details unreported. Neither the efficacy summaries nor the safety counts alone support broader conclusions beyond what was recorded.

13. Why This Trial Matters Statistically

This trial illustrates an important distinction in randomized studies: describing change within each group is not the same as estimating the effect of assignment to one group rather than another. The registry reports eye-specific means and confidence intervals, but the direct two-arm comparison is missing or not estimable. Without that contrast, the central comparative question cannot be answered from the posted outcome statistics.

The record also makes the analysis unit important. It states that each eye was analyzed independently, while participants can contribute measurements from both eyes. The posted counts are eye-specific, and the record does not explain how any relationship between the two eyes was handled. That information is relevant to understanding the precision of the estimates.

Population definitions matter as well. The primary outcome description includes subjects who completed assigned training, whereas the registry lists an enrollment of 48 and gives different denominators in the safety and efficacy summaries. The record does not describe participant flow or explain these counts. Readers therefore need to distinguish the trial’s listed enrollment from the observations contributing to each posted outcome summary.

Finally, the absence of posted secondary outcomes, analysis results, and missing-data details limits what can be concluded about the full statistical plan. These reporting gaps are informative: they show why estimates, uncertainty, analysis populations, and methods for handling incomplete data are all needed to interpret a randomized clinical trial. The posted descriptive changes are useful as summaries, but they do not replace the prespecified comparative analysis needed to answer the trial’s principal statistical question.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main limitation for interpreting efficacy is that the registry does not report a direct comparison between training in the blind field and training in the intact field. It provides separate eye-specific mean changes and confidence intervals, but no between-group estimate, interval, or test. As a result, the comparative effect cannot be quantified from the posted results.

The primary analysis population is described as subjects who completed assigned training. The record does not report participant flow, arm counts at randomization, or the number who did not complete training. It also does not report missing-data methods or whether an intention-to-treat analysis was performed. The relationship between the listed enrollment of 48, the eye-level outcome counts of 23 per result, and the safety counts of 25 and 23 is not explained.

The endpoint is analyzed separately for each eye, but details of how measurements from two eyes of the same participant were handled are not reported. The record also does not provide standard deviations, individual-level outcome distributions, or the method used to calculate the confidence intervals. No clinical-importance threshold for PMD change is given.

Other trial details relevant to statistical interpretation are not reported in the registry record, including the randomization procedure, masking implementation, analysis plan, interim analyses, and the trial’s sample-size justification. The registry lists no secondary outcomes and no analyses. These limits mean the available summaries are best read as descriptive outcome information rather than a complete account of the trial’s comparative statistical evidence.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Visual Restoration for Hemianopia — NCT03350919.
• Registry record last updated 2020-12-24; status last verified 2020-12.
• Results first posted 2020-12-24.
• 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.

18. Record Summary

NCT03350919 is a completed randomized, quadruple-masked, parallel-group trial sponsored by the University of Rochester. It compared software training in the blind visual field with software training in the intact visual field among adults with visual deficits related to brain damage. Listed enrollment is 48.

The primary outcome was change in 24-2 Humphrey PMD from baseline to the post-training assessment after a six-month training interval. The record reports mean changes and 95% confidence intervals for left and right eyes in each group, with 23 observations per eye-specific result. It states that each eye was analyzed independently and that subjects completing assigned training were included.

No direct two-arm comparison or analysis result is posted. The safety summary records no serious adverse events, other adverse events, or deaths in either group. Participant flow, missing-data methods, the trial’s sample-size justification, and further analysis details are not reported in the registry record.