The Support, Educate, Empower (SEE) Program: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Support, Educate, Empower (SEE) Program (NCT04735653), a randomized na study. The primary outcome is Mean Percent Electronically Monitored Medication Adherence Over Six Months.

2. Clinical Question

The trial asks whether personalized glaucoma coaching is associated with higher eye-drop medication adherence than enhanced standard care among people with glaucoma. The primary outcome is mean percent adherence measured electronically over up to six months. The registry record describes the coaching as personalized education and motivational-interviewing-based counseling delivered by a trained non-physician coach. The comparator received glaucoma educational materials by mail. The record also identifies glaucoma-related distress as a secondary outcome, but it does not report a result for that outcome.

This is a comparison of trial groups, not a direct account of why individual participants did or did not take medication on time. The registry record identifies randomized allocation and a parallel-group model. It does not give enough detail to establish how participants were recruited, which eligibility criteria determined inclusion, or how broadly the results apply beyond the participants represented in the outcome data.

3. Trial Design

The registry record identifies NCT04735653 as a randomized, single-masked, parallel-group trial of the Support, Educate, Empower (SEE) personalized glaucoma coaching program. Its status is completed. The listed enrollment is 236, and the listed study dates run from 2021-04-27 to primary completion on 2024-07-31. The sponsor is the University of Michigan. The phase is listed as not applicable.

The intervention was personalized glaucoma coaching. The arm description characterizes it as a six-month program in which a coach trained in motivational-interviewing-based counseling uses a web-based tool to produce tailored education. The aim described in the record is to help participants identify barriers to medication adherence and explore possible solutions. The control arm is called enhanced standard care. Its description specifies non-tailored educational materials about glaucoma sent by mail every two months during the six-month study period. The comparator description also says that medication adherence was monitored electronically for all participants, though the remainder of that description is truncated in the registry record.

The primary outcome is mean percent electronically monitored medication adherence over up to six months. The outcome population is described as participants who provided data for that outcome. The posted arm summaries include 113 participants in the coaching group and 114 in the control group. These are the outcome-specific data counts; they should not be treated as interchangeable with the listed enrollment or the safety denominators.

The record identifies glaucoma-related distress as a secondary outcome, but a result for it is not reported in the registry record. The original sample-size justification, details about who was masked, full eligibility criteria, recruitment procedures, and the original statistical analysis plan are not reported in the registry record. Those omissions limit assessment of design decisions beyond the information explicitly recorded.

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: 236 participants were consented. 235 were randomized. The unrandomized individual was determined to be ineligible post-consenting.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Data is shown for participants who provided data for this outcome.

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 Mean Percent Electronically Monitored Medication Adherence Over Six Months, measured over Up to 6 months.

The primary endpoint is mean percent electronically monitored medication adherence over up to six months. The registry defines adherence as the percentage of prescribed doses taken on time, based on an electronic bottle-in-bottle monitoring method. An adherent event is use of an eye-drop medication within a specified time window relative to a dose on the previous day. The endpoint therefore reflects recorded timing of medication use, rather than self-reported adherence.

The registry describes several rules for calculating the measure. Participants were censored at withdrawal, loss to follow-up, or disengagement, with adherence calculated only over their reduced time in the study. Periods when monitors malfunctioned or were unused because of vacation or hospitalization were excluded. These rules mean the outcome is based on eligible monitoring periods, not necessarily the full nominal follow-up period for every participant.

A continuous mean is useful for comparing the average adherence level between groups while retaining variation in the measured values. The posted summary gives a mean and standard deviation for each group. The group means are 77.6% for personalized coaching and 58.0% for enhanced care control. These summaries describe the observed outcome data; by themselves they do not explain the causes of adherence patterns or establish how the result might translate to different patients or settings.

The registry record lists glaucoma-related distress as a secondary outcome. Its measurement details and result are not reported in the registry record, so interpretation of that outcome is not possible from the posted results.

7. Results Posted in the Registry

• SEE Intervention - Personalized Glaucoma Coaching: mean 77.60 versus 58.00 on control; difference 19.60 (95% CI 13.72 to 25.48), standardized effect d = 0.87; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies randomized allocation and a parallel-group design. In a randomized trial, allocation is intended to make the groups comparable on average at the start, although randomization does not guarantee balance in every characteristic. The record labels masking as single, but who was masked is not reported in the registry record. The intervention and control are behavioral programs with different educational approaches, while electronic monitoring is used to measure the primary outcome.

What does the primary summary describe?

The primary endpoint is continuous: it is a percentage adherence measure summarized by a mean and standard deviation in each arm. The posted mean is 77.6% in the personalized coaching group and 58.0% in the enhanced-care control group. The standard deviations are 19.7 and 25.2, respectively. A mean represents the average observed value in the group. A standard deviation describes how spread out individual values are around that group mean. It is not a confidence interval and does not describe the uncertainty in the mean in the same way.

The unadjusted recalculation compares the group means by subtracting the control mean from the coaching mean, producing a mean difference of 19.6 percentage points. This arithmetic is a descriptive contrast based on the posted summary statistics; it is not the sponsor’s reported analysis. The recalculated 95% confidence interval runs from 13.72 to 25.48. It is an interval estimate of the mean difference under the assumptions used for the recalculation, rather than a range containing a fixed share of individual participants’ adherence values.

What does the Wilcoxon result test?

The registry’s reported method is the Wilcoxon (Mann-Whitney) test, with a superiority analysis and p < 0.001. This rank-based method compares the ordering of observations between groups. It can detect a difference in distributions or the tendency for values in one group to rank higher than values in another. It is not, without additional assumptions, a direct test of the difference in arithmetic means. For that reason, the test result and the recalculated mean difference answer related but not identical questions.

A p-value describes how incompatible the observed data are with a specified null model, assuming the analysis assumptions hold. It is not the probability that the intervention has no effect, nor does it convey the practical size of a difference. The confidence interval for the mean difference helps describe the estimate’s scale and uncertainty, while the Wilcoxon p-value records the result of the sponsor-listed rank-based test. The registry record does not give a test statistic, exact analysis details, or an estimate and interval from that rank-based analysis.

Why can over-dispersion matter?

General statistical context: over-dispersion occurs when variability in the data is greater than a model assumes. A model that understates variability can produce confidence intervals that are too narrow and tests that appear more decisive than the data warrant. The posted endpoint is summarized with standard deviations, and the group standard deviations are not identical. The registry record does not report a model-based analysis that addresses over-dispersion, and the primary sponsor-listed method is rank-based. This general issue should not be taken as evidence that over-dispersion occurred or that the reported test is invalid.

How should the recalculation be distinguished from the reported analysis?

The recalculated mean difference and its confidence interval are simple, unadjusted calculations from the posted group summaries. They do not include covariate adjustment or any other model-based correction. The registry’s own analysis is the Wilcoxon (Mann-Whitney) test, reported as p < 0.001. The record does not report the sponsor’s mean-difference estimate or a confidence interval for that estimate. The recalculation is therefore useful as a transparent description of the posted group means, but it is not a replacement for the sponsor’s own inferential analysis.

The record does not state the original analysis plan, whether assumptions were checked, whether the test was prespecified, or whether the analysis adjusted for any baseline variables. It also does not report an intention-to-treat designation or another named analysis population. These details matter when assessing whether the posted analysis matches the planned estimand and how participants with incomplete monitoring contributed to the analysis.

What do confidence intervals and p-values add?

General statistical context: a confidence interval communicates a range of estimates compatible with the data and the calculation’s assumptions. Its width reflects precision: a narrower interval is more precise than a wider one on the same scale. A p-value summarizes evidence against a null hypothesis under a specified test, but does not show the magnitude of the effect. Neither statistic resolves design limitations, missing data, measurement error, or the relevance of the result outside the study. Those questions require information about trial conduct and population characteristics that is not reported in the registry record.

10. Confidence Intervals and Precision

The unadjusted recalculated mean difference in electronically monitored adherence is 19.6 percentage points, with a 95% confidence interval from 13.72 to 25.48. The interval is entirely above zero, consistent with higher mean adherence in the coaching group in these posted data. Its endpoints express uncertainty about the mean contrast under the recalculation’s assumptions, not a range of individual participant outcomes.

Precision is not the same as certainty about every aspect of the result. A confidence interval addresses sampling variability as represented by the calculation. It does not incorporate unknown biases, measurement limitations, or the possibility that participants with missing or excluded monitoring data differ from those with usable data. The registry states that some participant observation time could be reduced and that some monitor periods were excluded, but it does not report the amount of missing or excluded data or a sensitivity analysis addressing it.

The confidence interval is from an unadjusted recalculation of posted summary statistics. The sponsor-listed analysis is a Wilcoxon (Mann-Whitney) superiority test with p < 0.001; a confidence interval for that analysis is not reported in the registry record. Because the two methods summarize and test the data differently, the recalculated interval should not be presented as the interval from the sponsor’s reported test.

The standard deviations of 19.7 in the coaching arm and 25.2 in the control arm describe variability among the posted outcome values. They help contextualize the difference between group means but are not measures of precision for the treatment contrast. The registry record does not report adjusted estimates, subgroup-specific intervals, or longer-term adherence results. The available interval therefore characterizes the posted unadjusted primary-outcome contrast, not every clinically or statistically relevant uncertainty.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed mean difference, not the original trial’s design calculation. For this continuous outcome, each figure is the number of participants per arm, rather than a total across arms or a count of events. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row uses a stated fraction of the observed effect as the effect to detect. The original trial’s sample-size assumptions and justification are not reported in the registry record.

For planning based on 100% of the observed effect, the illustration gives 21 participants per arm for 80% power and 29 participants per arm for 90% power. For 75% of the observed effect, it gives 38 participants per arm for 80% power and 50 participants per arm for 90% power. For 50% of the observed effect, it gives 84 participants per arm for 80% power and 113 participants per arm for 90% power.

The 80% power column describes a planned design’s chance of detecting the assumed effect at the stated significance level if the planning assumptions and effect are correct. The 90% column uses a higher target probability of detection for that assumed effect, and its corresponding participant requirement is larger in these illustrations. The rows also show the general planning principle that smaller effects need more participants to detect reliably: reducing the assumed effect fraction is accompanied by larger per-arm requirements.

These values are not observed power. They are hypothetical planning requirements calculated from the observed effect under stated assumptions. The actual trial’s design rationale, allowance for missing data, and intended effect size are not reported in the registry record, so the illustration cannot establish whether the trial was planned around these assumptions.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 19.600; pooled standard deviation σ = 22.630.
2. n = 2 × (1.96 + 0.842)² × 22.630² / 19.600² = 20.9, rounded up to 21 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 19.6                      # observed difference in means
sigma <- 22.6299310157538                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(4735653)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 21 (80% power), 29 (90% power)
 75% of observed effect: n per arm = 38 (80% power), 50 (90% power)
 50% of observed effect: n per arm = 84 (80% power), 113 (90% power)
power.t.test: 21.9 
Simulated power at n = 21 per arm: 0.784 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The registry’s own reported analysis is a Wilcoxon (Mann-Whitney) superiority test comparing the personalized coaching and enhanced-care control groups. It reports p < 0.001, but gives no test statistic, mean-difference estimate, or confidence interval for that analysis. The posted group summaries are means and standard deviations, which permit a separate unadjusted recalculation of the mean contrast.

That recalculation gives a mean difference of 19.6 percentage points, with a 95% confidence interval from 13.72 to 25.48. It is based only on the posted summary statistics and is not an adjusted sponsor analysis. The methods answer related but distinct questions: the recalculated estimate expresses the arithmetic difference between group means, whereas the reported rank-based test compares the ordering of observations between groups.

No discrepancy is flagged between the posted results and the recalculation. The recalculation therefore offers an additional descriptive estimate, but it should not be mistaken for an estimate reported by the sponsor. The sponsor’s recorded test remains the trial’s own analysis in the registry record. Details such as its full implementation, analysis population, missing-data handling, and any prespecified analysis plan are not reported in the registry record.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists three outcome measures and three analyses, and it identifies glaucoma-related distress as a secondary outcome. A result for that secondary outcome is not reported in the registry record. The record does not identify the complete set of hypotheses tested, distinguish confirmatory from exploratory analyses, or describe any adjustment for multiple comparisons. It is therefore not possible to determine from the registry entry whether multiplicity was considered in the trial’s analysis or how it might affect interpretation across outcomes.

For the primary outcome, the record says the posted data are for participants who provided data for that outcome. The group summaries list 113 participants in the coaching arm and 114 in the control arm. The overall listed enrollment is 236, but participant flow from enrollment to outcome data is not reported in detail. The record does not say how many participants withdrew, were lost to follow-up, disengaged, or had unusable monitoring, or whether those circumstances differed between groups.

The endpoint description specifies that adherence was calculated only through withdrawal, loss to follow-up, or disengagement, and that monitor malfunction or periods of non-use due to vacation or hospitalization were excluded. These rules explain how some observation periods were treated, but they do not provide the amount of excluded time or the assumptions used for incomplete data. The sponsor’s missing-data strategy, any imputation method, and sensitivity analyses are not reported in the registry record.

The analysis population is also not named. The registry does not state whether the primary analysis followed an intention-to-treat principle, included only participants with complete endpoint data, or used another defined population. That omission matters because the posted outcome-specific data counts may not describe everyone assigned to each group. It also limits assessment of whether the result estimates the effect of assignment to the programs or the contrast among participants contributing measured outcomes.

The registry says masking was single, but does not identify who was masked. The original statistical analysis plan, prespecification of the primary comparison, baseline adjustment, and procedures for handling protocol deviations are not reported in the registry record. These missing details do not negate the posted result; they constrain how fully its analytic robustness and alignment with the planned trial objectives can be evaluated.

15. 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 safety figures report zero serious adverse events among 117 participants at risk in the coaching group and zero among 118 at risk in the enhanced-care control group. Deaths are reported as zero in the coaching group and two in the control group. Other adverse events are listed for 18 participants at risk in the coaching group and 7 in the control group. The registry lists no top serious-event terms.

These entries are counts alongside the registry’s stated at-risk denominators. They should be read as the safety information recorded for the trial, not as a complete account of the nature, timing, severity, or cause of events. The category labeled other adverse events is not further described in the information recorded here, and the record does not establish whether events were considered related to either program. The two deaths in the control group are a reported count; cause and timing are not reported in the registry record.

The safety figures are distinct from the primary adherence outcome and its outcome-specific participant counts. No rate comparison or causal safety conclusion can be drawn from these counts alone. The registry record does not provide detailed event narratives, event-specific exposure time, or an analysis comparing safety between groups. The figures therefore summarize what was posted without establishing that either program caused or prevented an event.

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

The primary-outcome results show higher mean electronically monitored medication adherence in the personalized coaching group than in the enhanced-care control group during the stated follow-up period. The posted means are 77.6% and 58.0%, and the unadjusted recalculated mean difference is 19.6 percentage points, with a 95% confidence interval from 13.72 to 25.48. The registry’s reported Wilcoxon (Mann-Whitney) test gives p < 0.001. Together, these results describe a clear group contrast in the observed trial data.

This result concerns the trial’s electronically monitored measure. The registry defines adherence using dose timing within a specified window and describes exclusions and censoring for some monitoring periods. It is not a direct measure of every aspect of medication use, nor does it indicate why participants took or missed doses. The result also does not supply a value for the secondary outcome of glaucoma-related distress, which is not reported in the registry record.

The contrast does not, by itself, establish how the coaching program would perform in settings or populations not represented by the outcome data. Information about eligibility, participant characteristics, recruitment, and the distribution of missing data is limited or not reported in the registry record. The posted analysis does not provide an adjusted estimate, and the original analysis plan and missing-data strategy are not reported. These limitations make it important to distinguish the observed comparison from broader claims about causation or generalizability.

The result should not be interpreted as proof that every participant’s adherence improved, or that the same difference would occur for every participant. Group means can conceal variation among individuals, reflected in the reported standard deviations of 19.7 and 25.2. The confidence interval describes uncertainty in the recalculated mean contrast under its assumptions, not the range of individual responses. The p-value also does not measure the probability that the result is due to chance or the probability that the intervention is effective.

In short, the posted data support a statistically strong difference in electronically monitored adherence between the trial groups, with the group mean higher for coaching. They do not settle questions about longer-term outcomes, the secondary distress outcome, the reasons for missing monitoring data, or how well the findings transfer beyond this trial. Those questions require evidence not reported in the registry record.

17. Why This Trial Matters Statistically

This trial illustrates why a clinical result is best read through several complementary statistical summaries. The group means communicate the average adherence level in each arm, while the standard deviations convey participant-level variation. The unadjusted mean difference gives the contrast on the endpoint’s original percentage scale. Its confidence interval describes the precision of that estimated contrast under the recalculation’s assumptions. The sponsor-listed Wilcoxon (Mann-Whitney) test adds a rank-based comparison that is related to, but not identical with, a comparison of means.

The example also highlights the distinction between an endpoint’s measurement rules and a simple complete-period measure. The registry describes electronic monitoring, dose-timing criteria, censoring at withdrawal, loss to follow-up, or disengagement, and exclusion of specified periods when monitors malfunctioned or were unused. Such rules affect which monitoring observations contribute to the adherence calculation. Without counts or sensitivity analyses for excluded and missing data, the size of their influence cannot be assessed from the registry record.

Planning values need similarly careful interpretation. The sample-size illustration shows how required per-arm numbers change when the assumed effect is smaller or the target power is higher. It is based on the observed effect and stated assumptions, not on the trial’s original planning rationale. A post-result planning illustration should not be confused with evidence that the trial had a particular probability of detecting its observed result.

Finally, the record gives a primary outcome contrast but does not report the listed secondary distress outcome, full multiplicity handling, or a detailed analysis plan. Statistical interpretation therefore depends not only on the strength of the primary result, but also on knowing which outcomes were prespecified, how incomplete data were treated, and whether the analysis matched the design. These are central questions for evaluating the evidential scope of a clinical trial.

18. Limitations

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

The registry record leaves important details of the trial population and conduct unspecified. Although it describes participants with glaucoma and refers in its summary to poor adherence, full eligibility criteria, baseline characteristics, recruitment procedures, and setting details are not reported in the registry record. This limits assessment of how representative the outcome-data participants were and how readily the finding might apply elsewhere.

The primary endpoint depends on electronic monitoring and specified adherence windows. The record says periods involving malfunction, vacation, or hospitalization were excluded, and that data collection was curtailed after withdrawal, loss to follow-up, or disengagement. It does not report the extent of these exclusions, the amount of usable monitoring in each group, or how the missingness patterns were assessed. Consequently, the effect of incomplete monitoring on the group comparison cannot be evaluated.

The registry lists a Wilcoxon (Mann-Whitney) superiority test but does not report its statistic or a confidence interval for that analysis. The mean-difference estimate and interval are unadjusted recalculations from the posted summaries, not sponsor-reported estimates. The original analysis plan, analysis population, any covariate adjustment, and sensitivity analyses are not reported in the registry record. The record also does not identify who was masked despite the single-masked designation.

Glaucoma-related distress is listed as a secondary outcome, but its measurement details and results are not reported in the registry record. The record lists three outcome measures and three analyses without explaining the full multiplicity strategy. Safety information is also limited to event counts and at-risk denominators, with no event narratives or causal attribution reported. These gaps constrain interpretation of outcomes beyond the primary adherence comparison and prevent a comprehensive assessment of the trial’s reporting and analytic choices.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: The Support, Educate, Empower (SEE) Program — NCT04735653.
• Registry record last updated 2025-11-03; status last verified 2025-10.
• Results first posted 2025-11-03.
• 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.

22. Record Summary

NCT04735653 is a completed randomized, single-masked, parallel-group trial sponsored by the University of Michigan. The listed enrollment is 236, with study start on 2021-04-27 and primary completion on 2024-07-31. It compares the six-month personalized SEE glaucoma coaching program with enhanced standard care.

The primary endpoint is mean electronically monitored medication adherence over up to six months. Posted means are 77.6% for coaching and 58.0% for enhanced care control. The unadjusted recalculated mean difference is 19.6 percentage points, with a 95% confidence interval from 13.72 to 25.48. The registry-listed Wilcoxon (Mann-Whitney) superiority analysis reports p < 0.001.

Glaucoma-related distress is listed as a secondary outcome, but its result is not reported in the registry record. Detailed information about missing-data handling, the original analysis plan, and the original sample-size rationale is also not reported in the registry record.