Morning Light Treatment to Improve Glucose Metabolism: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Morning Light Treatment to Improve Glucose Metabolism (NCT03188263), a randomized na study. The primary outcome is Change in Glucose Levels (Area-under-the-curve) Measured by Performance on Oral Glucose Tolerance Test (OGTT).

2. Clinical Question

3. Trial Design

The registry describes NCT03188263, titled Morning Light Treatment to Improve Glucose Metabolism, as a randomized, triple-masked, parallel-group pilot study. Northwestern University is listed as the sponsor. The study status is completed, with enrollment recorded as 10. The registry lists the start date as 2017-09-05 and the primary completion date as 2019-12-31. The phase entry is NA.

The intended study population was people with prediabetes, with circadian dysregulation also listed among the conditions. The study aimed to test a head-worn light device used at home. The two registered interventions were Bright Light and Dim Light. Bright Light was described as 230 μW/m2 and 500 lux; Dim Light was described as reduced to 3 μW/m2 and 7 lux. These are the registered intervention descriptions, not evidence that both conditions contributed data to the primary outcome.

Although the design is registered as randomized with parallel groups, the primary outcome record states that no participants were enrolled into the dim light condition. The posted primary result is consequently confined to the Bright Light arm. It reports six participants, a mean change of -11.3 mg*h/dL, and a standard deviation of 11.0 mg*h/dL. The registry's overall enrollment entry and the outcome-specific group record describe different quantities; the record does not explain that difference.

The study's stated purpose was pilot testing, with its data intended to support the development of a larger clinical trial. The registry lists change in circadian phase measured by dim light melatonin onset as a secondary outcome. It also records two outcome measures and two analyses, but a complete statistical analysis plan, including prespecified analysis populations and multiplicity procedures, is not reported in the registry record.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: There were no participants enrolled into the dim light condition.

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.

Population: There were no participants enrolled into the dim light condition.

6. Primary Endpoint

The registry lists the primary outcome as Change in Glucose Levels (Area-under-the-curve) Measured by Performance on Oral Glucose Tolerance Test (OGTT), measured over Change from baseline to after 4 weeks of light treatment.

The primary endpoint is change in glucose levels summarized by the area under the curve during an oral glucose tolerance test. The registry defines the change as the value after four weeks of light treatment compared with baseline. The measurement is expressed in mg*h/dL, and the posted summary is a mean with a standard deviation.

An oral glucose tolerance test measures blood glucose over time after consumption of a 75-gram dextrose solution. In this study's endpoint description, samples are taken at -15 min, 0 min, 30 min, 60 min, 90 min, 120 min, and 180 min. The total area under the curve summarizes the glucose measurements across the test rather than relying on a single time point. This can represent the overall glucose response during the procedure, while retaining less detail about the shape and timing of that response than the full set of measurements.

Change from baseline is used to describe how the endpoint differs after treatment relative to each participant's starting measurement. For a continuous outcome, the mean gives a measure of the average change in the observed group, while the standard deviation describes variation among participant-level changes. The posted primary result is for Bright Light only. Because the Dim Light condition has no participants enrolled in the primary outcome record, the endpoint data do not yield a comparison of average change between the registered conditions.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the mean change describe?

The primary endpoint is continuous: it measures change in glucose area under the curve in mg*h/dL. The posted Bright Light summary gives a mean change of -11.3 mg*h/dL among six participants. The mean is the arithmetic average of the participant-level changes included in that summary. It offers a concise description of the group's central tendency but does not imply that every participant had a change equal to the mean or even a change in the same direction.

The standard deviation of 11.0 mg*h/dL describes how spread out the observed changes are around the mean. It is not the standard error of the mean, and it is not a confidence interval. A standard deviation describes variation among observations; a confidence interval describes uncertainty in an estimated quantity under a specified statistical procedure. The registry reports the standard deviation but no confidence interval for this outcome.

What does the Sign test assess?

The registry names the analysis method as a Sign test and associates it with the Bright Light group. In general, a Sign test uses the directions of paired or one-sample differences, rather than the sizes of those differences, to assess whether positive and negative changes are consistent with a specified null position. Because it relies on signs, it does not use the full magnitude information in the mean change and standard deviation. Its interpretation depends on the exact data structure and null hypothesis used.

The record does not provide the test statistic, details of the null hypothesis, a confidence interval, or an explicit description of the analysis population. The p-value is reported as .046. The result should therefore be described as the recorded Sign test result for the Bright Light group, without inferring unreported implementation details. A within-group Sign test does not answer whether Bright Light differs from Dim Light.

Why is there no two-arm comparison?

A comparative trial estimate would normally use outcome information from both groups to quantify the difference between their outcomes, along with uncertainty for that difference. The primary outcome record states that no participants were enrolled into Dim Light. The posted group data likewise have a Bright Light result and no Dim Light mean or spread. A two-arm continuous-outcome comparison is consequently not estimable from these posted summaries.

General statistical context: a comparator group helps distinguish changes associated with the intervention from changes that could also occur over time or under another study condition. When comparator outcome information is absent, a within-arm change cannot substitute for a randomized between-arm contrast. The registry identifies randomization as part of the design, but the outcome data available for this endpoint do not support estimating the planned contrast.

How should uncertainty and p-values be read?

A p-value is calculated under a statistical model or test procedure and its null hypothesis. It measures how unusual the observed data, or data more extreme, would be if that null model held. It does not state the probability that the treatment works, the probability that the result will replicate, or the size of an effect. Its meaning also depends on the design and analysis that produced it.

A confidence interval would show a range of estimates compatible with the data and the method used. It can make the precision of an estimate more apparent than a p-value alone. No confidence interval is posted for the primary outcome analysis. The reported p-value is therefore not a measure of precision and should not be read as a substitute for an interval estimate, especially given that there is no comparator estimate.

What does over-dispersion mean in general?

General statistical context: over-dispersion refers to variability greater than a model expects. It is especially relevant for outcomes modeled as counts or event rates, where an overly restrictive variance assumption can make standard errors too small and statistical evidence appear more precise than it is. The primary endpoint in this study is continuous and reported as a mean with a standard deviation, so an over-dispersion model is not identified in the registry record as the method for this analysis. The broader point is that any statistical model needs assumptions appropriate to the outcome and its variability.

How do planning illustrations differ from observed results?

Sample-size planning is a forward-looking exercise: it describes how many participants per arm might be required under specified assumptions to detect a chosen effect with a stated significance level and power. The planning assumptions recorded for this study use a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation, with the target effect expressed as a fraction of the observed effect. Those assumptions are not the study's own design justification, which is not reported in the registry record.

Planning illustrations are not observed power and do not change what the trial measured. They are conditional on the assumptions used, including the effect size chosen for planning. Because the observed endpoint data come from Bright Light without a Dim Light outcome result, any planning calculation based on the observed effect should not be mistaken for evidence of a comparative effect. Sample-size requirements can inform future-study planning, but they cannot repair the absence of comparator outcome data in this record.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Bright Light")
n   <- c(6)
m   <- c(-11.3)
sp  <- c(11)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Bright Light  6  -11.300  11.000  -22.844  0.244

11. Analyses Reported by the Sponsor

The registry's reported analysis is a Sign test for the Bright Light group, with p = .046. The posted descriptive result gives a mean change of -11.3 mg*h/dL and a standard deviation of 11.0 mg*h/dL for six participants. These are not interchangeable summaries: the mean and standard deviation describe the magnitude and spread of changes, while the Sign test evaluates a question based on the directions of differences under its test assumptions.

A simple recalculation from a mean and standard deviation cannot recreate a Sign test, because those summaries do not retain the participant-level signs needed by that test. More importantly, a two-arm comparison cannot be recalculated from this endpoint record because no Dim Light participants were enrolled for the primary outcome. There is no recalculated between-arm estimate to compare with the registry's within-arm test. The registry record does not flag a discrepancy between a sponsor result and a recalculation; the limitation is that the available summaries do not support a like-for-like recalculation of the reported test or a comparative estimate.

The sponsor's posted test should therefore be read as the registry's analysis of Bright Light data, not as evidence of a difference between treatment conditions. The record does not report a confidence interval or the additional analysis details needed to independently reproduce the Sign test.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures and two analyses, and it names change in circadian phase measured by dim light melatonin onset as a secondary outcome. The primary outcome is change in glucose area under the curve. The multiplicity plan—how the study handled the possibility of testing more than one outcome or analysis—is not reported in the registry record. It is therefore not possible to determine whether the reported p-value was adjusted, whether a hierarchy of outcomes was prespecified, or how the different analyses were intended to contribute to interpretation.

Missing outcome information requires particular care. For the primary endpoint, the registry explicitly states that no participants were enrolled into the Dim Light condition. That is not the same as a reported Dim Light result of no change, nor does it establish why the condition has no endpoint participants. The record does not explain the difference between the overall enrollment entry of 10 and the primary outcome group record, which reports six Bright Light participants and none in Dim Light.

The registry does not describe how missing measurements were handled, whether endpoint data were incomplete for any other reason, or whether any imputation was used. It also does not report the analysis population definition, such as which participants were included under the analysis approach. These details matter because analysis populations and missing-data methods can affect which observations contribute to an estimate and how its uncertainty is calculated. No assumptions about these procedures can be recovered from the posted group summary alone.

Because the comparator outcome is absent, the principal limitation is not simply an uncertainty interval that is wide or a result that may be sensitive to missing values. The direct comparison required to estimate a between-arm difference is unavailable in the posted primary outcome record. The registry record does not report an interim analysis plan or a complete statistical analysis plan.

13. 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 adverse-event record reports zero serious adverse events among 10 participants at risk in the Bright Light group. For Dim Light, it reports zero participants at risk and zero serious adverse events. The record also lists zero other adverse events among 10 participants at risk in Bright Light and zero participants at risk in Dim Light. Deaths are recorded as zero for both groups. These entries describe the registry's recorded counts; they do not supply a comparative safety estimate.

The Dim Light group has no participants at risk in this record, so its zero event count should not be read as evidence that participants in that condition experienced no events. There are no participants in the recorded denominator from whom such a conclusion could be drawn. For Bright Light, the event count is reported alongside the stated at-risk count, but the registry record does not provide event timing, follow-up completeness, or further detail about the adverse-event collection process.

Zero recorded events in a small pilot does not establish that an intervention has no risk. Counts alone also do not show whether an event was related to treatment. The registry lists no serious-event terms. These figures are limited descriptive safety data, not evidence for a definitive comparison between the two registered conditions.

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

The posted primary result describes a mean reduction on the glucose area-under-the-curve change scale in the Bright Light group: -11.3 mg*h/dL, with a standard deviation of 11.0 mg*h/dL among six participants. The registry also records a Sign test p-value of .046 for that group. Taken together, these are a small set of arm-specific descriptive and test results. They do not provide a direct estimate of how Bright Light compares with Dim Light.

The absence of primary outcome participants in Dim Light prevents estimating the randomized contrast for this endpoint from the posted data. A change within Bright Light can reflect the treatment condition, but it can also reflect time, variation in measurements, participant-level factors, or other influences. Without a comparator outcome, those explanations cannot be separated through the registered between-group design.

The Sign test result does not answer the comparative question. A p-value is not a measure of effect size, does not quantify precision, and does not establish the probability that a treatment effect exists. The record does not provide a confidence interval for the primary result. The standard deviation describes variability among observed changes but does not show how precisely the mean change estimates a broader population value.

The intended population was people with prediabetes and increased risk for type 2 diabetes, and the project was described as a pilot to inform a larger trial. Results from this small, incomplete comparative endpoint record should not be treated as a definitive estimate for a broader population or as proof of a causal effect. The registry's stated hypothesis is a study rationale, not a finding established by the posted result.

General statistical context: an estimate can be numerically different from zero while still being uncertain, and statistical significance does not by itself establish practical importance. Conversely, a non-significant result would not prove that no effect exists. Interpretation depends on the estimated effect, its precision, design quality, endpoint completeness, and the comparison needed to address the scientific question. Here, the missing comparator result is a fundamental constraint on what the primary outcome can establish.

15. Why This Trial Matters Statistically

This study illustrates a basic distinction between a registered trial design and the outcome data available for analysis. The registry identifies a randomized parallel-group comparison of Bright Light with Dim Light, but the primary outcome record includes Bright Light observations only and states that no participants were enrolled in Dim Light for that endpoint. As a result, the posted data describe change in one group rather than the contrast the trial design was intended to support.

That distinction is important when interpreting within-group tests. A Sign test can assess a feature of changes in the Bright Light group under its assumptions, but it cannot tell whether those changes differ from the outcome under Dim Light. Even a small p-value does not transform a single-group analysis into a randomized comparison. The endpoint's mean, standard deviation, and p-value answer related but distinct statistical questions.

The record also demonstrates why reporting the analysis population, missing-data approach, confidence intervals, and multiplicity plan matters. Without those details, readers cannot fully assess which observations contributed to the analysis, how uncertainty was quantified, or how the reported test fits among the study's outcomes and analyses. The record's pilot purpose makes these questions especially relevant to planning further research, because future design choices depend on the quality and limits of the observed data.

Finally, the adverse-event counts show how an empty group denominator changes what a zero count can mean. No events recorded in a group with no participants at risk is not evidence of an absence of events among exposed participants. Careful reading of denominators, outcome availability, and analysis purpose is essential for avoiding stronger conclusions than the registry record supports.

16. 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 most consequential limitation specific to this trial's posted primary outcome is the absence of Dim Light participants. The registry says no participants were enrolled into that condition for the endpoint, and there is no comparator mean or standard deviation. A between-arm effect cannot be estimated from the endpoint record, despite the randomized parallel-group design designation.

The Bright Light result is based on six participants. The registry enrollment entry is 10, but the record does not explain the difference between overall enrollment and the group count contributing to the primary outcome. It does not report whether this reflects enrollment, eligibility, treatment exposure, outcome availability, or another issue, so no explanation should be inferred.

The primary outcome analysis is reported as a Sign test, but the record does not give the test statistic, exact null hypothesis, participant-level data, confidence interval, or full analysis-population definition. The mean and standard deviation alone are not sufficient to reproduce a Sign test. Missing-data handling, the multiplicity plan, and the statistical analysis plan are also not reported in the registry record.

The study is described as a pilot, and the primary outcome summary is limited to one arm. Consequently, the result offers little basis for assessing the precision of a comparative effect or the variability of outcomes across both registered conditions. The posted adverse-event record likewise has no Dim Light participants at risk, limiting safety comparisons. These limitations are specific to the available registry record and do not establish why data are absent or whether other analyses exist outside it.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Morning Light Treatment to Improve Glucose Metabolism — NCT03188263.
• Registry record last updated 2021-01-20; status last verified 2020-12.
• Results first posted 2020-12-29.
• 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.

20. Record Summary

NCT03188263 is a completed randomized, triple-masked, parallel-group pilot sponsored by Northwestern University. It was designed to test home-based morning Bright Light against Dim Light in people with prediabetes, with the stated aim of informing a larger clinical trial. The primary endpoint is change in glucose area under the curve from baseline to after four weeks of treatment.

The posted primary outcome reports six Bright Light participants, a mean change of -11.3 mg*h/dL, a standard deviation of 11.0 mg*h/dL, and a Sign test result of p = .046. The registry states that no participants were enrolled in Dim Light for this endpoint. Thus, the posted result is a within-group summary and does not estimate a Bright Light versus Dim Light difference. The record does not report a confidence interval, a complete analysis plan, or the reasons for the absent comparator outcome.