Investigating the Effects of Evening Light Exposure on Melatonin Suppression, Alertness and Nocturnal Sleep: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Investigating the Effects of Evening Light Exposure on Melatonin Suppression, Alertness and Nocturnal Sleep (NCT01586039), a randomized na study. The primary outcome is Melatonin Suppression.

2. Clinical Question

The study title identifies alertness and nocturnal sleep as additional interests. The registry lists sleep structure, sleep quality, subjective alerting response, and objective alerting response as secondary outcomes. Those outcomes broaden the clinical and physiological questions beyond melatonin, but their numerical results are not reported in the primary endpoint record. The registry record also does not report participant characteristics that would allow the reader to judge how closely the study population resembles a particular clinical or community population.

3. Trial Design

The registry classifies the study as randomized, single-masked, and factorial, with a within-subject design described in the study summary. The four light conditions combine two light sources—CFL and blue-depleted LED—with two illuminance levels, 90 lux and 50 lux. The primary comparisons are between the two sources at the same illuminance level. The factorial classification describes the arrangement of experimental conditions; it does not mean that an interaction result is available in the posted primary endpoint results.

The registry lists visible light as the intervention and records enrollment of 33. The primary endpoint results report n=8 for each of the four conditions. These numbers describe different aspects of the record, and the endpoint counts should not be used to infer how many people completed every condition or contributed to every outcome. The adverse-event record uses its own at-risk counts, which are not identical to every primary endpoint count.

The registry identifies the sponsor as Brigham and Women's Hospital and gives the study status as completed. The recorded study start is 2012-08, and primary completion is 2014-03. Phase is listed as NA. The record does not describe the randomization sequence, the exact role of masking, how condition order was assigned, or whether a washout interval separated exposures. Those procedural details are not reported in the registry record.

A within-subject design generally means that a participant experiences more than one condition, allowing comparisons to account for stable differences between participants. That general feature can be useful when outcomes vary substantially from one person to another. However, the posted results do not explain how repeated measurements were handled in the analysis, whether all participants completed all conditions, or how any incomplete condition data were treated. The test is described as a two-sided t-test, but the registry record does not specify whether it was paired or give the model 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 Melatonin Suppression, measured over 6-h constant posture interval of the light exposure.

The primary endpoint is melatonin suppression, reported as a percentage, during the 6-h constant posture interval of light exposure. The registry defines the measure using the melatonin area under the curve (AUC) during the exposure interval relative to the AUC measured in dim light on the previous day over the corresponding interval. AUC is a way of representing the overall concentration-time profile across an interval, rather than relying on a single measurement. The record states that higher values indicate more light-induced melatonin suppression.

Melatonin is relevant to the study's stated question because the investigators sought to compare non-visual responses to evening light. Measuring suppression over an interval captures a response across the exposure period. The comparison is not a direct measure of sleep quality or alertness; those are listed as additional outcomes. Nor does the endpoint, by itself, establish how any measured response relates to later sleep or health.

For each light condition, the registry reports a mean and standard error. At 90 lux, the mean is 71.4 for CFL and 70.1 for blue-depleted LED. At 50 lux, the mean is 54.3 for CFL and 32.5 for blue-depleted LED. The reported standard errors are 5.7, 7.2, 5.9, and 7.6, respectively, in the same order. These summaries describe the endpoint as recorded; they are not individual participant values. The registry does not provide confidence intervals for the arm means or for the between-condition contrasts.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The posted primary endpoint summaries are means of a continuous measure. A mean describes the average endpoint value within a condition, while the standard error describes the precision of that estimated mean under the analysis used. A standard error is not the same quantity as the standard deviation of individual responses. The registry reports the standard errors but does not report standard deviations or confidence intervals for the means.

What does the two-sided t-test address?

A t-test assesses whether the data are compatible with a null hypothesis of no difference in the quantity being compared, using a t-distribution to account for uncertainty in the estimated mean or contrast. A two-sided test considers departures in either direction. The registry reports this method for both source comparisons: p=0.73 at 90 lux and p=0.001 at 50 lux. The record does not provide the test statistic, degrees of freedom, exact contrast estimate, or confidence interval.

In a within-subject study, a comparison may be based on the difference between conditions for each participant. A paired t-test would analyze those within-person differences. That approach can account for the fact that measurements from the same person are related. The registry describes the study as within-subject but labels the reported method only as a two-sided t-test. It does not specify whether the analysis was paired, how repeated observations were modeled, or how participants with incomplete measurements were handled. Those details matter for interpreting the test and reproducing it.

How should the p-values be read?

A p-value is calculated under a null hypothesis and assumptions about the data and analysis. It describes how unusual data at least as discrepant as those observed would be if that null and those assumptions held. It is not a measure of the probability that a result arose by chance, and it is not the probability that the null is true. A relatively large p-value, such as 0.73, does not show that two conditions are identical. Evidence for similarity or equivalence would require an appropriate design and analysis, including a prespecified margin for what would count as a sufficiently small difference; such a margin is not reported in the registry record.

A smaller p-value, such as 0.001, does not by itself indicate whether an effect is large enough to matter. The estimated contrast and its uncertainty are needed to assess size and precision. The posted means indicate the direction of the difference in each comparison, but the registry does not give contrast confidence intervals. It is therefore not possible to judge the range of effect sizes compatible with the reported tests from the posted summaries alone.

What does the factorial structure contribute?

The four conditions combine light source and illuminance. Considering each source comparison within an illuminance level addresses whether the two sources differ at that level. A formal interaction analysis would address whether the source contrast itself changes across illuminance levels. Although the registry classifies the model as factorial, it does not report an interaction estimate or test. The factorial label should not be mistaken for evidence that every possible main effect or interaction was analyzed or posted.

In general, a factorial analysis can use all conditions together and estimate effects associated with each factor, as well as whether factors interact. The specific model, terms, covariance structure, and handling of repeated measurements determine what those estimates mean. Those details are not reported in the registry record. No particular factorial model can be inferred solely from the classification.

How do multiple outcomes affect interpretation?

The registry identifies one primary endpoint and lists sleep structure, sleep quality, subjective alerting response, and objective alerting response as secondary outcomes. It also records five outcome measures and five analyses. Testing multiple outcomes or comparisons can increase the chance of obtaining at least one small p-value even when no effects are present. Methods such as family-wise error control or false-discovery-rate procedures can address different multiplicity goals, but the registry record does not report whether any adjustment was used or define a testing hierarchy. The stated p-values should therefore be understood as the reported tests, without assuming a multiplicity procedure.

The record does not report a prespecified statistical analysis plan, a sample-size justification, interim analyses, or the analysis populations. A planning illustration based on an observed effect, when used, is not observed power and does not establish the precision of the actual trial. The registry gives assumptions for such planning illustrations, including a two-sided significance level of 0.05, equal allocation, and a normal approximation, but does not report a planning value in the primary endpoint results. The trial's design rationale and any planned power target are not reported in the registry record.

Finally, an analysis of a continuous endpoint depends on how the endpoint is distributed and how the test handles its variability. The registry describes a t-test, but does not report distributional checks, transformations, adjustments for covariates, or sensitivity analyses. These omissions do not show that such procedures were or were not used; they mean the record does not give enough detail to evaluate them. Statistical interpretation should therefore stay close to the posted summaries and reported tests, and avoid treating unreported analytic choices as established details.

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("Compact Fluorescent Light 90 Lux", "Blue-depleted LED Light 90 Lux", "Compact Fluorescent Light 50 Lux", "Blue-depleted LED Light 50 Lux")
n   <- c(8, 8, 8, 8)
m   <- c(71.4, 70.1, 54.3, 32.5)
sp  <- c(5.7, 7.2, 5.9, 7.6)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
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  Compact Fluorescent Light 90 Lux  8  71.400  16.122  57.922  84.878
CARM  Blue-depleted LED Light 90 Lux  8  70.100  20.365  53.075  87.125
CARM  Compact Fluorescent Light 50 Lux  8  54.300  16.688  40.349  68.251
CARM  Blue-depleted LED Light 50 Lux  8  32.500  21.496  14.529  50.471

11. Analyses Reported by the Sponsor

The registry reports the sponsor's primary endpoint results as condition-specific means and standard errors, together with two-sided t-test p-values. The 90-lux comparison has p=0.73, and the 50-lux comparison has p=0.001. It does not report a numerical estimate or confidence interval for either between-condition contrast. The arm means provide descriptive direction, but they are not a replacement for a reported contrast estimate with its uncertainty.

No recalculated contrast is available in the posted results, so there is no recalculated estimate to compare numerically with the sponsor's tests. Reconstructing an unadjusted contrast test from the arm summaries would require assumptions about the analysis, especially because the study is described as within-subject. The standard errors of separate means do not specify the within-person covariance needed for a paired contrast. It would therefore be inappropriate to present a newly calculated contrast or p-value from these summaries alone.

The registry record does not identify a disagreement between the sponsor's result and a recalculated result. That absence is not evidence of numerical agreement: no recalculated contrast is available for comparison. The sponsor's reported tests should be read as the analyses recorded, while recognizing that their precision and model details cannot be fully evaluated from the posted endpoint summaries.

12. Multiplicity, Missing Data and Analysis Populations

The registry identifies melatonin suppression as the primary outcome and lists sleep structure, sleep quality, subjective alerting response, and objective alerting response as secondary outcomes. It records five outcome measures and five analyses. The primary results described in the registry include two source comparisons, one at 90 lux and one at 50 lux. The record does not describe how these tests, the secondary outcomes, or any other analyses were organized into a family of hypotheses.

When many outcomes or comparisons are tested, the chance of at least one apparently unusual result can rise. Adjustment procedures can control different forms of error, but they also depend on which hypotheses are grouped together and the inferential goal. The registry record does not report whether multiplicity was addressed, whether a hierarchy was prespecified, or whether the reported p-values were adjusted. The p-values should be interpreted as the reported comparisons rather than assumed to represent multiplicity-adjusted evidence.

The record does not report a missing-data method, the number of participants with missing endpoint measurements, or the reasons for any missing measurements. It also does not specify whether the primary analysis included all randomized participants, only participants with endpoint data, or another analysis population. The primary endpoint results give n=8 in each condition, while the adverse-event record reports different at-risk counts for one condition. These counts arise in different parts of the record and do not establish how missing measurements were handled.

The study is described as within-subject, but the registry does not state whether participants needed data from every condition to enter the primary analysis or whether partial repeated measurements could be retained. It also does not describe imputation, sensitivity analyses, or checks for differences between participants with and without endpoint data. These are not details that can be inferred from the reported means and p-values.

Although the registry includes assumptions for planning illustrations, the record does not report the trial's own sample-size justification or target power. A planning illustration based on an observed effect is not observed power and should not be treated as a substitute for an original design rationale. Interim analyses and a prespecified statistical analysis plan are also not reported in the registry record.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The adverse-event record reports no serious adverse events in any of the four light conditions, and no deaths in any condition. For other adverse events, the record reports zero in CFL 90 lux, one in blue-depleted LED 90 lux, zero in CFL 50 lux, and zero in blue-depleted LED 50 lux. The at-risk counts are 8, 9, 8, and 8, respectively, in the same condition order. These are the registry's recorded counts, not estimates of comparative risk.

The one reported other adverse event occurred in the blue-depleted LED 90-lux group. The record does not provide a term for that event, its severity, timing, duration, or an assessment of whether it was related to the light exposure. No serious-event terms are listed. The at-risk count of 9 in that group differs from the primary endpoint n of 8 for the same condition; the record does not explain the difference. The two counts should not be treated as interchangeable.

Small event counts provide limited information about uncommon harms. Zero reported serious events does not establish that no such events could occur, and the registry does not supply enough detail to assess event ascertainment or follow-up. These figures describe what the registry records; they do not establish a causal relationship between a condition and an event.

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

The primary results describe melatonin suppression under four evening-light conditions. At 90 lux, the reported means for CFL and blue-depleted LED are 71.4 and 70.1, with a reported two-sided t-test p-value of 0.73. At 50 lux, the means are 54.3 and 32.5, with p=0.001. Since the registry defines higher values as more light-induced suppression, the 50-lux means are directionally consistent with less suppression under the blue-depleted LED condition than under CFL. These findings relate only to the reported endpoint and comparisons.

The results do not establish equivalence between light sources at 90 lux. A p-value of 0.73 is not an equivalence test, and the registry does not report an equivalence margin or confidence interval for the contrast. The 50-lux p-value does not, on its own, describe the size or practical importance of a difference. The absence of reported contrast estimates and confidence intervals limits what can be said about precision.

The primary endpoint is not a direct measure of sleep quality, sleep structure, alertness, or health. Although those outcomes are listed in the registry, their numerical results are not included in the primary endpoint results. The melatonin findings therefore do not answer whether the light conditions differed on those other outcomes, nor do they establish that an observed melatonin response led to a later sleep effect.

The record also does not report an interaction test, detailed repeated-measures analysis, multiplicity adjustment, missing-data method, or analysis populations. That limits interpretation of differences across illuminance levels and assessment of the overall evidence. The most defensible reading is a description of the reported arm summaries and sponsor tests, with uncertainty about the size and precision of the contrasts and without extending the conclusions to unreported outcomes or broader effects.

These results should not be turned into a general claim about all evening lighting or all individuals. The trial compares specified light sources under the conditions recorded in the registry. The record does not provide enough detail about participant characteristics or exposure procedures to establish how the findings apply outside those study conditions.

15. Why This Trial Matters Statistically

This trial illustrates why a study's design labels and its posted analysis details need to be read together. The registry calls the design randomized, single-masked, factorial, and within-subject. Those descriptions point to a structured comparison of light sources and illuminance levels, with repeated conditions described in the study summary. Yet the primary results give only condition means, standard errors, and two p-values. They do not report a full model, a contrast estimate, or confidence intervals.

The distinction between a condition mean and a within-person contrast is especially important. In a within-subject design, the information in a comparison depends on how the same participant's outcomes relate across conditions. Arm-level summaries alone do not convey that relationship. A p-value can incorporate information from an analysis not reconstructible from those summaries, but without analysis details the reader cannot fully inspect how the test was formed.

The result also demonstrates why a non-small p-value is not proof of no effect, and why a small p-value is not a measure of effect size. The 90-lux test does not establish that the sources are equivalent. The 50-lux test is evidence against the test's null under its assumptions, but the record does not supply a contrast interval to show the precision or range of plausible sizes. Both distinctions are central to careful interpretation of clinical and laboratory comparisons.

Finally, a factorial arrangement invites questions about whether the effect of one factor varies with another. Separate source comparisons at each illuminance level do not answer that question by themselves. The lack of a posted interaction result, multiplicity details, missing-data methods, and analysis population highlights how registry summaries can be informative yet incomplete for statistical appraisal. Transparent reporting of estimates, uncertainty, and model choices would make the evidence easier to evaluate.

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 primary endpoint record is limited to condition means, standard errors, group counts, and two-sided t-test p-values. It does not report contrast estimates or confidence intervals, so the magnitude and precision of the source comparisons cannot be fully assessed. The means indicate direction descriptively, but do not provide the uncertainty around the between-condition differences.

The study is described as within-subject, but the registry does not say whether the t-tests were paired or provide details of the repeated-measures model. It does not report the randomization sequence, condition order, washout procedures, or how exposure sessions were scheduled. These omissions prevent assessment of how order or carryover was handled. They do not establish that such issues occurred; they mark details absent from the record.

Although the design is classified as factorial, no interaction estimate or test is reported. The separate comparisons at 90 lux and 50 lux cannot alone establish whether the source contrast changes with illuminance. The registry also does not report a multiplicity strategy, missing-data procedure, analysis population, prespecified analysis plan, sample-size justification, or interim analyses. The degree to which the reported tests reflect a prespecified analysis cannot therefore be determined from the registry record.

The primary endpoint group counts are n=8 for each condition, while the adverse-event record gives an at-risk count of 9 for blue-depleted LED at 90 lux. The record does not explain that difference, and the counts refer to different endpoint contexts. Enrollment is recorded as 33, but the record does not describe the flow from enrollment to each analysis or the completeness of each participant's measurements.

Finally, the primary results do not include numerical findings for the four listed secondary outcomes. Participant characteristics and the detailed event information for the one reported other adverse event are also not reported in the registry record. These limitations constrain both the statistical assessment and the scope of conclusions that can be drawn.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Investigating the Effects of Evening Light Exposure on Melatonin Suppression, Alertness and Nocturnal Sleep — NCT01586039.
• Registry record last updated 2021-09-14; status last verified 2021-09.
• Results first posted 2021-03-05.
• 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

The registry record concerns NCT01586039, “Investigating the Effects of Evening Light Exposure on Melatonin Suppression, Alertness and Nocturnal Sleep.” It lists a randomized, single-masked, factorial study sponsored by Brigham and Women's Hospital, with enrollment of 33 and completed status. The recorded start is 2012-08, and primary completion is 2014-03.

The primary endpoint is percentage melatonin suppression during the 6-h constant posture interval of light exposure. The four conditions compare CFL and blue-depleted LED at 90 lux and 50 lux. The registry reports means, standard errors, and two-sided t-test p-values for the source comparisons: p=0.73 at 90 lux and p=0.001 at 50 lux. It lists sleep structure, sleep quality, subjective alerting response, and objective alerting response as secondary outcomes.

The record does not report contrast confidence intervals, a detailed repeated-measures model, multiplicity handling, missing-data methods, or analysis populations. Serious adverse events and deaths are recorded as zero in each condition; one other adverse event is recorded for blue-depleted LED at 90 lux, with event details not reported.