Environmental Treatment for Seasonal Affective Disorder (SAD): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Environmental Treatment for Seasonal Affective Disorder (SAD) (NCT00139997), a randomized na study. The primary outcome is Percentage SIGH SAD (Structured Interview Guide For The Hamilton Depression Rating Scale), SAD Version.

2. Clinical Question

3. Trial Design

The registry describes a multicenter, randomized controlled trial with a parallel-group model and quadruple masking. It lists seasonal affective disorder as the condition and Yale University as the sponsor. The study status is completed, with an enrollment of 26. The record lists the start as 2002-01 and primary completion as 2005-08. Phase is recorded as NA.

Participants were assigned to LED phototherapy or an inactive intervention. The LED group used Litebook treatment devices for 30 min before 8 am. The comparator is described as equivalent exposure to an inactivated negative ion generator. The masking designation is quadruple, but the registry record does not specify which participant, clinician, assessor, or other roles were masked, or how masking was maintained.

The primary outcome was a continuous measure: percentage SIGH SAD at Week 4, reported as a mean and standard deviation. The analysis population is described as intention-to-treat with last observation carried forward. The posted arm counts are 15 for LED phototherapy and 11 for inactive intervention. The registry lists one secondary outcome, SIGH SAD, and two outcome measures overall. It lists one analysis.

The registry record does not report the original sample-size justification, the trial’s own recruitment target, any interim-analysis plan, or the original statistical analysis plan. Those details should not be inferred from the post hoc planning illustrations or from the reported outcome analysis.

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

Analysis population for the primary outcome, as defined in the registry: These participants are analyzed ITT- last observation carried forward.

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 Percentage SIGH SAD (Structured Interview Guide For The Hamilton Depression Rating Scale), SAD Version, measured over Week 4.

The primary endpoint is percentage SIGH SAD at Week 4. SIGH SAD is the Structured Interview Guide for the Hamilton Depression Rating Scale, Seasonal Affective Disorder version. The registry description says that additional items address atypical symptoms associated with SAD, including increased sleep, weight, appetite, and fatigue. The scale has a minimum score of 0 and a maximum score of 63; higher scores indicate greater depression intensity. A score of less than 9 is described in the record as consistent with normal mood and the absence of major depression.

The endpoint is not simply the raw Week 4 score. It is calculated as the SIGH SAD score at trial end multiplied by 100 and divided by the score at randomization. This expresses the endpoint score relative to the participant’s starting score. A lower percentage is consistent with a lower score at trial end relative to randomization, while a higher percentage indicates a higher relative score. This interpretation follows the stated calculation and direction of the scale; it does not by itself establish why scores changed.

A continuous symptom measure allows the comparison to retain information about the score level rather than reducing each participant to a categorical outcome. The endpoint is reported as a group mean and standard deviation. The registry record does not provide an endpoint-specific rationale for choosing Week 4 or further details about how the percentage score was handled when a baseline score was zero.

7. Results Posted in the Registry

• Inactive Intervention: mean 60.40 versus 34.50 on control; difference 25.90 (95% CI 5.10 to 46.70), standardized effect d = 0.93; z-test p = 0.015.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous, so the registry summarizes it using group means and standard deviations. A mean gives the average score in a group. A standard deviation describes the spread of participant scores around that group average; it is not the uncertainty around the mean and is not itself a confidence interval. In these results, the mean percentage SIGH SAD score is 34.5 for LED phototherapy and 60.4 for inactive intervention, with standard deviations of 30.47 and 23.61, respectively.

What does the mean difference describe?

The reported comparison is a mean difference, with inactive intervention contrasted against LED phototherapy. Its estimate is 25.9. A positive value in this stated direction means the inactive-intervention group had the higher average percentage score. Since higher SIGH SAD scores represent greater depression intensity, and the endpoint expresses the trial-end score relative to randomization, the direction of the comparison is important. The difference is in percentage-score units; it should not be described as a difference in raw SIGH SAD points.

A mean difference is a group-level contrast. It does not say that every participant in one group had a higher score than every participant in the other group, nor does it describe the range of individual responses. Participant-level distributions and changes are not reported in the registry record.

What does a t-test assess?

The registry reports a two-sided t-test and p = 0.028. In general, a t-test assesses whether a difference in group means is compatible with a null hypothesis of no mean difference, under the assumptions of the test. A two-sided test considers departures in either direction rather than only a difference favoring one specified group. A p-value is a measure of how unusual the observed test result would be under the test’s null model; it is not the probability that the null hypothesis is true, nor does it measure the size or importance of the group difference.

The registry’s reported analysis does not include a test statistic, an estimated difference, or confidence limits. The interval accompanying the summary-statistic comparison is part of an unadjusted recalculation. It is useful for describing uncertainty in that recalculated contrast, but it is not a substitute for details that the registry does not provide about the sponsor’s analysis.

Why can a recalculation differ from the registry analysis?

The posted recalculation uses the summarized group results and gives p = 0.015. A recalculation from aggregate means, standard deviations, and group sizes cannot necessarily reproduce an analysis conducted from participant-level data or under additional modeling choices. The registry record does not describe any such choices, so it would be speculative to attribute the difference between p = 0.015 and p = 0.028 to a particular adjustment, assumption, or analysis population. The record does not flag a discrepancy. The recalculation should be clearly distinguished from the registry’s own two-sided t-test.

In general, over-dispersion occurs when variability is greater than a statistical model expects. If a model understates variability, its uncertainty estimates may be too narrow. This is a general statistical consideration, not a finding established for this trial. The registry record does not report a model-based analysis or an assessment of over-dispersion.

What does intention-to-treat with last observation carried forward mean?

Intention-to-treat generally aims to analyze participants according to their randomized group, preserving the comparison created by random assignment. The registry specifically describes the primary outcome population as ITT with last observation carried forward. In general, last observation carried forward fills a missing later measurement with that participant’s last recorded observation. This approach can affect the results when measurements are missing, and its suitability depends on the context and assumptions about the missing outcomes.

The registry record does not report how many participants lacked a Week 4 measurement, whether all randomized participants contributed to the reported means, or whether the carried-forward values were used for every missing observation. It also does not report sensitivity analyses using alternative missing-data assumptions. Those omissions limit assessment of how missing observations may have affected the comparison.

What cannot be inferred from the analysis listing?

The record lists two outcome measures, one secondary outcome, and one analysis. These counts do not explain whether the primary analysis was prespecified in a protocol, whether other analyses were considered, or whether a multiplicity adjustment was used. The original statistical plan, interim analyses, and sample-size justification are not reported in the registry record. As a result, the available information supports a description of the posted comparison but not a detailed audit of the analysis plan.

General statistical interpretation separates the estimated difference, its uncertainty, and the p-value. The estimate conveys direction and magnitude on the endpoint scale; an interval conveys a range of values compatible with the analysis assumptions; and the p-value summarizes evidence against a specified null model. None of these alone establishes clinical importance, causality beyond the randomized comparison as conducted, or the likely result in another population.

10. Confidence Intervals and Precision

The recalculated mean difference is 25.9, with an interval from 5.105 to 46.7. This interval describes uncertainty around the estimated difference under the recalculation’s assumptions. Its endpoints are both on the percentage SIGH SAD scale and use the stated comparison direction: inactive intervention relative to LED phototherapy. The interval is not a prediction range for individual participants, and it does not indicate that every participant’s outcome lies within those bounds.

In general, a confidence interval combines an estimate with information about its sampling uncertainty. A wider interval indicates less precision than a narrower interval on the same scale. The lower and upper bounds communicate which values remain compatible with the analysis, given its assumptions. An interval should be interpreted together with the study’s design, endpoint, sample size, and analysis population; it does not account automatically for bias, missing-data uncertainty, or unreported analytic choices.

The registry’s own analysis identifies its test as 95%, but reports no confidence-interval bounds for that analysis. The interval accompanying the unadjusted recalculation is presented separately from the registry-reported test, and the record does not attach a stated confidence level to that recalculated interval. The two summaries should not be merged as though they were a single sponsor-reported analysis.

Enrollment was 26, divided into groups of 15 and 11. Such group sizes make the precision of the estimate important to consider; the interval is more informative about uncertainty than the point estimate alone. The registry record does not provide additional precision measures or participant-level data that would allow a fuller assessment of the distribution of outcomes.

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 planning figures are participants per arm, not total events. The sample-size assumptions specify a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as a stated fraction of the observed effect. This is a planning illustration based on the observed result, not the original trial’s design assumptions, which are not reported in the registry record.

For planning at 100% of the observed effect, the illustration gives 19 participants per arm for 80% power and 25 participants per arm for 90% power. For planning at 75% of the observed effect, it gives 33 participants per arm for 80% power and 44 participants per arm for 90% power. For planning at 50% of the observed effect, it gives 73 participants per arm for 80% power and 97 participants per arm for 90% power.

In general, power is the probability that a study will detect a specified effect under stated assumptions if that effect is present. The 80% column describes the planned group size for an analysis with 80% power under those assumptions; the 90% column gives the corresponding group size for 90% power. Higher planned power requires more participants when the other assumptions are held fixed. Smaller effects also require more participants to distinguish them from random variation, which is why the illustrations increase as the fraction of the observed effect decreases.

These figures are not observed power. They do not show what power the completed trial achieved, and they do not establish what enrollment the trial originally planned. They are planning illustrations using the observed comparison as a reference, with assumptions that may differ from the trial’s actual design.

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 = 25.900; pooled standard deviation σ = 27.818.
2. n = 2 × (1.96 + 0.842)² × 27.818² / 25.900² = 18.1, rounded up to 19 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 <- 25.9                      # observed difference in means
sigma <- 27.818019939121                     # 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(139997)
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 = 19 (80% power), 25 (90% power)
 75% of observed effect: n per arm = 33 (80% power), 44 (90% power)
 50% of observed effect: n per arm = 73 (80% power), 97 (90% power)
power.t.test: 19.1 
Simulated power at n = 19 per arm: 0.795 (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 analysis is a two-sided t-test and reports p = 0.028. It does not report the sponsor analysis’s estimated mean difference or confidence limits in the posted analysis entry. Separately, an unadjusted recalculation from the posted arm summaries gives a mean difference of 25.9, with an interval from 5.105 to 46.7 and p = 0.015.

The recalculation is a summary-statistic analysis and should not be presented as the sponsor’s test. The two p-values are different, and the registry record does not explain the reason. It does not report sufficient detail about the sponsor’s analytic implementation to reconcile them. The record does not flag a discrepancy, so the difference should be described without asserting that it reflects a data error or a particular modeling decision.

The registry result is the trial’s reported analysis; the recalculation is a supplementary, unadjusted check based on the posted group summaries. They address the same broad group comparison, but they are not interchangeable. Interpretation should retain that distinction and should not treat the recalculated p-value as if it had been reported by the sponsor.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures overall, including one secondary outcome, and one analysis. The primary endpoint is percentage SIGH SAD at Week 4. The listed secondary outcome is SIGH SAD, described as the Structured Interview Guide for the Hamilton Depression Rating Scale, SAD Version. The record does not provide a numerical result for that secondary outcome in the posted results.

When a study examines multiple endpoints or conducts multiple statistical tests, the chance of obtaining at least one apparently unusual result can increase. This is the general issue of multiplicity. Researchers may address it through a prespecified hierarchy, an adjustment to the testing procedure, or other planned methods. The registry record does not report whether a multiplicity strategy was specified or used, how the two outcome measures relate to the one listed analysis, or whether additional analyses were conducted.

The record’s analysis count does not establish that only one analysis was planned or performed. Nor does the count establish that the reported endpoint was selected without reference to the results. The original protocol and statistical analysis plan are not reported in the registry record, so prespecification and any adjustment for multiple testing cannot be confirmed from the posted information.

The primary population is described as intention-to-treat with last observation carried forward. In general, intention-to-treat analysis seeks to retain participants in their randomized groups, while last observation carried forward substitutes an earlier observation when a later measurement is unavailable. The registry does not report the number of missing Week 4 outcomes, the number of carried-forward values, or the reasons for missingness. It also does not report sensitivity analyses that test whether conclusions change under alternative assumptions about missing data.

The record does not provide enough information to assess whether missingness differed by group or whether the last-observation method was prespecified. These unknowns do not invalidate the posted result, but they limit how fully its robustness can be evaluated. No unreported missing-data procedure or multiplicity adjustment should be assumed.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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 serious adverse event figures list 0 events among 15 participants at risk in the LED phototherapy group and 0 among 11 participants at risk in the inactive-intervention group. Thus, no serious adverse events are recorded for either group in this entry. The record lists no serious adverse event terms.

Other adverse events are listed for 2 participants at risk of 15 in the LED phototherapy group and 1 participant at risk of 11 in the inactive-intervention group. These are counts in the registry entry; they do not provide descriptions of the events, their severity, timing, or relationship to the intervention. The registry record does not report deaths in the posted entry.

These figures are limited to the events and categories reported. A count of zero serious events in a group does not establish that an intervention has no serious risks, particularly when the trial has a small enrollment. The record does not specify the safety observation period or provide additional detail about event collection, adjudication, or handling of participants who discontinued.

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

The posted results show different average Week 4 percentage SIGH SAD scores between the randomized groups: 34.5 for LED phototherapy and 60.4 for inactive intervention. The comparison estimate is 25.9 for inactive intervention relative to LED phototherapy. Because higher SIGH SAD scores reflect greater depression intensity, this direction describes a higher average relative score in the inactive-intervention group. The registry’s own two-sided test reports p = 0.028, while the separate unadjusted recalculation reports p = 0.015.

The results do not establish that every participant in one group experienced a particular outcome, or that the difference was the same for all participants. The group means and standard deviations cannot describe individual trajectories or the full distribution of scores. The percentage endpoint is a relative score derived from the randomization score and trial-end score, rather than a raw endpoint score alone.

The findings also do not establish the size of any effect in other populations or settings. The registry record does not report enough eligibility and participant-characteristic details to assess representativeness. It does not provide the original sample-size rationale, protocol, complete statistical plan, detailed missing-data counts, or a reported numerical secondary outcome. These gaps constrain interpretation of robustness and generalizability.

A statistically reported difference is not by itself a determination of clinical importance. The endpoint description supplies context about the SIGH SAD scale, but the registry results do not report a threshold for a clinically important between-group difference. Nor do the posted data establish the mechanism underlying the difference or how long any group separation persisted beyond the Week 4 endpoint.

Accordingly, the result is best read as a comparison of the recorded endpoint in this randomized trial, with uncertainty and reporting limitations. It is not a guarantee of an individual outcome, a complete account of safety, or medical advice.

17. Why This Trial Matters Statistically

Statistically, this trial is useful as an example of how an apparently straightforward mean comparison depends on careful endpoint interpretation. The primary outcome is not the raw SIGH SAD score at Week 4, but a percentage calculated relative to the score at randomization. That distinction affects the units and meaning of the reported means and mean difference. Keeping the endpoint definition visible helps prevent percentage-score contrasts from being mistaken for raw scale-point changes.

The record also illustrates why sponsor-reported analyses and recalculations from aggregate summaries should be identified separately. The registry’s own two-sided t-test reports p = 0.028, while the unadjusted summary-statistic recalculation reports p = 0.015. The record does not explain the difference. Reporting both transparently avoids presenting the recalculation as the trial’s official analysis or inventing a reason for the distinct results.

Precision matters alongside statistical testing. The group means and the estimated contrast are accompanied by variability and an interval, while the trial enrolled 26 participants. A p-value alone would not communicate the uncertainty around the estimated difference or the scale on which it is expressed. The record’s limited detail on missing data and analysis planning also illustrates why reporting completeness shapes the strength of statistical interpretation.

The planning illustrations provide a further teaching point: sample sizes depend on assumptions about the effect to detect, power, allocation, and statistical method. These post hoc illustrations are not evidence about the original design or the probability that the completed trial detected an effect.

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.

Several limitations are specific to the information reported for this trial. Enrollment was 26, with 15 participants in the LED phototherapy group and 11 in the inactive-intervention group. The registry does not report participant characteristics or detailed eligibility criteria, limiting assessment of which people with SAD are represented.

The endpoint is a percentage of the Week 4 score relative to the randomization score. The registry description gives the calculation but does not report how participants with a randomization score of zero were handled. It also does not provide participant-level score distributions or individual change data. The posted means and standard deviations therefore cannot reveal the range or pattern of individual responses.

The registry’s analysis entry reports a two-sided t-test and p = 0.028 but gives no estimated difference or confidence limits for that analysis. The separate unadjusted recalculation reports different inferential results, and the record does not explain the difference. The original analysis plan, sample-size justification, and any interim-analysis plan are not reported in the registry record.

Although the primary population is described as ITT with last observation carried forward, missing-outcome counts, reasons for missingness, and sensitivity analyses are not reported. The record also does not state whether multiplicity was addressed or whether the listed analysis represents all analyses conducted. The masking designation is quadruple, but the roles covered by masking and the procedures used are not specified.

Safety reporting is also limited: the entry gives event counts but not event descriptions, timing, observation period, or deaths. The registry record does not report a numerical secondary-outcome result. These limitations do not change the values posted, but they restrict independent assessment of the analysis and the broader interpretation of the trial.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Environmental Treatment for Seasonal Affective Disorder (SAD) — NCT00139997.
• Registry record last updated 2013-03-06; status last verified 2013-01.
• Results first posted 2013-03-06.
• 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

This completed, randomized, parallel-group trial compared LED phototherapy with an inactive intervention for seasonal affective disorder. Enrollment was 26, and the registry identifies Yale University as the sponsor. The primary endpoint was percentage SIGH SAD at Week 4, analyzed as a continuous outcome in an ITT population with last observation carried forward.

The mean endpoint values were 34.5 for LED phototherapy and 60.4 for inactive intervention. The registry reports a two-sided t-test with p = 0.028. An unadjusted recalculation from the posted summaries gives a mean difference of 25.9 and p = 0.015. The original sample-size justification, detailed statistical plan, missing-data counts, and numerical secondary-outcome results are not reported in the registry record.