Treatment of Patients With Blepharitis and Facial Rosacea: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of Patients With Blepharitis and Facial Rosacea (NCT00560703), a randomized phase 2 study. The primary outcome is Change in Bulbar Conjunctival Hyperemia.

2. Clinical Question

The endpoint is a graded assessment of redness on an ordered scale from 0, clear, to 4, severe, with half scores permitted. A more negative change represents a fall in the score over follow-up. The registry’s posted group means are -0.61 for COL-101 and -0.6 for placebo. The unadjusted mean difference, defined as COL-101 minus placebo, is -0.01. Its confidence interval crosses zero, and the unadjusted p-value is 0.957. Those figures describe the observed contrast and its uncertainty; they do not establish whether a difference of this size would matter clinically. The registry also posts an ANCOVA p-value of 0.578, without an adjusted effect estimate or confidence interval. The adjustment details and the clinical threshold for an important score change are not reported in the registry record.

3. Trial Design

The registry identifies NCT00560703 as a completed phase 2, randomized, quadruple-masked, parallel-group trial. It records enrollment as 70 and lists Galderma R&D as the sponsor. The recorded start is 2007-11 and primary completion is 2009-05. The two posted treatment groups are COL-101 (doxycycline, USP) capsules and placebo, described as a sugar capsule. The registry record does not state the method used to generate the random sequence or how allocation was concealed.

The posted primary outcome is change in bulbar conjunctival hyperemia from baseline to Week 12. The COL-101 group has a reported sample size of 46, with a mean change of -0.61 and standard deviation of 0.797. The placebo group has a reported sample size of 24, with a mean change of -0.6 and standard deviation of 0.707. These are the group sizes reported for this outcome; the registry entries do not explain whether any participants were excluded from the analysis or why outcome denominators might differ from enrollment.

Quadruple masking indicates that the registry records masking at that level, but which roles were masked and how masking was maintained are not reported in the registry record. Likewise, the protocol’s original sample-size calculation, planned allocation ratio, and original design assumptions are not reported. The planning illustrations attached to the posted results are calculations based on the observed effect and should not be mistaken for the trial’s original design rationale.

The registry lists two outcome measures, but no secondary outcome results are posted. The treatment period is not fully described by the endpoint timeframe alone: the primary measure is assessed from baseline to Week 12, while dosing details, adherence, co-interventions, and follow-up beyond that endpoint are 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.

Before assignment: Subjects were screened for eligibility within 6 weeks prior to the baseline visit. This 12-week study employed a two-treatment, parallel-group design with subjects randomized in a 2:1 ratio to active drug or placebo.

Overall Study

Reasons for not completing this period:

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 Change in Bulbar Conjunctival Hyperemia, measured over Baseline to Week 12.

The primary endpoint is change in bulbar conjunctival hyperemia from baseline to Week 12. Hyperemia refers to visible redness in the bulbar conjunctiva, and the registry describes grading from 0, clear, through 4, severe. Half scores are acceptable. The stated grades progress from normal appearance, through localized injection and pink or red coloration, to marked dark redness. The outcome is reported as a mean with a standard deviation for each group.

As a general statistical point, a baseline-to-follow-up change score expresses how the recorded outcome shifted during the assessment period. A negative change on this scale corresponds to a lower score at follow-up than at baseline, and therefore less redness on the grading scale. The posted mean changes are -0.61 with COL-101 and -0.6 with placebo.

The measure is used to quantify a clinical feature relevant to blepharitis in the enrolled population. Its ordered grades are categories, while the posted analysis summarizes them with means. A mean gives a concise group-level comparison, but it does not show how many individuals improved, stayed the same, or worsened. As a general measurement consideration, the numerical distances between ordered categories need not represent equal differences in the underlying clinical state. The registry record does not report a threshold for a clinically important change or additional information about the measure’s validation.

7. Results Posted in the Registry

• COL-101 (Doxycycline, USP) Capsules: mean -0.61 versus -0.60 on control; difference -0.01 (95% CI -0.37 to 0.35), standardized effect d = -0.01; z-test p = 0.957.

8. Statistical Interpretation

9. Statistical Methods Explained

What does the unadjusted mean difference represent?

The unadjusted comparison subtracts the placebo mean change from the COL-101 mean change. The posted values are -0.61 for COL-101 and -0.6 for placebo, giving a mean difference of -0.01. “Unadjusted” means this calculation uses the group summaries without incorporating covariates or other model-based adjustments. It is a straightforward comparison of the posted group averages, not necessarily the same estimand or method used in the sponsor’s ANCOVA.

A mean difference retains the units of the outcome: score points on the bulbar conjunctival hyperemia scale. Its sign depends on the order of subtraction. With COL-101 minus placebo, a negative value means the average change was numerically lower in the COL-101 group. Since a decrease on this scale reflects a lower hyperemia score, direction can be described, but the small point estimate should be considered together with its uncertainty and the absence of a reported clinical-importance threshold.

What does the standard deviation describe?

A standard deviation describes how dispersed individual observations are around a group mean. The registry reports standard deviations of 0.797 for COL-101 and 0.707 for placebo. These values are measures of variability in change scores, not standard errors of the means and not confidence limits. A standard deviation can help characterize the spread of observations, but group-level summaries cannot reconstruct the individual score distribution.

As a general statistical point, a standard error describes uncertainty in an estimated quantity, such as a group mean or difference, whereas a standard deviation describes variation among observations. Greater variability generally makes a mean difference less precise when other features are held constant. The registry’s posted recalculation reports the uncertainty for the group contrast through a confidence interval and p-value.

How should the confidence interval be read?

A confidence interval gives a range of effect estimates that are reasonably compatible with the data under the statistical method used. In repeated use of the same procedure under its assumptions, intervals constructed at the stated confidence level would cover the target value at that rate. It is not a probability statement that the particular true effect lies within the observed interval.

The unadjusted 95% confidence interval for the mean difference crosses zero. Zero represents no mean difference under the defined subtraction. Because the interval includes values on either side of zero, the data do not precisely distinguish a small difference favoring either group from no difference. The interval’s endpoints are not repeated; its relationship to zero is the key interpretive feature. The registry does not post an ANCOVA confidence interval.

What does a p-value say?

The unadjusted p-value is 0.957. In general, a p-value measures how surprising data at least as incompatible with a specified null hypothesis would be if that hypothesis and the model assumptions were true. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. A large p-value does not prove equality. In this trial, the p-value should be read alongside the estimated mean difference and its confidence interval.

The registry’s ANCOVA p-value is 0.578. ANCOVA commonly compares outcomes while accounting for selected covariates, but the covariates, model specification, and adjusted estimate are not reported in the registry record. It is therefore not possible to reconstruct or fully interpret the reported ANCOVA result from the registry entries alone. The unadjusted recalculation and ANCOVA p-value arise from different analysis approaches and should not be treated as interchangeable.

Why does the endpoint’s scale matter?

The outcome is described as an ordered categorical scale, with half scores permitted, and is summarized using a mean and standard deviation. As a general statistical consideration, ordered categories establish rank but do not necessarily establish equal spacing between successive grades. Mean-based comparisons can be useful summaries, but they rely on treating score changes as numerically interpretable. A reader may also want distributions, category frequencies, or a model suited to ordinal outcomes; those details are not posted as primary results in the registry record.

Over-dispersion is a general concern when the observed variability exceeds what a chosen model assumes. In such circumstances, standard errors and confidence intervals from a model that understates variability can be too narrow. No over-dispersion assessment or model diagnostic is reported in the registry record, and the posted outcome summaries alone do not permit such an assessment.

What can and cannot be inferred from the posted methods?

The registry identifies randomization, quadruple masking, a parallel design, and an ANCOVA analysis. Those labels provide useful design and analysis context, but do not disclose the randomization procedure, masked roles, covariates, missing-data approach, or analysis population. The sponsor’s reported analysis is not fully reproducible from the information posted. The unadjusted recalculation is consequently best understood as a transparent comparison based on the available group summaries, rather than a replacement for a prespecified trial analysis.

In general, a statistically non-significant comparison is not an equivalence test. Equivalence requires a prespecified margin and an analysis showing that the uncertainty lies within that margin; non-inferiority likewise requires a margin and appropriate direction-specific reasoning. The registry’s legacy label mentions non-inferiority or equivalence, but the margin and supporting effect estimate are not reported. No conclusion about either criterion follows from the label or from the ANCOVA p-value alone.

10. Confidence Intervals and Precision

Precision concerns how tightly the available data estimate the group contrast. The unadjusted mean difference is -0.01 score points, and the reported 95% confidence interval crosses zero. Because the interval includes both negative and positive values, it does not isolate the direction of the underlying difference. Its span indicates that the point estimate should not be read as an exact or highly certain treatment contrast.

The confidence interval is based on a recalculation from the posted group means, standard deviations, and group sizes. It is not the registry’s ANCOVA interval: no adjusted interval is posted. The p-value from this unadjusted recalculation is 0.957, while the registry reports p = 0.578 for ANCOVA. These values answer related but methodologically distinct questions, and neither p-value conveys the magnitude or clinical relevance of the effect.

As a general point, confidence interval width depends on factors such as sample size, variability, and the analysis method. The posted standard deviations are 0.797 in the COL-101 group and 0.707 in the placebo group. The registry does not report a clinically important difference against which to judge whether the interval excludes effects that would matter. It also does not post the adjusted estimate needed to assess precision under the sponsor’s ANCOVA.

Thus, the unadjusted result is centered close to no mean difference, but the interval does not establish equivalence. Precision should be judged from the range of effects still compatible with the data, not from the point estimate or p-value alone.

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 illustrations based on the observed unadjusted effect, not the trial’s original design calculation. They use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to the stated fraction of the observed effect: 100%, 75%, or 50%. The registry’s original design assumptions and sample-size justification are not reported in the registry record.

The figures are participants per arm for 80% power and participants per arm for 90% power. In plain language, the 80% column gives the planned number in each arm for a hypothetical study to have an 80% chance of detecting the specified effect under the listed assumptions, if that effect and assumptions hold. The 90% column uses a higher target chance of detection, which requires more participants per arm than the 80% target for the same assumed effect. These are planning targets, not a description of the trial’s achieved power.

The rows plan for 100%, 75%, and 50% of the observed effect. The figures rise as the assumed effect becomes smaller: detecting a smaller difference against the same background variability requires more information, commonly through a larger sample. The displayed values are participants per arm, not total events. This is not observed power; it is an illustration of sample-size requirements under stated hypothetical assumptions.

The observed unadjusted effect is close to zero, so planning directly around that estimate yields very large illustrative sample-size figures. Such calculations can be highly sensitive to an observed effect that is small or imprecise. They should not be taken as a recommended design or as evidence that the original trial was designed to detect the observed contrast.

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 = -0.010; pooled standard deviation σ = 0.768.
2. n = 2 × (1.96 + 0.842)² × 0.768² / -0.010² = 92,526.6, rounded up to 92,527 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 <- -0.01                      # observed difference in means
sigma <- 0.767740496905195                     # 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(560703)
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 = 92,527 (80% power), 123,867 (90% power)
 75% of observed effect: n per arm = 164,492 (80% power), 220,208 (90% power)
 50% of observed effect: n per arm = 370,107 (80% power), 495,468 (90% power)
power.t.test: 92527.6 
Simulated power at n = 92,527 per arm: 0.798 (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 reports the sponsor’s analysis as ANCOVA, with p = 0.578. It does not provide the adjusted effect estimate, confidence interval, covariates, or further model details. The separate unadjusted recalculation from the posted means and standard deviations gives a mean difference of -0.01, a 95% confidence interval crossing zero, and p = 0.957.

The difference between the p-values is not, by itself, evidence of an error. ANCOVA may account for information beyond the unadjusted group summaries, while the recalculation compares those summaries directly. Because the adjustment details and adjusted estimate are not posted, the two results cannot be reconciled in more detail from the registry record. The recalculation should be read as an unadjusted analysis, not as a substitute for the sponsor’s analysis.

The reported analysis carries a legacy non-inferiority-or-equivalence label. However, the registry does not report the margin, the adjusted effect, or the confidence interval needed to evaluate those frameworks. The label and p-value alone do not establish non-inferiority or equivalence. The registry entries do not flag a discrepancy between the sponsor result and the recalculation.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists two outcome measures, while no secondary outcome results are posted. The primary endpoint is identified as change in bulbar conjunctival hyperemia from baseline to Week 12. The registry record does not report a multiplicity adjustment, a hierarchy for testing outcomes, or a prespecified strategy for interpreting multiple comparisons. Without those details, it is not possible to assess how the primary analysis fits into a broader family of planned tests.

Missing-data handling is not reported in the registry record. The posted primary outcome groups have reported sample sizes of 46 for COL-101 and 24 for placebo, but the record does not explain how missing Week 12 measurements, withdrawals, or other unavailable data were handled. It also does not say whether the posted summaries include all randomized participants, only participants with observed endpoint data, or another analysis population. No assumptions for any imputation or sensitivity analysis are posted.

The registry identifies the analysis method as ANCOVA, but the covariates and model specification are not reported. The record does not describe the analysis population as intention-to-treat, per-protocol, or another defined group. The randomization and masking labels do not answer these analysis questions. These omissions limit assessment of whether the reported analysis followed the planned approach and how robust the result is to missing outcomes or alternative analysis populations.

These are limits of what can be learned from the registry record, not evidence that a particular procedure was or was not used. Protocol and analysis-plan details, including multiplicity control, missing-data procedures, and population definitions, are not reported in the registry record.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: non-hodgkin's lymphoma (4.2%), migraine headache (2.2%).

The posted serious adverse event figures list 1 affected participant among 46 at risk in the COL-101 group and 1 among 24 at risk in the placebo group. These are the group-specific counts recorded for serious adverse events. The registry also lists non-Hodgkin's lymphoma at 4.2% and migraine headache at 2.2% among the top serious terms, but does not provide enough detail in the registry record to assign those terms to a treatment group or connect each term to a specific participant count.

The record lists deaths as null for both groups, so death information is not reported in the registry record. It does not provide the event narratives, timing, duration, severity criteria, or attribution to treatment. The figures should therefore be read as descriptive safety reporting, not as a causal comparison. The other adverse event counts are not needed to interpret the serious-event figures and do not resolve the missing detail about serious events.

Differences in the posted counts should be interpreted cautiously, particularly because the group sizes differ and the registry does not provide a detailed safety analysis or uncertainty estimates. These figures alone cannot establish whether a treatment caused an event or compare safety reliably between groups.

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

The primary posted result is an unadjusted mean change of -0.61 with COL-101 and -0.6 with placebo, yielding a mean difference of -0.01. On the registry’s grading scale, decreases correspond to lower hyperemia scores. The estimate is close to zero, and its unadjusted confidence interval crosses zero. The posted data therefore describe similar average changes in the two groups in this analysis, while leaving uncertainty about the true difference.

That description is not the same as proving that the treatments are equivalent. Equivalence requires a clinically justified margin and a suitable analysis demonstrating that the plausible effects lie within that margin. The registry record does not report the margin or enough ANCOVA details to make that assessment. Nor does the unadjusted p-value of 0.957 prove there is no effect. A non-significant result can reflect a small effect, uncertainty, or both.

The result also does not establish whether a score difference would be meaningful to patients or clinicians. The registry does not state a threshold for clinical importance. Group means do not show the full distribution of individual responses, and no secondary outcome findings are posted. The primary measure is an ordered grading scale summarized by means, so interpretation also benefits from information about category distributions and measurement properties that are not included in the registry entries.

The sponsor’s ANCOVA p-value is 0.578, but its adjusted effect estimate and confidence interval are not posted. It should not be read as a measure of effect size, and it cannot be directly substituted for the unadjusted recalculation. The registry’s legacy analysis label does not supply the missing non-inferiority or equivalence margin.

In short, the posted primary summaries show a near-zero unadjusted average difference with uncertainty that includes both directions. They do not establish clinical benefit, lack of benefit, equivalence, or non-inferiority. Those conclusions require context and analysis details that are not reported in the registry record.

17. Why This Trial Matters Statistically

This trial illustrates why a point estimate, a p-value, and a confidence interval answer different questions. The unadjusted mean difference of -0.01 describes the estimated size and direction of the average contrast. The p-value of 0.957 describes compatibility with a null comparison under the unadjusted method and its assumptions. The confidence interval crossing zero communicates uncertainty about the direction and magnitude. No single statistic should be used as a stand-in for all three.

It also shows why a non-significant result is not the same as equivalence. The registry includes a legacy label associated with non-inferiority or equivalence, but the margin and supporting adjusted estimate are not reported. Without those, the label cannot support a conclusion that the groups are sufficiently similar for a prespecified purpose.

Another statistical issue is that an ordered clinical scale is summarized with means and standard deviations. This is a common compact summary, but it can hide how scores are distributed across categories and among individuals. The endpoint’s scale and the unreported clinical-importance threshold are essential context for translating a numerical difference into a clinical interpretation.

Finally, the record presents an ANCOVA p-value alongside an unadjusted recalculation. Their difference emphasizes that estimates depend on the analysis method and available details. Without the adjusted effect, covariates, missing-data strategy, and analysis population, readers cannot fully reproduce or compare the sponsor’s model. The study is therefore a useful example of both what registry summaries can communicate and the limits created when key analysis information is not reported.

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 trial-specific details needed for a fuller assessment are not reported in the registry record. These include the random-sequence method, allocation concealment, the roles included in quadruple masking, and how masking was maintained. The registry identifies a randomized, parallel design but does not provide these operational details.

The primary endpoint is an ordered score with half scores allowed, analyzed as a mean and standard deviation. The registry does not provide category-level distributions, a clinically important change threshold, or additional validation information. Those omissions limit evaluation of how the numerical change maps to meaningful differences in redness.

The registry posts the sponsor’s ANCOVA p-value without its effect estimate, confidence interval, covariates, or model specification. It does not report missing-data procedures, analysis-population definitions, multiplicity handling, or sensitivity analyses. Although two outcome measures are listed, no secondary results are posted. The original sample-size justification and design assumptions are also not reported, so the observed-effect planning illustration cannot be treated as the trial’s design rationale.

Safety reporting is limited in the registry entries: death information is not reported, and serious-event terms are not linked to treatment groups with event narratives or attribution. The adverse event figures are descriptive and do not provide a complete comparative safety analysis. These limits constrain interpretation of both the primary result and the broader trial record.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Treatment of Patients With Blepharitis and Facial Rosacea — NCT00560703.
• Registry record last updated 2021-02-18; status last verified 2011-12.
• Results first posted 2012-01-11.
• 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

NCT00560703 is recorded as a completed phase 2 randomized, quadruple-masked, parallel-group trial sponsored by Galderma R&D. The registry records a start of 2007-11, primary completion in 2009-05, and enrollment of 70. The comparison is COL-101 (doxycycline, USP) capsules versus a sugar-capsule placebo in patients with blepharitis and facial rosacea.

The primary outcome is change in bulbar conjunctival hyperemia from baseline to Week 12. The posted group means are -0.61 for COL-101 and -0.6 for placebo; the unadjusted mean difference is -0.01, with a confidence interval crossing zero and p = 0.957. The registry also reports an ANCOVA p-value of 0.578, without an adjusted effect estimate or confidence interval.

Interpretation is limited by unreported analysis details, missing-data procedures, population definitions, clinical-importance criteria, and the absence of posted secondary results. Serious adverse event counts are recorded for both groups, but detailed event and attribution information is limited.