Bioequivalence Study of Tretinoin Gel 0.05% to Brand Tretinoin Gel: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bioequivalence Study of Tretinoin Gel 0.05% to Brand Tretinoin Gel (NCT02249767), a randomized phase 3 study. The primary outcome is Percent Change in Baseline Acne Lesions at Week 12.

2. Clinical Question

In statistical terms, the central questions are how lesion-change averages differed across the randomized groups and how precisely those differences were estimated. The vehicle arm provides a comparison for each tretinoin arm, while the direct generic-versus-brand question is suggested by the study title but is not included among the posted recalculated comparisons. The record therefore supports interpretation of the reported arm summaries and the two comparisons with vehicle, not an unreported direct comparison or a formal bioequivalence conclusion.

3. Trial Design

The registry lists this as a completed phase 3, randomized, parallel-group study. It records quadruple masking, while the brief summary calls the study double-blind. Both descriptions indicate masking, but the record does not explain the difference in terminology or identify the roles covered by the quadruple-masking label. The study had three treatment groups: generic tretinoin, brand tretinoin, and placebo vehicle.

The brief summary describes treatment of the full face once daily for 84 days. Acne lesions were graded by a single blinded observer at screening and at Weeks 0, 2, 4, 8, and 12, with baseline identified as Week 0. The primary outcome concerns the change at week 12. The registry lists enrollment as 574; that figure is distinct from the smaller per-protocol group sizes used for the posted primary outcome summaries.

The primary endpoint analysis population is identified as per protocol. The record also gives intent-to-treat counts and the stated exclusions from that population: Generic Tretinoin had 222 ITT, with 22 excluded for not following protocol, and 200 in the per-protocol population; Brand Tretinoin had 220 ITT, with 11 excluded, and 209 in the per-protocol population; Placebo had 107 ITT, with 7 excluded, and 100 in the per-protocol population. The reason for using the per-protocol population for the primary endpoint, and the detailed criteria for protocol adherence, are not reported in the registry record.

The record gives a study start of 2013-11 and primary completion of 2014-06. It does not report the original sample-size rationale, randomization procedure, allocation concealment details, or the trial's prespecified statistical analysis plan.

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: This is Per protocol population (PP) Analysis. Generic Tretinoin Intent to Treat (ITT) started 222- 22 excluded as not following protocol= 200 PP. Brand Tretinoin 220 ITT - 11 excluded not following Protocol=209 PP. Placebo 107 ITT - 7 excluded = 100 PP Percent change from baseline in acne lesions

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 Percent Change in Baseline Acne Lesions at Week 12, measured over Baseline and 12 weeks.

The primary endpoint is percent change from baseline in inflammatory and non-inflammatory acne lesions at week 12, reported in the registry with the unit “Percent reduction.” It is a continuous measure: each participant’s lesion change is represented numerically, and the arm summaries report a mean and standard deviation. The outcome is assessed between baseline and 12 weeks.

In general, a continuous endpoint can preserve information about the size of change rather than reducing every participant’s experience to a simple category. A mean describes the average value in a group, while the standard deviation describes how spread out individual values are around that average. The group summaries can therefore show both the average reported change and the variability in that change.

The posted mean values are negative. Read alongside the registry’s percent-reduction description, they represent reported reductions from baseline; the sign is part of the way the change is expressed and should not be confused with a count of lesions. The endpoint description combines inflammatory and non-inflammatory lesions, and the posted primary summaries do not give separate mean results for those lesion types.

Using a baseline-to-week-12 endpoint provides a common comparison time across the randomized groups. However, the registry record does not report how the percentage was calculated for each participant, how multiple lesion locations were combined, or how missing week 12 measurements were handled. Those details matter when interpreting exactly what the group mean represents, and should not be inferred from the outcome label alone.

7. Results Posted in the Registry

• Generic Tretinoin: mean -32.34 versus -26.38 on control; difference -5.96 (95% CI -13.57 to 1.65), standardized effect d = -0.19; z-test p = 0.125.
• Brand Tretinoin: mean -37.32 versus -26.38 on control; difference -10.94 (95% CI -18.38 to -3.50), standardized effect d = -0.36; z-test p = 0.004.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous, and the registry identifies its summary parameter as a mean with standard deviation. General statistical interpretation begins by separating the estimated average from the uncertainty around a comparison. The mean describes the center of the observed group values; the standard deviation describes variability among those values. Neither measure alone explains how the comparison was calculated or whether an adjustment was made for other factors.

What does a mean difference represent?

For each posted comparison, the recalculated mean difference contrasts a treatment-arm mean with the vehicle-arm mean. A negative estimate means the treatment-arm mean is lower on the reported numerical scale. Because the outcome is expressed as percent change and the posted means are negative, a more negative value corresponds to a larger reported reduction from baseline. The estimate is not a ratio, and it does not state how many participants experienced a particular response.

The recalculated comparison for generic tretinoin versus vehicle is -5.96. The comparison for brand tretinoin versus vehicle is -10.94. These are unadjusted calculations from the posted summary statistics. The registry record does not describe a covariate-adjusted analysis, a model-based estimate, or the exact inferential method used in any original sponsor analysis. Thus, the recalculations should be understood as straightforward contrasts of the reported means rather than estimates that account for additional variables.

What does a confidence interval add?

In general, a confidence interval conveys the uncertainty around an estimated quantity. Its endpoints mark values compatible with the data under the method used to calculate the interval; a wider interval indicates less precision than a narrower interval. The generic-versus-vehicle interval runs from -13.57 to 1.646, and the brand-versus-vehicle interval runs from -18.38 to -3.496. The generic comparison’s interval includes zero, while the brand comparison’s interval lies below zero. The interval calculation method and confidence level are not reported in the registry record.

A confidence interval should not be read as a range containing a fixed proportion of individual participant outcomes. Individual variation is described in part by the standard deviations, whereas the interval concerns uncertainty in the estimated mean difference. Nor does an interval alone establish clinical importance: that judgment would require a meaningful threshold, which is not reported in the registry record.

How should the p-values be understood?

A p-value is a measure of how unusual data at least as discrepant as the observed data would be under a specified null hypothesis and statistical model. It is not the probability that the null hypothesis is true, nor does it measure the size or importance of a treatment effect. The posted unadjusted p-values are 0.125 for generic tretinoin versus vehicle and 0.004 for brand tretinoin versus vehicle. They are recalculated from the posted summaries, not identified as the trial’s own reported test results.

The record does not report whether the original analysis used a particular significance threshold, whether it adjusted for repeated testing, or how the trial’s primary comparison was defined. The planning illustration separately lists a two-sided significance level of 0.05, but that is an assumption for the planning calculations and cannot be treated as the original trial’s analysis threshold.

Why does over-dispersion matter in general?

General explanation: when individual observations vary substantially around the group mean, the estimate of a group average is less precise than it would be with less variability, all else equal. A standard deviation helps describe that spread. If a statistical model assumes less variation than is present, uncertainty can be understated and intervals can become too narrow. The posted standard deviations are substantial relative to the mean differences, so the group spread is important context when reading the estimates. The registry does not report a model that assumes a particular dispersion structure, and no claim about model misspecification can be made from the summaries alone.

What does randomization contribute?

In general, random allocation aims to create groups that are comparable on average at the start of a study, including with respect to factors that may not be measured. It does not guarantee exact balance in any particular trial, and it does not resolve bias that may arise from post-randomization exclusions, missing outcomes, or analysis choices. The record identifies the trial as randomized but does not report the randomization sequence method, allocation concealment, baseline balance details, or the analysis population for any endpoint beyond the stated per-protocol primary outcome.

The endpoint was assessed by a single blinded observer, according to the brief summary. Blinding can reduce the possibility that knowledge of treatment assignment influences assessment, but the record does not explain how masking was maintained or whether the observer remained unaware of assignment at every assessment. The registry’s quadruple-masking designation and its double-blind summary are both recorded; their differing labels are not reconciled.

What is not established by these methods?

The reported primary outcome summaries do not establish the full analysis plan. The registry record does not report the planned handling of missing values, any adjustment for baseline lesion counts, the procedure for combining lesion locations, or a prespecified equivalence margin. It also does not report an inferential analysis directly comparing generic with brand tretinoin. Those omissions limit the conclusions that can be drawn from the simple unadjusted contrasts, even though the posted group means and their recalculated comparisons are interpretable as descriptive statistical summaries.

10. Confidence Intervals and Precision

Precision concerns how narrowly an estimate is located, not whether the result is important in practice. The recalculated generic-versus-vehicle mean difference is -5.96, and its confidence interval extends from -13.57 to 1.646. Because zero is within that interval, the posted summary statistics do not pinpoint the direction of the generic-versus-vehicle mean difference with the same clarity as an interval that lies entirely on one side of zero. The interval also spans values in both directions, so the point estimate should not be treated as an exact value.

The recalculated brand-versus-vehicle difference is -10.94, with a confidence interval from -18.38 to -3.496. This interval is entirely below zero, but it still covers a range of possible mean differences. The point estimate is therefore not the only plausible value consistent with the posted data and calculation. The unadjusted p-value for this contrast is 0.004; the corresponding generic comparison has an unadjusted p-value of 0.125.

In general, larger variability and fewer observations tend to make an estimated mean difference less precise, while more information tends to improve precision. The posted standard deviations are 30.35 for generic tretinoin, 28.81 for brand tretinoin, and 32.33 for vehicle. The endpoint analysis groups are reported as 200, 209, and 100, respectively. The adverse-event risk sets are separate figures and should not be substituted for these endpoint group sizes.

The registry record does not identify the confidence level or interval method, and these intervals are recalculated rather than described as sponsor-reported intervals. Precision should therefore be interpreted in the context of the posted summaries and the unadjusted calculation, not as evidence about the trial’s unreported prespecified analysis. A confidence interval also does not resolve whether the estimated difference is clinically meaningful or meets a bioequivalence criterion; no relevant threshold is reported in the registry record.

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 for 80% power and participants per arm for 90% power, not total events. They are planning illustrations based on the observed effect, not the trial’s original sample-size design. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect.

For the generic-versus-vehicle comparison, planning for 100% of the observed effect gives 426 participants per arm for 80% power and 570 participants per arm for 90% power. Planning for 75% of the observed effect gives 757 participants per arm for 80% power and 1,013 participants per arm for 90% power. Planning for 50% of the observed effect gives 1,702 participants per arm for 80% power and 2,278 participants per arm for 90% power.

For the brand-versus-vehicle comparison, planning for 100% of the observed effect gives 118 participants per arm for 80% power and 158 participants per arm for 90% power. Planning for 75% of the observed effect gives 210 participants per arm for 80% power and 281 participants per arm for 90% power. Planning for 50% of the observed effect gives 472 participants per arm for 80% power and 632 participants per arm for 90% power.

In general, the 80% power column describes a planning target with an 80% chance of detecting the specified effect under the stated assumptions if that effect is present; the 90% column uses a higher target. Smaller effects require more participants under the same assumptions, which is why the lower effect fractions correspond to larger per-arm figures. These are planning illustrations, not observed power.

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 = -5.960; pooled standard deviation σ = 31.022.
2. n = 2 × (1.96 + 0.842)² × 31.022² / -5.960² = 425.3, rounded up to 426 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 <- -5.96                      # observed difference in means
sigma <- 31.0218065000008                     # 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(2249767)
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 = 426 (80% power), 570 (90% power)
 75% of observed effect: n per arm = 757 (80% power), 1,013 (90% power)
 50% of observed effect: n per arm = 1,702 (80% power), 2,278 (90% power)
power.t.test: 426.2 
Simulated power at n = 426 per arm: 0.805 (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. Multiplicity, Missing Data and Analysis Populations

The primary endpoint is reported for the per-protocol population. The registry gives the group counts used for that endpoint and also describes intent-to-treat counts and exclusions for not following protocol: 222 ITT and 200 per protocol for generic tretinoin, 220 ITT and 209 per protocol for brand tretinoin, and 107 ITT and 100 per protocol for placebo. The record does not give a detailed definition of protocol adherence or explain the rationale for selecting the per-protocol population for the primary endpoint.

No corresponding ITT results for the primary endpoint are posted. The registry record does not report how participants with missing week 12 measurements were handled, whether any values were imputed, or whether sensitivity analyses assessed the effects of missing data. It also does not report the number of missing endpoint observations by group. These details are important because an analysis limited to participants who met protocol criteria may differ from an analysis that retains all randomized participants.

Multiplicity refers generally to the increased chance of a false-positive finding when multiple outcomes or comparisons are assessed without appropriate planning or adjustment. There are two posted recalculated comparisons against vehicle, with unadjusted p-values of 0.125 and 0.004. The registry record does not report a multiplicity adjustment, a testing hierarchy, or whether these comparisons were prespecified as co-primary or secondary tests. The p-values should therefore be read as unadjusted, not as evidence that any correction was applied.

The record lists one outcome measure and no posted secondary results, but that does not establish the complete set of outcomes or analyses that may have been planned. The trial’s original statistical analysis plan, missing-data strategy, and sample-size justification are not reported in the registry record. No assumptions about those choices can be inferred from the posted per-protocol summaries or from the planning illustrations.

14. 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 registry reports serious adverse events for risk sets of 230 in the generic tretinoin group, 229 in the brand tretinoin group, and 115 in the placebo vehicle group. The serious adverse event count was 0 in each group. Deaths were also reported as 0 in each group, and no serious adverse event terms are listed.

These figures are counts as recorded, not estimates of the likelihood of a future event. A recorded count of zero means that no serious adverse events were reported in these risk sets during the registry’s safety reporting; it does not establish that the risk is absent. The registry record does not provide further detail about event definitions, duration of safety follow-up, or how events were collected.

The risk sets used for safety reporting differ from the per-protocol group sizes given for the primary endpoint. The record does not explain the relationship between those populations, so the safety counts should not be treated as the denominator for the primary endpoint analysis. The other adverse-event counts were 49 for generic tretinoin, 46 for brand tretinoin, and 26 for placebo vehicle; these are separate from the serious adverse event counts and are not characterized further in the posted record.

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

The posted primary endpoint summaries describe average percent change in acne lesions from baseline to week 12 among participants in the per-protocol population. The generic tretinoin mean was -32.34, the brand tretinoin mean was -37.32, and the vehicle mean was -26.38. The recalculated unadjusted comparisons suggest different estimated contrasts with vehicle: -5.96 for generic tretinoin and -10.94 for brand tretinoin.

The confidence interval for generic tretinoin versus vehicle includes zero, whereas the interval for brand tretinoin versus vehicle lies below zero. Those statistical features describe the uncertainty associated with these recalculated comparisons. They do not establish clinical importance, and the record does not provide a threshold for judging whether a difference is meaningful to patients or clinicians.

The title identifies the study as a bioequivalence study, but the posted results do not include a formal generic-versus-brand comparison, an equivalence margin, or a prespecified decision rule. Separate comparisons of generic and brand products against vehicle are not a substitute for testing whether generic and brand fall within an agreed equivalence range. The registry record therefore does not support a formal conclusion about bioequivalence from the information posted.

The results also do not establish how the findings would apply outside the enrolled population, treatment schedule, and assessment period described in the record. The registry describes normal, healthy male and female children and adults ages 12 to 40 years with at least Grade 2 acne, treatment once daily for 84 days, and a week 12 endpoint. It does not report longer-term results or separate endpoint summaries for subgroups or lesion types.

These observations are statistical descriptions of the posted data, not medical advice. They should not be interpreted as a recommendation to use or avoid any treatment. The record’s missing analysis details, per-protocol primary population, and absence of a reported formal equivalence criterion are central to understanding what can and cannot be concluded.

16. Why This Trial Matters Statistically

This trial illustrates why a study title and its posted statistical evidence need to be considered separately. The study is titled as a bioequivalence study, but the posted recalculated comparisons are each against vehicle. Formal equivalence is a question about the difference between the generic and brand products in relation to a prespecified acceptable range. Without a reported direct comparison, equivalence margin, and decision rule, the title alone cannot establish that the posted results demonstrate bioequivalence.

The trial also shows how the choice of analysis population affects the interpretation of an endpoint. The primary outcome is reported for a per-protocol population, while the record also gives ITT counts and protocol-related exclusions. Readers need to know which participants contributed to each analysis and how missing observations were handled. The registry provides some population counts, but the missing-data method and corresponding ITT endpoint results are not reported.

Another statistical lesson is that point estimates, confidence intervals, and p-values answer different questions. The mean differences quantify estimated contrasts, the intervals describe uncertainty, and the unadjusted p-values summarize compatibility with a null hypothesis under the recalculation method. None alone measures clinical importance. The interval for generic tretinoin versus vehicle includes zero; the interval for brand tretinoin versus vehicle lies below zero. Those distinctions can be stated without overstating what the results mean.

Finally, planning illustrations depend on the effect chosen for detection. The figures increase when the target is a smaller fraction of the observed effect, under the stated assumptions. They are not a reconstruction of the trial’s original design rationale. This distinction prevents post hoc planning calculations from being mistaken for evidence about the original study’s design or its achieved power.

17. 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 outcome is reported for the per-protocol population, and the registry describes exclusions for not following protocol. The record does not detail the criteria used to classify adherence, give an ITT analysis of the endpoint, or report a sensitivity analysis. The consequences of excluding participants therefore cannot be evaluated from the posted results alone.

The endpoint description refers to inflammatory and non-inflammatory lesions together. Separate results for these lesion types, separate values by assessment location, and the method for combining lesion counts are not reported in the registry record. The record also does not explain the participant-level percentage-change formula or the treatment of any missing baseline or week 12 assessments.

The recalculated contrasts are unadjusted comparisons with vehicle. The posted results do not include a recalculated direct comparison of generic with brand tretinoin, and the registry record does not report an equivalence margin or formal bioequivalence decision rule. Multiplicity handling, the trial’s own statistical analysis method, and its original sample-size justification are also not reported.

The masking description is not fully consistent: the registry identifies quadruple masking, while the brief summary calls the study double-blind. The record does not reconcile those terms or explain which study roles were masked. It also does not provide details on randomization procedures or allocation concealment.

Finally, the registry’s posted safety figures give event counts and risk sets but little additional context about event definitions, collection, or follow-up. The per-protocol endpoint groups and safety risk sets differ, and their relationship is not explained. These reporting limitations restrict interpretation of both efficacy summaries and safety counts without implying that unreported methods or results had any particular form.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Bioequivalence Study of Tretinoin Gel 0.05% to Brand Tretinoin Gel — NCT02249767.
• Registry record last updated 2020-02-10; status last verified 2020-01.
• Results first posted 2020-01-13.
• 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.

21. Record Summary

The registry record describes a completed phase 3 randomized, parallel-group study of generic tretinoin gel 0.05%, brand tretinoin gel 0.05%, and vehicle for acne. The brief summary describes once-daily treatment of the full face for 84 days in normal, healthy male and female participants ages 12 to 40 years with at least Grade 2 acne. Lesions were assessed by a single blinded observer at baseline and scheduled follow-up visits.

The primary endpoint was percent change in inflammatory and non-inflammatory acne lesions at week 12. Posted means and standard deviations are for the per-protocol population. Recalculated unadjusted comparisons with vehicle were -5.96 for generic tretinoin and -10.94 for brand tretinoin. A direct generic-versus-brand estimate and formal equivalence criteria are not reported in the registry record.

Serious adverse event counts and deaths were reported as 0 in each arm. The registry does not report the original sample-size rationale, detailed missing-data methods, multiplicity strategy, or full statistical analysis plan.