A Study to Evaluate the Safety and Efficacy of Ruxolitinib Phosphate Cream Applied Topically to Adults With Atopic Dermatitis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Evaluate the Safety and Efficacy of Ruxolitinib Phosphate Cream Applied Topically to Adults With Atopic Dermatitis (NCT03011892), a randomized phase 2 study. The primary outcome is Mean Percentage Change From Baseline in Eczema Area and Severity Index (EASI) Score at Week 4 in Participants Treated With Ruxolitinib 1.5% Cream Twice a Day (BID) Compared With Participants Treated With Vehicle Cream BID.

2. Clinical Question

The registry record describes a randomized, triple-masked, parallel-group Phase 2 trial with 307 participants enrolled. Its primary question concerns a between-arm difference in the mean percentage change in EASI, not the proportion of participants who reached a specified response threshold. The result therefore describes group averages for the recorded endpoint and time point. It does not, by itself, characterize every participant’s response or establish how the result relates to outcomes beyond the recorded study period.

The record also lists several secondary questions, including EASI changes at other time points, EASI-50, investigator global assessment, and itch numerical rating scale outcomes. Those measures address related but distinct aspects of disease assessment. The primary comparison remains the prespecified focus identified in the registry record; interpretation of other outcomes depends on their own estimates and on the analysis details reported for them.

3. Trial Design

This was a Phase 2 randomized, triple-masked, parallel-group trial in adults with atopic dermatitis. The registry lists 307 participants as enrolled and the study status as completed. The recorded start date is 2017-01-09 and the primary completion date is 2018-01-10. The sponsor is Incyte Corporation.

The double-blind treatment period lasted 8 weeks for vehicle BID and for the ruxolitinib cream arms: 0.15% QD, 0.5% QD, 1.5% QD, and 1.5% BID. The triamcinolone comparator received triamcinolone 0.1% BID for 4 weeks, followed by vehicle BID for 4 weeks. Thus, the primary comparison is specifically between ruxolitinib 1.5% BID and vehicle BID, while the study also included dose-ranging comparisons and a topical active comparator.

After the double-blind period, participants who met the recorded criteria—protocol compliance and no safety concerns—could receive open-label ruxolitinib 1.5% BID for 4 weeks. This extension followed the Week 8 assessment and was not the randomized, masked primary comparison. The record describes transitions to open-label treatment from the listed double-blind arms.

For the primary outcome, the analysis population is described as the intent-to-treat population, defined in the record as participants randomized to vehicle BID or ruxolitinib 1.5% BID with available data. The exact original sample-size rationale, detailed randomization procedure, and fuller operational details of masking are not reported in the registry record. The existence of a randomized parallel design should not be taken to imply additional design or analysis details that the record does not state.

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: 307 participants randomized to 6 treatment groups (ruxolitinib cream 1.5% BID, or QD, 0.5% or 0.15% QD, vehicle BID, \& TAC 0.1% BID). Participants applied blinded study drug for 8 weeks, TAC 0.1% BID for 4 weeks followed by vehicle BID for 4 weeks in DB period. After DB period, participants who met criteria (compliant with the protocol and no safety concerns) offered OL treatment with ruxolitinib 1.5% BID for 4 weeks. Out of 261 participants who completed DB Period, 252 entered the OL Period.

Only the first 6 of 12 groups are shown.

Double Blind (DB) Period

Reasons for not completing this period:

Open Label (OL) Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants from Intent to treat (ITT) population, all participants who are randomized to vehicle BID or ruxolitinib 1.5% BID, with available data were included in the analysis.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Intent-to-Treat (ITT) population included all participants who are randomized to the study.

Only the first 6 of 7 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Mean Percentage Change From Baseline in Eczema Area and Severity Index (EASI) Score at Week 4 in Participants Treated With Ruxolitinib 1.5% Cream Twice a Day (BID) Compared With Participants Treated With Vehicle Cream BID, measured over Baseline and Week 4.

The primary endpoint is mean percentage change from baseline in EASI score at Week 4. EASI combines the proportion of body regions affected with the intensity of key signs of atopic dermatitis. It evaluates four regions—head and neck, upper limbs, trunk, and lower limbs—and weights those regions when forming a total score. The total EASI scale ranges from 0 to 72; a higher score indicates worse disease status. A negative change from baseline indicates improvement in the score.

Using percentage change allows the endpoint to express the Week 4 score relative to each participant’s baseline score, while the group mean summarizes the average change in the analyzed arm. The comparison between arms asks whether those average changes differ, rather than whether all participants experienced the same change. The registry reports the primary arm summaries as least-squares means with standard errors and identifies the sponsor’s comparison as a difference in least-squares means.

EASI is used as a composite clinical measure because it incorporates both the extent of affected body area and the severity of signs. That makes it different from a single symptom rating or a responder outcome such as EASI-50. A group mean can be informative about the overall direction and magnitude of change, but it does not show the distribution of individual outcomes, the proportion meeting a response threshold, or the participant-level relationship between EASI and other outcomes.

7. Results Posted in the Registry

• DB: Ruxolitinib 1.5% BID: mean -71.57 versus -11.90 on control; difference -59.67 (95% CI -77.79 to -41.55), standardized effect d = -1.34; z-test p = <0.001.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Mean Percentage Change From Baseline in EASI Score at Week 4 in Participants Treated With Ruxolitinib QD Compared With Participants Treated With Vehicle Cream BID

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • DB: Ruxolitinib 0.15% QD: mean -44.92 versus -11.90 on control; difference -33.02 (95% CI -51.20 to -14.84), standardized effect d = -0.75; z-test p = <0.001.
  • DB: Ruxolitinib 0.5% QD: mean -52.80 versus -11.90 on control; difference -40.90 (95% CI -59.15 to -22.65), standardized effect d = -0.92; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for DB: Ruxolitinib 0.15% QD than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for DB: Ruxolitinib 0.5% QD than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Mean Percentage Change From Baseline in EASI Score at Week 4 in Participants Treated With Ruxolitinib QD/BID Cream Compared With Participants Treated With Triamcinolone 0.1% Cream BID Followed by Vehicle

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • DB: Ruxolitinib 0.15% QD: mean -44.92 versus -59.54 on control; difference 14.62 (95% CI -3.40 to 32.64), standardized effect d = 0.33; z-test p = 0.112.
  • DB: Ruxolitinib 0.5% QD: mean -52.80 versus -59.54 on control; difference 6.74 (95% CI -11.35 to 24.83), standardized effect d = 0.15; z-test p = 0.465.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between DB: Ruxolitinib 0.15% QD and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between DB: Ruxolitinib 0.5% QD and the control. The interval gives the range of effects that remain compatible with the data.

Percentage Change From Baseline in EASI Score at Week 4

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • DB: TAC 0.1% BID/Vehicle Cream BID: mean -59.80 versus -15.50 on control; difference -44.30 (95% CI -66.29 to -22.31), standardized effect d = -0.84; z-test p = <0.001.
  • DB: Ruxolitinib 0.15% QD: mean -45.40 versus -15.50 on control; difference -29.90 (95% CI -54.70 to -5.10), standardized effect d = -0.50; z-test p = 0.018.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for DB: TAC 0.1% BID/Vehicle Cream BID than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for DB: Ruxolitinib 0.15% QD than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes extend the primary Week 4 EASI question to other time points, response thresholds, investigator global assessment, and itch numerical rating scale. The posted secondary analyses include mean EASI percentage changes for QD ruxolitinib comparisons with vehicle and comparisons with the triamcinolone-then-vehicle group, along with a separate descriptive EASI percentage-change outcome. This range of endpoints gives context about related disease measures, but the outcomes are not identical and should not be collapsed into one claim.

In broad terms, the posted continuous EASI estimates show negative mean changes across the listed treatment arms. The reported QD-versus-vehicle MMRM contrasts are negative, while the posted comparisons of two QD arms with the triamcinolone-then-vehicle group have confidence intervals that include zero. These are descriptions of particular reported comparisons, not conclusions about every secondary outcome or every treatment arm. The registry lists other secondary outcomes for which corresponding numerical results are not included in the record.

The secondary pattern is related to, but does not replace, the primary result: it concerns additional doses, schedules, endpoints, or time points rather than the registered primary contrast alone. Secondary analyses can help assess consistency, but the registry record does not report a multiplicity adjustment strategy. In general, when many outcomes and comparisons are examined, unadjusted p-values carry less weight because chance findings become more likely across a collection of tests. The primary analysis should retain its distinct status, and secondary findings should be interpreted cautiously, particularly where model details and a complete analysis plan are not reported.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Mean Percentage Change From Baseline in EASI Score at Week 4 in Participants Treated With Ruxolitinib QD Compared With Participants Treated With Vehicle Cream BID
• Mean Percentage Change From Baseline in EASI Score at Week 4 in Participants Treated With Ruxolitinib QD/BID Cream Compared With Participants Treated With Triamcinolone 0.1% Cream BID Followed by Vehicle
• Mean Percentage Change From Baseline in EASI Score at Week 2 and Week 8
• Percentage of Participants Who Achieve a ≥ 50% Improvement From Baseline in EASI (EASI-50) at Weeks 2, 4, and 8
• Percentage Change From Baseline in EASI Score at Week 4
• Time to Achieve EASI-50
• Percentage of Participants Achieving an Investigator's Global Assessment (IGA) Score of 0 to 1 Who Have an Improvement of ≥ 2 Points From Baseline at Weeks 2, 4, and 8
• Mean Change From Baseline in the Itch Numerical Rating Scale (NRS) Score at Weeks 2, 4, and 8

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: each participant’s EASI percentage change is numerical, and the group-level result is expressed as a mean. The registry reports least-squares means and standard errors for the primary arms. A least-squares mean is a model-based estimate of a group mean; unlike a simple arithmetic average, its value can depend on the statistical model and the observations incorporated by that model. The record identifies the sponsor’s primary comparison method as MMRM, or mixed model for repeated measures.

What does MMRM contribute?

In general, a repeated-measures model can analyze outcomes collected over multiple visits while accounting for the fact that measurements from the same participant are related. It can use the observed measurements within the specified model rather than reducing the analysis to one isolated visit-level comparison. The registry names MMRM for the sponsor’s primary comparison, but the full model specification—including covariance structure, fixed effects, and missing-data assumptions—is not reported in the registry record. Those details matter when trying to reproduce or fully assess a model-based estimate.

The posted primary arm results include a mean and standard error for each group. A standard error describes the statistical uncertainty in an estimated mean; it is not the spread of individual participant outcomes. The standard deviation, by contrast, describes variability among participant values. Confusing these quantities can make group precision appear more or less extensive than it is. The primary summaries give standard errors, while the listed recalculation derives an approximate pooled standard deviation from those standard errors and arm sizes.

How should the recalculation be understood?

The recalculated comparison is a simple, unadjusted calculation from the posted summary statistics. It estimates the difference between the two group means and uses an approximate pooled standard deviation for a standardized effect calculation. Because the source summaries are least-squares means with standard errors, this recalculation is not the same procedure as the sponsor’s MMRM. It does not reproduce the sponsor’s model, account for unreported model features, or provide an independent participant-level analysis. The word “recalculated” identifies the method and its limitations, not a separate trial result.

The recalculated p-value is unadjusted. In general, an unadjusted test evaluates a particular contrast without correcting for the number of outcomes or comparisons examined. If many tests are considered, the chance of at least one apparently noteworthy result can increase. The record lists multiple secondary outcomes and analyses, but the multiplicity strategy is not reported in the registry record. Consequently, unadjusted secondary p-values should be read as exploratory evidence rather than as results with a documented family-wise error-control procedure.

What do the confidence intervals describe?

A confidence interval expresses uncertainty around an estimate under the assumptions of the analysis. For the sponsor’s primary comparison, the registry reports a 95% confidence interval for the difference in least-squares means. In general, a narrower interval indicates greater precision than a wider one, although precision is not the same as clinical importance. An interval that excludes zero is consistent with a nonzero difference under the corresponding statistical procedure, but interpretation should also consider the outcome scale, model assumptions, and multiplicity.

The primary endpoint is a change score, and the contrast is expressed in the same percentage-change units. The sign of the difference is meaningful only with the stated arm order and endpoint direction in mind: negative EASI change indicates improvement, and the sponsor’s contrast is reported as ruxolitinib versus vehicle. The record’s comparison estimate and its confidence interval are therefore interpreted relative to that coding, not as a change in an individual participant’s score.

What is not established by the statistical methods?

The registry identifies randomization, triple masking, a parallel design, and the MMRM primary comparison. It does not report the original sample-size justification, the detailed statistical analysis plan, interim analyses, the full model specification, or a multiplicity adjustment strategy. It is not appropriate to infer those elements from the reported p-value or from the total enrollment. General statistical principles can explain how these methods work, but they cannot fill in unreported protocol choices.

The primary analysis population is described as the intent-to-treat population with available data among participants randomized to the two primary arms. This description communicates which randomized groups contributed and that available data were used. It does not, by itself, answer how each type of missing observation was handled or whether all randomized participants contributed to every model term. Such distinctions can affect interpretation of an analysis, especially when follow-up observations are incomplete.

Finally, the study includes several doses, schedules, comparators, and outcome measures. These design features make the full pattern of evidence relevant, but they also mean that the primary result should be distinguished from secondary contrasts. The sponsor’s primary MMRM result is the central analysis for the registered primary question; simpler recalculations and unadjusted secondary analyses provide context rather than changing that hierarchy.

11. Confidence Intervals and Precision

The sponsor’s primary estimate is a difference in least-squares means of -59.66, with a 95% confidence interval from -77.85 to -41.47. The interval lies entirely below zero. In the context of the contrast as reported, the point estimate and interval are consistent with a more negative mean EASI percentage change in the ruxolitinib 1.5% BID arm than in the vehicle BID arm. The interval communicates uncertainty around the estimated group difference, not a range containing a stated proportion of individual participant responses.

The recalculated estimate is -59.67, with a 95% confidence interval from -77.79 to -41.55. Its interval is also entirely below zero and is close to the sponsor’s interval. The sponsor’s interval comes from the MMRM analysis, whereas the recalculated interval comes from a simpler unadjusted approach using posted summaries. Their similarity is reassuring about the arithmetic relationship between the summaries and reported contrast, but it does not make the underlying methods identical.

Precision should be interpreted alongside the analysis population and outcome definition. The primary analysis includes available data from the two specified randomized groups, and the record does not fully describe missing-data assumptions or the model specification. The confidence interval also does not address whether the difference is clinically important; the registry record does not provide a threshold for interpreting the size of the between-group contrast. Nor does the interval address uncertainty from unreported design choices or multiplicity across all outcomes.

In general, confidence intervals are more informative than a p-value alone because they show a range of plausible effect estimates under an analysis. Here, the reported interval gives a direct scale for the Week 4 mean difference. It should be read as statistical precision for that contrast, not as proof of persistence, generalizability, or an effect on outcomes not measured in the primary analysis.

12. 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 sample-size figures are planning illustrations based on the observed effect, not the trial’s original design calculation. For these continuous-outcome comparisons, the figures are participants per arm, not a total across arms. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row uses the stated fraction of the observed effect as the effect to detect. The 80% power column gives the estimated participants per arm for an 80% chance of detecting that assumed effect under those planning assumptions; the 90% column gives the corresponding number for a 90% chance.

The illustrations are not observed power. They are hypothetical planning calculations, not evidence about the probability that the completed trial’s result is true or reproducible. They also do not establish what sample size the investigators originally intended, because the trial’s own design assumptions are not reported in the registry record.

Rows that plan for smaller effects require more participants per arm in the listed calculations. This reflects a general statistical principle: when an effect is harder to distinguish from variability, greater sample size can improve the chance of detecting it under specified assumptions. The estimates use the observed effect as their starting point, so they are sensitive to that trial estimate and should not be treated as independent evidence about the effect’s likely size.

The study includes multiple treatment contrasts, but the figures apply to the particular comparison and assumed effect represented in each planning illustration. They do not account for unreported features of the original design or establish a design suitable for another population, endpoint, or analysis plan.

13. 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 = -59.670; pooled standard deviation σ = 44.550.
2. n = 2 × (1.96 + 0.842)² × 44.550² / -59.670² = 8.8, rounded up to 9 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 <- -59.67                      # observed difference in means
sigma <- 44.5502485532219                     # 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(3011892)
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 = 9 (80% power), 12 (90% power)
 75% of observed effect: n per arm = 16 (80% power), 21 (90% power)
 50% of observed effect: n per arm = 36 (80% power), 47 (90% power)
power.t.test: 9.8 
Simulated power at n = 9 per arm: 0.757 (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.

14. Analyses Reported by the Sponsor

The sponsor reports the primary comparison using MMRM, with a difference in least-squares means of -59.66, a 95% confidence interval from -77.85 to -41.47, and p < 0.001. The recalculated unadjusted comparison from posted summaries is -59.67, with a 95% confidence interval from -77.79 to -41.55 and an unadjusted p-value of <0.001. The estimates and intervals are close, and both indicate a negative between-arm difference under the stated contrast.

The two results are not interchangeable. The sponsor’s estimate uses the reported MMRM analysis, while the recalculation uses arm-level summaries and an approximate unadjusted method. It cannot reproduce the model or account for details that are not reported in the registry record. The sponsor’s analysis is the relevant analysis for the registry’s primary question; the recalculation is a transparent check of the relationship between the posted summaries and the reported contrast.

The record does not flag a discrepancy between the recalculation and sponsor result. Differences in the last reported digits are compatible with the distinct calculations and rounding. The sponsor’s p-value and confidence interval should therefore remain the primary inferential result, with the simple recalculation used as supporting context rather than as a substitute.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary endpoint and multiple secondary outcomes, and it reports 24 analyses. Examining multiple endpoints, visits, treatment arms, or contrasts creates a multiplicity issue: in general, the more statistical tests considered, the greater the opportunity for a chance result to appear notable. The registry record does not report whether the study used a prespecified hierarchy, adjusted significance thresholds, or another multiplicity-control strategy. The primary result can be identified as the registered primary comparison, but unadjusted secondary p-values should not be interpreted as if a documented correction had been applied.

The primary population is described as the intent-to-treat population, consisting of participants randomized to vehicle BID or ruxolitinib 1.5% BID with available data. This definition identifies the randomized groups and indicates the use of available observations. It does not specify the full handling of missing values, the assumptions about why observations may be missing, or sensitivity analyses for missing data. Those details are not reported in the registry record. MMRM is named as the sponsor’s method, but the complete model specification and missing-data procedures are also not reported.

The posted primary arm summaries have different group sizes, and the record’s stated analysis population includes participants with available data. Without fuller accounting of follow-up and missingness, it is not possible to determine from the registry record alone how missing observations affected the model-based estimate. This limitation does not negate the reported result; it defines what can and cannot be evaluated from the recorded methods.

The record does not report interim analyses, the original sample-size justification, or a complete statistical analysis plan. Their absence from the record should not be interpreted as evidence that a particular procedure was or was not used. It means that readers cannot verify those design and analysis details from the registry entry.

16. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: myocardial infarction (2.0%).

The registry reports serious adverse events by listed treatment group and gives a number at risk for each group. It records zero serious adverse events for vehicle BID, the ruxolitinib 0.15% QD arm, the 0.5% QD arm, the 1.5% QD arm, and the 1.5% BID group. The triamcinolone 0.1% BID for 4 weeks followed by vehicle BID group has one serious adverse event. The registry records zero deaths in each listed group.

The number at risk is 52 for vehicle BID, 51 for the triamcinolone-then-vehicle group, 51 for each of the three QD ruxolitinib strengths, and 259 for the listed 1.5% BID group. These denominators are not alike. In particular, the 1.5% BID safety denominator differs substantially from the randomized primary analysis group size, so it should not be assumed to represent only the double-blind randomized arm. The record also lists myocardial infarction among serious adverse event terms at 2.0%, but does not provide an arm-specific attribution for that term in the reported summary.

These figures describe the adverse-event information recorded, not a complete causal assessment or a comparison adjusted for exposure. Differences in denominators, treatment periods, and the open-label extension limit direct comparisons across groups. The registry figures alone do not establish whether a treatment caused an event or characterize risks beyond the reported study record.

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

The primary result describes a Week 4 difference in mean EASI percentage change between adults randomized to ruxolitinib 1.5% cream BID and vehicle BID. The sponsor’s MMRM estimate is negative, and its 95% confidence interval lies below zero. In the context of the endpoint definition—where negative EASI change indicates improvement—the recorded group-level result favors a more negative mean change in the ruxolitinib arm during the primary comparison.

That statistical description is narrower than a general claim about treatment. It does not say that every participant improved, establish the size or importance of an individual response, or answer how the result applies beyond the participants and treatment conditions in this trial. It also does not address all possible outcomes, long-term follow-up, or treatment comparisons that were not part of the primary question. The record’s open-label period after Week 8 is distinct from the randomized and masked primary comparison.

A p-value is not the probability that the null hypothesis is true, nor does it measure the probability that the finding will replicate. The confidence interval describes uncertainty for the mean contrast under the sponsor’s model; it is not an individual prediction interval. Statistical evidence and clinical importance are different questions. The registry record does not provide a threshold for judging the clinical importance of the estimated between-arm difference.

The record also does not provide the full model specification, detailed missing-data assumptions, multiplicity strategy, or original sample-size rationale. These omissions constrain how completely the analysis can be reproduced or evaluated. The appropriate reading is therefore specific: the sponsor reports a statistically precise Week 4 mean difference for the primary comparison, while broader conclusions require information and outcomes not established by this registry record.

18. Why This Trial Matters Statistically

This trial is statistically informative because it separates a clearly identified primary comparison from a set of dose-ranging and secondary questions. The primary endpoint is a continuous EASI measure, and the sponsor reports a model-based difference in least-squares means with a confidence interval. Reading the estimate together with its interval makes the direction and precision of the recorded group difference more transparent than relying on a p-value alone.

The presence of dose-ranging arms is relevant to interpretation because the study assessed more than one ruxolitinib strength and dosing schedule, alongside vehicle and an active topical comparator. Yet those additional arms do not make every contrast equally central. The registered primary question concerns ruxolitinib 1.5% BID versus vehicle BID at Week 4; the other comparisons and outcomes should be understood as secondary evidence with their own uncertainty and, where unadjusted, multiplicity concerns.

The comparison between the sponsor’s MMRM result and the simple recalculation also illustrates an important statistical distinction. A close numerical match can support confidence that the reported contrast is consistent with the arm summaries, while the model-based and summary-statistic methods remain different analyses. A recalculation cannot recover unreported model details or substitute for participant-level analysis.

Finally, the record makes clear why transparent reporting matters. The named model and analysis population are useful, but missing-data procedures, multiplicity strategy, and original design assumptions are not reported. Statistical interpretation depends not only on an effect estimate, but also on the analysis choices and uncertainty that produced it.

19. 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 registry record identifies the primary comparison and sponsor analysis but does not report the complete MMRM specification. Details such as the covariance structure and the full set of model terms are therefore unavailable for independent reconstruction. The record states that the primary analysis included the intent-to-treat population with available data, but does not give a complete account of missing observations or sensitivity analyses under alternative missing-data assumptions.

The trial includes multiple doses, schedules, comparators, and secondary outcomes. A multiplicity adjustment strategy is not reported, so the secondary p-values cannot be assumed to control the overall chance of false-positive findings. In addition, some listed secondary outcomes do not have numerical results in the reported information. The available secondary evidence is therefore incomplete for an overall assessment of every registered outcome.

The simple recalculation is based on summary statistics and uses an approximate pooled standard deviation. It does not replicate the sponsor’s MMRM or use participant-level observations. It is useful as a check, but cannot assess the effect of covariate adjustment, correlation across repeated measurements, or unreported analysis decisions.

The safety summary has unequal numbers at risk across listed treatment groups, and the 1.5% BID safety denominator is notably different from the size of the randomized primary analysis group. The record does not provide enough detail in this summary to make direct, exposure-adjusted safety comparisons. It also does not provide an arm-specific breakdown for the serious adverse event term listed at 2.0%.

The original sample-size justification, interim-analysis details, and complete statistical analysis plan are not reported in the registry record. These limits affect reproducibility and the assessment of design choices; they do not, on their own, determine whether the reported estimate is valid.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study to Evaluate the Safety and Efficacy of Ruxolitinib Phosphate Cream Applied Topically to Adults With Atopic Dermatitis — NCT03011892.
• Registry record last updated 2021-04-09; status last verified 2021-03.
• Results first posted 2021-04-09.
• 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.

23. Record Summary

The registry record describes a completed Phase 2 randomized, triple-masked, parallel-group study of topical ruxolitinib in adults with atopic dermatitis. Enrollment was 307. The primary endpoint was mean percentage change from baseline in EASI score at Week 4, comparing ruxolitinib 1.5% cream BID with vehicle BID.

The sponsor’s MMRM analysis reported a difference in least-squares means of -59.66, with a 95% confidence interval from -77.85 to -41.47 and p < 0.001. A simple unadjusted recalculation from posted summaries was close to that estimate. The record also lists dose-ranging, active-comparator, open-label, secondary-outcome, and safety information, with important analysis details—including the multiplicity approach and full missing-data procedures—not reported.