Tralokinumab in Combination With Topical Corticosteroids for Moderate to Severe Atopic Dermatitis - ECZTRA 3: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Tralokinumab in Combination With Topical Corticosteroids for Moderate to Severe Atopic Dermatitis - ECZTRA 3 (NCT03363854), a randomized phase 3 study. The primary outcome is Participants With Investigator's Global Assessment (IGA) Score of 0 (Clear) or 1 (Almost Clear) at Week 16.

2. Clinical Question

The registry record describes a Phase 3 randomized, double-blind, placebo-controlled trial of tralokinumab with TCS for moderate to severe atopic dermatitis. Its primary objective was to assess whether the combination was superior to placebo with TCS for treating the condition. The endpoint focuses on a clinician-rated assessment of disease severity at a specified time, rather than a change in symptoms alone.

The central statistical question is about a group-level contrast: whether the share reaching the stated IGA threshold differed between the two initial treatment groups at Week 16, and how much uncertainty surrounds that contrast. The registry record also lists secondary questions about itch, disease severity, quality of life, topical treatment use, flares, days without topical treatment, and safety. Those outcomes add context, but they do not replace the primary comparison.

3. Trial Design

ECZTRA 3 (NCT03363854) is recorded as a Phase 3 randomized, double-blind, parallel-group trial sponsored by LEO Pharma. It enrolled 380 participants with atopic dermatitis. The registry record gives a start date of 2018-02-22, a primary completion date of 2019-03-08, and a completed status. The official title describes a randomized, double-blind, placebo-controlled trial of tralokinumab combined with topical corticosteroids in people with moderate to severe atopic dermatitis.

The primary comparison concerns initial treatment from Week 0 to Week 16: tralokinumab Q2W plus TCS versus placebo plus TCS. The posted primary endpoint analysis uses the full analysis set, described as 378 treated participants among the 380 randomized to initial treatment. The endpoint results list 252 participants in the tralokinumab group and 126 in the placebo group. These are the analysis denominators recorded for that endpoint.

The study also had a continuation period from Week 16 to Week 32. The arm descriptions indicate different continuation regimens according to initial treatment and response status, including continued tralokinumab, placebo continuation, or transition from placebo to tralokinumab. These later pathways are not the same comparison as the Week 16 primary endpoint, and their groups should not be read as if they were the original randomized comparison.

The stated secondary objectives covered disease severity and extent, itch, health-related quality of life, and safety. The brief summary describes safety assessment over 32 weeks. The original sample-size justification, detailed allocation procedure, and other original design assumptions 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: After the participant gave informed consent, they went through a 2- to 6-week screening period for washout of previous atopic dermatitis medication and disallowed medication. The participant was assigned treatment at Week 0.

Only the first 6 of 8 groups are shown.

Initial Treatment Period

Continuation Treatment Period

Safety Follow-up Period

Analysis population for the primary outcome, as defined in the registry: Full analysis set (FAS). Of the 380 participants randomised to initial treatment, 378 were treated. Therefore, the FAS consisted of 378 participants (252 + 126).

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 Participants With Investigator's Global Assessment (IGA) Score of 0 (Clear) or 1 (Almost Clear) at Week 16, measured over Week 16.

The primary endpoint is the number of participants with an IGA score of 0, clear, or 1, almost clear, at Week 16. IGA is described as a five-point assessment of atopic dermatitis severity, ranging from 0 for clear to 4 for severe. For this endpoint, the lower two categories are combined into a threshold outcome: each participant either meets that threshold at the specified assessment or does not.

This kind of endpoint is useful because it expresses an outcome in a form that is readily interpretable as the proportion of a group reaching a defined clinical state. It is not a measure of average score change, and it does not show how far participants outside the threshold may have moved on the scale. The binary classification also means that participants close to the cutoff can fall into different categories despite having similar underlying scores.

The registry record reports 98 participants meeting the endpoint among 252 in the tralokinumab-plus-TCS group, and 33 among 126 in the placebo-plus-TCS group. The sponsor's analysis describes the between-group result as a risk difference at Week 16. The endpoint is specific to that visit and definition; it does not by itself summarize symptom change across the full study, the continuation period, or safety.

7. Results Posted in the Registry

• Tralokinumab Q2W+TCS: 38.9% (98/252) versus 26.2% (33/126) on control; risk difference 12.7 percentage points (95% CI 2.9 to 22.5), risk ratio 1.48 (1.07–2.07); unadjusted two-proportion z-test p = 0.014.

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.

Reduction of Worst Daily Pruritus Numeric Rating Scale (NRS) (Weekly Average) of at Least 4 From Baseline to Week 16

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Tralokinumab Q2W+TCS: 45.4% (113/249) versus 34.1% (43/126) on control; risk difference 11.3 percentage points (95% CI 0.9 to 21.6), risk ratio 1.33 (1.01–1.76); unadjusted two-proportion z-test p = 0.037.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Tralokinumab Q2W+TCS than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change in Scoring Atopic Dermatitis (SCORAD) From Baseline to Week 16

• 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:
  • Tralokinumab Q2W+TCS: mean -37.70 versus -26.80 on control; difference -10.90 (95% CI -15.20 to -6.60), standardized effect d = -0.55; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Tralokinumab Q2W+TCS than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change in Dermatology Life Quality Index (DLQI) Score From Baseline to Week 16

• 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:
  • Tralokinumab Q2W+TCS: mean -11.70 versus -8.80 on control; difference -2.90 (95% CI -4.24 to -1.56), standardized effect d = -0.47; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Tralokinumab Q2W+TCS than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary results cover several kinds of outcome: an itch-response threshold, changes in SCORAD and DLQI, and additional measures involving topical corticosteroid use, flares, treatment-free days, and anti-drug antibodies. The results shown for the itch threshold and the two continuous scales are directionally consistent with the primary endpoint in that the reported comparisons favor the tralokinumab-plus-TCS group on the stated outcomes. The other listed secondary outcomes broaden the questions being asked, but their full numerical results are not all described in the registry record summarized in these analyses.

These measures capture different aspects of atopic dermatitis and treatment experience. The IGA threshold is a categorical clinician-rated status at a visit; the itch measure is a threshold response; SCORAD and DLQI are scale changes. Their estimates are not interchangeable, and agreement in direction does not make them repetitions of the same endpoint. They can provide complementary context about disease severity, symptoms, and quality of life.

Secondary results generally carry less confirmatory weight when many outcomes are examined and no multiplicity adjustment is reported. With multiple statistical tests, the chance of at least one small p-value under a collection of no-effect hypotheses can increase. The registry record does not report a multiplicity-control procedure, so the secondary p-values should be read as unadjusted evidence rather than as a family of confirmatory tests. This does not make the outcomes irrelevant; it means their role is more supportive and exploratory in the absence of a reported adjustment plan.

The registry record lists 17 posted secondary outcomes, but this count alone does not reveal which were prespecified for formal testing, how missing observations were handled, or whether any outcomes were selected for reporting. The secondary pattern should therefore be considered alongside the primary result and the limitations of the reporting record, without treating every nominal result as independent confirmation.

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

• Reduction of Worst Daily Pruritus Numeric Rating Scale (NRS) (Weekly Average) of at Least 4 From Baseline to Week 16
• Change in Scoring Atopic Dermatitis (SCORAD) From Baseline to Week 16
• Change in Dermatology Life Quality Index (DLQI) Score From Baseline to Week 16
• Frequency of Anti-drug Antibodies (ADA)
• Amount of Topical Corticosteroid (TCS) Used Through Week 16 Assuming no TCS Used From the Non-returned Tubes
• Amount of Topical Corticosteroid (TCS) Used Through Week 16 Assuming All TCS Used From the Non-returned Tubes
• Number of Atopic Dermatitis Flares Through Week 16
• Number of Days Without Topical Treatment Use From Baseline to Week 16

9. Statistical Interpretation

10. Statistical Methods Explained

Different statistical summaries answer different questions. For a binary outcome such as meeting the IGA threshold, the group counts and denominators describe how many participants met the definition in each treatment group. A risk difference compares the two group proportions by subtraction and expresses the contrast on an absolute scale. Unlike a relative measure, it directly describes the separation between proportions. The posted recalculation also includes relative measures, but the main sponsor-reported comparison is the risk difference.

What does the sponsor's method do?

The sponsor reports a Cochran-Mantel-Haenszel analysis for the primary risk difference. In general, this method can estimate a treatment-group contrast while accounting for specified strata. The registry record names the method but does not report the strata details or enough information to reconstruct all elements of the analysis. The reported estimate and interval should therefore be read as the sponsor's analysis, rather than as a calculation from the marginal group counts alone.

The recalculated result is a simple unadjusted analysis based on the posted counts. It does not reproduce the named sponsor method. Its purpose is to provide a transparent comparison derived from the displayed group summaries, not to replace the trial's reported analysis. The two estimates are close but not identical; differences can arise because the analyses use different methods. The registry record does not identify a discrepancy requiring a different interpretation of the sponsor's result.

How should a confidence interval be understood?

In general, a confidence interval describes the range of effect values reasonably compatible with the data and the assumptions of the analysis, at its stated confidence level. It communicates precision as well as direction. A narrower interval indicates a more precise estimate, while a wider interval leaves more uncertainty about the effect's size. The interval is not a guarantee that the true value lies within its endpoints, and it does not assign probabilities to individual values inside it.

For the binary endpoint, the reported interval is for the risk difference. Its interpretation depends on the scale used for that difference. A value of zero represents no difference between groups on the risk-difference scale; values on either side indicate the direction of the contrast. The sponsor-reported interval and the recalculated interval come from their respective analyses and should not be combined into a single interval.

What do the secondary analyses measure?

The secondary outcomes include binary and continuous measures. A binary outcome counts participants meeting a definition, such as the listed itch-response threshold. Continuous outcomes such as SCORAD and DLQI are summarized as least-squares means of change, with standard errors. In general, a least-squares mean is a model-based group estimate; a standard error describes the estimated sampling uncertainty of that mean. The sponsor reports repeated-measurement models for the SCORAD and DLQI comparisons. The registry record does not give the full model specifications.

The recalculated continuous comparisons use posted summary statistics and are identified as approximate where the standard deviations are estimated from standard errors. These calculations should not be mistaken for a re-fit of the sponsor's repeated-measurement model. Model-based analyses can use information and assumptions not recoverable from a simple comparison of group summaries.

Why does over-dispersion matter in some analyses?

In general, over-dispersion means that observed variability is greater than a statistical model expects. When present, a model that assumes less variability than the data contain can produce uncertainty estimates that are too small. The primary endpoint is binary and reported with risk differences; the registry record does not describe an over-dispersion adjustment for that analysis. This general issue is relevant when choosing a model for count or other outcomes, but no such adjustment should be attributed to this trial without a registry description.

The planning illustrations use the observed effect as a starting point and show what group sizes would correspond to selected fractions of that effect under stated assumptions. They are not results from the trial's original design calculation. The original sample-size rationale and detailed statistical plan are not reported in the registry record. No interim-analysis plan, multiplicity procedure, or comprehensive missing-data method is reported in the registry record either, so these features cannot be inferred from the endpoint estimates alone.

11. Confidence Intervals and Precision

The sponsor-reported primary risk difference is 12.4, with a 95% confidence interval from 2.9 to 21.9. The interval gives a range of effect sizes compatible with the reported analysis and its assumptions. Its lower endpoint is above zero, so the interval does not include the no-difference value on the risk-difference scale. The result is therefore directionally consistent with a higher endpoint proportion in the tralokinumab-plus-TCS group in the sponsor's analysis.

The interval nevertheless permits a range of magnitudes, from a more modest separation to a larger one. The point estimate should not be treated as an exact or certain effect. The confidence interval is the more complete description of precision because it displays uncertainty around the estimate rather than only its central value. It does not indicate whether the difference is important to patients, and it does not capture uncertainty arising from limitations that the analysis cannot quantify.

The recalculated unadjusted risk difference is 0.127, with a 95% interval from 0.029 to 0.225. This interval comes from the simple unadjusted calculation and is not the same analysis as the sponsor's Cochran-Mantel-Haenszel result. Both intervals exclude zero, but their numerical scales and methods should be kept distinct when describing the posted evidence.

Precision is also shaped by the analysis population and the endpoint's binary definition. The registry record identifies the full analysis set and the endpoint denominators, but does not report the missing-data handling in sufficient detail to assess how missing Week 16 outcomes were treated. Confidence intervals describe sampling uncertainty under an analysis; they do not resolve uncertainty about unreported analytic choices or the generalizability of the trial population.

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 illustrations are planning calculations based on the observed effect, not the trial's original design justification. For these comparisons, the entries are participants per arm, not total events. The stated 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. Time-to-event planning displays, if present for another outcome, use total events; that convention does not apply to these participant-per-arm entries.

The 80% power column describes the number of participants per arm that the planning calculation associates with an 80% chance of detecting the specified effect under the stated assumptions, if that effect and model assumptions hold. The 90% power column gives the corresponding planning quantity for a higher target probability of detection. For example, the primary endpoint illustration gives 213 participants per arm for 80% power and 284 for 90% power when planning for 100% of the observed effect. For 75% of the observed effect, the corresponding entries are 369 and 494; for 50%, they are 807 and 1,080.

Smaller effects are harder to distinguish from random variation, so the illustrations require more participants per arm as the target effect becomes smaller. These calculations are not observed power and do not establish what the trial was originally designed to detect. The original design assumptions and sample-size justification are not reported in the registry record. Treat the values as conditional planning examples, not as evidence that a particular future study will achieve its target under different conditions.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.262; treatment proportion p₁ = 0.389; average p̄ = 0.325.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6626 = 1.2987.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6565 = 0.5525.
4. n = (1.2987 + 0.5525)² / (0.1270)² = 212.5, rounded up to 213 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
p0 <- 0.261904761904762                          # control proportion
p1 <- 0.388888888888889                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(3363854)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 213 (80% power), 284 (90% power)
 75% of observed effect: n per arm = 369 (80% power), 494 (90% power)
 50% of observed effect: n per arm = 807 (80% power), 1,080 (90% power)
power.prop.test: 212.5 
Simulated power at n = 213 per arm: 0.804 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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's primary result is a Cochran-Mantel-Haenszel risk difference of 12.4, with a 95% confidence interval from 2.9 to 21.9 and p = 0.015. This is the trial's reported analysis. The recalculated comparison uses the posted group counts without adjustment and gives a risk difference of 0.127, with a confidence interval from 0.029 to 0.225 and an unadjusted p-value of 0.014.

The recalculation is useful as a check against the displayed counts, but it does not reproduce the sponsor's method. The estimates are not exactly the same because they are based on different analyses; the registry record does not provide the full details needed to reconstruct the sponsor's stratified analysis. The calculated p-value is also explicitly unadjusted, so it should not be substituted for the reported p-value when describing the trial's primary analysis.

Both analyses point in the same direction and their intervals exclude zero, but the sponsor's named method and reported result remain the principal trial analysis. Since the registry record does not flag a discrepancy, there is no separate discrepancy issue to resolve. The simple recalculation should be presented as complementary, not as a correction to the sponsor's result.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study tests several endpoints, time points, or treatment contrasts. In general, the more tests considered, the greater the chance of obtaining at least one apparently noteworthy result by chance if the tests are treated independently and no adjustment is made. The registry record lists 17 secondary outcomes and does not report a method for controlling multiplicity. It is therefore not possible to determine from the registry record whether the secondary analyses were adjusted, arranged in a formal testing hierarchy, or intended as exploratory comparisons.

The primary endpoint is identified separately, and the sponsor reports a superiority analysis for it. The record does not provide a complete statistical analysis plan or enough detail to establish all prespecified testing decisions. The original sample-size justification, interim-analysis plan, and any broader strategy for controlling type I error are not reported in the registry record. Their absence from the record should not be filled by assumptions about common trial practice.

The primary analysis population is described as the full analysis set: 378 treated participants among 380 randomized to initial treatment. The endpoint groups are recorded as 252 and 126. The registry record does not describe how participants without an available Week 16 assessment were handled, whether any values were imputed, or how intercurrent events were treated. It also does not report sensitivity analyses for missing outcomes. Without those details, the effect of missing data on the estimate cannot be assessed from the posted summary alone.

These reporting gaps matter because multiplicity and missing-data choices can affect how strongly an estimate supports a claim. They do not erase the reported group counts or sponsor analysis, but they limit independent evaluation of the full analysis process. The p-values should be interpreted in the context of the methods that are stated, while recognizing that unreported decisions remain unknown.

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: invasive ductal breast carcinoma (2.4%), appendicitis (1.4%), gastroenteritis clostridial (1.4%), ligament rupture (1.4%), hypoglycaemia (1.4%).

For the initial treatment period, the registry record lists 2 participants with serious adverse events among 252 receiving tralokinumab Q2W plus TCS, and 4 among 126 receiving placebo plus TCS. It also records 118 participants with other adverse events in the tralokinumab group and 36 in the placebo group. No deaths are listed for either initial treatment group. These are separate categories in the safety record; serious adverse events should not be combined with the broader other-adverse-event counts.

The continuation-period groups have different treatment histories and are divided by initial response and treatment pathway. Their counts are not direct substitutes for the randomized initial-period comparison. The safety follow-up is also reported separately. The listed serious-event terms include invasive ductal breast carcinoma, appendicitis, gastroenteritis clostridial, ligament rupture, hypoglycaemia, and wrist fracture; the registry record gives percentages for these terms but does not provide a basis for treating them as comparative risks between initial groups.

Adverse-event counts describe what was recorded during specified periods. They do not by themselves establish causality or provide a complete assessment of comparative safety. The registry record does not provide further detail about event adjudication or exposure-adjusted analyses.

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

The posted primary result describes a Week 16 difference in the proportion of participants reaching an IGA score of 0 or 1. In the sponsor's analysis, the estimated risk difference is 12.4, with a 95% confidence interval from 2.9 to 21.9 and p = 0.015. This is evidence about the specified endpoint, groups, analysis population, and time point. It is not a guarantee that an individual participant will have the same outcome as the group average.

The result does not describe every aspect of atopic dermatitis. A threshold endpoint does not show the complete distribution of IGA scores or the amount of improvement among participants who remain outside the threshold. It also does not summarize the entire course from Week 0 to Week 32, and it should not be confused with the continuation-period comparisons, which involve response-based treatment pathways.

The primary finding does not establish that every secondary outcome has the same degree of evidential support. Secondary outcomes measure different concepts and multiplicity handling is not reported in the registry record. Nor do the recorded adverse-event counts alone determine whether an event was caused by treatment. The registry results are a statistical account of group outcomes, not an individualized prediction or a substitute for clinical judgment.

Interpretation also depends on the population represented by the trial and the completeness of reporting. The record describes participants with moderate to severe atopic dermatitis, but does not provide all details needed to assess representativeness. Missing-data handling and some analysis-plan details are not reported. These limits do not change the values recorded for the analysis, but they constrain how broadly and confidently those values can be applied beyond the trial comparison.

18. Why This Trial Matters Statistically

This trial illustrates why the definition and scale of an endpoint matter. The primary outcome is not simply a change in average disease score; it is the proportion reaching a defined IGA category at a particular visit. The risk difference gives an absolute comparison between groups, while a confidence interval shows how uncertain the estimated contrast remains. Together, these summaries are more informative than a p-value alone.

The record also illustrates the distinction between a sponsor-reported analysis and a simple recalculation from posted group counts. A Cochran-Mantel-Haenszel analysis and an unadjusted calculation can yield slightly different estimates because they are not the same procedure. Clear labeling of method and analysis population helps readers avoid treating a recalculation as though it were the original model-based result.

Finally, the trial's primary and secondary outcomes address related but distinct questions. Itch, clinician-rated severity, and quality of life can offer complementary evidence, while multiple testing and missing-data decisions affect the strength of interpretation. Where these decisions are not documented in the registry record, statistical caution is warranted. The case shows how to read an estimated effect, its precision, and the boundaries of what the registry results establish without turning a statistical comparison into a broader claim than the data support.

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 does not report the original sample-size justification or the detailed assumptions used to plan the trial. The planning illustrations are based on the observed effects and should not be mistaken for the trial's original design calculations. The record also does not provide a full statistical analysis plan, including details of the strata used in the primary Cochran-Mantel-Haenszel analysis.

Important analysis details are not reported in the registry record: the approach to missing Week 16 outcomes, any imputation or sensitivity analyses, and the strategy for multiplicity across secondary outcomes. These omissions make it harder to independently assess robustness and the confirmatory status of secondary findings. The presence of several posted outcomes increases the importance of knowing which comparisons were prespecified and how inference across them was handled.

The primary endpoint is a binary threshold at Week 16. It does not capture the entire score distribution or describe the time course of response. The continuation period uses response-based pathways and different treatment sequences, so those groups do not preserve the simplicity of the initial randomized comparison. Interpretation of later-period results requires attention to that selection and treatment history.

The primary analysis population is described as treated participants in the full analysis set, rather than all participants randomized. The registry record does not explain in detail how that population was defined beyond the stated treated count. Finally, the safety summaries are counts over specified periods and groups; the record does not provide exposure-adjusted comparisons or enough information to establish event causality. These are limits of interpretation, not evidence that the reported results are incorrect.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Tralokinumab in Combination With Topical Corticosteroids for Moderate to Severe Atopic Dermatitis - ECZTRA 3 — NCT03363854.
• Registry record last updated 2025-03-11; status last verified 2021-02.
• Results first posted 2021-01-14.
• 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

ECZTRA 3 (NCT03363854) was a completed Phase 3 randomized, double-blind, parallel-group trial of tralokinumab plus topical corticosteroids versus placebo plus topical corticosteroids in moderate to severe atopic dermatitis. The registry record lists 380 randomized participants and describes the primary analysis population as 378 treated participants.

The primary endpoint was IGA score 0 or 1 at Week 16. The sponsor-reported Cochran-Mantel-Haenszel risk difference was 12.4, with a 95% confidence interval from 2.9 to 21.9 and p = 0.015. The registry also lists secondary efficacy and safety outcomes and a continuation period with response-based treatment pathways.

Multiplicity handling, detailed missing-data methods, the original sample-size justification, and a full statistical analysis plan are not reported in the registry record. The planning illustrations are separate calculations based on observed effects and stated assumptions.