Filgotinib in Combination With Methotrexate in Adults With Moderately to Severely Active Rheumatoid Arthritis Who Have an Inadequate Response to Methotrexate: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Filgotinib in Combination With Methotrexate in Adults With Moderately to Severely Active Rheumatoid Arthritis Who Have an Inadequate Response to Methotrexate (NCT02889796), a randomized phase 3 study. The primary outcome is Percentage of Participants Who Achieved an American College of Rheumatology (ACR) 20% Improvement (ACR20) Response at Week 12.

2. Clinical Question

3. Trial Design

This was a randomized, double-blind, parallel-group, multicenter Phase 3 study. The registry record lists enrollment of 1759 participants and a completed study. The study assessed filgotinib administered for 52 weeks in combination with methotrexate in adults whose rheumatoid arthritis remained moderately to severely active despite methotrexate.

Participants were assigned to filgotinib 200 mg, filgotinib 100 mg, adalimumab, or placebo, with a stable dose of methotrexate. Matching placebos were used for the study drugs. The primary comparison was filgotinib versus placebo at Week 12; adalimumab provided an active-control comparison. The registered primary objective focused on the percentage achieving ACR20 at Week 12.

The study also included placebo treatment sequences lasting up to 24 weeks. In the sequences described as placebo to filgotinib, participants were rerandomized after 24 weeks to receive filgotinib 200 mg or 100 mg. A separate placebo group was described as never receiving filgotinib. These treatment sequences matter when interpreting later follow-up and safety counts: the placebo exposure did not have the same course in every group.

The primary analysis population included randomized participants who received at least 1 dose of study drug. Enrollment is the overall registered count, whereas the primary outcome results are reported by treatment group in that analysis population. The registry record does not report the original sample-size justification or enough detail to reconstruct the complete randomization and allocation procedures. The registered masking is double; further operational details about how masking was maintained 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: 2582 participants were screened.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The Full Analysis Set included participants who were randomized and received at least 1 dose of study drug.

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: The Safety Analysis Set included all participants who received at least 1 dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants Who Achieved an American College of Rheumatology (ACR) 20% Improvement (ACR20) Response at Week 12, measured over Week 12.

The primary endpoint was the percentage of participants achieving an American College of Rheumatology 20% improvement response, or ACR20, at Week 12. ACR20 is a composite response: participants needed at least 20% improvement from baseline in both tender joint count, measured across 68 joints, and swollen joint count, measured across 66 joints. They also needed improvement in at least 3 of 5 additional measures.

Those additional measures cover physician and participant assessments of disease activity, participant pain assessment, physical function measured with the Health Assessment Questionnaire-Disability Index, and high-sensitivity C-reactive protein. The assessments combine symptoms, joint findings, ability to function, and an inflammation measure rather than relying on a single symptom or laboratory value. A response threshold makes it possible to report the proportion of participants meeting a defined improvement criterion.

ACR20 is used as a measure of improvement in rheumatoid arthritis signs and symptoms. It is a categorical endpoint: each participant is classified as meeting or not meeting the response definition at the specified visit. It does not describe the size of improvement beyond the threshold, and it does not by itself describe every aspect of disease status or longer-term outcome. For the primary analysis, participants with missing outcomes were treated as non-responders, so the reported response proportions incorporate that stated missing-outcome rule.

7. Results Posted in the Registry

• Filgotinib 200 mg: 76.6% (364/475) versus 49.9% (237/475) on control; risk difference 26.7 percentage points (95% CI 20.8 to 32.6), risk ratio 1.54 (1.39–1.70); unadjusted two-proportion z-test p = <0.001.
• Filgotinib 100 mg: 69.8% (335/480) versus 49.9% (237/475) on control; risk difference 19.9 percentage points (95% CI 13.8 to 26.0), risk ratio 1.40 (1.26–1.56); unadjusted two-proportion z-test p = <0.001.
• Adalimumab: 70.5% (229/325) versus 49.9% (237/475) on control; risk difference 20.6 percentage points (95% CI 13.9 to 27.3), risk ratio 1.41 (1.26–1.58); unadjusted two-proportion 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.

Percentage of Participants Who Achieved Disease Activity Score for 28 Joint Count Using C-Reactive Protein [DAS28 (CRP)] < 2.6 at Week 12

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Filgotinib 200 mg: 34.1% (162/475) versus 9.3% (44/475) on control; risk difference 24.8 percentage points (95% CI 19.8 to 29.8), risk ratio 3.68 (2.71–5.01); unadjusted two-proportion z-test p = <0.001.
  • Filgotinib 100 mg: 23.8% (114/480) versus 9.3% (44/475) on control; risk difference 14.5 percentage points (95% CI 9.9 to 19.1), risk ratio 2.56 (1.85–3.54); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Filgotinib 200 mg 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 proportion with the event is statistically higher for Filgotinib 100 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants Who Achieved DAS28 (CRP) ≤ 3.2 at Week 12

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Filgotinib 200 mg: 49.7% (236/475) versus 23.4% (111/475) on control; risk difference 26.3 percentage points (95% CI 20.4 to 32.2), risk ratio 2.13 (1.76–2.56); unadjusted two-proportion z-test p = <0.001.
  • Filgotinib 100 mg: 38.8% (186/480) versus 23.4% (111/475) on control; risk difference 15.4 percentage points (95% CI 9.6 to 21.2), risk ratio 1.66 (1.36–2.02); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Filgotinib 200 mg 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 proportion with the event is statistically higher for Filgotinib 100 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes broaden the assessment beyond the primary ACR20 response. They include disease activity thresholds based on DAS28 using C-reactive protein, change in physical function, structural change, health status, fatigue, and ACR50 response at several visits. The registry lists 79 secondary outcomes. The posted numerical results for the DAS28 thresholds at Week 12 follow a pattern in which the filgotinib groups have higher response proportions than placebo. This is directionally consistent with the primary ACR20 comparison, while reflecting different definitions of response and disease activity.

Agreement in direction across related outcomes can add context, but these measures are not interchangeable. ACR20 is a composite improvement threshold; DAS28 thresholds classify disease activity using a joint count and C-reactive protein measure. Change-from-baseline measures, structural outcomes, and participant-reported scales address other aspects of disease and have different interpretations. Their visits also differ, so findings at one time point should not be treated as a single combined result.

Secondary outcomes carry less confirmatory weight when many outcomes and analyses are considered and the multiplicity strategy is not reported. With many statistical tests, the chance of obtaining apparently notable findings by chance can increase. The registry record does not report how multiplicity was controlled, which secondary analyses were designated confirmatory, or whether the reported analyses followed a prespecified hierarchy. Therefore, secondary findings are best read as supporting context or exploratory evidence unless their inferential status is established by the full analysis plan. They should not replace the defined primary comparison or be treated as independent confirmations merely because several measures move in a similar direction.

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

• Change From Baseline in the Health Assessment Questionnaire-Disability Index (HAQ-DI) Score at Week 12
• Percentage of Participants Who Achieved Disease Activity Score for 28 Joint Count Using C-Reactive Protein [DAS28 (CRP)] < 2.6 at Week 12
• Change From Baseline in Modified Total Sharp Score (mTSS) at Week 24
• Percentage of Participants Who Achieved DAS28 (CRP) ≤ 3.2 at Week 12
• Change From Baseline in 36-Item Short Form Survey (SF-36) Physical Component Summary (PCS) Score at Week 12
• Change From Baseline in Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue Score at Week 12
• Percentage of Participants Who Achieved ACR 50% Improvement (ACR50) at Weeks 2, 4, 12, and 24
• Percentage of Participants Who Achieved ACR50 at Weeks 36, and 52

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is binary at the participant level: a participant either met the ACR20 response definition at Week 12 or did not. A response percentage is the proportion meeting that definition in an arm. The registry reports the arm percentages with 95% confidence intervals and reports between-group tests using logistic regression. The sponsor's analyses identify the comparison as a superiority analysis.

What does the risk difference describe?

A risk difference subtracts one response proportion from another, keeping the result on the proportion or percentage-point scale. It directly describes the difference in the observed response proportions for the two groups being compared. It is not the same as a risk ratio, which compares proportions as a ratio, or an odds ratio, which compares odds. Different measures can describe the same binary outcome while answering different questions. The recalculated values use the unadjusted arm counts and proportions; they are not the sponsor's logistic-regression estimates.

Why use logistic regression?

Logistic regression is a method for modeling a binary outcome such as response versus non-response. A regression analysis can estimate a treatment comparison within a specified model. The registry identifies the sponsor's method as logistic regression and reports a difference in response rates, confidence intervals, p-values, and a superiority designation. The registry record does not report enough detail to identify all model terms, any covariate adjustment, or other implementation choices. Those details should not be inferred from the method label alone.

What does a p-value say?

A p-value describes how compatible the observed data are with a stated null hypothesis under the assumptions of the analysis. A small p-value, such as the reported p < 0.001 for both filgotinib comparisons, is evidence against the null comparison used in that analysis. It is not the probability that the null hypothesis is true, nor does it quantify the probability that an individual participant benefits. A p-value also does not establish whether a difference is important in clinical terms. The estimated difference and its confidence interval are needed to understand magnitude and precision.

How should the confidence interval be read?

A confidence interval describes statistical uncertainty around an estimate under the analysis method. A 95% confidence interval is constructed by a procedure that, under its assumptions, would cover the target value in 95% of repeated applications. It is not a range containing 95% of individual participant outcomes. For a difference, the interval helps show which values are reasonably compatible with the data and method. For an arm percentage, it conveys uncertainty about that response proportion. The registry reports intervals for both the arm response percentages and the sponsor's between-group estimates.

What is special about the missing-outcome rule?

The primary endpoint description states that participants with missing outcomes were set as non-responders. This rule classifies missing endpoint observations as failures to meet the response definition rather than excluding them from the response numerator or imputing a response based on other information. Such a rule affects the estimand represented by the reported proportion: the result is the response rate under that specified handling of missing primary outcomes. The registry record does not report a broader account of missingness patterns, reasons for missing outcomes, or sensitivity analyses that test alternative assumptions.

What do the secondary analyses add?

Secondary outcomes include other response thresholds, disease activity measures, physical function, fatigue, health status, and structural change. Each outcome has its own scale, visit, analysis population, and uncertainty. Binary outcomes can be described as response proportions; change-from-baseline outcomes require interpretation on their measurement scales. The registry record lists 80 outcome measures and 174 analyses. This breadth offers additional perspectives, but results across many outcomes require careful attention to multiplicity and to whether findings were prespecified. The registry record does not report the multiplicity strategy or a complete statistical analysis plan.

Planning illustrations use the observed difference as a starting point and show hypothetical per-arm sample sizes under stated assumptions. They are not the trial's original design calculations. The original sample-size justification, its assumptions, and whether any interim analyses were planned are not reported in the registry record. The sample-size illustrations therefore help explain how detectable effects relate to sample size, not how the trial was actually planned or how much power it had after completion.

11. Confidence Intervals and Precision

The arm-level confidence intervals show uncertainty around the observed Week 12 response proportions. For filgotinib 200 mg, the reported response is 76.6%, with a 95% confidence interval from 72.7 to 80.5. For filgotinib 100 mg, it is 69.8%, with an interval from 65.6 to 74.0. The placebo response is 49.9%, with an interval from 45.3 to 54.5. The active-control response is 70.5%, with an interval from 65.3 to 75.6. Each interval concerns an arm proportion; it is not an interval for the difference between arms.

For the sponsor-reported comparisons, the 95% confidence interval for the filgotinib 200 mg response-rate difference is 20.6 to 32.8 percentage points. For filgotinib 100 mg, it is 13.6 to 26.2 percentage points. The intervals are wholly on the positive side of no difference, in keeping with the reported p-values of < 0.001. Their widths also show that the estimates are not exact: a range of effect sizes remains compatible with the data and analysis.

Confidence intervals communicate sampling uncertainty, not every source of uncertainty. They do not account for unreported analysis choices, possible departures from model assumptions, or uncertainty about whether the results generalize to other populations. The registry reports group sizes and intervals, but readers should not treat the boundaries as hard limits on the true value. Nor should arm-level interval overlap alone be used as a substitute for the reported between-group analysis. The comparison-specific interval is the direct summary of uncertainty for the estimated difference.

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 calculation. For these binary outcomes, the figures 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. Each planning scenario uses the stated fraction of the observed effect as the effect to detect.

The 80% power column gives the number of participants per arm that the calculation estimates would be needed to have an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column expresses the same planning question with a higher probability of detecting that effect. For example, for planning at 100% of the observed filgotinib 200 mg effect, the illustration gives 50 participants per arm for 80% power and 67 participants per arm for 90% power. These are hypothetical planning figures, not observed trial results.

Rows planning for 75% or 50% of the observed effect use smaller effects than the full observed effect. Detecting a smaller difference generally requires more participants, which is reflected in the larger per-arm figures for those scenarios. The illustrations are sensitive to their assumptions, including the assumed response rates and the effect size selected. They should not be read as evidence that the completed study had a particular achieved power: this is not observed power.

The trial's own original sample-size justification and planning assumptions are not reported in the registry record. The calculation therefore cannot establish the rationale for the actual enrollment or replace the trial's original design documentation.

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 = 0.499; treatment proportion p₁ = 0.766; average p̄ = 0.633.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6818 = 1.3363.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6550 = 0.5513.
4. n = (1.3363 + 0.5513)² / (0.2674)² = 49.8, rounded up to 50 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 <- 0.498947368421053                          # control proportion
p1 <- 0.766315789473684                          # 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(2889796)
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 = 50 (80% power), 67 (90% power)
 75% of observed effect: n per arm = 93 (80% power), 124 (90% power)
 50% of observed effect: n per arm = 215 (80% power), 287 (90% power)
power.prop.test: 49.8 
Simulated power at n = 50 per arm: 0.806 (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 posted primary analyses use logistic regression and report differences in response rates. For filgotinib 200 mg versus placebo, the reported difference is 26.7 percentage points, with a 95% confidence interval from 20.6 to 32.8 and p < 0.001. For filgotinib 100 mg versus placebo, it is 19.9 percentage points, with a 95% confidence interval from 13.6 to 26.2 and p < 0.001. Both are designated superiority analyses in the registry record.

The unadjusted recalculations use the posted counts and arm proportions to estimate risk differences directly. They give 0.267 for the 200 mg comparison and 0.199 for the 100 mg comparison, with recalculated 95% confidence intervals of 0.208 to 0.326 and 0.138 to 0.260. The recalculated p-values are each < 0.001. The recalculations are broadly consistent with the sponsor's reported comparisons, but they are not the same analysis: they do not reproduce the sponsor's logistic-regression method or any model details not reported in the registry.

For interpretation of the primary analysis, the sponsor's reported results remain the reference analysis. The recalculations are useful as a transparent check based on summary statistics, but should not be presented as a corrected or competing sponsor result.

15. Multiplicity, Missing Data and Analysis Populations

The registry record reports a primary endpoint and numerous secondary outcomes and analyses. It does not report a multiplicity adjustment, testing hierarchy, gatekeeping strategy, or other procedure for controlling the overall false-positive risk across those analyses. When many hypotheses are tested, interpreting each p-value as though it were the only test can overstate the strength of the combined evidence. This concern is especially relevant to secondary outcomes and repeated visits. The registry record also does not identify which secondary analyses, if any, were intended to support formal confirmatory claims.

For the primary ACR20 endpoint, the registry explicitly states that participants with missing outcomes were classified as non-responders. This gives a clear rule for missing primary response data, although it does not show how many outcomes were missing, why they were missing, or whether the missingness differed by group. Other missing-data methods and sensitivity analyses are not reported in the registry record. The primary rule should therefore be understood as part of the definition of the reported primary result, not as evidence that missing-data concerns are absent.

The primary analysis population is described as the Full Analysis Set, including participants who were randomized and received at least 1 dose of study drug. The registry record does not provide enough information to determine how every protocol deviation or post-randomization event was handled in each analysis. The original statistical analysis plan, the detailed analysis populations for all outcomes, and the approach to any interim analyses are not reported in the registry record. Those omissions limit reconstruction of the full inferential plan and make it important to distinguish the stated primary analysis from less fully documented secondary analyses.

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: pneumonia (0.9%), chest pain (0.5%), road traffic accident (0.5%), arthralgia (0.5%), ischaemic stroke (0.5%).

The registry reports serious adverse events by group. Among the groups treated with filgotinib throughout the primary treatment assignment, serious adverse events affected 35 of 475 participants in the 200 mg group and 40 of 480 in the 100 mg group. The corresponding counts were 22 of 325 for adalimumab and 21 of 475 for placebo. The placebo-to-filgotinib 200 mg group had 7 of 190 participants affected, and the placebo-to-filgotinib 100 mg group had 8 of 191. These counts should be read with the group sizes and treatment sequences in mind; the placebo-to-filgotinib groups had a different exposure course from groups assigned to active treatment throughout.

The registry also lists deaths by group: 3 in the filgotinib 200 mg group, 1 in the 100 mg group, 1 in the adalimumab group, 1 in each of the two placebo-to-filgotinib groups, and 2 in the placebo group. The listed serious adverse event terms include pneumonia at 0.9%, and chest pain, road traffic accident, arthralgia, ischaemic stroke, and pulmonary embolism at 0.5% each. The registry record does not provide enough detail in these summary figures to establish causality or compare the full safety profiles. Counts alone are not a complete safety assessment.

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

The primary result describes the proportion of participants meeting a specified improvement threshold at Week 12 under the stated missing-outcome rule. The posted response proportions and sponsor-reported comparisons show that the observed ACR20 response was higher in each filgotinib group than in the placebo group in this trial. The confidence intervals express uncertainty around those estimates, and the reported p-values address the statistical comparison under the sponsor's analysis.

The result does not mean that every participant in a group had the same experience. ACR20 is a threshold-based composite, so it does not show how far each participant improved, which component drove an individual response, or whether a response persisted. Nor does a Week 12 result by itself describe the entire 52-week treatment period. Other outcomes, visits, and safety information require their own interpretation.

The results also do not settle comparisons beyond those reported. The registry identifies adalimumab as an active-control arm, but the primary objective is the filgotinib-versus-placebo comparison; a difference between filgotinib and placebo should not be interpreted as a direct comparison with adalimumab. The posted summaries do not establish how outcomes would vary across patient subgroups or settings not described in the record.

Statistical evidence is not a complete judgment of clinical importance. The size and precision of an estimated difference are relevant, but assessment of importance also depends on the outcome, patient priorities, duration, harms, and the broader evidence. These results are a report of trial outcomes, not individualized medical advice or a recommendation about treatment.

18. Why This Trial Matters Statistically

This trial provides a useful example of how a binary clinical response can be analyzed and communicated. ACR20 reduces several dimensions of rheumatoid arthritis signs and symptoms to a defined responder status, allowing group response proportions and between-group differences to be reported. The primary endpoint also demonstrates why the outcome definition and the missing-data rule matter: both determine what the reported response percentage represents.

The registry offers both arm-level response proportions and sponsor-reported regression comparisons. Reading them together distinguishes descriptive results in each group from an inferential estimate of the treatment contrast. The recalculated unadjusted risk differences offer a second, transparent scale for understanding the summary data, while the sponsor analysis remains the reported regression analysis. Confidence intervals add information about precision that a p-value alone cannot provide.

The study also illustrates the challenges of interpreting a broad outcome program. Measures of disease activity, function, fatigue, health status, and structural change answer different questions, and numerous secondary analyses make multiplicity relevant. Finally, the changing placebo treatment sequences show why treatment exposure and follow-up need to be considered when reading later outcomes and safety counts. Together, these features make the registry record a practical case for separating the primary result, secondary evidence, planning illustrations, and safety summaries rather than treating them as a single conclusion.

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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

Several trial-specific limitations of interpretation follow from what is and is not described in the registry record. The primary endpoint captures a threshold response at Week 12, while the study treatment period was described as 52 weeks. ACR20 does not represent every dimension of disease activity, and the primary result alone does not establish the pattern of response over later visits or the size of improvement beyond the threshold.

The primary analysis counts missing outcomes as non-responses, but the number and reasons for missing primary outcomes are not reported in the registry record. No sensitivity analysis for alternative missing-data assumptions is described. The record also does not give the original sample-size justification, detailed analysis model terms, full statistical analysis plan, multiplicity strategy, or interim-analysis plan. These omissions restrict the ability to assess design assumptions and reconstruct all analysis decisions.

Although the study included an active-control arm, the stated primary objective concerns filgotinib versus placebo. The primary results therefore should not be treated as a formal comparative conclusion between filgotinib and adalimumab. The registry lists different placebo treatment sequences, including groups rerandomized to filgotinib after 24 weeks and a group described as never receiving filgotinib. This complicates simple interpretation of later placebo-related outcomes as a single exposure group.

Finally, the registry summaries do not provide participant-level data, detailed subgroup results, or enough information to examine whether the findings vary across the enrolled population. Safety counts and listed serious-event terms are summary descriptions, not a complete account of event timing, relatedness, or comparative risk. These constraints call for keeping conclusions tied to the outcomes, groups, visits, and methods actually reported.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Filgotinib in Combination With Methotrexate in Adults With Moderately to Severely Active Rheumatoid Arthritis Who Have an Inadequate Response to Methotrexate — NCT02889796.
• Registry record last updated 2021-06-09; status last verified 2021-05.
• Results first posted 2021-01-19.
• 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

NCT02889796 is a completed randomized, double-blind, parallel-group, multicenter Phase 3 study of filgotinib with methotrexate in adults with moderately to severely active rheumatoid arthritis and an inadequate response to methotrexate. The registry lists enrollment of 1759. The primary endpoint was ACR20 response at Week 12, with missing outcomes classified as non-responses in the Full Analysis Set.

The posted primary results show ACR20 response proportions of 76.6% for filgotinib 200 mg, 69.8% for filgotinib 100 mg, and 49.9% for placebo. Sponsor-reported logistic-regression comparisons versus placebo were statistically significant for both filgotinib groups. Interpretation should distinguish those primary results from secondary analyses, hypothetical sample-size planning calculations, and the separate safety summaries.