A Study to Compare Upadacitinib (ABT-494) to Placebo in Adults With Rheumatoid Arthritis (RA) Who Are on a Stable Dose of Conventional Synthetic Disease-Modifying Anti-Rheumatic Drugs (csDMARDs) and Have an Inadequate Response to csDMARDs: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Compare Upadacitinib (ABT-494) to Placebo in Adults With Rheumatoid Arthritis (RA) Who Are on a Stable Dose of Conventional Synthetic Disease-Modifying Anti-Rheumatic Drugs (csDMARDs) and Have an Inadequate Response to csDMARDs (NCT02720523), a randomized phase 2, phase 3 study. The primary outcome is Percentage of Participants With an American College of Rheumatology 20% (ACR20) Response at Week 12.

2. Clinical Question

The central comparison is the Week 12 percentage meeting the American College of Rheumatology 20% response criteria in each upadacitinib dose group versus placebo. ACR20 is a composite response definition: it requires improvement in both tender and swollen joint counts, together with improvement in at least three of five additional disease-activity, pain, function, and inflammation measures. The outcome therefore reflects improvement across multiple aspects of rheumatoid arthritis rather than a change in a single measure.

The trial enrolled participants who continued stable csDMARDs and had not responded adequately to that treatment. Its question is consequently about adding upadacitinib, at the tested doses, in this selected study population during the initial placebo-controlled period. The registry record describes a longer treatment period after the initial 12 weeks, but the primary comparison concerns Week 12.

The response percentages describe the observed proportions in the randomized groups. They do not by themselves establish how well the results apply to people outside this study population, or what happened over longer follow-up. Those questions require evidence beyond the Week 12 endpoint.

3. Trial Design

This was a randomized, parallel-group Phase 2b/3 study sponsored by AbbVie. The registry record lists quadruple masking and an enrollment of 197. It describes Japanese participants with moderately to severely active rheumatoid arthritis who were receiving a stable dose of csDMARDs and had an inadequate response to csDMARDs. The study status is completed; its recorded start date is 2016-03-22 and its primary completion date is 2017-08-03.

During Period 1, which lasted 12 weeks, participants received placebo or upadacitinib at 7.5 mg, 15 mg, or 30 mg once daily. The primary outcome was assessed at Week 12. Period 2 descriptions show that participants initially assigned placebo switched to upadacitinib, while participants assigned active treatment continued upadacitinib. The 30 mg descriptions state that participants were to switch to 15 mg after regulatory approval of the rheumatoid arthritis indication in Japan. The registry descriptions specify treatment for 248 weeks for some groups; for the 30 mg groups, the available description says treatment continued until that approval and then switched to 15 mg, but does not give a complete duration after the switch.

The brief study summary says that following marketing approval of upadacitinib for rheumatoid arthritis in Japan, the study would become a post-marketing clinical study and include a long-term extension period. The registry record does not establish in the material what occurred during that proposed extension.

Specific randomization procedures, allocation ratios, masking roles, and the original protocol’s statistical analysis plan are not reported in the registry record. The original sample-size justification and any planned interim analyses are also not reported in the registry record. These omissions limit how much can be concluded about the design choices behind the trial; they do not change the definition of the reported Week 12 comparison.

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: Participants were randomized to 1 of 3 doses of upadacitinib or placebo in Period 1. Participants who completed week 12 could continue in the study; participants in the upadacitinib groups continued on the same dose of upadacitinib, participants in the placebo group switched to upadacitinib 7.5, 15, or 30 mg per the pre-specified randomization.

Only the first 6 of 7 groups are shown.

Period 1: Placebo-controlled Period

Reasons for not completing this period:

Period 2: Long-term Extension Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set; participants who prematurely discontinued from study drug prior to Week 12 or for whom ACR data were missing at Week 12 were considered non-responders.

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 Full Analysis Set (FAS) included all randomized participants who received at least 1 dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With an American College of Rheumatology 20% (ACR20) Response at Week 12, measured over Baseline and Week 12.

The primary endpoint is the percentage of participants meeting the ACR20 response criteria at Week 12. A participant had to meet all of the following: at least 20% improvement from baseline in the 68-tender joint count; at least 20% improvement in the 66-swollen joint count; and at least 20% improvement in at least three of five additional measures. Those five measures are physician global assessment of disease activity, patient global assessment of disease activity, patient assessment of pain, the Health Assessment Questionnaire–Disability Index (HAQ-DI), and high-sensitivity C-reactive protein.

This composite is used to classify whether a participant reaches a defined level of improvement across joint counts and other disease-related domains. Because success requires meeting multiple criteria, the endpoint is not equivalent to improvement in joint counts alone, nor does it state how much each individual measure changed. It converts the combined criteria into a binary outcome: response or non-response.

The registry specifies a full analysis set for the primary endpoint. Participants who stopped study drug before Week 12 or whose ACR data were missing at Week 12 were counted as non-responders. The percentage therefore incorporates that endpoint-specific rule rather than describing only participants with an observed Week 12 assessment. Other details of missing-data handling are not reported in the registry record.

ACR20 offers a standardized way to compare response proportions between randomized groups at a defined time. It does not, by itself, describe durability of response, changes in every component, or longer-term outcomes.

7. Results Posted in the Registry

• Upadacitinib 7.5 mg: 75.5% (37/49) versus 42.9% (21/49) on control; risk difference 32.7 percentage points (95% CI 14.3 to 51.0), risk ratio 1.76 (1.23–2.53); unadjusted two-proportion z-test p = 0.001.
• Upadacitinib 15 mg: 83.7% (41/49) versus 42.9% (21/49) on control; risk difference 40.8 percentage points (95% CI 23.5 to 58.1), risk ratio 1.95 (1.38–2.76); unadjusted two-proportion z-test p = <0.001.
• Upadacitinib 30 mg: 80.0% (40/50) versus 42.9% (21/49) on control; risk difference 37.1 percentage points (95% CI 19.4 to 54.9), risk ratio 1.87 (1.31–2.65); 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 With an American College of Rheumatology 50% (ACR50) Response 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:
  • Upadacitinib 7.5 mg: 40.8% (20/49) versus 16.3% (8/49) on control; risk difference 24.5 percentage points (95% CI 7.3 to 41.7), risk ratio 2.50 (1.22–5.13); unadjusted two-proportion z-test p = 0.007.
  • Upadacitinib 15 mg: 65.3% (32/49) versus 16.3% (8/49) on control; risk difference 49.0 percentage points (95% CI 32.1 to 65.9), risk ratio 4.00 (2.06–7.79); 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 Upadacitinib 7.5 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 Upadacitinib 15 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants With an American College of Rheumatology 70% (ACR70) Response 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:
  • Upadacitinib 7.5 mg: 20.4% (10/49) versus 2.0% (1/49) on control; risk difference 18.4 percentage points (95% CI 6.4 to 30.3), risk ratio 10.00 (1.33–75.17); unadjusted two-proportion z-test p = 0.004.
  • Upadacitinib 15 mg: 34.7% (17/49) versus 2.0% (1/49) on control; risk difference 32.7 percentage points (95% CI 18.7 to 46.6), risk ratio 17.00 (2.35–122.82); 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 Upadacitinib 7.5 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 Upadacitinib 15 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants Achieving Low Disease Activity (LDA) Based on DAS28(CRP) 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:
  • Upadacitinib 7.5 mg: 53.1% (26/49) versus 18.4% (9/49) on control; risk difference 34.7 percentage points (95% CI 17.0 to 52.4), risk ratio 2.89 (1.51–5.51); unadjusted two-proportion z-test p = <0.001.
  • Upadacitinib 15 mg: 69.4% (34/49) versus 18.4% (9/49) on control; risk difference 51.0 percentage points (95% CI 34.2 to 67.9), risk ratio 3.78 (2.03–7.01); 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 Upadacitinib 7.5 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 Upadacitinib 15 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary results follow the same broad direction as the primary endpoint: response or disease-activity outcomes at Week 12 were more favorable in the upadacitinib groups than in the placebo group for the secondary results with detailed figures reported. This includes the ACR50 and ACR70 response measures and low disease activity based on DAS28(CRP). The different outcomes do not measure the same thing, so their consistency is supportive context rather than a replication of the primary endpoint.

ACR50 and ACR70 apply more stringent response criteria than ACR20, while low disease activity based on DAS28(CRP) uses a disease-activity classification. A pattern across these measures can help show whether the primary finding sits alongside other outcomes that address response and disease activity. It cannot establish that every registered secondary outcome changed in the same way. Numerical results for the other listed outcomes, including HAQ-DI, SF-36 physical component score, clinical remission based on DAS28(CRP), and ACR20 at Week 1, are not reported in the registry record.

The registry lists multiple secondary outcomes and multiple analyses. When many outcomes or comparisons are examined, the chance of finding at least one nominally small p-value can increase. A multiplicity adjustment or hierarchical testing plan can help address that issue, but such a plan is not reported in the registry record. The posted secondary p-values should therefore be interpreted as less definitive than a result supported by a clearly documented testing strategy.

Secondary findings can provide context for the primary ACR20 comparison, but they do not substitute for it, erase uncertainty in the primary estimate, or establish longer-term effects. The endpoint definitions, analysis populations, and missing-data handling may differ across outcomes; complete details for those matters are not reported in the registry record.

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

• Change From Baseline in Disease Activity Score 28 (DAS28) (CRP) at Week 12
• Change From Baseline in Health Assessment Questionnaire - Disability Index (HAQ-DI) at Week 12
• Percentage of Participants With an American College of Rheumatology 50% (ACR50) Response at Week 12
• Percentage of Participants With an American College of Rheumatology 70% (ACR70) Response at Week 12
• Change From Baseline in Short-Form 36 (SF-36) Physical Component Score (PCS) at Week 12
• Percentage of Participants Achieving Low Disease Activity (LDA) Based on DAS28(CRP) at Week 12
• Percentage of Participants Achieving Clinical Remission (CR) Based on DAS28(CRP) at Week 12
• Percentage of Participants With an American College of Rheumatology 20% (ACR20) Response at Week 1

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a binary response: each participant either met the ACR20 criteria at Week 12 or did not. The registry reports the percentage in each randomized group and a 95% confidence interval. It also reports the sponsor’s response-rate differences for comparisons of each active dose with placebo.

What does a response-rate difference mean?

In general, a response-rate difference subtracts the proportion meeting the response definition in one group from the proportion in another. It is expressed in percentage points when the response proportions are expressed as percentages. A positive difference for an active-versus-placebo comparison means the observed response percentage is higher in the active group. This measure describes an absolute difference between groups; it is not a relative ratio.

For this endpoint, the sponsor reported differences of 32.7, 40.8, and 37.1 percentage points for the 7.5 mg, 15 mg, and 30 mg groups, respectively, compared with placebo. The registry identifies the sponsor’s method as Cochran-Mantel-Haenszel. The specific stratification factors or other implementation details are not reported in the registry record, so the method’s full application cannot be reconstructed from the record alone.

Why use the Cochran-Mantel-Haenszel method?

In general, the Cochran-Mantel-Haenszel method combines information across strata to estimate or test an association while accounting for stratification. Its interpretation depends on how the strata were defined and how the analysis was carried out. Since those details are not reported in the registry record, it is appropriate to identify the reported method without assuming which factors were included or how they were handled.

The sponsor also reported a Cochran-Armitage test across the active-dose groups, with p < 0.001. In general, this test evaluates evidence for a trend across ordered groups. A trend test does not, on its own, answer every pairwise dose comparison or establish that the response changes by a particular amount between adjacent doses. No more detailed description of that trend analysis is reported in the registry record.

What does the p-value tell the reader?

In general, a p-value measures how compatible the observed data are with a specified statistical null hypothesis, under the assumptions of the test. A small p-value does not measure the probability that the null hypothesis is true, the size or clinical importance of a difference, or the likelihood that the result will recur. The reported p-values are attached to the sponsor’s stated comparisons and methods. The null hypothesis details, any prespecified decision thresholds, and the broader testing plan are not reported in the registry record.

The reported sponsor p-values for the three primary comparisons are each p < 0.001. The simple unadjusted re-calculations include p = 0.001 for 7.5 mg and p < 0.001 for 15 mg and 30 mg. The numbers can differ because the sponsor’s reported method and a simple unadjusted calculation need not use the same analysis. The sponsor’s reported analyses are the trial’s inferential results.

How should a confidence interval be read?

In general, a confidence interval gives a range of values compatible with the data and the method used, at a stated confidence level. It conveys uncertainty around an estimate; it is not a guarantee that the true value lies in that particular interval. The sponsor’s 95% intervals for the primary response-rate differences are 14.3 to 51.0, 23.5 to 58.1, and 19.4 to 54.9 percentage points. The intervals are all above zero, but their breadth means the exact size of the difference remains uncertain.

In general, an interval’s width depends on the amount of information in the data and the method used. With binary outcomes, the observed response proportions and group sizes contribute to that uncertainty. It is not appropriate to read a confidence interval as a range of individual participant outcomes, or as evidence that all people in a group experienced a similar response.

What is the role of missing-data rules?

The primary endpoint specifies that participants who discontinued study drug before Week 12 or had missing ACR data at Week 12 were classified as non-responders. In general, a missing-as-nonresponse rule assigns an outcome category to participants without the required assessment rather than omitting them from the response count. This can affect the estimated response percentage, especially if missingness or discontinuation differs between groups. The registry record does not report the number of missing assessments, reasons for discontinuation, or alternative sensitivity analyses, so their influence cannot be assessed from the record.

The full analysis set is the stated primary population. The registry record does not give a detailed operational definition of that population beyond the endpoint description. A separate per-protocol analysis, sensitivity analysis, or model-based missing-data analysis is not reported in the registry record.

These explanations of statistical concepts are general. The interpretation of the trial-specific estimates rests on the stated endpoint, analysis method, reported group results, and the limits of detail in the registry record.

11. Confidence Intervals and Precision

The 95% confidence intervals provide a useful indication of the precision of the primary response-rate differences. For 7.5 mg versus placebo, the sponsor reported a difference of 32.7 percentage points with a 95% interval from 14.3 to 51.0. For 15 mg versus placebo, the estimate was 40.8 with an interval from 23.5 to 58.1. For 30 mg versus placebo, the estimate was 37.1 with an interval from 19.4 to 54.9. All are reported as percentage-point differences.

Each interval excludes zero, which is consistent with the sponsor’s reported superiority findings for these comparisons. But the intervals remain broad enough to allow materially different magnitudes of difference. The point estimates should therefore not be mistaken for exact values. The confidence intervals quantify statistical uncertainty under the analysis method; they do not capture every possible source of uncertainty, such as limits on generalizing beyond the studied population or uncertainty about unreported analysis details.

The primary groups included 49 participants in each of the placebo, 7.5 mg, and 15 mg groups, and 50 in the 30 mg group. These group sizes help explain why the estimates have noticeable uncertainty. The reported confidence intervals are also not individual ranges: they do not indicate where most participants’ responses lie, and they do not describe variation in the amount of improvement among responders.

In general, a 95% confidence interval is generated by a method that, over repeated use under its assumptions, would cover the parameter at the stated rate. It does not mean there is a 95% probability that the fixed parameter lies in the specific interval observed. Interpretation also depends on the method. The sponsor reports Cochran-Mantel-Haenszel estimates; the unadjusted re-calculations are based on posted summary statistics and are not the same 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 planning illustrations express sample size as participants per arm for the binary comparisons. They are not total event counts. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. These are planning assumptions for the illustrations; the trial’s original design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the estimated participants per arm for a design intended to have an 80% chance of detecting the assumed effect under those planning assumptions. The 90% power column gives the corresponding per-arm number for a design intended to have a 90% chance. These are design-planning concepts, not probabilities that the observed result is correct. The illustrations are not observed power.

For the primary comparison of 15 mg with placebo, planning for 100% of the observed effect gives 21 participants per arm for 80% power and 28 participants per arm for 90% power. Planning for 75% of the observed effect gives 40 and 53 participants per arm, respectively; planning for 50% gives 93 and 124. For the 7.5 mg comparison, planning for 75% of the observed effect gives 64 participants per arm for 80% power and 85 for 90%. For 30 mg, the corresponding 75% illustration gives 49 and 65 per arm.

The pattern illustrates a general planning principle: a smaller effect is harder to distinguish from random variation, so more participants are needed to achieve the same target power. The estimates depend on the observed effect used as their starting point, which may not be the effect a future study should expect. They should be read as conditional planning examples, not as evidence that the trial was designed around these assumptions or that it had a particular probability of success after completion.

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 = 0.429; treatment proportion p₁ = 0.755; average p̄ = 0.592.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6951 = 1.3623.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6556 = 0.5518.
4. n = (1.3623 + 0.5518)² / (0.3265)² = 34.4, rounded up to 35 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 <- 0.428571428571429                          # control proportion
p1 <- 0.755102040816326                          # 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(2720523)
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 = 35 (80% power), 46 (90% power)
 75% of observed effect: n per arm = 64 (80% power), 85 (90% power)
 50% of observed effect: n per arm = 146 (80% power), 195 (90% power)
power.prop.test: 34.4 
Simulated power at n = 35 per arm: 0.809 (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 reported primary comparisons using the Cochran-Mantel-Haenszel method. The reported response-rate differences were 32.7 percentage points for 7.5 mg versus placebo, 40.8 for 15 mg versus placebo, and 37.1 for 30 mg versus placebo. The corresponding 95% confidence intervals were 14.3 to 51.0, 23.5 to 58.1, and 19.4 to 54.9, with p < 0.001 for each comparison.

Simple unadjusted re-calculations from the posted group summaries give differences that round to the same displayed estimates and confidence limits. The calculations do not reproduce the sponsor’s analysis method, however. For example, the unadjusted p-value for the 7.5 mg comparison is 0.001, while the sponsor reports p < 0.001. This difference in p-value reporting is not a flagged discrepancy in the registry information. The sponsor’s reported analysis should be treated as the trial’s own inferential result; the re-calculation is a separate descriptive check based on summary data.

The record does not report enough detail to reconstruct all aspects of the sponsor’s analysis, including any stratification factors. It is therefore useful to distinguish the reported method from the supplementary unadjusted calculation rather than treating them as interchangeable. The estimates are close after rounding, but the source and analytical basis of each result remain different.

15. Multiplicity, Missing Data and Analysis Populations

The registry reports a primary Week 12 endpoint and multiple secondary outcomes, with a total of 12 outcome measures and 37 analyses listed. It reports p-values for the primary comparisons and for some secondary comparisons. The registry record does not report a multiplicity adjustment, a hierarchy of hypotheses, or a procedure specifying which tests were confirmatory. Without that information, the nominal p-values should not be read as if the full family of tests had necessarily been controlled for false-positive findings.

This is particularly relevant to secondary results. In general, testing many outcomes or contrasts increases the opportunity for chance findings. A planned adjustment or testing sequence can limit that risk, but the registry record does not report whether such a plan was used. The primary results are the central findings described by the registry, while secondary p-values offer supporting context and should be interpreted more cautiously when the testing plan is unavailable.

The primary endpoint population is described as the full analysis set. The registry also specifies that participants who discontinued study drug before Week 12 or had missing ACR data at Week 12 were considered non-responders. This is a clear rule for the primary binary endpoint. The record does not report counts or reasons for missing assessments, the amount of treatment discontinuation before Week 12, or sensitivity analyses that test other assumptions about missing outcomes.

Rules for missing data in the secondary outcomes are not reported in the registry record. Nor is a detailed definition of the full analysis set reported beyond its mention in the primary endpoint description. The protocol’s statistical analysis plan, interim analyses, and original sample-size justification are not reported in the registry record. These gaps make it difficult to assess whether each analysis followed a prespecified plan, how robust the findings are to alternative assumptions, and how the family of statistical tests was handled.

In general, missing-data assumptions and multiplicity procedures affect how confidently an estimate or p-value can be interpreted. Their absence from the registry record is a limit on interpretation, not evidence that a particular procedure was or was not used.

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: pneumocystis jirovecii pneumonia (6.1%), herpes zoster (4.5%), pneumonia (3.0%), interstitial lung disease (3.0%), cellulitis (2.0%).

The registry reports serious adverse event counts for Period 1 and for Period 1 plus Period 2. In Period 1, the reported number affected was 0 of 49 with placebo, 1 of 49 with upadacitinib 7.5 mg, 1 of 49 with 15 mg, and 5 of 50 with 30 mg. In the Period 1 plus Period 2 groups, the reported counts were 9 of 65 for 7.5 mg, 14 of 64 for 15 mg, and 18 of 66 for 30 mg. The registry reports 0 deaths in each Period 1 group and 0, 0, and 2 deaths in the Period 1 plus Period 2 groups, respectively.

The time periods differ: Period 1 is the initial 12-week treatment period, while the combined-period figures include longer follow-up. These counts should not be compared as if participants had identical observation time. After Period 1, placebo participants switched to upadacitinib, so the longer-period groups do not retain the initial placebo comparison.

The registry also lists serious adverse event terms with percentages, including Pneumocystis jirovecii pneumonia at 6.1%, herpes zoster at 4.5%, pneumonia and interstitial lung disease at 3.0% each, and cellulitis and transient ischaemic attack at 2.0% each. The record does not report the denominator or further context for these term percentages in the information. Counts alone do not establish comparative risk, cause, or an individual’s likelihood of an event.

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

The Week 12 results show higher observed ACR20 response percentages in each upadacitinib group than in the placebo group among Japanese adults with moderately to severely active rheumatoid arthritis who had an inadequate response to stable csDMARDs. The sponsor-reported response-rate differences were positive for all three dose comparisons, and each reported 95% confidence interval was above zero. Those estimates describe the trial’s randomized groups and the endpoint’s specified missing-as-nonresponse rule.

That finding does not mean every participant responded, or that the percentage difference precisely predicts what would occur in another group of patients. ACR20 is a composite threshold, not a measure of the amount of improvement for each person. The reported endpoint also does not establish the result for every symptom, function measure, or laboratory measure considered separately.

The Week 12 comparison does not answer how durable a response might be, what the comparative results would be over longer follow-up, or how outcomes would compare after the placebo group switched treatment. The registry describes a longer treatment period, but the results summarized for the primary endpoint are at Week 12. Longer-term comparative results are not reported in the registry record.

These findings do not establish effects in populations unlike the enrolled Japanese participants, including people with different disease severity or treatment histories. The registry record also does not provide all protocol and analysis-plan details needed to evaluate every design decision, missing-data sensitivity, or multiplicity procedure. The confidence intervals express statistical uncertainty under the reported method, not all uncertainty about generalizability or study conduct.

In general, statistical evidence about group-level outcomes is not a substitute for clinical judgment about an individual. These trial findings describe measured group outcomes; they do not offer treatment advice or guarantee a particular outcome for any person.

18. Why This Trial Matters Statistically

This trial is statistically informative because it places several upadacitinib doses alongside a randomized placebo group during a defined initial treatment period. The Week 12 ACR20 outcome is a binary composite, and the reported response percentages and response-rate differences make the group contrast interpretable in absolute terms. Confidence intervals help show both the direction and uncertainty of those estimates rather than leaving interpretation to point estimates alone.

The endpoint’s non-responder rule is also important. Participants who discontinued study drug before Week 12 or lacked Week 12 ACR data were classified as non-responders. That makes the primary outcome definition explicit on a key missing-data issue, while the absence of missingness counts and sensitivity analyses leaves uncertainty about how much that rule influenced the results.

The trial includes more than one active dose and reports a trend test across active-dose groups, which is a distinct statistical question from the three comparisons against placebo. Keeping those questions separate prevents a trend result from being overread as proof of pairwise differences among doses. Similarly, the secondary outcomes offer context but require care because the registry does not describe a multiplicity procedure.

Finally, the initial placebo-controlled period must be distinguished from the later treatment period. Placebo participants switched to active treatment after the first 12 weeks, and the longer-term safety groups cover different periods of exposure. That design feature matters when interpreting the scope of comparative evidence. The record supports a focused analysis of Week 12 group outcomes, while several broader questions about longer follow-up, missingness, and analysis planning remain unanswered.

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.

The enrolled population is specific: Japanese adults with moderately to severely active rheumatoid arthritis, inadequate response to csDMARDs, and a stable csDMARD dose. The findings may not directly generalize to people outside those characteristics. The primary results are measured at Week 12, so they do not describe persistence of response or longer-term comparative efficacy.

The placebo comparison applies to Period 1. Participants initially assigned placebo switched to upadacitinib in Period 2, limiting the longer-term direct comparison with an untreated placebo group. The registry describes different treatment durations across groups, including a 248-week period for some groups and a dose switch after regulatory approval for the 30 mg groups. The completed long-term results are not reported in the registry record.

There were 49 participants in each of three primary groups and 50 in the remaining group. The confidence intervals reflect uncertainty around the group estimates, and relatively limited group sizes make precision important when reading the point estimates. The original sample-size justification and design assumptions are not reported in the registry record; the planning illustrations based on observed effects do not fill that gap.

For the primary endpoint, the rule for discontinuation and missing Week 12 ACR data is stated, but the number and reasons for missing assessments and the results of sensitivity analyses are not reported. Missing-data rules for secondary outcomes are not reported in the registry record. Nor are the detailed definition of the full analysis set, multiplicity strategy, interim analyses, allocation ratios, or full statistical analysis plan.

Only some secondary outcomes have detailed numerical results in the registry information. The registry lists other measures, but their numerical results are not reported in the registry record. These limitations constrain conclusions about robustness, multiplicity, longer-term outcomes, and how broadly the Week 12 estimates apply.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study to Compare Upadacitinib (ABT-494) to Placebo in Adults With Rheumatoid Arthritis (RA) Who Are on a Stable Dose of Conventional Synthetic Disease-Modifying Anti-Rheumatic Drugs (csDMARDs) and Have an Inadequate Response to csDMARDs — NCT02720523.
• Registry record last updated 2023-06-07; status last verified 2023-06.
• Results first posted 2019-10-21.
• 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

NCT02720523 is a completed, randomized, quadruple-masked, parallel-group Phase 2b/3 study sponsored by AbbVie. It enrolled 197 Japanese adults with moderately to severely active rheumatoid arthritis and an inadequate response to stable csDMARDs. During the initial 12-week period, participants received placebo or upadacitinib 7.5 mg, 15 mg, or 30 mg once daily.

The primary endpoint was the Week 12 ACR20 response percentage. The sponsor reported higher response percentages in each upadacitinib group than with placebo, with Cochran-Mantel-Haenszel response-rate differences of 32.7, 40.8, and 37.1 percentage points and p < 0.001 for each comparison. The registry also reports selected secondary outcomes and safety counts. Placebo participants switched to upadacitinib after Period 1, so the primary comparison is the initial 12-week placebo-controlled result.

Details of the original sample-size justification, multiplicity plan, interim analyses, and several missing-data and analysis-plan details are not reported in the registry record. The longer-term comparative results are also not reported in the registry record.