A Study in Patients With Rheumatoid Arthritis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study in Patients With Rheumatoid Arthritis (NCT00966875), a randomized phase 2 study. The primary outcome is Dose-Response Relationship Measured by the Percentage of Participants With American College of Rheumatology (ACR) 20 Response in bDMARD-Naive Population.

2. Clinical Question

The study was registered as a Phase 2 dose-ranging trial. Its stated purposes included examining safety, whether LY2439821 could help participants with active rheumatoid arthritis, and how much LY2439821 should be given. The primary endpoint focused on the proportion meeting a defined response threshold rather than on an average change in a single measurement. The registry record also lists outcomes in the TNFα-inadequate responder population and continuous disease activity measures, so the study addressed more than one clinical subgroup and more than one way of characterizing response.

The primary result is a dose-group comparison at a specified time point, not an estimate of what would happen to every participant or over every duration of treatment. The study’s results describe the enrolled trial populations and the recorded endpoint definitions. They do not, by themselves, establish how the findings apply to people outside those populations or to treatment decisions.

3. Trial Design

This was a randomized, quadruple-masked, parallel-group Phase 2 study. The registry records enrollment of 448 participants and a completed status. The study began in 2009-08, with primary completion in 2011-03. Participants had active rheumatoid arthritis and were receiving concomitant DMARD therapy. The registry distinguishes a bDMARD-naive population from a TNFα-inadequate responder population.

In Part A, the bDMARD-naive population was assigned among placebo and five LY2439821 dose groups: 3 mg, 10 mg, 30 mg, 80 mg, and 180 mg. The TNFα-inadequate responder population had placebo, 80 mg, and 180 mg groups. The stated Part A schedule administered treatment at Weeks 0, 1, 2, 4, 6, 8, and 10. Part B was optional; the arm descriptions specify 160 mg LY2439821 beginning at Weeks 16, 18, and 20, followed by dosing every 4 weeks through Week 60. Participants assigned to placebo in Part A also had a Part B description involving 160 mg LY2439821.

The primary endpoint was evaluated at Week 12 in the bDMARD-naive population. The registry defines its analysis population as the Part A full analysis set: randomized participants who received at least one dose and were bDMARD-naive. The primary outcome is the percentage of participants with ACR20 response. Secondary outcomes include ACR20 response in the TNFα-inadequate responder population, response thresholds for ACR and DAS28 in the bDMARD-naive population, changes in DAS28, and ACR20, ACR50, and ACR70 response outcomes in Parts A and B.

The registry record does not report the randomization ratio, allocation concealment details, masking procedures, eligibility criteria beyond the population descriptions, or the rationale for the optional Part B structure. The original sample-size justification and the trial’s full statistical analysis plan 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: Participants were randomized to Part A and had the option of continuing in the open-label extension, Part B after completing Part A.

Only the first 6 of 9 groups are shown.

Part A

Reasons for not completing this period:

Part B

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Part A FAS: All randomized participants who received at least 1 dose of study drug and were bDMARD-naive.

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

Only the first 6 of 10 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Dose-Response Relationship Measured by the Percentage of Participants With American College of Rheumatology (ACR) 20 Response in bDMARD-Naive Population, measured over Week 12.

The primary endpoint was the percentage of participants in the bDMARD-naive population who achieved an ACR20 response at Week 12. ACR20 is a composite response definition. It required at least 20% improvement from baseline in both tender joint count and swollen joint count, together with improvement in at least three of five other core measures: HAQ-DI, C-reactive protein, patient assessment of arthritis pain, patient global assessment of disease activity, and physician global assessment of disease activity.

This type of endpoint combines several aspects of a participant’s status rather than relying on a single symptom or laboratory measure. The threshold gives a clear responder classification for comparing the proportions in randomized groups. It is different from an average change score: it records whether each participant met the criteria, then summarizes the share meeting them in each group. Consequently, the percentage depends on the particular response definition and the treatment of missing observations.

For this primary outcome, missing values were imputed using non-responder imputation. Under that approach, participants with missing response information are counted as not having achieved the response. The registry states that the percentage was calculated as the number of ACR20 responders divided by the number of participants treated, multiplied by 100. The analysis population was the Part A full analysis set of randomized participants who received at least one dose and were bDMARD-naive. These definitions determine what the reported percentages mean; they should not be interpreted as a continuous measure of the amount of improvement or as an outcome at other time points.

7. Results Posted in the Registry

• 3 mg LY2439821 (bDMARD-naive Population): 45.0% (18/40) versus 33.3% (18/54) on control; risk difference 11.7 percentage points (95% CI -8.2 to 31.6), risk ratio 1.35 (0.81–2.25); unadjusted two-proportion z-test p = 0.250.
• 10 mg LY2439821 (bDMARD-naive Population): 51.4% (18/35) versus 33.3% (18/54) on control; risk difference 18.1 percentage points (95% CI -2.7 to 38.9), risk ratio 1.54 (0.94–2.53); unadjusted two-proportion z-test p = 0.089.
• 30 mg LY2439821 (bDMARD-naive Population): 54.1% (20/37) versus 33.3% (18/54) on control; risk difference 20.7 percentage points (95% CI 0.3 to 41.1), risk ratio 1.62 (1.00–2.62); unadjusted two-proportion z-test p = 0.049.
• 80 mg LY2439821 (bDMARD-naive Population): 56.1% (32/57) versus 33.3% (18/54) on control; risk difference 22.8 percentage points (95% CI 4.8 to 40.8), risk ratio 1.68 (1.08–2.62); unadjusted two-proportion z-test p = 0.016.

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 ACR20 Response in Tumor Necrosis Factor Alpha-Inadequate Responder (TNFα-IR) Population

• 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:
  • 80 mg LY2439821 (TNFα-IR Population): 40.0% (26/65) versus 23.4% (15/64) on control; risk difference 16.6 percentage points (95% CI 0.8 to 32.4), risk ratio 1.71 (1.00–2.91); unadjusted two-proportion z-test p = 0.043.
  • 180 mg LY2439821 (TNFα-IR Population): 39.0% (23/59) versus 23.4% (15/64) on control; risk difference 15.5 percentage points (95% CI -0.7 to 31.7), risk ratio 1.66 (0.96–2.87); unadjusted two-proportion z-test p = 0.062.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for 80 mg LY2439821 (TNFα-IR Population) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 180 mg LY2439821 (TNFα-IR Population) and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in Disease Activity Score (DAS28)-Part A

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • 3 mg LY2439821 (bDMARD-naive Population): mean -1.31 versus -0.82 on control; difference -0.49 (95% CI -0.92 to -0.06), standardized effect d = -0.48; z-test p = 0.025.
  • 10 mg LY2439821 (bDMARD-naive Population): mean -1.56 versus -0.82 on control; difference -0.75 (95% CI -1.29 to -0.20), standardized effect d = -0.63; z-test p = 0.007.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for 3 mg LY2439821 (bDMARD-naive Population) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for 10 mg LY2439821 (bDMARD-naive Population) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline in DAS28 - Part B

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • 3 mg/160 mg LY2439821(bDMARD-naive Population): mean -1.83 versus -2.14 on control; difference 0.31 (95% CI -0.19 to 0.81), standardized effect d = 0.28; z-test p = 0.226.
  • 10 mg/160 mg LY2439821(bDMARD-naive Population): mean -2.32 versus -2.14 on control; difference -0.18 (95% CI -0.78 to 0.42), standardized effect d = -0.14; z-test p = 0.562.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 3 mg/160 mg LY2439821(bDMARD-naive Population) and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 10 mg/160 mg LY2439821(bDMARD-naive Population) and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes extend the primary ACR20 question in two directions: they examine a separate TNFα-inadequate responder population and use additional response thresholds or disease activity measures. The outcome list includes ACR20, ACR50, and ACR70 responses, dose levels associated with specified fractions of maximum response, and changes in DAS28. This range gives a broader descriptive view of the trial’s endpoints, but it also means that the results address multiple questions rather than one single confirmatory contrast.

The posted results include ACR20 percentages for the TNFα-inadequate responder groups and mean DAS28 changes in Part A and Part B. The numerical results for the remaining listed secondary outcomes are not reported in the registry record. The secondary findings that are available are relevant as additional descriptions, but they do not replace the primary Week 12 ACR20 result in the bDMARD-naive population. Differences in population, endpoint definition, time point, and analysis method mean that results should be interpreted within their own context rather than combined as though they were the same measure.

Unadjusted secondary comparisons carry less weight because the record includes many outcomes, dose groups, time points, and analyses. When many tests are conducted, some results may appear notable by chance even if there is no underlying difference. The registry record does not report a multiplicity adjustment or a hierarchy that would distinguish confirmatory from exploratory secondary analyses. The sponsor reports Fisher exact tests for the TNFα-inadequate responder ACR20 comparisons and ANCOVA for selected Part A DAS28 comparisons, but the record does not provide a complete statistical analysis plan for all secondary outcomes.

Accordingly, secondary results are best treated as supporting or exploratory evidence about patterns across endpoints and populations. They can add context to the primary result, but they should not be used to claim a consistent treatment effect across outcomes without the complete prespecified analysis plan and multiplicity strategy.

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

• Percentage of Participants With ACR20 Response in Tumor Necrosis Factor Alpha-Inadequate Responder (TNFα-IR) Population
• Smallest Doses That Achieve 10%, 50%, and 90% of the Maximum ACR 20 Response in bDMARD-Naive Population
• Smallest Doses That Achieved 10%, 50%, and 90% of the Maximum Disease Activity Score (DAS) 28 Response in bDMARD-Naive Population
• Smallest Doses That Achieve 10%, 50%, and 90% of the Maximum ACR50 Response in bDMARD-Naive Population
• Change From Baseline in Disease Activity Score (DAS28)-Part A
• Change From Baseline in DAS28 - Part B
• Percentage of Participants With ACR20/50/70 Response - Part A
• Percentage of Participants With of ACR20/50/70 Response - Part B

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is binary at the participant level: each participant either meets the ACR20 definition or does not. The group result is expressed as a percentage. A risk difference compares the response proportions between two groups by subtracting one group’s proportion from the other. In the recalculated comparisons, a positive estimate means the observed responder proportion was higher in the active-dose group than in placebo. The risk-difference interval describes uncertainty around that unadjusted contrast. If an interval includes zero, values compatible with no difference are among the values covered by that interval; this does not prove that the groups are identical.

The posted primary percentages and the recalculated risk differences are related descriptions, but they are not interchangeable analyses. The recalculations use the posted summary statistics and are simple pairwise comparisons. They do not account for the trial’s full design or necessarily reproduce the sponsor’s model. The registry reports a logistic regression analysis across placebo and all five active-dose groups, with p = 0.031. Logistic regression is a method for modeling a binary outcome; in a multi-group setting it can evaluate an overall group or dose-response pattern. The record does not provide enough detail to identify every model term or assumption used in that analysis.

Other reported analyses use different methods because their endpoints or comparisons differ. Fisher exact tests are listed for the TNFα-inadequate responder ACR20 comparisons. This test evaluates association in a two-group table of binary outcomes and can be used when the observed counts are limited. For the Part A DAS28 change outcome, the registry lists ANCOVA comparisons. ANCOVA models a continuous outcome while adjusting for one or more covariates; the covariates used in the sponsor’s analysis are not reported in the registry record. The posted DAS28 results are group means and standard deviations, whereas the ACR20 results are percentages of responders.

For ACR20, missing response values were handled by non-responder imputation. This assigns missing observations to the non-responder category for the endpoint calculation. It is a conservative convention in the sense that missing participants do not contribute as responders, but it does not eliminate the possibility that missingness affects interpretation. The amount and pattern of missing data, reasons for missingness, and sensitivity analyses using alternative assumptions are not reported in the registry record. The specific missing-data handling for every secondary outcome is also not reported.

The primary analysis population is described as the Part A full analysis set: all randomized participants who received at least one dose and were bDMARD-naive. This definition is more specific than simply stating the number enrolled. The arm-specific denominators for the primary outcome range from 35 to 57 participants. Differences between enrolled participants and participants included in a particular analysis can arise from population definitions and available observations, but the registry record does not give a complete accounting of those differences.

Several statistical concepts help distinguish the results. The percentage in an arm is a descriptive estimate of the response frequency in that group. A risk difference expresses the contrast on an absolute scale, while a risk ratio compares the frequencies proportionally and an odds ratio compares the odds. These scales answer related but not identical questions. The recalculated intervals and p-values are unadjusted; they do not incorporate a correction for the many doses, secondary endpoints, and analyses in the record.

General statistical explanation: a p-value describes how incompatible the observed data are with a specified null model, under the assumptions of the analysis. It is not the probability that the null hypothesis is true, and it does not measure the magnitude or clinical importance of an effect. A confidence interval gives a range of estimates compatible with the data and the interval procedure. Wider intervals indicate less precision than narrower intervals, all else being equal. Neither a p-value nor an interval resolves questions about bias, missing data, or whether the endpoint is important to patients.

The trial’s original sample-size assumptions, planned target effect, and statistical analysis plan are not reported in the registry record. Planning illustrations based on observed effects are separate from the study’s original design. They can show how sample requirements change under stated assumptions, but they cannot establish that the original trial was adequately powered or recover its design rationale.

11. Confidence Intervals and Precision

The recalculated risk differences are estimates from the observed group summaries, and their intervals express uncertainty in those estimates. For the primary bDMARD-naive comparisons, the posted intervals are narrower for some contrasts and broader for others. The 3 mg and 10 mg intervals include zero; the 30 mg, 80 mg, and 180 mg intervals do not. These interval boundaries should be read as measures of precision for the unadjusted pairwise estimates, not as proof that any particular dose has a definite effect. The registry’s separate logistic regression result addresses the overall six-group analysis and should not be conflated with these pairwise intervals.

Arm sizes were modest, ranging from 35 to 57 participants in the primary outcome groups. In general, estimates from smaller groups tend to be less precise, although the observed event proportions and analysis method also affect interval width. The observed percentages themselves are point estimates; an interval conveys that the underlying response frequency or group contrast is not known exactly from the sample.

The registry record does not state the confidence level for the recalculated intervals, so no confidence percentage is assigned to them in this interpretation. The sponsor’s reported primary analysis gives a p-value but does not include an effect estimate or interval in the posted analysis details. As a result, the sponsor-reported overall test and the recalculated pairwise contrasts contribute different information: one is a model-based multi-group test, while the other set supplies unadjusted estimates and intervals for individual dose-versus-placebo contrasts.

Precision is also limited by the scope of the analysis population and outcome. The primary result concerns ACR20 at Week 12 among the defined bDMARD-naive participants. It does not directly quantify uncertainty for other endpoints, other populations, or later time points. Interpreting numerical precision alongside the endpoint definition, analysis population, missing-data rule, and multiplicity context gives a more complete view than relying on a single interval or p-value.

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 figures are stated as participants per arm for the binary outcomes in this trial. They are not total event counts. The illustrations assume a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Each planning row uses the stated fraction of the observed effect as the effect to detect. The original trial’s own design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the estimated participants needed in each arm under those planning assumptions to have an 80% chance of detecting the specified effect if that effect is present. The 90% power column uses a higher chance of detection and therefore generally calls for more participants per arm. These are planning calculations, not observed power. They are conditional on the assumed effect, allocation, significance level, and approximation; they do not describe the probability that the trial’s result is true.

Rows for 100% of the observed effect use the full recalculated contrast as the planning target. Rows for 75% or 50% of the observed effect represent smaller targets. Smaller effects are harder to distinguish from random variation, so they require more participants per arm under the same assumptions. For example, for the primary 80 mg comparison, the illustration gives 74 participants per arm for 80% power at 100% of the observed effect, 130 at 75%, and 287 at 50%. The figures are not recommendations for a future study; they show how the assumed effect size changes the planning requirement.

The illustrations are generated from observed trial effects and should not be mistaken for the actual enrollment target or a retrospective judgment of adequacy. The registry reports total enrollment of 448, but the primary bDMARD-naive arm sizes differ from one another. A planning calculation using equal allocation between two arms therefore does not reproduce the full multi-dose trial design.

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 = 0.333; treatment proportion p₁ = 0.450; average p̄ = 0.392.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6903 = 1.3530.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6854 = 0.5768.
4. n = (1.3530 + 0.5768)² / (0.1167)² = 273.6, rounded up to 274 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 <- 0.333333333333333                          # control proportion
p1 <- 0.45                          # 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(966875)
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 = 274 (80% power), 366 (90% power)
 75% of observed effect: n per arm = 481 (80% power), 643 (90% power)
 50% of observed effect: n per arm = 1,065 (80% power), 1,426 (90% power)
power.prop.test: 273.6 
Simulated power at n = 274 per arm: 0.799 (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 analysis is a logistic regression across the placebo group and five LY2439821 dose groups in the bDMARD-naive population. The registry reports p = 0.031 for that analysis. This is the sponsor’s analysis of the multi-group primary dose-response outcome, rather than a collection of separate placebo-versus-dose tests.

The recalculated results are simple, unadjusted pairwise risk differences based on posted summary statistics. They provide an absolute-scale contrast and an interval for each dose versus placebo, but they do not reproduce the sponsor’s logistic regression. The posted arm percentages and recalculated proportions also show a discrepancy. Because that discrepancy is flagged, the sponsor’s own analysis takes precedence over the recalculations for interpreting the primary result.

The recalculations can still help describe the size and precision of individual contrasts, provided they are identified as unadjusted. Their p-values should not be substituted for the sponsor’s overall test, and they do not account for multiplicity across the dose groups or other outcomes. The registry record does not report enough model detail to independently reconstruct the sponsor’s analysis or determine whether the posted overall test was adjusted for any other comparisons.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity is a central interpretive issue because the registry records 37 outcome measures and 108 analyses. The primary endpoint compares five active doses with placebo, while secondary outcomes cover another population, multiple response thresholds, dose-response summaries, change scores, and Part B results. Testing many hypotheses increases the chance of observing apparently unusual results by chance. An adjustment or a prespecified testing hierarchy can address that issue, but a multiplicity plan is not reported in the registry record.

The sponsor-reported primary logistic regression has p = 0.031 for the six-group analysis. The recalculated dose-versus-placebo p-values are unadjusted, as are the simple recalculated intervals. They should not be read as though they had automatically controlled the error rate across all comparisons. The registry also reports Fisher exact tests for the TNFα-inadequate responder ACR20 comparisons and ANCOVA for selected DAS28 comparisons; a comprehensive account of the sponsor’s testing strategy is not reported.

Missing data are handled explicitly for the primary ACR20 endpoint through non-responder imputation. Participants with missing response values are counted as non-responders in the primary calculation. The registry record does not report the amount or reasons for missing data, the assumptions underlying that rule, or sensitivity analyses using alternative missing-data approaches. Missing-data methods for all secondary outcomes are not reported in the registry record.

The primary population is the Part A full analysis set, defined as randomized participants who received at least one dose and were bDMARD-naive. The registry does not provide a detailed participant flow for every analysis or a complete accounting of exclusions and missing observations. The original sample-size justification and full statistical analysis plan are also not reported. These gaps limit assessment of whether the reported analyses followed a prespecified strategy and how robust the results are to alternative analysis choices.

16. Safety Results

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

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: meningitis (3.1%), pancreatitis acute (3.0%), death (3.0%), urinary tract infection (3.0%), aortic aneurysm (3.0%).

The registry reports serious adverse event counts by arm and study part, with the number at risk and number affected. In Part A among bDMARD-naive participants, serious adverse events were reported in 0 of 40 participants in the 3 mg group, 1 of 35 in the 10 mg group, 1 of 37 in the 30 mg group, 4 of 57 in the 80 mg group, 1 of 37 in the 180 mg group, and 1 of 54 in the placebo group. The TNFα-inadequate responder Part A groups had 5 of 65 affected in the 80 mg group, 6 of 59 in the 180 mg group, and 1 of 64 in the placebo group.

Part B counts are reported separately for participants continuing or switching to 160 mg LY2439821. They should not be pooled with Part A counts without accounting for the different study periods and group definitions. The registry also lists serious-event terms and percentages, including meningitis at 3.1%, pancreatitis acute at 3.0%, death at 3.0%, urinary tract infection at 3.0%, aortic aneurysm at 3.0%, and pneumonia at 2.9%. The group-level death counts are not reported in the registry record.

These counts describe reported events within the recorded risk groups. They do not by themselves establish whether an event was caused by study treatment, and comparisons are limited by group size and the study structure. They should be read as safety reporting, not as individualized risk estimates.

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

The primary result describes the percentage of participants meeting the study’s ACR20 definition at Week 12 in the bDMARD-naive population. The posted percentages were higher in each active-dose group than in placebo, and the sponsor reported an overall logistic regression p-value of 0.031 across the six groups. The pattern across doses rises at lower doses and is broadly similar among the higher-dose groups. These are observations from this trial, not a guarantee of an outcome for any individual participant.

The endpoint is a threshold-based composite. Meeting ACR20 means meeting the specified improvement criteria across joint counts and other core measures; it does not mean every measure improved by the same amount, nor does it describe the magnitude of change beyond the response threshold. The DAS28 change outcomes are separate measures and should not be treated as interchangeable with ACR20 response.

The primary analysis concerns a particular population, treatment schedule, and time point. The trial also included a TNFα-inadequate responder population and an optional Part B, but those results answer different questions. The findings do not establish effects at all doses, for all participants with rheumatoid arthritis, or beyond the time points reported. The registry record does not report all details needed to assess generalizability or reconstruct the sponsor’s full statistical model.

Statistical evidence and clinical importance are also distinct. A p-value does not measure the size or value of an outcome, and an observed difference does not alone show that the difference is meaningful to participants. Interpretation depends on the endpoint, precision, missing-data handling, multiplicity, safety reporting, and applicability of the trial population. The record supports a description of the posted trial results; it does not by itself support medical advice or a broad conclusion about treatment.

18. Why This Trial Matters Statistically

This trial offers a useful statistical example of dose-ranging analysis: several active dose groups are compared with placebo, while the primary question concerns an overall response pattern across doses. The sponsor’s logistic regression result and the recalculated pairwise risk differences illustrate how different analyses can address related but distinct questions. An overall multi-group test is not the same as five separate estimates of the absolute difference from placebo.

It also shows why effect estimates and precision matter alongside p-values. The recalculated risk differences are accompanied by intervals that vary in width, while the sponsor’s reported primary analysis supplies a p-value without an effect estimate or interval in the posted analysis details. Reading those pieces together prevents a single test result from being mistaken for a complete account of the evidence.

The endpoint and population definitions are equally important. ACR20 is a composite responder outcome, the primary population excludes participants outside the specified bDMARD-naive full analysis set, and missing primary responses were handled as non-responders. Secondary outcomes include different populations and measurement scales. Those distinctions affect what can be concluded from each result.

Finally, the record’s many outcome measures and analyses make multiplicity, transparent analysis planning, and careful labeling of unadjusted calculations especially relevant. The registry illustrates both the value and limits of posted results: they permit readers to examine observed group patterns, but missing details about the original design assumptions and full analysis plan restrict independent evaluation of every statistical choice.

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 primary outcome group sizes range from 35 to 57 participants, so precision for individual dose comparisons is limited. The study distributes the bDMARD-naive population across placebo and five active doses, rather than concentrating all participants in a single active-versus-control comparison. The simple planning illustrations use equal allocation between two arms, which does not reproduce that multi-arm structure.

The primary endpoint is a composite threshold, and its calculation uses non-responder imputation for missing values. The registry record does not report the amount or pattern of missingness, reasons for missing observations, or sensitivity analyses. The record also does not report the original sample-size justification, the full statistical analysis plan, the randomization ratio, or a multiplicity strategy. These omissions limit assessment of prespecification and robustness.

The registry identifies a discrepancy between posted primary summary results and recalculated summaries. The sponsor’s reported logistic regression therefore takes precedence; the recalculated pairwise estimates are descriptive, unadjusted analyses and should not be treated as substitutes. In addition, only selected secondary outcome statistics are available in the record described, and the remaining listed secondary outcome values are not reported in the registry record.

Safety counts are given by arm and study part, but they do not establish causality, and group-level death counts are not reported. Part B was optional and involved a subsequent 160 mg regimen, so Part B group summaries reflect a different stage and group definition from the primary Part A comparison. These features constrain direct comparisons across periods and populations.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study in Patients With Rheumatoid Arthritis — NCT00966875.
• Registry record last updated 2016-05-26; status last verified 2016-04.
• Results first posted 2016-05-26.
• 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

NCT00966875 is a completed randomized, quadruple-masked, parallel-group Phase 2 study of LY2439821 in patients with active rheumatoid arthritis on concomitant DMARD therapy. The registry records enrollment of 448, a start of 2009-08, and primary completion in 2011-03. The primary endpoint was Week 12 ACR20 response in the bDMARD-naive Part A full analysis set.

Across placebo and five LY2439821 dose groups, the registry reports an overall logistic regression p-value of 0.031. The posted ACR20 percentages were 34.1 for placebo and 46.2 to 55.3 across the active-dose groups. Selected secondary results cover the TNFα-inadequate responder population and DAS28 change. Interpretation is constrained by the discrepancy between posted summaries and recalculations, unadjusted secondary analyses, and statistical-plan details not reported in the registry record.