A Study Of Tocilizumab in Patients With Moderate to Severe Active Rheumatoid Arthritis Who Have an Inadequate Response to or Are Unable to Tolerate Biologic and Non-Biologic Disease-modifying Antirheumatic Drugs (DMARDs): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study Of Tocilizumab in Patients With Moderate to Severe Active Rheumatoid Arthritis Who Have an Inadequate Response to or Are Unable to Tolerate Biologic and Non-Biologic Disease-modifying Antirheumatic Drugs (DMARDs) (NCT00891020), a randomized phase 3 study. The primary outcome is Percentage of Participants Experiencing at Least One Serious Adverse Event (SAE) During the 24 Week Treatment Period.

2. Clinical Question

3. Trial Design

This was a Phase 3 randomized, open-label, parallel-group study of tocilizumab in people with moderate to severe active rheumatoid arthritis and an inadequate response to, or inability to tolerate, biologic or non-biologic DMARDs. The registry lists enrollment as 886. The study began in 2009-05, reached primary completion in 2010-07, and is recorded as completed. The sponsor was Hoffmann-La Roche.

Randomization assigned participants among three regimens: tocilizumab 8 mg/kg as monotherapy, tocilizumab 4 mg/kg with a non-biologic DMARD, or tocilizumab 8 mg/kg with a non-biologic DMARD. In the combination groups, the non-biologic DMARD was selected by the investigator. The treatment descriptions specify intravenous infusions every 4 weeks, for a total of 6 infusions over 24 weeks. The monotherapy description says participants who completed the treatment period could be offered entry to a long-term extension at the investigator’s discretion. The available description for one combination arm is incomplete, so further details of that arm’s follow-up are not reported in the registry record.

Masking was listed as none: participants and study personnel were not blinded according to the design record. That matters when interpreting subjective outcomes and decisions made during treatment, although the primary endpoint is the occurrence of a serious adverse event. The record describes risk-mitigation strategies developed with the FDA, but does not give their details in the available trial summary.

The primary endpoint covered serious adverse events during the 24-week treatment period. Secondary outcomes included other safety measures and measures of clinical response, including remission, DAS28, ACR response, and RAPID3. The registry lists 10 outcome measures and 9 posted secondary outcomes. The trial’s detailed allocation procedures, original sample-size rationale, and 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: 1129 patients were screened for the study. Of these 1129 patients, 886 patients were enrolled and 243 patients were screen failures.

24 Week Treatment Period

Reasons for not completing this period:

Long-term Extension Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Safety Population. Patients were included in the TCZ dose group according to the first infusion they actually received. Patients were assigned as using a nonbiologic DMARD if they took at least 1 dose of nonbiologic DMARD during the treatment period. Patients who did not take at least 1 dose of nonbiologic DMARD were considered as on monotherapy.

5. Baseline Characteristics

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

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants Experiencing at Least One Serious Adverse Event (SAE) During the 24 Week Treatment Period, measured over 24 Weeks.

The primary endpoint is the percentage of participants who experienced at least one serious adverse event during the 24-week treatment period. The registry defines a serious adverse event as an event that, at any dose, was fatal, life-threatening, required inpatient hospitalization or prolonged existing hospitalization, resulted in persistent or significant disability or incapacity, was a congenital anomaly or birth defect, or was medically significant or required intervention to prevent one of those outcomes.

This is a participant-level safety measure: a participant is counted if at least one qualifying event occurred during the stated period. It is not a count of all events, because one participant could have more than one event. Nor does the endpoint describe the severity or duration of every adverse event that did not meet the serious-event definition. The registry lists the measure as a percentage of participants and names the 24-week treatment period as its timeframe.

In general, serious adverse events are monitored because they capture important harms, including events requiring major intervention or hospitalization. They can help characterize safety during a trial, but a comparison of percentages does not, by itself, establish that a regimen caused an event. Interpretation also depends on the study population, exposure, event definitions, and completeness of follow-up. Further details on endpoint adjudication and handling of participants with incomplete observation are not reported in the registry record.

7. Results Posted in the Registry

• Tocilizumab 4 mg/kg + DMARD: 8.0% (29/364) versus 5.8% (8/138) on control; risk difference 2.2 percentage points (95% CI -2.6 to 7.0), risk ratio 1.37 (0.64–2.93); unadjusted two-proportion z-test p = 0.406.
• Tocilizumab 8 mg/kg + DMARD: 8.4% (32/381) versus 5.8% (8/138) on control; risk difference 2.6 percentage points (95% CI -2.2 to 7.4), risk ratio 1.45 (0.68–3.07); unadjusted two-proportion z-test p = 0.326.

8. Statistical Interpretation

9. Statistical Methods Explained

For a binary endpoint such as experiencing at least one serious adverse event, each participant is classified as having or not having the event during the stated period. The primary result is expressed as the percentage of participants with an event in each randomized group. In general, this makes the result relatively easy to interpret, but it reduces the details of event timing, recurrence, and duration to a yes-or-no participant-level outcome.

What does a risk difference express?

The reported pairwise effect measure is the risk difference: the event proportion in a combination group compared with the event proportion in the monotherapy reference group. A value above zero corresponds to a higher event proportion in the combination group; a value below zero corresponds to a lower proportion. Zero represents no difference in the proportions. The posted confidence intervals for both contrasts extend across zero, so they do not rule out no difference. The detailed formula and method used for the registry’s own analysis are not reported in the registry record.

How should the p-values be read?

The posted unadjusted p-values are 0.406 for the comparison of 4 mg/kg with a DMARD against monotherapy, and 0.326 for the comparison of 8 mg/kg with a DMARD against monotherapy. A p-value describes how compatible the observed data are with a specified null model under the assumptions of the test; it is not the probability that the null hypothesis is true. The exact test, null hypothesis, and prespecified interpretation threshold for the trial’s own analyses are not reported in the registry record. These p-values therefore should not be treated as a substitute for the effect estimates and their uncertainty.

What does the confidence interval add?

In general, a confidence interval communicates the range of effect values that are compatible with the data under the method used to construct it. It conveys precision as well as direction: an interval that spans zero allows for no difference, while its endpoints also indicate that a range of positive and negative differences remains plausible. The confidence level and interval-construction method for the posted comparisons are not reported in the registry record. The intervals are presented as posted; they should not be mistaken for evidence that all values inside them are equally likely.

Why does the analysis population matter?

The primary outcome description identifies a safety population. Participants were assigned to a tocilizumab dose group according to the first infusion they actually received. A participant was considered to be using a non-biologic DMARD if they took at least one dose of one during the treatment period. This is a treatment-exposure-based description, rather than simply a statement that every participant is analyzed according to original randomization. The remaining population text is incomplete, and the registry record does not report a full analysis-population specification or how deviations from assigned treatment were handled in each comparison.

In general, randomized assignment supports comparisons by initially assigned group, while classification by treatment received can change the interpretation if participants do not receive the assigned regimen. The record does not state how many participants changed treatment, stopped early, or had outcome status unavailable. Those details are important to assess whether the presented safety results reflect the full randomized groups, the treatments actually received, or another analysis set.

What does over-dispersion mean for a binary endpoint?

General explanation: over-dispersion occurs when observed variability is greater than a model assumes. It is often discussed for count data or clustered observations; a simple binary participant-level outcome is not automatically subject to an over-dispersion adjustment. If outcomes are correlated within sites or otherwise depart from model assumptions, uncertainty estimates may require methods that account for that structure. The registry record does not report a clustering adjustment or over-dispersion assessment for these comparisons, so no such method should be inferred.

The trial is described as randomized, parallel, and open-label, but randomization procedures and the statistical method for the primary endpoint are not reported in the registry record. The registry also does not report whether estimates were adjusted for baseline characteristics, whether the comparisons were prespecified, or whether an interim analysis was conducted. These omissions limit reconstruction of the intended analysis; they do not change the fact that the posted endpoint percentages and comparison results can be described as recorded.

10. Confidence Intervals and Precision

The precision of a comparison is reflected by its confidence interval as well as by its point estimate. For 4 mg/kg tocilizumab with a DMARD versus monotherapy, the posted risk-difference interval is -0.026 to 0.070. For 8 mg/kg tocilizumab with a DMARD versus monotherapy, it is -0.022 to 0.074. Each interval includes zero and reaches into both negative and positive values, so the direction of the difference is uncertain.

The intervals also span a range of possible effects rather than concentrating narrowly around zero. This means that the observed results do not pin down a small difference with high precision. They remain compatible with lower as well as higher event proportions in the combination group relative to monotherapy. A confidence interval is not a guarantee that the true effect lies within its endpoints, and the confidence level and construction method are not reported in the registry record.

Precision depends in part on the amount of information available, including the number of participants and the frequency of the outcome. The displayed arm sizes are 138, 364, and 381, while the primary outcome is a relatively uncommon event in the posted groups. The registry record does not report an original sample-size justification for this safety comparison or an analysis designed to establish equivalence. Thus, the results should not be interpreted as demonstrating that the regimens have the same serious-adverse-event risk merely because the intervals include zero.

The p-values, 0.406 and 0.326, do not express precision on their own. The intervals more directly show the range of contrasts compatible with the data under the calculation method. Neither the p-values nor the intervals replace consideration of event definitions, observation time, analysis population, and missing outcome information.

11. 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 illustrations based on the observed pairwise effect, not the trial’s original design calculations. They assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect. The figures are participants per arm, not total participants or total events. These are planning illustrations, not observed power.

For the comparison of 4 mg/kg with a DMARD against monotherapy, planning for 100% of the observed effect gives 2,136 participants per arm for 80% power and 2,859 participants per arm for 90% power. Planning for 75% of the observed effect gives 3,658 and 4,897 participants per arm, respectively. Planning for 50% gives 7,917 and 10,599 participants per arm.

For the comparison of 8 mg/kg with a DMARD against monotherapy, planning for 100% of the observed effect gives 1,528 participants per arm for 80% power and 2,045 participants per arm for 90% power. Planning for 75% gives 2,602 and 3,483 participants per arm, respectively. Planning for 50% gives 5,594 and 7,488 participants per arm.

In plain terms, the 80% column is the estimated number of participants needed in each arm under the stated assumptions for an 80% chance of detecting the chosen effect if it is present; the 90% column uses a higher target chance. Smaller effects require more participants, as the planning values illustrate. The trial’s own sample-size rationale and assumptions are not reported in the registry record, so these calculations should not be treated as the study’s original recruitment target or as evidence of its actual power.

12. 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 = 0.058; treatment proportion p₁ = 0.080; average p̄ = 0.069.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.3580 = 0.7017.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.3577 = 0.3010.
4. n = (0.7017 + 0.3010)² / (0.0217)² = 2,135.3, rounded up to 2,136 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 <- 0.0579710144927536                          # control proportion
p1 <- 0.0796703296703297                          # 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(891020)
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 = 2,136 (80% power), 2,859 (90% power)
 75% of observed effect: n per arm = 3,658 (80% power), 4,897 (90% power)
 50% of observed effect: n per arm = 7,917 (80% power), 10,599 (90% power)
power.prop.test: 2135.3 
Simulated power at n = 2,136 per arm: 0.802 (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.

13. Multiplicity, Missing Data and Analysis Populations

The study lists several outcomes beyond the primary serious-adverse-event endpoint, including serious and non-serious adverse events of special interest, clinical remission, DAS28 change, ACR response, dose increases, and RAPID3 change. The registry lists 10 outcome measures and 9 posted secondary outcomes. Multiple outcomes and multiple time points can create multiple opportunities to observe a result that appears notable by chance, particularly when many comparisons are examined.

The registry record does not report a multiplicity adjustment, a hierarchy of outcomes, or a strategy for controlling the overall type I error rate. It also does not say whether the pairwise comparisons described for the primary endpoint were prespecified. The displayed p-values are unadjusted. They should therefore be interpreted as results for the individual comparisons, without assuming that they account for the full set of outcomes or analyses.

Information about missing outcome data is also limited. The registry describes a safety population and gives rules for dose-group and DMARD classification, but does not report the number of participants with missing safety follow-up, the handling of incomplete observation, or any imputation method. For a safety endpoint measured during treatment, time off treatment and duration of observation can affect the opportunity to record events. Details of exposure-adjusted analyses or follow-up after treatment discontinuation are not reported in the registry record.

The registry does not report whether interim analyses were conducted, whether stopping rules applied, or how protocol deviations were handled in the primary analysis. It also does not provide a complete statistical analysis plan, including model specification and any covariate adjustment. These gaps limit assessment of the full inferential strategy. They do not justify assuming that interim testing, imputation, or multiplicity correction was or was not performed.

The population description says that dose groups were based on the first infusion actually received and that DMARD use required at least one dose during treatment. That rule helps explain the safety-group classification but does not provide a full account of how participants were analyzed relative to randomized assignment. Further analysis-population details are not reported in the registry record.

14. 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: pneumonia (2.1%), cellulitis (1.6%), large intestine perforation (0.8%), clostridial infection (0.7%), herpes zoster (0.7%).

The primary endpoint percentages were 5.8% for 8 mg/kg monotherapy, 8.0% for 4 mg/kg with a DMARD, and 8.4% for 8 mg/kg with a DMARD. Each percentage refers to participants with at least one serious adverse event during the 24-week treatment period, according to the stated endpoint definition. These percentages do not count the total number of serious events, and a participant with multiple events remains a participant with at least one event for this measure.

A separate safety listing reports serious-event counts of 11 among 138 participants in the monotherapy group, 45 among 364 in the 4 mg/kg combination group, and 40 among 381 in the 8 mg/kg combination group. The registry record does not explain how these counts relate to the primary endpoint percentages; they should not be treated as interchangeable summaries. It also lists pneumonia at 2.1%, cellulitis at 1.6%, large intestine perforation at 0.8%, clostridial infection at 0.7%, herpes zoster at 0.7%, and small intestinal obstruction at 0.7% among the top serious-event terms. The listed terms do not provide a complete account of all safety events or their clinical context.

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

The results answer a specific descriptive question: what percentage of participants in each treatment group experienced at least one serious adverse event during the 24-week treatment period, as defined in the registry. The recorded percentages are 5.8% with 8 mg/kg monotherapy, 8.0% with 4 mg/kg plus a DMARD, and 8.4% with 8 mg/kg plus a DMARD. The pairwise comparisons show uncertainty broad enough to include no difference and differences in either direction.

These findings do not establish that one treatment regimen causes more or fewer serious adverse events than another. They also do not demonstrate equivalence. A result that is uncertain is not the same as evidence that the risks are identical. Interpretation is constrained by the interval width, the limited endpoint timeframe, and details of the analysis and follow-up that are not reported in the registry record.

The study was open-label and the treatment groups differ in both dose and use of a concomitant non-biologic DMARD. The comparisons therefore do not isolate the effect of tocilizumab dose or the separate effect of adding a DMARD. The reported percentages also do not describe all adverse events, all consequences of treatment, or long-term safety. The registry mentions a long-term extension option for some participants, but the results of that extension are not part of the primary endpoint.

Clinical response outcomes are listed as secondary measures, but their results are not included in the available trial details. The primary endpoint is a safety measure, not a test of rheumatoid arthritis improvement. It should not be used to infer symptom benefit or lack of benefit. In general, trial findings apply most directly to the population, treatments, outcome definitions, and follow-up that were studied; the registry record does not provide enough detail to extend this specific comparison beyond those terms.

16. Why This Trial Matters Statistically

This study offers a useful statistical example of why an observed difference and the uncertainty around it should be considered together. The serious-adverse-event percentages are higher in the two DMARD combination groups than in the monotherapy group, but the pairwise confidence intervals extend across zero and the unadjusted p-values are 0.406 and 0.326. A simple ranking of the percentages would omit the uncertainty and could overstate what the data establish.

The three-arm structure also matters. The treatment groups do not differ on a single feature: the combination groups include a non-biologic DMARD, and the tocilizumab doses differ across arms. As a result, the comparisons describe regimens rather than separately identifying the contribution of each treatment component. This is a general issue in interpreting multi-arm studies when interventions vary along more than one dimension.

The endpoint illustrates the value and limits of a participant-level binary safety measure. It gives a clear summary of whether a participant experienced at least one serious event, while not conveying event recurrence, timing, or the full range of adverse events. The separate serious-event listings and the endpoint percentages also require careful reading because the registry record does not explain their relationship.

Finally, the planning illustrations show how sample-size needs depend on the effect size chosen for detection. Smaller target effects require more participants per arm under the stated assumptions. These planning results are distinct from the trial’s own design rationale, which is not reported in the registry record. Keeping descriptive results, uncertainty, and planning assumptions separate is essential to a transparent statistical reading.

17. 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 registry record gives only partial detail about the primary analysis. The method used to calculate the trial’s primary endpoint comparison, the confidence level and construction method for the posted intervals, any covariate adjustment, and the prespecified decision threshold are not reported in the registry record. The original sample-size justification is also not reported, so the planning illustrations cannot explain how enrollment was determined.

The safety-population description uses treatment actually received for dose-group classification and at least one DMARD dose for DMARD classification. The record does not give a complete account of the analysis population, crossover or treatment changes, duration of exposure, discontinuations, or missing safety follow-up. It also does not report how incomplete follow-up or protocol deviations were handled.

The open-label design may be relevant to the observation and reporting of adverse events, but the registry record does not describe event adjudication or the extent of any independent review. The treatment arms differ in dose and concomitant DMARD use, limiting interpretation of a pairwise comparison as the effect of one component alone. The primary endpoint covers 24 weeks, and it cannot describe longer-term event patterns.

The primary endpoint values and the separate safety-group counts are reported in different parts of the record, without an explanation of their relationship. The reason for that distinction is not reported in the registry record, so the counts and percentages should be presented separately rather than reconciled by assumption. Detailed results for the listed secondary outcomes, the full statistical analysis plan, multiplicity handling, and any interim analyses are likewise not reported.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: A Study Of Tocilizumab in Patients With Moderate to Severe Active Rheumatoid Arthritis Who Have an Inadequate Response to or Are Unable to Tolerate Biologic and Non-Biologic Disease-modifying Antirheumatic Drugs (DMARDs) — NCT00891020.
• Registry record last updated 2012-10-25; status last verified 2012-10.
• Results first posted 2012-04-17.
• 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.

21. Record Summary

NCT00891020 is a completed, randomized, open-label, parallel Phase 3 study sponsored by Hoffmann-La Roche. It enrolled 886 participants with moderate to severe active rheumatoid arthritis and compared three tocilizumab regimens: 8 mg/kg monotherapy, 4 mg/kg with a non-biologic DMARD, and 8 mg/kg with a non-biologic DMARD. The treatments were described as intravenous infusions every 4 weeks for 6 infusions over 24 weeks.

The primary endpoint was the percentage experiencing at least one serious adverse event during the 24-week treatment period. Registry-reported percentages were 5.8%, 8.0%, and 8.4%, respectively. Pairwise comparisons against monotherapy had unadjusted p-values of 0.406 and 0.326, with confidence intervals spanning zero. The registry record does not report the full analysis plan, original sample-size rationale, missing-data handling, or multiplicity strategy.