Adalimumab in JIA-associated Uveitis Stopping Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Adalimumab in JIA-associated Uveitis Stopping Trial (NCT03816397), a randomized phase 4 study. The primary outcome is Time to Treatment Failure.

2. Clinical Question

3. Trial Design

This was a Phase 4, randomized, parallel-group trial. The registry records quadruple masking. Its brief description also calls the study double-masked, controlled, stratified, and block-randomized; the exact meaning of the different masking descriptions, as well as the stratification factors and block procedures, is not reported in the registry record.

The trial is titled Adalimumab in JIA-associated Uveitis Stopping Trial. The registry lists an enrollment of 87. The posted primary-outcome groups are Continue Adalimumab, with 43, and Stop Adalimumab, with 44. Participants in the continuation arm received their current weight-based adalimumab dose subcutaneously every other week. Participants in the stopping arm received a volume-matched placebo on the same schedule.

The registry lists the study start as 2020-03-15 and primary completion as 2025-01-23, with status Completed. The primary outcome timeframe is from baseline until 48 weeks post-randomization. The trial description says the aim was to determine the feasibility of discontinuing adalimumab in patients whose uveitis was quiescent. The posted primary endpoint, however, is time to treatment failure, defined by recurrence criteria rather than by a general feasibility measure.

The registry record does not report the original sample-size justification, the detailed randomization procedures, the allocation ratio as a formal protocol specification, or the trial’s full analysis plan. Those details should not be inferred from the posted group counts or from the planning illustrations based on the observed effect.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Two patients dropped out of the trial before reaching the primary endpoint (one in each group). The length of time that these patients participated in the trial before dropout is included in the primary outcome analysis given the nature of survival analyses. Patients still in active follow-up who had not yet reached the primary endpoint (13 in the adalimumab group and six in the placebo group) were censored at the interim analysis date, Feb 14, 2024. No patients were lost to follow-up.

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 Time to Treatment Failure, measured over From baseline until 48 weeks post-randomization.

The primary endpoint was time to treatment failure, measured in days from baseline, with the stated timeframe extending to 48 weeks post-randomization. It is a time-to-event outcome: the analysis considers both whether failure occurred and when it occurred. This is useful when the timing of recurrence matters and when participants may have different amounts of observed follow-up.

The registry defines failure as recurrence of ocular inflammation in at least one eye. Listed criteria include 3+ anterior chamber cells at a single visit; more than 0.5+ anterior chamber cells for at least 28 days; a two-step increase in anterior chamber cells at two visits at least 7 days apart; or specified posterior eye inflammation or macular edema at a visit. The definition also permits failure to be declared for persistent and severe joint inflammation that necessitates unmasking to manage arthritis recurrence.

The outcome is reported as a median time and inter-quartile range, alongside a hazard ratio from a Log Rank analysis. The Stop Adalimumab arm has a posted median of 119 days and an inter-quartile range of 84 to 243 days. A median for the Continue Adalimumab arm is not reported in the registry record. The record notes that two patients dropped out before reaching the primary endpoint, one in each group, and that their observed participation time was included in the analysis. This treatment of partial follow-up is consistent with the time-to-event framing, but the registry record does not give further details about censoring procedures.

7. Results Posted in the Registry

• Continue Adalimumab vs Stop Adalimumab: hazard ratio 8.70 (95% CI 3.60–21.20), reported p < 0.001 (as reported in the registry).

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a time-to-event measure. For each participant, the relevant information is the time from baseline to treatment failure, or the time observed without a recorded failure. Time-to-event methods can use partial follow-up rather than treating every participant with incomplete observation as if no event occurred. General explanation: a participant who leaves observation before an event can contribute information up to the time they were last observed, provided the analysis handles that observation appropriately. The registry says the observed time for two dropouts was included, but does not report all details of its censoring implementation.

What does a hazard ratio describe?

A hazard is the instantaneous rate at which an event occurs among participants who have not yet had that event, conditional on remaining at risk at that time. A hazard ratio compares these rates between groups over the analyzed follow-up. It is a relative measure, not a difference in median days, and it does not state the probability that an individual participant will experience failure. The reported hazard ratio is 8.7 for Continue Adalimumab versus Stop Adalimumab. Its interpretation depends on the comparison ordering recorded by the sponsor.

A hazard ratio also summarizes a comparison over time. General explanation: when the relative event rate changes substantially during follow-up, a single hazard ratio may not represent the contrast equally well at every time point. The registry record does not report an assessment of whether the proportional-hazards assumption was examined or whether the hazard ratio was considered constant over time. No conclusion about that assumption can be drawn from the posted estimate alone.

What does the Log Rank method do?

The Log Rank method compares observed event timing between groups across the follow-up period. It uses information about when failures occur, rather than comparing only the final proportion with failure. The sponsor’s analysis is listed as a Log Rank analysis and reports a hazard ratio, a 95% confidence interval, and p < 0.001. The registry record does not give the test statistic, degrees of freedom, detailed model specification, or additional adjustment variables.

General explanation: the Log Rank comparison is most straightforward to interpret when the relative event rates are reasonably stable over time. It may be less representative as a single summary if the group contrast changes over follow-up. The record does not report diagnostic plots or other checks addressing that issue. The posted analysis should therefore be described as the sponsor’s reported Log Rank result, without assuming unreported checks or modeling choices.

How should the median and inter-quartile range be read?

The median time is the point at which the estimated event-time distribution reaches its midpoint. The inter-quartile range describes the span from the lower to the upper quartile. For Stop Adalimumab, the registry reports a median of 119 days and an inter-quartile range of 84 to 243 days. The median for Continue Adalimumab is not reported in the registry record. These quantities are descriptive summaries; they are not substitutes for the full time-to-event comparison and do not show how precise the hazard-ratio estimate is.

General explanation: time-to-event medians can be estimated using methods that account for incomplete follow-up, so the median need not be a simple midpoint of observed event times. The registry gives the outcome parameter as a median and the dispersion as an inter-quartile range, but does not describe the specific estimator used for those summaries. The method used to calculate the median and interval is therefore not reported in the registry record.

What does the endpoint definition contribute?

The definition sets clinical criteria for counting an event. Here, treatment failure includes specified signs of ocular inflammation, including anterior chamber findings, vitreous haze, active retinal or choroidal inflammation, and macular edema. It also includes a route for declaring failure because of persistent and severe joint inflammation when unmasking is needed to manage arthritis recurrence. This means the endpoint is not limited to eye findings alone. The record does not report adjudication procedures, whether outcome assessors were independent, or how borderline findings were resolved.

The registry describes the trial as masked, while its brief description uses the term double-masked and the design record lists quadruple masking. Masking can reduce the influence of expectations on care or outcome assessment, but the exact roles covered by each masking label are not stated. No further inference about the implementation of masking is warranted.

Finally, the record reports one primary outcome measure and one analysis, and no secondary outcomes are posted. The statistical analysis plan, any interim analyses, and any prespecified sensitivity analyses are not reported in the registry record. The posted result can be interpreted as the reported primary comparison, while its full analysis context remains incompletely documented.

10. Confidence Intervals and Precision

The hazard ratio is 8.7, with a 95% confidence interval from 3.6 to 21.2. A confidence interval describes a range of values compatible with the data and the method’s assumptions; it is not a guarantee that the true effect lies in that particular interval. In repeated use of the same method under its assumptions, intervals at this confidence level would contain the underlying parameter at the stated rate.

The interval does not include the value representing equal hazards, consistent with the reported p < 0.001 for the Log Rank analysis. At the same time, the interval spans a broad range of hazard-ratio values. The direction of the comparison is clear in the sponsor’s reported estimate, but the precise magnitude is less certain than the p-value alone might suggest. Statistical evidence against equal hazards and precision of the effect-size estimate are related but distinct questions.

The median time for Stop Adalimumab is 119 days, with an inter-quartile range from 84 to 243 days. The registry does not report a confidence interval for that median, nor does it report a median for Continue Adalimumab. The inter-quartile range describes the spread of the event-time summary, not uncertainty around the median estimate. Confidence intervals for group medians or a full survival curve are not reported in the registry record.

Precision is also affected by the amount and timing of event information, not only by the number assigned to each group. The registry’s planning illustrations express requirements in total events for time-to-event comparisons. The observed event count and a detailed accounting of censoring are not reported in the registry record, so those factors cannot be independently evaluated from the posted summary.

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 stated as total events, not participants per arm. For the time-to-event illustration, the calculation uses the Schoenfeld approach. It assumes 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.

For planning at 100% of the observed effect, the figures are 7 total events for 80% power and 9 total events for 90% power. At 75% of the observed effect, the figures are 12 total events for 80% power and 16 total events for 90% power. At 50% of the observed effect, they are 27 total events for 80% power and 36 total events for 90% power.

The 80% power column describes the total event count estimated to give an 80% chance of detecting the specified effect under the calculation’s assumptions, if that effect were the true one. The 90% power column describes the corresponding target for a 90% chance. These are planning probabilities under a hypothetical design, not probabilities that the posted result is correct. As the target effect gets smaller, more events are needed to distinguish it from the null under the stated assumptions.

These are planning illustrations based on the observed effect, not the trial’s own original design assumptions. The original sample-size justification and design assumptions are not reported in the registry record. This is not observed power; it does not measure the probability that the completed trial detected an effect after the results were known.

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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 8.700, so ln(HR) = 2.1633.
2. d = 4 × (1.96 + 0.842)² / [ln(8.700)]² = 6.7, rounded up to 7 events in total 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
hr <- 8.7                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(3816397)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 7 (80% power), 9 (90% power)
 75% of observed effect: events = 12 (80% power), 16 (90% power)
 50% of observed effect: events = 27 (80% power), 36 (90% power)
Simulated power with 7 events: 0.468 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

13. Analyses Reported by the Sponsor

The posted hazard ratio is the sponsor’s reported estimate from a Log Rank analysis: 8.7 for Continue Adalimumab versus Stop Adalimumab, with a 95% confidence interval from 3.6 to 21.2 and p < 0.001. The posted interpretation uses this sponsor-reported time-to-event result. No independently recalculated hazard ratio is used in the statistical summary.

The registry record does not flag a discrepancy between the sponsor’s analysis and an alternative recalculation. Accordingly, there is no conflicting estimate to reconcile. The sponsor’s own analysis is the relevant reported result, and it should not be replaced with an estimate reconstructed from the group medians or inter-quartile range. Those summaries do not contain enough information to reproduce the Log Rank hazard ratio.

The record does not include enough detail to reproduce the sponsor’s analysis fully. In particular, the detailed model specification, diagnostics, and full censoring procedures are not reported in the registry record. The estimate and its uncertainty should therefore be attributed to the reported sponsor analysis rather than treated as an independently verified calculation.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure and one analysis, with no secondary outcomes posted. It does not report a plan for controlling multiplicity across outcomes, comparisons, subgroups, or repeated analyses. The absence of posted secondary outcomes does not establish that no other analyses were performed; it means they are not reported in the registry record.

The record does not report interim analyses, stopping rules, or any adjustment for repeated looks at the data. These aspects of the analysis plan are not reported in the registry record, so the possibility and handling of interim monitoring cannot be assessed. The reported p-value should be read as the sponsor’s stated result, without assuming an adjustment that is not described.

For missing or incomplete follow-up, the registry says two patients dropped out before reaching the primary endpoint, one in each group, and that their observed time in the trial was included because of the nature of survival analyses. It also states that time from all participants is included in the analysis. The available population description is incomplete, and details about censoring, the treatment of participants still in active follow-up, and the handling of other missing data are not fully reported in the registry record.

The analysis population is also not described in enough detail to determine whether the sponsor used an intention-to-treat, per-protocol, or other defined population. The randomization and the posted arm sizes are recorded, but the analysis set and any exclusions are not. No conclusions about analysis-population membership or sensitivity to missing data should be drawn beyond the specific dropout information stated in the record.

15. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: surgery for torted cyst of morgagni (2.3%), shortness of breath at rest; suspected to be anxiety-related (2.3%), over 5 times the upper limit of normal for alanine aminotransferase (2.3%), planned orthognathic surgery (2.3%).

The registry reports serious adverse events for 4 of 43 participants in Continue Adalimumab and 0 of 44 participants in Stop Adalimumab. Other adverse events are reported for 34 of 43 participants and 29 of 44 participants, respectively. Deaths are recorded as 0 in each group. These are descriptive counts from the posted safety record and do not by themselves establish the cause of an event or a comparative treatment effect.

The listed serious-event terms include surgery for torted cyst of Morgagni, shortness of breath at rest suspected to be anxiety-related, over 5 times the upper limit of normal for alanine aminotransferase, and planned orthognathic surgery. Each is listed as 2.3%. The registry does not give further clinical context for these terms in the posted summary, so their relationship to treatment cannot be determined from these data alone.

Safety counts depend on how events were defined, collected, and attributed. The registry record does not provide additional detail about event severity grading, exposure-adjusted rates, or causal assessment. The counts should therefore be read as the reported number of participants with events, not as proof that either strategy is safer or more hazardous.

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

The posted primary result is a statistically clear difference in time to treatment failure under the sponsor’s Log Rank analysis. The reported hazard ratio is 8.7 for Continue Adalimumab versus Stop Adalimumab, with a 95% confidence interval from 3.6 to 21.2 and p < 0.001. The confidence interval is above the value representing equal hazards, while its width indicates uncertainty about the exact size of the relative difference.

The result concerns the trial’s specific population: patients described as having quiescent JIA-associated uveitis or chronic anterior uveitis. It also concerns the registry’s composite failure definition, which includes specified ocular recurrence findings and a joint-inflammation circumstance requiring unmasking. It does not answer every question about uveitis, arthritis, or adalimumab use, and it does not provide an estimate for populations not described in the record.

The Stop Adalimumab arm’s median time to failure is reported as 119 days, with an inter-quartile range of 84 to 243 days. A median for Continue Adalimumab is not reported in the registry record. The hazard ratio and the median are different summaries: one compares event rates over follow-up, while the other describes a point in an event-time distribution. Neither should be used to infer an unreported median or an individual participant’s outcome.

The findings do not establish why the groups differed, whether the effect was constant over time, or how the comparison varied across subgroups. Those analyses are not reported in the registry record. Nor does the posted p-value establish the probability that the result is due to chance or that the reported effect size is exact. The result is a summary of the randomized comparison as analyzed, with uncertainty and reporting limitations that remain relevant.

Safety information is descriptive: serious and other adverse events and deaths are counted by group. These counts do not, on their own, establish causation or a comprehensive balance of benefits and harms. No medical decision should be inferred from the statistical summary alone.

17. Why This Trial Matters Statistically

Statistically, this trial illustrates why time-to-event methods can be appropriate when the outcome is recurrence and the timing of recurrence matters. A simple comparison of whether failure ever occurred would discard information about when it happened and could handle incomplete follow-up poorly. The posted analysis instead uses a Log Rank comparison and reports a hazard ratio.

The result also shows why relative effect estimates should be read together with their confidence intervals. The reported hazard ratio and p-value indicate a difference in the analyzed comparison, while the interval from 3.6 to 21.2 leaves substantial uncertainty about its magnitude. A small p-value does not remove that uncertainty or provide a direct measure of clinical importance.

The endpoint definition is another important statistical feature. Treatment failure includes several forms of ocular recurrence and a joint-inflammation condition that can prompt unmasking. A composite definition can capture clinically relevant pathways to failure, but the interpretation depends on exactly which events qualify and how they are assessed. Further details about adjudication and analysis procedures are not reported in the registry record.

Finally, the record distinguishes the completed trial’s reported analysis from hypothetical planning illustrations based on the observed effect. The illustrations concern total events needed under specified assumptions; they are not evidence of the original design rationale or the trial’s observed power. Keeping those concepts separate prevents retrospective planning calculations from being mistaken for prespecified design information.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The posted outcome summary has important reporting gaps. A median time to treatment failure is available for Stop Adalimumab, but not for Continue Adalimumab. The registry record also does not provide the full event-time distributions, confidence intervals for the medians, or a detailed accounting of observed events and censoring. Those omissions limit interpretation of the time pattern behind the reported hazard ratio.

The record names a Log Rank method but does not report the model specification, proportional-hazards checks, or other diagnostics. It is therefore not possible to assess from the posted information whether a single hazard ratio adequately summarizes the group contrast over follow-up. Details on endpoint adjudication and how uncertain or borderline recurrence findings were handled are also not reported in the registry record.

Information about missing data and analysis populations is incomplete. The record notes two dropouts, one in each group, and says their observed participation time was included. It does not fully describe the censoring approach, the status of participants with ongoing follow-up, the analysis population, or sensitivity analyses. The original sample-size justification, any interim analyses, and the multiplicity plan are likewise not reported.

The registry’s brief description calls the study double-masked, while the design entry records quadruple masking. The exact masking roles and implementation are not stated. Finally, the safety summary gives counts and selected serious-event terms without further context on attribution or exposure-adjusted comparisons. These limitations do not change the posted estimate, but they restrict how fully its assumptions and broader implications can be assessed.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Adalimumab in JIA-associated Uveitis Stopping Trial — NCT03816397.
• Registry record last updated 2025-05-11; status last verified 2025-05.
• Results first posted 2025-03-26.
• 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.

22. Record Summary

The Adalimumab in JIA-associated Uveitis Stopping Trial was a Phase 4 randomized, parallel-group study, recorded as completed. Its primary question concerned time to treatment failure after continuing adalimumab or stopping it and receiving placebo in patients with quiescent JIA-associated uveitis or chronic anterior uveitis. The endpoint covered the period through 48 weeks post-randomization.

The sponsor-reported Log Rank analysis gave a hazard ratio of 8.7 for Continue Adalimumab versus Stop Adalimumab, with a 95% confidence interval from 3.6 to 21.2 and p < 0.001. The Stop Adalimumab median was 119 days, with an inter-quartile range of 84 to 243 days; the other arm’s median is not reported in the registry record. The interpretation remains limited by incomplete reporting of analysis details, censoring, and endpoint assessment.