Study Evaluating Etanercept for the Treatment of Active, Severe, and Advanced Axial Ankylosing Spondylitis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study Evaluating Etanercept for the Treatment of Active, Severe, and Advanced Axial Ankylosing Spondylitis (NCT00420238), a randomized phase 4 study. The primary outcome is Normalized Net Incremental Area Under the Curve (AUC) for the Bath Ankylosing Spondylitis Disease Activities Index (BASDAI) Between Randomization and Week 12.

2. Clinical Question

3. Trial Design

The registry describes a Phase 4, randomized, double-blind, parallel-group, placebo-controlled study of etanercept for active, severe, and advanced axial ankylosing spondylitis. The official title identifies the participants as adults and describes the study as multicentre. The listed enrollment is 82. The registry marks the study as completed, with a start date of 2007-01 and primary completion in 2009-02.

The intervention groups were etanercept and placebo. The primary outcome analysis included 39 participants in the etanercept group and 43 in the placebo group. These are the arm counts reported for that analysis, rather than a reconstruction of participant flow. The registry’s arm descriptions contain no additional treatment detail. Dose, treatment schedule, adherence, and permitted background therapies are not reported in the registry record.

The primary endpoint was measured at baseline and Weeks 2, 4, 8, and 12. The outcome was a normalized net incremental area under the curve for BASDAI, summarized as least-squares means with 95% confidence intervals. The registry specifies a modified intent-to-treat population comprising randomized subjects who received at least one dose of blinded study drug, with last observation carried forward for missing outcome values.

Secondary outcomes listed in the record include BASDAI 50 response, ASAS 20, ASAS 50, and ASAS 70 response at specified follow-up weeks, and partial remission at specified weeks. The registry lists 65 outcome measures and 91 analyses, with 64 secondary outcomes posted. The original sample-size justification, allocation ratio assumptions, randomization details, masking procedures, and trial-specific 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: Subjects completed a screening period of up to 6 weeks duration prior to entering the treatment period; 95 subjects were screened, and 82 subjects were randomized.

12 Week Double-Blind Treatment Period 1

Reasons for not completing this period:

12 Week Open-Label Treatment Period 2

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Modified Intent-To-Treat (mITT) population: includes all randomized subjects who received at least 1 dose of blinded study drug. Last observation carried forward (LOCF).

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 Normalized Net Incremental Area Under the Curve (AUC) for the Bath Ankylosing Spondylitis Disease Activities Index (BASDAI) Between Randomization and Week 12, measured over Baseline, Week 2, Week 4, Week 8, Week 12.

The primary endpoint combines repeated BASDAI measurements into a single summary of change over time from randomization to Week 12. BASDAI assesses discomfort, pain, and fatigue using a 100 millimeter visual analog scale, with 0 representing none and 100 representing very severe. The endpoint is the normalized net incremental area under the curve, calculated from baseline and measurements at Weeks 2, 4, 8, and 12.

In general, an area-under-the-curve measure can represent both the size and duration of changes across repeated assessments, rather than focusing on a single follow-up measurement. The registry says the area was calculated using the linear trapezoidal method. Areas above baseline and under the BASDAI curve are positive, while areas below baseline and above the curve are negative; these signed areas are combined and divided by study duration. A negative result therefore indicates improvement relative to baseline. Normalization expresses the accumulated change relative to the study duration, using millimeters as the stated unit.

The endpoint is useful for describing the course of a patient-reported symptom measure across the assessment period. It retains information from multiple scheduled assessments in one continuous outcome. The registry does not report why this endpoint was selected over a single-week score, nor does it describe a clinically important threshold for interpreting its magnitude.

7. Results Posted in the Registry

• Etanercept: mean -19.79 versus -10.96 on control; difference -8.83 (95% CI -16.07 to -1.59), standardized effect d = -0.53; z-test p = 0.017.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a continuous measure, so its values can vary across a range rather than falling into a small set of categories. The registry reports least-squares means and 95% confidence intervals for each group. A least-squares mean is a model-based group estimate; the registry does not state in the primary outcome description which covariates, if any, were used to calculate the arm-level estimates. The posted comparison analysis is identified separately as ANCOVA, or analysis of covariance, for the mean difference in final values. The covariates used in that analysis are not reported in the registry record.

What does the area-under-the-curve endpoint summarize?

The BASDAI endpoint uses assessments at baseline and Weeks 2, 4, 8, and 12 to summarize the pattern of values across the follow-up period. The registry specifies the linear trapezoidal method, which estimates the area between successive measured time points by treating the change between them as linear. The net incremental area combines positive and negative areas relative to baseline, then divides by study duration. The registry defines a negative value as improvement. Because this is an accumulated measure normalized by duration, it represents the average direction and extent of change across the interval rather than only the value at Week 12.

In general, repeated-measure summaries can use information from multiple visits and may reflect both the magnitude and persistence of change. They also depend on how measurements between visits are represented. The linear trapezoidal method is a calculation convention for that purpose; it does not mean that a participant’s actual symptoms necessarily changed in a straight line between assessments.

What does the reported ANCOVA comparison mean?

ANCOVA is a general statistical method for comparing group means while accounting for one or more other variables in a model. The registry identifies the reported analysis as ANCOVA and gives a mean difference in final values with a 95% confidence interval and p-value. However, the registry does not identify the model covariates or provide enough detail to reconstruct the analysis. The reported ANCOVA result should therefore be read as the sponsor’s posted model-based comparison, not as a fully reproducible specification.

The recalculated comparison is an unadjusted re-calculation from the posted summary statistics. It gives a mean difference of -8.83, a 95% confidence interval from -16.07 to -1.591, and p = 0.017. This calculation does not replace the trial’s reported ANCOVA analysis. It is useful as a check of the broad comparison using the arm-level summaries, but it does not reproduce unspecified model adjustments. The posted analysis reports the same point estimate, with a 95% confidence interval from -16.5 to -1.51 and p = 0.019.

Why are confidence intervals and p-values both reported?

In general, a confidence interval describes the uncertainty around an estimated effect, while a p-value measures how compatible the observed data are with a specified null hypothesis under the statistical model. Neither measure states the probability that the treatment is effective, nor does either establish clinical importance. The registry reports a 95% confidence interval for the group estimates and for the ANCOVA mean difference. The recalculation also reports a 95% confidence interval. The fact that the intervals for the differences do not cross zero is consistent with the reported p-values being below the conventional 0.05 level, but the registry does not state the threshold used for its own inferential decisions.

Over-dispersion is a concern mainly for outcomes whose variability exceeds what a chosen count model expects. In general, allowing for over-dispersion can prevent uncertainty from being understated when count data are more variable than a simple model assumes. The primary endpoint in this trial is continuous, and the registry does not report an over-dispersion analysis for it. This concept should not be read as a feature of the reported primary analysis.

How were missing measurements handled?

The primary outcome population is described as modified intent-to-treat, including randomized subjects who received at least one dose of blinded study drug. The registry states that last observation carried forward was used. In general, this method fills a missing later measurement with the participant’s last observed value. It relies on an assumption that the earlier value is an adequate substitute for the missing later value; that assumption may not hold if outcomes continue to change after the last observation. The registry does not report the amount or pattern of missing data, reasons for missingness, or sensitivity analyses that would show how alternative missing-data assumptions affect the primary result.

The record does not report the trial’s prespecified statistical analysis plan, the sample-size justification used when the study was designed, or details of interim analyses. It also does not report whether the primary analysis included additional adjustments or how deviations from the intended analysis were handled. These omissions limit independent assessment of the analysis choices, but they do not change the values that the registry reports.

10. Confidence Intervals and Precision

The estimated between-group mean difference was -8.83 millimeters, with the endpoint’s negative direction corresponding to greater improvement. The sponsor-reported ANCOVA analysis gave a 95% confidence interval from -16.5 to -1.51. The unadjusted recalculation gave a 95% confidence interval from -16.07 to -1.591. Both intervals lie below zero, but they span a range of possible effect magnitudes rather than identifying one exact value.

In general, a 95% confidence interval expresses the uncertainty associated with an estimate under the assumptions of the analysis. It is not a guarantee that the true effect lies within the interval, nor does it mean that 95% of participants experienced an effect in that range. The interval concerns the estimated group difference. Its width reflects the data and analysis precision; a narrower interval would indicate greater precision, all else equal.

The group-level 95% confidence intervals were -25.03 to -14.54 for etanercept and -15.95 to -5.97 for placebo. These intervals describe uncertainty in each group estimate, not the between-group comparison itself. The difference has its own confidence interval. Overlap between separate group intervals is not a substitute for examining the confidence interval or test for the difference.

The registry’s reported interval and the unadjusted recalculated interval are similar but not identical. The distinction is relevant because the sponsor reports an ANCOVA analysis, while the recalculation uses posted summary statistics without model adjustment. The registry does not provide enough information about the ANCOVA covariates to explain the difference in interval limits. The available intervals indicate uncertainty around the size of the estimated difference; they do not supply a clinical-importance threshold, which is not reported in the registry record.

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 participants per arm for 80% power and participants per arm for 90% power. They are illustrations based on the observed effect, not the trial’s original design justification. The trial’s own design assumptions are not reported in the registry record. The listed assumptions for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; each planning scenario uses a stated fraction of the observed effect.

For 100% of the observed effect, the planning figures are 57 participants per arm for 80% power and 76 participants per arm for 90% power. For 75% of the observed effect, they are 100 participants per arm for 80% power and 134 participants per arm for 90% power. For 50% of the observed effect, they are 225 participants per arm for 80% power and 301 participants per arm for 90% power. These are prospective planning illustrations based on a range of assumed effects, not counts of participants observed in the trial.

In general, 80% power means that a study designed under the stated assumptions would have an 80% chance of detecting an effect of the specified size if that effect were real. The 90% column describes a higher target chance under those assumptions. Smaller effects require more participants to detect reliably, which is why the planning figures rise as the assumed fraction of the observed effect falls.

These values are not observed power. They do not tell whether the completed trial was adequately designed, and they should not be treated as a reconstruction of its original enrollment target. The sample-size assumptions for time-to-event outcomes refer to total events, while all other planning entries refer to participants per arm; the primary outcome in this trial is continuous, so the listed planning figures are participants per arm.

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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -8.830; pooled standard deviation σ = 16.703.
2. n = 2 × (1.96 + 0.842)² × 16.703² / -8.830² = 56.2, rounded up to 57 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
delta <- -8.83                      # observed difference in means
sigma <- 16.7031057552183                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(420238)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 57 (80% power), 76 (90% power)
 75% of observed effect: n per arm = 100 (80% power), 134 (90% power)
 50% of observed effect: n per arm = 225 (80% power), 301 (90% power)
power.t.test: 57.1 
Simulated power at n = 57 per arm: 0.797 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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. Analyses Reported by the Sponsor

The registry reports the sponsor’s own primary comparison as an ANCOVA mean difference in final values of -8.83, with a 95% confidence interval from -16.5 to -1.51 and p = 0.019. The recalculated unadjusted comparison from posted summary statistics is also -8.83, with a 95% confidence interval from -16.07 to -1.591 and p = 0.017. The point estimates match, while the confidence limits and p-values differ.

The distinction is methodological: the reported result is identified as ANCOVA, whereas the recalculation is an unadjusted comparison based on group summaries. The registry does not identify the covariates or full model specification for the ANCOVA, so the reported analysis cannot be independently reproduced from the record. The recalculation is a transparency check, not a replacement for the sponsor’s analysis or a claim that its method is preferable.

The registry does not flag a discrepancy between the results. The sponsor’s posted analysis is therefore the primary reported analysis to cite, and the recalculation can be read as a separate summary-statistic comparison. Both show a negative estimated difference, consistent with the endpoint’s definition that negative values indicate improvement. Their uncertainty intervals both remain below zero, while the exact interval limits depend on the analysis used.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 65 outcome measures, 91 analyses, and 64 posted secondary outcomes. The secondary outcomes include BASDAI 50 response, ASAS 20, ASAS 50, ASAS 70, and partial remission at specified assessment weeks. The record does not report a multiplicity adjustment, testing hierarchy, or strategy for controlling the chance of false-positive findings across multiple outcomes and time points. Consequently, the primary result should be distinguished from the broader collection of secondary analyses, and the record does not support a conclusion about whether multiplicity was addressed in the trial’s full analysis plan.

The primary outcome description specifies a modified intent-to-treat population: randomized subjects who received at least one dose of blinded study drug. It also specifies last observation carried forward for missing values. The registry does not report how many measurements were missing, the reasons for missingness, or whether missingness differed between groups. It does not describe sensitivity analyses using alternative assumptions about missing outcomes.

In general, last observation carried forward treats the last measured value as a substitute for a later missing value. This can be problematic when outcomes may change after the last assessment, because the imputed value may not represent the unobserved outcome. The degree of concern depends on the amount and pattern of missing data and on the assumptions about how outcomes evolve, information not reported in the registry record.

The record does not state whether all randomized participants were included in other analyses, how protocol deviations were handled, or whether analysis populations differed by outcome. The primary outcome population definition is therefore important when interpreting the estimate: the result is not explicitly described as an analysis of every randomized participant regardless of treatment exposure. The original statistical analysis plan and the trial’s procedures for interim analyses are not reported in the registry record.

15. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hyperkalaemia (2.6%), neutropenia (2.6%), lung neoplasm malignant (2.6%), ankylosing spondylitis (2.3%).

For the reported etanercept group, the registry lists 2 participants with serious adverse events among 39 at risk. For placebo, it lists 1 participant with a serious adverse event among 43 at risk. These are the posted serious-event counts and their corresponding at-risk group sizes for those arms. No event-rate calculation is added here, and the counts alone do not describe severity, timing, duration, or whether an event was considered related to treatment.

The registry also lists later group labels, Etanercept/Etanercept and Placebo/Etanercept, with 0 of 38 and 1 of 39 participants with serious adverse events, respectively. The record excerpt does not explain the treatment periods represented by these labels, so they should not be combined with or treated as equivalent to the initial randomized-arm comparison. Death counts are not reported in the registry record.

The listed serious adverse event terms include Hyperkalaemia, Neutropenia, and Lung neoplasm malignant at 2.6% each, and Ankylosing spondylitis at 2.3%. The registry does not provide context in this summary for how these terms map to the treatment periods or the full adverse-event coding. Serious adverse event counts are a safety description, not a complete assessment of harms. The record does not report enough detail to assess event timing, exposure-adjusted rates, causality, or comparative risk.

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

The posted results show a more negative mean normalized net incremental BASDAI area under the curve for etanercept than for placebo over the baseline-to-Week-12 assessment period. Since negative values are defined as improvement, the estimated difference of -8.83 millimeters favors a greater average reduction in BASDAI for the etanercept group on this endpoint. The sponsor-reported ANCOVA analysis and the unadjusted recalculation both give a confidence interval for the difference that lies below zero.

This is evidence about a particular outcome, population, and follow-up interval. It does not establish the magnitude of benefit for every participant, nor does it show that the same result applies to other patient populations, other treatment settings, or longer durations. The registry does not report a clinical-importance threshold for the endpoint, so the statistical difference cannot by itself determine whether the estimated size is meaningful to patients.

The endpoint aggregates multiple assessments and uses a specified calculation method, but it is not a direct summary of every aspect of ankylosing spondylitis. The listed secondary outcomes cover additional response definitions and assessment times, but their results are not included in the registry record for this analysis. No conclusions about those outcomes should be inferred from the primary endpoint result.

The registry’s safety figures report serious adverse events and selected event terms, but they do not provide the detail needed for a comprehensive comparison of harms. The absence of a reported death count is not evidence that no deaths occurred; death counts are not reported in the registry record. The trial result should therefore be read as a focused statistical comparison, not as a complete account of all benefits and risks.

Finally, the primary analysis used a modified intent-to-treat population and last observation carried forward. Missing-data quantity and sensitivity analyses are not reported. These features are relevant to how robustly the estimated difference represents the randomized groups, and they limit what can be concluded from the registry record alone.

17. Why This Trial Matters Statistically

This trial illustrates why the definition and direction of a continuous endpoint matter. A negative normalized net incremental BASDAI area is defined as improvement, so interpreting the sign of the mean difference requires understanding the endpoint construction. The comparison is not simply a difference in a single visit score: it summarizes repeated measurements from baseline through Week 12 using a specified area-under-the-curve method.

It also demonstrates the distinction between a model-based reported analysis and an unadjusted recalculation from group summaries. The reported ANCOVA analysis and the recalculated comparison share the same point estimate but have different confidence limits and p-values. Reading both makes clear that the estimated direction is similar while the details of uncertainty depend on the analysis method. The record does not provide the covariates needed to reconstruct the ANCOVA.

The result also shows why statistical evidence and clinical importance are separate questions. The reported difference is estimated with a confidence interval that lies below zero, but the registry does not state a threshold for a clinically important BASDAI difference. Precision, direction, and the clinical meaning of the magnitude therefore require separate consideration.

Finally, this record brings population definition, missing-data handling, multiplicity, and safety reporting into view. The modified intent-to-treat definition and last-observation-carried-forward method shape interpretation, while the absence of information on missingness, multiplicity procedures, and parts of the analysis plan limits assessment of robustness. The value of the example is methodological: it encourages careful reading of what an endpoint, model, and registry entry do—and do not—establish.

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.

Several details needed to fully evaluate the trial are not reported in the registry record. The original sample-size justification and trial-specific design assumptions are absent, so the planning illustrations based on the observed effect should not be mistaken for the enrollment rationale used by the investigators. The full statistical analysis plan, interim-analysis procedures, and any prespecified multiplicity strategy are also not reported.

The registry identifies the reported comparison as ANCOVA but does not state the covariates or full model specification. This limits independent reproduction of the sponsor’s analysis and prevents a direct explanation of the differences between its confidence interval and the unadjusted recalculation. The record also does not report whether additional analyses were conducted for robustness.

The primary analysis used a modified intent-to-treat population and last observation carried forward. The number of missing measurements, reasons for missingness, and patterns by group are not reported. Without those details or sensitivity analyses, the effect of missing-data assumptions on the estimate cannot be assessed from the registry record.

The endpoint summarizes BASDAI over the baseline-to-Week-12 interval, and no longer-term primary result is included in the registry record. A clinical-importance threshold for this particular endpoint is not reported. Arm dosing and treatment schedule are also absent, and the registry’s brief arm descriptions do not fill in those details.

Safety information is limited to posted serious-event counts, selected terms, and later group labels whose treatment-period context is not described. Death counts are not reported in the registry record, and the summary does not provide the detail needed to evaluate event timing, relatedness, or exposure-adjusted comparisons. These limitations constrain interpretation of both effectiveness and safety from the registry results alone.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Study Evaluating Etanercept for the Treatment of Active, Severe, and Advanced Axial Ankylosing Spondylitis — NCT00420238.
• Registry record last updated 2010-06-29; status last verified 2010-06.
• Results first posted 2010-06-15.
• 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

NCT00420238 is a completed Phase 4 randomized, double-blind, parallel-group, placebo-controlled study of etanercept in adults with active, severe, and advanced axial ankylosing spondylitis. The registry lists enrollment of 82, a start date of 2007-01, and primary completion in 2009-02.

The primary endpoint was normalized net incremental BASDAI area under the curve from randomization to Week 12. The reported ANCOVA mean difference in final values was -8.83, with a 95% confidence interval from -16.5 to -1.51 and p = 0.019. The unadjusted recalculation gave the same estimate, with a 95% confidence interval from -16.07 to -1.591 and p = 0.017.

The primary analysis used a modified intent-to-treat population and last observation carried forward. The registry also lists secondary response outcomes and reports serious adverse event counts, but important details about missing data, multiplicity, model specification, and safety context are not reported in the registry record.