Study Of Celecoxib Or Diclofenac For Efficacy and Safety In Chinese Patients With Ankylosing Spondylitis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study Of Celecoxib Or Diclofenac For Efficacy and Safety In Chinese Patients With Ankylosing Spondylitis (NCT00762463), a randomized phase 3 study. The primary outcome is Change From Baseline in Participant's Assessment of Global Pain Intensity at Week 6.

2. Clinical Question

The registry describes a randomized, double-blind, parallel-group study in Chinese patients with ankylosing spondylitis. Its primary endpoint was a participant assessment of overall pain intensity in the previous 48 hours. The clinical question is therefore specific to the reported pain measure, time point, and analysis population; it does not by itself address every symptom, later outcome, or safety question examined during the study.

3. Trial Design

The registry describes a Phase 3, randomized, double-blind, parallel-group study with an active comparator. It lists enrollment as 240 and the study status as completed. The initial double-blind comparison lasted 6 weeks and evaluated celecoxib 200 mg QD against diclofenac SR 75 mg QD. The study title also describes a 6-week extension phase: participants could receive celecoxib 400 mg QD or maintain double-blind phase therapy.

The primary outcome was measured at Week 6. The registry’s primary analysis population was per protocol rather than all randomized participants: participants had to receive at least one dose, have a Week 6 global pain intensity assessment, and have no major protocol deviations. The posted primary outcome counts were 111 for celecoxib and 108 for diclofenac SR. These counts describe the primary per-protocol analysis, not the registry’s overall enrollment.

The registry also lists a range of secondary measures, including participant and physician global assessments, functional measures, laboratory measures, and concomitant paracetamol use. It reports outcomes at several time points, including Weeks 2, 4, 6, and 12. The extension-phase treatment options and later measurements mean that not every result necessarily refers to the same treatment period or analysis population; the time frame and group labels matter when interpreting each result.

The registry record does not report the original sample-size justification, the details of the randomization procedure, the allocation concealment method, or the specific procedures used to maintain masking. The number of participating centers is not reported in the registry record. The record gives study start and primary completion as 2009-07 and 2010-06, respectively, but those dates do not explain the analysis plan or establish how missing observations were handled.

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.

Double-blind Phase (Baseline to Week 6)

Reasons for not completing this period:

Enrollment in Extension Phase (Week 6)

Reasons for not completing this period:

Extension Phase (Week 6 to Week 12)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Per-Protocol (PP): All randomized participants who received at least one dose of study medication, and had global pain intensity assessment at Week 6 and no major protocol deviations.

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 Change From Baseline in Participant's Assessment of Global Pain Intensity at Week 6, measured over Baseline, Week 6.

The primary endpoint was change from baseline in participant-assessed global pain intensity at Week 6. Participants rated overall pain intensity during the previous 48 hours in response to a question about their global pain intensity. The measure was a 100-mm visual analog scale, ranging from 0 for no pain to 100 for the worst pain. A change less than 0 indicated improvement, so more negative changes represent larger decreases in the recorded pain score.

A participant-reported global measure captures the participant’s overall assessment rather than a clinician’s rating or a laboratory value. That makes it relevant to a question about perceived pain, while also limiting the conclusion to this particular experience, scale, and reporting period. It does not directly measure every aspect of ankylosing spondylitis or establish how participants functioned in other contexts.

The registry reports least-squares means for the primary outcome and identifies the standard error as the dispersion measure. The primary outcome population was per protocol: randomized participants who received at least one dose, had a Week 6 assessment, and had no major protocol deviations. The posted group means were changes from baseline, not the participants’ Week 6 scores alone. The registry record does not report additional details about the scale’s administration beyond the stated question and range.

7. Results Posted in the Registry

• Diclofenac SR 75 mg: mean -27.10 versus -23.80 on control; difference -3.30 (95% CI -8.84 to 2.24), standardized effect d = -0.16; z-test p = 0.243.

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.

Change From Baseline in Erythrocyte Sedimentation Rate (ESR) at Week 6

• 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 errors.
• Result:
  • Diclofenac SR 75 mg: mean -1.80 versus -2.50 on control; difference 0.70 (95% CI -2.90 to 4.30), standardized effect d = 0.05; z-test p = 0.703.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Diclofenac SR 75 mg and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in C-Reactive Protein (CRP) at Week 6

• 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 errors.
• Result:
  • Diclofenac SR 75 mg: mean -2.72 versus -4.16 on control; difference 1.44 (95% CI -2.98 to 5.86), standardized effect d = 0.09; z-test p = 0.523.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Diclofenac SR 75 mg and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With Concomitant Use of Paracetamol

• 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:
  • Diclofenac SR 75 mg: 0.0% (0/120) versus 2.5% (3/120) on control; risk difference -2.5 percentage points (95% CI -5.3 to 0.3); unadjusted two-proportion z-test p = 0.081.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Diclofenac SR 75 mg and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover several kinds of measures, including inflammatory laboratory values, participant and physician assessments, a functional index, and concomitant paracetamol use. The posted secondary results span different time points and outcome scales, so they address related but nonidentical questions. Their pattern does not provide a clear, consistent separation between the treatment groups in the simple recalculations.

For the secondary comparisons with recalculated estimates, the confidence intervals span the null value, and their unadjusted p values do not establish a clear difference. This broad pattern is useful context for the primary result, but it does not resolve the discrepancy between the sponsor’s primary analysis and the simple recalculation. Nor does it establish that the treatments have identical effects across all outcomes.

Secondary results generally carry less confirmatory weight when many endpoints or time points are examined without a reported multiplicity strategy. Each additional comparison creates another opportunity for a chance finding, and a collection of separate tests does not automatically constitute a single coherent test of treatment effect. The registry record does not report whether a multiplicity adjustment or hierarchical testing plan was used. Accordingly, the unadjusted secondary comparisons are best treated as exploratory context rather than independent confirmation.

The registry lists 24 secondary outcomes as posted. Results should be interpreted according to each outcome’s population and time frame; the record does not establish that every endpoint used the same participants or analysis approach. A broad pattern across secondary measures cannot substitute for a prespecified primary analysis or for details about how missing observations and multiple comparisons were handled.

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

• Change From Baseline in Participant's Assessment of Global Pain Intensity at Weeks 2 and 4
• Change From Baseline in Participant's Assessment of Global Pain Intensity at Week 12
• Change From Baseline in Participant's Global Assessment of Disease Activity at Weeks 2, 4, and 6
• Change From Baseline in Participant's Global Assessment of Disease Activity at Week 12
• Change From Baseline in Physician's Global Assessment of Disease Activity at Weeks 2, 4, and 6
• Change From Baseline in Physician's Global Assessment of Disease Activity at Week 12
• Change From Baseline in Bath Ankylosing Spondylitis Functional Index (BASFI) at Weeks 2, 4, and 6
• Change From Baseline in BASFI at Week 12

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: it is a change in a score measured in millimeters. The registry reports least-squares means and standard errors for each group. A least-squares mean is an estimate from a statistical model, rather than necessarily the raw arithmetic average. The registry identifies ANCOVA as the method for its reported primary between-group analysis, but the covariates and other model details are not reported in the registry record.

What does the group contrast represent?

A difference between group means expresses how far apart the estimated changes are, using the stated group order and units. The simple recalculation uses the posted group summaries and expresses the diclofenac SR result relative to celecoxib: -3.3 mm. Because negative change indicates improvement, a negative contrast in that order means the posted change estimate was more negative for diclofenac SR. This is a description of the calculated contrast, not a conclusion that one treatment is better.

The registry’s ANCOVA result is expressed as an LS mean difference of 3.3. The reported result belongs to the registry’s own model and is labeled as a noninferiority-or-equivalence analysis. Its contrast orientation and model adjustment need not match a simple unadjusted calculation from group summaries. The record does not provide enough details to reconstruct the model or explain the difference between the reported and recalculated uncertainty. Since the discrepancy flag is true, the sponsor’s analysis takes precedence over the recalculation.

What do standard errors and confidence intervals tell us?

A standard error describes the uncertainty around an estimated mean under the statistical model. It is not the same as the spread of individual participant scores. The registry reports standard errors of 1.98 for celecoxib and 2.02 for diclofenac SR on the primary outcome. The simple recalculation uses these summary statistics to estimate the uncertainty of a between-group difference.

A confidence interval gives a range of values compatible with the estimate under the method used. It helps distinguish a comparatively precise estimate from one that remains compatible with a wider range of differences. The recalculated interval for the primary contrast spans zero, as does the registry’s reported 95% interval. However, the two intervals and their p values are not interchangeable: one comes from a simple recalculation and the other from the sponsor’s ANCOVA analysis.

Why can the two analyses disagree?

An ANCOVA can account for model terms, including baseline values, while a simple comparison based on posted summary estimates does not reproduce all such adjustments. The exact model specification, covariates, missing-data procedures, and analysis details are not reported in the registry record. Differences in contrast ordering can reverse the sign, but not by themselves explain different interval widths or p values. The registry’s own analysis therefore remains the relevant reported result, while the recalculation is a diagnostic comparison of the posted summaries.

The primary analysis population is also important. It is defined as per protocol and excludes randomized participants who did not meet the stated criteria. This means the outcome estimate pertains to that defined population. The registry record does not report a corresponding intention-to-treat primary estimate or describe whether sensitivity analyses addressed departures from the per-protocol set.

How should the p values be read?

A p value describes how compatible the data are with a specified null model under the analysis used. It does not quantify the probability that a treatment works, the probability that the null hypothesis is true, or the clinical importance of an estimate. The recalculated p value of 0.243 belongs to an unadjusted comparison and does not encode the noninferiority question. The sponsor’s p value of 0.009 belongs to the reported legacy noninferiority-or-equivalence analysis. Without the noninferiority margin and full model details, that p value cannot be interpreted as a standard superiority finding.

General statistical context: a noninferiority analysis asks whether the treatment contrast is compatible with a prespecified amount of inferiority no greater than an accepted margin. The margin and its rationale are central to that interpretation. Neither is reported in the registry record, so the posted label and p value alone do not permit an independent evaluation of the criterion.

Why do secondary comparisons need caution?

The record includes multiple secondary outcomes and time points. When many questions are tested, the chance of seeing an apparently notable result by chance can rise if no adjustment is made. The registry record does not report a multiplicity adjustment for the posted secondary comparisons. The recalculated secondary p values are unadjusted and should not be read as confirmatory evidence. The outcome measures also differ in meaning and scale, so their estimates should not be collapsed into a single overall treatment effect.

General statistical context: over-dispersion is a concern when count or binary data vary more than a model assumes, because standard errors may then be too small. The primary outcome is continuous and is analyzed through mean estimates, so this issue is not the central concern for its reported summary. For the binary paracetamol-use outcome, the registry record does not provide a model-based analysis or enough information to assess alternative variance assumptions. The simple comparison should therefore be read as a descriptive, unadjusted calculation.

11. Confidence Intervals and Precision

The primary group estimates are -23.8 mm for celecoxib and -27.1 mm for diclofenac SR. The recalculated between-group difference, expressed as diclofenac SR relative to celecoxib, is -3.3 mm, with an interval from -8.844 to 2.244. Because the interval includes zero, the summaries are compatible with no between-group difference as well as with differences in either direction within the stated range. The width of the interval matters: the point estimate alone does not show how much uncertainty remains.

The registry’s own ANCOVA result reports an LS mean difference of 3.3 and a 95% confidence interval from -2.2 to 8.8. This interval also spans zero, but it comes from the sponsor’s analysis and is associated with a noninferiority-or-equivalence label. Its interpretation depends on the analysis specification and the noninferiority margin, neither of which is fully reported. The sponsor’s analysis takes precedence over the simple recalculation, given the flagged discrepancy.

Confidence intervals describe uncertainty under their analysis assumptions; they do not guarantee that the true effect lies within the interval or show the range of individual participant responses. Nor does an interval spanning zero establish equivalence. Equivalence and noninferiority require a prespecified margin and a comparison with that margin. Since the margin is not reported in the registry record, the reported interval cannot independently establish whether the noninferiority criterion was met.

Precision is also specific to the analyzed population and endpoint. The primary result uses the per-protocol population, and the registry record does not report how missing outcome data were handled. The estimates should therefore be read as results from the posted analysis rather than as a complete account of all randomized participants or of treatment effects at other time points.

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 estimates are participants per arm for 80% power and participants per arm for 90% power. They are not total enrollment figures. The calculation uses the observed effect as a reference and also considers planning for 75% and 50% of that observed effect. For the primary comparison, the displayed estimates range from 632 participants per arm for planning for 100% of the observed effect at 80% power to 3,381 participants per arm for planning for 50% of the observed effect at 90% power.

The listed assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The target effect is the stated fraction of the observed effect. These are planning illustrations based on the posted result, not the trial’s own original design assumptions; those original assumptions are not reported in the registry record.

The 80% power column means the planned study would have an 80% chance of detecting the specified effect under the calculation’s assumptions if that effect were the true effect. The 90% power column expresses the analogous target at a higher probability of detection. They are prospective design quantities, not measures of what this completed trial achieved. This is not observed power.

Smaller target effects require more participants for the same stated power and significance level. That pattern is expected because a smaller difference is harder to distinguish from random variation. These estimates should not be treated as recommendations for a new study without considering its endpoint, design, analysis, and scientifically relevant effect size.

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

Worked calculation for this trial

1. Observed difference Δ = -3.300; pooled standard deviation σ = 20.926.
2. n = 2 × (1.96 + 0.842)² × 20.926² / -3.300² = 631.2, rounded up to 632 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 <- -3.3                      # observed difference in means
sigma <- 20.9257174694896                     # 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(762463)
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 = 632 (80% power), 846 (90% power)
 75% of observed effect: n per arm = 1,123 (80% power), 1,503 (90% power)
 50% of observed effect: n per arm = 2,525 (80% power), 3,381 (90% power)
power.t.test: 632.2 
Simulated power at n = 632 per arm: 0.799 (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.

14. Analyses Reported by the Sponsor

The registry reports an ANCOVA LS mean difference of 3.3, with a 95% confidence interval from -2.2 to 8.8 and p = 0.009. It labels this as a legacy noninferiority-or-equivalence analysis. The simple recalculation from the posted group summaries instead gives a difference of -3.3 for diclofenac SR relative to celecoxib, an interval from -8.844 to 2.244, and an unadjusted p value of 0.243.

The sign difference is consistent with the analyses using opposite group orders. However, their intervals and p values also differ, so reversing the contrast alone does not reconcile the results. ANCOVA may use model adjustments that are not reproduced by a simple comparison of posted summaries, but the registry record does not report enough detail to identify the reason for the discrepancy or reproduce the sponsor’s model.

Because the discrepancy flag is true, the sponsor’s own analysis takes precedence. Its p value should be read in the context of the stated noninferiority-or-equivalence analysis, not as an ordinary test of superiority. The noninferiority margin, the covariate specification, and fuller details of the model are not reported in the registry record. Without those details, the sponsor’s result cannot be independently evaluated against its intended decision rule from the registry entry alone.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple outcomes, including several secondary assessments and repeated time points. Testing many outcomes can increase the chance of at least one result appearing notable by chance. The registry record does not report whether the analyses used a family-wise error adjustment, a hierarchical testing strategy, or another multiplicity plan. Consequently, unadjusted secondary p values should be interpreted cautiously and should not be treated as though they came from a single prespecified test.

The primary outcome population is defined as per protocol: randomized participants who received at least one dose, had a Week 6 assessment, and had no major protocol deviations. The posted primary outcome counts are 111 for celecoxib and 108 for diclofenac SR. The registry’s overall enrollment is 240, but the record does not provide a complete account of how every randomized participant contributed to every outcome analysis. An intention-to-treat primary result is not reported in the registry record.

Missing-data handling is not reported in the registry record. The record does not say whether missing outcome measurements were imputed, whether participants with missing assessments were excluded, or whether sensitivity analyses assessed the impact of missingness. The per-protocol definition requires a Week 6 assessment, but it does not by itself describe the treatment of missing data in other analyses.

The record also does not report the original sample-size justification, interim analyses, or a complete statistical analysis plan. The sponsor’s primary result is identified as ANCOVA, but the covariates and other model details are not reported. These gaps limit independent assessment of prespecification, multiplicity control, analysis-population choices, and the robustness of the reported findings.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

For the initial celecoxib 200 mg group, the registry records 0 serious adverse events among 120 participants at risk. For the initial diclofenac SR 75 mg group, it also records 0 serious adverse events among 120 participants at risk. The extension-phase groups listed in the safety results each have 0 serious adverse events among their reported participants at risk.

These figures describe the serious adverse events recorded in the listed groups and should not be read as proof that no serious event could occur. The record does not list serious adverse event terms. Death counts are shown as unavailable rather than as zero, so deaths are not reported in the registry record.

The registry separately reports other adverse events for the initial groups: 11 for celecoxib and 9 for diclofenac SR. These are not serious adverse event counts and should not be combined with the serious-event figures. The listed extension groups have their own treatment histories and at-risk counts; those figures should be read as separate entries rather than as a single continuous comparison. The registry record does not provide additional detail about event timing, duration, or relatedness to treatment.

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

The posted primary group estimates show decreases from baseline in participant-assessed global pain intensity in both groups at Week 6. The sponsor’s ANCOVA result is the registry’s own reported analysis and is labeled as a noninferiority-or-equivalence analysis. A simple unadjusted recalculation gives a different contrast and p value, and the registry flags a discrepancy. The sponsor’s analysis therefore takes precedence, but the registry record does not provide the margin and full model details needed to independently assess its noninferiority interpretation.

The result does not establish superiority of one treatment over the other. A p value from a noninferiority-or-equivalence analysis is not automatically a superiority test, and the reported confidence interval cannot establish equivalence without a stated margin. The simple recalculation also does not settle the question: it is an unadjusted analysis of posted summary statistics and has an interval spanning no difference.

The finding concerns a specific endpoint, a 6-week time point, and a per-protocol population. It does not by itself establish what happened for participants excluded from that analysis, what the longer-term experience was, or how the treatments compare on every secondary outcome. The secondary estimates are unadjusted and subject to the limitations of multiple comparisons.

The safety record lists zero serious adverse events in the reported groups, but death counts are not reported in the registry record. These entries do not establish the absence of all possible harms. Taken together, the registry supports a description of the posted estimates and their uncertainty, but missing information about analysis details, missing data, and multiplicity limits broader conclusions.

18. Why This Trial Matters Statistically

This trial illustrates why the analysis population and the contrast definition matter. The primary results are based on a per-protocol population, while the registry also gives overall enrollment and safety counts. Those numbers refer to different groups of participants and should not be treated as interchangeable. A reader needs the endpoint, time point, and population alongside the estimate to understand what it represents.

It also demonstrates the distinction between a sponsor’s model-based analysis and a simple recalculation from posted summaries. The reported ANCOVA result is explicitly labeled as a noninferiority-or-equivalence analysis, while the recalculation is an unadjusted comparison. Their estimates and p values differ, and the registry does not report enough detail to reproduce the sponsor’s model or evaluate its noninferiority margin. The discrepancy is a reason to keep the two analyses clearly distinguished, not to substitute one for the other.

Finally, multiple secondary outcomes and time points make multiplicity and prespecification important. Without a reported testing strategy, isolated unadjusted findings carry limited confirmatory weight. The registry record also leaves important questions about missing data and analysis details unanswered. These are practical statistical issues: they shape how confidently results can be interpreted, even when group estimates and intervals are available.

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.

The registry record leaves the original sample-size justification unreported. The planning illustrations based on the observed effect do not show what assumptions guided the trial’s original design. The record also does not report an interim-analysis plan, a complete statistical analysis plan, or the randomization and allocation concealment procedures.

The primary endpoint is analyzed in a per-protocol population, and an intention-to-treat primary estimate is not reported in the registry record. The record does not explain how missing assessments were handled or whether sensitivity analyses assessed their potential influence. The primary ANCOVA covariates and other model specifications are also not reported.

The reported primary analysis is labeled as noninferiority-or-equivalence, but the margin and its rationale are not reported in the registry record. That limits independent evaluation of the sponsor’s conclusion. The simple recalculation differs from the sponsor analysis and cannot replace it, particularly because the posted summary statistics do not reproduce the full model.

Secondary results involve multiple endpoints and time points, while multiplicity control is not reported. The secondary comparisons are therefore less secure as confirmatory evidence. The safety record lists serious-event counts but does not report death counts, serious event terms, or additional event details such as timing and relatedness. These reporting limitations restrict interpretation of both the efficacy estimates and the safety information.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study Of Celecoxib Or Diclofenac For Efficacy and Safety In Chinese Patients With Ankylosing Spondylitis — NCT00762463.
• Registry record last updated 2021-02-02; status last verified 2021-01.
• Results first posted 2011-08-23.
• 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

The registry describes a completed Phase 3 randomized, double-blind, parallel-group study of celecoxib 200 mg QD versus diclofenac SR 75 mg QD in Chinese patients with ankylosing spondylitis. Enrollment is listed as 240. The initial comparison lasted 6 weeks, followed by a 6-week extension phase with the treatment options described in the study title.

The primary endpoint was change from baseline in participant-assessed global pain intensity at Week 6, measured on a 100-mm visual analog scale. The primary outcome analysis used a per-protocol population. The registry reports an ANCOVA result labeled as noninferiority-or-equivalence and a separate simple recalculation differs from that result; the sponsor’s analysis takes precedence, while the margin and full model details are not reported in the registry record.

The record includes multiple secondary outcomes and reports zero serious adverse events in the listed groups. Death counts are not reported in the registry record. Missing-data handling, multiplicity procedures, and the original sample-size justification are also not reported.