Roflumilast in Chronic Obstructive Pulmonary Disease (COPD) Patients Treated With Fixed Dose Combinations of Long-acting β2-agonist (LABA) and Inhaled Corticosteroid (ICS): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Roflumilast in Chronic Obstructive Pulmonary Disease (COPD) Patients Treated With Fixed Dose Combinations of Long-acting β2-agonist (LABA) and Inhaled Corticosteroid (ICS) (NCT01443845), a randomized phase 4 study. The primary outcome is Rate of Moderate or Severe COPD Exacerbations Per Patient Per Year..

2. Clinical Question

The question concerns an added treatment and an event rate, not whether the background LABA/ICS combination differs between groups. Moderate exacerbations were defined as those requiring oral or parenteral glucocorticosteroids. Severe exacerbations were defined as those requiring hospitalization and/or leading to death. Both categories contributed to the primary outcome during the double-blind treatment period.

The registry record describes the trial’s aim as demonstrating an additional benefit of roflumilast when added to fixed-dose LABA/ICS in reducing exacerbations. The statistical interpretation is narrower: it considers the observed rates, the sponsor-reported comparison, and the uncertainty around that comparison. The reported results do not, by themselves, establish that the groups differed, nor do they settle questions about outcomes outside the registry’s stated measures and time frame.

3. Trial Design

This was a phase 4, randomized, triple-masked, parallel-group trial with 2,354 participants enrolled. The study is listed as completed. The registry record gives a start date of 2011-09-30 and a primary completion date of 2015-11-30.

Participants were assigned to roflumilast or placebo. The study title specifies that participants were being treated with a fixed-dose combination of LABA and ICS. The official title identifies roflumilast 500 μg. The record describes the target population as people with severe to very severe COPD. It does not report further detail about the randomization procedure, allocation concealment, masking procedures, or the characteristics of participants at baseline.

The primary outcome was an exacerbation rate per patient per year from baseline to Week 52. The primary population is identified as the Intent-to-Treat Study Population. The registry reports 1,174 participants in the placebo group and 1,178 in the roflumilast group for this outcome. The numbers in the outcome groups need not be treated as a complete description of every analysis population: for example, the continuous FEV1 outcome reports different group counts.

The registry also lists outcomes for severe exacerbations, exacerbations treated with antibiotics, and change in predose forced expiratory volume in 1 second (FEV1). The record includes these as secondary outcomes, but the results of those outcomes do not alter the definition of the primary endpoint. Details of the trial’s original sample-size justification, planned interim analyses, and full statistical analysis plan are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: The study consisted of 2 weeks of single-blind placebo lead-in with fixed-dose combination (FDC) long-acting β2-agonist(s) (LABA)/inhaled corticosteroid(s) (ICS) treatment followed by 52 weeks of double-blind treatment in addition to maintenance FDC LABA/ICS.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat Study Population

5. Baseline Characteristics

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

Population: The demographic information provided here is for the safety population. The safety population (numbering 2352 patients) of the trial does not include 2 patients from the enrolled population (numbering 2354 patients), due to a protocol deviation for one patient and an adverse event for a second patient, both in the placebo arm

6. Primary Endpoint

The registry lists the primary outcome as Rate of Moderate or Severe COPD Exacerbations Per Patient Per Year., measured over Baseline to Week 52.

The primary endpoint is the rate of moderate or severe COPD exacerbations per patient per year during the double-blind treatment period, from baseline to Week 52. A moderate exacerbation was defined as one requiring oral or parenteral glucocorticosteroids. A severe exacerbation was defined as one requiring hospitalization and/or leading to death. The endpoint therefore combines events of different levels of severity into one rate measure.

An event rate expresses the frequency of events in relation to participant time at risk. It is useful when the question concerns repeated events, because one person may experience more than one exacerbation during follow-up. The registry gives the outcome unit as COPD exacerbations per patient per year. The reported group rates are 1.27 for placebo and 1.17 for roflumilast, with 95% confidence intervals of 1.17 to 1.39 and 1.06 to 1.28, respectively.

For a comparative analysis, the rate ratio relates the event rate in one group to the rate in the other. A rate ratio of 1 represents equal rates; values on either side represent different relative rates. General statistical interpretation also depends on the confidence interval, which describes the uncertainty compatible with the analysis. Since events can recur and event counts may vary more than a simple Poisson model allows, count-rate analyses often use methods that accommodate over-dispersion. The sponsor reports using negative binomial regression for its primary comparison.

This endpoint captures exacerbation frequency over the stated period. It does not describe every aspect of COPD, nor does it independently summarize other outcomes such as lung function, hospitalization, or safety.

7. Results Posted in the Registry

• Roflumilast: 1.17 versus 1.27 (COPD exacerbations per patient per year) on control; rate ratio 0.92 (95% CI 0.81–1.05, derived from the posted confidence limits); p ≈ 0.208.

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.

Rate of COPD Exacerbations That Led to Hospitalization or Death (ie, Severe COPD Exacerbations)

• Endpoint type: Event rate: events per unit of follow-up, compared as a rate ratio.
• Test: Poisson rate ratio, with the standard error taken from the posted confidence limits.
• Result:
  • Roflumilast: 0.28 versus 0.29 (COPD exacerbations per patient per year) on control; rate ratio 0.97 (95% CI 0.76–1.22, derived from the posted confidence limits); p ≈ 0.772.
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Roflumilast and the control. The interval gives the range of effects that remain compatible with the data.

Rate of Moderate or Severe COPD Exacerbations or COPD Exacerbations Treated With Antibiotics

• Endpoint type: Event rate: events per unit of follow-up, compared as a rate ratio.
• Test: Poisson rate ratio, with the standard error taken from the posted confidence limits.
• Result:
  • Roflumilast: 1.31 versus 1.45 (COPD Exacerbations per Patient per Year) on control; rate ratio 0.90 (95% CI 0.80–1.02, derived from the posted confidence limits); p ≈ 0.092.
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Roflumilast and the control. The interval gives the range of effects that remain compatible with the data.

Mean Change in Predose Forced Expiratory Volume in 1 Second (FEV1)

• 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:
  • Roflumilast: mean 0.04 versus -0.01 on control; difference 0.05 (95% CI 0.04 to 0.07), standardized effect d = 0.34; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Roflumilast than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes cover severe exacerbations, exacerbations that were moderate or severe or treated with antibiotics, and mean change in predose FEV1. Across these outcomes, the pattern is not uniform: the rate comparisons do not all show the same statistical evidence, while the FEV1 comparison reports a clearer difference in the posted analysis. This variation reinforces that the secondary endpoints address distinct questions rather than multiple interchangeable measurements of one result.

The primary endpoint remains the main reference point for the trial’s stated question. Secondary findings can add context about event severity, antibiotic-treated episodes, or lung function, but they do not replace the primary comparison. A secondary result should not be used alone to recast the primary outcome when the primary analysis does not show clear statistical evidence of a difference.

Unadjusted secondary p-values carry less weight when several outcomes are examined. When many comparisons are made, the chance of at least one apparently notable result can increase, even if there are no true differences. Statistical approaches to multiplicity can address this issue, but the registry record does not report whether such an approach was planned or applied. The secondary results should therefore be interpreted as reported findings with that uncertainty in mind, not as a definitive set of independent confirmatory tests.

The registry lists three secondary outcomes and posts results for those outcomes. It does not report enough detail to establish the hierarchy among secondary analyses, the complete family of analyses considered, or whether all analyses were prespecified. Those limitations make it difficult to assign confirmatory status to any one secondary result.

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

• Rate of COPD Exacerbations That Led to Hospitalization or Death (ie, Severe COPD Exacerbations)
• Rate of Moderate or Severe COPD Exacerbations or COPD Exacerbations Treated With Antibiotics
• Mean Change in Predose Forced Expiratory Volume in 1 Second (FEV1)

9. Statistical Interpretation

10. Statistical Methods Explained

The trial compares two randomized groups on an event-rate outcome. The primary endpoint counts moderate or severe COPD exacerbations during the double-blind period and expresses their frequency per patient per year. Because an individual can experience repeated exacerbations, the analysis needs to account for event counts rather than treating each participant as having only one possible event.

What does a rate ratio express?

A rate ratio compares the event rate in one group with the event rate in the other. A value of 1 corresponds to equal rates. A value below 1 indicates that the numerator group’s estimated rate is lower relative to the comparison group; a value above 1 indicates a higher estimated rate. The ratio is a relative measure and should be read alongside the underlying group rates. The registry reports rates of 1.27 for placebo and 1.17 for roflumilast, both in COPD exacerbations per patient per year. The sponsor reports a rate ratio of 0.92 for roflumilast compared with placebo.

Why use negative binomial regression?

The sponsor reports negative binomial regression for the primary rate comparison. This is a model for count outcomes and can accommodate over-dispersion: general statistical terminology for count variability greater than a basic Poisson model would expect. If over-dispersion is present but not adequately represented, uncertainty estimates from a simpler model may be too small. The registry record does not report the model’s covariates, the specific dispersion estimates, or other implementation details.

The simple re-calculation is different in purpose and assumptions. It uses posted summary statistics and a Poisson-based approximation; it is not a reconstruction of the sponsor’s regression. Its unadjusted p-value is 0.208, whereas the sponsor reports p = 0.163. This difference illustrates why a summary-statistic calculation and a model-based analysis may not agree exactly. The record marks a discrepancy, so the sponsor’s analysis takes precedence.

How should the confidence interval and p-value be read?

The sponsor’s 95% confidence interval for the primary rate ratio is 0.81 to 1.04. In general, a confidence interval gives a range of values compatible with the observed data and the assumptions of the analysis at the stated confidence level. It is not a guarantee that the true value lies in that particular interval, and it does not give the probability that a specific value is true. Since the interval includes 1, it includes equal rates as well as values on both sides of equality.

The p-value of 0.163 is calculated under a null hypothesis of no difference, together with the assumptions of the sponsor’s analysis. It describes how unusual data at least as inconsistent with that null as the observed data would be if the null and model assumptions held. It is not the probability that the null hypothesis is true, nor is it a measure of the size or clinical importance of a difference.

What do the secondary analysis methods add?

The record reports negative binomial regression for the rate comparisons of the severe-exacerbation outcome and the outcome that also includes exacerbations treated with antibiotics. For the FEV1 outcome, the sponsor reports a Mixed Model for Repeated Measures (MMRM) analysis and a mean difference in final values. In general, repeated-measures models analyze measurements collected across time while accounting for their relationship within participants. The registry does not report the detailed model specification or missing-data assumptions.

There is also a distinction between the registry’s reported analyses and the simple unadjusted re-calculations. The latter use summary statistics and assumptions stated for planning illustrations; they do not adjust for any covariates and do not reproduce the sponsor’s full modeling. The re-calculated comparisons are therefore useful as a check on broad scale and planning, not as a substitute for the trial’s reported analysis.

Interpretation should keep the primary endpoint, secondary outcomes, and analysis methods distinct. A result on one outcome does not automatically determine the result for another. The record does not report a complete analysis plan, multiplicity strategy, interim-analysis plan, or details of how missing observations were handled. Those omissions limit how fully the posted results can be independently reconstructed.

11. Confidence Intervals and Precision

The sponsor reports a 95% confidence interval of 0.81 to 1.04 for the primary rate ratio of 0.92. The interval includes 1, so the data are compatible with equal rates as well as with values indicating a lower or higher rate in the roflumilast group. Its endpoints convey uncertainty around the estimate; the point estimate should not be read as an exact measure of the underlying rate relationship.

The group-specific rate intervals are 1.17 to 1.39 for placebo and 1.06 to 1.28 for roflumilast. These intervals describe uncertainty in each group’s rate estimate, while the rate-ratio interval addresses the between-group comparison. They are not interchangeable. Precision for the comparative question is most directly represented by the interval around the rate ratio.

Confidence intervals are generally narrower when estimates are more precise and wider when they are less precise. Their interpretation depends on the analysis model and the data assumptions. Here, the sponsor used negative binomial regression, while the simple recalculation uses a Poisson-based approximation. The record marks a discrepancy between the recalculation and sponsor analysis, so the sponsor-reported interval is the primary basis for interpreting the trial’s comparative result.

The interval does not resolve whether any difference would be clinically important, and it does not account for unreported sources of uncertainty such as unreported modeling choices or multiplicity procedures. Details about these matters are not reported in the registry record.

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 sample-size illustrations are planning calculations based on the observed effect, not the trial’s original design justification. For this trial, the outcomes are event rates or a continuous FEV1 measure, so the figures are participants per arm, rather than total events. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each scenario considers an effect equal to the stated fraction of the observed effect.

The 80% power column gives the number of participants per arm that the planning calculation estimates would be needed for an 80% chance of detecting the specified effect under those assumptions, if the effect and assumptions were true. The 90% power column gives the corresponding planning number for a 90% chance. These are design illustrations, not guarantees about what a future study would achieve.

The rows consider planning for 100%, 75%, and 50% of the observed effect. A smaller effect is harder to distinguish from random variation, so the planning calculations require more participants per arm for the smaller-effect scenarios. For example, the primary outcome’s calculation gives 1,923 participants per arm for 80% power when planning for 100% of the observed effect, and 7,531 participants per arm when planning for 50% of the observed effect. These figures are illustrations rather than recommendations.

This is not observed power. It is a planning calculation using the observed effect as an input. The registry’s original sample-size assumptions and justification are not reported in the registry record.

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, λ₀ and λ₁ are event rates per patient-year, and T is follow-up per participant in years. This assumes Poisson counts and equal follow-up.

Worked calculation for this trial

1. Control rate λ₀ = 1.2700; treatment rate λ₁ = 1.1700 events per patient-year; follow-up T = 1.00 years.
2. n = (1.96 + 0.842)² × (1/1.2700 + 1/1.1700) / (1.00 × [ln(0.921)]²) = 1,922.8, rounded up to 1,923 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
lambda0 <- 1.27                      # control rate, events per patient-year
lambda1 <- 1.17                      # treatment rate
tt      <- 0.996550402453047                       # follow-up per participant, years
n_per_arm <- function(power, frac = 1) {
  l1 <- lambda0 * (lambda1 / lambda0)^frac   # smaller effect = fraction of the observed log rate ratio
  ceiling(round((z_a + qnorm(power))^2 * (1 / lambda0 + 1 / l1) / (tt * log(l1 / lambda0)^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 = ",")))

# Simulation 1: Poisson counts. Simulation 2: over-dispersed counts (variance = 1.5 x mean), analysed with the matching variance.
set.seed(1443845)
n    <- n_per_arm(0.80)
reps <- 20000
m0 <- lambda0 * n * tt; m1 <- lambda1 * n * tt
x0 <- rpois(reps, m0); x1 <- rpois(reps, m1)
z  <- log(x1 / x0) / sqrt(1 / x0 + 1 / x1)
cat(sprintf("Simulated power, Poisson counts, n = %s per arm: %.3f\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE)))
phi <- 1.5
y0 <- rnbinom(reps, mu = m0, size = m0 / (phi - 1)); y1 <- rnbinom(reps, mu = m1, size = m1 / (phi - 1))
z2 <- log(y1 / y0) / sqrt(phi * (1 / y0 + 1 / y1))
cat(sprintf("Simulated power, over-dispersed counts (phi = 1.5), same n: %.3f\n", mean(abs(z2) > z_a, na.rm = TRUE)))

Output from running the code

100% of observed effect: n per arm = 1,923 (80% power), 2,575 (90% power)
 75% of observed effect: n per arm = 3,383 (80% power), 4,528 (90% power)
 50% of observed effect: n per arm = 7,531 (80% power), 10,082 (90% power)
Simulated power, Poisson counts, n = 1,923 per arm: 0.800
Simulated power, over-dispersed counts (phi = 1.5), same n: 0.625

The first simulation draws Poisson counts and should land close to the target power. The second allows extra variability (variance 1.5 times the mean, a purely illustrative value), which lowers the power at the same sample size, so rate trials usually need more participants than the Poisson formula suggests.

14. Analyses Reported by the Sponsor

The sponsor’s primary analysis is a negative binomial regression comparison. It reports a rate ratio of 0.92, a 95% confidence interval from 0.81 to 1.04, and p = 0.163. This is the analysis to use when describing the trial’s reported primary result.

The simple re-calculation uses posted group summary statistics and a Poisson-based approximation. Its unadjusted p-value is 0.208. The re-calculation and sponsor analysis are not identical: they use different approaches, and the record flags a discrepancy. The re-calculated comparison is therefore not a correction to, or replacement for, the sponsor’s analysis.

The sponsor analysis takes precedence for interpretation. The re-calculation can provide a rough independent check based on summary data, but it cannot reproduce details of a negative binomial model that are not available in the registry record. Model covariates and other implementation details are not reported there. The recalculation should not be treated as evidence that overrides the sponsor’s reported estimate and interval.

15. Multiplicity, Missing Data and Analysis Populations

The registry record reports a primary outcome and several secondary outcomes, along with multiple analyses. It does not report a multiplicity adjustment strategy or a testing hierarchy that would explain how the overall chance of false-positive findings across outcomes was controlled. Without that information, secondary p-values should be read as nominal results rather than as evidence adjusted for the number of comparisons.

The record identifies the primary analysis population as the Intent-to-Treat Study Population. It does not report comparable population details for every secondary analysis, nor does it provide a complete account of exclusions or analysis-set definitions. The outcome group counts differ for the FEV1 measure, but the reason for that difference is not reported in the registry record.

Missing-data methods are not reported in the registry record. In particular, the record does not describe how missing outcome measurements, incomplete follow-up, or withdrawal from treatment were handled in the statistical models. These details matter because different assumptions about unobserved outcomes can affect estimates and their uncertainty.

The registry also does not report the trial’s full statistical analysis plan, planned interim analyses, original sample-size justification, model covariates, or detailed model specifications. Their absence does not establish that no such procedures existed; it means that they cannot be characterized from the registry record. The reported primary result can still be described as posted, but a complete independent audit of the analysis is not possible from these details alone.

16. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pneumonia (4.0%), lobar pneumonia (0.6%), acute myocardial infarction (0.6%), acute respiratory failure (0.6%), respiratory failure (0.6%).

The registry records serious adverse events for 162 participants in the placebo group and 180 participants in the roflumilast group. The groups at risk are reported as 1,174 and 1,178 participants, respectively. The record also reports other adverse events affecting 188 participants in the placebo group and 317 participants in the roflumilast group. These are counts, and they should be read in the context of the corresponding group sizes and the registry’s event definitions.

The most frequently listed serious adverse event term is pneumonia, at 4.0%. Other listed terms include lobar pneumonia, acute myocardial infarction, acute respiratory failure, and respiratory failure, each reported at 0.6%, and non-cardiac chest pain at 0.5%. These entries are registry safety summaries, not a complete account of every event or a causal comparison.

Deaths are not reported in the registry record’s safety figures. The posted counts alone do not establish that an event was caused by a study treatment, and they do not provide a full assessment of safety differences. The registry record does not include enough detail to characterize event timing, seriousness criteria, or individual clinical circumstances.

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

The primary results show observed exacerbation rates of 1.27 per patient per year with placebo and 1.17 with roflumilast. The sponsor’s estimated rate ratio is 0.92, but its 95% confidence interval includes 1 and the reported p-value is 0.163. Statistically, the posted primary analysis does not provide clear evidence of a difference between the groups under the sponsor’s model.

This does not mean that the groups were proven to have identical rates. The interval includes a range of possible relative values, and the point estimate is uncertain. A non-significant result is not proof of no effect; it indicates that the analysis did not establish a difference at the level of evidence summarized by the reported test and interval.

Nor does the result establish that the observed difference is clinically important. The registry record does not state a threshold for clinical importance, and statistical significance is not the same as clinical importance. The event-rate outcome is specific to moderate or severe exacerbations during the stated treatment period. It does not capture every dimension of COPD or every possible benefit or harm.

The secondary results offer additional outcome-specific information, but they do not erase the uncertainty around the primary result. Multiple secondary comparisons, with no multiplicity strategy reported in the registry record, require cautious interpretation. The safety summary also has limits: event counts do not establish causation and do not describe every aspect of risk.

The conclusions apply to the randomized trial population and the reported follow-up and outcomes. The registry record does not provide detailed information about all eligibility criteria, missing-data handling, or model specifications. The posted results therefore support a bounded statistical description, rather than a complete account of effects in every population or setting.

18. Why This Trial Matters Statistically

This trial illustrates why the choice of statistical model matters for clinical event-rate outcomes. Exacerbations can recur within participants, and count data may show over-dispersion. The sponsor’s use of negative binomial regression is therefore an important part of understanding the reported rate ratio and its uncertainty. A simpler summary-statistic calculation does not necessarily reproduce a model-based result.

It also demonstrates why an effect estimate should be read with its confidence interval rather than on its own. The primary estimate is below 1, but the reported interval includes 1. The result is not adequately described by saying only that the estimated rate was lower in one group; uncertainty is central to the interpretation.

The record also separates the primary question from multiple secondary outcomes. Secondary endpoints can provide useful context, yet unadjusted results across several comparisons are less conclusive when the multiplicity strategy is not reported. The primary analysis remains the clearest anchor for the trial’s stated aim.

Finally, the difference between the sponsor analysis and the unadjusted re-calculation is a practical reminder that posted group summaries are not enough to reconstruct every feature of a regression analysis. Where a discrepancy is identified, the sponsor’s own analysis takes precedence, while the simplified calculation should be treated as a limited check. This distinction helps prevent planning illustrations or approximate recalculations from being mistaken for the trial’s original statistical design or reported inference.

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.
• Rate comparisons use a Poisson approximation unless the sponsor posted confidence limits; real event counts are often over-dispersed.

Several trial-specific details needed for a fuller statistical assessment are not reported in the registry record. These include the original sample-size justification, the detailed statistical analysis plan, planned interim analyses, model covariates, and full specifications for the negative binomial and repeated-measures models. The registry therefore does not permit a complete reconstruction of the sponsor’s analyses.

Missing-data handling is also not reported. This matters for the continuous FEV1 outcome, whose analysis group counts differ from those for the primary endpoint, as well as for event outcomes with incomplete follow-up. The reason for the differing FEV1 counts and the assumptions used for missing measurements are not stated.

The posted secondary outcomes involve multiple comparisons, but the record does not describe a multiplicity adjustment or a hierarchy of testing. Secondary p-values consequently warrant caution. The registry also does not report sufficient detail to map every reported analysis to a named outcome with certainty beyond the analyses directly associated with the posted outcomes.

The simple recalculations are based on summary statistics and simplified assumptions; they do not reproduce the sponsor’s model. The record marks a discrepancy for the primary comparison, so the sponsor’s analysis takes precedence. Finally, the registry safety summary presents selected adverse-event counts and terms, while deaths are not reported in those figures. These summaries do not establish causality or provide a complete safety assessment.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Roflumilast in Chronic Obstructive Pulmonary Disease (COPD) Patients Treated With Fixed Dose Combinations of Long-acting β2-agonist (LABA) and Inhaled Corticosteroid (ICS) — NCT01443845.
• Registry record last updated 2017-04-13; status last verified 2017-03.
• Results first posted 2017-01-25.
• 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

This completed phase 4 trial randomized 2,354 participants with severe to very severe COPD treated with fixed-dose LABA/ICS to roflumilast or placebo. It was triple-masked and used a parallel-group design.

The primary endpoint was the rate of moderate or severe COPD exacerbations per patient per year from baseline to Week 52. The sponsor reported a negative binomial rate ratio of 0.92, with a 95% confidence interval of 0.81 to 1.04 and p = 0.163. The interval includes 1, so the primary analysis does not provide clear statistical evidence of a group difference.

Secondary outcomes addressed severe exacerbations, exacerbations treated with antibiotics, and FEV1 change. Multiplicity procedures, missing-data methods, and the trial’s original sample-size justification are not reported in the registry record.