Cardiovascular Events in Chronic Obstructive Pulmonary Disease Patients Initiating Olodaterol or Other Long-acting beta2 Agonists: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cardiovascular Events in Chronic Obstructive Pulmonary Disease Patients Initiating Olodaterol or Other Long-acting beta2 Agonists (NCT03405363), a registered phase not stated study. The primary outcome is Occurrence of First Hospitalisation or Hospital Outpatient Clinic Visit for Atrial Fibrillation or Flutter (AF).

2. Clinical Question

This is a question about outcomes observed in a treatment-initiation cohort, not a randomized assignment. The comparison can describe what was recorded in the study population, but the registry entries do not establish that the groups were otherwise comparable. The exact comparative statistical question, including any adjusted contrast, is not reported in the registry record.

3. Trial Design

The registry identifies this as a completed cohort study of cardiovascular events among patients with COPD initiating olodaterol or other long-acting beta2 agonists. It lists Boehringer Ingelheim as the sponsor, a start date of 2018-01-31, a primary completion date of 2020-02-03, and enrollment of 65406. Allocation is not stated. The record also cautions that the arms are not a like-for-like control comparison, so the study should not be described as a randomized trial.

The comparison is between new users of olodaterol and new users of other LABAs. The population is described as a propensity score-trimmed population derived from the Danish Patient Registry. The registry does not report the trimming rules, how the propensity score was estimated, which variables were included, or the group-specific enrollment counts. It therefore does not allow readers to assess the details of the resulting population or the balance of measured characteristics between groups.

The exposure description limits assessment to the first episode of continuous use, defined as consecutive dispensings separated by up to 14 days. Follow-up termination is described as occurring at the outcome of interest, database disenrollment, 14 days after estimated discontinuation of the last dispensing for the index LABA, switching to another LABA, dispensing a second LABA, death, or the end of the study period. The endpoint timeframe is listed as up to 4 years and 11 months. The relationship between that timeframe and the dates listed elsewhere in the registry record is not explained.

The record lists 6 outcome measures and 8 analyses. It identifies all-cause mortality as a secondary outcome, but the statistical analysis plan, analysis populations beyond the population description, and the methods used to compare groups are not reported in the registry record. The study status is completed; that status alone does not provide the missing outcome estimates or analysis details.

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: Only subjects that met all inclusion and none of the exclusion criteria were included. The matched cohort study design and propensity score methodology were used to establish two comparable treatment cohorts, the Olodaterol cohort and the other LABA cohort.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Propensity score-trimmed population of patients with Chronic Obstructive Pulmonary Disease (COPD) derived from the Danish Patient Registry

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: Propensity score-trimmed population of patients with Chronic Obstructive Pulmonary Disease (COPD) derived from the Danish Patient Registry initiating olodaterol (either alone or in free- or fixed-dose combination with a LAMA) or an alternative LABA (alone or in a free- or fixed-dose combination with a LAMA).

6. Primary Endpoint

The registry lists the primary outcome as Occurrence of First Hospitalisation or Hospital Outpatient Clinic Visit for Atrial Fibrillation or Flutter (AF), measured over Up to 4 years and 11 months.

The primary endpoint is the occurrence of a first hospitalisation or hospital outpatient clinic visit for atrial fibrillation or flutter (AF). Cases were identified through at least one relevant ICD-10 code, either as a primary inpatient discharge diagnosis or as a diagnosis code in a hospital outpatient specialist visit in the Danish National Patient Registry. The measure is a count of events, and the listed timeframe is up to 4 years and 11 months.

Using a first recorded occurrence focuses the endpoint on whether an eligible AF event was identified during the defined exposure and follow-up period, rather than on repeated visits by the same person. Hospital and specialist-visit records provide a way to identify clinically recorded diagnoses in administrative data. As a general statistical consideration, such an endpoint reflects events captured and coded in those settings; it does not by itself describe events that were not recorded in those sources.

The endpoint description specifies the exposure and follow-up rules, but the registry record does not report the number of AF events in either group, a group comparison, or a measure of uncertainty. It also does not report validation details for the diagnosis-code definition. Thus, the endpoint can be described from the registry definition, while its observed frequency and comparative precision cannot be assessed from the reported primary results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a count endpoint represent?

The primary endpoint is a count of first recorded AF events, identified from hospital inpatient or hospital outpatient specialist records using ICD-10 diagnosis codes. The event definition, data source, exposure window, and follow-up stopping rules determine what can be counted. A count is not itself a risk or rate: interpreting it comparatively requires information about how many people were at risk and how long they were observed, as well as a specified method for comparing groups. The registry record does not report group-specific primary counts, denominators, person-time, or a primary comparative estimate.

As a general concept, count outcomes can be analysed with methods that model event counts or rates. The appropriate approach depends on the study design, observation time, distribution of outcomes, and the estimand of interest. Over-dispersion occurs when observed variability in counts exceeds what a chosen model expects. If not handled appropriately, standard errors and confidence intervals may be poorly calibrated. These are general considerations; the registry does not report which count model, if any, was used for this endpoint.

How should the exposure definition be read?

The exposure groups are defined by new use of olodaterol or another LABA. The record limits the exposure window to the first continuous-use episode, with consecutive dispensings separated by up to 14 days. Follow-up can end at the event, database disenrollment, estimated discontinuation, a switch or second LABA dispensing, death, or the stated end of the study period. These rules define the period in which events are attributed to the exposure under the registry description. The record does not explain how discontinuation dates were estimated or how censoring was handled analytically.

In general, censoring occurs when outcome follow-up ends before an event is observed. Statistical methods for time-to-event data account for differing follow-up by incorporating the timing of events and censoring. A count-based analysis may instead use an observation-time offset or another approach, depending on the intended estimand. The record does not state whether the primary comparison used time-to-event methods, a count model, or another method. The endpoint measure is labelled NUMBER, but that label alone does not identify the analysis model.

What does propensity score trimming contribute?

The registry calls the primary population propensity score-trimmed and says it was derived from the Danish Patient Registry. In general, a propensity score estimates the probability of receiving a treatment conditional on measured characteristics. Trimming can exclude observations in regions where treatment groups have limited overlap, with the aim of defining a population in which comparisons may be more supportable. It can also change which patients the resulting estimate represents.

Those general properties do not establish what was done in this study. The registry does not report the characteristics used to estimate the score, the trimming criteria, the number of records retained or excluded, or post-trimming balance. It also does not specify whether the final comparison was adjusted using the score or another method. The population label therefore indicates that trimming was part of the stated population definition, but does not allow independent assessment of its implementation or consequences.

What information would an effect estimate and interval provide?

Generally, an effect estimate communicates the direction and magnitude of a contrast under a defined analysis. Its meaning depends on whether it is a difference, ratio, rate contrast, hazard-based measure, or another quantity. A confidence interval communicates uncertainty in that estimate under the model and sampling assumptions. A narrower interval indicates greater precision within those assumptions; it does not remove bias from confounding, outcome misclassification, or selection.

The registry does not report the primary effect measure, its confidence interval, or a p-value. It is therefore not possible to explain a particular estimate or assess its precision. Statistical significance, if reported, would not by itself establish that an association is causal or clinically important. Conversely, an imprecise result would not establish equivalence. These are general principles and are not study-specific findings.

What do the planning assumptions mean?

The registry describes a sample-size planning illustration based on the observed effect, a stated fraction of that effect, equal allocation, a two-sided significance level of 0.05, and a normal approximation. It specifies that time-to-event calculations use the total number of events, while other calculations report participants per arm. These are planning conventions, not descriptions of observed study results.

In general, a sample-size calculation estimates the amount of information a future study might need to detect a defined effect under stated assumptions. Changing the effect to detect changes the planning requirement. A calculation based on an observed effect is especially dependent on that estimate, which may be uncertain. The registry does not report the observed effect used, the planning outputs, or the study's own design justification. The illustration should not be described as observed power, which is not a substitute for reporting the estimated effect and its uncertainty.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 6 outcome measures and 8 analyses, and identifies all-cause mortality as a secondary outcome. It does not report which analyses correspond to which outcomes, which analysis was designated primary beyond the listed AF endpoint, or whether the analyses were prespecified. A statistical plan describing the handling of multiple outcomes, comparisons, subgroups, or repeated analyses is not reported in the registry record.

As a general statistical issue, examining multiple outcomes or contrasts can increase the chance of a nominally notable result arising by chance. Methods for addressing multiplicity may adjust the interpretation or testing threshold, or may distinguish confirmatory analyses from exploratory ones. No multiplicity procedure or rationale is reported, so it is not possible to determine how the set of analyses was handled.

The record describes a propensity score-trimmed population derived from the Danish Patient Registry, but does not report missing-data procedures, the extent of missingness, or how unavailable covariates were handled. It also does not specify whether all eligible records contributed to each analysis or whether outcome data were incomplete. The analysis population is therefore only partly described: the source and trimming label are available, but detailed inclusion rules and analysis-specific populations are not reported in the registry record.

These omissions limit assessment of selective reporting, the influence of missing information, and the comparability of populations across analyses. They do not establish that the analyses were improperly conducted; they mean that the relevant methods cannot be evaluated from the registry record.

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

The record establishes that the study addressed cardiovascular outcomes in COPD patients initiating olodaterol or other LABAs, and it defines a primary AF endpoint based on hospital and specialist-visit diagnoses. It also records deaths by exposure group in the adverse-event data and lists all-cause mortality as a secondary outcome. These details describe the study question and some reported counts.

They do not provide the primary AF event counts by group, a comparative estimate, or uncertainty around that estimate. The mortality counts are not accompanied by denominators or a stated comparative analysis. As a result, the posted information does not support quantifying the difference in AF or mortality between treatment groups. It also does not establish that the outcomes were equal, since the absence of reported comparative results is not evidence of no difference.

The cohort design and the registry's warning that the arms are not a like-for-like control comparison matter to interpretation. Treatment initiation was not described as randomized, and allocation is not stated. An observed contrast in such a cohort can be influenced by differences between patients receiving the respective treatments. Propensity score trimming is mentioned, but its methods and balance results are not reported; it cannot be assumed to eliminate confounding.

More generally, statistical association and causal effect are different claims. A causal interpretation requires defensible assumptions about treatment assignment, confounding, measurement, follow-up, and analysis. The registry entries available do not permit those assumptions to be evaluated. The study question is relevant to treatment-associated outcome monitoring, but the record as described does not establish that either treatment increases or decreases risk. It also does not establish comparative safety or efficacy.

12. Why This Trial Matters Statistically

This study illustrates why design and analysis details are essential when interpreting comparisons from observational treatment-initiation cohorts. The groups are defined by new use of olodaterol or other LABAs, but the record states that they are not a like-for-like control comparison and does not state how allocation occurred. Those features make it important to distinguish a descriptive comparison from a causal conclusion.

The primary endpoint is defined through administrative diagnosis records and a specific follow-up scheme. Clear endpoint definitions help readers understand which events count, while the available record also leaves important questions about coding validation, group-specific event counts, and the statistical comparison unanswered. Without the event estimates and their precision, the outcome definition alone cannot show the size or certainty of an association.

The propensity score-trimmed population is another important statistical feature. Trimming may affect the population represented by an estimate, and its value depends on implementation and the degree of measured covariate balance. Since the registry does not report those details, readers cannot assess how the procedure affected comparability or generalizability.

The record also illustrates the difference between planning calculations and observed results. Planning assumptions can inform a future design, but do not substitute for estimates from the completed study and are not observed power. For this study, the record describes an illustration but does not give its outputs or the underlying observed effect. Transparent reporting of the analysis plan, results, and uncertainty would be needed to assess the statistical evidence more fully.

13. 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 central limitation for interpreting this record is that the primary AF outcome is defined but its group-specific event counts and comparative results are not reported. No primary effect estimate, confidence interval, or p-value is available. The registry's death counts, 859 for olodaterol and 1872 for other LABAs, lack group-specific denominators and a reported analysis, so they cannot be interpreted as comparative risks.

The observational cohort design limits causal interpretation, particularly because allocation is not stated and the registry explicitly says the arms are not a like-for-like control comparison. Propensity score trimming is mentioned, but its variables, rules, exclusions, and resulting balance are not reported. Residual confounding and changes in the population represented by a comparison therefore cannot be assessed from the record.

Outcome ascertainment depends on ICD-10-coded hospital inpatient or hospital outpatient specialist records. The registry does not report validation of the AF definition or information on events outside those recorded settings. Follow-up is described with several potential termination events, but the analytic handling of censoring and discontinuation is not reported. The relationship between the listed endpoint timeframe of up to 4 years and 11 months and the study dates is not explained.

The analysis plan, multiplicity approach, missing-data procedures, group-specific enrollment, and analysis populations are also not described in sufficient detail to evaluate. These are limits of what can be assessed from the registry record; they do not show that a particular method was or was not used.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: Cardiovascular Events in Chronic Obstructive Pulmonary Disease Patients Initiating Olodaterol or Other Long-acting beta2 Agonists — NCT03405363.
• Registry record last updated 2021-04-30; status last verified 2021-04.
• Results first posted 2021-04-30.
• 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.

17. Record Summary

The registry record describes a completed cohort study of COPD patients initiating olodaterol or other long-acting beta2 agonists, sponsored by Boehringer Ingelheim. It lists enrollment of 65406, a start date of 2018-01-31, and a primary completion date of 2020-02-03. The primary endpoint is first hospitalisation or hospital outpatient specialist visit for AF, identified using ICD-10 diagnosis codes in the Danish National Patient Registry. The stated timeframe is up to 4 years and 11 months.

The record describes a propensity score-trimmed population and exposure rules based on the first continuous-use episode. It lists all-cause mortality as a secondary outcome and records deaths of 859 in the olodaterol group and 1872 in the other LABA group. Primary AF event counts, comparative estimates, uncertainty intervals, and detailed analysis methods are not reported in the registry record. The record cautions that the arms are not a like-for-like control comparison.