This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for A Study Based on Medical Records That Looks at the Duration of Use of Two Types of Inhalers With Different Medicines in Patients With Chronic Obstructive Pulmonary Disease (COPD) (NCT03979807), a registered phase not stated study. The primary outcome is Rate of Discontinuation of Index Treatment (Olodaterol/Tiotropium Bromide or Umeclidinium/Vilanterol).
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03979807 |
| Phase | Phase not stated |
| Condition | Pulmonary Disease, Chronic Obstructive |
| Allocation | Not Stated |
| Design model | Not Stated |
| Masking | Not Stated |
| Enrollment | 11296 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Boehringer Ingelheim |
| Results posted | Yes |
2. Clinical Question
The clinical question is whether discontinuation rates differ between initiators of olodaterol/tiotropium bromide and initiators of umeclidinium/vilanterol among people with chronic obstructive pulmonary disease (COPD). The registry describes both as dual bronchodilators from the same drug class, delivered through different inhaler devices: a dry-powder inhaler (DPI) and a Soft Mist Inhaler (SMI). It does not specify which combination corresponds to each device in the outcome description.
The comparison concerns persistence with the initially selected treatment, not a randomized assignment to inhaler or medication. The outcome is a rate of discontinuation per person-time at risk, estimated in propensity score-matched cohorts. The registry describes the comparison population as initiators of the two combinations identified in the Truven MarketScan Claims database between November 2015 and March 2018 and assembled in a 1:2 matched cohort.
In practical statistical terms, the question is whether the observed event rates differ in these matched claims-based groups under the registry’s definition of a refill gap. A rate comparison uses both the number of discontinuation events and the person-time accumulated before an event or censoring. It therefore addresses the pace of recorded discontinuation over observed follow-up rather than simply the proportion of people who discontinued by a single common date.
The registry record does not report the numerical rates, event counts, person-time totals, an effect estimate comparing the cohorts, or an uncertainty interval. Consequently, it is not possible to determine from the reported outcome information which cohort had a higher rate or how precisely any difference was estimated.
3. Trial Design
Despite being listed as a clinical study, this is a medical-record analysis rather than a randomized treatment comparison. The brief description says the study compares persistence with two medicines in the same drug class delivered by different devices. The primary population is described as initiators of olodaterol/tiotropium bromide and initiators of umeclidinium/vilanterol identified in the Truven MarketScan Claims database between November 2015 and March 2018.
The registry describes the primary analysis population as a 1:2 propensity score-matched cohort. Matching is intended to construct comparison groups based on measured characteristics, but the registry record does not report which characteristics entered the propensity score, the matching procedure’s technical details, or measures of balance after matching. It also does not report the number of participants in each matched cohort. Total enrollment is 11296; that figure should not be treated as the matched analysis sample size without further information.
The study is marked completed. Its registry start date is 2019-06-10 and its primary completion date is 2019-06-11. Those dates are distinct from the claims-identification period stated for the cohort. The record does not explain the relationship between the registry dates and the underlying data assembly or analysis schedule.
Allocation is not stated, and the registry notes that the arms are not a like-for-like control comparison. Because treatment was not described as randomly assigned, differences between groups may reflect characteristics associated with treatment choice as well as any device-related differences. The record does not report a protocol-level sample-size justification, a randomization procedure, or an analysis plan in sufficient detail to establish how these issues were addressed.
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 in the study. Initiators of Olodaterol/Tiotropium Bromide were matched to patients receiving Umeclidinium/Vilanterol using a 1:2 propensity score matching, resulting in a matched cohort of 11296 overall.
Overall Study
| Stage | Olodaterol/Tiotropium Bromide - Matched Cohort | Umeclidinium/Vilanterol - Matched Cohort |
|---|---|---|
| STARTED | 3876 | 7420 |
| COMPLETED | 3876 | 7420 |
| NOT COMPLETED | 0 | 0 |
Analysis population for the primary outcome, as defined in the registry: 1:2 propensity score matched cohort of initiators of Olodaterol/Tiotropium Bromide and initiators of Umeclidinium/Vilanterol, identified from Truven MarketScan Claims database between November 2015 and March 2018.
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: 1:2 propensity score matched cohort of initators of Olodaterol/Tiotropium Bromide and Initiators of Umeclidinium/Vilanterol, identified from Truven MarketScan Claims database between November 2015 and March 2018.
| Characteristic | Unit | Olodaterol/Tiotropium Bromide - Matched Cohort | Umeclidinium/Vilanterol - Matched Cohort | Total |
|---|---|---|---|---|
| Participants analysed | 3876 | 7420 | 11296 | |
| Age, Continuous | Years | 64.63 (SD 10.50) | 64.58 (SD 10.53) | 64.60 (SD 10.52) |
| Sex: Female, Male: Female | Participants | 1785 | 3409 | 5194 |
| Sex: Female, Male: Male | Participants | 2091 | 4011 | 6102 |
| Race and Ethnicity Not Collected | Participants | — | — | 0 |
6. Primary Endpoint
The registry lists the primary outcome as Rate of Discontinuation of Index Treatment (Olodaterol/Tiotropium Bromide or Umeclidinium/Vilanterol), measured over From first day after the cohort entry date to the earliest occurence of the outcome (discontinuation or refills), or any censoring criteria (365 days of follow-up without discontinuation, death, disenrollment end of data)..
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | Events per 1000 person-years |
| Endpoint classification | Event-Rate Endpoint |
The primary endpoint is the rate of discontinuation of the index treatment, reported in events per 1000 person-years. The index treatment is olodaterol/tiotropium bromide or umeclidinium/vilanterol, according to the cohort entered. The registry says that rates are calculated as the number of events divided by person-years at risk in the matched cohorts.
The endpoint uses a 60-day allowable gap. The description refers to discontinuation in terms of persistence and says persistence is defined by no refill claim within 60 days after the end of treatment supply, excluding treatment switch and death. It also says that adding another treatment to the index treatment does not count as discontinuation. Since this wording links the outcome label and the persistence definition in a potentially confusing way, interpretation should follow the registry’s stated operational rule rather than assume a different definition.
Follow-up begins on the first day after cohort entry and continues until the earliest listed outcome or censoring criterion. These include 365 days of follow-up without discontinuation, death, disenrollment, or the end of available data; the description also mentions refills in defining the observation period. The endpoint is useful for claims-based persistence research because it incorporates the time participants contribute before discontinuation or censoring. It is not the same as a simple percentage discontinuing by a common follow-up date.
The registry record does not report the event counts, person-years, or numerical rates needed to interpret the observed endpoint quantitatively.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry identifies a rate outcome and describes a matched comparison, but it does not report the numerical primary outcome result. There is no posted rate for either treatment cohort, no event count, no person-time total, and no comparative effect estimate or confidence interval. A direction or magnitude of difference therefore cannot be established from the outcome information in the registry record.
This absence is important because an event rate is interpreted using both its numerator and its exposure-time denominator. A rate alone is not a count of people who discontinued, nor does it directly give the probability that a participant discontinued over a specified period. Without the cohort-specific rates and their uncertainty, readers cannot assess the size or precision of the comparison. The record also does not report a p-value for the primary outcome.
The matched design may help make the treatment cohorts more comparable on measured characteristics included in the propensity score. However, the characteristics used, the quality of matching, and the post-match balance are not reported. Matching does not by itself guarantee comparability on unmeasured characteristics or remove all confounding. In this nonrandomized analysis, treatment choice may be associated with patient or care characteristics that also relate to persistence.
The 1:2 matching description indicates a design ratio, not the number of people analyzed. The total enrollment of 11296 is reported separately, and the registry does not state how many people remained in each matched cohort or contributed person-time to the primary analysis. Nor does it report how much follow-up was available in each group or how often censoring occurred.
The outcome definition has a direct bearing on interpretation. A refill gap is a claims-based proxy for treatment discontinuation, not a direct observation of inhaler use. The registry specifies a 60-day allowable gap and says a treatment switch and death are excluded from the persistence definition. Addition of another treatment does not count as discontinuation. The result, if reported, would therefore describe discontinuation under these rules and within the claims data, rather than every possible form of stopping or changing treatment.
The registry lists one primary outcome measure and one analysis, but it does not report a complete statistical analysis plan or any multiplicity procedure. It is not reported in the registry record whether interim analyses were conducted, how missing information was handled, or whether additional analyses were prespecified. These omissions limit assessment of the analytic context, but they do not alter the central point: without the numerical primary outcome result, the observed comparison cannot be quantified.
9. Statistical Methods Explained
The primary measure is an event rate: discontinuation events divided by person-years at risk, expressed as events per 1000 person-years. This type of measure accounts for differing amounts of observed follow-up. A participant who contributes time before an event or censoring contributes to the denominator for that period, while an observed discontinuation contributes an event to the numerator.
What does a rate measure?
As a general statistical explanation, an event rate describes how frequently events occur relative to accumulated time at risk. It is not automatically the same as the proportion of participants with an event. A proportion uses a fixed group of people as its denominator, while a rate uses person-time. Rates can be helpful when follow-up time varies, but their interpretation depends on how events and time at risk are defined.
For this analysis, the registry states that the rate is based on events divided by person-years at risk. It does not provide the cohort-specific numerators or denominators, so the reported rate values cannot be checked or interpreted from the registry record. No rate ratio or other direct comparison is reported either.
How is discontinuation defined?
The registry uses a 60-day allowable gap after the end of treatment supply. Its description says persistence is assessed by whether there is no refill claim within that interval, excluding treatment switch and death. It also states that adding another treatment to the index treatment does not count as discontinuation. The description uses persistence and discontinuation in a way that can be difficult to parse; the stated claims-based gap rule is the relevant operational detail, and no further clarification is reported.
Claims records capture billing and refill events, not every aspect of medication-taking behavior. As a general limitation of this type of measure, a recorded refill pattern may not establish whether or how a medicine was used between claims. The endpoint should therefore be read as a claims-defined measure of persistence or discontinuation, not as a direct measure of inhaler use.
What does matching contribute?
The cohorts are described as propensity score matched in a 1:2 ratio. In general, a propensity score represents the estimated probability of receiving a particular treatment given measured characteristics. Matching on such a score aims to create groups with more comparable distributions of the measured characteristics included in the model. It is a design and adjustment strategy for observational data, not random allocation.
The registry does not identify the characteristics used to estimate the propensity score, explain the matching algorithm, or report balance diagnostics. It is therefore not possible to assess which differences the matching addressed or how closely the resulting groups aligned on measured characteristics. As a general point, matching cannot balance characteristics that were not measured or not incorporated, and it does not establish that treatment groups are equivalent in all respects.
How should censoring be understood?
Follow-up is described as beginning on the first day after cohort entry and ending at the earliest outcome or censoring criterion. Listed criteria include 365 days of follow-up without discontinuation, death, disenrollment, or end of data. Censoring means that the observation period ends without a recorded event at that point; it does not mean that the person is known never to experience the event. Rate analyses rely on how event times and time at risk are defined, so transparent reporting of censoring and follow-up is useful.
The registry does not provide the amount of person-time, the number censored for each reason, or an explanation of how the analysis handled censoring beyond the stated criteria. It also does not report a distribution of follow-up time. Those details are not available for evaluating the amount of information contributing to the rate comparison.
What uncertainty can be assessed?
A confidence interval is, in general, a range of values compatible with the data and the statistical model at a stated confidence level. Its width conveys precision: a narrower interval indicates greater precision under that model, while a wider interval indicates less precision. The registry record does not report a confidence interval for the primary rate or a comparative estimate, so the precision of the result cannot be assessed.
The record also does not report a p-value for the outcome or a detailed model specification. The analysis is listed as a rate outcome, but the precise inferential method is not reported in the registry record. No adjustment variables beyond the general matching description are specified. The registry’s sample-size assumptions describe an illustration using a two-sided significance level of 0.05, equal allocation, a normal approximation, and an effect set as a fraction of the observed effect. Such planning illustrations are not observed power and do not substitute for an original design justification. The registry record does not report a sample-size justification for the study’s own design.
Over-dispersion is a general concern when event counts vary more than a simple model assumes; if present, it can make standard errors from an overly restrictive model too small. The registry does not report a model or a dispersion assessment, so it is not possible to say whether over-dispersion was evaluated or how it was handled. This is a general methodological consideration, not a reported feature of this analysis.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists one primary outcome measure and one analysis. It does not report a multiplicity adjustment or state whether any additional outcomes or subgroup analyses were part of a prespecified analysis plan. With only the listed primary outcome available, it is not possible to assess the broader family of statistical comparisons that may have been considered.
Missing-data methods are not reported in the registry record. There is no description of imputation, exclusion rules for incomplete records, or the amount and pattern of missing information. The analysis population is described as a 1:2 propensity score-matched cohort, but the numbers entering matching, retained after matching, and included in the outcome analysis are not reported. The total enrollment of 11296 does not establish the size of the matched analysis population.
The endpoint description does specify several censoring criteria: 365 days of follow-up without discontinuation, death, disenrollment, and end of data. These are part of the stated follow-up definition, but the numbers censored for each reason and the handling of censoring in the statistical analysis are not reported. The record also does not provide the person-time totals or follow-up distribution by cohort.
Interim analyses are not reported in the registry record. Neither is a detailed statistical analysis plan, including the treatment of missing data or the approach to any repeated or secondary comparisons. These gaps make it difficult to evaluate multiplicity and data completeness beyond the fact that one primary analysis is listed. They do not provide evidence that a particular method was or was not used; they indicate that the relevant details are not reported in the registry record.
11. What the Result Does, and Does Not, Mean
The registry record establishes that a claims-based comparison of discontinuation rates was planned or conducted in matched cohorts of initiators of two dual-bronchodilator combinations for COPD. It also supplies an operational description based on a 60-day refill gap and person-time at risk. These features define what the comparison is intended to measure.
However, the numerical outcome result is not reported in the registry record. No conclusion about the direction, size, or precision of a difference between the matched cohorts can be drawn from the listed endpoint description alone. The absence of numerical results should not be interpreted as evidence that the rates were equal, or that no events occurred. The record does not report the data needed to make either claim.
Even if rates were available, a nonrandomized matched comparison would require interpretation in light of treatment selection and potential confounding. Matching can address measured characteristics included in the propensity score, but the record does not identify those characteristics or report balance. Residual differences, including unmeasured ones, may remain. The comparison therefore would not by itself establish that the inhaler device caused any difference in persistence.
The outcome is also specific to the claims-based definition. A 60-day gap after supply ends is used, treatment switching and death are excluded as described, and adding another treatment does not count as discontinuation. The measure does not directly establish how consistently an inhaler was used, why a refill was absent, or whether a treatment change was clinically intended.
The registry’s adverse-event fields list zero counts for participants at risk, affected participants, and deaths in both cohort entries, and no serious-event terms are listed. Those fields do not provide enough context to conclude that no adverse events occurred; the record does not report further safety details. Overall, the record describes the question and outcome definition more fully than it reports the results needed to answer the question quantitatively.
12. Why This Trial Matters Statistically
This comparison is statistically notable because it focuses on treatment persistence using routinely collected claims data rather than assigning treatment randomly. Such data can support comparisons of real-world treatment patterns, but the interpretation depends heavily on how cohorts are assembled, how treatment exposure is defined, and how differences between treatment groups are handled.
The 1:2 propensity score-matched design makes the observational nature of the analysis especially relevant. Matching may improve comparability on measured characteristics, but its value cannot be judged without knowing which characteristics were included and whether the matched groups were balanced. The registry does not report those details. A matching label alone is not evidence that all important confounding was removed.
The choice of a rate also matters. By using events per person-time, the outcome accounts for time at risk rather than only counting participants with a recorded discontinuation. Yet a rate requires clear reporting of both events and exposure time to be interpretable. Neither the numerical rates nor their component counts are reported in the registry record, leaving the statistical question unresolved for readers.
The refill-gap rule illustrates how a statistical endpoint operationalizes a clinical concept. Persistence is not directly observed in claims; it is represented through refill records and a specified allowable gap. The treatment-switch and add-on rules further shape which patterns count as discontinuation. These definitions can make analyses reproducible, but they also determine what the result does and does not represent.
Finally, the record distinguishes total enrollment from the matched analysis population in concept, but does not report the size of the latter. Clear reporting of cohort counts, matching diagnostics, event rates, person-time, uncertainty intervals, missing-data handling, and model details would be needed for a fuller statistical assessment. The current record supports discussion of the design and measurement choices, but not a quantitative conclusion about the comparison.
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 principal limitation for interpreting this analysis is that the registry does not report the numerical primary outcome. Without rates, events, person-time, a comparative estimate, or an uncertainty interval, the observed result cannot be characterized. The registry lists one outcome measure and one analysis but provides no numerical estimate with which to evaluate its magnitude or precision.
The study is observational, and allocation is not stated. The record itself notes that the arms are not a like-for-like control comparison. Propensity score matching is described, but the variables used, matching details, balance diagnostics, and matched cohort sizes are not reported. Residual confounding and differences related to treatment selection therefore cannot be assessed.
The total enrollment is 11296, but the number of people included in each matched cohort and in the primary analysis is not reported. The record also lacks cohort-specific event counts, person-years, follow-up summaries, and censoring counts. The stated end-of-follow-up criteria are not accompanied by information about how often each occurred or how censoring was addressed statistically.
The outcome depends on claims-based refill information and a 60-day allowable gap. Such a definition may not capture every aspect of treatment use or the reason a refill is absent. The registry wording around persistence and discontinuation is not fully clear, although it states the gap rule and specifies how switching and adding treatment are treated.
Further limitations include unreported missing-data procedures, interim analyses, multiplicity handling, model details, and sample-size justification for the study design. The adverse-event fields list zeros but provide no serious-event terms or further context. These limitations concern what can be evaluated from the registry record; they do not establish that a particular method was absent from the underlying work.
14. Related Tutorials
15. Related Statistical Calculators
16. Sources
- ClinicalTrials.gov: A Study Based on Medical Records That Looks at the Duration of Use of Two Types of Inhalers With Different Medicines in Patients With Chronic Obstructive Pulmonary Disease (COPD) — NCT03979807.
- Registry record last updated 2021-11-15; status last verified 2021-11.
- Results first posted 2020-07-23.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
17. Record Summary
This completed medical-record study concerns persistence with two dual-bronchodilator combinations in people with COPD. The registry describes a 1:2 propensity score-matched comparison of initiators of olodaterol/tiotropium bromide and umeclidinium/vilanterol identified in the Truven MarketScan Claims database between November 2015 and March 2018. Total enrollment is 11296.
The primary outcome is the rate of discontinuation per 1000 person-years, using a 60-day allowable refill gap and specified censoring criteria. Treatment switching and death are excluded from the persistence definition, and adding another treatment does not count as discontinuation.
The registry record does not report numerical rates, event counts, person-time, a comparative estimate, or an uncertainty interval. It also does not report matched cohort sizes, matching covariates, balance diagnostics, missing-data methods, or a detailed statistical analysis plan. The design and outcome definition can therefore be described, but the primary comparison cannot be quantified from the reported results.