Evaluation of Potential Off-label Use of Dabigatran Etexilate in Europe: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of Potential Off-label Use of Dabigatran Etexilate in Europe (NCT02240654), a registered phase not stated study. The primary outcome is Percentage of Participants With Potential Off-label Use Estimated Among New Users of Dabigatran Etexilate in Each of the Data Sources..

2. Clinical Question

3. Trial Design

Despite being indexed as a clinical study, this is a descriptive, observational, multi-country cross-sectional study, not a randomized trial. It aimed to characterize on-label and potential off-label prescribing, together with other medical characteristics, at the time of the first captured dabigatran etexilate prescription in each database. Treatment was not assigned by investigators, and the registry does not describe a concurrent control group.

The stated data sources were Cegedim Strategic Database in France, the Danish National Databases in Denmark, and the Clinical Practice Research Datalink in the UK. The registry lists an enrollment of 36573. That registry entry does not establish that this number represents a common analytic denominator across the countries, definitions or databases.

New users were defined as patients initiating dabigatran etexilate during the study period who had not used it during the previous year. The outcome description gives country-specific periods since approval of the SPAF indication: France, 01 August 2011 to 30 June 2014; Denmark, 01 August 2011 to 30 November 2013; and the UK, 01 August 2011 to 30 August 2015. SPAF is described as prevention of stroke and systemic embolism in adults with non-valvular atrial fibrillation.

The registry lists the study as completed, with a start date of 2014-11-28 and primary completion date of 2016-05-20. It does not report an assigned intervention, a prospective follow-up comparison, or a statistical analysis plan. The design supports description of prescribing patterns within the recorded data, but does not by itself establish why prescribing occurred or what would have happened under a different treatment strategy.

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: Because it is not possible to rule out duplicate patients across the 2 French panels, all numbers were calculated stratified by data source, and for France further stratified by panel - i.e., not overall.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The study population included new users of dabigatran etexilate in the study period. New users were defined as those patients who initiated treatment with dabigatran etexilate during the study period and who had not used it during the previous year.

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 study population included new users of dabigatran etexilate in the study period. New users were defined as those patients who initiated treatment with dabigatran etexilate during the study period and who had not used it during the previous year.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Potential Off-label Use Estimated Among New Users of Dabigatran Etexilate in Each of the Data Sources., measured over Time period since approval of the SPAF indication in, France: 01 August 2011 to 30 June 2014; Denmark: 01 August 2011 to 30 November 2013; UK: 01 August 2011 to 30 August 2015..

The primary endpoint is the percentage of participants with potential off-label use, estimated separately among new users in each data source. The underlying definition concerns use for a disease or medical condition other than the labelled indications, as documented in the respective data source and interpreted in light of label changes during the study period.

The registry describes two approaches. A broad definition identifies on-label prescribing using disease-indication codes, such as codes for atrial fibrillation. A restrictive definition excludes patients with conditions for which the drug is not indicated, with valvular atrial fibrillation given as an example. The estimates therefore depend partly on which diagnoses or proxies are recorded and on which classification rule is applied.

The outcome description defines prevalence as the number of index prescriptions meeting the off-label definition divided by the total number of index prescriptions. Although the endpoint unit is described as a percentage of participants, its calculation is stated in terms of index prescriptions. The registry record does not clarify how that wording was reconciled in the analysis.

This measure is useful for describing prescribing classification in the data available at treatment initiation. It is not a measure of clinical benefit, harm, adherence, or treatment effect. The record lists a 95% confidence interval as the dispersion measure, but does not report the endpoint estimates or their intervals. Accordingly, the magnitude and precision of potential off-label use cannot be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The study uses an observational, cross-sectional approach. In general, an observational study describes exposures or characteristics as recorded in practice rather than assigning them experimentally. A cross-sectional analysis characterizes a population at a defined point or encounter; in this study, the relevant point is the first captured prescription of dabigatran etexilate in each database. This design is suited to describing prescribing status at initiation, but does not establish a causal effect of treatment.

How is a new user defined?

The registry defines a new user as a patient who initiated dabigatran etexilate during the study period and had not used it during the previous year. This look-back condition helps distinguish treatment initiation from continued use within the observable data. In general, its reliability depends on whether prior treatment is captured in the database and whether the patient’s earlier history is observable. The registry record does not report details about the completeness of that history or how records without adequate look-back information were handled.

How is potential off-label use classified?

The outcome definition is based on a disease or medical condition other than a labelled indication, as documented in the data source, with changes to the label during the study period taken into account. The registry describes a broad approach that uses disease-indication codes, and a restrictive approach that excludes conditions for which the drug is not indicated. These are operational rules for classification: they translate diagnostic records and proxies into a category for analysis. They do not independently verify the clinical reasoning behind each prescription.

In general, diagnostic coding can be incomplete, delayed, or inconsistent across records and data systems. A diagnosis not recorded in the available data may not be equivalent to a diagnosis being absent in clinical care. Conversely, a code or proxy may not fully represent the indication at prescribing. Such measurement issues can lead to misclassification. The registry does not report validation results, coding performance, or the number of prescriptions that could not be classified.

What does a percentage estimate represent?

A proportion is calculated by dividing the number of records that meet the defined condition by the relevant denominator. The registry describes the prevalence calculation using index prescriptions meeting the off-label definition divided by total index prescriptions. It labels the endpoint as a percentage of participants. Because the record does not explain the relationship between prescription-based calculation and participant-based wording, the exact unit of analysis is not fully clear from the registry description. No numerator or denominator is reported, so the percentage cannot be independently checked or interpreted for any country or definition.

What is the role of the confidence interval?

The registry identifies a 95% confidence interval as the dispersion measure. In general, a confidence interval expresses uncertainty around an estimate under a specified statistical method and its assumptions. For a proportion, the method must account for the data structure and the number of observed records. The registry record does not identify the interval calculation method, and it gives no interval limits or point estimates. The interval therefore cannot be used to judge precision or compare estimates.

What can comparisons across data sources tell us?

Descriptive estimates from France, Denmark and the UK could characterize prescribing in those sources and periods. However, a comparison across databases is not automatically a comparison of like-for-like populations. General factors include differences in coding systems, available clinical information, database coverage and calendar time. The study periods differ by country, and the registry identifies country-specific data sources. Without reported estimates and details on harmonization, neither the size nor direction of any cross-source difference is available.

The registry does not report an inferential analysis plan, adjustment methods, subgroup analyses, or a method for handling clustering or over-dispersion. In general, over-dispersion occurs when variability in the data exceeds what a chosen model assumes; accounting for it can affect uncertainty estimates. Whether it was relevant cannot be determined from the registry record. No model should be attributed to the study in the absence of a reported method.

The registry also does not report a comparative treatment effect. General concepts such as randomized allocation, adjusted causal contrasts, or event-time analysis do not describe the endpoint. The endpoint is a classification of prescribing at initiation, and the available registry entry does not provide the results needed to evaluate its observed frequency or statistical uncertainty.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure, potential off-label use among new users, and zero analyses. The description nevertheless refers to two classification definitions: a broad approach using indication codes and a restrictive approach excluding conditions for which the drug is not indicated. It does not report whether both definitions were analyzed, whether one was designated primary for inference, or how estimates across countries and definitions were handled statistically.

Multiplicity refers, in general, to the increased chance of observing an apparently notable result when many statistical tests or outcomes are examined. The registry record does not report a multiplicity adjustment or a prespecified hierarchy of comparisons. Since no analysis results or inferential tests are reported, the effect of multiplicity on conclusions cannot be assessed.

Missing-data procedures are not reported in the registry record. In general, missing or incomplete diagnosis information could affect whether a prescription can be classified as on-label or potentially off-label. The record does not state how incomplete records, unavailable covariates, or unclassifiable prescriptions were treated, nor whether the frequency or pattern of missingness was examined.

The population is defined through new use and a previous-year non-use period, but the registry does not describe analytic population exclusions or the handling of patients whose prior treatment history was not observable. It also does not provide country-specific denominators. Consequently, the relationship between the recorded enrollment and the population contributing to each endpoint estimate is not reported in the registry record.

The study is observational and does not have randomized analysis populations such as intention-to-treat or per-protocol groups described in the record. No adjusted analysis population, sensitivity analysis, or subgroup analysis is reported. These omissions limit assessment of how robust a potential prevalence estimate would be to missing information or alternative analytic choices.

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

The endpoint is intended to describe the frequency of prescriptions classified as potentially off-label among new dabigatran etexilate users in each data source. It is a measure of prescribing classification based on documented diagnoses or proxies and the relevant label, not a direct measure of clinical outcomes. The registry provides the study question and classification framework, but does not report the estimated percentages.

As a result, the record does not establish whether potential off-label use was common or uncommon, whether estimates differed across France, Denmark and the UK, or how precisely any estimate was measured. It also does not report the counts or denominators needed to judge the information underlying the intended percentages. The listed confidence-interval measure cannot resolve this because no interval estimates are present.

Even with numerical estimates, an observational description would not establish why a clinician prescribed the medicine or whether a particular prescription was clinically appropriate. The classification depends on recorded information and the chosen definition. The registry record does not report a review of individual clinical decisions, nor does it measure treatment effectiveness or safety for the endpoint described.

The study also cannot support a causal comparison between dabigatran etexilate and another treatment: no assigned comparator is described. Comparisons among databases, if made using other reported results, would need to account for source-specific recording practices and different observation periods. The registry entry itself does not supply findings that would permit such a comparison.

Thus, the main contribution documented in the record is methodological: it specifies a multi-country effort to characterize potential off-label prescribing among new users and explains alternative classification rules. The actual prevalence findings and their uncertainty are not reported in the registry record. No treatment recommendation or conclusion about individual prescribing follows from the information recorded.

12. Why This Trial Matters Statistically

Studies of prescribing patterns can address a different question from studies of treatment efficacy. This study’s stated focus is how new prescriptions are classified relative to labelled indications in routine data. That makes its endpoint a matter of measurement as much as counting: the result depends on what diagnoses are recorded, how indication codes or proxies are interpreted, and which definition is applied.

The multi-country design also highlights a general issue in observational statistics. Data sources can differ in coverage and recording practices, while the study periods themselves differ across countries. Descriptive estimates can be informative within each source, but comparisons require careful attention to how populations and classifications align. The registry record describes the sources and periods but does not report harmonization details or estimates with which to examine cross-source consistency.

The study is statistically important to interpret as a prevalence description rather than a comparative trial. It has no randomized control group, and its primary endpoint is not a clinical benefit or harm outcome. These distinctions matter because a prescribing percentage cannot be read as a treatment effect, nor can an association or classification alone establish the reason for a prescribing decision.

The record’s lack of outcome estimates is consequential. A stated endpoint and confidence-interval measure do not substitute for reported results, denominators, or uncertainty intervals. Without those findings, readers cannot determine prevalence, precision, or differences between the included data sources. The record therefore documents a relevant descriptive research question and its proposed framework, while leaving the numerical evidence unavailable in the registry record.

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 registry entry is that it does not report the primary outcome estimates, confidence intervals, counts, or country-specific denominators. Although the registry lists a 95% confidence interval as the dispersion measure, there are no interval values with which to assess precision. It also lists zero analyses, so the reported record does not contain an inferential result.

The endpoint description uses participant language while defining the calculation in terms of index prescriptions. The registry does not clarify whether each participant contributed only one index prescription or how the analysis reconciled those units. This limits clarity about the denominator and interpretation of the percentage.

Classification depends on recorded diagnoses or proxies and on the chosen broad or restrictive definition. The registry does not report validation of those classifications, coding completeness, or the number of records that could not be classified. It also does not describe how label changes were operationalized beyond stating that they were taken into account.

The data sources and calendar periods differ across countries. The registry does not report steps taken to make coding and population definitions comparable, or analyses addressing potential source-specific differences. It also does not report missing-data procedures, analytic exclusions, sensitivity analyses, or subgroup methods. These are limits on what can be assessed from the registry entry; they do not establish that any particular problem occurred.

Finally, this is an observational cross-sectional description with no assigned comparator. It cannot on its own establish a causal treatment effect or determine the clinical appropriateness of individual prescriptions. The registry entry also does not specify a statistical analysis plan, sample-size justification, or interim analyses; these details are not reported in the registry record.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: Evaluation of Potential Off-label Use of Dabigatran Etexilate in Europe — NCT02240654.
• Registry record last updated 2017-12-15; status last verified 2017-11.
• Results first posted 2017-10-16.
• 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

NCT02240654 is a completed, descriptive, observational, multi-country study of potential off-label dabigatran etexilate use among new users. The listed sources are Cegedim Strategic Database in France, the Danish National Databases, and the Clinical Practice Research Datalink in the UK. New use required initiation during the study period with no use during the previous year.

The primary endpoint was the percentage of new users with potential off-label use, using a broad indication-code definition and a restrictive definition that excludes conditions for which the drug is not indicated. Country-specific observation periods are stated in the endpoint description. The registry lists enrollment as 36573, a start date of 2014-11-28, and primary completion on 2016-05-20.

The registry describes the endpoint and identifies a 95% confidence interval as the dispersion measure, but does not report estimates, confidence intervals, denominators, or outcome counts. It lists one outcome measure and zero analyses. Thus, numerical results and their precision are not reported in the registry record.