Drug Persistence/Adherence in Patients Treated With Dabigatran or VKA for Stroke Prevention in Non Valvular Atrial Fibrillation (SPAF): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Drug Persistence/Adherence in Patients Treated With Dabigatran or VKA for Stroke Prevention in Non Valvular Atrial Fibrillation (SPAF) (NCT02240667), a registered phase not stated study. The primary outcome is Percentage of Patients Treated With Anticoagulation Initially Started at the 12 Month.

2. Clinical Question

The study also planned to collect prescriptions, adverse events and reasons for definitive discontinuation during quarterly visits over 12 months. Adherence was to be assessed at the 6-month visit using the Morisky Score. These are related but distinct questions: persistence concerns continued treatment over time, while adherence concerns how consistently a patient follows treatment. The registry record does not report the primary endpoint estimates, so the comparative persistence result cannot be assessed in the registry record recorded.

3. Trial Design

The registry describes a 12-month follow-up study of patients with atrial fibrillation receiving dabigatran etexilate or a vitamin K antagonist for stroke prevention. Quarterly visits were planned. Prescriptions, adverse events and, where applicable, reasons for definitive treatment discontinuation were to be collected. At the 6-month visit, adherence was to be assessed using the Morisky Score. The stated study status is completed; the record gives a start date of 2014-09-18 and a primary completion date of 2017-12-31.

The registry reports enrollment of 1506, while its brief summary describes 1600 patients as the intended follow-up cohort. Those are different registry statements and should not be conflated. The record does not explain the difference. It also does not state how patients came to receive one treatment rather than the other. The comparison is explicitly not randomized, and the allocation method is not stated.

For the primary endpoint, the record identifies a population in the TS matched 1:1 using propensity scores, with the aim of improving comparability between the dabigatran etexilate and VKA groups. Matching is an analytical approach to address measured differences; it does not make treatment assignment random, nor does it guarantee that all relevant differences have been removed. The registry does not report the matching variables or diagnostics in the registry record recorded.

The study therefore combines follow-up of treatment use with a matched comparison of persistence. The interpretation depends on the endpoint definition, how follow-up and discontinuations were handled, and how well the matched groups were comparable. Details of these procedures beyond the stated definition and matching approach 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 patients were enrolled from 1000 sites from Germany. Approximately 2/3 (about 670) will be community-based primary care internists/general practitioners and approximately 1/3 (about 330) community-based cardiologists across Germany.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Patients in the TS who were matched 1:1 based on propensity score matching in order to ensure comparability of outcome variables between both treatment groups (dabigatran etexilate and VKA).

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 Treated set (TS) includes all patients with main diagnosis of nonvalvular AF (=IC 2) with at least one evaluable efficacy endpoint (being on treatment is an efficacy endpoint).

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Patients Treated With Anticoagulation Initially Started at the 12 Month, measured over 12 month (Visit 5).

The primary endpoint is the percentage of patients treated with the anticoagulant initially started at the 12-month visit. It is defined as a Kaplan-Meier estimate at 12 months for persistence, reported separately for dabigatran etexilate and VKA. Persistence is the time from the documented treatment start at visit 1 until documented permanent discontinuation of dabigatran etexilate or VKA therapy.

This endpoint captures continuation of the original treatment over time, rather than simply whether a patient had a prescription at one isolated visit. The Kaplan-Meier approach is generally used for time-to-event outcomes when follow-up times can differ and some patients have no event observed during their available follow-up. In this setting, the event is permanent discontinuation as defined by the registry. The estimate at 12 months represents the estimated proportion still persistent at that time under the analysis definition; it is not by itself a measure of adherence to each dose.

Persistence is relevant to a study of long-term treatment use because discontinuation is a distinct outcome from adherence at a particular assessment. The registry also lists adherence at the 6-month visit as a secondary outcome, using the Morisky Score. The primary endpoint estimates and their confidence intervals are not reported in the registry record, so the level and precision of persistence in either group cannot be described from this record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint uses a Kaplan-Meier estimate of persistence at 12 months. In general, Kaplan-Meier methods estimate the probability of remaining free of an event over time while accounting for the fact that participants may have different lengths of observed follow-up. A participant whose follow-up ends without a documented event may be censored, meaning that their later event status is unknown to the analysis rather than treated as if the event occurred. The registry defines the event as permanent discontinuation, but the specific censoring rules and handling of incomplete follow-up are not reported in the registry record.

What does persistence mean in this analysis?

The registry defines persistence as the period between treatment initiation and permanent discontinuation. Initiation is the documented treatment start at visit 1; discontinuation is the documented permanent discontinuation of dabigatran etexilate or VKA. The endpoint is therefore tied to recorded treatment events. Temporary interruptions, switches, or other treatment histories cannot be classified from the posted endpoint definition alone unless the registry's operational rules specify how they were handled; such rules are not reported in the registry record.

The endpoint is expressed as a percentage of participants and estimated at the 12-month time point. In general, a 95% confidence interval describes statistical uncertainty around an estimate under the analysis assumptions. A narrower interval indicates greater precision than a wider one, but neither the interval nor the Kaplan-Meier estimate is reported in this registry record. It is not possible to describe the precision of either group's estimate or judge how compatible the data are with different persistence levels.

Why use propensity-score matching?

Propensity-score matching was used to form a primary endpoint population with 1:1 matching between patients receiving dabigatran etexilate and VKA. In general, a propensity score represents the estimated probability of receiving a particular treatment given measured characteristics. Matching patients with similar scores is intended to improve comparability on the characteristics used to construct the score. It does not reproduce randomization: unmeasured factors can still differ, and the quality of the match depends on the variables and implementation.

The record does not report the variables included in the propensity score, balance diagnostics, matching algorithm, or the number of patients retained in the matched population. Thus the stated matching objective can be described, but the degree to which it achieved balance cannot be assessed from the registry record. The registry also does not state whether the primary estimate was further adjusted after matching.

How should the planning illustrations be read?

In general, a sample-size calculation asks how much information might be needed to detect a specified effect under stated assumptions. The planning illustrations described for this study use a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation; the target effect is stated as a fraction of the observed effect. These are planning illustrations based on an observed effect, not observed power and not the trial's own design justification. The record does not report a sample-size justification from the study protocol. The assumptions are not evidence that the realized study had the stated power, and a planning calculation cannot substitute for the observed estimate and its confidence interval.

The registry describes a time-to-event endpoint, while the planning assumptions distinguish time-to-event calculations from other outcomes. In general, time-to-event sample-size methods depend on the number of events as well as allocation and the target effect. No additional planning values are reported in the registry record details, so no numerical sample-size conclusion can be drawn.

What is not available for interpretation?

The primary endpoint result, its 95% confidence interval, and a comparative test result are not reported in the registry record. The record identifies one analysis entry, but details of its statistical model beyond the stated Kaplan-Meier estimation and propensity-score matching are not reported. There is no basis to infer adjustment variables, interim analyses, prespecified decision rules or other model procedures. In general, statistical methods should be interpreted alongside the analysis population, event definitions, follow-up completeness and uncertainty; several of those details are not reported.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Dabigatran (Dabigatran vs VKA)", "VKA (Dabigatran vs VKA)")
n      <- c(476, 476)
events <- c(426, 426)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Dabigatran (Dabigatran vs VKA)  476  426  89.5  86.4  92.1  86.4  91.9  86.4  91.9
ARM  VKA (Dabigatran vs VKA)  476  426  89.5  86.4  92.1  86.4  91.9  86.4  91.9

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary persistence endpoint and secondary outcomes that include adherence categories at the 6-month visit and reasons for definitive treatment discontinuation. It reports three outcome measures, but the available record details do not specify a multiplicity adjustment, a hierarchy of secondary hypotheses, or how results across outcomes were intended to be interpreted. Those features are not reported in the registry record. In general, testing multiple outcomes can increase the chance of observing an apparently notable result by chance; whether and how that issue was addressed cannot be determined from this record.

Missing-data procedures are also not reported in the registry record. The persistence analysis uses a Kaplan-Meier estimate, for which handling incomplete observation and censoring is important. The registry defines initiation and permanent discontinuation, but does not report censoring rules, loss-to-follow-up handling, or sensitivity analyses for missing information. The planned quarterly visits and 6-month adherence assessment identify collection time points, not the completeness of those data.

The primary population is described as the TS population matched 1:1 by propensity score. The registry does not report the matched population size, the number excluded or unmatched, or the amount of missing covariate information. It does not state whether an intention-to-treat or per-protocol framework was used; those labels should not be inferred from the endpoint description. The relationship between the enrollment figure and the intended cohort described in the study summary is not explained.

Consequently, the registry record does not support assessment of how missingness, exclusions or multiplicity may have affected the reported results. In general, these details matter because they can influence both the estimate and its uncertainty, particularly when follow-up or analysis populations differ between treatment groups.

12. Safety Results

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

Deaths

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

Serious events with the highest arm-level rate: cardiac failure (0.8%), acute coronary syndrome (0.6%), atrial fibrillation (0.5%), cardiac arrest (0.5%), sepsis (0.4%).

The registry lists serious adverse events for the dabigatran group as 45 among 771 at risk, and for the VKA group as 39 among 650 at risk. It also lists deaths as 10 and 5, respectively. The registry's other adverse-event count is listed as 0 in each group. These are descriptive registry figures; they should not be read as proof that treatment caused or prevented an event.

The record also lists the following serious-event terms with percentages: Cardiac failure, 0.8%; Acute coronary syndrome, 0.6%; Atrial fibrillation, 0.5%; Cardiac arrest, 0.5%; Sepsis, 0.4%; and Angina pectoris, 0.3%. The registry entries do not specify the denominator or further context for these term percentages, so they should not be used to infer group-specific risks.

In general, event counts require context such as the event definition, observation time, completeness of ascertainment and analysis population. The record does not report a statistical comparison for these safety figures or establish causality. The at-risk counts shown for these summaries should not be assumed to be the same as the matched population used for the primary persistence endpoint.

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

The central question is whether patients remained on their initially started anticoagulant through the 12-month visit, with persistence estimated using Kaplan-Meier methods. The registry provides the endpoint definition, describes propensity-score matching, and states that the study is completed. It does not provide the actual primary endpoint estimates or their 95% confidence intervals in the registry record details. Accordingly, the direction, magnitude and precision of any difference in persistence cannot be stated.

The study is not a randomized comparison, and treatment allocation is not stated. Matching was intended to improve comparability between groups, but it cannot ensure that groups are alike in every relevant respect. Any comparative estimate would need to be interpreted with those design features in mind. The registry entries do not report matching diagnostics or enough detail to evaluate the success of that process.

The endpoint also has a specific scope: it concerns continued use of the initially started anticoagulant, based on documented initiation and permanent discontinuation. It is not the same as adherence measured at a single visit. The registry planned a Morisky Score assessment at 6 months, but the results for that outcome are not reported in the registry record. Nor are the listed reasons for definitive discontinuation accompanied by result values.

The serious adverse-event and death counts provide descriptive safety information, but they do not establish a causal relationship or a comparative safety effect. The registry record does not report a safety analysis or uncertainty estimates for those figures. More generally, a completed study record is not itself evidence of a particular treatment outcome. Estimates, uncertainty, definitions and analysis populations are needed to judge what the observed data support.

In general, statistical findings describe evidence in the analyzed population under the methods used. They do not alone establish what would happen for every patient or explain the reasons behind treatment persistence or discontinuation. With the primary numerical result absent from this record, the justified interpretation is limited to the study question, endpoint definition and documented design.

14. Why This Trial Matters Statistically

Persistence is a time-dependent outcome, so the choice to estimate it with Kaplan-Meier methods is statistically relevant. Unlike a simple status measured at one visit, a time-to-discontinuation analysis uses the timing of documented events and can accommodate different follow-up durations under its assumptions. The endpoint's explicit definition also matters: it distinguishes permanent discontinuation from other aspects of medication-taking behavior.

The registry pairs this endpoint with a non-randomized comparison and propensity-score matching. That makes the distinction between statistical adjustment and random assignment important. Matching can improve balance on measured characteristics, but it cannot eliminate bias from unmeasured differences. It also changes the analysis population to the matched patients described in the registry, rather than necessarily representing everyone enrolled or intended for follow-up.

The study asks about both persistence and adherence, which are related but not interchangeable. Persistence is defined by time to permanent discontinuation; adherence was to be assessed at the 6-month visit using the Morisky Score. Keeping those outcomes conceptually separate helps avoid interpreting one as a substitute for the other. The planned quarterly follow-up and collection of reasons for definitive discontinuation add context to the study question, though the corresponding results are not reported in the record details.

Finally, the registry illustrates why statistical interpretation requires more than a completed status or an endpoint label. The primary estimate and uncertainty are necessary for judging the observed result, while analysis-population details and missing-data handling help assess its reliability. The registry record describes important design features but does not report the primary numerical result or several implementation details. That limits what can be concluded from the record without diminishing the value of clearly defining the outcome and comparison.

15. 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 main limitation for interpreting this registry entry is that it does not report the primary persistence estimates or their 95% confidence intervals. The record therefore does not establish the observed level of persistence in either group or the precision of a comparison. It also does not report a p-value or a numerical comparative effect in the registry record.

The comparison was not randomized, and the allocation method is not stated. Propensity-score matching is described, but the registry does not report the score variables, matching details, balance diagnostics, or matched population size. Residual differences between groups cannot be ruled out, and the success of matching cannot be evaluated from the registry record.

Several implementation details are not reported, including censoring and loss-to-follow-up rules, missing-data procedures, sensitivity analyses and whether additional adjustment was used. The record also does not state a statistical analysis plan, interim analyses, or a sample-size justification. Those details should not be inferred from the completed status or from planning illustrations.

The registry gives both enrollment of 1506 and an intended cohort of 1600 in its brief summary, without explaining the relationship. The primary endpoint population is the matched TS population, while the adverse-event summaries list separate at-risk counts. The record does not provide enough information to reconcile these populations or denominators.

Finally, the serious-event counts are descriptive and do not include a reported comparative safety analysis. The listed serious-event term percentages lack denominator context in the recorded details. These limitations constrain conclusions about comparative effectiveness, safety, and generalizability.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Drug Persistence/Adherence in Patients Treated With Dabigatran or VKA for Stroke Prevention in Non Valvular Atrial Fibrillation (SPAF) — NCT02240667.
• Registry record last updated 2019-04-19; status last verified 2019-04.
• Results first posted 2019-04-19.
• 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.

19. Record Summary

This completed registry study concerns dabigatran etexilate and vitamin K antagonists in patients with atrial fibrillation receiving anticoagulation for stroke prevention. The study summary describes quarterly follow-up over 12 months, collection of prescriptions, adverse events and reasons for definitive discontinuation, and a Morisky Score adherence assessment at 6 months.

The primary endpoint is persistence with the initially started anticoagulant at the 12-month visit, estimated using Kaplan-Meier methods. The primary endpoint population is described as the TS population matched 1:1 by propensity score. The comparison is not randomized, and treatment allocation is not stated.

The registry reports enrollment of 1506 and an intended follow-up cohort of 1600 but does not explain the difference. The primary endpoint estimates and confidence intervals are not reported in the registry record details. Serious adverse-event and death counts are listed, but no comparative safety analysis is reported. The entry supports a description of the study question and design, while leaving the principal numerical result and several analysis details unreported.