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 Safety and Effectiveness of Apixaban in Very Elderly Patients With Non-valvular Atrial Fibrillation (NVAF) Compared to Warfarin Using Administrative Claims Data (NCT05438888), a registered phase not stated study. The primary outcome is Incidence Rate (Per 1,000 Participant-Year) of Composite of Stroke and Systemic Embolism (SE): Balanced Cohorts.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT05438888 |
| Phase | Phase not stated |
| Condition | Non-valvular Atrial Fibrillation |
| Allocation | Not Stated |
| Design model | Not Stated |
| Masking | Not Stated |
| Enrollment | 77814 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Pfizer |
| Results posted | Yes |
2. Clinical Question
Population
Eligible participants with non-valvular atrial fibrillation whose information was registered in the MDV database and observed for the study. The study describes its focus as very elderly patients. A specific age threshold and further eligibility details are not reported in the registry record.
Intervention
Apixaban treatment, represented by the apixaban cohort. The registry describes participants as patients with non-valvular atrial fibrillation treated with apixaban.
Comparator
Warfarin treatment, represented by the warfarin reference cohort. The registry describes participants as patients with non-valvular atrial fibrillation treated with warfarin. Allocation to treatment is not stated, and the record cautions that the arms are not a like-for-like control comparison.
Primary question
Among the balanced cohorts, what is the incidence rate per 1,000 participant-years of the first occurrence of the composite of stroke and systemic embolism after treatment initiation? The endpoint includes ischemic and hemorrhagic stroke as well as systemic embolism. A numerical estimate for either cohort, and a numerical comparison between them, are not reported in the registry record.
The objective also includes examining whether higher age-related risk factors affect the relative risk of apixaban compared with warfarin through subgroup analyses. The registry identifies this as an objective, but subgroup definitions, estimates and results are not reported in the registry record. It is therefore not possible to tell from the record which subgroups were assessed or how their results compare.
This is a question about outcomes observed in treatment cohorts identified from administrative claims, not a question assigned by a stated randomization process. The comparison can describe associations in the observed data, but its interpretation depends on how treatment groups were formed, how their characteristics were balanced, and how completely claims capture the outcomes. Details beyond the stated weighting approach are not reported in the registry record.
3. Trial Design
This completed study is registered as NCT05438888 and is sponsored by Pfizer. The registry reports an enrollment of 77814. Its stated objective is to investigate the safety and effectiveness of apixaban compared with warfarin in very elderly patients with non-valvular atrial fibrillation, using administrative claims data. The study start date is 2022-07-01, and the primary completion date is 2022-10-19.
The two cohorts are defined by treatment: a warfarin reference cohort and an apixaban cohort. The record does not state how treatment was allocated and identifies the comparison as not randomized. It also cautions that the arms are not a like-for-like control comparison. Thus, treatment assignment should not be interpreted as a random process. Differences between groups may reflect patient characteristics or care patterns as well as treatment.
The primary outcome is an incidence rate for the first occurrence of a composite of stroke and systemic embolism after the index date. The index date is defined as initiation of warfarin or apixaban. The follow-up period starts on the next day and ends at the first of the target outcome, treatment discontinuation, switching treatment, or withdrawal from the database. The endpoint description gives a data observation period from 26-Feb-2013 to 31-Dec-2021, described as approximately 8 years 10 months; extracted data were evaluated in approximately 3.5 months of the study.
The primary analysis is described as using inverse probability treatment weighting with stabilized weights to balance participant characteristics among reporting groups. The population is described as eligible participants registered on the MDV database whose data were observed in the study. Further details on eligibility, the weighting model, the definition of balanced cohorts, and the handling of incomplete records are not reported in the registry record. The record gives no numerical primary outcome estimates or confidence limits.
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: Data of eligible participants were extracted from Medical Data Vision Company Limited (MDV Co. Ltd.) database for duration of 26-Feb-2013 to 31-Dec-2021. Extracted data was evaluated for objectives of this study in approximately 3.5 months of this study.
Overall Study
| Stage | Warfarin Cohort | Apixaban Cohort |
|---|---|---|
| STARTED | 39936 | 37878 |
| COMPLETED | 39936 | 37878 |
| NOT COMPLETED | 0 | 0 |
Analysis population for the primary outcome, as defined in the registry: Eligible participants registered on MDV database, whose data was observed in the study. Analysis was performed using inverse probability treatment weighting with stabilized weights (s-IPTW) method to balance participant characteristics among reporting groups. Here, "Overall Number of Participants Analyzed" is the number of participants after application of s-IPTW method to raw numbers and is different from the actual participants included in the reporting arm.
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: All eligible participants whose data were extracted from database and observed in the study.
| Characteristic | Unit | Warfarin Cohort | Apixaban Cohort | Total |
|---|---|---|---|---|
| Participants analysed | 39936 | 37878 | 77814 | |
| Age, Continuous | Years | 85.3 (SD 4.15) | 85.7 (SD 4.22) | 85.5 (SD 4.19) |
| Sex: Female, Male: Female | Participants | 18804 | 19171 | 37975 |
| Sex: Female, Male: Male | Participants | 21132 | 18707 | 39839 |
| Race and Ethnicity Not Collected | Participants | — | — | 0 |
6. Primary Endpoint
The registry lists the primary outcome as Incidence Rate (Per 1,000 Participant-Year) of Composite of Stroke and Systemic Embolism (SE): Balanced Cohorts, measured over Follow-up period during data observation period from 26-Feb-2013 to 31-Dec-2021 (approximately 8 years 10 months); extracted data evaluated in approximately 3.5 months of this study.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | 95% Confidence Interval |
| Unit | Events Per 1000 Participant-Years |
| Endpoint classification | Event-Rate Endpoint |
The primary endpoint is the incidence rate per 1,000 participant-years of the first occurrence of a composite of stroke and systemic embolism after the index date. Stroke includes ischemic and hemorrhagic stroke. Events were labelled using International Classification of Diseases, Tenth Revision, diagnosis codes. The endpoint therefore combines different event types into a single outcome measure; the registry does not report the separate contribution of each event type to the primary result.
An incidence rate expresses the number of events relative to accumulated time at risk, rather than simply the proportion of participants who experienced an event. Participant-years account for differing lengths of follow-up, which is relevant when observation ends at different times for reasons such as an outcome, treatment discontinuation, treatment switching, or database withdrawal. This is a general explanation of the measure; the registry does not give the cohort-specific event counts or exposure-time totals.
A composite can capture several clinically important outcomes in one measure, but its overall rate does not reveal whether the component outcomes occurred at similar rates or had similar importance. The record also does not report separate primary estimates for ischemic stroke, hemorrhagic stroke, or systemic embolism. Numerical rates and confidence intervals for the endpoint are not reported in the registry record, so the magnitude and precision of the observed association cannot be described from this information.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record describes a comparison of apixaban and warfarin cohorts after balancing participant characteristics with inverse probability treatment weighting using stabilized weights. It does not provide the primary incidence rate for either cohort, a contrast between the cohorts, or numerical confidence intervals. Without those estimates, the direction, magnitude, and precision of the primary comparison cannot be assessed. No conclusion about the observed relative or absolute outcome pattern can be drawn from the endpoint description alone.
The endpoint is an event rate per 1,000 participant-years. In general, a rate places events in relation to observed time at risk, while a confidence interval expresses uncertainty around an estimated quantity under the assumptions of the analysis. The registry identifies a 95% confidence interval as the dispersion measure, but no interval limits are reported in the registry record. It is therefore not possible to judge how narrowly or broadly the primary rate or comparison is estimated.
The weighting approach is intended to balance measured participant characteristics between treatment groups. In general, inverse probability treatment weighting reweights observations according to the probability of receiving the treatment actually received, based on measured characteristics. Stabilized weights are a form of weighting intended to moderate the weights. These methods can address measured differences represented in the weighting model, but they do not by themselves ensure that the cohorts are comparable in every relevant respect. The registry does not report enough detail to assess which characteristics were included, how balance was evaluated, or whether extreme weights were examined.
The nonrandomized nature of the comparison matters for interpretation. Treatment choices in administrative claims data can be associated with patient health, prescribing patterns, access to care, or other factors. If such factors are related to both treatment and outcome, residual confounding may remain after weighting. This is a general concern for observational comparisons, not evidence that any particular bias occurred in this study. The registry itself notes that the arms are not a like-for-like control comparison.
The registry describes subgroup analyses concerning age-related risk factors, but it reports no subgroup definitions or estimates. It also names secondary outcomes involving cardiogenic cerebral embolism, ischemic stroke, intracranial hemorrhage, gastrointestinal bleeding, and intraocular bleeding in balanced cohorts, without numerical results in the registry record. The record therefore establishes the planned scope of outcomes but not their observed values. It cannot support a comparison of the primary endpoint with secondary outcomes or subgroup patterns.
Enrollment is reported as 77814, but the registry does not state the number of participants contributing to each weighted cohort or the amount of follow-up observed for the primary analysis. The enrollment figure should not be treated as the denominator for the event-rate calculation without those details. The registry also does not report event counts for the primary endpoint. As a result, the reader cannot determine from this record how many outcome events informed the estimates or how much information was available for the weighted comparison.
Any planning illustration based on an observed effect is distinct from an observed result. The record describes planning assumptions of a two-sided significance level of 0.05, equal allocation, a normal approximation, and effects defined as a stated fraction of the observed effect. These are assumptions for a planning calculation, not a report of observed power. Numerical planning results and the underlying observed effect are not reported in the registry record.
9. Statistical Methods Explained
What does an incidence rate represent?
As a general statistical concept, an incidence rate relates the number of events to the amount of time participants contribute while at risk. The primary outcome is expressed per 1,000 participant-years. This differs from a simple percentage: participants can contribute different amounts of follow-up, and a rate uses the total time observed in its denominator. The registry defines follow-up as beginning the day after treatment initiation and ending at the first of the target outcome, treatment discontinuation, switching treatment, or withdrawal from the database. It does not report the event counts or total participant-time used in the rate calculations.
The endpoint is the first occurrence of a composite of stroke and systemic embolism. For this endpoint, stroke includes ischemic and hemorrhagic stroke, identified using International Classification of Diseases, Tenth Revision, diagnosis codes. A composite can make an overall outcome measure that includes multiple event types, but the resulting rate does not show how much each component contributes. Separate component results for the primary endpoint are not reported in the registry record.
What does the weighting approach do?
The registry states that the primary analysis used inverse probability treatment weighting with stabilized weights to balance participant characteristics among reporting groups. In general, inverse probability weighting uses treatment probabilities estimated from recorded characteristics to give participants weights in an analysis. The purpose is to create a weighted comparison in which measured characteristics are more balanced than they may be in the original cohorts. Stabilization is a weighting refinement; it does not turn an observational comparison into a randomized trial.
Weighting can only balance characteristics that have been measured and incorporated appropriately. It cannot directly adjust for unrecorded characteristics, measurement errors, or factors absent from the model. The registry record does not report the covariates used, how treatment probabilities were estimated, how balance was assessed, or how the analysis addressed unusually influential weights. Consequently, the adequacy of the weighting cannot be evaluated from the record alone.
Why does the observational setting matter?
In a randomized study, treatment assignment is determined by a random process; this record does not describe such allocation. In this claims-based comparison, participants are grouped by treatment received. As a general matter, this can leave differences between groups that are associated with both treatment choice and outcomes. Weighting is one way to address measured differences, but a weighted association still requires cautious interpretation as an observational result. The registry explicitly warns that the arms are not a like-for-like control comparison.
The source is described as administrative claims data from the MDV database. Claims-based outcome identification depends on the recorded diagnosis information and the definitions used to classify events. The registry says diagnosis codes were used for stroke events, but it does not report validation procedures, coding algorithms in detail, or how outcome misclassification was assessed. These details are relevant to judging the quality of outcome measurement, but they are not reported in the registry record.
How should the confidence interval be read?
The primary outcome names a 95% confidence interval as its measure of dispersion. In general, a confidence interval communicates uncertainty around an estimate under the assumptions of the statistical method. A narrower interval indicates greater precision in the estimate than a wider interval, but interpretation also depends on the design, data quality, and model assumptions. Since the interval limits are not reported in the registry record, the precision of the primary estimate cannot be assessed.
A confidence interval is not a guarantee that the true value lies within its particular limits, nor does its width resolve potential confounding or errors in data capture. It describes statistical uncertainty conditional on the analysis and its assumptions. The registry does not state further details of the variance estimation for the weighted rates, so those methods cannot be evaluated.
What do the planning assumptions mean?
The registry describes planning illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation, a normal approximation, and the effect to detect set as a stated fraction of the observed effect. In general, these are inputs for illustrating how sample requirements can vary with a target effect and statistical assumptions. For time-to-event calculations, the record says the planning quantity is the total number of events; other calculations are expressed as participants per arm.
Such planning illustrations are not observed study results and should not be described as observed power. They do not show that the study had a particular probability of detecting an effect after the data were collected. Numerical planning outputs are not reported in the registry record, and the record does not report the original sample-size justification or whether these illustrations reflect the original study design.
The registry does not report an interim-analysis plan, a complete statistical analysis plan, or the details of the primary rate model. Those omissions prevent a fuller assessment of model specification, testing procedures, and any prespecified analysis decisions.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists a primary composite endpoint and secondary outcomes for cardiogenic cerebral embolism, ischemic stroke, intracranial hemorrhage, gastrointestinal bleeding, and intraocular bleeding in balanced cohorts. It also describes subgroup analyses related to age-related risk factors. Numerical results for these secondary and subgroup analyses are not reported in the registry record.
The record does not report whether any adjustment for multiple comparisons was planned or applied. When several outcomes or subgroups are examined, the chance of observing an apparently unusual result somewhere among the analyses can increase. As a general statistical issue, multiplicity procedures may be used to control or describe that risk, but no such procedure can be attributed to this study from the available record. The complete testing hierarchy and distinction between confirmatory and exploratory analyses are not reported.
Missing-data methods are also not reported in the registry record. The study population is described as eligible participants registered in the MDV database whose data were observed, and follow-up can end with withdrawal from the database. The record does not state how missing or incomplete treatment, covariate, follow-up, or outcome information was handled, or whether any sensitivity analyses assessed its impact.
The analysis population is described in general terms, and the primary analysis uses stabilized inverse probability treatment weighting. The record does not provide cohort-specific analysis counts, exclusions, or the number of participants contributing to each outcome. Although enrollment is reported as 77814, that figure alone does not establish the denominator for the primary rate or each secondary analysis. Further details on analysis populations and participant flow are not reported in the registry record.
11. What the Result Does, and Does Not, Mean
The registry identifies an important comparative question: the incidence rate of a composite of stroke and systemic embolism among apixaban and warfarin cohorts of very elderly patients with non-valvular atrial fibrillation. It also specifies the endpoint definition, follow-up rules, claims-data source, and a weighting approach intended to balance participant characteristics. These details clarify what the planned comparison concerns.
The record does not report numerical primary outcome rates, a treatment contrast, confidence interval limits, or event counts. It therefore does not establish from the information recorded whether one cohort had a higher or lower observed rate, how large any difference was, or how precise an estimate might be. Secondary outcome and subgroup results are also not reported in the registry record. The stated study objective is not itself evidence of an outcome finding.
Even if numerical estimates are considered from another source, a claims-based nonrandomized comparison should be interpreted as observational. In general, weighting can improve balance on measured characteristics but cannot rule out confounding from unmeasured factors or fully resolve errors in claims-based outcome classification. The registry’s warning that the arms are not a like-for-like control comparison reinforces the need to distinguish an adjusted association from a causal effect.
The findings described by this record cannot establish how the comparison applies to people outside the defined database population, nor can they answer questions about outcomes or populations not captured by the stated endpoints and eligibility description. The registry does not report a specific age threshold, detailed eligibility criteria, or enough numerical results to judge the breadth or precision of any conclusion.
Accordingly, this record supports a description of the study question and its stated analytical approach, but not a conclusion about comparative safety or effectiveness. Numerical outcome findings, detailed methods, and limitations of the underlying data would be needed for a fuller assessment.
12. Why This Trial Matters Statistically
This study is statistically notable for using administrative claims data to compare treatment cohorts rather than describing a randomized allocation. That distinction affects what the estimates can mean. A weighted comparison may address measured differences between groups, but the result remains sensitive to how treatment was selected and to characteristics not captured in the weighting approach. The registry’s explicit caution that the arms are not a like-for-like control comparison is central to interpretation.
The endpoint is an event rate per 1,000 participant-years, which reflects both event occurrence and observed time at risk. Follow-up ends at specified events or changes in treatment and database observation. These rules define which time contributes to the denominator and therefore matter to the rate’s meaning. The record describes the rules but does not report event counts, total exposure time, or numerical rates.
The use of a composite endpoint also has statistical implications. Combining stroke and systemic embolism offers a single measure spanning several outcomes, while leaving the component-specific pattern unclear unless separate results are reported. The registry names other outcomes and age-related subgroup analyses, but numerical estimates for them are not reported in the registry record. Without those results, the relationship between the composite and its components cannot be evaluated.
Finally, reporting a weighting strategy and a confidence-interval measure is not enough to assess the analysis in full. Details of covariate selection, balance diagnostics, uncertainty estimation, missing-data handling, and multiplicity are not reported. Those methods affect how robustly an observational rate comparison can be interpreted. The registry record therefore makes the analytical aims clear while leaving important elements of the evidence assessment unresolved.
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 registry record does not report numerical results for the primary endpoint, including cohort-specific incidence rates, a comparative estimate, confidence interval limits, or event counts. This prevents assessment of the direction, magnitude, and precision of the primary comparison. Numerical findings for the named secondary outcomes and subgroup analyses are also not reported.
Important design and analysis details are absent. The record does not state treatment allocation, and it identifies the comparison as nonrandomized and the arms as not a like-for-like control comparison. It describes stabilized inverse probability treatment weighting but does not report the covariates in the weighting model, balance diagnostics, weight assessment, or variance-estimation details. The extent to which measured characteristics were balanced cannot therefore be judged.
The population description is limited to eligible participants in the MDV database whose data were observed, with the study focused on very elderly patients. A precise age threshold and detailed eligibility criteria are not reported in the registry record. Cohort-specific participant counts, exclusions, and exposure-time totals are also absent, so the reported enrollment of 77814 does not describe the analysis denominator for the primary rate.
Administrative claims and diagnosis codes are used to identify outcomes, but validation details and assessment of misclassification are not reported. Missing-data methods, multiplicity procedures, and the complete statistical analysis plan are not reported either. These limitations are gaps in the registry information; they do not establish that any particular bias or analysis problem occurred.
14. Related Tutorials
15. Related Statistical Calculators
16. Sources
- ClinicalTrials.gov: Safety and Effectiveness of Apixaban in Very Elderly Patients With Non-valvular Atrial Fibrillation (NVAF) Compared to Warfarin Using Administrative Claims Data — NCT05438888.
- Registry record last updated 2024-05-20; status last verified 2023-10.
- Results first posted 2024-05-20.
- 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
NCT05438888 is a completed, Pfizer-sponsored administrative-claims study of apixaban compared with warfarin in very elderly patients with non-valvular atrial fibrillation. The registry reports enrollment of 77814, a study start date of 2022-07-01, and a primary completion date of 2022-10-19.
The primary endpoint is the incidence rate per 1,000 participant-years of first occurrence of a composite of stroke and systemic embolism in balanced cohorts. Stroke includes ischemic and hemorrhagic stroke, and the analysis is described as using stabilized inverse probability treatment weighting. Follow-up begins the day after treatment initiation and ends at the first of the target outcome, treatment discontinuation, switching treatment, or database withdrawal.
Numerical primary and secondary outcome results are not reported in the registry record. Allocation is not stated, and the record cautions that the cohorts are not a like-for-like control comparison. Further details about eligibility, weighting, missing data, and multiplicity are not reported.