Highlights
-
•
Effects of edoxaban versus VKAs on PROs were assessed using a WR approach
-
•
PACT-Q2 captured patient-perceived convenience/satisfaction with anticoagulants
-
•
WR adds interpretation of meaningfulness and drivers of satisfaction/convenience
-
•
Convenience/satisfaction was meaningfully better with edoxaban versus VKA
-
•
Key factors driving the favorable outcomes with edoxaban versus VKAs were identified
The previous report on patient-reported outcome findings of the ENVISAGE-TAVI atrial fibrillation (AF) (NCT02943785) trial demonstrated improved patient experience for edoxaban versus vitamin K antagonists (VKAs). This post hoc analysis aimed to provide insights on the patient-reported outcome findings of ENVISAGE-TAVI AF using a win ratio (WR) approach to understand key drivers of treatment differences. This analysis included patients who received edoxaban or VKAs and had evaluable Perception of Anticoagulant Treatment Questionnaire 2 (PACT-Q2) assessments from ENVISAGE-TAVI AF. The PACT-Q2 assesses treatment convenience (13 items) and satisfaction (7 items). PACT-Q2 data at months 3 and 12 were analyzed using the WR. Patient-to-patient pairs (one from each group) were compared based on predefined outcome rules; a “win,” “loss,” or “tie” was determined for edoxaban in each pairwise comparison. The WR (95% confidence interval [CI]) for edoxaban was calculated as the total number of pairs with a win divided by that of pairs with a loss. WR >1 indicates a more favorable patient experience for edoxaban versus VKAs. Edoxaban was associated with a higher probability of improved overall treatment convenience and satisfaction compared with VKAs at months 3 (WR [95% CI], 1.87 [1.58 to 2.22]) and 12 (WR [95% CI], 2.01 [1.70 to 2.38]). This difference was driven by 18 of 20 PACT-Q2 items, showing a significantly higher probability of meaningfully better outcomes with edoxaban. In conclusion, this WR analysis demonstrated that meaningfully better treatment convenience and satisfaction were more likely with edoxaban than with VKAs in patients with AF after transcatheter aortic valve replacement.
Although the efficacy and safety of oral anticoagulants have been studied in patients with atrial fibrillation (AF) after transcatheter aortic valve replacement (TAVR), ,, data on patient-reported outcomes (PROs) in this population are limited. In the ENVISAGE-TAVI AF (NCT02943785) trial, patients treated with edoxaban after successful TAVR reported significantly improved treatment satisfaction and convenience compared with those receiving vitamin K antagonists (VKAs) as measured by the Perception of Anticoagulant Treatment Questionnaire 2 (PACT-Q2) “Treatment Satisfaction” and “Convenience” dimensions scores. However, a statistically significant difference between treatments in mean PRO scores is not necessarily clinically meaningful to patients.
Interpreting differences in PRO scores across treatments is challenging because validated clinically meaningful thresholds are often unavailable, and aggregated scores obscure individual component contributions. These limitations complicate identification of key drivers of treatment differences, underscoring the need for advanced analytical approaches.
The win ratio (WR) is a statistical method that compares 2 treatment groups by considering patient pairs, with each pair consisting of 1 patient from each group. It offers enhanced clinical interpretability by assessing treatment differences at the patient-pair level and quantifying the probability that one treatment provides a meaningfully better experience than another. Unlike traditional mean score comparisons, which may demonstrate statistical differences without clear relevance to patient-perceived benefit, the WR incorporates patient-centered definitions of meaningful difference, thereby bridging the gap between statistical and clinical significance.
We applied the WR analytical approach to reevaluate the effect of edoxaban versus VKAs on PROs by prioritized items and to further identify the drivers of treatment satisfaction and adherence differences using data from ENVISAGE-TAVI AF.
Methods
Study design
ENVISAGE-TAVI AF was a multicenter, prospective, randomized, open-label, adjudicator-masked trial that compared edoxaban with VKAs in patients with prevalent or incident AF after successful TAVR. , The study design and primary results were previously published. , The trial was conducted in accordance with the Declaration of Helsinki and the International Council for Harmonisation; local laws and regulations were followed. Written informed consent was obtained before enrollment.
Study population
This study included all randomized patients (intention-to-treat [ITT] population) with available PACT-Q2 data at month 3 or 12. The analytic populations for months 3 and 12 were derived following a prespecified hierarchical imputation approach. When the end of treatment (EOT) assessment was collected within the relevant visit window, it was substituted; otherwise, the last observation, if available, was carried forward (see “Statistical” section for details).
Patient-reported outcomes
The PACT-Q assesses patients’ expectations of and satisfaction with anticoagulant treatment and treatment convenience using 2 modules. The PACT-Q1 module assesses patients’ baseline expectations before anticoagulation treatment initiation. The PACT-Q2 module assesses treatment “Convenience” (11 items), “Burden of Disease and Treatment” (2 items), and “Anticoagulation Treatment Satisfaction” (7 items; Supplementary Table S1 ) while receiving anticoagulants. Both PACT-Q modules use a 5-point Likert scale (score range 1 to 5 for each item) for response options. The “Convenience” and “Burden of Disease and Treatment” domains are combined into the “Convenience” dimension, and the “Anticoagulation Treatment Satisfaction” domain constitutes its own dimension. For each PACT-Q2 dimension, the sum score of associated items is rescaled to a range of 0 to 100, with a higher score indicating a better outcome.
The PACT-Q1 module was administered before treatment initiation, and the PACT-Q2 module was administered postbaseline at month 3, month 12, and EOT. The aggregated scores were previously reported. This study focused on treatment convenience and satisfaction as assessed by the PACT-Q2 module.
Statistical analysis
The analyses included all randomized patients (ITT population) with available PACT-Q2 data at month 3 or 12, following imputation for missing values. For missing month 3 data, the EOT value was used if available and recorded within 3 months of randomization (plus 14 days). If the EOT value was missing but a baseline assessment was available, the baseline value was carried forward. Otherwise, the data remained missing ( Supplementary Figure S1 ). For missing month 12 data, the EOT value was used if recorded after month 3 and within 12 months of randomization (plus 14 days). If the EOT value was unavailable, the month 3 value was carried forward to month 12 if present; otherwise, the data remained missing ( Supplementary Figure S2 ).
In the WR analysis, item outcomes of the PACT-Q2 were first compared separately at months 3 and 12 within every possible patient pair (1 from each treatment group). The comparative outcome (i.e., “win,” “loss,” or “tie”) on a given item for each pair was determined based on predefined rules that were meaningful to patients or other stakeholders. For example, a patient in the edoxaban group reporting “not at all” to Item B1 of the PACT-Q2 (“How difficult is it to take your anticoagulant?”) would have a “win” (score = +1) on this item for this paired comparison if the patient’s counterpart in the VKA group reported a less favorable experience (i.e., “a little” or “worse”). If item responses were the same between groups, then a “tie” (score = 0) on that item was assigned for that pair. If item responses were more favorable with the VKA group, then a “loss” (score =–1) was designated for that pair. Each item’s outcome (win, loss, or tie) was subsequently used to jointly inform the outcome at the domain, dimension, and instrument levels within the pair according to prespecified rules. For example, a patient on edoxaban was assigned a win for a given domain if the patient had a net win compared with their counterpart on VKAs (i.e., the sum of scores across all items in that domain ≥+1, assuming an equal weight for each item), loss if the patient had a net loss (sum ≤–1), or tie (sum = 0). The comparative results on a given dimension and the entire measure within a patient pair were determined similarly based on predefined criteria using the comparative outcomes at the domain and dimension levels, respectively.
“Meaningful” differences were operationally defined at the item level according to the direction of the Likert response scale. For each pairwise comparison, a “win” was assigned when the edoxaban patient selected a more favorable response category than their VKA counterpart, reflecting a patient-perceivable improvement without requiring validated minimally important differences, which are unavailable for the PACT-Q2.
The numbers of pairwise wins ( N W ) and losses ( N L ) for edoxaban were tallied and summarized by item, domain/dimension, and overall measure. The WR for edoxaban was calculated as N W / N L . The P -value and 95% confidence interval (CI) of a WR were estimated using the method developed by Dong et al ( Supplementary Materials ). A WR of >1 indicates edoxaban being favored over VKAs.
Results
Patient characteristics and treatment expectations
Of the 1,426 patients enrolled in the ENVISAGE-TAVI AF trial, 1,161 patients (81.4% of the ITT population) were included in the WR analysis (606/713 [85.0%] ITT patients assigned to edoxaban and 555/713 [77.8%] ITT patients assigned to VKAs). The baseline patient characteristics and treatment expectations were similar between both edoxaban and VKA treatment groups ( Table 1 ). The mean ± standard deviation age of patients was 81.9 ± 5.4 years, and 45.9% of patients were female. Patients had a mean ± standard deviation CHA 2 DS 2 -VASc score of 4.5 ± 1.4 ( Table 1 ).
Table 1
Baseline patient characteristics and treatment expectations
| Characteristic |
Edoxaban
( n = 606) |
VKA
( n = 555) |
Overall
( n = 1,161) |
|---|---|---|---|
| Age (year) , mean ± SD | 82.1 ± 5.3 | 81.8 ± 5.4 | 81.9 ± 5.4 |
| Sex , n (%) | |||
| Male | 318 (52.5%) | 310 (55.9%) | 628 (54.1%) |
| Female | 288 (47.5%) | 245 (44.1%) | 533 (45.9%) |
| Ethnicity , n (%) | |||
| Hispanic or Latino | 158 (26.1%) | 147 (26.5%) | 305 (26.3%) |
| Not Hispanic or Latino | 435 (71.8%) | 401 (72.3%) | 836 (72.0%) |
| Missing | 13 (2.1%) | 7 (1.3%) | 20 (1.7%) |
| Race , n (%) | |||
| Asian | 81 (13.4%) | 76 (13.7%) | 157 (13.5%) |
| Black or African American | 1 (0.2%) | 2 (0.4%) | 3 (0.3%) |
| Other | 12 (2.0%) | 13 (2.3%) | 25 (2.2%) |
| White | 503 (83.0%) | 461 (83.1%) | 964 (83.0%) |
| Missing | 9 (1.5%) | 3 (0.5%) | 12 (1.0%) |
| Body weight (kg) , mean ± SD | 74.7 ± 17.4 | 76.2 ± 17.4 | 75.4 ± 17.4 |
| BMI (kg/m 2) , mean ± SD | 27.5 ± 5.5 | 27.9 ± 5.5 | 27.7 ± 5.5 |
| Dose adjustment at randomization , n (%) | 280 (46.2%) | 249 (44.9%) | 529 (45.6%) |
| Antiplatelet therapy within 30 days before randomization , n (%) | 270 (44.6%) | 250 (45.0%) | 520 (44.8%) |
| Creatinine clearance by Cockcroft–Gault formula (ml/min) , mean ± SD | 58.1 ± 3.8 | 59.6 ± 24.6 | 58.8 ± 24.2 |
| CHA 2 DS 2 -VASc score , mean ± SD | 4.5 ± 1.4 | 4.5 ± 1.3 | 4.5 ± 1.4 |
| Disease history , n (%) | |||
| Hypertension | 548 (90.4%) | 506 (91.2%) | 1,054 (90.8%) |
| Congestive heart failure | 501 (82.7%) | 477 (85.9%) | 978 (84.2%) |
| Diabetes mellitus | 228 (37.6%) | 207 (37.3%) | 435 (37.5%) |
| Stroke | 104 (17.2%) | 100 (18.0%) | 204 (17.6%) |
| Myocardial infarction | 83 (13.7%) | 82 (14.8%) | 165 (14.2%) |
| Labile INR | 39 (6.4%) | 44 (7.9%) | 83 (7.1%) |
| Previous procedures received , n (%) | |||
| PCI | 147 (24.3%) | 148 (26.7%) | 295 (25.4%) |
| CABG | 59 (9.7%) | 54 (9.7%) | 113 (9.7%) |
| PACT-Q1 score , mean ± SD | |||
| A1—Confidence in prevention of blood clots | 3.8 ± 1.0 | 3.8 ± 1.0 | 3.8 ± 1.0 |
| A2—Expectations of symptom relief | 3.0 ± 1.2 | 2.8 ± 1.2 | 2.9 ± 1.2 |
| A3—Expectations of side effects | 2.6 ± 1.1 | 2.7 ± 1.2 | 2.6 ± 1.1 |
| A4—Importance of ease of use | 4.2 ± 0.9 | 4.1 ± 1.0 | 4.1 ± 1.0 |
| A5—Worries about making mistakes | 2.1 ± 1.3 | 2.3 ± 1.3 | 2.2 ± 1.3 |
| A6—Importance of independency | 4.0 ± 1.1 | 3.8 ± 1.2 | 3.9 ± 1.2 |
| A7—Worries about cost | 2.5 ± 1.4 | 2.5 ± 1.4 | 2.5 ± 1.4 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
