Essentials Medication nonadherence is prevalent in chronic disease. The FIRST registry reports adherence to rivaroxaban for the treatment of VTE in the United Kingdom. Overall adherence was good, and nonadherence was mostly unintentional. Older age was the strongest predictor of good adherence. 1 BACKGROUND

Adherence can be defined as the extent to which patients take medications as prescribed by health care providers.1 The importance of adherence to anticoagulation therapy has grown in significance in recent years with the advent of short-acting direct oral anticoagulants (DOACs).2, 3 Previous research has shown that subtherapeutic anticoagulation may place patients at an increased risk of recurrence of venous thromboembolism (VTE) and long-term complications such as postthrombotic syndrome (PTS).4-9 Nonadherence has significant costs for health care providers. In Europe, it is estimated that 125 billion euros are spent each year on avoidable hospitalizations, emergency care, and adult outpatient appointments as a result of poor medication adherence.10, 11

The clinical impact of nonadherence presents a greater risk with DOACs than with vitamin K antagonists (VKAs) due to their shorter half-lives and lack of laboratory monitoring, which results in less contact with health care professionals.12 For VKAs, the international normalized ratio (INR) and the time in therapeutic range (TTR), a tool used to measure an individual’s INR control on a VKA, provide the patient and prescriber with instant feedback on the quality of anticoagulation and can indicate when a patient has been nonadherent.13-16 In the DOAC era, there is no TTR to guide treating clinicians or patients. In a sense, patients are left to their own devices.

Adherence was not problematic in the landmark VTE trials for rivaroxaban, edoxaban, or apixaban.17-19 The level of acceptable adherence >80% (defined by the World Health Organisation),20 was reported as 93.5% in EINSTEIN (pooled results) overall (mean study follow-up 207 ± 95.9 days). This is not surprising in the setting of a phase III clinical trial with frequent intense follow-up and a selected population of highly motivated individuals who have put themselves forward to participate in a clinical trial. It is widely acknowledged that adherence in clinical trials is not replicated in clinical practice, and therefore clinical outcomes comparable with those reported from seminal trials may not be achieved.21

In the absence of a routine direct sampling method such as INR monitoring, the measurement of medication adherence for DOACs is challenging both in clinical practice and clinical research. There are a number of different direct and indirect methodologies used in clinical research to report medication adherence, including medication event monitoring systems (electronic pill caps), proportion of days covered (PDC), and self-report, all of which have their advantages and disadvantages.1, 2 Self-report is the most accessible and inexpensive method, by questionnaire or by assessment of the number of doses missed over a defined period.

Rivaroxaban was the first factor Xa inhibitor to be licensed in the United Kingdom in 2012 and is now well established in many centers as first-line treatment for VTE.22, 23 Rivaroxaban is prescribed in primary and secondary care in the United Kingdom, and the frequency and delivery of follow-up varies among centers. To date, adherence data for rivaroxaban has mostly been reported in the setting of stroke prevention with atrial fibrillation (AF). This research has shown that adherence to DOACs decreases over time.24, 25 A US claims database study reported that the percentage of patients with a PDC ≥80% decreased from 73% at 3 months to 55% at 9 months for patients with AF prescribed rivaroxaban.24 There is a concern that as time passes from an acute VTE event and patients are no longer symptomatic, secondary prevention of VTE becomes analogous to stroke prevention in the setting of AF where adherence is known to be poor.

Findings from the FIRST registry demonstrated that rivaroxaban was effective for the majority of patients in daily care.49 However, 7/1262 (0.6%) reported an episode of VTE recurrence. Five out of those seven cases were as a result of nonadherence to anticoagulation therapy. In this analysis we sought to investigate the extent of nonadherence within the FIRST registry, the characteristics of those who were nonadherent, and possible reasons for medication nonadherence.

2 METHOD

The FIRST registry was a UK-only, multicenter, noninterventional, observational study investigating the long-term complications of VTE for patients treated with rivaroxaban without bridging therapy. The study methodology has previously been reported.49 The study population for this analysis was the safety population, which included any participant who had received ≥1 dose of rivaroxaban.

An adherence screening tool (AST) was used to quantify and explain reasons for nonadherence (Supporting Information S1) and the Anti-Clot Treatment Scale (ACTS)26 was used to assess treatment satisfaction with rivaroxaban as a possible reason for nonadherence.

Both questionnaires were provided to patients simultaneously and were completed independently of the local investigator. The questionnaires were administered at 1 month for all participants regardless of treatment duration; end of treatment, typically at 3 or 6 months for those on short-term treatment, or annually for those on long-term treatment.

The AST consisted of 16 questions (Supporting Information S1). Patients were stratified as nonadherent if they self-reported having missed ≥1 dose of rivaroxaban in the week preceding questionnaire completion. Fourteen questions explore the possible explanations for nonadherence, while the final question explores the worries or concerns of the participant about rivaroxaban treatment.

The ACTS is a 17-item questionnaire validated to measure patient-reported satisfaction with anticoagulation treatment.26 It comprises a 12-item ACTS burdens subscale and a 3-item ACTS benefits subscale reported using a 5-item Likert scale. The tool includes two global questions regarding overall satisfaction. A high benefits score is indicative of high perceived treatment satisfaction. Convention is to reverse the burdens score for analysis; as such, a high burdens score is indicative of low treatment burdens and therefore greater satisfaction. To account for any missing participant responses, a scale-specific mean imputation method was used. In this study, 9 of 12 items from the burdens subscale and 2 of 3 items from the benefits subscale needed to have been completed for inclusion in the analysis.

2.1 Statistical analysis

Patient characteristics are reported as descriptive statistics. For quantitative and ordinal data, mean and standard deviation or median and interquartile range (IQR) were calculated as appropriate. For such data, comparison between subgroups was made using a t test or Mann-Whitney test. Categorical data were compared between subgroups using the chi-square or Fisher’s exact test. Statistical significance was set at P < .05. A logistic regression was completed to estimate the effect of patient characteristics and satisfaction on adherence to rivaroxaban treatment, with the results reported as odds ratios with 95% confidence intervals.

2.2 Ethical approval

Ethical approval was obtained from the West of Scotland Research Ethics Service (14/WS/1120). Each National Health Service (NHS) Trust participating in the study also obtained local research and development approval before opening. All participants provided written informed consent to participate, and confidentiality and anonymity were maintained.

3 RESULTS

In total, 1262 patients were recruited to the FIRST registry, of which 1239 had received ≥1 dose of rivaroxaban and were eligible for this analysis.

Overall, 1030 of 1239 (83.1%) participants completed at least one AST questionnaire while prescribed rivaroxaban. Since the number of patients that completed the adherence questionnaire declined after 2 years of follow-up, this analysis will focus on those who completed questionnaires up to 2 years after the index VTE event (1028/1239; 83.0%). Participants were less likely to have completed the questionnaire if they were younger; were of Black, Asian, or mixed descent; or had an upper limb or distal lower limb deep vein thrombosis (DVT) as their index event (Supporting Information S2).

Since those on longer-term anticoagulation were likely to complete more questionnaires and therefore had more opportunity to report nonadherence, stratification of the study population was based on the response to the first questionnaire completed. As such, nonadherence was reported by 113 of 1028 (11.0%) participants on their first questionnaire, after a median duration of 28 days (IQR, 21-45) on rivaroxaban after diagnosis.

Considering all questionnaires completed, 155 of 1028 (15.1%) reported nonadherence at least once during their treatment with rivaroxaban.

The characteristics of adherent and nonadherent patients are described in Table 1. Adherent patients were more likely to be older and White. In total, 1660 questionnaires were completed by 1028 participants (1.6 questionnaires per participant). The median duration of rivaroxaban exposure for participants was 168 days (IQR, 89-377). Overall, adherence to rivaroxaban was high, although there was a reduction in the proportion of adherent participants observed 2 years after the initiation of rivaroxaban (Table 2.).

TABLE 1. Comparison of participant characteristics between the adherent or nonadherent subgroups

Adherent

n = 915

Nonadherent

n = 113

P N Sex, n (%) Female 352 (38.5) 36 (31.9) .35 Male 561 (61.3) 77 (68.1) … Transgender 2 (0.2) 0 (0.0) … Age, mean (SD) – 60.1 (14.9) 53.7 (15.8) <.001 Race, n (%) White 817 (91.3) 88 (80.7) .001 Black 60 (6.7) 13 (11.9) … Asian 16 (1.8) 5 (4.6) … Mixed 2 (0.2) 3 (2.8) … Unknown 20 4 … Diagnosis, n (%) Distal DVT 261 (28.5) 30 (26.5) .35 Proximal DVT 407 (44.5) 60 (53.1) … PE 211 (23.1) 20 (17.7) … DVT and PE 22 (2.4) 3 (2.7) … Upper limb 14 (1.5) 0 (0.0) … Personal history, n (%) No personal history of VTE 711 (77.9) 80 (70.8) .14 1 previous VTE 171 (18.7) 30 (26.5) … >1 previous VTE 31 (3.4) 3 (2.7) … Unknown 2 2 … Cancer-associated VTE, n (%)a No cancer 888 (97.3) 110 (97.3) .63 Cancer 25 (2.7) 3 (2.7) Days of rivaroxaban exposure, median (IQR) – 168 (90-379) 116 (84-352) .15 Note Participants were stratified as adherent or nonadherent based on the response from the first questionnaire completed. This approach was adopted since as time elapsed from the index event, those on longer-term anticoagulation were likely to complete more questionnaires and therefore had more opportunity to report nonadherence. Abbreviations: DVT, deep vein thrombosis; IQR, interquartile range; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism. aCancer status was not reported for two participants. TABLE 2. Number of adherence screening tool questionnaires completed, stratified by adherent or nonadherent at each time point Time Questionnaire Completed 1 month 3 months 6 months 1 year 2 years N participants 878 291 173 205 113 Adherenta  , n (%) 779 (88.7) 262 (90.0) 156 (90.2) 183 (89.3) 97 (85.8) Nonadherenta, n (%) 99 (11.3) 29 (10.0) 17 (9.8) 22 (10.7) 16 (14.2) aThe proportion of adherent and nonadherent participants at each time point are presented. Inferences about the trend of nonadherence for individuals cannot be drawn since questionnaires were not all completed at the same time points by the same participants.

Forgetting to take the rivaroxaban and carelessness were consistently reported in a higher proportion in the nonadherent subgroup (P < .001; Table 3). Nonadherent participants reported that they were less able to manage a change in their routine with respect to rivaroxaban. This difference was observed at 1 month (P < .001) and at 1 year (P < .05) but was not observed at 2 years. At 1 and 2 years after the initiation of rivaroxaban, a higher proportion of nonadherent patients reported forgetting to refill their prescription for rivaroxaban (P < .05). At 1 year, a higher proportion of nonadherent patients reported not being aware of the long-term benefits of rivaroxaban (P < .05), and that they did not have a routine for taking their rivaroxaban (P < .001).

TABLE 3. The results of the adherence screening tool stratified by adherence at 1 month, 1 year, and 2 years 1 month 1 year Two years Adherent Nonadherent P Adherent Nonadherent P Adherent Nonadherent P 779 99 183 22 97 16 Do you ever forget to take your rivaroxaban? n (%) No 708 (91.4) 25 (25.3) <.001 156 (87.2) 5 (22.7) <.001 82 (85.4) 3 (18.8) <.001 Yes 67 (8.6) 74 (74.7) 23 (12.8) 17 (77.3) 14 (14.6) 13 (81.2) Do you find it difficult to take your rivaroxaban (eg, swallowing your tablet)? n (%) No 768 (99.2) 98 (99.0) … 179 (98.9) 22 (100.0) – 95 (100.0) 16 (100.0) … Yes 6 (0.8) 1 (1.0) 2 (1.1) 0 (0.0) 0 (0.0) 0 (0.0) Are you confident that you are taking your rivaroxaban in the correct way? (%) No 106 (13.7) 10 (10.2) .42 31 (17.0) 4 (18.2) … 16 (17.0) 2 (12.5) .93 Yes 668 (86.3) 88 (89.8) 151 (83.0) 18 (81.8) 78 (83.0) 14 (87.5) When you feel better, do you sometimes stop taking your rivaroxaban? n (%) No 765 (99.1) 95 (96.9) .17 179 (98.4) 21 (95.5) .91 95 (99.0) 15 (93.8) .66 Yes 7 (0.9) 3 (3.1) 3 (1.6) 1 (4.5) 1 (1.0) 1 (6.2) Do you take your rivaroxaban only when you feel you need to? n (%) No 761 (98.3) 97 (99.0) .95 175 (97.2) 21 (95.5) .50 95 (99.0) 15 (93.8) .66 Yes 13 (1.7) 1 (1.0) 5 (2.8) 1 (4.5) 1 (1.0) 1 (6.2) Do you know the long-term benefits of taking your rivaroxaban as told to you by your doctor, nurse, or pharmacist? n (%) No 155 (20.0) 22 (22.2) .70 25 (13.9) 8 (36.4) .007 16 (16.8) 2 (13.3) … Yes 620 (80.0) 77 (77.8) 155 (86.1) 14 (63.6) 79 (83.2) 13 (86.7) Do you think the rivaroxaban you have been prescribed has been effective in treating your condition? n (%) No 54 (7.6) 7 (7.7) … 9 (5.2) 0 (0.0) .58 3 (3.4) 2 (13.3) .32 Yes 657 (92.4) 84 (92.3) 165 (94.8) 22 (100.0) 85 (96.6) 13 (86.7) Sometimes if you feel worse when you take your rivaroxaban, do you stop taking it? n (%) No 764 (98.8) 94 (98.9) … 176 (99.4) 22 (100.0) … 93 (100.0) 15 (93.8) .32 Yes 9 (1.2) 1 (1.1) 1 (0.6) 0 (0.0) 0 (0.0) 1 (6.2) Sometimes do you stop taking your rivaroxaban so your body can take a break from its effects? n (%) No 767 (99.2) 95 (96.9) .12 180 (99.4) 21 (95.5) .52 95 (100.0) 15 (93.8) .31 Yes 6 (0.8) 3 (3.1) 1 (0.6) 1 (4.5) 0 (0.0) 1 (6.2) Sometimes do you forget to refill your prescription for your rivaroxaban on time? n (%) No 735 (97.1) 87 (95.6) .65 167 (92.3) 16 (72.7) .004 92 (95.8) 12 (75.0) .01 Yes 22 (2.9) 4 (4.4) 14 (7.7) 6 (27.3) 4 (4.2) 4 (25.0) Do you have a routine to help you take your rivaroxaban regularly? n (%) No 121 (15.6) 23 (23.2) .08 25 (13.7) 11 (52.4) <.001 15 (15.6) 4 (25.0) .57 Yes 653 (84.4) 76 (76.8) 157 (86.3) 10 (47.6) 81 (84.4) 12 (75.0) When there is a change in your routine, are you confident you can continue to take your rivaroxaban on time? n (%) No 57 (7.4) 25 (25.5) <.001 19 (10.4) 7 (31.8) .005 12 (12.5) 4 (25.0) .35 Yes 713 (92.6) 73 (74.5) 163 (89.6) 15 (68.2) 84 (87.5) 12 (75.0) Are you careless at times about taking your rivaroxaban? n (%) No 745 (97.3) 72 (72.7) <.001 179 (98.4) 15 (71.4) <.001 90 (94.7) 8 (53.3) <.001 Yes 21 (2.7) 27 (27.3) 3 (1.6) 6 (28.6) 5 (5.3) 7 (46.7) Do you believe that you need to take your rivaroxaban regularly? n (%) No 29 (3.8) 5 (5.1) .72 7 (3.9) 1 (4.5) … 5 (5.3) 0 (0.0) .77 Yes 744 (96.2) 94 (94.9) 172 (96.1) 21 (95.5) 89 (94.7) 16 (100.0)

The concerns that participants reported regarding rivaroxaban are outlined in Figure 1. Possible side effects and long-term effects of rivaroxaban were the two most frequently reported concerns reported by participants.

Patient-reported concerns about their treatment with rivaroxaban

3.1 ACTS Analysis

Treatment satisfaction with rivaroxaban was high. Patient-reported benefits were consistently high, and patient-reported burdens decreased over the study period (Figure 2).

The results of the ACTS Burdens and Benefits Scale reported at each time point up to 2 years for adherent and nonadherent participants. ACTS burdens (/60), a higher score represents lower overall patient reported burdens; ACTS benefits (/15), a higher score represents higher overall patient reported benefits

Finally, we sought to assess the relationship between adherence and satisfaction. Table 4 summarizes burdens and benefits scores stratified by adherence at each time point.

TABLE 4. Results of the ACTS burdens and benefits scores stratified by patients who were found to be adherent or nonadherent at each time point ACTS subscale N

ACTS score

Median (IQR)

Adherence N (%)

ACTS score

Median (IQR)

1 mo 878 Burdens 826