Background

Oral anticoagulants (OACs) are routinely used in the UK to prevent stroke in atrial fibrillation (AF) and to prevent or treat venous thromboembolism (VTE). The 2014 NICE guidance [1, 2] recommended the use of both warfarin (a vitamin K antagonist) and direct (or direct-acting or non-vitamin K antagonist, previously referred to as novel) OACs; dabigatran, rivaroxaban, apixaban, and edoxaban. This guidance was updated between 2018 and 2021 [3,4,5]. With exceptions for specific conditions, the new guidance recommends any of the four direct OACs for stroke prevention in AF; for VTE treatment, apixaban and rivaroxaban are recommended; for secondary prevention, continued treatment or apixaban are recommended; and for primary prevention, various options are recommended depending on the reason for hospitalisation.

Clinicians and patients still have choices about which recommended OAC to prescribe or take. Although warfarin was standardly used before the approval of direct OACs, it requires regular patient monitoring to maintain coagulation in the target range [1, 6]. Patients are also required to avoid foods containing vitamin K, and to have a low alcohol intake. Common side effects are bleeding (leading to bruising, nosebleeds, and headaches), mild rash and hair loss [7], and warfarin is known to interact with other medications [7]. These factors can lead to poor adherence. Direct OACs only require annual rather than daily to quarterly blood tests [8], and there are fewer indications of interactions with other medications [9, 10]. They are more expensive than warfarin, but if they are more effective due to better adherence, they could be more cost-effective.

This systematic review of systematic reviews (overview from this point on) was commissioned, in 2018, by the Department of Health and Social Care (DHSC) in England with the aim of identifying, assessing, and summarising the relevant research evidence that had been published since the 2014 NICE guidance. Since our full report was completed in 2018, new evidence has become available, filling some of the gaps that we had identified. Although the NICE guidance has been updated, this overview incorporates new research and is a useful guide for practitioners. It compares direct OACs with warfarin, in adults, to prevent stroke in AF, or to prevent or treat VTE. We focussed on efficacy and safety, self-monitoring, genotype-guided dosing, and patient and clinician experiences of OACs. Given the large volume of research, an overview of existing systematic reviews was considered to be the most appropriate method [11, 12]. Here we provide a summary of the 2018 report [13], and of the systematic reviews published since that report.

Research questions addressed

RQ1. What evidence syntheses have been conducted to address the efficacy of UK-approved OAC therapy with respect to:

  1. (a)

    Warfarin versus direct OACs in different patient cohorts?

  2. (b)

    The evidence for an optimised pathway on genotyping?

  3. (c)

    The evidence for an optimised pathway on self-monitoring?

RQ2. What evidence syntheses have been conducted to address the safety of UK-approved OAC therapy with respect to:

  1. (a)

    Renal function and the long-term use of direct OACs?

  2. (b)

    Complications associated with warfarin and direct OACs, including bleeding and stroke risk?

RQ3. What are patient and clinician experiences of UK-approved OAC therapy concerning:

  1. (a)

    The impact of direct OACs and warfarin on patient lifestyle?

  2. (b)

    Medicines adherence and compliance of direct OACs and warfarin?

  3. (c)

    Clinician perceptions of direct OACs and warfarin?

  4. (d)

    Monitoring international normalised ratio (INR)s in patients receiving vitamin K antagonists and the effect on patient adherence?

Methods

The protocol for this systematic overview was registered on PROSPERO (CRD42017084263). Academic, clinician, NHS, and DHSC stakeholders were consulted throughout the review. The overview is reported in accordance with the PRISMA statement, and the checklist is in Additional file 1. We initially searched MEDLINE, Embase, ASSIA, and CINAHL, in October, 2017, and we updated these searches on 30 November, 2021. Search terms covered oral anticoagulants, systematic review and their synonyms, and the medical condition (where appropriate). The searches were limited to articles published since 2014, when the previous NICE guidance on OACs was issued [1]. The original and updated MEDLINE strategies are in Additional files 2 and 3.

To be included, articles had to meet the following criteria:

  • Published since 2014

  • Available in English

  • Focussed on OECD settings

  • Focussed on adults eligible for oral anticoagulation

  • Focussed on OACs for the prevention of stroke related to AF, or the acute treatment, or primary or secondary (after a VTE) prevention, of VTE

  • Focussed on warfarin, dabigatran, rivaroxaban, edoxaban, or apixaban

  • Be a systematic review, with a search of more than one database

  • Assess therapeutic doses of warfarin, compared with a direct OAC, or a comparison between direct OACs

  • Report health or cost outcomes or stakeholder experiences

At least two reviewers independently screened titles and abstracts, until agreement was over 90%, then single reviewers completed screening of abstracts and potentially relevant full texts. For the update search, we used priority screening [14] (a new feature of EPPI-Reviewer web) [15] to identify and prioritise those articles that were most likely, based on the screening results for the initial search, to meet the inclusion criteria. We stopped screening when we were including, for assessment on full text, fewer than one in fifty articles. Data were extracted, quality assessed and synthesised by single reviewers, and checked by and agreed with another reviewer. EPPI-Reviewer© software [15, 16] was used to manage the process.

We extracted pre-defined descriptive characteristics from the reviews, including.

  • Year of publication;

  • Date range of included primary studies;

  • Setting (community, hospital, etc.);

  • Main topic focus (efficacy, safety, experiences, or cost);

  • Target population (health condition, or at-risk group);

  • Participant characteristics (age, gender, etc.);

  • Intervention characteristics (type of oral anticoagulant, or self-monitoring);

  • Number, countries, and designs of primary studies included in the review;

  • Type of outcomes measured (health outcomes, hospitalisation, health-related quality of life, stakeholder experiences, etc.); and

  • Author and year of primary studies (to assess overlap between reviews).

We coded the review characteristics and assigned reviews to each research question or subsection. We narratively synthesised or summarised the data. For self-monitoring or adherence, the primary outcome was time in therapeutic range (TTR) or proportion of days covered (PDC). For genotyping, where the outcomes were not reported for patients with AF or VTE separately from patients with other conditions, we described the reviews, rather than extracting efficacy or safety data. Summary tables were produced (see Table 2, Table 3, Table 4, and Additional file 4).

The included systematic reviews were assessed for risk of bias using AMSTAR2 [17] or the JBI Critical Appraisal Checklist for Systematic Reviews and Research Syntheses [18], as appropriate. Where we included multiple reviews addressing the same question, overlap in primary studies was assessed [19]. Overlap means that the same primary studies are included in more than one review, giving their results more weight than others, which can bias the conclusions [11].

Results

Figure 1 shows the flow diagram for the articles included in this overview. The initial search identified 1840 unique articles, and the update identified an additional 3644 unique articles. After the initial search, we consulted with NHS commissioners and together decided to focus on one recent, rigorous, and comprehensive systematic review to address the two questions of effectiveness and safety (RQ1 and RQ2) [20]. This systematic review included network meta-analyses on the prevention of stroke in AF, and the primary and secondary prevention and treatment of VTE. The remaining 424 articles that were identified as possibly relevant, based on their titles and abstracts, are listed with the full report [13]. The update search and screening identified an additional 468 articles as possibly relevant, and these are listed in Additional file 5, with brief details of five overviews that were identified; four [21,22,23,24] on AF and one [25] on pulmonary embolism. Many of the additional reviews focussed on patients with specific conditions, such as renal disease, cancer, or diabetes, or undergoing various procedures, or elderly or obese patients. From the initial search, 50 articles were screened on full text and 23 of these were included. From the update search, we screened 1953 articles on title and abstract and 42 on full text, and included 26 additional reviews. In addition to the review on efficacy and safety, we included a total of 17 reviews for genotyping, 23 reviews for self-monitoring or adherence, and 15 reviews for stakeholder experiences (six reviews were relevant to both self-monitoring and stakeholder experiences, and one to both self-monitoring and genotyping). Based on full text, the most common reason for exclusion was not reporting systematic review methods; the reasons for exclusion are reported in Additional file 6. Table 1 summarises the findings for all the research questions.

Fig. 1
figure 1

Flow diagram

Full size image
Table 1 Summary of results for all research questions
Full size table

Efficacy and safety

The systematic review [20] on efficacy, safety, and cost-effectiveness was assessed as having low risk of bias in 11 of the 16 AMSTAR2 domains; full results are in Additional file 7. The summary characteristics are described in Table 2 below. As the four topics were distinct, there was no overlap in primary studies between the sections.

Table 2 Efficacy and safety review characteristics and main findings
Full size table

Randomised controlled trials (RCTs) were included in the systematic review [20] if they assessed adults who were eligible for OACs. Apixaban, edoxaban, rivaroxaban, dabigatran, and betrixaban (not licenced in the UK) were compared with warfarin, low-molecular-weight heparin, or antiplatelets (aspirin or clopidogrel). There were no direct comparisons (i.e., compared within the same trial) between direct OACs; they were evaluated using a common comparator across trials, usually warfarin, low-molecular-weight heparin, or placebo, in Bayesian fixed-effect network meta-analyses. The outcomes were stroke, symptomatic VTE, bleeding events, and death. Two of the 23 trials on patients with AF were conducted in the UK. Most of the included trials were rated by the review authors as at low risk of bias, except for the prevention of stroke in patients with AF, where the trials were rated as at a mix of low, high, and unclear risk. Most trials were relatively short (3 to 42 months, for stroke prevention in AF; 4 to 183 days, for VTE primary prevention; 12 to 48 weeks, for VTE treatment; and 3 to 52 months, for VTE secondary prevention) and funded by drug manufacturers.

For the prevention of stroke in patients with AF, direct OACs were generally more effective and safer than warfarin. Comparing direct OACs, apixaban (5 mg twice daily) had the best efficacy and safety profile and was most likely to be cost-effective. For patients at risk of VTE (undergoing hip or knee surgery or with a medical condition), there was no strong evidence that efficacy or safety differed between direct OACs and low-molecular-weight heparin. Rivaroxaban was most likely to be cost-effective, but with high uncertainty. For the treatment of VTE, there was little evidence that efficacy differed between direct OACs and warfarin, but apixaban (5 mg twice daily) and rivaroxaban (15 mg twice daily, then 20 mg once daily) could reduce the risk of (major and clinically relevant) bleeding, and apixaban (5 mg twice daily) was most likely to be cost-effective. For secondary prevention of VTE, there was little evidence that efficacy differed between direct OACs and warfarin, but apixaban (2.5 or 5 mg twice daily) and dabigatran (150 mg twice daily) could reduce the risk of bleeding. Aspirin was most likely to be cost-effective, but this was uncertain.

Genotyping

From the initial search, none of the 10 reviews [26,27,28,29,30,31,32,33,34,35] on genotyping focussed exclusively on patients with AF or VTE. These reviews assessed patients receiving OACs for any condition, including cardiomyopathy, heart-valve replacement, and rheumatic heart disease, as well as AF or VTE (see Additional file 4 for details). The update searches identified seven reviews [36,37,38,39,40,41,42] that met the inclusion criteria (see Additional file 4 for details). One [38] of these reported data for patients with AF. This review met 12 and partly met one of the 16 relevant AMSTAR criteria and found no significant difference in efficacy of stroke prevention, between each direct OAC and genotype-guided warfarin dosing (the gene was not specified, but references indicated it was CYP2C9). The authors assessed the primary evidence as having moderate-to-high risk of bias.

All reviews assessed genotype-guided dosing for warfarin. We were unable to obtain the full text for one review that reported information on genotyping (ABCB1) for direct OACs, but without specifying the condition[43]. Overall, the reviews were rated as at moderate risk of bias. The AMSTAR2 risk of bias assessment for all reviews is in Additional file 7. Most of the 155 studies within the reviews (see Additional file 8) included patients with AF or VTE, but all reviews, except one [38], presented findings for patients with all conditions without subgroup analyses by AF or VTE. Therefore, we were unable to extract any results specifically for patients with AF or VTE from the remaining reviews. Overlap across the 17 reviews was slight (CCA: 4.5%); however, six studies were in 10 or 11 reviews.

Self-monitoring

From the initial search, we identified six reviews [44,45,46,47,48,49] that addressed self-monitoring, and from the update search, a further 15 reviews [38, 50,51,52,53,54,55,56,57,58,59,60,61,62,63] that addressed self-monitoring or adherence. The summary characteristics are presented in Table 3, below. Quality assessment results are in Additional file 7.

Table 3 Self-monitoring review characteristics and main findings
Full size table

Interventions to improve adherence included the following: education and/or decision aids (four reviews) [44, 55, 62, 64]; self-testing with clinician dosing or self-management with dosing according to rules (eight reviews) [38, 44, 46, 48, 53, 54, 61, 63]; and pharmacist management (four reviews) [45, 47, 49, 66]. The most common outcomes reported were TTR, proportion of INR measurements in range, or PDC (a measure used for direct OACs as they do not require as frequent blood monitoring as warfarin). Adherence [50, 57, 59, 60], persistence [52, 56], discontinuation [51], and switching [58] were also reported. Most reviews assessed patients with any condition (including AF or VTE); nine reviews focussed on patients with AF; one focussed on AF or VTE; none focussed on patients with VTE.

Overall, the reviews were rated as at low-to-moderate risk of bias (see Additional file 7), with few major flaws. One review [44] met 15 of 16 AMSTAR2 domains, one [48] fully or partly met 14, and one [54] fully or partly met 13. Five reviews met just under half of their relevant criteria [51, 57, 58, 61, 62], and five reviews met just over half of their relevant criteria [47, 50, 60, 63, 64]. The remaining reviews met most of their relevant criteria. Eight reviews mainly included RCTs; 15 reviews mainly included other study designs. The authors of six reviews rated primary-study quality as high or good [45, 49, 56, 57, 60, 61]. The authors of one review [47] reported scores ranging from 16 to 28 (mean 19.5; where 28 is the highest score) using the Downs and Black Checklist [67]. The remaining reviews reported a range of risks of bias (from low to high) or flaws in study design across included studies.

For patients with AF, low-quality evidence from a few studies in one review [44] suggested that education with or without a decision aid could improve control of coagulation, compared with usual care without education, at least in the short term (less than 1 year). One review [62] described the evidence on decision aids as unclear; in two studies, they improved adherence and in one they did not. Another review [64] found improved adherence, but the evidence was at high risk of bias. Compared with usual care, self-management appeared to improve coagulation, in most of the low-to-moderate-quality trials in three reviews [45, 47, 49]. The findings of a meta-analysis of two RCTs suggested that self-management plus education improved TTR, although the difference was not significant, compared with usual care without education (mean difference 6.31%, 95% confidence interval − 5.63 to 18.25) [44]. Together, four reviews [56,57,58,59] focussing on AF patients found higher PDC or less switching with apixaban than rivaroxaban, followed by dabigatran, then warfarin; one review [60] found that dabigatran had highest adherence, followed by apixaban, then rivaroxaban, and lastly vitamin K antagonists. One review [51] reported lower discontinuation at 2 years with direct OACs compared with VKA, but not at 1 year, and another review [52] reported the lowest likelihood of non-persistence with apixaban.

For patients requiring an OAC for a range of conditions (including AF and VTE), education improved adherence and TTR[55], as did decision aids[64]. Self-testing improved INR values in range, compared with usual care, but the low-to-moderate-quality evidence was contradictory for TTR, with both longer and shorter TTR reported across studies, within reviews[46, 48]. Three reviews reported improvements in TTR with self-testing over usual care [38, 53, 54]. Self-management was associated with both more and fewer values in range, and longer and shorter TTR [44, 46, 48], and TTR with remote management was not significantly different from with in-person management [61, 63]. A meta-analysis of four high-quality RCTs found a higher percentage of TTR with pharmacist management than with usual (primary) care (mean difference 3.66, 95% confidence interval 2.20 to 5.11; although this was not significant for time in extended therapeutic range) [49]. The findings of another meta-analysis were consistent [66]. Similarly, low- or uncertain-quality evidence suggested that pharmacist management could improve TTR [45, 47].

The extent of overlap (corrected covered area[19], CCA; see Additional file 8 for details) between studies included in the reviews was slight across four reviews for education or decision aids (4.9%), high across eight reviews for self-testing with or without self-management (13.7%); very high across four reviews for pharmacist management (21.2%); and moderate across eight reviews for adherence measures (7.0%).

Stakeholder experiences

From the original search, we included nine reviews [44, 45, 49, 68,69,70,71,72,73] that focussed on the experiences of patients, and three of these also examined the experiences of physicians [68, 70, 72]. From the update, we included four reviews [50, 55, 74, 75] that focussed on patients, one [76] on clinicians, and one [51] was an analysis of reasons reported in medical records. The summary characteristics are presented in Table 4. Quality assessment results are in Additional file 9.

Table 4 Stakeholder experiences review characteristics and themes
Full size table

From the original search, we included six reviews that focussed on patients with AF [44, 68,69,70,71,72] and three that included patients with a range of conditions including AF and VTE [45, 49, 73]. One review [73] focussed on direct OACs in patients with AF, VTE, or other conditions, specifically in patients with renal disease. From the update, we included three reviews that focussed on AF [51, 75, 76], two on AF and VTE [50, 74], and one on any condition [55]; four focussed on direct OACs (three compared with warfarin) [50, 51, 74, 76], and two included any OAC [55, 75].

Seven reviews investigated clinicians’ [76] or patients’ (and clinicians’ in two reviews) [68, 70] perceptions and attitudes to warfarin [70], warfarin, and direct OACs [68, 71, 74] or direct OACs [50, 73, 76]. Two reviews [69, 72] investigated patients’ (and physicians’ in one review) [72] experiences of vitamin K antagonists and direct OACs in discrete-choice experiments. One review [51] examined the reasons for discontinuation, given in patient records. The other five reviews examined patients’ knowledge gaps [75], or views of education or behavioural interventions [44, 55], or pharmacist management [45, 49]. Across the reviews, the number of included primary studies ranged between eight and 140, with between 341 and 11,607Footnote 1 patients, where reported.

Seven reviews [44, 49, 51, 69, 70, 74, 76] met 10 of the 11 JBI criteria [18], and five [45, 50, 55, 59, 72] met nine, while one [71] met seven, one [68] met six, and one [73] met five criteria (see Additional file 9). Overall, we assessed the reviews to be at low risk of bias.

Our synthesis of the original reviews showed that patients and physicians were most concerned with drug efficacy, followed by safety, except in one review [70] where geriatricians reported that safety was most important. Convenience or daily management factors were found to be important for adherence, although there was no consistent pattern across the reviews. A wide range of factors were reported as influencing patients’ decisions about starting, switching, or continuing OACs with no consistency about which were the most important for which groups of patients, at which point in their treatment. Figure 2 indicates the themes that were identified across these reviews. The evidence from the reviews identified in the update was consistent with these themes.

Fig. 2
figure 2

Themes influencing OAC decisions

Full size image

For patients, knowledge and the need for information influenced their decisions about which OAC to start with and whether or not to switch treatment. One review [44] suggested that improving knowledge could improve quality of life, anxiety, and depression. Past experience of stroke, bleeding, and/or OACs, and the experiences and support of families, also influenced patients’ decisions. Patients expressed a need for support and information in managing their therapy, and two reviews [45, 49] suggested that patients were more satisfied with pharmacist management than usual care. Three reviews found higher satisfaction with direct OACs than with warfarin, or with versus without education [50, 55, 74]. One review found gaps in knowledge about OACs and health conditions [75].

For clinicians, their knowledge of the patient, past experience, and expertise plus scientific evidence influenced their decisions. Poor communication between professionals (specialists and primary care physicians) who were involved in the patient’s care could complicate decision-making where their approaches differed [70]. One review [76], identified in the update, found that clinicians thought that direct OACs were safer, particularly for patients who missed appointments, but they had concerns about bleeding and reversal.

The review of medical notes [51] listed reasons for discontinuation that included bleeding, gastrointestinal events, frailty, and fall risk.

Where reviews included both clinicians and patients [68, 70, 72], two factors were found to influence decisions on oral anticoagulants: expectations of their role in decision-making (who was responsible for making the decision) and the quality of communication between clinician and patient. The use of decision aids did not improve patient satisfaction nor reduce decision conflict [44]. Both clinicians and patients wanted improved communication.

The 15 reviews included a total of 237 primary studies. Of these, 38 (16%) studies were included in two reviews, three were included in three reviews, and one was included in four reviews, suggesting slight overlap overall (CCA 1.4%).

Discussion

Summary of the evidence

One high-quality systematic review found that for the prevention of stroke in patients with AF (in accordance with NICE guidance), most direct OACs were more effective and safer than warfarin; apixaban 5 mg twice daily was most likely to be best and cost-effective (based on high- and low-quality primary studies). For the primary prevention of VTE (based on high-quality primary studies), there was no strong evidence that direct OACs should replace low-molecular-weight heparin (which is recommended by NICE for hip or knee replacement) [3]. For both the treatment and secondary prevention of VTE, there was little evidence that direct OACs were better than warfarin, but some of them may reduce the risks of bleeding (the 2020 NICE guidance recommends apixaban or rivaroxaban) [4]. For VTE treatment, apixaban (5 mg twice daily) was most likely to be cost-effective, and for secondary prevention, aspirin was most likely to be cost-effective (based on high-quality primary studies).

None of the reviews of genotyping, identified by the first search, reported results separately for patients with AF or VTE. In the update, one low risk-of-bias review focussed on stroke prevention in patients with AF and found no significant differences with genotype-guided warfarin dosing, compared with usual care. There remains a lack of evidence on genotype-guided dosing for patients with VTE. Most of the 23 reviews of self-monitoring were at moderate or low risk of bias. The evidence suggested that in patients with AF, education with or without a decision aid improved time in therapeutic range (based on high risk-of-bias primary studies), while self-management and self-testing had little effect (based on low-to-high-risk-of-bias primary studies); three of the reviews published in 2020 to 2021 found improvements in TTR based on evidence at high or medium risk of bias. Pharmacist management improved time in range (based on low risk-of-bias primary studies). Both adherence (PDC) and satisfaction were higher with direct OACs. Most of the 15 reviews of stakeholder experiences were at a low risk of bias. Based on primary studies of low-to-high or unclear risk of bias, efficacy was the main driver of the choice of OAC, followed by safety. Other factors were important (see Fig. 2), but these varied by patient and by time-point of decision. Patients were more satisfied with pharmacist management than with usual care, and with direct OACs than with warfarin.

Key messages

What is known about oral anticoagulants.

  • The 2014 NICE guidelines recommended both warfarin and direct oral anticoagulants (OACs) to prevent and treat stroke related to atrial fibrillation (AF), and for venous thromboembolism (VTE).

  • These guidelines were updated between 2018 and 2021, making direct OACs the first choice in some cases. This is in accordance with the evidence presented in our original 2018 overview.

What this systematic overview adds.

  • For AF, direct OACs were more effective and safer than usual care, and apixaban 5 mg twice daily had the best profile. For VTE, overall, direct OACs were no better than low-molecular-weight heparin (prevention), warfarin (treatment), and warfarin or aspirin (secondary prevention).

  • One review found no difference between genotype-guided warfarin dosing and direct OACs for stroke prevention in patients with AF. Reviews of genotype-guided dosing for direct OACs, in patients with AF or VTE, are needed.

  • In patients with AF, education with or without a decision aid improved time in therapeutic range, while self-monitoring or self-testing made little difference. Evidence was lacking for patients with VTE. Pharmacist management could improve time in therapeutic range. Adherence was better with direct OACs than with warfarin.

  • Efficacy was the main driver of the choice of OAC, followed by safety, except for geriatricians, where safety was more important than efficacy. For patients, it seems that knowledge, past experience, disease-related issues, and support needs influence OAC choices and adherence. Patients were more satisfied with pharmacist management, and with direct OACs, than with usual care.

Strengths and limitations

Although this overview, commissioned to inform policy, was carried out within a short timescale, our processes were robust and key decisions were discussed with stakeholders. We updated our search to identify new evidence published since the original review was completed. A new review on genotyping in patients with AF, and a new review on satisfaction with direct oral anticoagulants, filled two evidence gaps. The main conclusions of the original review remain unchanged. This overview meets the reporting requirements of PRISMA (see Additional file 1). The included systematic reviews were rated as at low-to-moderate risk of bias, but the primary studies within the reviews had some limitations, for example, short follow-up; enrolling younger, healthier patients than would be found in usual practice; funding from drug manufacturers; and review authors assessed some of the primary studies as having a high risk of bias. Renally impaired patients were not specifically addressed by the included efficacy review; we identified a few reviews that may address this question (see the full report[13] and Additional file 5).

Recommendations for research

None of the reviews examined the effects of switching from warfarin to direct OACs on the workload of monitoring clinics, which was an aim of the original review. Evidence is needed from trials that include the types of patients found in clinical practice, with long-term follow-up, that directly compare direct OACs, and that are not funded by drug manufacturers. Systematic reviews are needed on genotype-guided dosing of direct OACs for patients with AF or VTE, and on adherence to direct OACs for the treatment and prevention of VTE. The reasons for the limited effectiveness of decision aids could be investigated further. The views of older patients, particularly over 80 years of age, should be investigated. Further exploration, by gender, age, and ethnicity, of self-monitoring, adherence, and patient and clinician views of interventions would add value.

Conclusion

Based on a comprehensive systematic overview of available reviews, the evidence suggests that direct OACs are safer and more effective than warfarin to prevent stroke in patients with AF, especially apixaban 5 mg twice daily. For VTE, there was no strong evidence that any direct OAC should replace those OACs currently recommended by NICE. Effectiveness of the treatment is the most important consideration for patients and clinicians, although older patients might be given the safest option. Genotype-guided dosing may not affect the risk of stroke in patients with AF. Pharmacist management of warfarin may be effective and patients preferred it over management in primary care. Patients were more satisfied with direct OACs than with warfarin. Educating patients about their condition and the use of OACs could improve their adherence and coagulation control.