Direct Oral Anticoagulants in Older and Frail Patients with Atrial Fibrillation: A Decade of Experience

Abstract

Introduction

Both the prevalence of atrial fibrillation (AF) and its subsequent use of direct oral anticoagulants (DOACs) are rapidly increasing in patients of older age. In the absence of contra-indications, guidelines advocate anticoagulation based on the CHA2DS2-VASc score for all AF patients aged 75 and above. However, some practitioners are hesitant to prescribe anticoagulants to older and frail patients due to perceived elevated bleeding risks. This review delves into the comparative treatment outcomes of DOACs versus vitamin K antagonists (VKAs) in older patients with AF, particularly focusing on those of advanced age, frailty, increased risk of falling, chronic kidney disease (CKD), or with a history of major bleeding. Additionally, considerations on the use of off-label DOAC doses, the role of left atrial appendage (LAA) closure and future developments in factor XIa-inhibitors will be discussed.

Results

While strong evidence supports the use of DOACs in the vital older patients with nonvalvular AF, it remains scant in frail patient groups. There is some evidence from non-randomized studies suggesting that the effect of DOACs compared with VKAs is consistent between frail and nonfrail patients. However, recent findings from a single randomized trial showed increased bleeding risks but comparable thromboembolic outcomes in frail individuals switching from VKAs to DOACs. In patients with an increased risk of falling, data suggest no relevant interaction of increased risk of falling on the effectiveness and safety of DOACs compared with warfarin. Resuming oral anticoagulants in patients with Af after major bleeding seems to be beneficial. Off-label low-dose DOAC is often prescribed to patients who were underrepresented in larger randomized trails because of an elevated risk of bleeding or overexposure to DOACs, but its effect on clinical outcomes remains uncertain.

Conclusions

DOACs are the recommended oral anticoagulant for vital older patients with AF. The scarcity of data backing DOAC use in frail individuals, those with renal impairments, or significant bleeding history underscores the necessity for further investigation. However, existing evidence suggests at least similar effectiveness and safety and potential benefits for DOACs in these patient subsets. Therefore, there is no reason to suggest these patients should be treated differently than the established guidelines regarding anticoagulation.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40266-024-01138-5.

Key Points

Treatment with anticoagulants is indicated for vital older patients and DOACs are the preferred oral anticoagulant.

Results from non-randomized comparisons suggest that the treatment effect of DOACs compared with VKAs is similar between frail and nonfrail patients. Therefore, there is no reason to suggest these patients should be treated differently than nonfrail patients.

One randomized trial suggests that frail patients, who are treated with a VKA, have an increased bleeding risk after switching to a DOAC.

Introduction

Since their introduction, direct oral anticoagulants (DOACs) have significantly simplified anticoagulant therapy. The primary advantage of DOACs over vitamin K antagonists (VKAs) lies in their utility, as they can be administered in fixed doses, while being at least as effective and safe as warfarin for preventing ischemic strokes in most patients with atrial fibrillation (AF) [1–4]. However, this lack of a need for anticoagulant activity monitoring can also be a disadvantage for certain patients. Evidence suggests that the risk of bleeding increases steadily with greater DOAC exposure, whereas the relationship between thromboembolic events and exposure is less pronounced [5–8]. This observation indicates that the relative treatment effect of DOACs compared with VKAs may differ in patients who were underrepresented or systematically excluded in randomized trials because they were at very high bleeding risk or overexposed to DOACs. Such patients include the very old patients or frail and those with chronic kidney disease (CKD), hepatic insufficiency, alcohol abuse, uncontrolled hypertension, or a history of major bleeding.

In this review, we discuss the available evidence on the efficacy and safety of DOACs compared with VKAs in multiple subgroups of patients at very high risk of bleeding or overexposure to DOACs, that is, the very old patients or frail and those with chronic kidney disease (CKD), uncontrolled hypertension, or a history of major bleeding [9–11]. Furthermore, we address knowledge gaps concerning patients deemed unsuitable for oral anticoagulants (OACs), the role of left atrial appendage closure, and future developments in factor XIa-inhibitors.

Method

For this review, a literature search was conducted on Pubmed. The following terms were used: “atrial fibrillation”, “oral anti-coagulation,” ”VKAs,” “warfarin,” “phenprocoumon,” “acenocoumarol,” “DOACs,” “Apixaban,” “Dabigatran,” “Edoxaban,” “Rivaroxaban,” “older,” “elderly,” “octogenarian,” “nonagenarian,” “frailty,” “vulnerable,” “polypharmacy,” “comorbidity,” “falls,” “kidney disease,” “hypertension,” “history of major bleeding,” “low dose DOAC,” “left atrial appendage closure,” “factor XIa inhibitor.” A separate search was conducted for every condition. Additional articles were included by snowballing. These search strategies can be found in the Supplementary Table 1 and Fig. 1. Relevant articles published by the 8 March 2023 were used. RCTs, systematic reviews, meta-analyses, and observational studies were included. High-impact articles that were published after this date were also included. Figure 1 shows the consort diagram for the combined search strategy.

Fig. 1.

Consort diagram of article selection

Old age

Older adults have a higher risk of stroke and systemic embolism (SSE), major bleeding, intracranial hemorrhage, and mortality compared with younger patients, regardless of the type of oral anticoagulant used [12]. Nevertheless, subanalyses of randomized trials data indicate that the therapeutic advantage of DOACs is not significantly affected in the vital older patients, despite the elevated absolute risk [1, 2].

A meta-analysis of five randomized controlled trials (RCTs) assessed the efficacy and safety of DOACs relative to VKAs in patients stratified by age, namely those younger and older than 75 years old. The analysis included 24,768 patients aged over 75 years representing 38.4% of the total trial population. In this subgroup, SSE incidence was 3.8% among patients treated with DOACs and 4.8% of those receiving VKAs [relative risk (RR) 0.83, 95% confidence interval (CI) 0.69–1.00, P = 0.04, I² = 55%)] [13].

For major bleeding events, the older adult’s cohort demonstrated rates of 7.7% and 8.1% for DOACs and VKAs, respectively (RR 0.91, 95% CI 0.73–1.15, I² = 85%, P = 0.44). Notably, major bleeding events in the DOAC group were predominantly gastrointestinal. A significant reduction in intracranial bleeding was associated with DOAC use compared to VKAs, independent of age (RR 0.49, 95% CI 0.35–0.69, I² = 37%, P < 0.0001; Table 1) [13].

Table 1.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in older patients.

Author	DOAC vs VKA	Outcome	HR/RR (95%-confidence interval)
Gencer et al. 2022	Edoxaban	NCO	HR 0.87; (0.79–0.97)*
Kim et al. 2018	All DOACs	SSE	RR 0.81; (0.72–0.91)*
		Mortality	0.94; (0.88–1.01)
		Major bleeding	0.91; (0.73–1.15)
		ICH	0.48; (0.38–0.67)*
		GI bleeding	1.53; (1.27–1.85)*
Bonanad et al. 2021	All DOACs	SSE	RR 0.72; (0.61–0.84)*
		Mortality	0.82; (0.70–0.96)*
		Major bleeding	0.85; (0.69–1.04)
		ICH	0.47; (0.36–0.60)*
Deitelzweig et al. 2019	Apixaban	SSE	HR 0.58; (0.49–0.69)*
		Major bleeding	0.60; (0.54–0.67)*
		ICH	0.53; (0.41–0.68)*
	Dabigatran	SSE	0.77; (0.60–0.99)*
		Major bleeding	0.92; (0.78–1.07)
		ICH	0.51; (0.33–0.79)*
	Rivaroxaban	SSE	0.74; (0.65–0.85)*
		Major bleeding	1.16; (1.07–1.24)*
		ICH	0.78; (0.64–0.95)*
		GI bleeding	1.33; (1.20–1.47)
Mitchell et al. 2019	Dabigatran, Rivaroxaban, Apixaban	SSE	HR 0.89; (0.82–0.96)*
		Composite effectiveness	0.95; (0.91–0.99)*
		Composite safety	0.97; (0.93–1.00)
Granger et al. 2011	Apixaban	SSE	HR 0.79; (0.66–0.95)*
		Major bleeding	0.69; (0.60–0.80)*
Patel et al. 2011	Rivaroxaban	SSE	HR 0.79; (0.66–0.96)*
		Major or clinically relevant bleeding	1.03; (0.96–1.11)
Shen et al. 2020	All DOACs	SSE
		Observational studies	HR 0.87; (0.81–0.94)*
		RCTs	0.82; (0.67–0.96)*
		Mortality
		Observational studies	1.01; (0.92–1.11)
		RCTs	0.94; (0.87–1.00)
		Major bleeding
		Observational studies	0.87; (0.77–0.98)*
		RCTs	0.89; (0.66–1.12)
		ICH
		Observational studies	0.47; (0.37–0.57)*
		RCTs	0.47; (0.31–0.63)*
Silverio et al. 2021	All DOACs	SSE	HR 0.79; (0.70–0.89)*
		Major bleeding	0.94; (0.85–1.05)
Lee et al. 2022	All DOACs	SSE	RR 0.82; (0.75–0.90)*
		Mortality	0.85; (0.75–0.96)*
		Major bleeding	0.88; (0.79–0.98)*
		ICH	0.38; (0.28–0.52)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, NCO net clinical outcome, SSE Stroke and systemic embolism, GI Gastrointestinal, RCT Randomized controlled trial, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio

*Statistically significant

Mortality rates were also lower in DOAC users. In patients under 75 years old, mortality was 6% for DOAC users versus 6.9% for VKA users. Among those over 75 years old, mortality rates were 10.7% and 11.6% in DOAC and VKA users, respectively (RR 0.91 95% CI 0.83–1.00, I² = 43%, P = 0.05)[13]. A meta-analysis of patients older than 80 years old, corroborated these findings (mortality RR 0.82, 95% CI 0.70–0.96, P = 0.012) [14].

Above described findings are consistent in both RCTs and observational studies (Table 1) [15–19].

In conclusion, current evidence indicates that DOACs are at least as effective as, if not superior to, VKAs in the vital older adults with non-valvular AF.

Frailty

Frailty can be defined and quantified in various ways. Usually, a sum score of unintentional weight loss, exhaustion, low physical activity, functional disabilities, gait speed, muscular weakness, social isolation, psychiatric and cognitive symptoms, multimorbidity, and polypharmacy is used [20, 21]. Due to the differing definitions and the systematic exclusion or underrepresentation of the frailest patients in phase III studies, there is a scarcity of high-quality data on the efficacy and safety of anticoagulants in these patients.

This section discusses the results of subgroup analyses of the large phase III trials on the impact of frailty components on the treatment effects of DOACs versus VKAs, as well as the findings from observational studies focusing on frail patients see Table 2 [22–27].

Table 2.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in frail patients.

Author	DOAC vs VKA	Outcome	HR/RR (95%-confidence interval)
Jaspers Focks et al. 2016	Apixaban	SSE
		0–5 drugs	HR 0.86; (0.83–1.17)
		6–8 drugs	0.76; (0.57–1.03)
		> 9 drugs	0.76; (0.54–1.07)
		Major bleeding
		0–5 drugs	0.50; (0.38–0.66)*
		6–8 drugs	0.72; (0.56–0.91)*
		> 9 drugs	0.84; (0.67–1.06)
		Mortality
		0–5 drugs	0.86; (0.70–1.05)
		6–8 drugs	0.89; (0.74–1.06)
		> 9 drugs	0.94; (0.77–1.14)
Alexander et al. 2019	Apixaban	SSE
		No multi-morbidity	HR 0.79; (0.57–1.11)
		Moderate multi-morbidity	0.72; (0.56–0.93)*
		High multi-morbidity	0.93; (0.57–1.50)
		Mortality
		No multi-morbidity	0.73; (0.58–0.92)*
		Moderate multi-morbidity	0.96; (0.82–1.13)
		High multi-morbidity	0.89; (0.70–1.11)
		Major bleed
		No multi-morbidity	0.64; (0.49–0.84)*
		Moderate multi-morbidity	0.67; (0.55–0.82)*
		High multi-morbidity	0.82; (0.59–1.13)
Martinez et al. 2018	Rivaroxaban	SSE	HR 0.68 (0.49–0.95)*
		Major bleeding	1.04; (0.81–1.32)
		ICH	0.49; (0.23–1.04)
	Apixaban	SSE	0.78 (0.46–1.35)
		Major bleeding	0.72; (0.49–1.06)
		ICH	0.97; (0.28–3.33)
	Dabigatran	SSE	0.94 (0.60–1.45)
		Major bleeding	0.87; (0.63–1.19)
		ICH	0.14; (0.02–1.11)
Nicolau et al. 2020	Edoxaban	SSE
		CCI = 0	HR 0.88; (0.67–1.16)
		CCI = 1	0.73; (0.41–1.30)
		CCI = 2	0.90; (0.70–1.15)
		CCI = 3	0.73; (0.46–1.15)
		CCI > 4	1.02; (0.59–1.75)
		Major bleeding
		CCI = 0	0.83; (0.66–1.05)
		CCI = 1	0.79; (0.51–1.23)
		CCI = 2	0.71; (0.57–0.88)*
		CCI = 3	0.96; (0.69–1.34)
		CCI > 4	0.81; (0.54–1.24)
		GI-bleeding
		CCI = 0	1.28; (0.81–2.04)
		CCI = 1	1.53; (0.55–4.30)
		CCI = 2	0.89; (0.60–1.33)
		CCI = 3	1.90; (1.04–3.49)*
		CCI > 4	2.39; (0.94–6.11)
		NCO
		CCI = 0	0.92; (0.80–1.06)
		CCI = 1	0.76; (0.59–0.98)*
		CCI = 2	0.86; (0.76–0.97)*
		CCI = 3	0.88; (0.74–1.06)
		CCI > 4	1.09; (0.86–1.38)
Piccini et al. 2016	Rivaroxaban	SSE
		5–9 vs 0–4 Medications	HR 1.07; (0.89–1.29)
		≥ 10 vs 5–9 Medications	0.96; (0.57–1.62)
		≥ 10 vs 0–4 Medications	1.02; (0.76–1.38)
		Major bleeding
		5–9 vs 0–4 Medications	1.46; (1.22–1.76)
		≥ 10 vs 5–9 Medications	1.31; (0.78–2.18)
		≥ 10 vs 0–4 Medications	1.91; (1.52–2.41)
Hohmann et al. 2019	All DOACs	Frailty
		SSE	HR 0.91; (0.77–1.07)
		Major bleeding	0,73; (0.60–0.89)*
		ICH	0.52; (0.41–0.67)*
		CCI>4
		SSE	0.94; (0.79–1.13)
		Major bleeding	0.70; (0.56–0.87)*
		ICH	0.52; (0.39–0.69)*
		Polypharmacy >7 medications
		SSE	0.89; (0.75–1.05)
		Major bleeding	0.72; (0.59–0.87)*
		ICH	0.54; (0.42–0.70)*
Kim et al. 2021	Apixaban	Composite of mortality, SSE, major bleeding
		Non-frail	HR 0.61; (0.52–0.71)*
		Pre-frail	0.66; (0.61–0.70)*
		Frail	0.73; (0.67–0.80)*
	Dabigatran	Composite of mortality, SSE, major bleeding
		Non-frail	0.81; (0.68–0.97)*
		Pre-frail	0.98; (0.90–1.08)
		Frail	1.09; (0.96–1.23)
	Rivaroxaban	Composite of mortality, SSE, major bleeding
		Non-frail	0.88; (0.77–0.99)*
		Pre-frail	1.04; (0.98–1.10)
		Frail	0.96; (0.89–1.04)
Wilkinson et al. 2020	Edoxaban	SSE
		Fit	HR 1.03; (0.71–1.49)
		Pre-frail	0.82; (0.66–1.01)
		Mild-moderate frail	0.84; (0.61–1.15)
		Severe frail	0.54; (0.20–1.50)
		Major bleeding
		Fit	0.96; (0.71–1.30)
		Pre-frail	0.76; (0.64–0.90)*
		Mild-moderate frail	0.75; (0.57–0.98)*
		Severe frail	0.60; (0.29–1.26)
Zeng et al. 2022	All DOACs	SSE	HR 0.79; (0.69–0.90)*
		Mortality	0.90; (0.84–0.96)*
		Major bleeding	0.79; (0.64–0.97)*
		ICH	0.58; (0.52–0.65)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, NCO net clinical outcome, SSE stroke and systemic embolism, GI gastrointestinal, RCT randomized controlled trial, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio, CCI Charlson Comorbidity Index; moderate multi-morbidity: 3–5 comorbidities; high multi-morbidity: > 5 comorbidities

*Statistically significant

Frailty Indices

In a retrospective cohort study, the comparison between DOACs and warfarin in older adults with AF was conducted. Frailty was assessed using a Clinical Frailty Index (CFI), which estimates frailty based on 93 variables defined by diagnosis, health services, and medical codes. Nonfrailty was defined as CFI < 0.15, prefrailty as CFI 0.15–0.24, and frailty as CFI > 0.25. The findings suggested that in patients with high frailty indices, the use of rivaroxaban (n = 275,944, of whom 156,540 were defined as prefrail and 48,272 as frail) and dabigatran (n = 158,730, 88,576 defined as prefrail and 25,232 as frail) was associated with a similar risk of the composite of ischemic stroke, major bleeding, or all-cause death compared with VKAs. However, in the apixaban–warfarin cohort (n = 218,738, 125,322 defined as prefrail and 41,250 as frail) the incidence of the composite endpoint was significantly lower in the apixaban group in nonfrail, prefrail, and frail subgroups [hazard ratio (HR) frail 0.73, 95% CI 0.67–0.80; Table 2) [28].

In a post hoc analysis of the ENGAGE AF-TIMI 48 trial (fit: n= 4459, prefrail: n = 12,326, mild-moderate frail: n = 3722, severe frail: n = 360), edoxaban was associated with similar SSE rates in every frailty category and a reduction in major bleeding in prefrail and mild-to-moderate frailty groups compared with warfarin (HR 0.75, 95% CI 0.57–0.98; Table 2). Frailty was defined using a cumulative deficit model developed with 40 available variables. This model identified health deficits based the principle that more deficits correlate with increased frailty. A dose reduction from 60 to 30 mg was associated with reduced major bleeding compared with warfarin, except in those with severe frailty. Edoxaban 60 mg also reduced major bleeding in prefrail and mild-to-moderate frailty groups [29].

A meta-analysis using a claims-based frailty index concluded that DOAC use was significantly associated with reduced risks of SSE (HR 0.79, 95% CI 0.69–0.90, I² = 71%), major bleeding (HR 0.79, 95% CI 0.64–0.97, I² = 97%), intracranial bleeding (HR 0.52, 95% CI 0.35–0.76, I² = 67%) and all-cause mortality (HR 0.96, 95% CI 0.84–0.96, I² = 45%) compared with VKAs [30]. The nonrandomized design and their heterogeneous results on treatment effect estimates preclude us from drawing definite conclusions.

Polypharmacy

Several studies have investigated the effect of polypharmacy as a surrogate marker for frailty on the efficacy and bleeding risk of DOACs vs VKAs. These studies collectively show that DOACs are as effective and safe as warfarin when conditioned for polypharmacy (Table 2) [23, 26, 27].

A subanalysis of the ROCKET-AF trial (rivaroxaban versus warfarin) involving 14,264 patients stratified by the number of comedications (excluding the anticoagulant), indicated that rivaroxaban was as effective (SSE) and safe (major bleeding) as VKAs regardless of the number of comedications (P interaction 0.99 and 0.87, respectively; Table 2) [26].

Similarly, a post hoc sub analysis of the ARISTOTLE database (apixaban versus warfarin) involving 18,201 patients found that apixaban was at least as effective as VKAs in preventing SSE across all subgroups (< 5 drugs, 6–8 drugs, and > 9 drugs). Major bleeding was less frequent in the apixaban group, independent of the number of comedications [< 5 drugs: HR 0.50 (0.38–0.66) and 6–8 drugs: HR 0.72 (0.56–0.91); although, in patients using nine or more drugs (n = 4,756), the difference was not significant [23]].

Multimorbidity

Using data from the same ARISTOTLE trial, an analysis based on the number of comorbidities was used as a surrogate for frailty. Of the 16,800 patients analyzed, 6087 (36.2%) had zero to two comorbidities, 8491 (50.5%) had three to five comorbidities, and 2222 (13.2%) had six or more comorbidities. In each comorbidity subgroup, SSE, mortality, and major bleeding were less frequent in the apixaban-treated group compared with VKAs with no interaction in relation to effectiveness or safety of apixaban [22].

In a post hoc analysis of the ENGAGE-AF-TIMI 48 study, edoxaban was compared with VKAs in different subgroups of the Charlson Comorbidity Index (CCI), a score to predict mortality risk based on comorbidities (CCI of 0, n = 6756; CCI of 1, n = 1533; CCI of 2, n = 8875; CCI of 3, n = 2677; CCI > 4, n = 1264). Edoxaban was at least as effective as VKA in preventing SSE and cardiovascular-related death, independent of the CCI classification. Though major bleeding was consistently lower in the edoxaban-treated group, there was an association between higher CCI scores and increased risk of gastrointestinal bleeding in the edoxaban group (CCI of 3, HR 1.90, 95% CI 1.04–3.49, P for interaction of 0.16; Table 2) [25].

Switching from VKAs to DOACs in Frail Patients

A multicenter open-label RCT in the Netherlands investigated whether frail patients could benefit from switching from a VKA to a DOAC [43]. In this trial, frailty was defined as being aged 75 years or older with a Groningen Frailty Indicator (GFI) ≥ 3. A total of 662 patients were switched from VKA to DOACs, and 661 patients continued VKAs. The trial was stopped for futility due to the higher incidence of bleeding complications in the switch arm (HR for major of clinically relevant bleeding 1.69, 95% CI 1.23–2.32). It should be noted that there were few major bleedings in both treatment arms, and most clinically relevant nonmajor bleedings were patient-reported, suggesting a potential for reporting bias. Hence this trial concluded that careful consideration is needed when deciding between continuing VKAs or switching to DOACs in frail older patients [31].

In conclusion, DOAC’s are generally at least as safe and effective in nonfrail patients with AF as VKAs. However, this remains uncertain for frail patients with AF due to the lack of randomized data. Nonetheless, the available data showed there is no reason to assume that in frail patients the effectiveness and safety profiles of DOACs differ from those of VKAs in nonfrail patients [22–27]. Future randomized controlled trials are needed to definitively assess the comparative efficacy of DOACs and VKAs in this vulnerable population [28].

Increased Risk of Falling

Patients using anticoagulants who have an increased risk of falling are at higher risk of complications, with major (intracranial) bleeding being the most feared [32, 33]. Therefore, it is important to identify these patients. Criteria for identifying these patients include a history of falling, muscle weakness in the lower extremities, poor balance, cognitive decline, orthostatic hypotension, vertigo, use of psychotropic medication, and severe arthritis [33]. Addressing modifiable factors, potentially with the assistance of a geriatrician, can help to reduce the risk of falling [34, 35].

Balancing Risks and Benefits of Anticoagulation

For most patients, the risk and consequences of SSE outweigh the risk of major bleeding, supporting the initiation of anticoagulant therapy [32–35]. However, the choice of anticoagulant remains unclear. Subgroup analyses of phase III studies suggest a potential advantage of DOACs over VKAs, but this finding may not be generalizable to patients at high risk of falling in the community, as these were underrepresented in these studies [32, 33].

A prespecified substudy the ENGAGE-AF-TIMI 48 focused on patients with an increased risk of falling. Out of 21,105 patients, only 900 (4%) were identified as having an increased risk. Major bleeding occurred in 5.43% of patients in the edoxaban-treated group and 5.55% in the warfarin-treated group. No fatal bleeding was reported the edoxaban group, while four fatal bleeding events occurred in the warfarin group. There was no significant treatment interaction between edoxaban and warfarin regarding effectiveness and safety outcomes (Table 3) [33].

Table 3.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in patients with an increased risk of falling

Author	DOAC vs VKA	Outcome	HR/RR (95%-confidence interval)
Gencer et al. 2022	Edoxaban	NCO	HR 0.96; (0.73–1.27)
Gao et al. 2022	All DOACs	SSE	OR 0.80; (0.68–0.95)*
		Mortality	0.86; (0.68–1.09)
		Major bleeding	0.73; (0.53–1.01)
		ICH	0.24; (0.09–0.69)*
Steffel et al. 2016	Edoxaban	SSE	HR 0.96; (0.53–1.75)
		Mortality	0.91; (0.64–1.29)
		Major bleeding	0.96; (0.59–1.56)
		ICH	0.16; (0.04–0.71)*
Rao et al. 2018	Apixaban	SSE	HR 0.88; (0.40–1.93)
		Mortality	0.96; (0.63–1.44)
		Major bleeding	0.81; (0.48–1.36)
		ICH	0.19; (0.04–0.88)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, NCO net clinical outcome, SSE stroke and systemic embolism, RCT randomized controlled trial, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio, CCI Charlson Comorbidity Index;

*Statistically significant

In the ARISTOTLE trial, 753 (4%) patients had a history of falling. Among these patients, major bleeding was twice as common in the apixaban-treated group and five times more common in the warfarin-treated group compared to those without a history of falling. DOACs were as safe as warfarin for major bleeding and sometimes superior for superior for ICH (HR 0.19, 95% CI 0.04–0.88; Table 3) [32].

A meta-analysis of three cohort studies evaluated the effectiveness and safety of DOACs compared with warfarin in patients with an increased risk of falling. DOACs were associated with a significantly reduced risk of ICH (RR of 0.28, 95% CI 0.10–0.75) compared with warfarin. There were no significant differences in SSE, all-cause mortality, or major bleeding (Table 3) [36].

Current data suggest no significant interaction between an increased risk of falling on the effectiveness and safety of DOACs compared with warfarin, (Table 3) [32, 33, 37, 38]. DOACs are at least as effective as VKAs. However, adequately powered randomized trials are needed to assess the relative treatment effect of these anticoagulants in patients with increased risk of falling.

Chronic Kidney Disease

AF and chronic kidney disease (CKD) frequently coexist and exert substantial mutual influence. Patients with both conditions have an increased risk of both SSE and major bleeding. DOACs, which are partially eliminated through renal pathways, exhibit varying degrees of renal clearance: dabigatran is predominantly cleared renally (80%), whereas the renal elimination of edoxaban (50%), rivaroxaban (35%), and apixaban (35%) is substantially lower [39]. Dabigatran is contraindicated in patients with a GFR of < 30 ml/min/1.73 m² [35].

Older studies supported adequate anticoagulation in patients with stage 3 and 4 CKD (eGFR 15–60 ml/min/1.73m²) and AF due to a higher SSE risk outweighing major bleeding risk [34, 40]. Recent phase III studies support the superiority of DOACs over VKAs in moderate to severe CKD (eGFR 30–50 ml/min/1.73m²) (Table 4) [35, 41, 42], necessitating dose adjustments to maintain safety and effectiveness advantages [34, 35, 43]. Meta-analyses comparing edoxaban with warfarin in CKD patients (eGFR 30–50 ml/min/1.73m²) showed similar effectiveness of edoxaban 30 mg and warfarin [44].

Table 4.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in patients with chronic kidney disease.

Author	DOAC	Outcome	HR/RR (95%-confidence interval)
Ha et al. 2019	All DOACs	CrCl 20–60 ml/min
		SSE	RR 0.79; (0.66–0.93)*
		Mortality	0.88; (0.78–0.99)*
		Major bleeding	0.80; (0.61–1.04)
		ICH	0.49; (0.30–0.80)*
Wang et al. 2023	Edoxaban	CrCl 30–50 ml/min
		SSE	HR 0.85; (0.39–1.87)
		Mortality	0.65, (0.35–1.19)
		Risk of bleeding	0.75, (0.60–0.93)*
		NCO	0.75, (0.63–0.90)*
Stanifer et al. 2020	Apixaban	CrCl > 30 ml/min
		SSE	HR 0.80 (0.67–0.96)*
		Mortality	0.89 (0.79–0.97)*
		Major bleeding	0.69 (0.62–0.76)*
		ICH	0.42 (0.31–0.59)*
		CrCl 25–30 ml/min
		SSE	0.55 (0.20–1.51)
		Mortality	1.02 (0.64–1.67)
		Major bleeding	0.34 (0.14–0.80)*
		ICH	–
Rhee et al. 2022	All DOACs	CrCl < 60 ml/min
		SSE	HR 0.78; (0.73–0.85)*
		Mortality	0.83; (0.72–0.96)*
		Major bleeding	0.76; (0.64–0.89)*
		CrCl < 50 ml/min
		SSE	0.71; (0.59–0.85)*
		Mortality	0.82; (0.67–1.00)
		Major bleeding	0.74; (0.61–0.89)*
		CrCl < 30 ml/min
		SSE	0.60; (0.43–0.85)*
		Mortality	0.81; (0.64–1.02)
		Major bleeding	0.74; (0.59–0.93)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, NCO net clinical outcome, SSE stroke and systemic embolism, RCT randomized controlled trial, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio, CrCl creatinine clearance.  

*Statistically significant

In severe CKD (eGFR < 30 ml/min/1.73 m²) the decision to anticoagulate is complex [42]. The ARISTOTLE trial’s subgroup analysis of patients with creatinine clearance of 25–30 ml/min/1,73 m²) found apixaban superior to warfarin in reducing SSE and major bleeding risks. In the apixaban arm, no intracranial hemorrhages occurred, compared with four patients (2.4 per 100 patient-years) in the warfarin arm [45].

A study of stage 4 CKD (eGFR 15–29 ml/min/1.73 m²), involving 92 patients receiving warfarin and 90 patients receiving DOACs [29, 57], showed comparable major bleeding rates (15.6% versus 14.1%). However, patients on warfarin therapy were monitored and optimized closely, which may not always feasible in daily practice [46].

There is no strong evidence that anticoagulation is beneficial in patients with end-stage renal disease or those undergoing dialysis. The optimal approach to anticoagulation in this population remains uncertain, despite it is common practice. The RENAL AF study, investigating apixaban versus warfarin in patients on dialysis, showed that the incidence of bleeding was ten times higher than that of thromboembolic events in anticoagulated patients [47]. This finding raises questions whether the therapeutic benefits of anticoagulation outweigh the associated risks in this patient cohort. However, when used, DOACS seem to be safer than VKAs.

Meta-analyses showed lower risks of SSE and major bleeding, regardless of the severity of renal impairment in DOACs compared with VKAs (Table 3), particularly edoxaban and apixaban, which presented the highest rank probability for reducing SSE (P score edoxaban 94.5%) and major bleeding (P score apixaban 95.8%) [48]

Given the absence of direct randomized comparisons, it remains difficult to choose the optimal anticoagulant in those with CKD for whom the benefit is expected to outweigh the risk [35]. However, DOACs seem at least as effective and safe as VKAs.

History of Major Bleeding

One of the most mentioned reasons for hesitating to initiate or discontinue anticoagulants is a history of major bleeding, particularly intracranial bleeding [11, 49].

A meta-analysis involving 5685 patients with AF with prior major bleeding (comprising six retrospective cohort studies and one small prospective cohort study), compared outcomes between patients who resumed OAC (either VKA or DOAC) with those who OAC was discontinued it after an index bleeding event. Patients who resumed OACs directly experienced a 46% relative risk reduction in SSE and a 10.8% absolute risk reduction in all-cause mortality compared with those who discontinued OAC. However, resuming OACs was associated with an increased risk of recurrent bleeding (OR 1.85, 95% CI 1.48–2.30, I² = 0%); although, there was no increased risk for an index bleeding event. Net clinical benefit (NCB) analysis indicated that resuming OAC conferred a clinical advantage (NCB 0.11, 95% CI 0.09–0.14) [50]. Given that all included studies were observational, the results are likely to be affected by the clinician’s decision to (not) reinitiate anticoagulation.

The ANNEXA-4 study, investigating an antidote for coagulation factor Xa-inhibitors (rivaroxaban, apixaban, and edoxaban) in patients with major bleeding, emphasizes the importance of resuming anticoagulation therapy. Within 30 days postantidote use and DOAC discontinuation, 34 patients (10%) experienced an SSE, compared with none in the group that resumed anticoagulation. A major part of mortality was attributed to SSEs [51].

In a recent systematic review and meta-analysis encompassing 56,697 patients with AF with prior bleeding from five retrospective cohort and one clinical trial, DOACs were associated with a lower risk of ischemic stroke (HR 0.73, 95% CI 0.59–0.91), and all-cause mortality (HR 0.70, 95% CI 0.50–0.98) than VKAs. There were no significant differences in treatment effects between the RCT and cohort studies. DOACs also exhibited a significantly lower risk of major bleeding events, compared with warfarin (HR 0.75, 95% CI 0.67–0.84) and a significant risk reduction in ICH (HR 0.63, 95% CI 0.48–0.82; Table 5). Among patients with previous intracranial hemorrhage, DOACs were associated with a significant reduction of ischemic stroke and ICH relative to warfarin; although, differences in all-cause mortality were not statistically significant [52].

Table 5.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in patients with previous major bleeding

Author	DOAC	Outcome	HR/RR (95%-confidence interval)
De Caterina et al. 2016	Apixaban	Major bleeding
		Without history of major bleeding	RR 0.66; (0.56–0.78)*
		Without history of ICH	0.42 (0.29–0.59)*
		History of major bleeding	0.55 (0.20–1.48)
		History of ICH	0.42 (1.18–0.96)
Suah et al. 2022	All DOACs	Previous ICH
		SSE	HR 0.83; (0.69–0.99)*
		mortality	0.67; (0.44–1.03)
		Major bleeding	0.75; (0.57–0.99)*
		ICH	0.69; (0.54–0.89)*
		Previous GI bleeding
		SSE	0.61; (0.54–0.68)*
		Mortality	0.81; (0.52–1.26)
		Major bleeding	0.76; (0.65–0.90)*
		ICH	0.62; (0.24–1.61)
		Previous major bleedinga
		SSE	0.73; (0.59–0.91)*
		Mortality	0.70; (0.50–0.98)*
		Major bleeding	0.75; (0.67–0.84)*
		ICH	0.63; (0.48–0.82)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, SSE stroke and systemic embolism, RCT randomized controlled trial, GI gastrointestinal, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio

*Statistically significant

^aIncluding previous ICH and GI bleeding.

A subgroup analysis from the ARISTOTLE study (apixaban versus warfarin) suggests that a history of bleeding is associated with several risk factors for stroke and portends a higher risk of major—but not intracranial—bleeding, during anticoagulation. However, the beneficial effects of apixaban over warfarin for effectiveness and safety remained consistent regardless of history of bleeding [53]. Phase III studies underscored the superiority of DOACs over VKAs, with a 14% relative reduction in major bleeding, and a 52% relative risk reduction in intracranial bleeding [1–3, 54, 55].

In patients with prior gastrointestinal bleeding, resumption of anticoagulants showed increased risk of recurrent gastrointestinal bleeding with warfarin but not in DOACs [56, 57], except for rivaroxaban [57].

In conclusion, resuming OACs in patients with AF after major bleeding appears to be beneficial, with a cautious preference for DOACs. However, the optimal timing for resuming OAC therapy remains unclear and may depend on the bleeding location, risk of SSE and individual patient characteristics. Patients with a high risk of recurrent ICH may require delayed resumption of anticoagulation, while patients with a high stroke risk may resume therapy sooner [58]. Cerebral amyloid angiopathy (CAA) associated with ICH represents a contraindication for anticoagulation due to the high risk of ICH recurrence [35].

Uncontrolled Hypertension

Controlling hypertension is essential in patients with AF and is related to fewer episodes of AF, SSE, and major bleeding [34, 35, 59–62]. A subgroup analysis of the ROCKET-AF-trial (rivaroxaban versus warfarin) indicates that the risk of SSE increases by 7% in every 10 mmHg increase in blood pressure, while the risk of major bleeding remains unchanged in these blood pressure changes. Controlled or uncontrolled hypertension was not associated with a higher risk of major bleeding compared to no history of hypertension [60]. Refraining from anticoagulants in the setting of uncontrolled hypertension is therefore not justified [60, 63]. Subgroup analyses of other phase III studies suggest that DOACs are more effective and safer than VKAs, independent of the stage of hypertension [60, 61, 63] or a history of hypertension (Table 6) [59, 62].

Table 6.

Overview of RCT’s, meta-analyses and observational studies investigating DOAC vs VKA in patients with (uncontrolled) hypertension.

Author	DOAC	Outcome	HR/RR (95%-confidence interval)
Rao et al. 2015	Apixaban	History of HTN
		SSE	HR 0.82; (0.68–1.10)
		Mortality	0.92; (0.77–1.08)
		Major bleeding	0.68; (0.61–0.76)*
Vemulapalli et al. 2016	Rivaroxaban	Controlled HTN
		SSE	HR 0.82; (0.66–1.03)
		Mortality	0.74 (0.72–0.97)*
		Major bleeding	1.13; (0.94–1.36)
		ICH	0.64; (0.40–1.03)
		Uncontrolled HTN
		SSE	0.95; (0.73–1.24)
		Mortality	1.11; (0.91–1.36)
		Major bleeding	0.99; (0.77–1.26)
		ICH	0.59; (0.34–1.02)
Park et al. 2019	Edoxaban	140 to < 150mmHg
		SSE	HR 0.68; (0.40–1.16)
		Major bleeding	0.61; (0.41–0.91)*
		ICH (> 140mmHg)	0.48; (0.25–0.92)
		> 150mmHg
		SSE	0.65; (0.24–1.78)
		Major bleeding	0.73; (0.39–1.38)
Nagarakanti et al. 2015	Dabigatran	History of HTN
		SSE	HR 0.66; (0.52–0.84)*
		Major bleeding	1.00; (0.85–1.17)
		ICH	0.36; (0.23–0.57)*
Harskamp et al. 2022	All DOACs	Controlled HTN
		SSE	HR 0.82; (0.69–0.97)*
		Major bleeding	0.79; (0.56–1.11)
		ICH	0.46; (0.33–0.65)*
		Uncontrolled HTN
		SSE	0.87; (0.69–1.10)
		Major bleeding	0.92; (0.74–1.15)
		ICH	0.58; (0.40–0.85)*

DOAC direct oral anticoagulant, VKA vitamin K antagonist, SSE stroke and systemic embolism, RCT randomized controlled trial, ICH intracranial hemorrhage, RR relative risk, HR hazard ratio, HTN hypertension.

*Statistically significant

Low Dose DOACs

There is ongoing debate about the use of lower doses in selected patients, even if they do not meet the criteria for dose reduction according to the label, often defined as off-label dose reduction.

Currently, there is no high-quality evidence supporting off-label dosing of DOACs in any population. Numerous studies have attempted to determine the effect of off-label dosing in patients with AF but failed to do so validly due to inappropriate patient selection [64]. However, we believe there is a rationale for testing the effect of such a strategy in selected patients based on three observations.

First, many patients, including the very older adults and frail patients, were systematically excluded or underrepresented in the randomized trials comparing the DOACs with warfarin [65, 66]. Consequently, it is uncertain whether the fixed dosing regimens tested in these trials are appropriate for these populations [65, 67].

Second, the risk of bleeding increases with greater exposure to DOACs, whereas the relationship between thromboembolic events and exposure is less pronounced [5–8].

Third, treating patients at high risk of bleeding or overexposure to DOACs with conventional doses is likely to result in unacceptable high bleeding risks. For instance, in ELDERCARE AF, Japanese patients with AF who were considered ineligible for treatment with currently approved strategies were randomized to edoxaban 15 mg (versus 30 or 60 mg) once daily or placebo. Edoxaban 15 mg significantly reduced the risk of thromboembolic events (2.3% versus 6.7% per year; HR 0.34, 95% CI 0.19–0.61) while maintaining an acceptable risk of major bleeding (3.3% versus 1.8%; HR 1.87, 95% CI 0.90–3.89) [68]. Given the strong evidence of higher risks of major bleeding with higher absolute doses, using the conventional doses would likely have led to unacceptably high bleeding risks in these patients [6]. Another example is the more recent published FRAIL AF randomized trial. In this trial, frail patients treated in Dutch primary care practices and who were receiving VKA were randomized to switch to a DOAC at conventional doses (though 6.6% were treated with an off-label dose reduction) or to continue treatment with a VKA [31]. This trial was stopped for futility because the risk of major or clinically relevant nonmajor bleeding was 69% higher for the patients who switched to DOACs (15,3% versus 9.4% per year, HR 1.69, 95% CI 1.23–2.32), while the risk of thromboembolic events was not significantly different between the treatment arms (2.4% versus 2.0% per year, HR 1.26, 95% CI 0.60–2.61). A possible explanation for the higher risk of bleeding in the patients who were switched is relative overexposure to DOACs [21].

In summary, there is evidence supporting the potential utility of off-label dose reductions in selected patients, but high-quality evidence is not yet available. Low dosages of DOACs are, in most instances, preferable to no anticoagulant therapy at all.

Future Perspectives

Factor XIa Inhibitors

Patients with FXI deficiency do not experience spontaneous bleeding, and data suggest they have lower rates of cardiovascular events, including SSE. Unlike factor IX and factor VIII, FXIa has only a minor effect on clot consolidation during hemostasis. Therefore, FXIa inhibition might offer an opportunity to prevent SSE, without increasing bleeding risk when compared with DOAC therapy. This finding is theoretically promising, particularly for the older adults and frail patients.

Phase II dose-finding trials have been published for various factor Xia inhibitors. The PACIFIC-AF phase II randomized controlled trial compared asundexian 20 mg versus 50 mg versus apixaban 5 mg twice daily. Asundexian demonstrated lower rates of bleeding compared with apixaban. However, the study’s short follow-up period of 12 weeks limits the ability to fully assess both bleeding and thrombotic events. The OCEANIC-AF phase III study investigating asundexian was stopped prematurely due to lack of efficacy [69]. Although it is theoretically promising, in practice it presents significant challenges.

Besides asundexian, other factor XIa inhibitors such as milvexian and the parenteral abelacimab, fesomersen, osocimab and xisomab 3G3 are under investigation. Once monthly subcutaneous administration of abelacimab was found to be safe and well-tolerated in patients with AF [70]. Comparative studies with OACs to evaluate the efficacy and safety of XIa inhibitors relative to DOACs are not available, let alone studies involving older and frails patients.

Left Atrial Appendage Closure

Percutaneous left atrial appendage (LAA) closure is a potential alternative strategy to long-term anticoagulation for patients with AF. However, this procedure requires a period of at least 6 months of antiplatelet or anticoagulant treatment postdeployment, which carries a risk of recurrent bleeding. In the long term, LAA closure may be safer than OACs in terms of risk of bleeding but is associated with higher short-term complications, such as cardiac tamponade and device embolization (RR 1.69, 95% CI 1.01–3.19) [71, 72].

In patients with a high risk of bleeding or stroke, LAA closure was found to be as effective as DOACs for preventing SSE (HR 1.14, 95% CI 0.56–2.30) and safety concerning clinically relevant bleeding (HR 0.75, 95% CI 0.44–1.27) [73, 74]. However, the confidence intervals of these hazard ratios indicate significant uncertainty. Additionally, recent evidence suggests a high incidence of nonpatency following LAA closure (49–54%) [75, 76], which is believed to increase rather than decrease the risk of ischemic events [77].

The COMPARE LAAO trial aims to determine if LAA closure is superior in preventing SSE and to evaluate the cost effectiveness of this procedure in patients with AF with a high thromboembolic risk and contraindications for anticoagulation [78].

For the very frail patients with AF who are ineligible for OAC, the suitability of LAA closure as an invasive percutaneous option is questionable. The ESC guidelines recommend that patients with nonvalvular AF and previous ischemic stroke or TIA who also have a high risk of bleeding or other contraindications to OAC should be included in a RCT, if possible [34].

Conclusions

DOACs are the recommended oral anticoagulant for vital older patients with AF. The available information on frail individuals, those with renal impairments, or significant bleeding histories suggests the benefit-risk profile favors DOACs over VKAs in most of these subgroups. Therefore, we conclude there is no reason to assume these patients should be treated differently than patients without these conditions. We also conclude that these patients should be administered on-label dosages of DOACs. Off-label low dosages may be considered only in selected cases, and low dosages of DOAC are—in most cases—preferable to no anticoagulant therapy at all. Lastly, patients who are treated with a VKA with stable INR values are not advised to switch to DOAC therapy. New therapies are under investigation; however, the optimal treatment for this complex patient population has yet to be identified.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Funding

This research received no specific grant from any funding agency in the public or commercial or not-for-profit sectors.

Competing interests

J.R.S. has no competing interests. T.A.C.d.V. reports nonfinancial support from Daiichi Sankyo and personal fees from Bristol-Myers-Squibb, both outside the submitted work. He also reports that he is a member of the adjudication committee of the LIMIT & DANCE trails, which are sponsored by the Population Health Research Institute. M.E.W.H. received consultancy and/or speaker fees from Bayer Netherlands, Daiichi Sankyo, Boehringer Ingelheim, and BMS-Pfizer. Received an unrestricted research grant from the Dutch Federation of Anticoagulation Clinics. R.P. has no competing interests. J.R.d.G. received research grants (through institution) from Atricure Inc, Bayer, Boston, Scientific, Daiichi Sankyo, Johnson&Johnson, Medtronic. Received honoraria/speaker/consultancy fees from AtriaN Medical, Atricure Inc, Bayer, Berlin Chemie, Daiichi Sankyo, Menarini, Medtronic, Novartis, Servier. R.W.M.M. Jansen received honoraria for speaker fees from Bayer Netherlands, Daiichi Sankyo, Boehringer Ingelheim, and BMS-Pfizer.

Author’s contributions

J.S., T.d.V., M.H., R.P., J.d.G., and R.J. were involved in drafting and revising the manuscript for intellectual content. All authors gave final approval for this manuscript to be published.

Data availability

Data sharing not applicable to this article as no datasets were generated for this review article.

Ethics approval

Not applicable.

Code availability

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.