Introduction: Non-adherence to recommended secondary preventive anticoagulation in stroke patients with atrial fibrillation (AF) is a common phenomenon, although the introduction of direct oral anticoagulants (DOACs) has simplified anticoagulation management for physicians as well as for patients. Methods: We examined the adherence of secondary preventive anticoagulation in AF patients after re-integration in their social environment 6–12 weeks after stroke unit and rehabilitation clinic treatment and analyzed for predictors for adherence and non-adherence. We conducted a telephone survey in consecutive patients treated between January 2013 and December 2021 at our institutional stroke unit with an acute cerebrovascular ischemic event, and we analyzed discharge letters of rehabilitation clinics of those patients not anticoagulated at follow-up. All patients had known or newly diagnosed AF, and in all, we had recommended secondary preventive anticoagulation. Results: Follow-up information about anticoagulant intake could be obtained in 1,348 of 1,685 patients (80.0%) treated within the study period. Anticoagulation rate was 91.5%, with 83.6% of patients receiving DOACs and 7.9% receiving vitamin K antagonists (VKAs). Adherence to recommended anticoagulation was associated with intake of the recommended anticoagulant already at discharge (adjusted odds ratio [OR], 18.357; confidence interval [CI], 9.637–34.969), recommendation of a specific DOAC and dose (in contrast to “DOAC” as drug category) (adjusted OR, 2.971; CI, 1.173–7.255), a lower modified Rankin Scale at discharge (per point; adjusted OR, 0.813; CI, 0.663–0.996), younger age (per year; OR, 0.951; CI, 0.926–0.976), and the absence of peripheral vascular disease (adjusted OR, 0.359; CI, 0.173–0.746). In patients already anticoagulated at discharge, adherence was 98.5%, irrespective of a patient’s age, functional deficit at discharge, and peripheral vascular disease. Avoidable obstacles for non-adherence in patients not on anticoagulants at stroke unit discharge were (1) non-implementation of recommended anticoagulation by rehabilitation physicians predominantly in patients with moderate-severe or severe stroke disability (2.1%), (2) delegation of anticoagulation start from rehabilitation physicians to general practitioners/resident radiologists (1.3%), and (3) rejection of recommended anticoagulation because of patients’ severe stroke disability (0.5%). Non-avoidable obstacles were contraindications to anticoagulation (2.1%) and patients’ refusal (0.7%). Conclusions: Commencing drug administration already during stroke unit hospitalization and providing an explanation for the selection of the recommended anticoagulant in discharge letters ensures high adherence at patients’ re-integration in their social environment after acute stroke treatment. If drug administration cannot be commenced before discharge, education of rehabilitation physicians by stroke physicians and the involvement of stroke physicians into the post-stroke decision process might hinder avoidable obstacles.

Direct oral anticoagulants (DOACs) are able to prevent three-quarters of all ischemic strokes caused by atrial fibrillation (AF) [1‒5]. Their introduction in Germany in 2011 had led to an estimated increase in the proportion of anticoagulated high-risk patients with AF (CHA2DS2-VASc score ≥2) from 42% to 63% in 2016 [6], whereas patients with a history of stroke were more likely to receive anticoagulation than those without. However, even before DOACs were generally established, 61–77% of AF patients who had suffered from an ischemic stroke were found to be anticoagulated [7, 8]. Today, those anticoagulation rates would be considered disappointing as stroke patients are treated almost exclusively at stroke units, where anticoagulation is recommended for up to 94% of AF patients [9].

We asked whether high recommendation rates in stroke unit patients with AF, e.g., 96% at our institution in 2018 [10], have led to a comparable increase in patients’ anticoagulation rates after discharge, and we searched for predictors for adherence and non-adherence. To address this, we conducted a telephone survey in a large cohort of consecutive stroke patients with known or newly diagnosed AF and with recommended anticoagulation at discharge, and we analyzed discharge letters from rehabilitation clinics documented in our medical records of those patients not anticoagulated at follow-up.

Patient Selection

Consecutive patients treated at the institutional stroke unit, Department of Neurology, Otto von Guericke University, Magdeburg, between January 2013 and December 2021 with an acute cerebrovascular ischemic event who had known or newly diagnosed AF and who had been recommended secondary preventive anticoagulation, were selected for the present study. The available data set included variables from the institutional stroke registry as listed in Table 1 and the type and dose of the recommended or already implemented anticoagulant at discharge.

Table 1.

Demographics, baseline characteristics, and clinical findings in 1,685 AF patients discharged from hospital with recommended anticoagulation after an acute cerebrovascular ischemic event and in the sub-cohort of 1,348 patients in whom follow-up data on anticoagulation could be obtained

N (%)N (%)
Cohort 1,685 (100) 1,348 (100) 
Age, years, mean (SD) 77.3 (9.5) 76.6 (9.5) 
Sex, male, n (%) 815 (48.4) 680 (50.4) 
Hypertension, n (%) 1,599 (94.9) 1,276 (94.7) 
Diabetes, n (%) 694 (41.2) 542 (40.2) 
Hypercholesterolemia, n (%) 1,149 (68.2) 941 (69.8) 
Body mass index, kg/m2, mean (SD) 27.9 (5.5) 28.1 (5.5) 
History of myocardial infarction, n (%) 402 (23.9) 331 (24.6) 
Congestive heart failure, n (%) 207 (12.3) 160 (11.9) 
Peripheral vascular disease, n (%) 139 (8.2) 102 (7.6) 
History of cerebral ischemia, n (%) 338 (20.1) 269 (20.0) 
CHA2DS2VASc score at discharge, median (IQR) 6 (5–7) 6 (5–7) 
Modified RS at discharge, median (IQR) 3 (1–4) 2 (1–4) 
Discharged 
 Home, n (%) 576 (34.2) 517 (38.4) 
 To other hospital (rehab/regional hospital), n (%) 998 (59.2) 775 (57.5) 
 To nursing home, n (%) 111 (6.6) 56 (4.2) 
Recommended anticoagulant taken at discharge, n (%) 1,054 (62.6) 904 (67.1) 
Recommended anticoagulant, n (%) 
 Apixaban 837 (49.7) 649 (48.1) 
 Dabigatran 547 (32.5) 474 (35.2) 
 Edoxaban 107 (6.4) 99 (7.3) 
 Rivaroxaban 23 (1.4) 12 (0.9) 
 DOAC (drug category) 38 (2.3) 24 (1.8) 
 VKA 133 (7.9) 90 (6.7) 
Year of stroke unit treatment, n (%) 
 2013–2015 509 (30.2) 383 (28.4) 
 2016–2018 561 (33.3) 400 (29.7) 
 2019–2021 615 (36.5) 565 (41.9) 
N (%)N (%)
Cohort 1,685 (100) 1,348 (100) 
Age, years, mean (SD) 77.3 (9.5) 76.6 (9.5) 
Sex, male, n (%) 815 (48.4) 680 (50.4) 
Hypertension, n (%) 1,599 (94.9) 1,276 (94.7) 
Diabetes, n (%) 694 (41.2) 542 (40.2) 
Hypercholesterolemia, n (%) 1,149 (68.2) 941 (69.8) 
Body mass index, kg/m2, mean (SD) 27.9 (5.5) 28.1 (5.5) 
History of myocardial infarction, n (%) 402 (23.9) 331 (24.6) 
Congestive heart failure, n (%) 207 (12.3) 160 (11.9) 
Peripheral vascular disease, n (%) 139 (8.2) 102 (7.6) 
History of cerebral ischemia, n (%) 338 (20.1) 269 (20.0) 
CHA2DS2VASc score at discharge, median (IQR) 6 (5–7) 6 (5–7) 
Modified RS at discharge, median (IQR) 3 (1–4) 2 (1–4) 
Discharged 
 Home, n (%) 576 (34.2) 517 (38.4) 
 To other hospital (rehab/regional hospital), n (%) 998 (59.2) 775 (57.5) 
 To nursing home, n (%) 111 (6.6) 56 (4.2) 
Recommended anticoagulant taken at discharge, n (%) 1,054 (62.6) 904 (67.1) 
Recommended anticoagulant, n (%) 
 Apixaban 837 (49.7) 649 (48.1) 
 Dabigatran 547 (32.5) 474 (35.2) 
 Edoxaban 107 (6.4) 99 (7.3) 
 Rivaroxaban 23 (1.4) 12 (0.9) 
 DOAC (drug category) 38 (2.3) 24 (1.8) 
 VKA 133 (7.9) 90 (6.7) 
Year of stroke unit treatment, n (%) 
 2013–2015 509 (30.2) 383 (28.4) 
 2016–2018 561 (33.3) 400 (29.7) 
 2019–2021 615 (36.5) 565 (41.9) 

AF, atrial fibrillation; DOAC, direct oral anticoagulant; SD, standard deviation; IQR, interquartile range; RS, Rankin Scale; VKA, vitamin K antagonist.

Data Collection and Study Objectives

In patients who had been treated in our stroke unit and had given informed consent, we conducted a routine follow-up telephone survey 3–6 months after the index stroke using contact data documented in the inpatient medical record. Objectives of our routine survey were patients’ functional stroke outcome, the recurrence of a cerebrovascular or cardiovascular event since stroke unit discharge, the course of an antithrombotic treatment since discharge, and, if applicable, the reason and the person responsible for non-adherence to a recommended and/or commenced antithrombotic drug. During hospitalization, stroke unit physicians informed the patient or his/her legal representative about the objective of our survey, the voluntary nature of participation, the possibility of withdrawing consent to participate at any time, and the anonymized data analysis and data publication. If consent was given, documented in the medical record, and corroborated at the telephone call, follow-up information was gathered and documented in the patients’ medical records (A.G., T.F.). All diagnostic and therapeutic procedures were performed within standard clinical practice, and collected data were entered into our institution’s vascular and stroke registry in accordance with federal and national data protection laws so that the responsible ethics committee of the medical faculty of the university waived the need for ethics committee approval. In addition, for those patients in whom anticoagulation recommended in our discharge letter was not administered, we checked discharge letters of rehabilitation clinics documented in our inpatients’ medical records for the reason and the person responsible for non-adherence, e.g., general practitioner, resident neurologist, or rehab hospital physician (T.F.). The responsible ethics committee had approved the latter part of the study. The adherence to recommended anticoagulation and anticoagulant 6–12 weeks after discharge from our hospital, i.e., when patients had been re-integrated in their social environment, was the primary study objective.

Statistics

Statistical analysis was performed using the SPSS 28.0 software package. A selection bias of patients within the follow-up collective was investigated by conducting χ2 tests and binary logistic regression analysis. We used backward multivariate logistic regression analysis to investigate influence and independence of variables on anticoagulant adherence. Patients’ characteristics (sex, age) and discharge-related variables, i.e., destination of discharge, type of anticoagulant recommended at discharge, and anticoagulant intake/non-intake at discharge, were put into the first block. The year of hospitalization also was added to the first block to allow for compensation of time trends within the observation period. The independence of discharge-related variables from confounding factors as listed in Table 1 was estimated by backward stepwise removal (p > 0.10) and entry selection (p < 0.05) of confounding variables at the second block based on their likelihood ratios. The Kruskal-Wallis test with Bonferroni correction was used to investigate the relation between reasons for non-adherence and severity of patients’ disability. Significance was set at a p value <0.05.

From January 2013 to December 2021, 1,685 consecutive patients with known or newly diagnosed AF suffering from an acute ischemic event were discharged with a recommendation for secondary preventive anticoagulation. Follow-up information about anticoagulant intake 6–12 weeks after discharge could be obtained in 1,348 patients (80.0%) (shown in Fig. 1). Table 1 presents the demographics, baseline characteristics, and discharge variables of the patient cohort and the follow-up sample. Patients who could be followed-up were younger (per year; adjusted odds ratio [OR] 0.962; confidence interval [CI] 0.947–0.978), were discharged home (adjusted OR, 2.928; CI, 1.692–5,068) or to another hospital (adjusted OR, 3.491; CI, 2.212–5.511) in contrast to a nursing home, took recommended anticoagulant already at discharge (adjusted OR, 1.452; CI, 1.081–1.952), and had less functional deficit on modified Rankin Scale at discharge (per point; adjusted OR, 0.645; CI, 0.573 to 0.727). Those associations were independent of the year of hospitalization.

Fig. 1.

Patients’ selection. AF, atrial fibrillation; AC, anticoagulation; CV, cerebrovascular; FU, follow-up.

Fig. 1.

Patients’ selection. AF, atrial fibrillation; AC, anticoagulation; CV, cerebrovascular; FU, follow-up.

Close modal

One thousand two hundred and thirty-three of 1,348 AF patients (91.5%) for whom we could acquire follow-up information were on secondary preventive anticoagulation 6–12 weeks after their discharge from the stroke unit, with 1,127 patients (83.6%) being on DOACs and 105 (7.9%) on vitamin K antagonists (VKAs). Adherence to anticoagulation was associated with intake of the recommended anticoagulant already at discharge, recommendation of a specific DOAC and dose (in contrast to “DOAC” as drug category), a lower modified Rankin Scale at discharge, younger age, and the absence of peripheral vascular disease. Adherence was not associated with the year of hospitalization within the whole observation period (Table 2). However, adherence ranged between 89.5% and 91.6% (mean, 90.5%) from 2013 to 2020 without significant trend and increased to 97.7% in 2021 (p < 0.001). Commencement of recommended anticoagulation already during stroke unit hospitalization (65.4% in 2013–2020 vs. 77.8% in 2021) was the only variable associated with that increase (p < 0.001). All but 1 patient was recommended a specific anticoagulant drug and dose in 2021.

Table 2.

Multivariate binary logistic regression identifying variables associated with adherence to recommended anticoagulation 6–12 weeks after stroke unit discharge in 1,348 patients with AF

Adjusted OR95% CIp value
Age, per year 0.951 0.926–0.976 <0.001 
Sex, male 1.018 0.647–1.601 0.939 
Discharged home (vs. nursing home) 2.066 0.623–6.852 0.236 
Discharged to other hospital (vs. nursing home) 2.681 0.953–7.544 0.062 
Recommended anticoagulant taken at discharge 18.357 9.637–34.969 <0.001 
Recommendation of specific DOAC/dose (vs. DOAC as category) 2.917 1.173–7.255 0.021 
Recommendation of VKA INR 2–3 (vs. DOAC as category) 1.389 0.401–4.805 0.604 
Modified RS at discharge, per point 0.813 0.663–0.996 0.046 
Peripheral vascular disease 0.359 0.173–0.746 0.006 
Year of stroke unit hospitalization, per year increase 1.030 0.944–1.123 0.507 
Adjusted OR95% CIp value
Age, per year 0.951 0.926–0.976 <0.001 
Sex, male 1.018 0.647–1.601 0.939 
Discharged home (vs. nursing home) 2.066 0.623–6.852 0.236 
Discharged to other hospital (vs. nursing home) 2.681 0.953–7.544 0.062 
Recommended anticoagulant taken at discharge 18.357 9.637–34.969 <0.001 
Recommendation of specific DOAC/dose (vs. DOAC as category) 2.917 1.173–7.255 0.021 
Recommendation of VKA INR 2–3 (vs. DOAC as category) 1.389 0.401–4.805 0.604 
Modified RS at discharge, per point 0.813 0.663–0.996 0.046 
Peripheral vascular disease 0.359 0.173–0.746 0.006 
Year of stroke unit hospitalization, per year increase 1.030 0.944–1.123 0.507 

Anticoagulation and discharge related variables (= investigated variables) were forced into the analysis as well as patients’ characteristics and the year of hospitalization to investigate independency among each other.

Other variables were selected by backward stepwise removal and entry statistics according to their significance.

AF, atrial fibrillation; CI, confidence interval; DOAC, direct oral anticoagulant; INR, international normalized ratio; OR, odds ratio; RS, Rankin Scale; VKA, vitamin K antagonist.

Eight-hundred-ninety of 904 patients (98.5%) taking recommended anticoagulants already at discharge were on anticoagulation at follow-up in contrast to only 343 of 444 patients (77.3%) without anticoagulation at discharge. In 14 patients (1.5%) without anticoagulation at follow-up, although discharged on an anticoagulant, medication had been paused because of an acute cerebrovascular (4) or cardiovascular (1) event and had been stopped because of bleeding complications (5; epistaxis twice, hemoptysis, ecchymosis, gastrointestinal), patient refusal (2), and unknown reasons (2). One-thousand-two-hundred-and-nineteen of 1,320 patients (92.1%) with recommendation of a specific anticoagulant drug and dose in the discharge letter took anticoagulants at follow-up, in contrast to only 14 of 24 patients (58.3%) with recommendation of “DOAC” as drug category. The percentage of patients on anticoagulants decreased with increasing modified RS at discharge (shown in Fig. 2). Eighty-seven of 102 patients (85.3%) with peripheral arterial disease in contrast to 1,146 of 1,246 (92.0%) without were on anticoagulants at follow-up.

Fig. 2.

Anticoagulation at follow-up 6–12 weeks after discharge related to modified Rankin Scale at discharge in 1,348 stroke patients with atrial fibrillation and recommended anticoagulation (percentage and numbers). Column colors indicate anticoagulation status at discharge and at follow-up: black, AC at discharge and FU; gray, AC not yet at discharge but at FU; red, AC neither at discharge nor FU; blue, AC at discharge but no longer at FU. AC, anticoagulation; FU, follow-up.

Fig. 2.

Anticoagulation at follow-up 6–12 weeks after discharge related to modified Rankin Scale at discharge in 1,348 stroke patients with atrial fibrillation and recommended anticoagulation (percentage and numbers). Column colors indicate anticoagulation status at discharge and at follow-up: black, AC at discharge and FU; gray, AC not yet at discharge but at FU; red, AC neither at discharge nor FU; blue, AC at discharge but no longer at FU. AC, anticoagulation; FU, follow-up.

Close modal

Anticoagulation was withheld at stroke unit discharge in 444 of the 1,348 follow-up patients (32.9%) because of secondary hemorrhage into a brain infarct (42 cases; 3.1%), suspected increased risk of secondary hemorrhage into a (large) brain infarct (352; 25.4%), contraindication due to (transient) comorbidity/surgical intervention (40; 3.0%), patient’s/legal representative’s denial (7; 0.5%), and one case without an obvious cause. Reasons for non-adherence in those 101 patients (7.5%) not yet on anticoagulants at follow-up are presented in Table 3. Severe comorbidity independent of stroke disability (21 patients) and bleeding complications (7) contraindicated implementation in 28 patients. Anticoagulation was delegated from rehabilitation physicians to general practitioners and/or radiologists for organizing CT control (13 patients) or reasons not mentioned in discharge letters (4) and was yet not implemented 12 weeks after stroke unit discharge. Anticoagulation was rejected in 7 cases by rehabilitation physicians because of severe or moderate-severe disability according to discharge letters and by the patient or his/her legal representative in additional 9 cases. In 28 cases, recommended anticoagulation was not implemented by rehabilitation physicians without an explanation or discussion in the discharge letter, and patients were discharged from the rehabilitation clinic without/without recommendation for anticoagulants. Patients in whom recommendation for anticoagulation was not implemented were significantly more severely disabled than those with a contraindication for an anticoagulation, which also holds true for those in whom anticoagulation was rejected by a physician (p < 0.001; Table 3).

Table 3.

Reasons for non-adherence to recommended anticoagulation in 101 patients not yet anticoagulated 6–12 weeks after stroke unit discharge; information based on discharge letters of rehabilitation clinics (70 cases) and patients’ answers at follow-up telephone calls (89)

Reasons for non-adherence to AC at follow-upDisability at follow-up
mRS 0–3mRS 4mRS 5mRS, Md (IQR)
Patients, n (%) 101 (7.5) 40 (3.0) 33 (2.4) 28 (2.1)  
Contraindicated by physician 28 (2.1) 15 11 3 (2–4) 
Delegated to other physicians 17 (1.3) 10 3 (1–4) 
Rejected by patient/legal representative 9 (0.7) 2 (2–4) 
Rejected by physician 7 (0.5)  5 (4–5) 
Not implemented by physician 28 (2.1) 10 15 5 (4–5) 
No data available 12 (0.9) 3 (1–4) 
Reasons for non-adherence to AC at follow-upDisability at follow-up
mRS 0–3mRS 4mRS 5mRS, Md (IQR)
Patients, n (%) 101 (7.5) 40 (3.0) 33 (2.4) 28 (2.1)  
Contraindicated by physician 28 (2.1) 15 11 3 (2–4) 
Delegated to other physicians 17 (1.3) 10 3 (1–4) 
Rejected by patient/legal representative 9 (0.7) 2 (2–4) 
Rejected by physician 7 (0.5)  5 (4–5) 
Not implemented by physician 28 (2.1) 10 15 5 (4–5) 
No data available 12 (0.9) 3 (1–4) 

Percentage figures refer to the follow-up cohort of 1,348 patients.

AC, anticoagulation; Md, median; mRS, modified Rankin Scale; IQR, interquartile range.

The adherence to recommended dabigatran (73.4%; 348 of 474 patients) was significantly lower than that to apixaban (83.1%; 539 of 649) and edoxaban (84.8%; 84 of 99) (p < 0.001). Seventy-seven of 474 patients (16.2%) who were recommended dabigatran treatment received another DOAC compared to only 42 of 649 patients (6.5%) and 8 of 99 patients (8.1%) who were recommended apixaban and edoxaban treatment, respectively (p < 0.001). In contrast, non-adherence to dabigatran in favor of VKA anticoagulation/no anticoagulation (49 patients, 10.3%) was similar to that for apixaban (68, 10.5%) and edoxaban (7, 7.1%).

Between 2013 and 2021, recommendation of anticoagulation by stroke unit physicians in patients with AF resulted in an anticoagulation rate of 91.5% 6–12 weeks after stroke unit discharge, i.e., when patients were re-integrated in their social environment. Of those, 83.6% received DOACs and 7.9% VKA. The implementation of the anticoagulation already at discharge and a detailed explanation of the recommended anticoagulant drug and its dose in the discharge letters instead of a general recommendation for DOACs were major reasons for drug adherence. Former was particularly evident in the significant increase in adherence to 97.7% in 2021 as compared to the entire period from 2013 to 2020, which was associated with a substantial percentage increase in patients discharged already on the recommended anticoagulant. The dominant obstacle hindering the start of recommended anticoagulation already during stroke unit hospitalization was a suspected higher risk of secondary hemorrhage into a (large) brain infarct. However, after initial investigations suggested that anticoagulation with DOACs only several days after a cerebrovascular ischemic event might not be associated with a substantial risk increase of cerebral bleeding complication [11, 12], and after we have presented preliminary results of the present investigation at our once-quarterly internal stroke update education, we started to implement earlier anticoagulation in our clinical practice. An already-administered anticoagulation at discharge was terminated or temporarily paused only in a few patients and had, in those we could determine, reasonable grounds.

Nevertheless, in patients with a very large infarct or in those with secondary hemorrhage into a brain infarct, the initiation of anticoagulation may be possible only during rehab hospitalization. Further forward delegation of this decision (and thus the responsibility for a potential iatrogenic cerebral hemorrhage) from the rehabilitation physician to a general practitioner/resident radiologist hindered the implementation of a recommended anticoagulation in 1.3% of our patients. We cannot give reasons for this by means of our investigation. However, ensuring that CT control can be performed and assessed during rehab hospitalization, and thus by a physician familiar with acute/subacute stroke patients, may enable adequate anticoagulation to start.

Reluctance to the commencement of recommended anticoagulation was observed in severely and moderate-severely disabled patients discharged to a nursing home from our stroke unit or via a rehabilitation hospital. The indication and recommendation for anticoagulation were rejected because of an unfavorable benefit-risk ratio in 0.5% of patients and were not implemented without giving reasons in 2.1% of rehabilitation physicians or general practitioners. A palliative care as a rationale can be excluded; neither did those patients get a recommendation for anticoagulation at stroke unit discharge nor were therapeutic drugs withheld at rehabilitation clinic discharge. One may speculate that physicians in those severely disabled but yet not palliative-cared patients also did not assume a benefit of anticoagulation when considering the risk of bleeding complications [13]. However, this would ignore the low risk of major and intracranial bleeding under DOAC which was similar, for example, for apixaban to that of aspirin [14]. Remarkably, both delegation and reluctance to the commencement of recommended anticoagulation no longer occurred in 2021, which – even if not quantifiable – might have been triggered by the presentation of our preliminary results to our rehab clinic colleagues at our institutional once-a-year stroke update education event.

During the study period, apixaban (49.7% of recommendations) and dabigatran (32.5%) were the predominantly recommended DOACs (Table 1). Probable reasons for this may be the later market introduction of edoxaban and the higher intracranial bleeding risk of rivaroxaban compared to apixaban and dabigatran as demonstrated in the underlying pivotal trials [2, 4, 5]. Non-adherence to apixaban (10.5%), dabigatran (10.7%), and edoxaban (7.7%) because of VKA administration or non-anticoagulation did not differ. In contrast, dabigatran (16.2%) was significantly more often replaced by another DOAC than apixaban (6.5%) and edoxaban (8.1%). A decreasing renal function in the post-discharge period might be an explanation for this finding because dabigatran is predominantly excreted through the kidneys, in contrast to all other DOACs which are predominantly excreted via the liver [15]. A higher gastrointestinal bleeding risk of dabigatran compared to apixaban might be another reason [16, 17]. Familiarity of physicians with distinct DOACs or the only once-daily administration of rivaroxaban in contrast to all other DOACs – the latter more important in outpatient than inpatient treatment - might be further reasons but cannot be proven by our investigation.

Our study has limitations. A recall bias may not be excluded in all patients, although the interviewer knew the patient’s recommended and discharge medication, functional disability at discharge, and where the patient was discharged during the telephone call and was able to check the reliability of patient’s answers. The date we asked for adherence to anticoagulation – 6–12 weeks after stroke unit discharge – aimed to investigate the adherence at the intersection between acute stroke treatment in stroke unit/subsequent rehabilitation clinic and long-term secondary prevention provided by general practitioners and differed from those of other studies, ranging from 12 to 15 months after the ischemic event [7, 8]. Adherence rates in latter studies are additionally determined by a later termination of anticoagulation and thus have to be expected to be lower than ours. However, to reduce non-adherence, it seems rational to search for reasons for intersection obstacles as well as for later termination which may be different. Conclusions on specific DOACs may not be generalized because reliable data were obtained only for apixaban and dabigatran, both predominantly recommended during the study period at our institution. Our follow-up cohort showed a bias compared to the overall study collective. Most of the differing variables and differences in the follow-up cohort – younger age, discharge to home/rehabilitation clinic rather than nursing facility, mild poststroke deficit – are related to a longer survival and thus a higher probability to obtain follow-up information. However, because those variables were also associated with higher adherence, our results may underestimate the real rate of non-adherence and raise the importance of our suggestions to improve adherence.

The administration of the drug even before discharge from the stroke unit and the explanation for the selection of the recommended anticoagulant in the discharge letters of our stroke unit led to a high adherence to recommended anticoagulation after patients’ reintegration in their social environment. Thus, recent evidence for starting anticoagulation during acute stroke hospitalization without a substantial increase in cerebral bleeding complications [18] may increase adherence considerably. In those without commenced anticoagulation during stroke unit hospitalization, e.g., after a secondary parenchymal hemorrhage or in the presence of a large infarct (often with subsequent severe disability), continuous education of rehabilitation physicians by stroke physicians and/or the involvement of stroke physicians in the post-stroke decision process might encourage former to start secondary preventive anticoagulation before the patient is dismissed in his/her social environment and additionally may increase adherence. However, further investigations are needed to address non-adherence due to later secondary termination of an already-implemented anticoagulation and its underlying reasons.

The authors thank Annette Glänzel [A.G.] for stroke register documentation and telephone calls.

Diagnostic and therapeutic procedures all performed within standard clinical practice, data collection and pseudonymized documentation in our institution’s vascular and stroke registry and routine telephone survey in stroke unit patients by hospital staff is in accordance with good clinical practice and federal and national data protection laws, so that the Ethics Committee of the Otto von Guericke University Magdeburg waived the need for ethics committee approval. Data inspection/handling and telephone survey by the doctoral student (T.F.) need approval according to federal data protection laws and were approved by the Ethical Committee of the Otto von Guericke University Magdeburg (Registration number 158/19). During hospitalization, stroke unit physicians informed the patient and, in case of a vulnerable patient, his/her legal representative about the objective of our survey, the voluntary nature of participation, the possibility of withdrawing consent to participate at any time, and the anonymized data analysis and data publication. If written informed consent to participate was obtained and corroborated at the telephone call, follow-up information was gathered. If written informed consent could not be obtained during hospitalization, patient was informed at the telephone call, and consent was inferred by completion of the questionnaire/participation in the interview. The ways of getting informed consent were approved by the Ethics Committee.

The authors have no conflicts of interest to declare.

No funding relevant to the study was received.

M.G. planned and initiated the study and consolidated the register and discharge data. T.F. performed follow-up telephone interviews and collected follow-up data. T.F. and M.G. performed statistical analyses and wrote the first draft of the paper. A.A., J.N., S.S., A.H., M.B., and M.S. contributed ideas and edited the final draft. All authors read and approved the final manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.

1.
Secondary prevention in non-rheumatic atrial fibrillation after transient ischaemic attack or minor stroke. EAFT (European Atrial Fibrillation Trial) Study Group
.
Lancet
.
1993
;
342
(
8882
):
1255
62
. .
2.
Connolly
SJ
,
Ezekowitz
MD
,
Yusuf
S
,
Eikelboom
J
,
Oldgren
J
,
Parekh
A
, et al
.
Dabigatran versus warfarin in patients with atrial fibrillation
.
N Engl J Med
.
2009
;
361
(
12
):
1139
51
. .
3.
Giugliano
RP
,
Ruff
CT
,
Braunwald
E
,
Murphy
SA
,
Wiviott
SD
,
Halperin
JL
, et al
.
Edoxaban versus warfarin in patients with atrial fibrillation
.
N Engl J Med
.
2013
;
369
(
22
):
2093
104
. .
4.
Granger
CB
,
Alexander
JH
,
McMurray
JJ
,
Lopes
RD
,
Hylek
EM
,
Hanna
M
, et al
.
Apixaban versus warfarin in patients with atrial fibrillation
.
N Engl J Med
.
2011
;
365
(
11
):
981
92
. .
5.
Patel
MR
,
Mahaffey
KW
,
Garg
J
,
Pan
G
,
Singer
DE
,
Hacke
W
, et al
.
Rivaroxaban versus warfarin in nonvalvular atrial fibrillation
.
N Engl J Med
.
2011
;
365
(
10
):
883
91
. .
6.
Hohnloser
SH
,
Basic
E
,
Nabauer
M
.
Uptake in antithrombotic treatment and its association with stroke incidence in atrial fibrillation: insights from a large German claims database
.
Clin Res Cardiol
.
2019
;
108
(
9
):
1042
52
. .
7.
Gumbinger
C
,
Holstein
T
,
Stock
C
,
Rizos
T
,
Horstmann
S
,
Veltkamp
R
.
Reasons underlying non-adherence to and discontinuation of anticoagulation in secondary stroke prevention among patients with atrial fibrillation
.
Eur Neurol
.
2015
;
73
(
3–4
):
184
91
. .
8.
Luger
S
,
Hohmann
C
,
Niemann
D
,
Kraft
P
,
Gunreben
I
,
Neumann-Haefelin
T
, et al
.
Adherence to oral anticoagulant therapy in secondary stroke prevention: impact of the novel oral anticoagulants
.
Patient Prefer Adherence
.
2015
;
9
:
1695
705
. .
9.
Wutzler
A
,
Krogias
C
,
Grau
A
,
Veltkamp
R
,
Heuschmann
PU
,
Haeusler
KG
.
Stroke prevention in patients with acute ischemic stroke and atrial fibrillation in Germany: a cross sectional survey
.
BMC Neurol
.
2019
;
19
(
1
):
25
. .
10.
Gronemann
C
,
Hause
S
,
Assmann
A
,
Neumann
J
,
Schreiber
S
,
Heinze
HJ
, et al
.
Modification of in-hospital recommendation and prescription of anticoagulants for secondary prevention of stroke after launch of direct oral anticoagulants and change of national guidelines
.
Cerebrovasc Dis
.
2020
;
49
(
4
):
412
8
. .
11.
Seiffge
DJ
,
Werring
DJ
,
Paciaroni
M
,
Dawson
J
,
Warach
S
,
Milling
TJ
, et al
.
Timing of anticoagulation after recent ischaemic stroke in patients with atrial fibrillation
.
Lancet Neurol
.
2019
;
18
(
1
):
117
26
. .
12.
Labovitz
AJ
,
Rose
DZ
,
Fradley
MG
,
Meriwether
JN
,
Renati
S
,
Martin
R
, et al
.
Early apixaban use following stroke in patients with atrial fibrillation: results of the AREST trial
.
Stroke
.
2021
;
52
(
4
):
1164
71
. .
13.
Ben Freedman
S
,
Gersh
BJ
,
Lip
GY
.
Misperceptions of aspirin efficacy and safety may perpetuate anticoagulant underutilization in atrial fibrillation
.
Eur Heart J
.
2015
;
36
(
11
):
653
6
. .
14.
Connolly
SJ
,
Eikelboom
J
,
Joyner
C
,
Diener
HC
,
Hart
R
,
Golitsyn
S
, et al
.
Apixaban in patients with atrial fibrillation
.
N Engl J Med
.
2011
;
364
(
9
):
806
17
. .
15.
Lu
Y
,
Branstad
R
,
Karim
RM
,
Asinger
RW
.
Consideration of clinical variables for choosing new anticoagulant alternatives to warfarin for the management of non-valvular atrial fibrillation
.
J Clin Pharm Ther
.
2014
;
39
(
6
):
628
36
. .
16.
Noseworthy
PA
,
Yao
X
,
Abraham
NS
,
Sangaralingham
LR
,
McBane
RD
,
Shah
ND
.
Direct comparison of dabigatran, rivaroxaban, and apixaban for effectiveness and safety in nonvalvular atrial fibrillation
.
Chest
.
2016
;
150
(
6
):
1302
12
. .
17.
Yao
X
,
Abraham
NS
,
Sangaralingham
LR
,
Bellolio
MF
,
McBane
RD
,
Shah
ND
, et al
.
Effectiveness and safety of dabigatran, rivaroxaban, and apixaban versus warfarin in nonvalvular atrial fibrillation
.
J Am Heart Assoc
.
2016
;
5
(
6
):
e003725
. .
18.
Fischer
U
,
Koga
M
,
Strbian
D
,
Branca
M
,
Abend
S
,
Trelle
S
, et al
.
Early versus later anticoagulation for stroke with atrial fibrillation
.
N Engl J Med
.
2023
;
388
(
26
):
2411
21
. .