Type and doses of oral anticoagulants and adherence to anticoagulant treatment in elderly patients with atrial fibrillation: the ESPARTA study

In the ESPARTA study, patients aged ≥75 years old, with a diagnosis of NVAF attended in Internal Medicine units (hospitalization or outpatients) in Spain, treated with oral anticoagulants within the 3 months previous the inclusion, that had started treatment with oral anticoagulants before the inclusion period and that had provided signed informed consent were consecutively included. Patients were excluded if they had mitral stenosis or other significant valvulopathy that required performed or planned specific treatment (i.e., prosthetic valve, valvuloplasty), participated in clinical trials involving the use of DOACs in the 6 months before inclusion, or if they were unable to understand or fulfill the questionnaires of the study. The study was approved by the Clinical Research Ethics Committee of Hospital Universitario de la Princesa de Madrid and endorsed by the local ethic committees of the participating centers [14].

The study was performed in 63 Internal Medicine units throughout Spain, according to clinical practice. The participating sites were representative sample of all such sites in Spain. The inclusion period started on October 2015 and ended on March 2016. The study involved a single visit that coincided with one of the patient’s regular follow-up visits. The biodemographic and clinical data were collected from the medical history during the physician interview and recorded by each investigator in an electronic case report form [14].

The thromboembolic risk was determined by the CHADS₂ and CHA₂DS₂-VASc scales and the risk of bleeding by the HAS-BLED scale [15,16]. The type of oral anticoagulant (VKA vs DOACs) was recorded, as well as the dosage of DOACs: rivaroxaban 20 mg once daily, dabigatran 150 mg twice daily and apixaban 5 mg twice daily were considered as the regular doses of DOACs and rivaroxaban 15 mg once daily, dabigatran 110 mg twice daily and apixaban 2.5 mg twice daily as the low doses of DOACs [12]. The adequacy of anticoagulant treatment was analyzed according to the recommendations performed by the Spanish Medicine and Sanitary Products Agency about the use of oral anticoagulants in NVAF patients (Supplementary Table 2) [13]. In this report, recommendations about the indications/contraindications of anticoagulation, when to use VKA or DOACs or the dosage of DOACs were given [13].

Polymedication was defined as the regular intake of at least five tablets daily at the moment of the visit. Comorbidity was determined according to the Charlson Comorbidity Index. High comorbidity was defined as having a Charlson Comorbidity Index of 3 or more [17]. Cognitive function was assessed using the Pfeiffer’s test, defining the presence of cognitive impairment as having a score of 3 or more [18]. Fragility was determined with the FRAIL scale. The fragile patient was defined as having a score of 3 or more in the FRAIL scale [19]. Physicians asked the patient or their relatives whether patient had partial dependency (need help for at least one of the following activities: personal care, eating, getting dressed or moving), total dependency (need help for all previous activities) or no dependency. Adherence to treatment with oral anticoagulants was determined by using the Morisky–Green test. Good adherence was defined as an average adherence to oral anticoagulation of more than 80% [20].

The study included 854 patients, 17 (2%) of whom were excluded from the final analysis due to improperly recorded or missing data in the electronic case report form. Finally, 837 patients were included for assessment (mean age 83.0 ± 5.0 years old; 48.7% men; mean CHADS₂ score 3.2 ± 1.2; mean CHA₂DS₂-VASc score 5.0 ± 1.4; mean HAS-BLED score 2.1 ± 0.9; mean Charlson Comorbidity Index 2.7 ± 1.9). According to Charlson Comorbidity Index, Pfeiffer’s test and FRAIL scale, comorbidity was high in 48.1% of patients, 32.3% of patients had some degree of cognitive impairment and 43.3% of patients were fragile, respectively. 17.4% of patients lived alone. Overall, 70.8% of patients were taking VKA (mean time on treatment 73.3 ± 59.3 months) and 29.2% DOACs (mean time on treatment 16.6 ± 14.3 months). The clinical characteristics of patients were analyzed according to the type of oral anticoagulant (Table 1). Permanent AF, hypertension and the risk of bleeding were more common in patients taking VKA compared with those patients taking DOACs. There was a trend to have more cognitive impairment in the group of patients taking VKA. In the multivariate analysis, hypertension, permanent AF, high hemorrhagic risk and a higher HAS-BLED score were associated with the use of VKA, whereas prior bleeding and a higher CHADS₂ score were associated with the prescription of DOACs (Table 2).

According to Morisky–Green test, 27.9% of patients had a poor adherence to anticoagulant treatment (27.0% in those patients taking VKA and 29.9% in those patients taking DOACs). Patients with a good adherence to treatment had a higher thromboembolic risk. By contrast, cognitive impairment and fragility were more common in patients with a poor adherence to treatment (Table 3). In the multivariate analysis, having a high CHADS₂ score was associated with a good adherence to treatment, whereas higher scores in the Pfeiffer’s test and FRAIL scale were related with poor adherence to treatment (Table 4).

Overall, 244 patients were taking DOACs. According to the recommendations performed by the Spanish Medicine and Sanitary Products Agency, about the use of oral anticoagulants in NVAF patients, 57.0% of patients were taking DOACs adequately (65.5% of patients treated with rivaroxaban; 56.3% with dabigatran and 46.5% with apixaban; p = 0.029). Among patients taking DOACS, 37.7% were taking the regular doses and 62.3% the lower doses. Compared with the regular doses, the low doses of DOACs were more commonly prescribed in women and polymedicated patients as well as in patients with heart failure, prior cerebrovascular disease and renal insufficiency. In addition, patients taking the low doses of DOACs were older, partial or total dependent, and had a higher thromboembolic and bleeding risk. The scores of the Charlson Comorbidity Index, Pfeiffer’s test and FRAIL scale were higher in the group of patients taking the low doses of DOACs. By contrast, body weight was higher in patients taking the regular doses of DOACs. There was a trend to have more permanent AF in patients taking the low doses of DOACs (Table 5). In the multivariate analysis, having a high CHADS₂ score and being older were independently associated with the use of the low doses of DOACs, whereas a higher body weight and greater glomerular filtration rates were associated with the use of regular doses of DOACs (Table 6).

In our study, a wide sample of anticoagulated elderly patients with AF was analyzed. Despite these patients had a high thromboembolic risk (CHADS₂ score = 3.2; CHA₂DS₂-VASc score = 5.0), less than 30% of patients had a high bleeding risk. This is in the line with current recommendations about the need for anticoagulating the majority of elderly patients with AF [6,21]. On the other hand, although one of the concerns about anticoagulation is the risk of falls, it has been determined that though caution should be taken, the indication of anticoagulation should be independent of the risk of falls [7]. In our study, falls were independent of the type of oral anticoagulant, the adherence to treatment or the dosage of DOACs.

It has recently been reported that among elderly patients with NVAF, compared with warfarin, treatment with DOACs is associated with a lesser risk of stroke and intracranial bleeding [22]. Surprisingly, in our study, VKAs were more commonly prescribed in patients with a high hemorrhagic risk. By contrast, prior bleeding and a high thromboembolic risk were associated with the use of DOACs. This means that in elderly patients with a high risk of bleeding, DOACs are mainly used only when a hemorrhagic episode occurs.

It was quite unexpected that there was a trend to have more cognitive impairment among patients taking VKA compared with DOACs. This issue should be evaluated in-depth in future studies.

Since anticoagulation reduces the risk of stroke in AF patients, assuring a good adherence to oral anticoagulation is mandatory [21,23,24]. In our study, nearly 30% of patients had a poor adherence to treatment. A high thromboembolic risk was associated with good adherence to treatment, whereas cognitive impairment and fragility were related with poor adherence. However, in contrast to previous information [11], polymedication was not associated with medication adherence in our study. It has been reported that complexity of treatment (number of tablets and frequency of administration) and some patients’ clinical characteristics, such as fragility, cognitive impairment, co-morbidities or polymedication, have a relevant impact on medication adherence [11]. However, it should be noted that many adherence researchers remain skeptical of self-reported measures of adherence, especially for accurately identifying patients who are not adherent [25]. As a result, it is likely that the finding that the study population had 30% bad adherence may be a conservative estimate because it is possible that some nonadherent patients were misclassified as adherent patients. Consequently, simplification of treatment through a reduction in the number of tablets, or using drugs with a lesser risk of interactions is mandatory to improve medication adherence. In addition, to improve medication adherence in elderly patients with NVAF may be important to assure a good family and/or community support, particularly in fragile patients or in those patients with cognitive impairment [11].

Another important point that was analyzed was the use of DOACs. In our study, a total of 244 elderly patients with NVAF were taken DOACs. Importantly, 43% of patients did not follow adequately at least one the recommendations performed by the Spanish Medicine and Sanitary Products Agency [13,14]. In addition, the majority of patients (62.3%) received the low doses of DOACs. In the last years, a number of small studies have also analyzed the dosage of DOACs used in clinical practice. Thus, in a study that included 148 patients with AF treated with warfarin (n = 73), rivaroxaban (n = 39) and dabigatran (n = 36), dosage was not appropriate in 77, 23 and 42% of patients, respectively [26]. In other study that included 69 patients, inappropriate use of DOACs was common and more than a quarter of patients were taking a wrong dose of DOACs. Importantly, this led to higher rates of adverse events, including stroke [27]. Although DOACs are administered at fixed doses, dosage of DOACs should be adjusted according to different clinical characteristics. However, these characteristics vary among different DOACs: dosage of rivaroxaban should be adjusted according to renal function; apixaban according to renal function, body weight and age; dabigatran according to renal function, age, risk of bleeding and the use of verapamil; and edoxaban according to renal function, body weight and the use of potent P-gp inhibitors [12]. In our study, the proper use of DOACs was more common with rivaroxaban, likely due to a lesser complexity in the dosage adjustment.

On the other hand, our data showed that a high thromboembolic risk (assessed by CHADS₂ and the CHA₂DS₂-VASc scores) and being older were independently associated with the use of the low doses of DOACs, whereas a higher body weight and greater filtration glomerular rates were associated with the use of regular doses of DOACs. All these data suggest that dose adjustment of DOACs is not properly performed in some patients. It is likely that some physicians may confuse dose adjustment among different DOACs. This is important, since using lower doses of DOACs than recommended could reduce the efficacy of DOACs for the prevention of stroke or systemic embolism in patients with NVAF [12]. Although medical education programs are needed to improve the correct use of DOACs, using those DOACs easier to adjust might reduce the errors performed in this context.

The most important limitation of this study derives from its observational and cross-sectional design that does not allow controlling some potential confounders. However, this is the design that better represents clinical practice. In addition, patients were consecutively included and data were meticulously recorded what may significantly reduce the impact of this potential bias.

In conclusion, among elderly patients with NVAF treated with oral anticoagulants in the Internal Medicine Departments in Spain, VKA were more commonly prescribed in patients with a high hemorrhagic risk and DOACs in patients with prior bleeding or a high thromboembolic risk. Nearly 30% of patients had a poor adherence to anticoagulant treatment. Finally, among patients treated with DOACs, more than 60% were treated with the low doses. A high thromboembolic risk and being older were independently associated with the use of low doses of DOACs, whereas a higher body weight and greater filtration glomerular rates with regular doses. These data strongly suggest that physicians confuse dose adjustment among different DOACs. This occurred to a lesser extent with rivaroxaban.

To view the supplementary data that accompany this paper please visit the journal website at: Supplementary Material

This study was funded by Bayer Hispania. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Writing assistance was utilized in the production of this manuscript. Writing and editorial assistance was provided by Content Ed Net Madrid with financial support from Bayer Hispania.

The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.