Abstract
Aims
Information on utilization of oral anticoagulants (OACs) in nursing homes is scarce. This study aimed to (i) describe OAC use in German nursing home residents, (ii) examine factors influencing whether treatment is initiated with vitamin K antagonists (VKAs) or non‐VKA oral anticoagulants (NOACs) and (iii) assess which conditions predict switching to NOAC instead of continuing VKA.
Methods
Using claims data (2010–2014), we studied a cohort of new nursing home residents aged ≥65 years receiving OAC. Further, OAC use in patients with atrial fibrillation (AF) was examined over the years.
Results
Overall, 16 804 patients (median age: 85 years, 75% female, 44% with renal disease) were included. The majority received phenprocoumon as first OAC (58.0%), followed by rivaroxaban (28.1%). Over the study period, NOAC use increased substantially. Initiating NOAC instead of VKA was predicted by a previous stroke (adjusted odds ratio: 1.76; 95% confidence interval: 1.49–2.08). In contrast, renal disease predicted VKA initiation (0.66; 0.59–0.75) as did the presence of a prosthetic heart valve. Switching from VKA to NOAC was predicted by a stroke (2.55; 2.00–3.24), bleeding events and a recent hospitalization. During 2010–2014, the proportion of AF patients with a CHADS2 score ≥2 receiving OAC increased from 27% to 46%.
Conclusions
NOACs are increasingly used in German nursing homes, both for initial anticoagulation but also in VKA pre‐treated patients. Switching from VKA to NOAC was substantially influenced by aspects such as intended higher effectiveness and safety but probably also practicability due to less blood monitoring.
Keywords: anticoagulants, drug utilization, elderly
What is Already Known about this Subject
In recent years, use of non‐vitamin K antagonist oral anticoagulants (NOACs) has increased substantially.
Information on initiation and switch of oral anticoagulants (OACs) in nursing home residents – a population with a high prevalence of indications requiring anticoagulation but also an elevated risk for bleeding – is scarce.
What this Study Adds
NOACs are increasingly used in the nursing home setting.
Conditions such as stroke or a recent hospitalization influenced OAC choice.
Despite an observed increase over the study years, the overall proportion of nursing home residents with atrial fibrillation receiving OAC treatment was low.
Introduction
Oral anticoagulants (OACs) have been used for more than 60 years to prevent and treat thromboembolic diseases. Traditionally, vitamin K antagonists (VKA) have been the only available drug class. Use of these agents, however, is complicated by a delayed on‐ and offset, a narrow therapeutic margin, the sensitivity for food– and drug–drug interactions, and a high inter‐ and intra‐individual variability necessitating regular monitoring of the anticoagulant effect by the International Normalized Ratio (INR) to prevent over‐ or under‐coagulation 1. In Germany, by far the most frequently prescribed VKA is phenprocoumon, which despite a similar chemical structure displays substantial differences in pharmacokinetics compared to the worldwide most commonly used warfarin 2.
In recent years, non‐VKA oral anticoagulants (NOACs) were introduced. In August 2011, dabigatran etexilate (dabigatran) was the first NOAC to receive approval for the prevention of stroke and systemic embolism in adults with non‐valvular atrial fibrillation (AF), followed by rivaroxaban (December 2011) and apixaban (November 2012). In contrast, rivaroxaban was the first NOAC approved for the treatment of venous thromboembolism (VTE: deep vein thrombosis and pulmonary embolism), and its secondary prevention (December 2011 and November 2012), followed by dabigatran and apixaban (June and July 2014, respectively) 3. As opposed to VKAs, use of NOACs for prevention of systemic embolism in patients with prosthetic heart valves is not recommended or indicated.
NOACs offer convenient treatment with faster on‐ and offset, fixed dosage schemes, no regular laboratory controls and few drug and food interactions 4. In Germany, use of these agents has increased substantially in recent years 5, 6. However, there are concerns about the assessment of the anticoagulant status in critical situations and the overall medication adherence without coagulation monitoring, the lack of specific antidotes for most of these agents, the dosing in special groups of patients and the higher drug costs compared with VKA 1, 4.
Nursing home residents typically comprise elderly and very elderly persons with a high prevalence of conditions requiring prevention and/or treatment of thromboembolic diseases 7, 8. Comorbid conditions such as an impaired renal function are common, and patients are often treated with several different drugs 9, 10. In this vulnerable population little is known about OAC treatment patterns including the choice of drugs and the reasons for changing treatment regimens.
Therefore our objectives were to (i) describe OAC use in German nursing home residents, (ii) identify factors influencing whether VKAs or NOACs are used to initiate oral anticoagulation and (iii) assess which conditions predict a change of treatment.
Methods
Data source
Using data from the DAK‐Gesundheit, a large statutory health insurance (SHI) fund insuring about six million persons (over 7% of the German population), this cohort study was based on new nursing home residents from 1 January 2010 to 31 December 2014 aged 65 years or older who had been continuously insured in the year preceding nursing home entry. For each person demographic characteristics, information on in‐ and outpatient diagnoses, treatments and procedures as well as outpatient prescriptions and care level were available. Diagnoses are based on the German modification of the International Classification of Diseases, 10th revision (ICD‐10‐GM). Information includes the type of diagnosis (main/ancillary/secondary) for inpatient and the diagnostic certainty (confirmed/suspected/ruled out/status post) for outpatient diagnoses. Therapeutic procedures are coded according to the Operations and Procedures Coding System (OPS). Prescription data include drugs dispensed in a pharmacy and reimbursed by the SHI; they can be linked to the Anatomical Therapeutic Chemical (ATC) classification system.
Patient informed consent was not required by law as the study was based on anonymous data.
Study population and drug exposure
We included all patients receiving at least one prescription of the vitamin K antagonists phenprocoumon (ATC code: B01AA04) or warfarin (B01AA03), the factor Xa inhibitors rivaroxaban (B01AF01, B01AX06) or apixaban (B01AF02) or the direct thrombin inhibitor (factor IIa) dabigatran (B01AE07) during their nursing home stay. Products only approved for the prevention of VTE in patients undergoing elective hip or knee replacement (rivaroxaban 10 mg and dabigatran 75 mg) were not considered. Further, prescriptions of apixaban 2.5 mg and dabigatran 110 mg were excluded if a procedure indicating elective hip or knee replacement but no diagnosis of AF or VTE was found in the 6 months before the prescription. Patients entered the cohort on the day of their first OAC prescription in the nursing home (index prescription). OAC patients treated with different agents on the day of the index prescription were excluded. Patients who had not received an OAC in the 12 months preceding the index prescription (irrespective of whether during or before nursing home stay) were considered incident OAC users. Cohort exit was defined as the first of the following dates: (i) end of insurance, (ii) death or (iii) end of study period.
Covariates
For all patients we determined comorbidity and other conditions, including potential indications for OAC as well as risk factors for stroke and bleeding (definitions presented in Table S1). Chronic conditions were assessed from inpatient (main and ancillary) and confirmed outpatient diagnoses, whereas acute severe events such as major bleeding, myocardial infarction, VTE or stroke were based on hospital data only. Inpatient diagnoses were determined in the 6 months before the index prescription. Outpatient diagnoses, which are coded on a quarterly basis (i.e. a 3‐month interval), were assessed in the quarter of the index prescription and in the two preceding quarters. Comedication was ascertained in the 6 months before and on the day of the index prescription. The number of different drugs (based on the 7th level of the ATC code) other than OACs was assessed during this time period. Finally, the care level ranging from 0/1 (limited daily living skills/substantial need of care) to 3 (most severe need of care) 11 of each patient was obtained at the day of the index prescription.
For AF patients we calculated the CHA2DS2‐VASc score assigning one point each for heart failure, hypertension, diabetes and vascular disease (including myocardial infarction and peripheral arterial disease) and two points for stroke (ischaemic or unspecified, including transient ischaemic attacks, arterial embolism and thrombosis). Being aged 65–74 or ≥75 at the time of the index prescription scored one or two points, respectively, and – if other risk factors were present – one point was added for female sex 12, 13. The CHADS2 score was based on heart failure, hypertension, age ≥75, diabetes (one point each) and stroke (two points). Additionally, the HAS‐BLED score was determined based on hypertension, renal or liver impairment, stroke (ischaemic or unspecified), bleeding history (or a predisposition thereof) and age >65 years (one point each). Being diagnosed with alcohol‐related disorders or treated with drugs increasing the risk of bleeding further scored one point each 12, 13. Since INR values are not part of German claims data and their measurement would have only applied to VKA‐treated patients, ‘labile INR values’ was not included and the maximum value for the HAS‐BLED score in this study was eight rather than nine.
OAC treatment
In prevalent OAC users a potential switch of treatment after nursing home admission was examined including (i) change of substance (within the same drug class) and (ii) change of drug class. We therefore compared a patient's index prescription with the last OAC he or she had received before nursing home entry.
OAC treatment duration was estimated for each patient by summing the number of defined daily doses (DDD) of all eligible OACs prescribed during nursing home stay.
Statistical analysis
Descriptive statistics [percentages, median, interquartile range (IQR)] were used to characterize patients and OAC treatment with respect to the index OAC. Based on the years 2012–2014, two logistic regression models were used to identify (i) factors predicting treatment initiation with NOAC vs. VKA in incident OAC users and (ii) conditions associated with a change of the OAC class after admission to the nursing home in VKA pre‐treated patients (i.e. comparing the index OAC to the last OAC prescribed before nursing home entry). Explored predictors included age and sex, indication for use, known risk factors for stroke and bleeding, other chronic conditions including care level, whether the OAC was prescribed by a general practitioner (GP) or by another physician specialty, a hospital stay in the 7 days preceding the index prescription, the number of additional drugs prescribed and the year of the index prescription. Further, these predictors were also analysed in the subgroup of patients with AF.
From 2010 to 2014, we additionally determined yearly proportions of AF patients with a CHADS2 score ≥2 receiving OAC. For every study year and without applying other inclusion criteria, all nursing home residents with at least one AF diagnosis and conditions equalling a CHADS2 score ≥2 in that year constituted the denominator. Among those, all patients receiving at least one OAC in the respective year were included in the numerator. No temporal association between diagnoses/conditions and prescriptions was required.
Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).
Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 14, and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 15.
Results
During their nursing home stay, of 127 227 newly admitted residents, 16 804 (13.2%) received at least one eligible (i.e. not used for orthopaedic indications) OAC (Figure 1 and Table 1). Median age at the time of the index prescription was 85 (IQR: 80–89) and three quarters of patients were female. The majority received phenprocoumon (58.0%) as index OAC, followed by rivaroxaban (28.1%) and dabigatran (7.3%). Over the study years, a large increase of NOAC index prescriptions namely of rivaroxaban was observed and in 2014, more than 60% of patients had a NOAC index prescription (Figure S1).
Figure 1.
Flowchart for study cohort. AF, atrial fibrillation; OAC, oral anticoagulant; VTE, venous thromboembolism. aassessed by products' strength
Table 1.
Characteristics of nursing home residents treated with oral anticoagulants
| Phenprocoumon (n = 9754) | Warfarin (n = 71) | Rivaroxaban (n = 4719) | Dabigatran (n = 1232) | Apixaban (n = 1028) | Total (n = 16 804) | |
|---|---|---|---|---|---|---|
| Median age (IQR) | 85 (80–88) | 85 (80–89) | 85 (80–89) | 85 (79.5–88) | 86 (81–90) | 85 (80–89) |
| Sex | ||||||
| Male | 2569 (26.3%) | 21 (29.6%) | 1051 (22.3%) | 307 (24.9%) | 231 (22.5%) | 4179 (24.9%) |
| Female | 7185 (73.7%) | 50 (70.4%) | 3668 (77.7%) | 925 (75.1%) | 797 (77.5%) | 12 625 (75.1%) |
| Care level | ||||||
| 0–1 | 6118 (62.7%) | 44 (62.0%) | 2532 (53.7%) | 703 (57.1%) | 539 (52.4%) | 9936 (59.1%) |
| 2 | 3166 (32.5%) | 24 (33.8%) | 1793 (38.0%) | 457 (37.1%) | 417 (40.6%) | 5857 (34.9%) |
| 3 | 470 (4.8%) | 3 (4.2%) | 394 (8.4%) | 72 (5.8%) | 72 (7.0%) | 1011 (6.0%) |
| Potential indication | ||||||
| Atrial fibrillation | 7462 (76.5%) | 65 (91.6%) | 3399 (72.0%) | 1124 (91.2%) | 945 (91.9%) | 12 995 (77.3%) |
| CHA2DS2‐VASc score | ||||||
| Median (IQR) | 5 (4–6) | 6 (5–7) | 5 (5–6) | 5 (5–7) | 6 (5–7) | 5 (4–6) |
| Value ≥ 2 (male) or ≥ 3 (female) | 7451 (99.9%) | 65 (100%) | 3389 (99.7%) | 1121 (99.7%) | 944 (99.9%) | 12 970 (99.8%) |
| CHADS2 score | ||||||
| Median (IQR) | 3 (3–4) | 4 (3–4) | 3 (3–4) | 3 (3–4) | 4 (3–4) | 3 (3–4) |
| Value ≥ 2 | 7174 (96.1%) | 64 (98.5%) | 3283 (96.6%) | 1092 (97.2%) | 924 (97.8%) | 12 537 (96.5%) |
| HAS‐BLED score | ||||||
| Median (IQR) | 3 (3–4) | 4 (3–4) | 3 (3–4) | 3 (3–4) | 4 (3–5) | 3 (3–4) |
| Value ≥ 3 | 5610 (75.2%) | 51 (78.5%) | 2820 (83.0%) | 883 (78.6%) | 839 (88.8%) | 10 203 (78.5%) |
| Venous thromboembolism | 1249 (12.8%) | 2 (2.8%) | 974 (20.6%) | 67 (5.4%) | 43 (4.2%) | 2335 (13.9%) |
| No diagnosis of AF or VTE | 1463 (15.0%) | 5 (7.0%) | 654 (13.9%) | 79 (6.4%) | 67 (6.5%) | 2268 (13.5%) |
| Other conditions | ||||||
| Diabetes | 4135 (42.4%) | 37 (52.1%) | 1875 (39.7%) | 466 (37.8%) | 413 (40.2%) | 6926 (41.2%) |
| Hypertension | 8941 (91.7%) | 69 (97.2%) | 4354 (92.3%) | 1162 (94.3%) | 979 (95.2%) | 15 505 (92.3%) |
| Heart failure | 5678 (58.2%) | 51 (71.8%) | 2711 (57.5%) | 709 (57.6%) | 610 (59.3%) | 9759 (58.1%) |
| Previous myocardial infarction | 644 (6.6%) | 3 (4.2%) | 349 (7.4%) | 95 (7.7%) | 97 (9.4%) | 1188 (7.1%) |
| Previous stroke | 964 (9.9%) | 11 (15.5%) | 687 (14.6%) | 328 (26.6%) | 319 (31.0%) | 2309 (13.7%) |
| Previous bleeding | 1142 (11.7%) | 9 (12.7%) | 645 (13.7%) | 196 (15.9%) | 144 (14.0%) | 2136 (12.7%) |
| Renal disease | 4319 (44.3%) | 38 (53.5%) | 2019 (42.8%) | 419 (34.0%) | 518 (50.4%) | 7313 (43.5%) |
| Liver disease | 1265 (13.0%) | 11 (15.5%) | 631 (13.4%) | 158 (12.8%) | 135 (13.1%) | 2200 (13.1%) |
| Prosthetic heart valve | 462 (4.7%) | 5 (7.0%) | 68 (1.4%) | 33 (2.7%) | 12 (1.2%) | 580 (3.5%) |
| Dementia | 4977 (51.0%) | 32 (45.1%) | 2839 (60.2%) | 656 (53.3%) | 631 (61.4%) | 9135 (54.4%) |
| Recent hospital stay a | 2862 (29.3%) | 15 (21.1%) | 1828 (38.7%) | 403 (32.7%) | 454 (44.2%) | 5562 (33.1%) |
| Median number of different drugs (IQR) | 11 (8–14) | 11 (8–16) | 11 (8–14) | 11 (8–14) | 11 (8–14) | 11 (8–14) |
AF, atrial fibrillation; IQR, interquartile range; VTE, venous thromboembolism
Refers to the 7 days preceding the index prescription
Patient characteristics and indication
The highest prevalence of AF was found in apixaban users (91.9%) and the lowest in those treated with rivaroxaban (72.0%), whereas VTE as a potential indication ranged from 2.8% (warfarin users) to 20.6% (patients receiving rivaroxaban). Overall 794 (4.7%) OAC users had diagnoses of AF and VTE and for a total of 13.5% of patients neither AF nor VTE had been diagnosed. The proportion of patients with none of the respective diagnoses recorded was highest in phenprocoumon (15.0%) and lowest in dabigatran (6.4%) users. Different patterns with respect to indication and index OAC were also observed over the study years (Figure S2).
In AF patients receiving apixaban or warfarin, the median CHA2DS2‐VASc score was 6 (IQR: 5–7) and 5 in those treated with other index OACs. Overall, 78.5% of OAC patients with AF had a HAS‐BLED score ≥3. Values were slightly lower in users of dabigatran, rivaroxaban or phenprocoumon (median: 3; IQR: 3–4) than in patients with apixaban or warfarin index prescriptions (median: 4; IQR: 3–5 and 3–4, respectively).
Patients treated with warfarin had the highest prevalence of diabetes, hypertension and heart failure. In contrast, previous myocardial infarction was least common in these patients (4.2%) whereas the respective proportion among apixaban users was substantially higher (9.4%). A diagnosis of stroke in the 6 months preceding the index prescription ranged from 9.9% (phenprocoumon users) to 31.0% (apixaban‐treated patients). Previous bleeding varied between 11.7% (patients receiving phenprocoumon as index OAC) and 15.9% (dabigatran users). Renal disease was common (overall 43.5%) and most often found in users of warfarin (53.5%) and least often in patients treated with dabigatran (34.0%). The proportion of patients with dementia was lowest among warfarin users (45.1%) and highest in those receiving apixaban as index OAC (61.4%). A care level of 2 or higher ranged from 37.3% (phenprocoumon users) to 47.5% (patients receiving apixaban as index OAC).
AF patients displayed slightly higher proportions of comorbid conditions but overall similar differences with respect to the index prescriptions (Table S2).
Drug use patterns
About 96% of index OACs were prescribed by a GP (Table 2). The proportion of cardiologists issuing the index prescription was highest among patients receiving rivaroxaban, whereas nephrologists more often prescribed phenprocoumon. Overall, 40.7% of patients were incident OAC users. Among prevalent users, 9.3% switched to the other OAC class after entering the nursing home. Of those, over 90% were VKA pre‐treated. Switching from phenprocoumon to rivaroxaban was the most common sequence. Less than 3% of NOAC pre‐treated patients switched to another NOAC. Patients with dabigatran, apixaban or rivaroxaban index prescriptions received substantially more often the lower approved doses (86.7%, 72.2% and 60.9%, respectively). The overall median treatment duration based on DDDs was 200 days (IQR: 100–400). While 13.2% receiving phenprocoumon and 14.1% receiving warfarin index prescriptions were later prescribed NOACs (i.e. switching OAC drug class at least once), only 2.9% of patients with NOAC index prescriptions also received VKAs during nursing home stay. AF patients revealed a higher proportion of prevalent users but overall similar drug use patterns compared to the overall cohort (Table S3).
Table 2.
Use of oral anticoagulants (OAC) during nursing home stay
| Phenprocoumon (n = 9754) | Warfarin (n = 71) | Rivaroxaban (n = 4719) | Dabigatran (n = 1232) | Apixaban (n = 1028) | Total (n = 16 804) | |
|---|---|---|---|---|---|---|
| Physician prescribing index OAC | ||||||
| GP | 9409 (96.5%) | 66 (93.0%) | 4533 (96.1%) | 1171 (95.1%) | 980 (95.3%) | 16 159 (96.2%) |
| Cardiologist | 19 (0.2%) | 0 (0.0%) | 30 (0.6%) | 5 (0.4%) | 15 (1.5%) | 69 (0.4%) |
| Nephrologist | 75 (0.8%) | 0 (0.0%) | 4 (0.1%) | 2 (0.2%) | 0 (0.0%) | 81 (0.5%) |
| Other specialist | 140 (1.4%) | 5 (7.0%) | 95 (2.0%) | 28 (2.3%) | 15 (1.5%) | 283 (1.7%) |
| Unknown | 111 (1.1%) | 0 (0.0%) | 57 (1.2%) | 26 (2.1%) | 19 (1.8%) | 212 (1.3%) |
| Characteristics of index prescription | ||||||
| Incident OAC use | 3535 (36.2%) | 14 (19.7%) | 2329 (49.4%) | 441 (35.8%) | 513 (49.9%) | 6832 (40.7%) |
| Prevalent OAC use | 6219 (63.8%) | 57 (80.3%) | 2390 (50.6%) | 791 (64.2%) | 515 (50.1%) | 9972 (59.3%) |
| No change of substance | 6120 (98.4%) | 50 (87.7%) | 1777 (74.4%) | 611 (77.3%) | 361 (70.1%) | 8919 (89.4%) |
| Other substance, same class | 14 (0.2%) | 6 (10.5%) | 53 (2.2%) | 13 (1.6%) | 40 (7.8%) | 126 (1.3%) |
| Other class | 85 (1.4%) | 1 (1.8%) | 560 (23.4%) | 167 (21.1%) | 114 (22.1%) | 927 (9.3%) |
| Dose of index prescription | ||||||
| Lower available dose a | 14 (0.1%) | N/A | 2868 (60.8%)c | 1068 (86.7%) | 743 (72.3%) | 4693 (27.9%) |
| Higher available dose b | 9740 (99.9%) | 71 (100%) | 1851 (39.2%) | 164 (13.3%) | 285 (27.7%) | 12 111 (72.1%) |
| OAC use during nursing home stay | ||||||
| Only one substance | 8461 (86.7%) | 55 (77.5%) | 4491 (95.2%) | 1062 (86.2%) | 997 (97.0%) | 15 066 (89.7%) |
| Different substances, same drug class | 7 (0.1%) | 6 (8.5%) | 93 (2.0%) | 122 (9.9%) | 12 (1.2%) | 240 (1.4%) |
| Different drug classes | 1286 (13.2%) | 10 (14.1%) | 135 (2.9%) | 48 (3.9%) | 19 (1.9%) | 1498 (8.9%) |
| Median number of DDDs (IQR) | 200 (100–400) | 200 (67–433) | 216 (98–392) | 300 (120–580) | 150 (90–270) | 200 (100–400) |
DDD, defined daily dose; GP, general practitioner; IQR, interquartile range; N/A, not applicable; OAC, oral anticoagulant
Phenprocoumon 1.5 mg/rivaroxaban 15 mg/dabigatran 110 mg/apixaban 2.5 mg
Phenprocoumon 3 mg/warfarin 5 mg/rivaroxaban 20 mg/dabigatran 150 mg/apixaban 5 mg
Includes one prescription of rivaroxaban 2.5 mg
Predictors for treatment choice
A previous stroke was a strong predictor for initiating treatment with a NOAC instead of a VKA [adjusted odds ratio (aOR): 1.76; 95% confidence interval (CI): 1.49–2.08, Figure 2]. Other factors predicting NOAC treatment were a recent hospital stay (aOR: 1.37; 95% CI: 1.22–1.55), receiving the OAC prescription from other specialties than GP (1.54; 1.10–2.15), a care level of 2 (1.23; 1.09–1.39) or 3 (1.61; 1.29–2.01) compared to 0 or 1, and being diagnosed with dementia (1.22; 1.08–1.37). The overall highest impact, however, was observed for the year of cohort entry (2.45; 2.12–2.84 and 3.53; 3.05–4.09 for 2013 and 2014, respectively). In contrast, renal disease predicted treatment initiation with a VKA (aOR: 0.66; 0.59–0.75) as did the presence of a prosthetic heart valve (aOR: 0.31; 0.19–0.50).
Figure 2.
Predictors of initiating oral anticoagulation with NOAC vs. VKA in nursing home residents in the years 2012 to 2014 (n = 5513). ATC, anatomical therapeutic chemical; CI, confidence interval; GP, general practitioner; NOAC, non–vitamin K antagonist oral anticoagulant; OR, odds ratio; VKA, vitamin K antagonist; VTE, venous thromboembolism. Notes: Adjusted OR <1 predicts treatment initiation with VKA, adjusted OR >1 predicts treatment initiation with NOAC. A recent hospital stay refers to the 7 days preceding the index prescription. Index prescriptions with a missing physician specialty were excluded from the analysis
Switching from VKA to NOAC was strongly predicted by a previous stroke, bleeding events (2.55; 2.00–3.24 and 2.43; 2.00–2.95, respectively) and a recent hospital stay (2.74; 2.31–3.26, Figure 3). A lower impact was observed for female sex (1.31; 1.09–1.58), being treated with a higher number of different drugs (1.27; 1.07–1.51) and a care level greater than 1 (1.39; 1.17–1.65 and 1.62; 1.14–2.31 for care level 2 and 3, respectively). Receiving the index prescription in 2013 (1.57; 1.27–1.93) or 2014 (1.69; 1.38–2.09) compared to 2012 was associated with a higher probability to change treatment as well. The only factor significantly predicting a continued VKA treatment was the presence of a prosthetic heart valve (aOR: 0.38; 0.24–0.59). Predictors for treatment initiation with or switching to NOAC in AF patients did not differ from our general results (Figures S3 and S4, respectively).
Figure 3.
Predictors of switching oral anticoagulation from VKA to NOAC with nursing home entry vs. continuing VKA treatment in the years 2012 to 2014 (n = 4829). ATC, anatomical therapeutic chemical; CI, confidence interval; GP, general practitioner; NOAC, non‐vitamin K antagonist oral anticoagulant; OR, odds ratio; VKA, vitamin K antagonist; VTE, venous thromboembolism. Notes: Adjusted OR <1 predicts continuation with VKA, adjusted OR >1 predicts switching from VKA to NOAC with nursing home entry. A recent hospital stay refers to the 7 days preceding the index prescription. Index prescriptions with a missing physician specialty were excluded from the analysis
Oral anticoagulation in atrial fibrillation over the years
In the yearly cohorts, less than half of the AF patients with a CHADS2 score ≥2 received OAC (Figure 4) displaying a substantial increase from 27% in 2010 to 46% in 2014. Patients with a HAS‐BLED score ≥3 were less likely to receive OAC than those with a score <3. Basing the analysis on the CHA2DS2‐VASc score did not change these results (data not shown).
Figure 4.
Patients with atrial fibrillation and a CHADS2 score ≥ 2 by year and bleeding risk
Discussion
In this study we examined OAC use in new nursing home residents, a population with a high prevalence of indications requiring anticoagulation 7, 8 but also an elevated risk for bleeding. Thus, balancing risks and benefits, the decision for OAC treatment, the choice of substance and the factors influencing a change of regimen are of high relevance. With a median age of 85 years, the OAC patients in our study were substantially older than those participating in the respective phase III trials where dabigatran, rivaroxaban, or apixaban were compared with warfarin 16. Further, a high proportion of patients had renal disease and other comorbid conditions and frequently received other medication indicating that in this real world situation, patient characteristics do not match the trial settings 17.
Being admitted to a nursing home is often accompanied by a change of health professionals involved. Medical care now usually takes place in the nursing home instead of the physician's office, hampering regular laboratory tests such as INR monitoring. Based on a patient's medication plan, drugs are administered by the nursing home staff. Acute events and/or chronic conditions leading to nursing home admission might further result in a change of patients' medication influencing OAC choice.
As expected, with a prevalence of 78%, AF was the most common indication for oral anticoagulation and based on the CHA2DS2‐VASc score, OAC treatment was recommended for almost all AF patients. VTE was far less common and for a small proportion of patients diagnoses of AF and VTE were found. In contrast, nearly 14% of patients had none of the examined diagnoses recorded. For VKAs this might be explained by other approved indications such as the prevention of systemic embolism in patients with prosthetic heart valves. Although excluding products only approved for orthopaedic indications, NOACs might have been used in this context.
In our study, a steep increase of NOAC use was found over the years mirroring German prescription statistics 5, 18. The differing patterns observed when stratifying by potential reason for use probably reflect the diverging sequence of the agents' approval for the respective indications. While several guidelines such as those published by the European Society of Cardiology, the German Cardiac Society or the German Stroke Society give preference to the new substances in AF management 12, 19, 20, 21, the Drug Commission of the German Medical Association recommends their use mainly for patients for whom VKA do not represent an adequate treatment option 1, 22. Accordingly, the German College of General Practitioners and Family Physicians found no advantage in treating AF patients who can be adequately treated with VKAs with NOACs instead 23.
In our study, stroke was the only acute condition significantly predicting treatment initiation with NOACs instead of VKA, which was also observed by a recent study based on the UK's Clinical Practice Research Datalink identifying patient characteristics associated with NOAC initiation 24. In contrast, both prior stroke as well as bleeding, reflecting the current treatment's effectiveness and safety, equally impacted switching from VKA to NOAC vs. continuing VKA treatment in this present analysis.
About one third of patients had been hospitalized in the 7 days preceding the index OAC. A recent hospitalization highly predicted treatment initiation with a NOAC and switching to NOAC, probably mirroring the tendency that new products introduced in the hospital market subsequently influence drug therapy in the outpatient setting 25. The large influence of recent hospital stays on NOAC initiations was also observed by a Danish study where hospital physicians were responsible for the initiation of 73.5% of all new NOAC users 26. With respect to treatment changes, another Danish study found that approximately half of all changes were preceded by a hospitalization 27. Although overall rare, the initiation of OAC treatment by other specialists than a GP was associated with NOAC use.
The only factor significantly associated with both initiation and maintenance of VKA treatment in our study was the presence of a prosthetic heart valve, a condition in which NOAC use is not recommended or indicated. The other factor significantly predicting treatment initiation with VKA was renal disease, a common condition in nursing home residents 28. For VKAs, renal clearance is considered to be a minor determinant of anticoagulant response and the anticoagulant dose can be titrated and monitored using the INR. In contrast, information on NOAC use in patients with severe chronic kidney disease is scarce 12 and a regular monitoring of the renal function is required. This is especially important for the direct thrombin inhibitor dabigatran 29. The comparatively low proportion of patients with renal disease among dabigatran users probably mirrors the fact that dabigatran is contraindicated in patients with severe renal impairment and the special caution applied to this substance. Whether renal function was monitored regularly was beyond the scope of this present analysis but a recent cross‐sectional study in German nursing homes found that in one quarter of patients creatinine values had not been determined in 1 year 28.
Interestingly, an impaired renal function did not significantly influence the decision whether to change VKA regimen to NOAC or continue VKAs. This indicates that other factors such as stroke or bleeding were considered more important. In this context, the above‐mentioned UK study found that patients with chronic kidney disease were less likely to be prescribed NOACs compared with VKAs early after the formers' introduction but not when using data from 2015 24, suggesting that over the years physicians might have become more confident to use NOACs also in these patients.
An impaired renal function and/or an increased risk of bleeding 29, 30, 31 but also other factors such as older age 29, 31 and low body weight 31 probably influenced the overall lower NOAC strengths prescribed to our study population, thereby reflecting recommended dose reductions in these patients. For rivaroxaban, the lower strength might also refer to the initial phase of VTE treatment where 15 mg twice daily is followed by 20 mg once daily 30. The high proportion of patients with a thromboembolic event preceding a rivaroxaban index prescription supports this assumption. Since VKA dosing is adapted according to the INR values obtained, the VKA dose applied cannot directly be deduced from the dispensed strength.
Another common condition in German nursing home residents is polypharmacy 10 which increases the risk for drug interactions. VKAs are among the substance classes often involved in drug interactions leading to hospitalization 32 predominantly by an elevated risk of bleeding 33, 34. While pharmacodynamic effects also apply to NOACs, their decreased sensitivity to pharmacokinetic interactions is considered advantageous compared with VKAs. In our study, receiving more drugs did not significantly influence treatment choice for OAC initiation but was associated with switching oral anticoagulation to NOACs.
When considering a higher care level and the presence of dementia, both were significantly associated with NOAC initiation and a switch to NOAC treatment. The omission of laboratory tests and the fixed dosage schemes assumingly have advantages with respect to adherence and practicability for the patient and probably also for the personnel involved in adapting and implementing medication plans.
Considering the yearly cohorts of nursing home residents, only between 27% and 46% of the AF patients with a CHADS2 score ≥2 received OACs. These low proportions are remarkable, since the considerable stroke risk without OAC treatment often exceeds the risk of bleeding associated with OAC use also in the elderly, in patients with cognitive dysfunction, or in patients with frequent falls or frailty 12.
The overall rise in NOAC prescriptions and the increase in OAC use in AF patients observed in our study suggest that with the introduction of the new substances, physicians have become more confident to prescribe OAC in elderly nursing home residents including those with a high bleeding risk.
For this study some strengths and limitations have to be considered. A major strength is the large cohort of elderly nursing home residents which reflects real world practice. By using pharmacy dispensing data for exposure information, recall bias can be ruled out and information with respect to time, brand, and dispensed dose can be regarded as precise 35. We were able to include several comorbidities based on in‐ and outpatient diagnoses and comedication as well as other conditions such as care level.
Limitations are due to the nature of the underlying administrative data. Since laboratory parameters are not included, we had no information about INR values or the time patients spent within their recommended therapeutic range which has a major impact on the treatment benefit of VKAs 36 and therefore probably also influenced the decision to continue VKA or switch to NOAC. Furthermore, we had no information on other patients' conditions such as overall frailty or frequent falls. As a further drawback, our study did not include a review of individual patient files which for data protection reasons is not feasible in Germany. Moreover, no direct linkage is possible between prescriptions and diagnoses as reflected by the patients for whom both examined indications were recorded. This uncertainty regarding prescriptions and corresponding diagnoses might have been more pronounced in our yearly analyses of AF patients, as no temporal association between prescriptions and diagnoses/conditions was required.
Finally, we had no information whether the physician issuing the last prescription before nursing home entry was the one responsible for the index prescription.
In conclusion, the overall tendency in Germany to use NOACs instead of VKAs was also observed in the nursing home setting affecting both the initial prescription as well as a potential switch of treatment. Aspects such as intended higher effectiveness and safety but probably also practicability might have influenced our findings. However, factors such as the possibility to assess the anticoagulant status or use and adaptation in patients with impaired renal function, a very common condition in nursing home residents, should also be taken into account and NOAC treatment should be conducted with caution.
Although not the focus of this present study and despite an observed increase over the study years, the overall low proportion of nursing home residents with AF and a CHADS2 score ≥2 receiving OAC treatment was remarkable and requires further investigation.
Competing Interests
There are no competing interests to declare.
The authors thank the DAK‐Gesundheit for providing the data for this study.
Contributors
K.J. contributed to the study concept and design, analysed and interpreted the data and drafted the manuscript. F.H. contributed to the study concept and design, interpreted the data and revised the manuscript. S.H.‐R. contributed to the study concept, interpreted the data, revised the manuscript and provided clinical expertise. M.D. contributed to the study concept and design, analysed and interpreted the data and revised the manuscript.
Supporting information
Table S1 Definitions of diseases, therapeutic procedures and drugs
Table S2 Characteristics of nursing home residents with atrial fibrillation treated with oral anticoagulants
Table S3 Use of oral anticoagulants (OAC) in patients with atrial fibrillation during nursing home stay
Figure S1 Index of oral anticoagulants by year
Figure S2 Index of oral anticoagulants by year and potential indication
Figure S3 Predictors of initiating oral anticoagulation with NOAC vs. VKA in nursing home residents with atrial fibrillation in the years 2012–2014 (n = 3686)
Figure S4 Predictors of switching oral anticoagulation with NOAC vs. VKA in nursing home residents with atrial fibrillation in the years 2012–2014 (n = 4097)
Jobski, K. , Hoffmann, F. , Herget‐Rosenthal, S. , and Dörks, M. (2018) Use of oral anticoagulants in German nursing home residents: drug use patterns and predictors for treatment choice. Br J Clin Pharmacol, 84: 590–601. doi: 10.1111/bcp.13474.
References
- 1. Arzneimittelkommission der Deutschen Ärzteschaft (AKdÄ) . Leitfaden: Orale Antikoagulation bei nicht valvulärem Vorhofflimmern (2nd Edition) [online]. 2016. Available at http://www.akdae.de/Arzneimitteltherapie/TE/LF/PDF/OAKVHF.pdf (last accessed 9 March 2017)
- 2. Ufer M. Comparative pharmacokinetics of vitamin K antagonists: warfarin, phenprocoumon and acenocoumarol. Clin Pharmacokinet 2005; 44: 1227–1246. [DOI] [PubMed] [Google Scholar]
- 3. European Commission . Community register of medicinal products for human use [online]. Available at http://ec.europa.eu/health/documents/community‐register/html/register.htm (last accessed 26 May 2017).
- 4. Monaco L, Biagi C, Conti V, Melis M, Donati M, Venegoni M, et al Safety profile of the direct oral anticoagulants: an analysis of the WHO database of adverse drug reactions. Br J Clin Pharmacol 2017; 83: 1532–1543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Hein L. Antithrombotika und Antihämorrhagika In: Arzneiverordnungs‐Report 2013, eds Schwabe U, Paffrath D. Berlin, Heidelberg: Springer‐Verlag, 2013; 413–433. [Google Scholar]
- 6. Hein L, Wille H. Antithrombotika und Antihämorrhagika In: Arzneiverordnungs‐Report 2016, eds Schwabe U, Paffrath D. Berlin, Heidelberg: Springer‐Verlag, 2016; 351–368. [Google Scholar]
- 7. Sinnaeve PR, Brueckmann M, Clemens A, Oldgren J, Eikelboom J, Healey JS. Stroke prevention in elderly patients with atrial fibrillation: challenges for anticoagulation. J Intern Med 2012; 271: 15–24. [DOI] [PubMed] [Google Scholar]
- 8. Engbers MJ, van Hylckama Vlieg A, Rosendaal FR. Venous thrombosis in the elderly: incidence, risk factors and risk groups. J Thromb Haemost 2010; 8: 2105–2112. [DOI] [PubMed] [Google Scholar]
- 9. Onder G, Liperoti R, Fialova D, Topinkova E, Tosato M, Danese P, et al Polypharmacy in nursing home in Europe: results from the SHELTER study. J Gerontol A Biol Sci Med Sci 2012; 67: 698–704. [DOI] [PubMed] [Google Scholar]
- 10. Dörks M, Herget‐Rosenthal S, Schmiemann G, Hoffmann F. Polypharmacy and renal failure in nursing home residents: results of the inappropriate medication in patients with renal insufficiency in nursing homes (IMREN) study. Drugs Aging 2016; 33: 45–51. [DOI] [PubMed] [Google Scholar]
- 11. Büscher A, Wingenfeld K, Schaeffer D. Determining eligibility for long‐term care – lessons from Germany. Int J Integr Care 2011; 11: e019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B, et al 2016 ESC guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Europace 2016; 18: 1609–1678. [DOI] [PubMed] [Google Scholar]
- 13. Friberg L, Rosenqvist M, Lip GYH. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish atrial fibrillation cohort study. Eur Heart J 2012; 33: 1500–1510. [DOI] [PubMed] [Google Scholar]
- 14. Southan C, Sharman JL, Benson HE, Faccenda E, Pawson AJ, Alexander SPH, et al The IUPHAR/BPS Guide to PHARMACOLOGY in 2016: towards curated quantitative interactions between 1300 protein targets and 6000 ligands. Nucleic Acids Res 2016; 44: D1054–D1068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Alexander SP, Fabbro D, Kelly E, Marrion NV, Peters JA, Faccenda E, et al The Concise Guide to PHARMACOLOGY 2017/18: Enzymes. Br J Pharmacol 2017; 174: S272–S359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Savelieva I, Camm AJ. Practical considerations for using novel oral anticoagulants in patients with atrial fibrillation. Clin Cardiol 2014; 37: 32–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Pattullo CS, Barras M, Tai B, McKean M, Donovan P. New oral anticoagulants: appropriateness of prescribing in real‐world setting. Intern Med J 2016; 46: 812–818. [DOI] [PubMed] [Google Scholar]
- 18. Hein L. Antithrombotika und Antihämorrhagika In: Arzneiverordnungs‐Report 2015, eds Schwabe U, Paffrath D. Berlin, Heidelberg: Springer‐Verlag, 2015; 477–500. [Google Scholar]
- 19. Camm AJ, Lip GYH, De Caterina R, Savelieva I, Atar D, Hohnloser SH, et al 2012 focused update of the ESC guidelines for the management of atrial fibrillation. Eur Heart J 2012; 33: 2719–2747. [DOI] [PubMed] [Google Scholar]
- 20. Deutsche Gesellschaft für Kardiologie – Herz und Kreislaufforschung e.V. Leitlinien für das Management von Vorhofflimmern [online]. 2013. Available at https://leitlinien.dgk.org/files/Pocket_Leitlinien_Vorhofflimmern_Update2013.pdf (last accessed 11 December 2017).
- 21. Arbeitsgemeinschaft der Wissenschaftlichen Fachgesellschaften . S3‐Leitlinie Sekundärprophylaxe ischämischer Schlaganfall und transitorische ischämische Attacke [online]. 2015. Available at http://www.awmf.org/uploads/tx_szleitlinien/030-133k_S3_Sekun%C3%A4rprophylaxe_isch%C3%A4mischer_Schlaganfall_2015-02.pdf (last accessed 31 March 2017).
- 22. Arzneimittelkommission der deutschen Ärzteschaft (AKdÄ) . Leitfaden: Orale Antikoagulation bei nicht valvulärem Vorhofflimmern [online]. 2012. Available at http://www.akdae.de/Arzneimitteltherapie/TE/LF/PDF/OAKVHF.pdf (last accessed 8 October 2015).
- 23. Deutsche Gesellschaft für Allgemeinmedizin und Familienmedizin . Neue orale Antikoagulantien (bei nicht valvulärem Vorhofflimmern) [online]. Available at http://www.degam.de/files/Inhalte/Leitlinien-Inhalte/Dokumente/S1-Handlungsempfehlung/053-031%20Neue%20orale%20Antikoagulantien/053-031_NOAK_Langfassung_redakt%20Hinweis_02%2002%202017.pdf (last accessed 4 May 2017).
- 24. Loo SY, Dell'Aniello S, Huiart L, Renoux C. Trends in the prescription of novel oral anticoagulants in UK primary care. Br J Clin Pharmacol 2017; 83: 2096–2106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Grimmsmann T, Schwabe U, Himmel W. The influence of hospitalisation on drug prescription in primary care – a large‐scale follow‐up study. Eur J Clin Pharmacol 2007; 63: 783–790. [DOI] [PubMed] [Google Scholar]
- 26. Pottegård A, Poulsen BK, Larsen MD, Hallas J. Dynamics of vitamin K antagonist and new oral anticoagulants use in atrial fibrillation: a Danish drug utilization study. J Thromb Haemost 2014; 12: 1413–1418. [DOI] [PubMed] [Google Scholar]
- 27. Hellfritzsch M, Grove EL, Husted SE, Rasmussen L, Poulsen BK, Johnsen SP, et al Clinical events preceding switching and discontinuation of oral anticoagulant treatment in patients with atrial fibrillation. Europace 2017; 19: 1091–1095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Hoffmann F, Boeschen D, Dörks M, Herget‐Rosenthal S, Petersen J, Schmiemann G. Renal insufficiency and medication in nursing home residents. Dtsch Ärzteblatt Int 2016; 113: 92–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Boehringer Ingelheim . Summary of products characteristics Pradaxa [online]. Available at http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_‐_Product_Information/human/000829/WC500041059.pdf (last accessed 16 March 2017).
- 30. Bayer Pharma . Summary of products characteristics Xarelto [online]. Available at http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_‐_Product_Information/human/000944/WC500057108.pdf (last accessed 16 March 2017).
- 31. Bristol‐Myers Squibb/Pfizer EEIG . Summary of products characteristics Eliquis [online]. Available at http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_‐_Product_Information/human/002148/WC500107728.pdf (last accessed 16 March 2017).
- 32. Budnitz DS, Lovegrove MC, Shehab N, Richards CL. Emergency hospitalizations for adverse drug events in older Americans. N Engl J Med 2011; 365: 2002–2012. [DOI] [PubMed] [Google Scholar]
- 33. Jobski K, Behr S, Garbe E. Drug interactions with phenprocoumon and the risk of serious haemorrhage: a nested case–control study in a large population‐based German database. Eur J Clin Pharmacol 2011; 67: 941–951. [DOI] [PubMed] [Google Scholar]
- 34. Abbas S, Ihle P, Harder S, Schubert I. Risk of bleeding and antibiotic use in patients receiving continuous phenprocoumon therapy. Thromb Haemost 2014; 111: 912–922. [DOI] [PubMed] [Google Scholar]
- 35. Schneeweiss S, Avorn J. A review of uses of health care utilization databases for epidemiologic research on therapeutics. J Clin Epidemiol 2005; 58: 323–337. [DOI] [PubMed] [Google Scholar]
- 36. Connolly SJ, Pogue J, Eikelboom J, Flaker G, Commerford P, Franzosi MG, et al Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the quality of international normalized ratio control achieved by centers and countries as measured by time in therapeutic range. Circulation 2008; 118: 2029–2037. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1 Definitions of diseases, therapeutic procedures and drugs
Table S2 Characteristics of nursing home residents with atrial fibrillation treated with oral anticoagulants
Table S3 Use of oral anticoagulants (OAC) in patients with atrial fibrillation during nursing home stay
Figure S1 Index of oral anticoagulants by year
Figure S2 Index of oral anticoagulants by year and potential indication
Figure S3 Predictors of initiating oral anticoagulation with NOAC vs. VKA in nursing home residents with atrial fibrillation in the years 2012–2014 (n = 3686)
Figure S4 Predictors of switching oral anticoagulation with NOAC vs. VKA in nursing home residents with atrial fibrillation in the years 2012–2014 (n = 4097)