We assessed the pattern of use and the effectiveness of antithrombotic therapy in a cohort of high-risk elderly patients hospitalized for atrial fibrillation, with data derived from prescription, hospitalization, and mortality databases from 3 linked registries. Results showed that antithrombotic therapy was underused, even in patients with no comorbid conditions. In addition, patients exposed to antithrombotic therapy had a significantly lower mortality rate. The collection of epidemiological data by record linkage represents a flexible and readily available tool for monitoring and improving routine clinical care.
Atrial fibrillation is primarily a disease of older people and is frequently associated with other age-related conditions.1,2 The clinical importance and public health relevance of this disease are increasing because of an aging population.
The most worrisome risk of atrial fibrillation is thromboembolism, specifically ischemic stroke. Several clinical trials and international guidelines have documented that antithrombotic therapy, particularly oral anticoagulation, is crucial for the prevention of thromboembolic events and the substantial reduction of the risk of stroke and death.3-5 For patients who cannot receive oral anticoagulation therapy, antiplatelet treatment is an effective alternative.6,7
Some observational studies have indicated that antithrombotic therapy, specifically oral anticoagulation therapy, is often underused in clinical practice.8,9 However, the clinical consequences of this underuse have not been assessed. We conducted a study to evaluate the use of antithrombotic therapy among unselected high-risk elderly patients and the effectiveness of antithrombotic therapy during the follow-up period.
Patients and methods
We analyzed the records of patients who were discharged from the hospital using linked databases of prescription and mortality records. The cohort consisted of consecutive patients aged 65 years or older who were hospitalized with a primary or secondary diagnosis of atrial fibrillation in a 12-month period (from January 1, 2002, to December 31, 2002). The first hospital admission during that period was considered the index date. Follow-up was extended from the index date to day 360 or until the occurrence of death from any cause or the occurrence of a first stroke, transient ischemic attack, or embolic episode.
We also included in the analysis cardiovascular and noncardiovascular comorbidity, as documented by hospitalizations or long-term exposure to pharmacologic treatments, over the 24-month period preceding the index date. Cardiovascular comorbidities included risk factors for stroke (hypertension, congestive heart failure, transient ischemic attack, coronary heart disease, and diabetes),10 previous presence of peripheral vascular disease, an embolic episode, and atrial fibrillation. Noncardiovascular conditions included malignancy, previous hospitalization for major bleeding, and chronic obstructive pulmonary disease.
The pattern of exposure to antithrombotic therapy agents was evaluated in the period preceding and following the index hospitalization. Patients receiving long-term antithrombotic therapy were categorized according to the degree of compliance with recommended strategies (≤ 30%, 31%-70%, or ≥ 70%). The association between antithrombotic therapy and major events during follow-up was evaluated using a fully adjusted Cox proportional hazards model. Variables included in the multivariate analysis were: age, sex, previous hospitalization for atrial fibrillation, congestive heart failure, diabetes, coronary heart disease, peripheral vascular disease, hypertension, malignancy, chronic obstructive pulmonary disease, stroke, transient ischemic attack, hospitalization for major bleeding, an embolic episode, and previous exposure to antithrombotic therapy.
Table. Patient characteristics.
ATT indicates antithrombotic therapy; TIA, transiet ischemic attack; AF, atrial
*All patients hospitalized with atrial fibrillation.
†Patients discharged alive (n = 1812) and stratified by the prescription of ATT
We identified 1920 patients with a diagnosis of atrial fibrillation. The mean age was 78.8 years, and 54.9% of patients were women (Table). As expected, there was a high prevalence of hypertension (78.2%), congestive heart failure (50.5%), coronary heart disease (20.3%), and diabetes (20.5%). In addition, 6.9% of patients had peripheral artery disease, and 19.4% were hospitalized with atrial fibrillation before the index date.
Despite the cardiovascular risk profile, 49.3% of patients had previously received antithrombotic therapy: 390 (20.3%) had received antiplatelets, 459 (23.9%) had received anticoagulants, and 97 (5.0%) had received both types of agents. Overall, 556 patients (28.9%) were prescribed anticoagulants, and 487 (25.4%) were prescribed antiplatelets. As shown in the Table, several noncardiovascular conditions were present, particularly chronic obstructive pulmonary disease (20.1%) and malignancy (9.0%).
Patients who received antithrombotic therapy before the index date tended to be younger and were more likely to be men; more likely to have a history of hypertension, coronary heart disease, atrial fibrillation, and previous exposure to antithrombotic therapy; and more likely to have fewer noncardiovascular comorbidities (Table).
In-hospital mortality was 5.6%. Among 1812 patients discharged, 1002 patients received antithrombotic therapy. A total of 521 (28.7%) patients received only anticoagulants, 392 (21.6%) received only antiplatelet agents, and 89 (4.9%) received both drugs. Among those patients prescribed anticoagulants, 11% adhered to this treatment ≤ 30% of the time during follow-up, 20% adhered to treatment from 31% to 70% of the time, and 69% adhered to treatment > 70% of the time. Similarly, among the 481 patients prescribed antiplatelets, 21% adhered to treatment ≤ 30% of the time during follow-up, 25.4% adhered to treatment from 31% to 70% of the time, and 53.6% adhered to treatment > 70% of the time. When we excluded patients who had any noncardiovascular comorbidity and those who died during the first 30 days after discharge, 420 (41.3%) were prescribed anticoagulant agents, and 289 (28.4%) were prescribed antiplatelet agents.
A total of 336 patients died during the follow-up period. Death rates were significantly lower among patients who had taken antithrombotic therapy. The reduction in all-cause mortality was —77% (–65% to –85%; < .001) with anticoagulant agents and —34% (–14% to –50%; = .003) with antiplatelet agents. Antithrombotic therapy was associated with a —64% (–53% to –72%; < .001) reduction in all-cause mortality overall (Figure). The adjusted probability of having a major event for patients on antithrombotic therapy was 0.86 (0.50-1.5; = .60). Anticoagulant agents were associated with a probability of thromboembolic events of 0.52 (0.25-1.07). The annualized rate of bleeding episodes was 13 per 1000 patient-years.
Figure. Effect of antithrombotic therapy (A), anticoagulants (B), and
antiplatelets (C) on total mortality. Antithrombotic therapy, anticoagulants,
and antiplatelets in solid line. No treatment in dotted line. Antithrombotic
therapy: HR = 0.36 (95% CI, 0.28-0.47), < .001; anticoagulants:
HR = 0.23 (95% CI, 0.15-0.35), < .001; antiplatelets: HR = 0.66
(95% CI, 0.50-0.86), = .003.
The chief finding of this analysis is that antithrombotic therapy is significantly underused among unselected high-risk elderly patients. There may be several reasons for this result. Patients in clinical trials are significantly different from those in clinical practice11; in our study, the average age of patients was 79 years. Furthermore, the high rate of noncardiovascular comorbidity could explain this low exposure to antithrombotic therapy. Our study shows that patients with chronic obstructive pulmonary disease and malignancy received antithrombotic therapy less often. When patients with noncardiovascular conditions were excluded, however, the proportion of patients exposed to antithrombotic treatment remained low.
The most recommended antithrombotic agents are anticoagulants. However, physicians may perceive anticoagulation therapy as difficult to prescribe in clinical practice. Anticoagulation clinics report huge variability in safety issues,12 a fact that additionally contributes to this uncertainty. Although a definitive conclusion cannot be drawn from our study because the sample size was underpowered, we did not find a rate of major bleeding that would suggest a harmful effect of these agents.
Patients are also usually hesitant to take anticoagulation therapy. In one study, only 61% of patients said they would prefer anticoagulation therapy to no treatment.13 In our study, about 70% of patients prescribed anticoagulants received the treatment for at least 70% of their follow-up time, indicating that adherence is feasible. Undertreatment occurred not only with anticoagulant agents but also with antiplatelet agents: only one third of patients were prescribed anticoagulants, and among those who were not, only one third received antiplatelet agents.
Another finding of our study was that antithrombotic therapy was associated with a significant reduction in total mortality, and anticoagulation was associated with a reduction in thromboembolic events. Antithrombotic therapy reduced mortality significantly more than it reduced thrombotic events, probably because of a more important reduction in fatal (compared with nonfatal) strokes among elderly patients. In our cohort, the reduction in total mortality with both anticoagulant and antiplatelet agents appears to be even greater than that observed in randomized trials, suggesting also that the effectiveness attributable to these agents is greater with increasing clinical risk.
The results of our study showed that antithrombotic therapy was underused among high-risk elderly patients hospitalized with atrial fibrillation, including in those without comorbid conditions. Our findings strongly support the use of antithrombotic therapy in a population of high-risk patients. In fact, exposure to antithrombotic therapy was associated with improved survival. Although our study was performed in 1 country, the results are consistent with other studies. 3,7,14 Our results also showed that epidemiologic data collected by record linkage should become a major tool for clinical research aimed at surveying and improving routine clinical care.15
and an exam are available with this article