Abstract
Objective To assess whether family physicians are using the CHADS2 (congestive heart failure, hypertension, age ≥ 75, diabetes mellitus, and stroke or transient ischemic attack) score in the decision to initiate warfarin therapy to prevent stroke in patients with atrial fibrillation.
Design Retrospective analysis of the medical records of patients with atrial fibrillation.
Setting Data were gathered from records at 3 clinics in a primary care network in Edmonton, Alta.
Participants The medical records of patients with atrial fibrillation who were currently taking warfarin therapy.
Main outcome measures Percentage of patients whose CHADS2 scores indicated warfarin therapy for stroke prophylaxis compared with the actual percentage of patients taking warfarin therapy. Data on patients’ age, number of medications, and number of comorbid conditions were also recorded.
Results Among these patients, 7% had a CHADS2 score of 0, for which no warfarin therapy was indicated; 21% had a score of 1, for which either acetylsalicylic acid or warfarin was indicated; and 72% had a score of 2 or greater, for which warfarin therapy was indicated. About 80% of patients were taking medication to control their heart rate.
Conclusion The CHADS2 score is not being used in all cases to assess the need for warfarin therapy for preventing stroke in patients with atrial fibrillation. The CHADS2 score might be of limited use because it is not sensitive enough to stratify patients clearly into high-, intermediate-, and low-risk groups. Although guidelines for stroke prevention should be followed, the CHADS2 portion of the guidelines might not be the most effective way to assess patients’ risk of stroke.
Atrial fibrillation is a commonly encountered condition in family practice and a risk factor for thromboembolic stroke. Patients with atrial fibrillation can benefit from anticoagulation therapy to reduce their risk of stroke. The decision to initiate anticoagulation therapy is made by weighing pertinent risk factors with potential benefit. Warfarin, a vitamin K antagonist, is often used for anticoagulation and has been shown to be superior to acetylsalicylic acid (ASA)1 for preventing stroke. However, warfarin has a narrow therapeutic range, and the decision to initiate warfarin therapy needs to be made with great care, as complications can include intracranial hemorrhage.
The decision to initiate warfarin therapy for patients with atrial fibrillation is dictated by patients’ cumulative risk factors and consideration of the contraindications to warfarin therapy. The CHADS2 (congestive heart failure, hypertension, age ≥ 75, diabetes mellitus, and stroke or transient ischemic attack) score, a stroke-risk stratification schema, is a common, easy-to-use, evidence-based tool available to clinicians.2–5 Briefly, CHADS2 scores are calculated by allocating points to patients based on their past and current medical conditions as criteria for risk of future stroke.1
The CHADS2 score was first used and validated in 20016 and since then has been shown to be superior to the stroke-risk prediction tools4,6 previously available. The CHADS2 score categorizes patients’ risk of stroke: a score of 0 equals low risk, 1 equals moderate risk, and 2 or greater equals high risk.1 According to current guidelines, warfarin is indicated for patients at high risk (ie, CHADS2 score ≥ 2).4,5
This study examines physicians’ level of adherence to clinical practice guidelines for managing patients with atrial fibrillation. The results could aid in assessing whether physicians are following guidelines, specifically with regard to recommendations based on the CHADS2 score.
METHODS
The electronic medical records of 3 clinics in the Edmonton-Oliver Primary Care Network in Alberta were searched to identify retrospectively patients who were diagnosed with atrial fibrillation and also prescribed warfarin. A follow-up search identified patients with atrial fibrillation who were not prescribed warfarin for comparison. The study received ethics approval from the Human Research Ethics Board at the University of Alberta in Edmonton, and consent to view the records was obtained from physicians in the network before the search.
Both chronic and paroxysmal atrial fibrillation were considered because anticoagulation therapy is beneficial for both conditions.7 Records were excluded if patients were deceased or warfarin therapy was not ongoing at the time of the search. Patients’ age, medications, and comorbid conditions were abstracted into a database for statistical analysis. Number of medications, number of patients taking specific medications, number of comorbid conditions, and patients’ age were recorded. The CHADS2 scores were derived from the data collected.
The CHADS2 score categorizes patients into 3 risk groups.5 For each group, we determined the number and proportion of men and women, the average number of medications they were taking, and the average number of comorbid conditions they had. We then determined the percentage of patients in each risk group and documented how many patients were using medications to control their heart rate (pharmacologic control of heart rate is another component of atrial fibrillation treatment guidelines7 and was used as an independent measure of adherence to guidelines). To identify differences among risk groups, these factors were compared across groups using one-way ANOVA (analysis of variance) for continuous factors (ie, age, number of medications, number of comorbid conditions) followed by Tukey post-hoc analysis when significance was found, and χ2 tests of association for categorical factors (ie, sex, prescriptions for drugs to control heart rate).
Owing to software limitations in the electronic medical records, only 1 clinic was able to generate the number of patients with atrial fibrillation who were not prescribed warfarin. Comparison of patients taking warfarin and not taking warfarin was done within this clinic, rather than across all 3 clinics. Because of the small sample size, nonparametric analyses were used to compare these 2 groups: a Fisher exact test for dichotomous outcomes (ie, sex) and a Mann-Whitney test for continuous outcomes (ie, age, medications, comorbid conditions).
RESULTS
The search found the records of 415 patients with atrial fibrillation who were taking warfarin for prophylaxis of stroke. To ensure that pooling patient data from 3 different clinics would not skew our results significantly, we compared our main outcome variables among the 3 clinics and found no statistically or clinically significant differences (data not published). Thus, we could justify combining results from the clinics into a single analysis. Demographic data on patients in each group, as well as on the collective population sample, are shown in Table 1. Increases in mean patient age, number of medications, and number of comorbid conditions correlated with progression from low- to intermediate- to high-risk groups. Of particular interest was the finding that 28% of patients were not in a category that clearly indicated warfarin as appropriate therapy, and 7% of patients were in the low-risk group, for which warfarin was not indicated at all.
Table 2 shows the frequency with which drugs commonly used to control heart rate (β-blockers, digoxin, or nondihydropyridine calcium channel blockers) were prescribed.7 Overall, 80% of patients were taking at least 1 type of drug to control heart rate. A χ2 value of 2.65 was calculated, so we were unable to identify any significant differences across risk groups with respect to prescriptions for heart rate control medications.
Only 1 clinic’s records indicated which patients were diagnosed with atrial fibrillation but were not taking warfarin. Table 3 shows the results of comparing patients taking and not taking warfarin. No significant differences were found in age, number of medications, or number of comorbid conditions; the only significant difference was found in the number of men and women in the 2 groups (P = .03).
DISCUSSION
To our knowledge, this is the first study that assesses physicians’ adherence to using the CHADS2 score in a Canadian community-based setting. Our results show that a significant increase in age, number of medications, and number of comorbid conditions parallels an increase in risk of stroke (P < .05). It seems that increases in these factors correlate with stroke-risk group categorization using the CHADS2 score. There was no significant difference in the number of men and women in the 3 risk groups.
A substantial proportion (28%) of the patients sampled were taking warfarin for stroke prophylaxis despite having CHADS2 scores that put them in low- or intermediate-risk categories for which warfarin was not indicated (ie, low risk) or not unequivocally recommended (ie, intermediate risk).5,7 Approximately 21% of patients were in the intermediate-risk group, for which there was no clearly defined treatment protocol. Canadian guidelines5 and other literature7 suggest that patients at intermediate risk be prescribed either ASA or warfarin, which leaves them in a sort of gray zone. Their physicians are left to make a decision about anticoagulation therapy that might not be supported by evidence.
Because as many as 7% of patients with atrial fibrillation were taking warfarin when it was not indicated, we could conclude that physicians were not aware of or not using the guidelines around the CHADS2 score. Table 2 shows that approximately 80% of patients with atrial fibrillation were taking at least 1 type of drug for heart rate control, which is an important component of therapy for atrial fibrillation.7 When we analyzed the rates of prescriptions for heart rate control drugs in each risk group, we found them to be no different from expected frequencies (χ2 = 2.65; P = .266). While heart rate control is another component of atrial fibrillation treatment guidelines (yet is independent of the CHADS2 score), it seems that physicians are aware of treatment guidelines4 but perhaps are not following all aspects of them. The specific recommendations regarding warfarin therapy for patients with CHADS2 scores of 2 or less are not being considered or followed.
The usefulness of the CHADS2 score could be called into question. It seems that physicians are prescribing warfarin to patients with atrial fibrillation in spite of a stroke-risk calculation that does not support such treatment. Of course, the decision to initiate warfarin for these patients must be made on an individual basis,8 and several factors other than those considered in the CHADS2 schema need to be assessed. The simplicity of the CHADS2 score, as well as its validated superiority over other risk-stratification schemas,1,4 make it a favourable tool for guidelines.
The findings of our study reflect criticisms of the CHADS2 score found in other studies.1,3 Karthikeyan and Eikelboom3 noted that the CHADS2 score categorizes a substantial number of patients into the intermediate-risk group, and that further risk-factor considerations for this group have therefore been suggested.7 Karthikeyan and Eikelboom also argued that some patients in the intermediate-risk group are at the threshold for benefit from anticoagulation therapy and that such patients should not necessarily be prescribed warfarin. On the other hand, Lip1 discusses evidence that up to a third of patients in a given population of patients with atrial fibrillation have been categorized as intermediate risk and might be taking ASA therapy when anticoagulation with warfarin could be more appropriate. The ambiguity surrounding treatment of the intermediate-risk group further challenges the usefulness of the CHADS2 score.
Limitations
This study has limitations. First, many patients are started on warfarin in hospital and, once out in the community, continue treatment without further consideration of risk of stroke. Also, hospital discharge with ASA therapy might then prompt physicians to calculate risk using the CHADS2 score. Our study is not sensitive enough to assess whether the CHADS2 score is being used in the community as an initial and routine part of the approach to managing atrial fibrillation. We were unable to collect sufficient data to make an adequate comparison of patients with atrial fibrillation who were and were not prescribed warfarin across all 3 clinics. However, a general comparison of the patients in all 3 clinics suggests that those patients not taking warfarin did not have obviously different characteristics from those taking warfarin.
Conclusion
Our findings indicate that patients with atrial fibrillation are not being prescribed warfarin for stroke prophylaxis in accordance with Canadian clinical practice guidelines. The CHADS2 score is not being used or not being followed; however, the prevalence of prescriptions for drugs used to control heart rate suggests that physicians are at least considering the guidelines. The large proportion of patients at intermediate risk of stroke and the unclear guideline for the treatment of these patients might suggest that the CHADS2 score is not the best tool for assessing risk of stroke in patients with atrial fibrillation. Newly developed risk-stratification schemas1,3 should be evaluated with a view to improving clinical practice guidelines.
Acknowledgments
Support for this project was provided by a Canadian Institutes for Health Research Health Professional Student Research Award.
Notes
EDITOR’S KEY POINTS
-
The CHADS2 (congestive heart failure, hypertension, age ≥ 75, diabetes mellitus, and stroke or transient ischemic attack) score is a common, easy-to-use, evidence-based tool available to clinicians. Since its creation in 2001, CHADS2 has been shown to be superior to the other stroke-risk prediction tools previously available.
-
The goal of this study was to assess physicians’ level of adherence to clinical practice guidelines for preventing stroke among patients with atrial fibrillation.
-
This study found that 28% of the patients, all of whom were taking warfarin therapy, fell outside a category that clearly indicated warfarin as appropriate therapy (such patients were in low- and intermediate-risk groups), and that 7% of patients fell into the low-risk group, for which warfarin was not indicated.
-
The study revealed that physicians were not prescribing warfarin therapy for stroke prophylaxis to patients with atrial fibrillation in accordance with Canadian clinical practice guidelines and questioned the usefulness of the CHADS2 score.
POINTS DE REPÈRE DU RÉDACTEUR
-
Le score de CHADS2 (insuffisance cardiaque congestive, hypertension, âge ≥ 75 ans, diabète, et accident vasculaire cérébral ou ischémie cérébrale transitoire) est un outil d’utilisation facile, fondé sur des données probantes, qui est disponible aux cliniciens. Depuis sa création en 2001, on a démontré que le CHADS2 est supérieur aux prédicteurs du risque d’accident vasculaire cérébral existant auparavant.
-
Cette étude avait pour but d’évaluer le degré de conformité aux directives de pratique clinique pour la prévention des accidents vasculaires chez les patients souffrant de fibrillation auriculaire.
-
L’étude a trouvé que 28 % des patients, lesquels recevaient tous de la warfarine, n’appartenaient pas à la catégorie pour laquelle la warfarine était clairement considérée appropriée (ces patients étaient dans les groupes à risque faible ou intermédiaire), et que 7 % d’entre eux étaient dans le groupe à faible risque, pour lequel la warfarine n’était pas indiquée.
-
Les auteurs ont observé que les médecins ne suivaient pas les directives canadiennes de pratique clinique lorsqu’ils prescrivaient la warfarine en prophylaxie des accidents vasculaires cérébraux chez les patients souffrant de fibrillation auriculaire et ils ont remis en question l’utilité du score de CHADS2.
Footnotes
-
This article has been peer reviewed.
-
This article is eligible for Mainpro-M1 credits. To earn credits, go to www.cfp.ca and click on the Mainpro link.
-
Cet article a fait l’objet d’une révision par des pairs.
-
Contributors
Dr Klein initiated the project by developing the research question and hypothesis, oversaw data collection, directed the analysis, and edited versions of the manuscript. Mr Levine collected and analyzed data, contributed to interpretation of the data, and wrote the initial drafts of the manuscript.
-
Competing interests
None declared
- Copyright© the College of Family Physicians of Canada