QJM Advance Access originally published online on March 27, 2007
QJM 2007 100(5):263-269; doi:10.1093/qjmed/hcm016
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
‘Rules of thumb’ or reflective practice? Understanding senior physicians’ decision-making about anti-thrombotic usage in atrial fibrillation
From the 1Medical Department for the Elderly, The General Infirmary at Leeds, Leeds, UK and 2Elderly Services, St. James's University Hospital, Leeds, UK
Address correspondence to Dr N Anderson, Medical Department for the Elderly, The General Infirmary at Leeds, Leeds LS1 3EX
Received 14 November 2006 and in revised form 27 December 2006
 |
Summary |
|---|
 Top Summary IntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Background: The recently published Atrial Fibrillation (AF) Guidelines from the National Institute for Health and Clinical Excellence (NICE) highlight the importance of assessing stroke risk and using appropriate anti-thrombotic therapy.
Aim: To improve understanding of physicians’ behaviour and attitudes in respect to decision-making in AF and the use of anti-thrombotics.
Design: Semi-qualitative, questionnaire- and interview-based study.
Methods: Five clinical vignettes relating to treatment choices for AF and stroke prevention illustrating a range of risk and benefit were examined by 14 senior physicians (consultants or specialist registrars) in Cardiology, General Medicine and Geriatric Medicine, who then recommended anti-thrombotic treatment. A semi-structured interview explored their decision-making and prescribing in AF, with qualitative analysis of interview transcripts using grounded theory.
Results: There was marked variation in the choice of anti-thrombotic treatment. Respondents were more likely to prescribe warfarin to patients with a previous intracerebral haemorrhage than to a patient with a history of falls. A key theme on qualitative analysis revealed that decision-making in AF is often associated with uncertainty and concerns about knowledge of risk and benefit.
Discussion: In this study, doctors rarely agreed on the choice of anti-thrombotics in AF, and their perceptions of stroke and bleeding risk showed considerable variation. Uncertainty, doubt, concerns about knowledge and varied approach to the role of patients in decision-making are all significant themes in the considerable variability in anti-thrombotic prescribing.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Atrial fibrillation (AF) is the most common arrhythmic condition, with a rising prevalence affecting an estimated 2.5 million people across the USA and UK alone.1 A greater proportion of AF diagnoses are being made in older people, in part because improved medical care is increasing longevity among patients with chronic cardiac conditions predisposing to AF. The accompanying risk of ischaemic stroke means 15% of all strokes are attributable to AF, with annual stroke rates of up to 18% in some patients with non-valvular AF.2,3
Although the efficacy of anti-thrombotics for stroke risk reduction is well recognized, anticoagulation remains underused, with extensive variation in clinical practice. Research indicates that many patients who should be anticoagulated are not, with misperception of the risks and benefits associated with warfarin.4–6 NICE guidelines have brought AF into the public domain, highlighting the importance of patient-centred care and the use of evidence-based information to support treatment decisions.7 They stress use of appropriate anti-thrombotic therapy, and a stroke risk algorithm is suggested to assess thromboembolic risk prior to starting thromboprophylaxis. Physicians need to balance patient treatment preferences with guideline-directed treatment recommendations, as patient choice will often differ from what may be considered most appropriate treatment.8
Such stroke risk classification schemes permit an objective method for risk stratification in the AF population, assisting physicians in making individualized therapeutic decisions for their patients.3 The recent introduction of the ABCD score by Rothwell et al. is another improvement in individualising stroke prevention through reliable risk assessment, identifying high-risk individuals after transient ischaemic attacks.9
Why then are risks so misperceived and anti-thrombotics under-prescribed? Many AF patients present therapeutic dilemmas, posing major challenges of balancing risk and benefit, especially in those with a history of falls or a previous intracerebral haemorrhage (ICH). Physician fears about such problems may be exaggerated and unfounded.10,11 Research suggests that clinical decisions are often made using heuristics (‘rules of thumb’), using relatively small amounts of information, leading to suboptimal decision-making.12,13 In AF, the heuristic of avoiding haemorrhage as a consequence of treatment risks overlooking the risk of ischaemic stroke. Research has also suggested that many clinicians overestimate the risk of subdural haemorrhages in patients on warfarin who are prone to falling.14 A possible explanation for this is that a fall-related subdural haemorrhage in an elderly patient is a rare event, but one that a clinician will easily remember. In an attempt to minimize such assumed risks, physicians opt against prescribing warfarin for patients with a history of falls, even though this risk is heavily outweighed by the potential benefits of stroke risk reduction provided by warfarin.
Improving our understanding of decision making in this domain is important, given the current ‘epidemic’ of AF. This study aims to move beyond the existing literature on AF decision-making, to examine why risks are misperceived, and how clinicians cope with risk and benefit dilemmas. A qualitative, grounded theory approach was used in an attempt to understand physicians’ behaviours and attitudes in respect to decision-making about AF and the use of anti-thrombotics.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Participants
Senior physicians (defined as either Consultants or Specialist Registrars (SpRs) in this study) in Cardiology, General Medicine and Geriatric Medicine were invited by letter to take part in this study. All were practicing clinicians regularly making decisions about anti-thrombotic prescribing in AF, and worked at St. James's University Hospital in Leeds, a large UK teaching hospital that undertakes both secondary and tertiary care.
Data collection
Clinical vignettes were designed, that related to treatment choices for AF and stroke prevention, to act as a starting point for subsequent discussion. The vignettes provided varying AF-related stroke risks from 1.9% to 18.2% per annum, calculated using the CHADS2 score, which provides a range of accurate predictions for ischaemic stroke due to AF (Table 1).
|
The vignettes were presented as hard copy and included: (1) A young patient with lone AF, but with anxieties about stroke. (2,3) Two scenarios of patients with AF, varying vascular risks and ICH. (4) An older person with AF, multiple vascular risks and recurrent falls. (5) A patient with a prosthetic mechanical heart valve and previous ICH.
Here is an example of the information included (Vignette 4): ‘A 78-year-old lady is referred to see you in out-patients with a history of recurrent falls. She has recently been an in-patient with impaired diabetic control on the background of heart failure, a previous TIA and hypertension. Her GP has arranged an ECG which has shown AF, and he has commenced her on Digoxin, which she takes reliably along with her antihypertensives and heart failure medicines. She attends clinic with her daughter who is concerned about the falls. Would you consider anti-thrombotic treatment and if so, which?'
Participants examined the vignettes and were then asked to decide what anti-thrombotic treatment they would recommend for each of the cases. Participants were asked to explain the reasoning behind their treatment choice, and encouraged to explain why they had decided against the other treatment choices.
All participants were interviewed by the same investigator (NA), typically lasting 30 min. This semi-structured interview explored decision-making and prescribing in atrial fibrillation, using the vignettes as a reference point.
|
Data analysis |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Audiotaped interviews with participants were transcribed verbatim for analysis. Transcripts were examined in detail using grounded theory by two investigators (NA and RF), which mirrors much of the inductive nature of qualitative research.15 This permits a systematic but flexible way of exploring processes of how belief and experience may alter individuals’ behaviour, developing concepts and themes which are ‘grounded’ in the data. As many themes overlap in context, they were gathered into broader ‘concept clusters’ (Table 3), previously used to analyse decision processes, allowing categorization of broad concepts.16 Transcripts were read and re-read multiple times to demonstrate emergent themes, which were then clustered. There was almost complete agreement in the initial themes generated when the manuscripts were reviewed by the two investigators. As part of grounded theory analysis, these concept clusters were read alongside contemporary AF and Risk/Decision literature to permit a richer approach to the data analysis.15
|
Rigour and trustworthiness were ensured at all stages of the research process, using open and closed questioning to reveal and explore unexpected themes, by the use of methods triangulation.17 Further triangulation was conducted by reading emergent themes alongside contemporary AF and risk literature. Theoretical saturation was rapidly reached in the interviews, with principal clusters revealed in interviews with the first eight respondents. Rigour was further enhanced using theoretical verification, where results were presented at meetings involving study participants, verifying that the study findings clearly represented their attitudes to thromboprophylaxis decisions in AF.
|
Results |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Participants
Of 20 clinicians invited, 14 (70%) agreed to participate. Five were cardiologists (2 Consultants, 3 SpRs) and nine worked in geriatric or general medicine (4 Consultants, 5 SpRs).
Physician decisions in response to the vignettes were recorded, and are detailed in Table 2.
|
Decision vignettes
There was marked variation in the choice of anti-thrombotic treatment, with physicians’ agreement in prescribing decisions ranging from 43% to 100% (Table 2). 100% agreement was only reached for Vignette 4 (AF, multiple risk factors (CHADS2 score 6) and falls), with no respondents choosing to use warfarin despite an annual stroke risk of 18.2%. This contrasts with the Vignette 5 (recent intracerebral haemorrhage, AF and a prosthetic valve), where eleven (79%) physicians would anticoagulate with warfarin, despite an identical stroke risk to that in the falls scenario.
Similar variation was seen for Vignettes 2 and 3 (AF but previous ICH), where little or no evidence-based practice exists to guide clinicians. In Vignette 2, eight physicians (57%) chose to use aspirin or warfarin in a patient with AF, ICH and an annual stroke risk of 8.5%, compared with only two (14%) in Vignette 3 (annual risk 12.5%). The study was not powered to look for differences between grades or specialty of physician, but the following qualitative analysis revealed key themes relating to some of the challenges of making decisions about anti-thrombotics in atrial fibrillation. Selected excerpts of interview text are presented below to illustrate these concepts.
|
Qualitative analysis: concept clusters (Table 3) |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Discourses of risk and benefit
This was a key theme in the content of physicians’ narratives when attempting to make decisions about anti-thrombotics, a concept closely tied to uncertainty (next section). Rather than simply analysing vignettes and reaching decisions, respondents sought additional information, using language to qualify carefully and justify before reaching a conclusion about treatments. These narratives expressed caution and often doubt, and were a feature of even simple vignettes such as lone AF, as in this narrative:
‘Erm, you’d want to know a little bit more about whether he's got high blood pressure or what his cholesterol is like. I would probably do his blood pressure and cholesterol, do an echocardiogram and if there isn’t any contraindication, give him aspirin. I wouldn’t start him on warfarin unless the Echo came back as showing LVH or a dilated left atrium, or he was hypertensive and high cholesterol.’
Whilst the risk literature points to doctors as heuristic decision-makers, virtually all respondents in this study exhibited considerable reflection when making decisions. The falls vignette (Vignette 4) was a typical example, with all physicians trying to consider questions around anticoagulation and wider issues such as addressing falls risk, patients’ perceptions and then revisiting the AF questions. One physician's view typified the concerns about warfarin and falls:
‘Ideally you would want to treat this lady with warfarin but in view of the recurrent falls and the subsequent risk of life-threatening haemorrhage l would opt for the lesser anti-thrombotic of either aspirin or clopidogrel’.
In one of many such narratives (AF and ICH), heuristic ‘instinct’ gives way to thoughtfulness, but again with uncertainty:
‘... my instinct says certainly not warfarin for now, and I'd be erring on the side of nothing until I'd found out exactly what was going on with the primary ICH and take more advice.’
Discourses of uncertainty
As previous concepts indicate, even ‘simple’ decisions about AF and anti-thrombotics posed significant challenges to physicians, and were often accompanied by uncertainty. As part of the analysis, narratives were scored as ‘certain’ or ‘uncertain’ for each scenario depending on the language used and (with the exception of the falls scenario), ‘certainty’ was only really expressed by <20% of physicians in any one vignette. Even for those physicians who were more ‘certain’ in their narratives, treatment decisions revealed inconsistency when read across all scenarios, with often incorrect decision-making (e.g. prescribing warfarin for a young patient in lone AF). Deeper analysis illustrates how easily swayed some of these ‘certain’ decision-makers are. In the lone AF scenario, one senior physician comments:
‘Strictly speaking this chap could get by without any anticoagulation but in view of his anxiety ... I think I would go ahead and offer him long term warfarinization’..
Doubt and anxiety about difficult decisions were also clearly expressed, particularly in relation to warfarin:
‘I’d be a bit scared to put him on warfarin’.
Sharing decisions and responsibility
Patient choice and involvement in decision-making was another key theme, with all fourteen (100%) of respondents outlining this in their narratives. Related to this was the strong focus of using risk and benefit language to outline risks and facilitate discussions about anti-thrombotic treatment. Patients were viewed as a key part of the decision process, but analysis revealed a range of views as to the extent to which physicians would allow patients to be the ‘final’ decision-maker. For physicians that tended to be ‘certain’ in their decisions, they were clear about the extent to which they would involve patients, even for challenging cases, such as scenario 3 (AF and ICH):
‘(I) would review the situation in the out-patient clinic, weighing up the risks and the benefits and discussing these with the patient. My own personal view I would probably avoid warfarin lifelong so the option would either be aspirin or clopidogrel in the future.’
More commonly, patients were involved considerably more in decision-making, with subtle shifts in responsibility for making decisions, especially when uncertainty prevailed. Common to many, this was apparent even for relatively simple scenarios such as lone AF:
‘I would almost put the decision or the ball in his court and I would go down the lines of describing to him his absolute and relative risk reductions with aspirin and warfarin ... and I'd see what he'd prefer to do’.
For more complex decisions, the narratives suggested a small proportion of respondents would be content to exercise patient-centred decision making to a point where the physician appeared to abdicate responsibility:
‘The chances of anything happening to him if it's simple AF are pretty low. I couldn’t tell you the figures (lone AF scenario), but he might be prepared to accept that as a risk. I think the ultimate decision would lie with him’.
Availability of risk information
The final concept cluster harnesses overlapping themes of knowledge, responsibility and links with the use of other specialists to make decisions. Physicians in this study did not appear to be very heuristic decision-makers, reflecting on the extent of their own knowledge about AF, stroke risks and anti-thrombotics. Virtually all physicians expressed that many of their views on AF/stroke risk were experiential, and how risks might be misperceived. One physician summed this point well:
‘If you do have the hard figures down in front of you it might sway you one way or another. And I think my impression from doing this exercise is that we underestimate the risk of ischaemic stroke and overestimate the risk of bleeding’.
Others were more open, expressing their discomfort at making decisions based on poor understanding of risks:
‘If he had a metallic valve ... I don’t think I’d be comfortable to make that decision. I don’t know what the risk of having a stroke is ... I wouldn’t discount him being on warfarin, but only because it's a kind of imprint—but it's not anything to do with what I think the risks and benefits would be’.
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Despite the proven efficacy of anti-thrombotics in reducing AF-related stroke risk, studies continue to show under-use of anticoagulants, particularly in the highest-risk groups.18,19 There are continued calls for greater use of anticoagulants in AF, but by overlooking the attitudes that underpin decision-making in this area of practice, such calls are unlikely to make significant changes. Our results show similarities with previous work,19,20 demonstrating misperceptions about AF-related stroke and treatment-related bleeding risks, but providing a deeper analysis of attitudes and behaviours. We have shown considerable heterogeneity in decision-making and prescribing practice.
The extensive variability in the decision vignettes (Table 2) merits particular comment, as the themes described in this study are major contributors to decision-making. Even for the scenario of lone AF in a younger patient, it is important to note that only ten (71%) of the respondents agreed on a treatment choice, with others recommending anticoagulation. Clinicians’ fears about falling were expressed in the 100% agreement not to anticoagulate the patient in the scenario because of the perceived risk of ICH, despite her significant annual stroke risk of 18%—a misperception previously disproven.14 Our initial assumption was that most physicians would avoid any anti-thrombotics in patients with previous intracerebral haemorrhage, yet 57% of respondents would prescribe anti-thrombotics in a patient with lower annual stroke risk and ICH, in contrast to the fall scenario. This appears to be a clear signal that there continues to be misperceived and incomplete knowledge about stroke and bleeding risks, despite available decision-modelling evidence.
The heterogeneity of attitudes in this study is important, illustrating the prevalence of contributors to decisions such as uncertainty and doubt. Seen in this study among physicians with different seniority, experience and from different specialties, it has also been reported in earlier work on how junior doctors approach risk.16 However, this heterogeneity, and the emergent concept clusters in this study, challenge many of the assumptions of how doctors reach clinical decisions. The presumption of the doctor as a ‘fast and furious’ heuristic decision-maker providing knowledge-based, justifiable decisions (where decisions are justifiable and knowledge based) remains popular in the literature.21,22 Our results with senior physicians show that decisions about AF prescribing are uncertain and sometimes uncomfortable, and accompanied by imprecise risk knowledge. While physicians acknowledge their own uncertainties about risk, these tend to be carefully hidden in discussion with patients, producing what some have coined the ‘illusion of certainty’.23 Previous work suggests that junior doctors experience extensive discomfort when discussing risk,16 but the experienced physicians in this study appear far more willing to discuss risks and decisions, avoiding some of the conflict seen in junior doctors’ approaches.
Regarding study limitations, while vignettes were a useful starting point for this study, they can never encompass all information available to a clinician when reaching a decision, especially if the vignette is complex. This may have been partly responsible for the cautious approach seen in some physicians, and the use of language to qualify and justify their decisions. Generalizability is always a concern with qualitative research methods, and our findings should be interpreted in context. However, the rigour of the study merits comment, and the grounded theory approach allowed us to ‘read’ the study findings alongside contemporary literature, thereby triangulating results of this study with established work. Further validity was ensured through user verification, and the rapid theoretical saturation and consistent themes that emerged from a varied group of clinicians.
Although the need to improve the rates of anti-thrombotic usage in AF is clear, this study has shown how uncertainty and complexity in cases will always lead to variation in prescribing among doctors. The NICE Guidelines may have helped to bring AF into the public domain, and they offer clinicians a structured approach to AF, but these guidelines cannot assist clinicians with the more complex cases, which involve careful balancing of risks and benefits. This study challenges the assumption that calls for greater anticoagulation usage in AF will in fact deliver such change; perhaps we should be more willing to accept that the inherent uncertainty in medical practice means we may never overcome some of the difficulties that contribute to the perceived under-use of drugs such as warfarin. Patient choice also seems to influence the extent of anticoagulation use, and evidence suggests that individual patient preferences will often differ from the recommendations suggested by guidelines.8
What does this mean for busy clinicians, and can we help support decision-making and accessibility of risk information? Physicians should continue to practice reflectively by acknowledging the uncertainty and doubt in AF prescribing, and carefully seeking the views of patients about health beliefs and risks. Previous research has suggested that patient involvement in decision-making about anti-thrombotic treatment could be enhanced through the use of decision aids.24 Results show that patients using decision aids are more knowledgeable, believe they are more informed about the important clinical issues, and have more realistic expectations about the probability of possible outcomes. Nevertheless, patients in these studies report that their physicians have an important role in the decision-making process, suggesting that decision aids are a support rather than a replacement to the patient-physician relationship.
Increasing numbers of patients will be questioning their doctors about thromboprophylaxis treatment, expecting explanations of risks and benefits of treatment options. The supporting leaflet published by NICE for patients with AF describes the treatment options available, and that patients can expect to be offered.25 In doing so, it is likely that more informed choices for physicians and patients will contribute more to appropriate anti-thrombotic prescribing than blanket calls to increase anticoagulant usage in atrial fibrillation.
|
Acknowledgements |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
Ethics committee approval was secured for the study from the Leeds (East) Research Ethics Committee at St James's University Hospital, Leeds. We have no conflict of interest to declare. This study received no sponsorship or funding. We have all contributed to the formulation of study design, data interpretation and preparation of all parts of this manuscript. Acquisition of subjects and interviews was by Dr Anderson, and qualitative analysis by Dr Fuller.
|
References |
|---|
|
TopSummaryIntroductionMethodsData analysisResultsQualitative analysis: concept...DiscussionAcknowledgementsReferences |
|---|
1. Steinberg JS. (2004) Atrial Fibrillation: an emerging epidemic? Heart 90 239–40.
2. Wolf PA, Abbott RD, Kannel WB. (1991) Atrial Fibrillation as an independent risk factor for stroke: the Framingham study. Stroke 22 983–8.
3. Gage BF, Waterman AB, Shannon W, et al. (2001) Validation of Clinical Classification Schemes for Predicting Stroke: Results from the National Registry of Atrial Fibrillation. JAMA 285 2864–70.
4. Whittle J, Whickenheiser L, Venditti LN. (1997) Is warfarin underused in the treatment of elderly persons with atrial fibrillation? Arch Intern Med 1197 441–5.
5. Brass LM, Krumholz JM, Scinto JM, et al. (1997) Warfarin use among patients with atrial fibrillation. Stroke 28 2382–9.
6. Albers GW, Yim JM, Bittar N, et al. (1996) Status of antithrombotic therapy for patients with atrial fibrillation in university hospitals. Arch Intern Med 156 2311–16.[Abstract]
7. National Institute of Clinical Excellence. (2006) Atrial Fibrillation NICE Guideline [http://www.nice.org.uk/download.aspx?o=336576]. Accessed 19 September 2006.
8. Man-Son-Hing M, Gage BF, Montgomery AA, et al. (2005) Preference-Based Antithrombotic Therapy in Atrial Fibrillation: Implications for Decision Making. Medical Decision Making 25 548–59.
9. Rothwell PM, Giles MF, Flossman E, et al. (2005) A simple score (ABCD) to identify individuals at high early risk of stroke after transient ischaemic attack. Lancet 366 29–37.[CrossRef][ISI][Medline]
10. Man-Son-Hing M and Laupacis A. (2003) Anticoagulant-Related Bleeding in Older Persons With Atrial Fibrillation. Arch Intern Med 163 1580–6.
11. Bond AJ, Molnar FJ, Li M, Mackey M, Man-Son-Hing M. (2005) The risk of haemorrhagic complications in hospital in-patients who fall while receiving antithrombotic therapy. Thrombosis J 3 1.[CrossRef]
12. Elwyn G, Edwards A, Eccles M, et al. (2001) Decision analysis in patient care. Lancet 358 571–4.[CrossRef][ISI][Medline]
13. In Gilovitch T, Griffin D, Kahneman D (Eds.). Heuristics and biases: The psychology of intuitive judgement. (2002) Cambridge Cambridge University Press.
14. Man-Son-Hing M, Nichol G, Lau A, et al. (1999) Choosing antithrombotic therapy for elderly patients with atrial fibrillation who are at risk for falls. Arch Intern Med 159 677–85.
15. Strauss A and Corbin J. (1990) Basics of Qualitative Research: Grounded Theory Procedures and Techniques.California Sage.
16. Fuller R. (2004) Managing Health Risks: Junior Doctors’ views of risk and decision-making. J Health Organ Manag 18 155–78.[CrossRef][Medline]
17. Rice PL and Ezzy D. (1999) Qualitative Research Methods: A Health Focus.Oxford Oxford University Press.
18. Lip GY, Zarifis RD, Watson RD, et al. (1996) Physician variation in the management of patients with atrial fibrillation. Heart 75 200–5.
19. Monette J, Gurwitz JH, Rochon PA, et al. (1999) Physician attitudes concerning warfarin for stroke prevention in atrial fibrillation: results of a survey of long-term care practitioners. J Am Geriatr Soc 45 1060–5.
20. Devereaux PJ, Anderson DR, Gardner MJ, et al. (2001) Differences between perspectives of physicians and patients on anticoagulation in patients with atrial fibrillation: observational study. Br Med J 323 1218–25.
21. Sarasin FP. (1999) Decision analysis and the implementation of evidence-based medicine. Q J Med 92 669–71.[ISI]
22. Schattner A and Fletcher RH. (2003) Research evidence and the individual patient. Q J Med 96 1–5.[ISI]
23. Gigerenzer G. (2002) Reckoning with Risk: Learning to live with uncertainty.London Allen Lane.
24. Man-Son-Hing M, Laupacis A, O'Connor A, et al. (1999) A Patient Decision Aid Regarding Antithrombotic Therapy for Stroke Prevention in Atrial Fibrillation: A Randomized Controlled Trial. JAMA 282 737–43.
25. National Institute of Clinical Excellence. Atrial Fibrillation—Information for the public. [http://www.nice.org.uk/page.aspx?o=cg036publicinfo] Accessed 19 September 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||