Atrial fibrillation: diagnosis and management—summary of NICE guidance
BMJ 2021; 373 doi: https://doi.org/10.1136/bmj.n1150 (Published 21 May 2021) Cite this as: BMJ 2021;373:n1150
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Dear Editors
Furthermore, I am unable to determine if NICE was able to tease out ischaemic stroke from the “stroke/thromboembolism (TE)” composite outcome which is evident from many of the studies they were basing their recommendations on, considering that in many cases TE includes not just ischaemic stroke, but also pulmonary embolism, peripheral embolism. (Ref 3)
Therefore how this tool to predict stroke or TE as reflected in the evidence base somehow became the basis to predict stroke (without other TE) in the main guidelines was not adequately explained; this could represent a major lapse or leap of faith in translating the evidence which relies in part (if not in full) stroke/TE composite in their original analysis.
In the evidence base for the NICE guideline (Ref 3), NICE recognised the ideal risk stratification tool for predicting stroke or thromboembolic events in people with atrial fibrillation “would have high sensitivity but also have enough specificity to allow the people with lowest risk to avoid unnecessary anticoagulation, with the excess risk that would entail”.
This is not congruent with the conduct of the analysis, which appears to be more focused on case findings to treat rather than avoiding unnecessary treatment.
A reflection of this outlook is as written in the main guidelines document as above which is repeated here for emphasis:
“It also means that 77% of people who would not later have a stroke (without anticoagulation) would be wrongly identified as needing anticoagulation. However, this was thought to be acceptable given the perceived lesser harms from unnecessarily giving anticoagulants compared with not giving anticoagulants to people who need them, together with the inevitable trade-off between sensitivity and specificity.”
Another example is their justification of choosing CHA2DS2-VASc over ATRIA on the basis of the former’s better sensitivity which appears to trump better specificity (which would have reduced overtreating people who are not at risk of stroke/TE)
“The ATRIA stroke risk score was shown to have better overall accuracy, but although it had better specificity than CHA2DS2-VASc (fewer false-positive results) it had lower sensitivity, meaning that more people at risk would be missed (more false-negative results) compared with the CHA2DS2-VASc score. As already suggested, sensitivity was agreed by the committee to be more important than specificity because the risks of unnecessary anticoagulation are outweighed by the risks of not treating people who need anticoagulation. In addition, the ATRIA risk score may result in a time delay in calculating the results.”(Ref 1)
“The decision of the committee was therefore that CHADSVASC was slightly more useful because of its better ability to ensure that people truly at risk of stroke were anticoagulated.”(Ref 3).
So what is this concern about a time delay in calculating the results?
“In addition to ATRIA having potentially more harms than CHADSVASC in terms of ATRIA leading to more people at risk of stroke not being anticoagulated, the ATRIA was also believed to be more difficult to use. The committee discussed the time delays in getting a dip-stick assessment of proteinuria done and retesting eGFR for the ATRIA, although it was pointed out that ATRIA might, on occasions, be able to utilise data already in the patients’ notes rather than requiring the acquisition of new data.” (Ref 3)
So, NICE feels that the delays (or inconvenience) from getting a simple bedside/clinic test of doing a urine dipstick, AND getting an eGFR which is nowadays supplied with standard electrolyte/urea/creatinine (or chem 7 in the US), “on occasion” (or rather in most occasions for the majority of AF patients, as the choice of DOACs is often influenced by renal function), pretty much condemned many a person to lifelong, often inconvenient, throughly expensive (if using DOACs), sometimes deadly (from bleeding) anticoagulation, which for a significant number of them would have been totally unnecessary and avoidable if the right matrix and outlook for them had been used.
Again I reiterate, NICE did not appear to set a bar on the minimum event rate of stroke/TE, or an absolute number of stroke events per 100,000 people at risk per year, or a threshold of numbers of stroke in UK in beyond which it is unacceptable. They seemed to be more fascinated or obsessed by the trick of c-statistics on sensitivity and specificity; again the need for case finding gazumped the need to avoid unnecessary treatment or harm.
This may be what some of my colleagues’ expectation of what medicine is now, but I am certain this perspective is not shared by an informed patient when she/he finds out how the recommendations are really made.
References:
1. https://www.nice.org.uk/guidance/ng196/resources/atrial-fibrillation-dia...
2. Lip GYH, Nieuwlaat R, Pisters R, Lane DA, Crijns HJGM. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–72.
3. https://www.nice.org.uk/guidance/ng196/evidence/c-and-d-tools-to-predict...
Competing interests: No competing interests
Dear Editors
I would like to share with fellow readers the rationale as proposed by NICE to recommend using the CHA2DS2-VASc stroke risk score to assess stroke risk in people with symptomatic or asymptomatic paroxysmal, persistent/permanent atrial fibrillation, atrial flutter or continuing risk of arrhythmia recurrence after cardioversion back to sinus rhythm or catheter ablation.
They wrote:
“The committee decided to prioritise identifying people above or below a certain risk threshold (discrimination) in its interpretation of the evidence rather than estimating a person's risk of stroke in absolute terms.
The evidence suggested that a score of 2 or more is the ideal threshold for the CHA2DS2-VASC in terms of indicating the need for anticoagulation. (Men with a CHA2DS2-VASc score of 1 were regarded as being at intermediate risk, and a group in whom anticoagulation should also be considered.) The evidence showed that this threshold of 2 or more offered a good combination of high sensitivity (0.92) and adequate specificity (0.23).
The high sensitivity means that the tool would correctly identify almost everyone who would later have a stroke if they did not receive anticoagulants. Importantly, this will allow them to be prescribed anticoagulants to reduce their risk of stroke.
The adequate specificity means that 23% of the people who would not later have a stroke (even when not taking anticoagulants) would be correctly identified as not needing anticoagulation. This would prevent these people from having adverse events from anticoagulants. It also means that 77% of people who would not later have a stroke (without anticoagulation) would be wrongly identified as needing anticoagulation. However, this was thought to be acceptable given the perceived lesser harms from unnecessarily giving anticoagulants compared with not giving anticoagulants to people who need them, together with the inevitable trade-off between sensitivity and specificity.
The ATRIA stroke risk score was shown to have better overall accuracy, but although it had better specificity than CHA2DS2-VASc (fewer false-positive results) it had lower sensitivity, meaning that more people at risk would be missed (more false-negative results) compared with the CHA2DS2-VASc score. As already suggested, sensitivity was agreed by the committee to be more important than specificity because the risks of unnecessary anticoagulation are outweighed by the risks of not treating people who need anticoagulation. In addition, the ATRIA risk score may result in a time delay in calculating the results.
The committee also discussed that the evidence for the QStroke risk calculator suggested that it might be a useful tool. However, the evidence was limited, and they agreed that further research was needed.” (Ref 1)
Remarkably this meant their risk threshold is based on the CHA2DS2-VASC score of 2 or more, rather than an absolute risk of stroke annually for each calculated score.
It is important to remember that CHA2DS2-VASc recommended by Lip et al (ref 2) is as follows:
“We refined the 2006 Birmingham (or NICE) TE risk schema into a risk factor-based approach, by defining definitive risk factors (previous stroke/TIA/TE and age 75 years) and combination risk factors (heart failure/moderate-severe cardiac dysfunction, hypertension, diabetes, vascular disease, female gender, and age 65-74 years). If we wished to artificially categorize these subjects, high risk was defined as one definitive or two or more combination risk factors, intermediate risk was essentially defined as one combination risk factor, and low risk was defined as no risk factors being present. This refined (2009) Birmingham schema was also tested with a point-based scoring system, the CHA2 DS2 -VASc score (see Table 2 for definition), whereby scores of 0 = low, 1 = intermediate, and >_2 = high risk.”
Hence their original definition of risk had no actual bearing on any event rate threshold, rather their perceived (weighted) number of additional risks. Later in the paper they seemed to benchmark true success of low-risk (score of 0) classification as no TE events recorded.
References:
1. https://www.nice.org.uk/guidance/ng196/resources/atrial-fibrillation-dia...
2. Lip GYH, Nieuwlaat R, Pisters R, Lane DA, Crijns HJGM. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–72.
3. https://www.nice.org.uk/guidance/ng196/evidence/c-and-d-tools-to-predict...
Competing interests: No competing interests
Dear Editor
We read with great interest the summary of NICE guidance by Perry et al. [1]. The authors have summarised the guidance concisely, making it easy for readers to enhance their understanding and management of atrial fibrillation (AF).
We would like to add some points which haven’t been covered in the article.
Firstly, some patients who develop atrial fibrillation might already be on anti-platelets for secondary prevention of cardiovascular complications. In patients who develop atrial fibrillation, the antiplatelet drug should be discontinued upon commencement of anticoagulation therapy. One exception to this is post-cardiac stenting. In these patients, anti-platelet should be continued for up to 12 months along with an oral anti-coagulant and should be done after discussion with the cardiology team to ensure multi disciplinary input [2]. In our experience, we have found that antiplatelet agents are often not discontinued on initiation of anti-coagulation, and this increases the risk of intracranial and extra-cranial bleedings.
Steroids, like NSAIDs, also increase the bleeding tendency when given concomitantly with anticoagulants, and taking this into account, is important to decide the risk-benefit ratio [3].
Patients on rivaroxaban should be advised to take the drug with a meal, as failing to do so can lead to reduced bioavailability thereby reducing the efficacy of the drug on prevention of stroke [4]. We, therefore, recommend that this advice should be given to patients upon commencement of rivaroxaban.
Blood pressure control is essential before initiation of direct oral anticoagulant (DOAC); otherwise, it increases the risk of intracranial bleed. All other modifiable risk factors of cardiovascular disease like diabetes, high cholesterol, smoking, weight reduction, alcohol intake not within recommended levels, and screening for sleep apnoea should be undertaken during follow-up either in primary or secondary care [5].
In cases of cardio-embolic stroke due to occult AF, patients should be screened for paroxysmal AF in a step-wise manner, as done in our hospital to ensure its detection and treatment with a DOAC to prevent further strokes [6].
Lastly, stopping anticoagulation in patients with a risk of falls remains a dilemma for most clinicians. We believe patients should be taken into confidence and the risk-benefit ratio be discussed with them. Several studies indicate that stopping anti-coagulation after a fall is not warranted. A patient must have 295 falls in a year, for Warfarin not to be considered as optimal therapy [8]. Furthermore, NICE recommends not stopping anticoagulation in patients with atrial fibrillation who are at risk of falls [7].
REFERENCES
1. Perry M, Kemmis Betty S, Downes N, Andrews N, Mackenzie S. Atrial fibrillation: diagnosis and management—summary of NICE guidance BMJ 2021; 373:n1150 doi:10.1136/bmj.n1150.
2. National Institute for Health and Care Excellence. (2020). Acute Coronary Syndromes [NICE Guideline No. 185]. https://www.nice.org.uk/guidance/ng185
3. Cheung KS, Leung WK. Gastrointestinal bleeding in patients on novel oral anticoagulants: Risk, prevention and management. World J Gastroenterol. 2017;23(11):1954-1963. doi:10.3748/wjg.v23.i11.1954.
4. Mueck W, Stampfuss J, Kubitza D, Becka M. Clinical pharmacokinetic and pharmacodynamic profile of rivaroxaban. Clin Pharmacokinet. 2014;53(1):1-16. doi:10.1007/s40262-013-0100-7.
5. Lip GYH. The ABC pathway: an integrated approach to improve AF management. Nat Rev Cardiol. 2017; 14:627–628.
6. Kausar SA. The latest national clinical guideline for stroke. Clinical Medicine Aug 2017, 17 (4) 382-383; DOI: 10.7861/clinmedicine.17-4-382.
7. National Institute for Health and Care Excellence. (2021). Atrial Fibrillation: diagnosis and management [NICE Guideline No. 196]. https://www.nice.org.uk/guidance/ng196
8. Man-Son-Hing M, Nichol G, Lau A, Laupacis A. Choosing Antithrombotic Therapy for Elderly Patients With Atrial Fibrillation Who Are at Risk for Falls. Arch Intern Med. 1999;159(7):677–685. doi:10.1001/archinte.159.7.677
Authors
Dr. Mohammad Abdullah
Stroke Registrar
Dr. Muhammad Yasir Rafiq
Stroke Registrar
Dr. Shahid A Kausar
Consultant in Geriatrics, Stroke & Internal Medicine
Department of Stroke Medicine, Russells Hall Hospital, Dudley, DY1 2HQ
Competing interests: No competing interests
Dear Editor,
Thanks for this article, which provides a good summary of the latest NICE guidance re atrial fibrillation. However, readers should be aware that there is yet to be a SnomedCT code allocated to the ORBIT score, despite a request being submitted on 28/1/21 (and further submissions 1/4/21 and 18/5/21). Consequently, it cannot be added to GP systems.
Yours faithfully
J C A Derrick MB ChB FRCS(Ed)
Competing interests: No competing interests
The three selectors for deciding if a patient with atrial fibrillation needs oral anticoagulation based on their sex and CHA2DS2-VASc ( CV ) score in the clinical flow chart contain multiple errors
Dear Editor
The decision box labelled “Stroke prevention” in the flow chart in Figure 1 should provide three mutually exclusive CV integer value selections (moving left to right ) of >= 2, 1, and 0, respectively. Unfortunately the authors' decision box contains the values >=2, 1, and both 0 and 1, respectively which does not make sense.
I now describe the errors in these three selectors ( labelled A, B and C – moving left to right ) displayed in the box named “Stroke prevention” in the original flow chart in Figure 1 and how they can be fixed.
These explanations are shown in the graphic in my Figure 1.
http://acolman.co.uk/BMJ_Rapid_Response_Figures/ChadsVasc_Algorithm.pdf
Selection A – "CHA2DS2-VASc >= 2 - offer oral anticoagulant”
Error A – Explanation - This selection is for men and women with a CV score of >=2 only.
The "men" and "women" sex descriptors are missing here. It is presumed that this CV score refers to both men and women but this has to be explicitly stated in the logic. The management advice is the same for both men and women.
Error A – Solution - A statement indicating that “both men and women with a CV score of >=2 should be offered an oral anticoagulant”..
Selection B - "CHA2DS2-VASc = 1 in men – consider oral anticoagulant”
Error B - Explanation - This selection is for men and women with a CV score of 1 only.
There is only a management advice for men with a CV score of 1.
The statement for women with a CV score of 1 is missing.
It belongs here rather than in Section C.
Error B – Solution - A statement indicating the management of both men and women with a CV score of 1. The management for men and women is different.
For men – “consider an oral anticoagulant”.
For women – “do not offer an oral anticoagulant”.
Selection C - "CHA2DS2-VASc <= 1 in women or CHA2DS2-VASc = 0 in men – do not offer an oral anticoagulant”
Error C - Explanation -This selection is for men and women with a CV score of 0 only.
There is management advice that “men with a CV score of 0 should not be offered an oral anticoagulant”.
There is also management advice that “women with a CV score of 0 or 1 should not be offered an oral anticoagulant”. The management of a woman with a CV score of 1 does not belong here, it belongs to Selection B. The use of the Boolean operator “OR” and equation CV <=1 is unnecessary.
Error C – Solution - A statement indicating that “both men and women with a CV score = 0 should not be offered an oral anticoagulant”.
An alternative flow chart design for anticoagulant treatment advice based on sex and the CV score is presented in my
Figure 2. http://acolman.co.uk/BMJ_Rapid_Response_Figures/ChadsVasc_Algorithm.pdf
Summaries of national clinical guidelines like this one enable the busy clinician to extract useful clinical information from them more quickly than using the larger and more complex original guideline. However, this aim will not be achieved if they introduce errors in medical decision making not present in the original guideline. As these summaries, including their flow charts, are widely referenced and reproduced in future medical publications, these errors will be propagated through numerous publications until they are recognised.
A clinical flow chart is a piece of highly-structured information engineering and its accurate production requires the skills of a competent computer programmer.[1]
References
[1] Colman A, Richards B.
Clinical Algorithms: purpose, content, rules, and benefits.
International Journal on Biomedicine and Healthcare. 2014 ( 2 ), 28 –40.
http://www.ijbh.org/ijbh2014-2.pdf ( accessed August 2021 )
Competing interests: No competing interests