Putting risk prediction in atrial fibrillation into perspective
Author(s) -
Jonathan P. Piccini,
Daniel E. Singer
Publication year - 2012
Publication title -
european heart journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 4.336
H-Index - 293
eISSN - 1522-9645
pISSN - 0195-668X
DOI - 10.1093/eurheartj/ehs031
Subject(s) - medicine , atrial fibrillation , framingham risk score , stroke (engine) , cardiology , cha2ds2–vasc score , statistic , risk assessment , warfarin , clinical prediction rule , cohort , disease , ischemic stroke , statistics , ischemia , mechanical engineering , mathematics , computer science , computer security , engineering
This editorial refers to ‘Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study’, by L. Friberg et al. , doi:10.1093/eurheartj/ehr488 Oral anticoagulation dramatically reduces the risk of ischaemic stroke and improves all-cause survival in patients with atrial fibrillation (AF). However, anticoagulation increases the risk of major bleeding, including uncommon but frequently fatal intracranial haemorrhages. As a consequence, guidelines recommend a risk-based approach to anticoagulation for AF, assuming those patients at higher untreated risk of stroke will gain sufficient absolute risk reduction to outweigh potential harms.1The most widely used risk assessment tool in AF is the CHADS2 score, which is based on risk factors identified in the early trials of warfarin in non-valvular AF.1 The CHADS2 acronym is easy to remember and has become a fixture of guidelines, clinical practice, and trial design. Unfortunately, its predictive ability is mediocre. The standard assessment of the discriminating ability of a prediction rule is the C-statistic, which is roughly interpretable as the probability the score will be higher in patients who sustain a stroke vs. those who do not (the interpretation varies depending on analytic perspective). A C-statistic of 0.5 indicates that the score conveys no predictive information. In contrast, the gold standard of chronic disease cardiovascular risk prediction, the Framingham score, has demonstrated C-statistics approaching 0.8.2 Typical C-statistics for the CHADS2 score have been in the range of 0.60. Several alternative scores have been published with similar predictive ability but they have not caught on.3 Recently, the CHA2DS2-VASc score has been proposed as an alternative to the CHADS2 score and has been incorporated into the 2010 European Society of Cardiology (ESC) guidelines.1Friberg and colleagues have …
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom