| To Register SMDM Homepage | |
|
|
|
||||
Purpose: To demonstrate whether different methods of risk assessment have different implications for treatment decisions.
Methods: Stroke risk due to non-valvular atrial fibrillation (NVAF) can be reduced by warfarin or aspirin; warfarin being more effective, with a higher risk profile. Choice of therapy requires assessment of both risks and benefits. Guidelines and risk stratification schemes incorporate stroke risk assessment for individual patients. However, the derivation of risk varies across schemes; from community based cohorts to randomised controlled trials (RCTs). We compared the risks and guidance in widely-promulgated approaches using 193 patients from a population prevalence study of AF. Stroke risk was calculated using the Framingham equation, an AF-specific Framingham equation, the CHADS2 score (derived from a US Medicare dataset) and Scottish (SIGN) guidelines. We compared treatment guidance using SIGN and a decision analytical approach using the Framingham equation. In the latter we classified some patients as “risk too low to benefit from warfarin (RTL)”, if the risk of haemorrhagic stroke caused by warfarin approximated to, or exceeded, risk reduction of thromboembolic stroke.
Results: The schemes produce markedly different risk estimates. Both Framingham equations produce significantly lower risks for men and women than SIGN. The CHADS2 score provides an intermediate level of risk, significantly lower than SIGN for men and women, significantly higher than Framingham for men. 24/148 (16%) NVAF patients without past history of stroke/TIA would be advised warfarin using the decision analysis tool; 62 (42%) using SIGN (p < 0.01). 73 (49%) would be classified as RTL using the Framingham equation, of which 54 (74%) would be advised to take warfarin using SIGN guidelines.
Discussion: Stroke risk in NVAF patients differs markedly using different schemes. Community-based cohorts give lower risk estimates than CHADS2; both are lower than risks derived from RCT control groups. Using community-derived risks would lead to far fewer patients treated with warfarin than guidance derived from RCT control groups. Using the latter may lead to many low risk patients being treated with high risk therapy. This emphasises the importance of deriving risk from good epidemiological studies and raises the debate about appropriate source of data for risk stratification of patients to support effective clinical decisions. RCTs are excellent for effectiveness data, but may be severely limited for risk assessment.
See more of Opening Plenary Session (6)
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)