Department of Clinical Neurology, Oxford University
Level 6, West Wing, John Radcliffe Hospital, Oxford, OX3 9DU, UK
Model for predicting the risk of ipsilateral ischaemic stroke in patients with recently symptomatic carotid bifurcation stenosis
** Please read these notes on definitions and limitations prior to using the model **
The program is based on the results of Cox regression model and estimates the one-year and five-year risks of ipsilateral ischaemic stroke (symptoms lasting longer than 24 hours) on medical treatment in patients with recently symptomatic carotid bifurcation stenosis. The model was derived on patients who had been randomised to medical treatment in the European Carotid Surgery Trial (ECST),1,2 and is detailed elsewhere.3,4 The predictive values of the individual variables included in the model have also been reported previously.5
Risk predictions produced by models such as the ECST model should not be regarded as exact, are not automatically generalisable, and should only be used by experienced clinicians as an aid to clinical decision-making in the wider context of all other relevant information available about the individual patient. The ECST model can be used as a guide to the approximate level of risk of stroke on medical treatment without endarterectomy or angioplasty/stenting. However, the following points should also be borne in mind:
1) The model was derived on a population of patients with TIA or non-disabling stroke. The results will not apply to patients with disabling stroke.
2) The risk of recurrent stroke is very high indeed during the first few days after a presenting TIA or stroke in a patient with carotid stenosis.6 Few patients were recruited into the randomised trials of endarterectomy during this acute phase. The model has therefore been designed to predict the risk of stroke from day 7 or later. The very early risk of recurrence is not included. Also, because the population on which the model was derived were usually recruited within six months of their last symptomatic event, the model will only accept cases with a time since last event of 180 days or less.
3) Patients recruited into randomised trials tend to have a better prognosis than patients in routine clinical practice, which may lead the model to underestimate risk.
4) Although antiplatelet treatment (usually aspirin alone) was standard, the majority of patients in the ECST were not treated with statins, which is likely to lead the model to overestimate risk compared with current clinical practice.
5) Risk prediction models tend to over-predict at the extremes (i.e. some low risk patients are predicted to be very low risk and some high risk patients are predicted to be extremely high risk). The following limits have therefore been set on the predictions produced by the model: <5% to >20% stroke risk at one year; <10% to >50% stroke risk at five years.
6) Not all potentially relevant risk factors were included in the model as they were not available in the ECST. For example, analyses of the North American Symptomatic Carotid Endarterectomy Trial (NASCET)7 have shown that the risk of stroke on medical treatment is also increased in patients with leukoaraiosis and in patients with stenosis of the distal internal carotid artery in addition to carotid bifurcation stenosis.8
7) The risk prediction produced by the model is not a prediction of likely benefit from endarterectomy of angioplasty/stenting. The procedural risk and the 1-2% risk of annual risk of stroke after successful treatment7 must both be subtracted.
Degree of stenosis: this refers to the maximum degree of linear stenosis at or around the symptomatic carotid bifurcation by method of measurement used in the NASCET trial and the Carotid Endarterectomy Trialists’ Collaboration.7,9 Accepted values for this variable are limited to between 50% and 99% stenosis. There is no evidence of benefit from endarterectomy in patients with <50% stenosis. It is important in cases where stenosis has been measured using non-invasive methods of imaging that users ensure that measurement of the degree of stenosis has been calibrated against the NASCET method of measurement on angiography, or that they transform the result using the previously determined formulae.10
Near-occlusion: The degree of stenosis by the NASCET method of measurement is invalid in patients with near-occlusion i.e. narrowing or collapse of post-stenotic internal carotid artery.9 The program will not therefore accept a specific degree of stenosis if the patient is also entered as a near-occlusion. However, the program automatically assumes that a near-occlusion is a severe stenosis.
Time since last event: refers to the most recent ipsilateral vascular symptomatic event. As outlined above accepted values for this variable are limited to between 7 days and 180 days. It should also be noted that the majority of patients in the ECST had only one or two events prior to presentation. The risk predictions will not apply to patients who have had many recurrent TIAs over months or years without having a stroke.
Primary symptomatic event: refers to the most “severe” ipsilateral vascular event during the previous 6 months (major stroke > minor stroke > multiple cerebral TIAs > single cerebral TIA > monocular TIA or retinal artery occlusion).
Major stroke is defined as a non-disabling stroke with residual neurological symptoms after 7 days.
Minor stroke is defined as a stroke with symptoms lasting between 24 hours and 7 days.
TIA is defined as an event with symptoms lasting up to 24 hours.
Diabetes (type 1 or 2), previous myocardial infarction, peripheral vascular disease: these variables refer to recent or previous confirmed clinical diagnoses irrespective of whether these conditions are currently symptomatic.
Treated hypertension: reflects a past history (i.e. prior to the current presentation with TIA or stroke) of hypertension that was considered to merit blood pressure lowering drugs. It should be noted that when the ECST was performed “hypertension” was usually defined as a BP of 160/90 or higher.
Irregular or ulcerated plaque surface: This definition was based on the surface morphology of the symptomatic carotid plaque as visualised on conventional arterial angiography, which was the imaging investigation of choice in the ECST. More detailed definitions are given elsewhere.11,12 If the patient has not undergone conventional angiography then please enter “Don’t know” and the program will give you the results for both ulcerated/irregular plaque and for smooth plaque separately. Angiographically irregular and ulcerated plaque have been shown to be highly correlated with lipid-rich unstable or ruptured plaques on histology.12
A patient with strong evidence of lipid rich or unstable/ruptured/ulcerated plaque on non-invasive imaging could therefore be entered as having irregular or ulcerated plaque.
6. Lovett JK, Coull A, Rothwell PM, on behalf of the Oxford Vascular Study. Early risk of recurrent stroke by aetiological subtype: implications for stroke prevention. Neurology 2004; 62: 569-74. [Abstract] [PDF]
9. Rothwell PM, Eliasziw M, Gutnikov SA, Fox AJ, Taylor W, Mayberg MR, Warlow CP, Barnet HJM for the Carotid Endarterectomy Trialists’ Collaboration. Pooled analysis of individual patient data from randomised controlled trials of endarterectomy for symptomatic carotid stenosis. Lancet 2003; 361: 107-16. [Abstract] [PDF]
11. Rothwell PM, Gibson R, Warlow CP. The interrelation between plaque surface morphology, degree of stenosis and the risk of ischaemic stroke in patients with symptomatic carotid stenosis. Stroke 2000; 31: 615-621. [Abstract] [PDF]
Return to Stroke Prevention Research Unit's starting page www.stroke.ox.ac.uk