Predicting Recurrent Stroke in the Short Term

A novel, online tool that estimates the 90-day risk of recurrent stroke has been created by investigators at the Massachusetts General Hospital. The tool is based on the retrospective review of clinical and MR data from more than 1400 consecutive patients presenting with a stroke to the center. The methods used to develop the tool are available in an advanced online publication of Neurology.

The tool provides weighted scores for features on a patient's brain MR images (if obtained within 72 hours of stroke onset) and those for the apparent cause or causes of the index stroke.

Independent Variable	Score
MR imaging data
Multiple infarcts of different ages	1
Simultaneous infarcts in different circulations	1
Multiple acute infarcts	1
Isolated cortical infarcts	1
History of TIA or stroke within previous month	1
Causative classification of stroke on admission
Large-artery atherosclerosis	2
Cardio-aortic embolism	2
Small-artery occlusion	0
Other cause(s)	3
Undetermined cause(s)	1

The cumulative score is then correlated with an estimated 90-day risk of recurrent stroke on the basis of whether MR imaging data are available. (The validity of the risk estimate appears to be more robust if early MR data exist.)

Total Score	Estimated Risk of Recurrent Stroke Within 90 Days, %
	No 72-Hour MR Data	72-Hour MR Data Available
0	1.1	0.7
1	3.4	3.9
2	5.9	4.2
3	19.1	27.3
4-6	45.3	38.8

The authors note that models currently available to estimate the risk of recurrent stroke, like the Stroke Prognosis Instrument II and the Essen Stroke Risk Score, are "designed to assess long-term risk [2 and 1 years, respectively] and have not been validated for short-term risk prediction." These longer-term models also use clinical factors, like hypertension and diabetes, that don't appear to influence the risk of recurrent stroke during the immediate 90-day post-event period.

Because most stroke patients will not sustain a recurrent ischemic event, the authors argue, a tool that identifies higher-risk patients will enable the directed use of timely, preventive therapies.

Caveats for using the model are provided in the article and an accompanying editorial ("Here comes the sun?"). Notably the number of identified recurrent strokes in the retrospective study was small (60), which limits the predictive power of the tool. The online accessibility of the tool, however, should facilitate its use and real-world validation. If, in fact, the tool is predictive, it should aid the assessment of short-term stroke therapies, like various forms of anticoagulation.

MR = magnetic resonance; TIA = transient ischemic attack.

Image of MR machine from NIH's Stroke: Challenges, Progress, and Promise.