Although they are very promising and may help to identify those who can be safely discharged for outpatient investigations, they have not been validated beyond several communities in England and California. As with any clinical prediction rule, even when fully validated, the ABCD2 rule should be used with caution to augment (not replace) clinical gestalt.
Working a typical Friday night shift, you evaluate a 67-year old male presenting with 30-minutes of facial droop and left arm weakness which resolved three-hours ago. He has diabetes and hypertension. Normotensive in the ED, your clinical impression is TIA. In your hospital it is unlikely that any investigations or full neurological consultation will be obtainable over the weekend. You wonder if this patient can be safely discharged to return for further evaluation on Monday.
1. Call your friendly neighborhood Neurologist and recommend admission to Hotel St. Elsewhere to await Monday’s diagnostic evaluations.
2. Ask the patient’s overworked, underpaid primary care physician to admit this patient for Neurology consult, diagnostic testing, and observation.
3. Send the patient home if the head CT is unremarkable since the short-term risk of stroke is minimal (your clinical gestalt is <1%).
4. Calculate the patient’s 2-day stroke risk and then discuss admission versus outpatient follow-up in conjunction with the patient, family, primary care physician, and Neurologist.
Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack, S Claiborne Johnston, Peter M Rothwell, Mai N Nguyen-Huynh, et. al., Lancet 2007; 369: 283–92
Stroke is a devastating disease that many patients consider to be a worse outcome than heart attack or even death. Recently, stroke literature has promoted the analogy of a “brain attack” and the role for acute interventions. Unfortunately, this analogy has not extended to TIA because until recently evidence demonstrating interventional models to reduce short-term stroke risk were lacking. Accordingly, many clinicians either admitted TIA for “rapid” investigation and possible intervention (thrombolytics) should progression occur, or discharged these patients home hoping no short-term adverse events occurred. Research has indicated widely different practice patterns for the management of TIA worldwide varying from uniform admission of TIA patients to discharge home of nearly all such patients.
In 2007, three interventional trials demonstrated reduced 90-day stroke rates and short-term hospital length-of-stay when TIA management is initiated from the ED. Furthermore, a systematic review of early stroke risk after TIA demonstrated that TIA patients presenting to the ED have a higher baseline stroke risk than other populations. More importantly, the systematic review indicated that emergent management of TIA by stroke specialists consistently reduced short-term stroke risk. Recommended acute approaches to TIA management include carotid dopplers, anti-platelet and statin therapy. Even in the absence of large randomized controlled trials, these studies strongly suggest that TIA mandates emergent (not delayed outpatient) management. Unfortunately, resources are not unlimited so we need to focus interventional stroke management upon the highest risk populations. The ABCD2 rule can help us to identify those high risk individuals.
The ABCD2 investigators took 2 previously developed rules that required further validation before general use. They then combined the 2 rules into a single score that might be more accurate in identifying a high-risk stroke population. The study included over 2800 patients in 2 diverse systems: the Kaiser Permanente HMO Emergency Departments in California and a number of primary care and neurology clinics in Oxford, UK.
Our patient would receive a point for age, symptom duration, and diabetes plus two points for unilateral weakness equaling 5 points and moderate risk for a subsequent stroke at 2-days.
For patients with a low score the two-day stroke risk was 1%. Although the authors hypothesize that the 1% risk was low enough to support outpatient work-up decisions, this is not what they studied or actually did. Among ED patients in the study, an ABCD2 score of ≤ 1 was associated with no strokes, so if you are only comfortable labeling a patient as low-risk with a 0% stroke rate use a cut-off of 1.
Moderate-risk patients had a 4.1% two-day stroke risk.
High-risk patients had an 8% two-day stroke risk, suggesting that even if there was no treatment to prevent progression, these patients may be better off in the hospital for immediate evaluation and treatment should stroke occur. Only 21% of patients were classified as high-risk which would substantially reduce the admission rates if compared with admitting all TIA patients.
These numbers, if true for the patient above, would certainly be helpful in quantifying the risk-benefit ratio for admission versus discharge.
A second study testing similar rules to the two originally derived may not perform equally well on other populations. In other words, if you don’t work primary care in Oxford or a Kaiser-Permanente-run ED, the rule may not perform as well in your patients. In addition some of the data used was retrospective, so the options for including new and potentially better predictors were limited. Although checking to see if the score corresponds to your clinical decision-making may be reassuring, using it routinely to supplement your management should await subsequent validation of the ABCD2 score on other populations.
Clinical predication rules such as this ABCD2 score can help us make decisions in difficult situations, but we need extensive testing of these rules before we can rest our patients’ futures upon them. Further research may confirm the usefulness of ABCD2 in broad practice since several additional studies are underway. Stay tuned!