Letter by Bhupali and Labovitz regarding article, "Clinical scores for predicting recurrence after transient ischemic attack or stroke: how good are they?".

Letter by Bhupali and Labovitz Regarding Article, “Clinical Scores for Predicting Recurrence After Transient Ischemic Attack or Stroke: How Good Are They?” To the Editor: Lemmens et al were unable to validate several prediction models of stroke recurrence after transient ischemic attack (TIA) or minor stroke and suggest that methodological differences between studies explain the variation. However, many of the prediction tools were evaluated in settings that differed from the ones in which they were originally validated. We suspect that this change in intended use is why the performance of the tools was inconsistent at best. One of the studies cited relies on application of brain and carotid imaging in addition to the ABCD2 score. With the addition of imaging, more true vascular events are identified, which reduce the study population to those patients who are less likely to have had a vascular event. Consequently, applying the ABCD2 score to this population is less likely to identify a category of patients at increased risk of stroke. In another study, diagnostic TIA criteria required that patients be evaluated by a vascular neurologist. We suspect that vascular neurologists are more likely to apply the diagnosis of TIA to true vascular events, thus enriching the sample for patients at high risk of future stroke. At our institution, we examined the TIA:ischemic stroke ratio, the proportion of TIA among all TIA and ischemic stroke cases, and found that this ratio significantly varies between general neurologists and board-certified vascular neurologists, who are less likely to diagnose TIA. The difference means that risk of future stroke after a vascular neurologist’s diagnosis of TIA is probably higher than after an ED physician’s diagnosis. Therefore, the predictive value of the ABCD2 score is reduced when diagnostic TIA criteria includes evaluation by a stroke specialist. The ABCD score was originally developed for use in the emergency room by non-neurologists but has been applied in many other settings. The tools can only be expected to be valid in the setting for which they were designed. It is not a surprise to discover that tools such as the ABCD2 score, are not valid when applied in different populations and clinical settings.