|Home | About | Journals | Submit | Contact Us | Français|
In the most recent edition of The Neurohospitalist, Marsh and colleagues reported a significant limitation of serial National Institutes of Health Stroke Scale (NIHSS) assessments in patients with acute ischemic stroke.1 The investigators rightfully concluded that an improvement in 4 points on the NIHSS was less sensitive for detecting neurologic recovery when compared to a comprehensive neurologic examination. Recognition of this weakness is pertinent to all clinicians who rely on the NIHSS to identify neurologic recovery (or worsening), especially when using a prethresholded NIHSS improvement in the definition of recovery.
The NIHSS is certainly not a replacement for a full neurologic examination. This tool, however, has traditionally been utilized to quantitate neurologic change in patients with stroke and serves as a useful adjunct in clinical decision-making and for research purposes. In some of the earliest clinical trials using serial NIHSS assessments, a clinically significant change has been prespecified at the same 4-point minimum as in this study.2 This is reasonable given the (1) imperfect interrater reliability and (2) fluctuations in neurologic symptoms depending on patient alertness and time of day, which may render a 2-point threshold less specific. However, in my experience, a 2-point threshold may be more sensitive and maintain a high specificity when compared to these historic 4-point cutoffs.3 As the authors demonstrate,1 a 2-point improvement actually confers nearly identical prognostic information regarding improvement when compared to the comprehensive neurologic assessment. It may even overestimate the degree of improvement. Specifically, at 24 hours post-tissue plasminogen activator, 78% of patients improved by ≥2 points compared to 71% who improved according to non-NIHSS neurologic assessment. At discharge, this rose to 83% and 71%, respectively, and at ~1-month follow-up, 100% and 95%, respectively. Therefore, a 2-point threshold may perhaps be superior for quantitating improvement in this population when compared to the 4-point cutoff.
That being said, should the primary outcome of investigations such as these be improvement? The authors conclude that serial NIHSS assessments are inadequate for capturing changes in functional outcome when in fact their findings indicate inadequately captured improvement. The delivery of positive prognostic information to patients should certainly be pursued and explored with research efforts, but (as the authors affirm) serial NIHSS assessments may be most relevant when “considering functional decline.”1 In my experience3 and others,4 serial NIHSS assessments can successfully capture deterioration. Confirmatory research is called upon to determine whether prethresholded NIHSS changes can be implemented clinically to determine when it is appropriate to intervene and where it may be effective in identifying reversible causes of deterioration.
The NIHSS is not an ideal tool for detecting neurologic change after stroke. But a quantitative assessment should be implemented as an adjunct to the physical examination in order to monitor for progression. For the time being, I find it reasonable to perform serial NIHSS assessments in order to detect neurologic deterioration—especially when a lower threshold (eg, 2- vs 4-point worsening) is used to detect a clinically meaningful decline.5