Introduction: Previous studies show that the NIH Stroke Scale (NIHSS) underestimates volume of right hemisphere (RH) relative to left hemisphere stroke.
Hypothesis: Analysis of descriptions of the “Cookie Theft” picture from the NIHSS yields quantitative measures of severity of RH cortical deficits that, when added to NIHSS scores, better account for infarct volume than NIHSS score alone in acute RH stroke.
Methods: We evaluated 26 patients with acute ischemic RH stroke within 48 hours of onset with NIHSS, analysis of the “Cookie Theft” picture, and DWI, and compared them to 26 age-matched controls. Picture descriptions were evaluated for: total content units (CU; based on published norms), syllables per CU; left CU:right CU; % interpretive CU. A neurologist blinded to clinical data measured lesion volume and % damage to gray and white matter regions of interest (ROI) on DWI after registration to an atlas. Multivariable linear regression was used to generate a model that best predicted stroke volume. Pearson correlations were calculated between behavioral scores and % damage to each ROI. Group means were compared by t-tests.
Results: Lesion volume correlated more strongly with syllables per CU (r2 = .39; p=0.0006) than with NIHSS score (r2=.38; p=0.0008). Moreover, lesion volume was best accounted for by a model that included: total CU, syllables per CU, left CU:right CU, % interpretive CU; NIHSS score, and age (r2 =.80; p<<.00001). Syllables per CU, % interpretive CU, and NIHSS each independently (p<0.05) accounted for some variance in lesion volume. Mean values for picture description scores were significantly different in RH stroke versus age-matched controls by t-tests (p=0.0001 to 0.028), and each score correlated with % damage to a distinct RH region: total CU with % damage to inferior frontal gyrus (r=-.47; p= 0.017) and superior temporal gyrus (STG; r=-.63; p=0.0006); CU per minute with % damage to supramarginal gyrus (r=-.40; p=0.046); left CU:right CU with % damage to STG (r=-.46; p=0.018), and NIHSS score with % damage to insula (r=.55; p=0.0005) and putamen (r=.48; p=0.012).
Conclusions: Adding picture description scores improves NIHSS in accounting for lesion volume in RH stroke and yields complementary information about lesion localization.