Abstract Background and aims Visual problems are common after stroke but are frequently overlooked in acute care, where structured visual assessment is rarely routine. The Vision Impairment Screening Assessment (VISA) was developed to enable early identification of stroke-related visual problems by stroke clinicians. We aimed to validate the Norwegian version of VISA against comprehensive orthoptic examination and to describe the prevalence and pattern of visual abnormalities in a hyperacute stroke population. Methods StrokeVIS was a prospective observational study including 127 patients admitted for acute stroke treatment at Oslo University Hospital thrombectomy centre. All participants underwent a structured orthoptic examination serving as the reference standard. VISA was performed by stroke clinicians who had received instruction in the use of the screening tool. Diagnostic performance was evaluated using sensitivity, specificity, predictive values, and overall agreement. Secondary analyses described prevalence and compared VISA with the visual items of the NIHSS. Results 81 patients completed both VISA and orthoptic assessment. Orthoptic examination identified visual abnormalities in 89/127 patients (70.1%), most commonly ocular motility disturbance (46.5%) and visual inattention (38.3%). When compared with the reference standard, VISA demonstrated high sensitivity (94.5%) but low specificity (34.6%), with an overall agreement of 75.3%. The NIHSS visual items showed substantially lower sensitivity (60.7%) and agreement (62.0%), missing 1/3 of abnormalities detected by orthoptic examination. Conclusions The Norwegian version of VISA showed high sensitivity for detecting post-stroke visual problems in acute care, outperforming the NIHSS visual items. These findings support the use of structured visual screening to improve early identification of visual abnormalities in stroke pathways. Conflict of interest Stephen James Ryan: Nothing to disclose, Øystein Kalsnes Jørstad: Nothing to disclose, Anna-Katharina Elisabeth L. Jørstad: Nothing to disclose, Ansar Roy: Nothing to disclose, Morten C. Moe: Nothing to disclose, Mona Skjelland: Nothing to disclose, Fiona Rowe: Nothing to disclose, Anne Hege Aamodt: Nothing to disclose Figure 1 - belongs to Background and aims Table 1 - belongs to Methods Table 2 - belongs to Results
Ryan et al. (Fri,) studied this question.