Validation of the interdisciplinary Norwegian vision assessment tool KROSS in stroke patients admitted to hospital or rehabilitation services

Abstract

Abstract Introduction Identifying vision problems after stroke is important for providing appropriate referral and vision rehabilitation in healthcare services. In Norway, vision assessment is not a standard routine or integrated in stroke care, due to lack of knowledge, guidelines and validated Norwegian assessment tools for healthcare professionals (HCPs) without formal vision competence. This study aimed to validate and assess the reliability of the KROSS (Competence, Rehabilitation of Sight after Stroke) tool for identifying vision problems in stroke patients. Methods The KROSS tool has 21 items, including symptoms, observations, and assessment of visual acuity, visual field, eye movements and visual inattention. The primary outcome is to identify if a vision problem is present. Sixty-seven stroke survivors (69.8 years, 28 females) were assessed twice. The first KROSS assessment was by an HCP without formal vision competence and compared to a reference assessment by an optometrist/KROSS specialist within 2 days. Sensitivity, specificity, positive and negative predictive values (PPV/NPV) and inter-rater reliability (Gwet’s AC1/Cohen’s Kappa) were calculated with 95% confidence intervals. Results The KROSS tool demonstrated high sensitivity (98%) and specificity (83%), with excellent reliability (AC1 > 0.86/Kappa > 0.83) and observer agreement (93%) for the primary outcome. A vision problem was identified in 64% of patients, where 44% reported a vision symptom. The PPV and NPV for identifying a vision problem were 0.91 and 0.95 respectively. Sensitivity scores for visual acuity, reading, and visual inattention assessments were all excellent (> 80%) and specificity scores were high for all items (> 70%). Most items showed excellent or substantial agreement (AC1 > 0.7/kappa > 0.6). The lowest agreements were for motility (AC1 > 0.8/kappa > 0.4) and peripheral visual fields (AC1 > 0.8/kappa > 0.5). Conclusions This study shows that the KROSS tool shows promise as a valuable tool for integrating vision assessment into stroke health services. It has high sensitivity and specificity, and excellent reliability, indicating high accuracy for identifying a vision problem. This indicates that the KROSS tool can reliably be used by HCPs without formal vision competence to identify a vision problem. The fact that many stroke survivors were identified with vision problems using the KROSS tool, even if they did not complain of visual symptoms, supports the significance of including structured vision assessment in stroke care.