Accuracy of the Oxford Visual Perception Screen
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Post-stroke visual perception difficulties are common, though not always screened for given limited time and available assessments. The aim is to evaluate diagnostic accuracy of a new short assessment, the Oxford Visual Perception Screen (OxVPS) which has been developed to support clinicians in standardised assessment of visual perception difficulties in line with clinical guidelines.
Methods
In a cross-sectional prospective study, stroke survivors were recruited across three subacute rehabilitation units in the United Kingdom between June 2023 and March 2024. Inclusion criteria were 18 years and older, diagnosis of stroke within six weeks, able to concentrate for 15 minutes, and understand simple English instructions. Participants completed the index test, OxVPS, and reference standard, the Rivermead Perceptual Assessment Battery (RPAB) within two weeks. Diagnostic accuracy was determined via a Receiver Operator Curve analysis of complete cases which included estimation of the Area Under the Curve (AUC) and of sensitivity and specificity at the optimal cut-off point (Youden Index J). A secondary analysis was performed on all cases in which missing data were replaced by random forest imputations. Completion time and rate were compared with a t-test and Wilcoxon test.
Results
The AUC for complete cases (n=33) was estimated at 0.83 with a 95% confidence interval ranging from 0.68 to 0.98. At the optimal OxVPS cut-off point of 2, sensitivity was estimated at 85% (95% CI: 0.54-1) and specificity at 77% (95% CI: 0.54-1). The analysis on all data (n=62) estimated AUC at 0.88 (95% CI=0.80-0.96), sensitivity at 87% (95% CI=0.62-1) and specificity at 80% (95% CI=0.67-1). A shorter completion time (13 min, t(32)=-20.08, p<0.001) and higher completion rate (97%, V=118, p=0.03) of OxVPS compared to RPAB.
Conclusions
Our results indicate that OxVPS is a time-efficient assessment suitable for time-limited clinical settings that can correctly identify stroke survivors with and without visual perception difficulties.
