From January, 2018 to October 2019, 136 stroke inpatients from the First Affiliated Hospital of Soochow University were conveniently sampled. They were assessed with EBI by rater A twice within 48 hours after admission, with interval above twelve hours; and by rater B within 24 hours. They were also assessed with modified Barthel Index (MBI) and Function Independence Measure (FIM) within 48 hours after admission by rater C. The intraclass correlation coefficient (ICC) of retest and inter-rater of EBI were calculated, as well as the Cronbach's α coefficient, Spearman correlation coefficient between scores of items and total, and Spearman correlation coefficient of EBI to MBI or FIM. The construct validity of EBI was tested with factor analysis.

The retest ICC of items was 0.766 to 0.953; the retest ICC of motor and cognitive items was 0.938 and 0.845, respectively; the retest ICC of total was 0.891. The inter-rater ICC of items was 0.728 to 0.976, the inter-rater ICC of motor and cognitive items was 0.948 and 0.717, respectively; the inter-rater ICC of total was 0.866. The Cronbach's α coefficient of EBI was 0.885. The correlation coefficients were above 0.4 in all the items (P < 0.001) to the total, except the item of vision (r = 0.215, P < 0.05). The correlation of items of EBI to MBI was significant (r = 0.648 to 0.958, P < 0.01), as well as those of EBI to FIM (r = 0.722 to 0.976, P < 0.01). Four components were extracted with principal component analysis, accumulated to 72.19% of the variable; after the vision item was removed, three principal components were extracted, accumulated to 69.09% of the variable. The component 1 was mainly about sphincter control and some advanced brain functions (communication and social cognition), component 2 mainly about ADL related to lower extremities, and component 3 mainly about ADL related to upper extremities.

EBI is reliable and valid in the assessment of ADL for stroke patients.