Abstract
Purpose :
We published in 2015 a linear Rasch-based scale (Computer Vision Symptom Scale, aka CVSS17) for measuring the computer-related visual and ocular symptoms in workers using video-display terminals. Because Computer adaptive testing (CAT) is currently considered a more efficient and less time-consuming (for test-takers) method than traditional linear questionnaires, we decided to create a new CAT for assessing these symptoms in general population. Therefore, the aim of our study is to identify content for this new CAT and to calibrate the items included in it.
Methods :
For content identification, we used three different sources: (1) items from existing instruments, including a review of the item bank developed for the CVSS17; (2) open questionnaires and semi-structured interviews with digital devices users; and (3) review of social media comments. Items emerged from this phase were reviewed and revised by a group of experts before including them in the new item bank. After that, students and staff of the Faculty of Optics and optometry along with workers from a technological company (DXC Technology) were invited to answer the 76 item bank's items. Subjects aged over 18 and using digital devices at least four hours a day were included in the study while those unable to understand the item descriptors were excluded. Responses were analyzed using Partial Credit Model (PCM) with Winsteps 4.5.1, to assess the item bank’s reliability, DIF according to sex and age group (presbyopes or non-presbyopes), item point bi-serial correlation and person/item measure Infit and outfit characteristics.
Results :
76 unique items exploring 19 different symptoms were included in the item bank and the responses of 377 participants (70.82% women, median age: 23.00 years, range interquartile of age: 21.00 –47.00). PCM and differential item functioning (DIF) analyses supported the retention of 38 items in the final item bank. The 38 items fitted the Rasch model (mean square Infit and outfit 0.77–1.26) and there was no DIF. Item point bi-serial correlation was (0.41, 0.67) and item difficulty –in logits- ranged from -2.14 to 1.77.
Conclusions :
A new item bank for the Digital Eye Strain Computer Adaptive Test (DESCAT) was developed using a systematic approach and all the items included in the DESCAT showed adequate psychometric properties.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.