Visual Activities Questionnaire: Assessment of subscale validity for cataract surgery outcomes Vijaya K. Gothwal, PhD, Thomas A. Wright, Ecosse L. Lamoureux, PhD, Konrad Pesudovs, PhD PURPOSE: To examine the psychometric properties of the Visual Activities Questionnaire (VAQ) and each of its subscales in a modern cataract population using Rasch analysis and if flawed, to revise the VAQ and create a valid measure that maximizes its measurement properties. SETTING: Flinders Eye Centre, Flinders Medical Centre, Adelaide, Australia. METHODS: Patients with cataract in 1 or both eyes drawn from the surgical waiting list were mailed the 33-item VAQ for self-administration. The following were examined for the entire questionnaire and each subscale: whether items measured a single construct (unidimensionality), the behavior of response categories, the ability to differentiate between patients’ visual abilities (person separa- tion), matching of item difficulty to participant ability (targeting), and whether items function sim- ilarly across subgroups of participants (differential item functioning [DIF]). RESULTS: The VAQ was completed by 561 patients. Response categories were used as intended. The VAQ discriminated the visual ability of the population (person separation, 4.88) but had sub- optimum targeting, misfitting items, significant multidimensionality, DIF, and 4 dysfunctional sub- scales. Elimination of items causing multidimensionality resulted in a reduced 13-item VAQ that met all validity criteria for satisfactory instrument performance. Only 1 valid subscale (peripheral vision) could be preserved in the 13-item VAQ. CONCLUSIONS: The VAQ in its native form was multidimensional and contained subscales with poor psychometric properties. The revised unidimensional 13-item VAQ was more appropriate for application in cataract outcomes assessment. Ideally, more items should be included to improve the targeting of item difficulty to more able cataract patients. J Cataract Refract Surg 2009; 35:1961–1969 Q 2009 ASCRS and ESCRS Online-Only Material Patient-reported outcomes (using self- or interviewer- administered questionnaires) are increasingly being considered in the evaluation of health care outcomes, including ophthalmic interventions. Many ophthalmic questionnaires were developed using the classical test theory, 1 an approach with acknowledged limitations. A major limitation of the classical test theory is that the person’s ability and item difficulty cannot be esti- mated separately. A classical test theory score is depen- dent on the items administered, and the item means vary by the sample of people who respond to the ques- tionnaire. Thus, classical test theory parameter esti- mates are sample dependent and should not be compared across samples that vary in the distribution of individual (person) ability. The other limitations of the classical test theory include the inappropriate treat- ment of raw scores (or their linear transformations) and categorical item responses as interval data. Modern psychometric approaches, specifically Rasch measurement models, have been adopted to overcome the limitations of the classical test theory. 2 An instrument that fits the Rasch model is reliable and valid. 2 Rasch analysis transforms ordinal-level data into a linear interval scale and thus supports the use of valid change scores and parametric statistics. 2 Rasch analysis also tests the unidimensionality of the instrument (ie, whether items form a unitary latent trait). 2 Due to the benefits offered, Rasch analysis is now increasingly applied in the development or reva- lidation of ophthalmic questionnaires. 3,4 The Visual Activities Questionnaire (VAQ) was de- veloped using the classical test theory to assess diffi- culty elderly persons have in performing everyday visual activities. 5 The VAQ provides a mean or sum- mary score that reflects visual disability and 8 subscale scores. The VAQ has high reported reliability Q 2009 ASCRS and ESCRS Published by Elsevier Inc. 0886-3350/09/$dsee front matter 1961 doi:10.1016/j.jcrs.2009.05.058 ARTICLE