Psychometric properties of the Dutch version Adult Strabismus-20 Questionnaire (AS-20)

Strabismus often has an impact on a patient's quality of life (QoL). The Adult Strabismus-20 questionnaire (AS-20) is an instrument developed to assess the strabismus-specific QoL. The aim was to investigate several psychometric properties of the Dutch-AS-20 among adult patients with strabismus.

Patients with any type of strabismus completed the Dutch-AS-20, the EuroQol five-dimensional questionnaire (EQ-5D-5L) and the Amblyopia and Strabismus Questionnaire (A&SQ) (N = 286). Psychometric properties were evaluated with the use of item response theory (IRT) analysis (graded response model) and by determining the internal consistency, test-retest reliability after two weeks, targeting and measurement precision. Construct validity of the AS-20 was tested against the EQ-5D-5L and the A&SQ.

Principal component analysis indicated a psychosocial and function subscale. After collapsing underutilised response options for 10 items, subsequent fit analysis showed that the function subscale benefitted from removal of two items. Differential item functioning (DIF) was detected on four items for age and gender subgroups, but with minimal impact on the total score. Internal consistency (Cronbach's alpha 0.93 psychosocial and 0.87 function subscale) and test-retest reliability (ICC 0.91 psychosocial and 0.88 function subscale) were sufficient. Targeting and measurement precision of both subscales was appropriate for patients with low to moderate levels of visual and psychosocial functioning but suboptimal for patients with high levels of visual and psychosocial functioning. Correlations were between 0.4 and 0.7 for (subscales of) the Dutch-AS-20 and the A&SQ or EQ-5D-5L, confirming expected construct validity.

The 18-item Dutch-AS-20 has sound psychometric properties to assess the strabismus-specific QoL among Dutch adult patients with low to moderate levels of visual and psychosocial functioning. Care must be taken when interpreting the results due to lower measurement precision at the higher end of both subscales. In future longitudinal studies, properties to assess treatment effects should be evaluated.