Abstract
Purpose: :
Item responses to visual function questionnaires reflect the respondents’ magnitude estimates of the difference between their functional ability and the functional ability demanded by the activity described in the item, i.e., functional reserve. Functional reserve can be improved by increasing the person’s functional ability (e.g., improving visual acuity) or by providing the person with a tool that makes the activity easier to perform (e.g., using a magnifier). Improving functional ability is expected to increase functional reserve for all items by the same amount. In contrast, the effects of tool use are likely to be activity-specific (e.g., a stand magnifier makes reading easier but has no effect on mobility). Such item-specific effects introduce differential item functioning (DIF), which can distort outcome measures. This study evaluates the VA LV VFQ for outcome measurement distortion from intervention-specific DIF as a result of low vision rehabilitation (LVR).
Methods: :
The VA LV VFQ was administered by phone interview to 100 low vision patients before and 3 months after LVR. All patients had central vision loss in both eyes and visual acuity ranged from <20/100 to >20/500 in the better eye. Half the patients received outpatient LVR in the VA Low Vision Intervention Trial and the other half received 4 to 6 weeks of inpatient rehabilitation services at the Hines VA Blind Rehabilitation Center.
Results: :
Rasch analysis was used to estimate interval measures of overall functional ability from patient difficulty ratings of items. There was significant intervention-specific DIF for 5/9 reading items, 5/11 O&M items, 4/11 visual motor items, and 5/14 visual information items. The DIF for most reading items is negative and the DIF for most mobility items is positive. When the functional domains are analyzed separately, DIF persists for the same 5/9 reading items, 4/11 visual motor items, 4/14 visual information items, but only 1/11 O&M items.
Conclusions: :
The VA LV VFQ exhibits small but significant intervention-specific DIF from LVR. Intervention-specific DIF can introduce bias into LVR outcome measures. The observed DIF indicates that reading items can lead to underestimates of functional ability improvements from LVR and mobility items can lead to overestimates.
Keywords: low vision • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical (human) or epidemiologic studies: outcomes/complications