June 2017
Volume 58, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2017
Latent variable model for visual function in low vision
Author Affiliations & Notes
  • Robert W Massof
    Ophthalmology, Johns Hopkins Wilmer Eye Inst, Baltimore, Maryland, United States
  • Footnotes
    Commercial Relationships   Robert Massof, Janssen R&D (C)
  • Footnotes
    Support  NIH Grant EY22322
Investigative Ophthalmology & Visual Science June 2017, Vol.58, 2049. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Robert W Massof; Latent variable model for visual function in low vision. Invest. Ophthalmol. Vis. Sci. 2017;58(8):2049.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : Visual function questionnaires (VFQ) elicit difficulty ratings on a list of activities. Each patient’s ability and each item’s average required ability to perform the activity are estimated with Rasch analysis. Low vision is caused by losses in visual acuity (VA), contrast sensitivity (CS), and peripheral vision, which differentially affect the difficulty of different items. Most low vision patients are old and have chronic health problems that limit activities, which also differentially affect item difficulty. The aim is to develop a structural equation model (SEM) that explains ability values estimated from responses to subsets of items defining different functional domains.

Methods : The Activity Inventory (AI) is an adaptive VFQ with an item bank of specific tasks nested under activity goals. The AI was given to 3200 low vision patients prior to service. VA, CS, and a detailed health history also were recorded. Each task item is assigned to 1 of 4 functional domains. Factor analysis was performed on ability measures estimated from difficulty ratings of AI activity goals, AI tasks, and subsets of AI tasks for functional domains and on log VA and log CS. The SEM was constructed from health state indicators in the history and estimated ability measures for each of the 4 functional domains.

Results : Two factors explain most of the variance (average 70%). After rotation to make factor 1 coincident with log VA, reading loaded most heavily on factor 1 and mobility on factor 2, Two latent visual variables were defined for the SEM with path weights constrained to their respective factor loadings. Five latent SEM health state variables were estimated from health history indicators to modify the 4 functional domain measures. Visual information measures are least affected by health states and mobility is most affected. Cognition has its largest effects on reading and visual motor functions while the other health state variables have their largest effects on mobility and visual motor functions.

Conclusions : Ability is the weighted sum of two independent vision variables, one correlated with VA and the other correlated with mobility function. Factor 1 can be represented as the “what” visual pathway (ventral stream), which depends on VA, and factor 2 can be represented as the “where” visual pathway (dorsal stream), which depends on peripheral vision and scotomas. Patients’ physical and mental health states also influence ability measures.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×