Measuring the Effectiveness of Low Vision Rehabilitation

Measuring the Effectiveness of Low Vision Rehabilitation


Most people would agree that providing rehabilitation services to low vision patients is a good thing, and certainly better than doing nothing. But how effective is it and how does one measure its effectiveness in a way that meets the high standards of modern evidence-based practice? A recently published study, the Veterans Affairs (VA) Low Vision Intervention Trial (LOVIT), provides the first compelling evidence, using scientifically rigorous measurements, that low vision rehabilitation is highly effective.1,2

This article describes the measurements used in the LOVIT, explains them in a nontechnical way, and reviews the results of the study.


The term "low vision" refers to chronic visual impairments that limit a person's ability to function in everyday life. By definition, the visual impairments causing low vision cannot be corrected. However, with low vision rehabilitation, therapists provide patients with vision-enhancing and assistive devices, and teach them new strategies and adaptations to improve their ability to function independently despite impaired vision.3 Because the aim of low vision rehabilitation is to improve functional ability, functional ability is the variable that must be measured to test the claim that rehabilitation is effective.

Robert W. Massof, PhD, is professor of ophthalmology and neuroscience and director of Lions Vision Research and Rehabilitation Center and the Wilmer Eye Institute of the Johns Hopkins School of Medicine in Baltimore. He reports no financial interests in any technologies mentioned in this article. Dr. Massof can be reached via e-mail at

Each person has some level of functional ability, and that is what we want to measure. In part, the amount of functional ability a person has depends on visual impairments. Because we cannot observe functional ability directly (it is a latent trait), we must observe manifest surrogate variables and, by way of scientific theory, infer the value of functional ability. Examples of manifest surrogate variables for functional ability are: visual performance measures performed in the clinic, such as walking speed or reading speed; ordinal ratings by clinicians of the patient's ability to perform different activities (analogous to judging a gymnastics competition); and ordinal ratings by patients of how much difficulty they have performing different activities. For the outcome measures, the LOVIT employed the VA Low Vision Visual Functioning Questionnaire (VA LV VFQ),4,5 a self-report rating-scale questionnaire.

"With low vision rehabilitation, therapists teach patients to function independently."

The VA LV VFQ has 48 items. Each item is a question that asks if it is difficult to perform a specific activity. For example, one item asks, "Is it difficult to prepare meals?" Another item asks, "Is it difficult to identify money?" The patient responds "yes" or "no." If the patient responds "yes," the follow-up question is, "How difficult is it?" Accepted responses are "moderately difficult," "extremely difficult," or "impossible to do." The patient's response to each item is scored 1 for "no difficulty" through 4 for "impossible to do."

The items are grouped into 4 functional domains: reading, mobility, visual information (eg, watching TV), and visual motor (eg, grooming).1 The functional ability variable estimated from responses to all 48 items is called visual ability. Functional ability variables also are estimated from responses to the subsets of items that make up each of the 4 functional domains, eg, reading ability.


When the patient rates the difficulty of an item on the VA LV VFQ, what is being judged? Theoretically, each activity described by an item requires a person to possess a specific level of functional ability in order to perform the activity at all. If a person does not have that much ability, the person most likely will respond that the activity is impossible to do. At the other extreme, if the person has functional ability that far surpasses that required by the described activity, the person most likely will respond that the activity is not difficult.

Between these 2 extremes, the person can still do the activity, so the person must have more functional ability than is required by the activity, but that person is likely to report that the activity is moderately or extremely difficult to perform. Thus, in this case the difference between the functional ability of the person and that required by the activity must be less than the difference that gives rise to the response of no difficulty. The difference between the functional ability possessed by the person and the functional ability required by the activity is called functional reserve.6 Theoretically, we assume that patients' difficulty ratings of activities are judgments of their functional reserve for those activities. By using compensatory strategies and devices, rehabilitation can increase functional reserve.

The balance scale is a useful metaphor to help understand how measures are made using responses to rating scale questionnaires. Each item in the questionnaire and each response category is represented by a unique standard weight of unknown value. The weights correspond to the functional ability required by each item and the criterion functional reserve for each response category; the weight of the person corresponds to functional ability. The item weights may be presented in any order, but the response category weights are ordered, with category 1 the heaviest and category 3 the lightest. If the person is "heavier" than item 4 plus category 1, then the response would be "not difficult." If not, then the category 1 weight would be replaced with the lighter category 2 weight and the observation repeated. If, in the end, the person is "lighter" than the item 4 weight plus the category 3 weight, which is the lightest, then the response would be "impossible to do." The response for every person to every item is observed and recorded. Then, when response scores are summed across items for each person and across persons for each item, it is possible to order the people by weight and to order the items by weight using the sums.7

Obviously, there will be random variability between people in their choices of response criteria for each category, ie, the category weights vary around their means across people, and, undoubtedly, there is random variability between people in their interpretation of the items, ie, the item weights also vary around their means across people. Rasch analysis employs a probability model of the variability between people in the item variables and the response category criteria to estimate the functional ability of each person (represented by the person weight), the average functional ability required by each item (represented by the item weight), and the average response category criteria represented by the category weights) from the pattern of responses to the items across people.8 It is not scientifically valid to simply add up response category rank scores, as is frequently done with rating scale questionnaires. At best, such a scale is ordinal and nonlinearly related to functional ability. One must use Rasch analysis to estimate linear measures from the response patterns and to validate those measures.4,5,9,10

"By using compensatory strategies and devices, rehabilitation can increase functional reserve."


The LOVIT is a 2-center, randomized controlled trial conducted under the leadership of Joan Stelmack, OD, MPH, at the Hines VA Blind Rehabilitation Center.1,2 One hundred twenty-six low vision patients having visual acuity less than 20/100 but greater than 20/500 in the better eye because of macular damage were randomized into a rehabilitation treatment group or a waiting-list control group. The VA LV VFQ was administered to each subject by telephone interview at baseline and again 4 months later.

Patients assigned to the treatment group received low vision services from an optometrist who specializes in low vision and from a certified low vision therapist. Low vision services at both centers followed the study protocol and consisted of low vision examination and refractive error correction, counseling and education, low vision therapy and training, and prescribing and dispensing low vision devices. The low vision therapy consisted of 5 weekly 2-hour sessions at the outpatient low vision clinic, during which a therapist taught patients strategies for more effective use of their vision, as well as how to use their devices.

Patients were assigned 5 hours of homework exercises to be completed between therapy sessions. The homework was reviewed with the patient by the therapist. A therapist also made 1 home visit to teach patients about needed environmental adaptations and how to use their new skills and devices to perform tasks in the home. Patients in the waiting-list control group received empathetic and encouraging twice-monthly telephone calls to keep them engaged in the study. Upon completion of the 4-month wait, the same low vision rehabilitation services were provided to the patients in the control group.

Rasch analysis was performed on the VA LV VFQ responses and functional ability was estimated for each patient at baseline and at the 4-month follow-up for both the treatment and control groups. Functional ability measures are reported in logit units. A logit is simply the log odds ratio of the probability of a given person using a particular response category for a given item to the probability of the person using the next lower category, which is calculated as part of the Rasch analysis. Change in functional ability was computed for each patient by subtracting the baseline measure from the follow-up measure. Statistical tests were performed to compare the distribution of change scores for the treatment group to the distribution of change scores for the control group.


Reading ability was the primary outcome measure. On average, this improved by 2.06 logits for the treatment group and worsened by 0.37 logit for the control group. The difference in mean change scores between the treatment and control groups is highly significant. Based on linear regressions between logMAR visual acuity and logits that were observed in previous studies,11,12 the magnitude of the improvement in the treatment group is roughly equivalent to the difference in functional ability expected between patients who differ by 16 lines of visual acuity on the Early Treatment of Diabetic Retinopathy System chart, whereas the change in the control group is roughly equivalent to that expected from a 3-line loss of visual acuity. Mobility, visual information processing, visual motor, and overall visual ability were secondary outcome measures. Highly significant improvements for the treatment group were seen for these measures, with average change scores of 0.57, 1.19, 1.47, and 1.43 logits respectively. The corresponding average change scores for the control group were –0.27, –0.20, –0.04, and –0.20 logits. The functional improvements observed in the secondary measures also are quite large.


Every professional society concerned with low vision rehabilitation has preferred practice patterns that advocate the team approach, similar to that employed by the LOVIT. This list includes the American Academy of Ophthalmology, the American Optometric Association, and the American Occupational Therapy Association. In private practice, low vision examinations and patient education and counseling are provided by ophthalmologists and optometrists. Low vision therapy is provided by occupational therapists (OT), under physicians' orders. The OTs who provide this service usually are certified as low vision rehabilitation specialists by the Academy for Certification in Vision Rehabilitation and Education Professionals or the American Occupational Therapy Association.

Both the low vision services provided by the ophthalmologist or optometrist and the services provided by the OT are covered by Medicare. A 2002 Medicare program memorandum mandates nationwide coverage of rehabilitation CPT codes used by OTs if the patient has at least moderate low vision, which is defined as corrected visual acuity less than 20/60 in the better eye, or central or peripheral field losses that cause functional limitations (ICD-9-CM code 369.25). The 10 hours of low vision therapy plus the 2-hour home visit employed in the LOVIT is eligible for Medicare coverage if provided by an OT.

The VA classifies low vision devices as prosthetics and covers the total cost of devices dispensed to eligible low vision patients. Despite 3 District Court rulings to the contrary, Medicare refuses to cover low vision devices as durable medical equipment or as prosthetics. It classifies all low vision devices as corrective lenses, and it refers to the eyeglasses exclusion clause in the Medicare law as the reason for not covering them. This difference in coverage policy could impact the effectiveness of low vision rehabilitation services in private practice if patients do not acquire the recommended low vision devices. RP


  1. Stelmack JA, Tang XC, Reda DJ, et al. The Veterans Affairs Low Vision Intervention Trial (LOVIT): Design and methods. Clin Trials. 2007;4:650-660.
  2. Stelmack JA, Tang XC, Reda DJ, Rinne S, Mancil RM, Massof RW; LOVIT Study Group. Outcomes of the Veterans Affairs Low Vision Intervention Trial (LOVIT). Arch Ophthalmol. 2008;126:608-617.
  3. Warren M. Employing occupational therapists to assist the low-vision population. Retinal Physician. 2006;3(1):72-73.
  4. Stelmack JA, Szlyk JP, Stelmack TR, et al. Psychometric properties of the Veterans Affairs Low-Vision Visual Functioning Questionnaire. Invest Ophthalmol Vis Sci. 2004;45:3919-3928.
  5. Stelmack JA, Szlyk JP, Stelmack TR, et al. Measuring outcomes of vision rehabilitation with the Veterans Affairs Low Vision Visual Functioning Questionnaire. Invest Ophthalmol Vis Sci. 2006;47:3253-3261.
  6. Massof RW. A systems model for low vision rehabilitation. II. Measurement of vision disabilities. Optom Vis Sci. 1998;75:349-373.
  7. Massof RW. Likert and Guttman scaling of visual function rating scale questionnaires. Ophthal Epidemiol. 2004;11:381-399.
  8. Massof RW, Ahmadian L. What do visual function questionnaires measure? Ophthal Epidemiol. 2007;14:198-204.
  9. Massof RW. The measurement of vision disability. Optom Vis Sci. 2002;79:516-552.
  10. Pesudovs K. Patient-centered measurement in ophthalmology – a paradigm shift. BMC Ophthalmol. 2006;6:25.
  11. Massof RW, Fletcher DC. Evaluation of the NEI visual functioning questionnaire as an interval measure of visual ability in low vision. Vision Res. 2001;41:397-413.
  12. Nutheti R, Shamanna BR, Krishnaiah S, Gothwal VK, Thomas R, Rao GN. Perceived visual ability for functional vision performance among persons with low vision in the Indian state of Andhra Pradesh. Invest Ophthalmol Vis Sci. 2004;45:3458-3465.