RATIONALE: Cognition is an important function to evaluate when assessing the rehabilitation potential of low-vision patients. Impaired cognition affects a patient’s reading function, which is the most frequent chief complaint of low-vision patients. Impaired cognition also can limit patients’ attention and memory, which impedes their ability to learn new strategies and visual skills.

The Telephone Interview for Cognitive Status (TICS) was designed to be a cognitive impairment screening tool that does not require vision to complete. Previous validation studies have identified cutoff values of the TICS raw score for mild cognitive impairment and for dementia, which is useful for classifying patients. However, whether valid measures of cognitive function can be estimated from the TICS is unknown, even though the TICS raw score is often used as a measure of cognitive ability in regression models.

DESIGN: A prospective observational study of the baseline traits of new low-vision patients schedule to be seen at low-vision centers throughout the United States.

PARTICIPANTS: A total of 764 low-vision patients from 28 clinical centers in the United States were enrolled between April 5, 2008, and May 2, 2011. Inclusion criteria included patients new to low-vision rehabilitation, ≥18 yr, and ability to hear and respond to questions in English over the telephone. No visual field, visual acuity, or diagnostic criteria had to be met to participate in this study.

METHOD: Patients were recruited from the participating clinic’s appointment schedule via telephone by each of the 28 clinical centers to receive permission to contact patient about the study. A contact permission form was sent to the patient’s home for review, which patients signed and mailed back to Johns Hopkins. Once received, a research assistant called the patient to administer a battery of questionnaires by interview. The questionnaires included detailed intake history along with several standardized assessments including the TICS.

ANALYSIS: Rasch analysis, employing the Masters partial credit model, was performed on the TICS item scores across patients. Interval-scaled measures were estimated for each patient, item, and item score threshold. Measurement validity was evaluated with information weighted fit statistics and principal-components analyses of item score residuals. Measurement precision was evaluated with separation reliability indices. Summary statistics and comparisons of item to person measure distributions were used to evaluate the prevalence of cognitive impairment in the low-vision population.

RESULTS: Separation reliability for TICS items is 0.98, which means that 98% of the variance in the item measure distribution can be attributed to real differences between items in the cognitive ability required to respond to each. The separation reliability for persons is only 0.30, which means that 30% of the variance in the person measure distribution can be attributed to real difference between people in cognitive ability. The mean person measure was 2.68 with a standard deviation of 1.66.

The infit statistics were consistent with the expectations of a unidimensional measurement model for both item and person measures. The item measure standard deviation was 0 with a standard deviation of 0.99. On the basis of a TICS cutoff raw score criterion of 31, 34% of the low-vision patients were diagnosed as having at least mild cognitive impairment. On the basis of a TICS raw score criterion of 27, 10% of the low-vision patients were diagnosed as having dementia.

DISCUSSION: The distribution of item measures indicates the TICS has a wide range with good measurement precision on a continuous cognitive ability scale. However, the person measure distribution is heavily skewed toward normal cognitive function where there are few items to discriminate people, which explains the poor separation reliability.

The infit statistics are consistent with the measurement of a single cognitive ability variable; however, principal-components analysis of response residuals shows that the estimated measures explain only 46% of the variance in the observed responses. There appear to be at least three independent variables contributing to the single variable measured by the TICS; these components could be executive functioning, memory and attention, which would have to be measured separately.

IMPACT STATEMENT: Cognition is a determining factor when occupational therapists are evaluating rehabilitation potential for low-vision patients. An accurate standardized assessment is important for occupational therapists to use when assessing cognition. The TICS does not accurately measure cognition for the low-vision patient. A standardized measure needs to be created to individually assess the components of cognition, which are executive functioning, memory, and attention.