Perceived Visual Ability for Functional Vision Performance among Persons with Low Vision in the Indian State of Andhra Pradesh

Abstract

purpose. To determine the distribution of perceived visual ability for functional vision performance among persons with low vision in the Indian state of Andhra Pradesh.

methods. As part of a population-based epidemiologic study, the Andhra Pradesh Eye Disease Study (APEDS), a 16-item visual function questionnaire was designed and applied to 7363 persons older than 15 years, to record the levels of difficulty perceived by the subjects. Of these, 123 persons were found to have low vision. Rasch analysis was used to convert the ordinal difficulty ratings of these 123 persons into interval measures of perceived visual ability for functional vision.

results. Content validity of the questionnaire was demonstrated by good separation indices (3.17 and 5.44) and high reliability scores (0.91 and 0.97) for person and item parameters. Construct validity was shown with model fit statistics. Criterion validity of the questionnaire was shown by good discrimination among the general vision ratings. The functional situation that required the least visual ability was “reaching an object farther or closer than you thought”; the situation requiring the most visual ability was “recognizing small objects.” Bivariate regression analysis determined that for every unit of logMAR visual acuity, perceived visual ability for functional vision decreased by 2.9 logit, which could explain 32% of the variability in the person measure.

conclusions. The described assessment, across a range of visual problems, is a valid way to measure perceived ability for functional vision in persons with low vision. Perceived visual ability varies with every unit of logarithm of the minimum angle of resolution (logMAR) visual acuity.

Low vision is one of the priorities in the global initiative, VISION 2020—The Right to Sight. ¹ According to the World Health Organization, a person with low vision is one who has impairment of visual function even after treatment or refractive correction and has visual acuity of less than 6/18 to perception of light or a visual field of less than 10° in the better eye and who uses, or is potentially able to use, vision for planning or execution of a task. ² Low vision, has a considerable impact on a person’s functioning and independence. The prevalence of low vision in the Indian state of Andhra Pradesh was estimated to be 1.05% (95% confidence interval [CI], 0.82%–1.28%). ³  

Although data on the prevalence of low vision are now available even from developing countries such as India, little is known about the functional vision performance in this population. Functional vision is defined as the vision that can be used to perform a task(s) involving sight (i.e., how a person uses his/her vision). ⁴ However, a measurement of self-perception of functional ability is needed, ⁵ a measurement that can be related to the loss of vision. This would not be an estimate of low-vision persons’ capabilities but rather an estimate of their perceptions of their capabilities. Such assessments could be used as a guide for referral, an aid in planning enhancement of sight and rehabilitation, and an outcome-measure tool for rehabilitation services. The goal of low-vision rehabilitation is to make it easier for visually impaired persons to perform everyday activities by increasing the functional reserve. Functional reserve for a given activity can be increased either by increasing the visual ability of the patient or by decreasing the visual ability needed to perform the activity. ^{6 7 8}  

Various instruments exist for the measurement of visual disability: the Visual Functioning Index (VFI), ⁹ the Visual Activities Questionnaire (VAQ), ¹⁰ the Activities of Daily Vision Scale (ADVS), ¹¹ the Visual Performance Questionnaire (VPQ), ¹² the 14-item Visual Functioning Index (VF-14), ¹³ the Visual Disability Assessment (VDA), ¹⁴ Low Vision Quality of Life Questionnaire (LVQOLQ), ¹⁵ and the National Eye Institute Visual Functioning Questionnaire (NEIVFQ), ¹⁶ among others. ^{17 18} Most of these have been used extensively in studies of treatment outcome. ^{19 20 21 22 23 24 25 26} Moreover, all these instruments have been developed with a different objective from that of the present study. 

As part of a large epidemiologic study, the Andhra Pradesh Eye Disease Study (APEDS), a visual function questionnaire (VFQ) was developed that asks about difficulty with functional vision. As perceived ability is a latent trait (i.e., it cannot be observed), a latent variable analysis is needed to measure the variable underlying the trait. In this study we used the Rasch analysis. ⁷ Several other investigators have also used the Rasch analysis to estimate the interval measures of perceived visual ability. ^{5 28 29 30}  

In the present study, we sought to determine the distribution of perceived ability for functional vision performance in an adult population with low vision. In addition, we describe the validity of the APEDS-VFQ using the Rasch analysis in a low-vision population aged above 15 years and relate the perceived ability for functional visual performance to visual acuity measures in the better eye. 

Methods

Visual Function Questionnaire

Various aspects of the study design of the APEDS have been described previously. ^{31 32 33 34 35 36} In the APEDS, nine instruments were used to collect data from subjects. These instruments are described elsewhere. ³³ Instrument VI (see Appendix) is the visual function instrument designed for persons above 15 years of age, to record the levels of difficulty perceived by the subject in various aspects of visual function such as distance and near vision, vision in bright and dim light, adaptation to light and dark surroundings, color vision, visual field, and depth perception. A focus group comprising two medical experts, two public health experts, an anthropologist, and a demographer, reviewed the questionnaire items, after which the provisional list was finalized. The final VFQ consisted of 16 items, with a general question related to the self-assessment of vision in general. 

A 5-point Likert scale (0–4) was used to record level of difficulty for each of the 16 items. The subjects were instructed to rate on a scale of 0 to 4 the level of difficulty they experienced in performing each task. They were told that 0 meant “no difficulty” and 4 meant “cannot manage.” If they could not rate themselves, the item was rated 5 (“don’t know”). For the Rasch analysis the “don’t know” data were considered missing data. 

Subjects

A multistage sampling procedure was used to select the APEDS sample of 10,000 persons of all ages, with 5,000 each younger and older than 30 years. This grouping was based on the assumption that a 0.5% prevalence of an eye disease in either of these groups may be of public health significance. One urban and three rural areas from different parts of Andhra Pradesh were selected. Approximately 2950 persons were sampled in each of these four areas with the purpose of including at least 2500 participants in each area, such that the total sample would broadly reflect the urban–rural and socioeconomic distribution of the population of this state. The sampling strategies for the urban and rural areas of APEDS has been described earlier. ^{31 32 33 34 35 36} The major difference between the urban and rural sampling was that the former was selected from blocks stratified by socioeconomic status and religion, whereas the latter were selected from villages stratified by caste (traditional social grouping) as described previously. ^{31 32 33 34 35 36}  

The final VFQ was administered to 7363 (99.1% of the 7432 eligible) subjects over the age of 15 years from 94 clusters in one urban and three rural areas by using stratified, random, cluster, systematic sampling by an anthropologist and two experienced field workers. ³¹ Distance visual acuities, presenting as well as best corrected after refraction, were measured separately in each eye using logarithm of minimum angle of resolution (logMAR) charts. ³⁷ The research adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from participants before examination. The study was approved by the ethics committee of the L. V. Prasad Eye Institute, Hyderabad, India. The APEDS was conducted from October 1996 to February 2000. 

Of the 7432 eligible persons who were examined clinically, 135 (1.8%) had low vision. Of these 123 (91.1%) responded to the final VFQ. The mean best corrected visual acuity (BCVA) in the better eye was 0.92 ± 0.45 (SD) (logMAR, 20/166). At presentation, 91 (74%) persons were unaided by glasses. The mean presenting visual acuity in the better eye was 1.08 ± 0.41 logMAR (20/240). The mean (± SD) age of the 123 persons was 54.3 ± 15.1 years (range, 17–102); 50.4% were male, and 17.1% were from urban areas. The most frequent causes of low vision included retinal diseases (43.9%), amblyopia (20.7%), optic atrophy (17.5%), glaucoma (9.3%), and corneal diseases (8.1%). 

Rasch Analysis

The total raw score for the items on the APEDS-VFQ ranged from 54 to 304. The mean (± SD) total raw score on the items was 198.4 ± 69.5, and the average rating was 2.0 ± 1.2. Interval measures of perceived visual ability for functional vision performance were estimated from the ordinal ratings of difficulty by performing a Rasch analysis (Wright and Masters ³⁸ ) on the matrix of ratings by the 123 subjects for the 16 items. An unconditional maximum-likelihood estimation routine (student version of WinStep, ver. 3.33; Mesa Press, Chicago, IL) was used to perform the Rasch analyses. 

The Rasch model is a model of the probability of using a particular rating category as a function of functional reserve. Functional reserve is the difference between the person’s perceived visual ability for functional vision performance and the visual ability required for the particular task. Rasch analysis allowed us to estimate each person’s visual ability (α_n), the required ability of each item ρ_i, and the step measure (i.e., the functional reserve threshold) for each category. It also enabled us to test the validity (accuracy) and reliability (precision) of the measurement of the construct. 

Results

Table 1 summarizes the analysis of five response categories for the difficulty ratings. For each difficulty rating, the “count” column shows how many times the rating was used across all items and subjects. The step measure is the value of person ability minus item difficulty at which the probability of responding with category x equals the probability of responding with category x − 1. The step measure is an important parameter in the Wright and Masters rating scale model, ³⁸ which is used by WinStep. There is no step measure for category 0, because there is no lower category. The step measures should increase monotonically with response category rank. However, there was a reversal from steps 2 to 3 (Table 1) . Moreover, response 2 was not the most probable response for all the values of the person-minus-item measure. Hence, the rating categories 2 and 3 were combined and reapplied, with the Rasch having 0 as “no difficulty” and 3 as “cannot manage.” Table 2 shows analyses of the four response categories for difficulty ratings. The table shows that the step measure increased monotonically with response category rank. The expected measure in each category is the average functional reserve for the extreme categories and the functional reserve for the peak of the probability function. In our sample, the expected measure showed a consistent increase with the order of the ratings. 

Figure 1 is a histogram of the person measures. Person measure is the opposite sign of the person logit value. The person logit corresponds to the difference between each person’s perceived visual ability (α_n) and the mean item measure p̄. If the person logit is positive, the person’s perceived visual ability is greater than the average required visual ability of the 16 visual function situations. If the person logit is negative, the person’s perceived visual ability is less than the average required visual ability. In our sample, estimates of the perceived visual ability (in logits) for visual function performance were normally distributed (P = 0.889, Kolmogorov-Smirnov Z-test). The mean of the distribution (in logits) was −0.87 ± 2.0 (SD). 

The 16 visual function items in the questionnaire are listed in Table 3 in order of least to most visual ability required for functional vision performance according to the subjects’ difficulty ratings. The item number represents the order in which the functional vision questions were listed on the questionnaire. The values in the table are “item logits,” which correspond to difference between the mean item measure for the 16 items (p̄) and the item measure for each item (ρ_i). The item measure (ρ_i) corresponds to the visual ability required for the performance of visual function. Here item measure is the opposite sign of the item logit value. If the item logit is positive, the required visual ability for the performance of visual function is less than the mean required visual ability of all the items, and if the item logit is negative, the required visual ability is greater than the mean required visual ability. Table 3 shows that “reaching an object that is farther or closer than you thought,” “identifying colors,” and “recognizing the people near them” required the least visual ability, whereas “recognizing small objects,” “reading small print in the newspaper,” “recognizing people across the street,” and “recognizing the bus number” required the most visual ability. The items that required almost the same visual ability are “estimating the distance of a vehicle while crossing the road,” “noticing objects off to the side, when walking and looking straight ahead,” and “recognizing traffic signals/lights.” Figure 2 shows a patient ability/item difficulty map determined by Rasch analysis for the items in the APEDS-VFQ. Patients (Xs on the left) appear in ascending order of ability from the bottom of the map to the top, and items (item names on the right) appear in ascending order of difficulty from the bottom to the top. On the whole, the item difficulty is meeting with the ability of the persons, which is represented by the Xs located more where the items are located and the means of the two distributions, denoted in Figure 2 by M, were close to each other. 

Table 4 summarizes the global fit statistics for person ability and item difficulty parameters. Content validity is tested with the separation index, which is a measure of how broadly the person and item measures are distributed along the visual ability dimensions and is simply the estimated ratio of estimated true SD to the SE of the estimate. Separation indices of 3.17 for person measures and 5.44 for item measures were observed in our study. Using these indices with the formula of Wright and Masters, ³⁸ we determined that our sample has five statistically distinct levels of person measures and seven statistically distinct levels of item measures. In addition to having the good separation indices, it is important to have high reliability measures. The reliability of the separation is the ratio of the adjusted SD to the SD of the person or item measure distribution. The closer the reliability value is to 1.0, the less the variability in the measurement distribution can be attributed to measurement error. Table 4 reports high reliability values for the person (0.91) and item (0.97) parameters. 

Construct validity (how well the data fit the assumptions of the model) was evaluated by calculating “infit” and “outfit” statistics. The fit statistics are indices of measurement accuracy. The outlier-sensitive (outfit) statistic is sensitive to unexpected behavior by persons on items far from the subject’s ability level. Values close to 1.0 indicate that the variability in the responses is close to the variance expected by the model. Values greater than 1.0 indicate that the variability in the responses is greater than the variability expectations of the model, and values below 1.0 suggest that the variability in responses is influenced by a covariance term. Because the outfit statistic is sensitive to misfitting persons or items, an information-weighted (infit) statistic also was calculated. This statistic more closely represents the variance of the responses of persons whose person measure is close to the item measure. If the data fit the model, the expected value is 1.0. The normalized infit and outfit mean squares (Z _STD) have an expected value of 0 mean and unit SD. Values that exceed ±2 indicate that the mean square exceeded the model’s expectations by more than 2 SD. Table 3 reports the infit and outfit statistics for the 16 items. Six items in our sample were misfits. The most extreme misfit item was “recognizing small objects.” The mean squares (infit and outfit) of this item exceeded the model’s expectation by 3.2 SD. Other items (“reaching an object that is farther or closer than you thought,” “seeing objects in poorly lit surroundings,” “seeing objects in bright light due to glare,” and “climbing up or down the steps”) had infit and outfit values that exceeded the model’s expectation by more than 2 SD. Ambiguous wording (e.g., farther or closer than you thought) may have contributed to the high variability of responses to these items. 

Figure 3 shows the person measures against the Z _STD infit values for the 119 persons whose responses were included in the final analysis. Data points for the persons with the most visual ability for functional vision performance are located at the top of the graph and those for persons with the least visual ability are located at the bottom. Nineteen (16%) persons’ Z _STD infit values exceeded 2, indicating that their mean squares exceeded the model’s expectations by more than 2 SD. A retrospective review of the persons (n = 6) whose Z _STD infit values lay between 3 and 4 were visually impaired with glaucoma (n = 1), amblyopia n = (2), or retinal disorders (n = 3). Only one 75-year-old woman was observed with a Z _STD infit value of more than 4 (actual, 5.5). This person was blind in one eye due to endophthalmitis and was moderately visually impaired in the second eye due to a retinal problem. She reported difficulty (cannot manage) with noticing objects off to the side while walking but not for the other mobility items with which we might expect such a person to have a greater level of difficulty. Elimination of this misfitting subject does not influence the estimation of item or person measures. 

Criterion Validity

Criterion validity was established by comparing the instrument’s ability to discriminate or predict against the gold standard. If the instrument measures the perceived visual difficulty, then the person measure should differentiate between persons on the basis of vision in general. One analytical tool for testing criterion validity is the receiver operating characteristic (ROC). ³⁹ We performed an ROC analysis ⁴⁰ on the person measures to determine the instrument’s discrimination ability. We computed the area (AUC) under the ROC curve of the person-measure distribution for each pair of responses to the general question and then compared these values to chance performance (AUC = 0.5) to test for significance. This general question was not included in the Rasch analysis. The person measures discriminated between those who rated their vision in general as “good” compared with those whose self-rating was “fair” (AUC = 0.81; P < 0.001) and “poor” (AUC = 0.91; P < 0.0001). Similarly, the person measures discriminated between “fair” and “poor” (AUC = 0.74; P < 0.0001) response categories as well. As only two persons rated their vision in general as “very good,” this category was not considered for ROC analysis. 

Visual Acuity in the Better Eye and Person Measures

We would expect many functional vision performances to become difficult as the visual impairment worsens—that is, measures of visual impairments to be covariant with the perceived visual ability person measure. Figure 4 presents a scatterplot of BCVA in the better eye against the person measure. It demonstrates that BCVA in the better eye is covariant with perceived visual ability (r = −0.57, P < 0.0001). Bivariate regression analysis was performed to determine how much the logMar BCVA accounted for the variability in visual ability person measure. The model computed was α = 1.56 − (2.9 · logMar BCVA). The logMAR BCVA accounted for 31.9% of the variability in the visual ability person measure. We also calculated the correlation coefficient (r = −0.46, P < 0.0001) of presenting visual acuity in the better eye with the person measure. 

Discussion

As part of the comprehensive APEDS, in the present study we made an attempt to determine the distribution of perceived visual ability for functional vision performance in the adult population with low vision. Subjects above 15 years of age responded to a visual function questionnaire (VFQ) developed for use in the APEDS. For each low-vision subject a person score of perceived visual ability was determined. The APEDS-VFQ showed adequate content validity as demonstrated by separation indices (3.17 and 5.44) and high reliabilities (0.91 and 0.97) for person and item measures, respectively. Construct validity of the APEDS-VFQ was demonstrated by good mean square (MNSQ) fit statistics. The criterion validity of the APEDS-VFQ was demonstrated by its ability to discriminate between subjects of different abilities (i.e., those who responded that their vision in general was “good,” “fair,” or “poor”). Although the area under the ROC curves was extremely high in the discrimination of the persons of different abilities, the entire questionnaire has to be applied to know exactly the functional items that require most visual ability. 

As in ours, the five-point Likert-type questionnaire has been used in many studies. However, we found that subjects could not discriminate between more than four categories of difficulty, as shown in the Table 1 . The misbehavior in step measure suggested that there might be some peculiar items (neither difficult nor easy items in the instrument) for which subjects could not discriminate between moderate difficulty and a great deal of difficulty. And further suggested the use of a four-point rating scale rather than the five-point scale. 

Low-Vision Persons Requiring the Most and Least Visual Ability

The hierarchy of required visual ability for the 16 items (Table 3) shows that the most difficult tasks were recognizing small objects, reading small print, recognizing people across the road, and recognizing the bus number. These are all activities that require high resolution. At the easiest extreme of our hierarchy were the items reaching an object that is farther or closer than you thought, identifying colors, and recognizing people at a close distance. The tasks related to mobility in traffic had the same difficulty: estimating the distance of a vehicle (a bus coming toward them) while crossing the road, noticing objects off to the side when walking and looking straight ahead, and recognizing traffic signals/lights. 

The linear scale allows easy comparison of the relative difficulty of items and relative ability of persons. The comparison of item difficulty to person ability is shown in Figure 2 . More persons are distributed where the items are located. Adjusting from bright light to darkness is found at the calibration point for the mean of the item group, whereas adjusting from darkness to light and climbing up or down the steps is at the calibration point for the mean of the person group. This illustrates good targeting of item difficulty distribution to person ability distribution. However, there are persons who perceive that they lack the visual ability to perform even the tasks requiring the least visual ability in the instrument. 

Perceived Ability and Visual Acuity

The correlation between the perceived visual ability and BCVA in the better eye is −0.57 (P < 0.0001). Although it is not near −1, the correlation confirms the expected trend that as visual impairment progresses, visual ability for functional vision performance decreases. We found that the correlation between the perceived visual ability and the presenting visual acuity in the better eye was slightly low (r = −0.46) compared to the BCVA, which may suggest that the adults perceive their functional vision ability as slightly better than their functional ability actually is. Hence, it would be appropriate to consider the BCVA in the better eye to predict the perceived visual ability for functional vision performance. Furthermore, bivariate regression analysis revealed that on average, for approximately every unit of change in logMAR BCVA in the better eye, the person measure decreased by 2.9 logit, suggesting that persons who show such a change in the functional vision performance should be referred to low-vision or rehabilitation services. 

The results of the present study are similar to those reported by Massof and Fletcher ¹⁶ in their evaluation of the NEI-VFQ as an interval measure of visual ability in low vision. They reported a high linear correlation between visual acuity and person measure, but the NEI-VFQ Z _STD values were more than the Z _STD values observed in the study. Recently, we ³⁰ reported a strong correlation between visual acuity and perceived visual ability suggesting that visual acuity is a major factor for visually impaired children in their responses to the 19 items on the LVP-FVQ. Hazel et al. ⁴¹ reported that perceived visual performance is not solely dependent on visual variables. A psychological or emotive element also contributes to how well patients believe they can see. Investigators have also noted that those patients with low vision and the elderly can sometimes be poor at providing an accurate global description of their visual ability. 

Although the visual function questionnaire used to measure perceived visual ability is valid and reliable, it has certain limitations. Though the nonresponse rate was low (8.9%), it was observed that, all the factors being comparable, the average presenting (1.93 ± 0.6) as well as the best corrected (1.93 ± 0.6) visual acuities of nonrespondents were significantly (P < 0.0001) worse than those of the the respondents (1.08 ± 0.4; 0.92 ± 0.4). In addition, using anthropologists instead of experienced field workers to collect the data is unconventional and could introduce some bias in terms of how they perceived the answers from the subjects. The measurement of self-perception of functional ability relies heavily on the subject’s attitude and interest—that is, it is subjective. This would be affected by the observer’s conscious or unconscious bias. The subjective nature of this questionnaire could also lead to certain problems. Whereas some subjects had difficulty with the task at a particular visual level, other subjects had no trouble with that task, even at a lower visual level. This will complicate conclusions and management. 

However, once low-vision services are made an integral part of comprehensive eye-care services for those with visual impairment, this instrument could be used to measure the low-vision outcomes in terms of improvement in the perceived visual ability for functional vision performance, taking into account the limitations. 

Appendix

APPENDIX

APEDS Data Collection Form VI: Visual Function

Study ID Number: Date: 

Would you say that your vision in general (with glasses, or other correction if you wear them) is: 

1. Very good ; 2. Good ; 3. Fair ; 4. Poor 

 Supported by the Hyderabad Eye Research Foundation, Hyderabad, India; Christoffel-Blindenmission, Bensheim, Germany; and partly supported by the Australian Federal Government through the Cooperative Research Centres Program.

 Submitted for publication March 3, 2004; revised June 7 and June 28, 2004; accepted July 12, 2004.

 Disclosure: R. Nutheti, None; B.R. Shamanna, None; S. Krishnaiah, None; V.K. Gothwal, None; R. Thomas, None; G.N. Rao, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Summary of the Analysis of the Response Categories for Difficulty of the 16 Items

Difficulty Rating*	Count	Step Measure	Infit	Outfit	Expected Score Measure, †
0	89	None	1.06	0.87	(−4.66)
1	652	−3.55	0.93	0.89	−1.52
2	228	0.80	0.91	0.94	0.45
3	357	0.31	0.88	0.96	1.68
4	198	2.44	1.29	1.53	(3.63)

*  0, No difficulty; 1, little difficulty; 2, moderate difficulty; 3, great deal of difficulty; 4, cannot manage.

†  Data in parentheses are unbounded on one side and are estimates of the category boundaries.

Table 2.

 

Summary of the Analysis of the Response Categories for Difficulty of the 16 Items

Difficulty Rating*	Count	Step Measure	Infit	Outfit	Expected Score Measure, †
0	89	None	1.03	0.96	(−5.02)
1	652	−3.91	0.84	0.82	−1.75
2	585	0.43	0.96	0.95	1.96
3	198	3.48	1.21	1.29	(4.62)

*  See Table 1 .

†  Data in parentheses are unbound on one side and are estimates of the category boundaries.

Figure 1.

 

Distribution of person measures of perceived visual ability for functional vision performance (α), estimated from the application of the model of Wright and Masters ³⁸ to subjects’ ratings of the difficulty of 16 items. Person measures are expressed as logits, where a person logit = α_n –p̄.

Table 3.

 

Results of the Rasch Analysis

Item	Visual Function	n *	Item Logit	Error	Infit			Outfit
										MNSQ	Z STD	MNSQ	Z STD
14	Reaching an object	118	1.96	0.19	0.65	−2.9	0.60	−3.1
10	Identifying colors	116	1.90	0.19	0.82	−1.3	0.80	−1.4
03	Recognizing people	119	1.13	0.18	0.91	−0.7	0.91	−0.7
16	Estimating the distance	111	0.91	0.19	0.88	−0.9	0.87	−0.9
12	Noticing objects off to the side	116	0.90	0.18	1.20	1.4	1.22	1.0
11	Recognizing traffic signals	29	0.88	0.36	1.26	0.9	1.29	1.0
15	Climbing up or down the steps	118	0.67	0.18	0.72	−2.4	0.70	−2.4
08	Adjusting from darkness to bright	119	0.64	0.18	1.21	1.6	1.19	1.3
13	Noticing objects on floor near you	117	0.25	0.18	0.90	−0.8	0.87	−1.0
09	Adjusting from bright light to dark	118	−0.10	0.18	0.77	−2.1	0.77	−1.9
07	Seeing objects in poorly lit area	117	−0.43	0.18	0.69	−2.8	0.69	−2.7
06	Seeing objects in bright light	93	−0.67	0.20	1.38	2.5	1.40	2.4
05	Recognizing the bus number	40	−1.38	0.30	1.15	0.7	1.27	1.1
04	Recognizing people across the street	102	−1.92	0.19	1.28	1.9	1.26	1.5
01	Reading small print	30	−2.15	0.35	1.24	0.9	1.30	1.0
02	Recognizing small objects	61	−2.59	0.25	1.68	3.2	1.68	2.9

 MNSQ, mean square; Zstd-MNSQ normalized to approximate a theoretical mean 0 and 1 SD (Standard Deviation).

*  Number of subjects to whom the item was applicable.

Figure 2.

 

Patient ability/item difficulty map for the 16-items. To the left of the dashed line are the patients, represented by X, and on the right are the items, denoted by their content. More able patients and more difficult items are near the top of the diagram, and less able patients and easier items are near the bottom. M, mean; S, 1 SD from the mean; T, 2 SD from the mean.

Table 4.

 

Summary of Global Fit Statistics for Person Ability and Item Difficulty Parameters

Parameter	Separation Index	Reliability	Average Infit	Average Output	Model Measurement Error	SD
Person ability	3.17	0.91	0.99	0.96	0.55	0.06
Item difficulty	5.44	0.97	1.05	1.05	0.22	0.06

Figure 3.

 

Person measures of perceived visual ability for functional vision performance (α) versus the Z-transformed infit mean squares (Z _STD). Data points for the subjects with the most perceived visual ability for the functional vision are located at the top of the graph, and those for persons with the least perceived visual ability are located at the bottom. Normalized infit values that exceed ±2 (located in the shaded regions of the graph) indicate that the mean square exceeded the model’s expectations by more than 2 SD.

Figure 4.

 

Scatterplot of person measure estimates based on subject responses to 16 items that required difficulty ratings versus BCVA. Visual acuity is expressed as the logMAR. A logMAR of 0 corresponds to a Snellen acuity of 20/20 and a logMAR of 1 corresponds to 20/200. The regression line (solid line) has a shape of 1.56 and an intercept of 2.9. r = −0.57.

Table 5.

QUESTION	RESPONSE
	0. No difficulty
	1. Little difficulty
	2. Moderate difficulty
	3. Great deal of difficulty but can still manage
	4. Cannot manage
How much difficulty do you have in	5. Don’t know
1. Reading small prints in the newspaper/magazines?
2. Recognizing small objects? (for example, inserting a thread in the needle)
3. Recognizing people near you?
4. Recognizing people across the street?
5. Recognizing the bus number?
6. Seeing objects in bright light due to glare?
7. Seeing objects in poorly lit surroundings?
8. Adjusting from darkness to bright light?
9. Adjusting from bright light to darkness?
10. Identifying colors?
11. Recognizing traffic signals/lights?
12. Noticing objects off to the side, when you are walking and looking straight ahead?
13. Noticing objects on floor near you, when you are walking around the house and looking straight ahead?
14. Reaching an object, because it is farther or closer than you thought (e.g., to take a glass)?
15. Climbing up or down the steps because they were closer or further away than you thought?
16. Estimating the distance of a vehicle (a bus coming towards you) while crossing the road?

The authors thank Robert W. Massof, PhD, for guidance with the Rasch analysis, the subjects for participating in this study, Lalit Dandona and Rakhi Dandona for the design and execution of the APEDS, and Nagraj V. Naidu, Kovai Vilas, Pyda Giridhar, and Mudigonda N. Prasad for interviewing the subjects.