Repeatability of Reading Ability Indices in Subjects with Impaired Vision

Abstract

purpose. Reading speed and acuity measurements are important means of evaluating reading performance in subjects with impaired vision. The MNREAD charts are gradually becoming a popular means of measuring reading ability in low-vision clinics; yet, to date there are no standards of reliability for the visually impaired population. The present study was conducted to determine the coefficients of repeatability for reading speed, reading acuity, and critical print size in a visually impaired population.

methods. Twenty-seven visually impaired subjects (19 women and 8 men; mean age, 72.07; SD, 12.79 years) participated. Test–retest repeatability of the MNREAD charts was assessed at the subject’s preferred viewing distance over two sessions by the same examiner.

results. The coefficient of repeatability (CR) was found to be ±0.10 logMAR for reading acuity, ±0.20 logMAR for critical print size and ±0.10 logarithm of words per minute for reading speed. The CR was not dependent on visual acuity.

conclusions. As expected, the coefficients of repeatability are higher in visually impaired subjects than in normal subjects. Critical print size measurements have the poorest reliability and precision and should be interpreted with caution.

Reading is cited as the primary rehabilitation goal by most individuals with visual impairment. ^{1 2} It plays such an important role in the everyday life of an individual that low vision has sometimes been defined as the inability to read a newspaper with best optical correction at a normal reading distance. ³ Reading ability can be measured with several commercially available charts. One reading chart in particular, the MNREAD acuity chart, ⁴ has become a popular means of measuring reading ability in clinical and low-vision research. ^{5 6 7 8 9} MNREAD acuity charts are continuous-text reading charts that were developed at the Minnesota Laboratory for Low-Vision Research (Minneapolis, MN). They can be used to measure reading acuity, reading speed, and critical print size and are commercially available in several languages. ¹⁰ The repeatability of MNREAD acuity charts has been established in children ¹⁰ and normally sighted young adults. ¹¹ Both studies found that the MNREAD acuity charts provide a reliable measurement of reading acuity, reading speed, and critical print size in the population studied. Values from both studies can be used to assess whether a significant change has occurred in these subject groups only. 

Various studies have measured repeatability in reading ability in subjects with visual impairment. ^{12 13 14 15} Pesudovs et al. ¹² assessed the repeatability of reading speed measurements for reading text consisting of 30 words of 1.20 logMAR size in subjects with normal vision and subjects with ocular disease. They found differences of up to 16% between sessions, irrespective of whether the subjects had normal or impaired vision. Maaijwee et al. ¹⁴ assessed the repeatability of the Dutch version of the Radner Reading Chart in subjects with macular disease at various distances VA (0.1–0.8 logMAR). They found the CR for reading acuity, reading speed, and critical print size to be very similar to what had been previously found with MNREAD charts on normal subjects by Subramanian and Pardhan. ¹¹ This report was interesting because repeatability studies for distance visual acuity have shown that CR in subjects with visual impairment is higher than that in normal subjects, ^{16 17 18 19} and one would expect similar results at near. Kiser et al., ¹⁶ for example, found that the criterion for change of distance visual acuity was larger in subjects with visual impairment than in subjects with normal vision and concluded that, as a general rule, the criterion for change was 11 letters for visual acuity in subjects with low vision. Patel et al. ¹⁷ also concluded that a significant amount of variability exists in visual acuity measurements in subjects with AMD. The CR was found to be 12 ETDRS letters, although this ranged from 9 to 17 letters, with more variability being observed in subjects with advanced AMD. Woods and Lovie-Kitchin ¹⁹ found a slightly higher CR in subjects with low vision on the Bailey-Lovie Distance Visual Acuity charts. The CR varied from 17 letters for measurements performed in the laboratory to 29 letters for measurements performed in a clinical setting. 

Given that MNREAD acuity charts are being used increasingly to establish whether reading ability improves postrehabilitation/treatment ^{8 20 21 22} in subjects with visual impairment, it is necessary to establish repeatability of these measurements. As far as we are aware, no previous studies have looked at the repeatability of the commercial version of MNREAD acuity charts on a population with visual impairment. Ahn et al. ¹⁵ looked at the repeatability of reading speed on an initial version of the MNREAD acuity chart where all the letters subtended 6° when viewed at 19 cm: This version of the chart differs from the currently available commercial version. This study was therefore designed to evaluate the repeatability of the current version of MNREAD charts in a population with visual impairment and to establish 95% confidence intervals for reading acuity, reading speed and critical print size. 

Methods

Twenty seven subjects with visual impairment and age ranging between 22 and 89 years (mean, 72.07; SD, 12.79) participated in the study. The subjects had binocular distance visual acuity ranging from −0.04 to 1.38 logMAR (mean, 0.62; SD, 0.37) on the Bailey-Lovie Distance Visual Acuity charts with best correction. Three subjects had a distance visual acuity of 0.12 logMAR or greater (0.12, 0.1, and −0.04), which would normally not be classified as visually impaired. However, since these subjects had been referred to the Low Vision Clinic by their optometrists with subjective difficulties with reading, it was decided to include them in the study. Of the 27 subjects, 3 had early macular changes, one had Stargardt’s disease, 1 had bilateral optic atrophy, and the remainder had age-related macular degeneration (AMD). More information about the study participants can be found in Table 1 . Ethical approval was obtained from the University Ethics Committee, and the study adhered to the tenets of the Declaration of Helsinki. All subjects signed a consent form. 

Reading ability was tested using the MNREAD acuity charts. The MNREAD acuity chart contains 19 sentences of different print sizes ranging between +1.3 and −0.5 logMAR. Each sentence contains 60 characters. Test–retest repeatability was measured at the subject’s preferred viewing distance with the subject’s reading correction. All measurements were performed binocularly. 

Two different versions of the English MNREAD acuity charts were used, and the version of the chart used was randomized between subjects and sessions. Both sessions were performed on the same day, although subjects were given a break between the sessions. The MNREAD chart was placed on a stand and the subjects adjusted the stand to their preferred viewing distance. The subjects were instructed not to alter their working distance and the examiner watched each subject to ensure that this was the case. The subjects were instructed to read a sentence aloud, starting from the top of the chart after hearing “start.” A tape recorder was used to record each session, and the reading parameters were calculated at a later date. Testing was performed in a well-lit room; the light available at the chart’s surface was 102 cd/m². 

Data Analysis

Three parameters were calculated: reading acuity, reading speed, and critical print size. Reading acuity is the smallest print that can just be read and is scored letter by letter. The following formula, as recommended by the manufacturer, was used to determine reading acuity:  

\[\mathrm{Reading\ acuity}\ {=}\ 1.4\ {-}\ (\mathrm{sentences}\ {\times}\ 0.1)\ {+}\ (\mathrm{errors}\ {\times}\ 0.01)\]

 

No correction factor was added when the measurements were performed at 40 cm. At all other distances, a correction factor was used based on the formula suggested by the manufacturer.  

\[\mathrm{Correction}\ {=}\ \mathrm{log}_{10}\ {[}40/(\mathrm{viewing\ distance\ in\ cm}){]}\]

 

Reading speed (in words per minute) was determined for each sentence as the number of words read correctly divided by the time taken to read the sentence. Reading speed remains constant over a wide range of print sizes, resulting in a plateau. However, as the print size becomes smaller the reading speed decreases before the acuity limit is reached. ²³ Maximum reading speed was defined as the mean of the reading speeds across the plateau. Reading speed is typically expressed as words per minute. For the purpose of this study, we thought it best to perform a log₁₀ transformation on reading speed to maintain uniformity, as both reading acuity and critical print size are expressed in log units. In addition, if the reading speed is expressed in words per minute (wpm), a coefficient of repeatability (CR) of ±20 wpm, for example, is very different if this criterion is used for a person with a reading speed of 150 wpm rather than 50 wpm, a change of approximately 15% to 40%. To avoid this problem, we decided to use log transformation and express the CR as a percentage change. 

Critical print size is defined as the smallest print that supports the maximum reading speed and was identified based on the criterion that all the following (smaller) sentences were read at a speed that was 1.96 times the standard deviation below the average of the largest preceding sentences. ^{9 10} It should be noted that critical print size is measured in steps of 0.1 logMAR. A correction factor was used where appropriate to account for the nonstandardized working distance. 

To measure test–retest repeatability the CR, as described by Bland and Altman, ²⁴ was calculated for reading acuity, reading speed, and critical print size. This calculation involves determining the upper and lower limit of change that can be expected when a test is repeated on a clinically stable subject. The CR was calculated by obtaining the difference in measurements between sessions 1 and 2 and then squaring the result. The squared difference was added for all subjects. This value was then divided by n (n = 27), and the square root obtained to give the SD of the differences, which was then multiplied by 1.96 to give the CR at 95% level. 

Since the present study had a relatively small sample size, the precision of the CR was also calculated as described by Bland and Altman. ²⁴ Precision was obtained by first calculating the SE of the differences of the means (root s ²/n, where s is the SD of the differences of the means and n is the sample size). For the given sample size the degrees of freedom were 26, giving t = 2.056. The precision of the CR was therefore CR ± 2.056 × SE of the differences of the means. 

Results

Table 1lists the individual characteristics of the study participants in terms of their age, ocular disease, distance spectacle prescription and addition required for reading, reading distance in centimeters, distance visual acuity (DVA) reading acuity (RA), critical print size (CPS), log reading speed (log RS), and reading speed in words per minute (RS wpm). Table 2lists the means and standard deviations for reading acuity, reading speed, and critical print size obtained on two separate occasions with the MNREAD charts. Practice (learning) effects were measured by using t-tests to determine whether there were any differences between the scores (reading acuity, critical print size and reading speed) on occasions 1 and 2. No significant differences were found (P > 0.01), indicating that practice effects did not exist. Table 2also lists the mean differences for the MNREAD scores, their 95% confidence intervals, the CR (expressed both in log units and percentage change), and the precision of the CR. The CR was similar for both reading acuity and reading speed but was poor for critical print size. 

The Bland and Altman plots for reading acuity, reading speed, and critical print size were calculated to assess repeatability (Figs. 1 2 3) . The figures show a comparison of reading acuity, reading speed, and critical print size measured on occasions 1 and 2, respectively. In all three figures the bold lines indicate the 95% confidence intervals and the central dotted lines indicate the critical difference for the parameter measured. The results showed good repeatability for reading speed and reading acuity, with narrow confidence intervals and only one outlier for reading speed. Critical print size showed the poorest repeatability, with larger confidence intervals and two outliers. 

We investigated the possibility of any relationship between distance VA and the CR for the different parameters of reading that were investigated in the present study. Study participants were divided into two groups based on their binocular distance visual acuity: Visual acuity 0.5 logMAR or better (group 1) and a VA < 0.5 but ≥1.3 logMAR (group 2). This grouping was based on the WHO classification where subjects with visual acuity of 6/18 to 3/60 (0.5–1.3 logMAR) in the better eye are classified as having low vision. ²⁵ This resulted in an inclusion of 14 subjects in group 1 and 12 subjects in group 2. One subject was excluded who had a distance visual acuity of 1.38 logMAR. The CRs for the individual subgroups are as follows: RA: 0.10 (10 words/line), CPS: 0.2 (two lines), RS: 0.11 log wpm (group 1); and RA: 0.11 (11 words/line) CPS: 0.2 (two lines) RS: 0.09 log wpm (group 2), for reading acuity, critical print size, and reading speed, respectively. 

The data showed no significant differences between the two groups. 

Discussion

Our results show that the MNREAD acuity charts had good repeatability for reading acuity and reading speed. Critical print size had the lowest repeatability. These findings are similar to those obtained by Virgilli et al. ¹⁰ Critical print size is measured in much larger steps than reading acuity (0.1 logMAR vs. 0.01 logMAR) and this difference may explain why critical print size had poorer repeatability. 

The CR in subjects with visual impairment was higher than we reported in subjects with normal vision, where CR was 0.05 logMAR (5 words) for reading acuity, 0.1 logMAR (10 words/line) for critical print size, and 8.54 wpm for reading speed. ¹¹ In the present study, the CR for both reading acuity and critical print size doubled in patients with visual impairment. Moreover, for reading speed, the CR nearly tripled. These findings are similar to those found for distance VA, ^{16 17 19} but are higher than those found by Maaijwee et al. ¹⁴ with the Radner reading charts. It is difficult to say why these differences were found, as the subjects in the present study and in Maaijwee et al. ¹⁴ seem similar in distance VA and disease profiles, although some subjects in the present study had higher distance VA. Even when data are analyzed without these subjects, the CR was still higher. Radner reading charts are similar to MNREAD acuity charts, except that the length and position of the words in the Radner reading chart are similar for all sentences, unlike the MNREAD acuity charts where they vary. Also the Radner reading chart uses the arial font whereas the MNREAD acuity charts use times new roman. ¹⁴ Our results are also higher than those of Pesudovs et al., ¹² who reported CR for reading speed of ±16%, and Ahn et al., ¹⁵ who found a CR of ±18%. Further studies are needed to explore the reason for such differences and to determine whether position and length of words have an impact on repeatability. 

As far as we are aware, no research groups investigating the repeatability of visual measurements have looked at the precision of repeatability measures. Since we had a limited sample size in the present study, calculating precision provides additional information about the repeatability of the measurements. The precision of the CR appears to be good for both reading acuity and reading speed, with the 95% confidence intervals varying between 0.09 and 0.12 log units. The precision of the CR for critical print size varies from 0.16 to 0.23 log units and is poorer than that found for the other two parameters. 

Several studies have been conducted to investigate whether the CR is related to VA. Kiser et al. ¹⁶ looked at the correlation between distance VA and CR by dividing their patients into two groups: VA ranging between 20/200 and 20/500 and VA greater than 20/500. They did not find any differences in CR between the two groups of patients. Patel et al. ¹⁷ were also not able to establish any concrete evidence of the relationship between CR and distance VA, although some of their results suggested that a relationship may exist. Woods ²⁶ was able to establish a correlation between VA and CR. Visual acuity was reduced by artificial means in his study by the use of contact lenses, which may not be a good simulation of visual loss due to ocular disease. In the present study no relationship was found between VA and CR. These results should be interpreted with caution as group 1 had subjects whose range of visual impairment was quite narrow (most subjects had visual acuity levels close to 0.5 logMAR, and only three had VA of 0.12 logMAR or better), and group 2 had subjects with moderate visual impairment (eight subjects had VA between 0.6 and 1.0 logMAR and four subjects had VA between 1.02 and 1.30 logMAR). None of the subjects had severe visual impairment, which may have biased the results. The suggestion of a similar CR for patients with low versus high visual acuity should be tested in larger samples, to have more statistical power to support such a conclusion. 

Our study indicates that reading speed differences in subjects with visual impairment must be greater than 25% between sessions to indicate that a clinical change has taken place. Values lower than 25% are most likely to be due to measurement error. This finding has important implications for rehabilitation studies in which reading speed is used to determine whether a particular form of rehabilitation has actually improved reading ability. 

All the measurements performed in this study were conducted in a laboratory. The setting may have influenced the results, as Woods and Lovie-Kitchin ¹⁹ found that the CR for distance VA was higher when measurements were performed in a clinical setting (29 letters) when compared with a laboratory setting (17 letters). It may be that the CR for the reading parameters measured would be higher if the tests are conducted in a clinic. Further studies should be performed to verify this. 

There are some limitations to our study. A majority (n = 23) of subjects who took part had central visual field loss due to AMD or Stargardt’s disease. Repeatability may vary with the type of field loss. For example, subjects with retinitis pigmentosa with a concentric visual field loss may have a different CR than would subjects with AMD or Stargardt’s disease. The type of AMD may also influence repeatability measurements, particularly in subjects show have from neovascular AMD where the macular scotoma may be worse, causing greater variability in reading ability. In the present study, repeatability was measured on the same day using the patients reading glasses. Repeatability may be different if subjects use their low vision aids such as hand and stand magnifiers and may also be worse if measured over a longer testing interval, such as 1 to 2 weeks. 

We therefore recommend that future studies measure repeatability in subjects with a variety of ocular diseases such as retinitis pigmentosa and advanced glaucoma. Repeatability should also be measured both with and without the patient’s low-vision aids, and data should be collected over a 1- to 2-week testing interval, rather than on the same day. Other parameters that may affect repeatability, including the age of the patient, the patient’s level of education, and time spent reading, should also be investigated. 

 Submitted for publication September 3, 2008; revised January 21, February 26, and March 18, 2009; accepted June 17, 2009.

 Disclosure: A. Subramanian, None; S. Pardhan, 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.

 

Characteristics of Study Participants

No.	Age (y)	Condition	Distance Prescription	Add (OU)	Working Distance (cm)	DVA	RA	CPS	Log RS	RS (wpm)
1	60	Early macular changes	R: Plano	+2.50 DS	40	−0.04	0.01	0.25	2.32	211
						L: Plano
2	68	Early macular changes	R: −7.00/−1.25×90	+3.00 DS	40	0.10	0.10	0.15	2.28	191
						L: −7.00 DS
3	51	Early macular changes with extensive peripheral drusen	R: +0.25/−0.25×165	+3.50 DS	40	0.12	0.10	0.25	2.30	201
						L: −0.50/−0.50×45
4	78	Neovascular AMD OU	R: +3.00/−1.50×105	+2.00 DS	50	0.30	0.35	0.55	1.86	73
						L: +3.00/−2.00×85
5	77	Neovascular AMD OD, dry AMD OS	R: +1.50/−0.75×80	+3.00 DS	40	0.34	0.30	0.55	2.14	140
						L: +2.00/−1.50×95
6	22	Stargardt	Nil	Nil	45	0.42	0.61	0.95	2.35	228
7	65	Neovascular AMD OU	R: +1.50 DS	+4.00 DS	30	0.42	0.33	0.60	2.00	100
						L: +1.00 DS
8	72	Dry AMD OU	R: +1.75 DS	+2.50 DS	40	0.44	0.20	0.50	2.14	140
						L: +1.50/−0.75×80
9	82	Dry AMD OU	R: −2.50/−1.00×110	+2.00 DS	50	0.44	0.54	0.80	2.16	143
						L: −2.00/−1.25×80
10	71	Dry AMD OU	R: +3.25/−2.00×125	+2.50 DS	40	0.44	0.60	0.75	2.02	106
						L:+1.50/−0.75×60
11	81	Optic atrophy OU	R: −6.50/−1.75×100	+3.50	33	0.44	0.42	0.70	1.85	71
						L: −2.00 DS
12	75	Dry AMD OU	R: +1.75 DS	+2.75 DS	40	0.46	0.48	0.70	2.32	213
						L: +2.25/−2.00×25
13	77	Dry AMD OU	R: Plano/−1.50×110	+4.50 DS	25	0.50	0.69	1.10	1.92	84
						L: +2.00/−2.00×70
14	77	Neovascular AMD OU	R: +1.75 DS	+2.50 DS	40	0.50	0.30	0.40	1.56	37
						L: +0.75 DS
15	65	Dry AMD OU	R: +1.00/−1.00×150	+2.25 DS	45	0.60	0.79	1.00	2.03	107
						L: + 1.00/−1.00×140
16	78	Dry AMD OU	R: +2.75/−1.75×107.5	+2.50 DS	50	0.60	1.00	1.10	1.94	87
						L: +2.50/−1.50×45
17	75	Dry AMD OU	R: −0.75/−0.50×180	+5.00 DS	20	0.60	0.45	0.55	2.12	134
						L: −1.25/−0.25×180
18	74	Neovascular AMD OU	R: +4.25/−2.75×150	+6.00 DS	35	0.62	0.41	0.70	1.87	75
						L: +5.50/−3.00×50
19	89	Dry AMD OU	R: +4.75/−0.50×27.5	+2.50 DS	40	0.66	0.67	0.85	2.04	109
						L: +4.75 DS
20	75	Dry AMD OU	R: −0.25/−0.25×165	+4.00 DS	20	0.90	1.05	1.20	2.13	135
						L: −1.00/−0.50×45
21	70	Dry AMD OU	R: +0.25/−2.50×30	+3.00 DS	33	0.92	0.89	1.00	1.79	107
						L: +3.25/−2.50×150
22	78	Dry AMD OU	R: +0.75/−1.25×160	+2.75 DS	40	0.95	0.81	1.05	1.66	46
						L: −1.00/−0.50×180
23	82	Dry AMD OU	R: Plano	+3.25 DS	33	1.02	0.85	1.1	2.05	114
						L: +1.75/−1.00×70
24	80	Dry AMD OU	R: +1.25/−1.25×110	+2.00 DS	40	1.15	1.08	1.25	1.78	60
						L: +3.25 DS
25	75	Dry AMD OU	R: −6.50/−1.00×10	+5.50 DS	20	1.2	0.83	0.9	2.24	176
						L: −7.50/−0.75×170
26	84	Dry AMD OU	R: +4.25/−3.00×10	+3.75 DS	25	1.3	0.94	1.15	1.57	38
						L: +5.75/−1.75×170
27	65	Dry AMD OU	R: +3.00/−1.00×180	+3.00 DS	33	1.38	1.17	1.05	1.72	53
						L: +4.50/−0.75×170

Table 2.

 

Mean Scores from the MNREAD Charts

Parameter	Mean MNREAD1 (SD)	Mean MNREAD2 (SD)	Mean Difference (95% CI)	CR (% Change)	Precision of CR
Reading acuity (logMAR)	0.58 (0.32)	0.59 (0.32)	−0.01 (0.09 to −0.11)	±0.10 (+25 to −20)	0.08–0.12
Reading speed (log wpm)	2.01 (0.22)	2.00 (0.24)	0.02 (0.12 to −0.09)	±0.10 (+25 to −20)	0.08–0.13
Critical print size (logMAR)	0.76 (0.32)	0.80 (0.31)	−0.03 (0.16 to −0.23)	±0.20 (+58 to −37)	0.16–0.24

 CI, confidence interval.

Figure 1.

 

Repeatability of reading acuity. The mean of test–retest repeatability plotted against the differences between the test–retest scores. All data points lie at or within 2 SD of the mean, with some overlapping points.

Figure 2.

 

Repeatability of reading speed. The mean of test–retest repeatability plotted against the differences between the test–retest scores. All but one of the data points lie within 2 SD of the mean.

Figure 3.

 

Repeatability of critical print size (CPS). The mean of test–retest repeatability plotted against the differences between the test–retest scores. All but two of the data points lie within 2 SD of the mean, with some overlapping points.