Purpose. To investigate the relationship between clinical impairment measures and reading performance in a large population with age-related macular degeneration.
Methods. The following clinical measures were evaluated on 243 patients with age-related macular degeneration: better eye distance visual acuity (ETDRS chart); threshold near word reading acuity (Bailey-Lovie Word Reading chart); maximum reading speed and critical print size (MNREAD chart); letter contrast sensitivity (Pelli-Robson); and kinetic perimetry (Bjerrum screen) to determine the nearest non-scotomatous point to fovea (NNPF; in degrees) and the central scotoma area (mm2).
Results. Distance acuity correlated well to threshold near word acuity (r = 0.71), but word acuity was usually poorer. Critical print size was strongly related (p < 0.001) to near visual acuity (r2 = 0.31 and β = 0.47) and was poorer than threshold near word visual acuity by a mean difference of −0.41 (range, −1.10 to 0.34), which represents a mean acuity reserve of 2.5:1. On single regression, distance (p < 0.0001, r2 = 0.35, and β = −102.37) and near acuities (p < 0.0001, r2 = 0.52, β = −126.53), critical print size (p = 0.0001, r2 = 19, and β = 0.002), contrast sensitivity (p < 0.0001, r2 = 19, and β = 79.47), scotoma size (p = 0.006, r2 = 12, and β = −0.04), and NNPF (p = 0.001, r2 = 12, and β = −4.39) were all highly significantly related to reading speed although these predicted only a low percentage of variance. Best prediction of reading speed was obtained on multiple regression, where NNPF and near word acuity explained 60% of the variance (p < 0.0001).
Conclusions. Optimal prediction of reading speed with clinical parameters appears to be based on the combination of near word acuity and scotoma area, explaining 60% of the variance. Other factors not measured in this study are likely to account for the rest of the prediction.
‡DPhil, MCOptom, DipGlauc
Faculty of Life Sciences, The University of Manchester, Manchester, United Kingdom.
This research was supported by funding from The Health Foundation.
Received April 30, 2009; accepted December 21, 2009.
Reprint requests: Christine M. Dickinson, Faculty of Life Sciences, The University of Manchester, Moffat Building, PO Box 88, Manchester, M60 1QD; e-mail: email@example.com.
Age-related Macular Degeneration (AMD) is the leading cause of visual impairment in the developed world. Slow and difficult reading, as a result of a central scotoma, is the most common complaint among patients with AMD. This reduced reading ability has been the subject of many studies. Whittaker and Lovie-Kitchin1 reviewed a number of previous observations and predicted the minimum visual requirements for spot or survival reading at low speed [∼40 words per minute (wpm)] and high fluent or leisure reading (∼160 wpm). They concluded that reading cannot be achieved with threshold stimuli and that text needs to be significantly above threshold (that is, there must be a “reserve” of both acuity and contrast) to read optimally.
The presence of a significant central scotoma and the need to view with a non-foveal area seems to place an unavoidable restriction on reading speed, even in normally sighted individuals.2–4 Legge et al.5 found that low-vision patients with central field defects read words composed of 12 and 24° letters at a median speed of 25 wpm compared with 130 wpm for those without central field defects. Fletcher et al.6 concluded from their large group with AMD that readers with scotomas read at approximately half the rate of readers without such scotomata. Ergun et al.4 found that the size of the absolute scotoma measured by means of microperimetry, correlated significantly with reading speed in patients with neovascular AMD. In the presence of a scotoma, the reading speed is affected by the patients ' ability to adjust their eye position so that images of interest fall on a preferred area of functioning parafoveal retina.7 This technique is called eccentric viewing, and the new area of the retina where the patient fixates is called the preferred retinal location/locus (PRL) or pseudofovea.8 Eccentric fixation differs from eccentric viewing in that patients have the sensation of looking directly at the visual object during fixation with the eccentric PRL; i.e., the PRL has also replaced the fovea as the reference point for the oculomotor system. Because resolution decreases away from the fovea due (at least in part) to the crowding effect,9 it is expected that the PRL would be on the area of peripheral healthy retina that is closest to the fovea. However, it is known that in an individual patient, the location of the PRL could vary with the task,10,11 and may well be “chosen” depending on the visual abilities required to perform different activities.
Most of these previous findings were based on relatively small numbers of patients. We wanted to test the theoretical predictions, which had been made about reading ability by investigating the relationship between clinical impairment measures and reading performance in a large group of patients.
The characteristics of the study population have been described previously.12 In brief, a total of 243 patients with binocular AMD, 157 females (65%) and 83 males (35%), with a median age of 81 ± 7 years (range, 56 to 97 years) and a median duration of the condition of 3.0 years (range, 0.3 to 40 years) were recruited. The Central Manchester Local Research Ethics Committee gave the ethical clearance for the study (Reference number: CEN/00/150). Informed consent was obtained from each subject before testing, and the study complied with the principles set out in the Declaration of Helsinki.
The following clinical measures were obtained on two visits. All measures were taken when the patient was wearing the optimal prescription for each working distance.
At the first visit, all the clinical measures were only measured for the better eye. These included, better eye distance visual acuity (VA) at 4 m (ETDRS chart: Precision Vision, La Salle, IL), threshold near word reading acuity at 25 cm (Bailey-Lovie Word Reading Chart13), and letter contrast sensitivity (Pelli-Robson chart: Not allowing C for O misreading) were measured. Kinetic perimetry (Bjerrum screen) was used to quantify the central scotoma. The Bjerrum screen was modified with a white cross fixation target across the full chart. The tester constantly observed fixation, and the edge of the scotoma was determined repeatedly to ensure that it was consistent. A 4-mm white target was used and moved from seeing to non-seeing. The resultant field plot was overlaid with a grid marked in 1° squares, and the size of the scotoma (in mm2) measured. The position of the nearest non-scotomatous point to the fovea (NNPF) was also determined from the plot.
At a second visit, which was approximately 1 month later, the clinical measures were taken binocularly. These were distance visual acuity (ETDRS chart), maximum reading speed and critical print size (the smallest print size at which patients can read with their maximum reading speed) were measured. The latter two values were obtained using the MNREAD chart following the manufacturer's instructions (Precision Vision) at the reading distance of the patient (corresponding to the reading addition that they wear on their near vision glasses) and then the critical print size was converted to the equivalent at the standard 40-cm working distance for which the charts were designed.
ETDRS distance VA was measured at both visits to ensure that the condition had not progressed: the first visit measurement was used for the analysis. A t-test revealed no significant statistical difference between the first visit better eye VA and the second visit binocular VA (p = 0.399). According to Kabanarou and Rubin,14 the differences between monocular and binocular VA do not exceed the test-retest variability of the chart, which permitted us to correlate the monocular measures with the binocular reading performance.
Correlations, simple regression, and multiple regression analyses were performed to test the association among the different clinical measures from the better distance VA eye. We tested the association between distance and near visual acuities by means of Pearson correlation. The degree to which near VA and reading speed can predict critical print size was studied by means of simple regression analysis. Similarly, the degree to which contrast sensitivity, NNPF, and distance VA predict reading speed was studied by means of simple regression analysis. Finally, by means of multiple regression, we looked for the best-fitting regression to predict reading speed. The interactive model was used for this purpose, and we followed the 8 steps recommended by Hosmer and Lemeshow15 that cover the regression diagnosis and the multiple regression modeling. All analyses were carried out using SPSS version 16.
VA in AMD
Distance and Near Acuity
The mean VAs in our study for distance letter and near word were 0.93 logMAR (range, 0.42 to 1.60) and 1.03 logMAR (range, 0.38 to 1.60), respectively. Our mean VA value compares well with that of Fletcher et al.,6 who found 0.97 logMAR. Distance acuity shows a strong linear relationship (p < 0.0001) with near acuity (r2 = 0.56 and β = 0.47), but word acuity was usually poorer (i.e., larger print size) (Fig. 1) by a mean difference of −0.1 logMAR (range, −0.80 to 0.49): This difference seems to be more obvious for better acuity cases and decreases as the acuity deteriorates.
Critical Print Size and Acuity Reserve
Critical print size is poorer than threshold near word VA by a mean difference of −0.41 logMAR (range, −1.10 to 0.34). This indicates the acuity reserve required to achieve maximum reading rate: mean acuity reserve was found to be 2.5:1, although the maximum value was 11.6:1. This difference is more obvious for better acuity levels and decreases as the acuity deteriorates (Fig. 2). The mean critical print size in our study was 1.43 logMAR (range, 0.54 to 2.20), which compares with the mean value of Fletcher et al.6 who found 1.20 logMAR, and it has a strong linear relationship (p < 0.001) with near word VA (r2 = 0.31 and β = 0.47). Diagonal trendlines corresponding to acuity reserves of 1:1, 3:1, and 6:1 are shown in Fig. 2.
Reading Speed in AMD
Reading Speed and Critical Print Size
Critical print size has a strong linear relationship with reading speed (p < 0.001) although it only predicts a small amount of the variance (r2 = 0.19 and β = 0.002) (Fig. 3).
Reading Speed and Contrast Sensitivity
The mean values for reading speed and contrast sensitivity in this study were 73 wpm (range, 0 to 229) and 0.96 logCS (range, 0.05 to 1.65) respectively. Our mean reading speed value is lower than that of Fletcher et al.6 who found 112 wpm. Contrast sensitivity has a strong linear relationship with reading speed (p < 0.001), although it only predicts a small amount of the variance (r2 = 0.19 and β = 79.47). Fig. 4 shows the contrast sensitivity plotted against reading speed. Given that MNREAD print contrast is 85% (as specified by the manufacturer), the dashed line shows the reading speeds predicted by Whittaker and Lovie-Kitchin1 for 3:1, 4:1, and 10:1 contrast reserve. Although some patients with very poor contrast sensitivity were able to read print faster than predicted, the majority were considerably slower. The data obtained by Kabanarou and Rubin14 are also plotted in Fig. 4 and show an even shallower regression line.
Reading Speed and Nearest Nonscotomatous Point to Fovea
Reading speed decreases as the scotoma size increases (Fig. 5), and NNPF and scotoma area have a strong linear relationship with reading acuity (p = 0.001 and 0.006, respectively), although they each only seem to predict 12% of the variance (r2 = 0.14, β = −4.39 and −0.04 respectively). The mean value of NNPF in the study was 3° (range, 0 to 15) and the mean area of the scotoma was 64°.2 The Whittaker and Lovie-Kitchin1 predictions give an accurate assessment of maximal performance but the values obtained by Wensveen et al.3 using simulated scotomas in an elderly population seem to give a better guide to average performance when plotted on the same graph and compared with our data.
Reading Speed and Distance Acuity
LogMAR distance letter acuity has a strong linear relationship (p < 0.001) with reading speed (r2 = 0.35 and β = −102.37). Fig. 6 shows the relationship between distance acuity and reading speed; plotted in the same way as Pesudovs et al.,16 with the two lines being not far apart. Interestingly, near VA explains reading speed better (r2 = 0.52, p < 0.0001, and β = −126.53) than does distance VA.
Multiple Regression to Predict Reading Speed
Near word VA, critical print size, contrast sensitivity, NNPF, and scotoma area were all tested as independent variables in the model. Distance letter VA was highly correlated to near word VA (r = 0.71 and p < 0.0001) and thus both provided the same information. From these two variables, we discarded distance VA because on single regression this variable did not explain reading speed (r2 = 0.35) as well as near VA (r2 = 0.52). The rest of the correlations between the independent variables were all low to moderate (range 0.17 to 0.57) and thus all five were tested. The results for the best-fitting model are summarized in Table 1. The variables giving the best fit were near VA and scotoma area, explaining 60% of the reading speed (adjusted r2 = 0.60 with a significance of p < 0.0001). These results were obtained after removing eight observations that were considered outliers (the 95% CIs of their corresponding regression residuals did not include the 0 value). Next we obtained a good fit to the normal line of the P-P plot and well-distributed values on the residuals vs. fitted plots.
VA in AMD
In this study, word acuity (near VA) was usually poorer than letter acuity (distance VA) by the equivalent of a line of acuity on the ETDRS chart. Perhaps due to the scotoma, reading words is more difficult than letters as most points are above the line of equal acuity. This finding is not surprising as word VA would suffer from a crowding effect that would not be present on single letter VA.17 We also noticed that this difference is more obvious for patients with better acuity and that it decreases as the acuity deteriorates, so that for distance VAs of 1.40 logMAR or worse, there is no difference between the two. It might be argued that as the letters get larger, the crowding effect attenuates. From a statistical point of view, this effect of the more extreme cases (best and worst VAs) falling toward the middle of the range is most likely to be due to the well-known regression toward the mean.18
A further increase in the size of the words is necessary before they can be read at maximum speed, i.e., the critical print size and the increase relative to threshold is equal to the acuity reserve. Our data show that the critical print size is an average of four lines larger than threshold acuity on the MNREAD chart and has a mean value of 1.43 logMAR, which is comparable with the median 1.30 logMAR found by Legge et al.5 in their 16 low-vision patients, although they used unrelated words. However, we had a few cases with better critical print size than threshold VA, probably due to some performance differences between the Bailey-Lovie near VA and MNREAD charts. The mean acuity reserve found was quite similar (2.5:1) to that predicted by Whittaker and Lovie-Kitchin1 who estimated that for high fluent reading rates (∼150 wpm) the acuity reserve should be between 2:1 and 3:1. This 2.5:1 relationship appears to apply better for the poorer values of critical print size, but the individuals with a larger critical print size, who do not read quickly (i.e., they never achieve high fluent reading), have a lower acuity reserve.
Reading Speed in AMD
In this study, reading speed is very significantly related to near word VA. This finding suggests that increasing magnification for patients with poor acuity does not increase reading speed as it does for patients with better acuity. There is obviously another factor limiting the reading speed: an associated contrast sensitivity loss; the central scotoma; the difficulty in making eye movements across the enlarged text; or a reduced visual span are all possible explanations.5,13,19
In this study, contrast sensitivity was related to reading speed in our large group of longstanding cases (p < 0.001), although it only predicts a small amount of the variance (r2 = 0.19 and β = 79.47). Although some patients with very poor contrast sensitivity were able to read print better than predicted, the majority were considerably worse. This finding suggests that factors other than contrast sensitivity are more significant in determining reading performance at higher levels of contrast sensitivity. Contrast sensitivity therefore, only seems to be a limiting factor when it is very low. The low association that we found is quite similar to that of Bullimore and Bailey20 who found a correlation of r = 0.30, but it is less than correlations (r = 0.61 and 0.62) found in other studies.21,22 These differences could be due to the limited number of patients that took part in these studies. We also need to bear in mind that Bullimore and Bailey2 used a different technique for measuring contrast sensitivity (a Joyce Electronics oscilloscope with vertical sinewave gratings) from the one in present study (Pelli-Robson low contrast letter chart). Finally, the study of Bullimore and Bailey20 was the only one performed exclusively on patients with AMD, whereas the other two studies had subjects with various eye conditions that may affect the reading speed in different ways. Kabanarou and Rubin14 plotted Pelli-Robson contrast sensitivity against reading speed when investigating binocular gain and their findings for better VA eyes shows little reading speed variation across the different contrast sensitivities. This finding might be explained by their subjects having better VAs (0.3 to 1.0 logMAR) than in this study. It might also be argued that the different method used by these authors to calculate the reading speed (i.e., by means of an eye tracker and averaging reading speed 10 presentations repeated on a second visit) may have provided a training opportunity, which could account for their better reading speed range.
Scotoma size and NNPF were related to reading speed in this study, even though they each only seem to predict 12% of the variance. One possible explanation for this low r2 value could be because the better eye was solely defined by VA and it could have a larger scotoma or poorer CS (e.g., in a ring scotoma) so that the binocular reading performance was actually due to the eye with poorer VA. Although we did not have a microperimeter available in our clinical setting, our data agree well with those of Ergun et al.4 who also found a good association between reading speed and absolute scotoma size (p = 0.02) when measured with a microperimeter. Although microperimetry is nowadays the gold standard technique for studying the PRL location and behavior, the exact location of the fovea for these patients still cannot be precisely identified on the damaged macula, and this issue is a difficult problem to overcome with any visualization technique. If our patients had less severe disease with smaller scotomas, we could have attempted to plot the position of the physiological blind spot to monitor fixation in a similar way to Mesias et al.23 However, many of our cases had large scotomas that were contiguous with or encompassed the blind spot, thereby preventing accurate location. Comparing our reading values with those from Wensveen et al.3 using simulated scotomas, our drop of reading speed is not as steep as theirs. This latter finding is likely to be due to the different methods used for measuring reading speed. We used the conventional MNREAD charts while they used a dedicated programmed computer display of letters. Another reason for the differences in reading speed could be that our patients were longstanding cases of AMD and thus they might, arguably, have had time to adapt to the condition to enable reading with a central scotoma. Our finding of the lower extreme of NNPF range being zero means that those patients think that they are still fixating centrally (i.e., the oculomotor center and the PRL has moved) while the others must make a conscious effort to look to one side, and so still know how to fixate centrally when asked to do so. Using a PRL is typical of longstanding AMD (according to a study of Cross-land et al.,24 1 year after the onset of AMD >50% of patients have already learnt to use a PRL). Our data and those of Wensveen et al.,3 show slower reading speeds across the wide range of VAs when compared with the predictions of Whittaker and Lovie-Kitchin.1
Similar to previous studies,16 our data show that reading speed is strongly related to distance VA (p < 0.001), although it only seems to predict a small amount of the variance (r2 = 0.35). On the other hand, near word VA seems to better explain reading speed with an adjusted r2 = 0.52 (p < 0.0001), and it may be that the word acuity test used for near VA testing is a better predictor than the letter acuity test used for distance VA. This weak variance on distance VA would be due to large intersubject variability and may explain why studies with smaller samples of subjects with AMD have found stronger associations between logMAR VA and reading speed.
On multiple regression, the best-fitting model to explain the reading speed predicted 60% of the variance. The two independent variables found to significantly contribute to this prediction model were near word VA and scotoma area. The r2 value of 0.60 is modestly higher than that of the near word VA alone (r2 = 0.52) and shows that scotoma area does contribute to predicting reading speed. However, because the main contribution to reading speed is given by near word acuity this easy clinical measure could be used in practice to predict the reading speed of the patient. Other factors not measured in the study, such as cognitive and educational factors,25 slower temporal processing in the peripheral retina,19 fixation stability,26 reduced span in peripheral retina,27 and poor eye movement control,28 are well known to affect reading speed and are likely to explain the remainder of variance in reading speed.
Our descriptive data on VA, CPS, and reading speed concur well with the results for another large study of patients with AMD6 and thus seems to provide evidence for this relationship characterizing reading performance in the population with AMD. Theoretical models of reading performance in the presence of a central scotoma give useful guide to clinical expectations. The present data on a large unselected sample of subjects with AMD attending a low-vision clinic provided a unique opportunity to test these predictions. The findings show that distance letter and near word visual acuities are correlated and that near word VA and reading speed have a strong linear relationship with critical print size on simple regression. In addition, contrast sensitivity, scotoma size, and near and distance visual acuities are all strongly associated with reading speed on single regression but are not necessarily good predictors of reading speed on their own because a large proportion of the variance was unaccounted for in the analyses. Multiple regression does however account for the highest prediction of reading speed, i.e., in the form of near word VA and scotoma area.
The authors thank all the participants who took part in the study.
Plaza Pere Garau
20 3B, Palma de Mallorca 07007
1.Whittaker SG, Lovie-Kitchin J. Visual requirements for reading. Optom Vis Sci 1993;70:54–65.
2.Cummings RW, Whittaker SG, Watson GR, Budd JM. Scanning characters and reading with a central scotoma. Am J Optom Physiol Opt 1985;62:833–43.
3.Wensveen JM, Bedell HE, Loshin DS. Reading rates with artificial central scotomata with and without spatial remapping of print. Optom Vis Sci 1995;72:100–14.
4.Ergun E, Maar N, Radner W, Barbazetto I, Schmidt-Erfurth U, Stur M. Scotoma size and reading speed in patients with subfoveal occult choroidal neovascularization in age-related macular degeneration. Ophthalmology 2003;110:65–9.
5.Legge GE, Rubin GS, Pelli DG, Schleske MM. Psychophysics of reading. II. Low vision. Vision Res 1985;25:253–65.
6.Fletcher DC, Schuchard RA, Watson G. Relative locations of macular scotomas near the PRL: effect on low vision reading. J Rehabil Res Dev 1999;36:356–64.
7.von Noorden GK, Mackensen G. Phenomenology of eccentric fixation. Am J Ophthalmol 1962;53:642–60.
8.Timberlake GT, Mainster MA, Peli E, Augliere RA, Essock EA, Arend LE. Reading with a macular scotoma. I. Retinal location of scotoma and fixation area. Invest Ophthalmol Vis Sci 1986;27:1137–47.
9.Pelli DG, Tillman KA, Freeman J, Su M, Berger TD, Majaj NJ. Crowding and eccentricity determine reading rate. J Vis 2007;7:20.1–36.
10.Duret F, Issenhuth M, Safran AB. Combined use of several preferred retinal loci in patients with macular disorders when reading single words. Vision Res 1999;39:873–9.
11.Deruaz A, Whatham AR, Mermoud C, Safran AB. Reading with multiple preferred retinal loci: implications for training a more efficient reading strategy. Vision Res 2002;42:2947–57.
12.Smith HJ, Dickinson CM, Cacho I, Reeves BC, Harper RA. A randomized controlled trial to determine the effectiveness of prism spectacles for patients with age-related macular degeneration. Arch Ophthalmol 2005;123:1042–50.
13.Bailey IL, Lovie JE. The design and use of a new near-vision chart. Am J Optom Physiol Opt 1980;57:378–87.
14.Kabanarou SA, Rubin GS. Reading with central scotomas: is there a binocular gain? Optom Vis Sci 2006;83:789–96.
15.Hosmer DW, Lemeshow S. Applied Logistic Regression, 2nd ed. New York: Wiley; 2000.
16.Pesudovs K, Patel B, Bradbury JA, Elliott DB. Reading speed test for potential central vision measurement. Clin Experiment Ophthalmol 2002;30:183–6.
17.Whatham AR, Mermoud C, Deruaz A, Goldschmidt M, Zesiger P, Safran AB. Computer-based measurement of letter and word acuity. Ophthal Physiol Opt 2006;26:156–68.
18.Bland JM, Altman DG. Regression towards the mean. BMJ 1994;308:1499.
19.Cheong AM, Legge GE, Lawrence MG, Cheung SH, Ruff MA. Relationship between slow visual processing and reading speed in people with macular degeneration. Vision Res 2007;47:2943–55.
20.Bullimore MA, Bailey IL. Reading and eye movements in age-related maculopathy. Optom Vis Sci 1995;72:125–38.
21.Rubin GS, Legge GE. Psychophysics of reading. VI. The role of contrast in low vision. Vision Res 1989;29:79–91.
22.Leat SJ, Woodhouse JM. Reading performance with low vision aids: relationship with contrast sensitivity. Ophthal Physiol Opt 1993;13:9–16.
23.Messias A, Reinhard J, Velasco e Cruz AA, Dietz K, MacKeben M, Trauzettel-Klosinski S. Eccentric fixation in Stargardt's disease assessed by Tubingen perimetry. Invest Ophthalmol Vis Sci 2007;48:5815–22.
24.Crossland MD, Culham LE, Kabanarou SA, Rubin GS. Preferred retinal locus development in patients with macular disease. Ophthalmology 2005;112:1579–85.
25.Friedman SM, Munoz B, Rubin GS, West SK, Bandeen-Roche K, Fried LP. Characteristics of discrepancies between self-reported visual function and measured reading speed. Salisbury Eye Evaluation Project Team. Invest Ophthalmol Vis Sci 1999;40:858–64.
26.Crossland MD, Culham LE, Rubin GS. Fixation stability and reading speed in patients with newly developed macular disease. Ophthal Physiol Opt 2004;24:327–33.
27.Legge GE, Ahn SJ, Klitz TS, Luebker A. Psychophysics of reading. XVI. The visual span in normal and low vision. Vision Res 1997;37:1999–2010.
28.White JM, Bedell HE. The oculomotor reference in humans with bilateral macular disease. Invest Ophthalmol Vis Sci 1990;31:1149–61.