Low vision is defined as an uncorrectable loss of vision that restricts affected patients in their everyday life. The International Classification of Diseases assumes moderate to severe vision impairment with visual acuity below 6/18 but better than 3/60 in the better eye with the best possible correction ( 32 ). According to estimations of 2010 globally 191 million people suffer from low vision ( 28 ). Major causes for vision loss are cataract, age related macular degeneration, diabetic retinopathy, glaucoma or refractive error ( 2 ).

Reading impairment is a main complaint of patients suffering from low vision. Hence improving reading speed is the primary therapeutic goal ( 6 ). Low vision therapy has a long-standing tradition and is widely pro-vided. Its aim is it to enable and support low vision pa-tients in reading and other daily activities. In order to decide which low vision aids are beneficial for an individual patient, near and distance visual acuity, contrast sensitivity and visual field are taken into account. Usually however it is difficult to predict whether and how much an individual patient may benefit from low vision therapy ( 15, 17, 31 ). This may be due to the scarce evidence from studies about the efficacy of low vision treatment, which has been examined, with some exceptions ( 26, 27 ), in small observational studies ( 29 ). Improvement of reading speed ( 20, 21 ) and quality of life ( 11, 14 ) could be found. In a randomized trial Stelmack et al. showed that training and instruction enabled significantly better performance than provision of low vision aids alone ( 22, 27 ). The great majority of studies focused on patients with macular disease causing a central scotoma. However low vision may have a great variety of causes, which raises the question whether all kinds of low vision respond equally to low vision support, irrespective of the cause.

To investigate this we compared the effect of magnification and contrast enhancement on reading performance in three different types of low vision. Measurements were performed in healthy subjects with a computer based simulation of low vision diseases. This approach gave us the possibility to compare directly and within subjects the different low vision types and furthermore we could avoid bias induced by comorbidities that may be found in real low vision patients. Low vision types tested in this study were simulated blurred vision, simulated oscillopsia and simulated central scotoma.

The calibration of the low vision condition was chosen such that a letter requiring a Snellen visual acuity of about 20/100 was just visible (see methods for details). To achieve this with a central scotoma, a blanked area with a diameter of 3.16° (range 2.5° – 4.0°) was necessary. For the same vision a blur radius of 329 seconds of arc (range 292 – 438 seconds of arc) was necessary. To induce the same reduction of vision a horizontal movement speed of 1.3Hz (range 1.0 – 1,6Hz) and a vertical movement speed of 2.45Hz (range 2.0 – 2.9Hz) was required. Next we measured reading speed and we found that in all types of simulated low vision reading speed improved with magnification (oscillopsia: F(5,55)=82, p<0.001, BIC=284; blur: F(5,55)=90, p<0.001, BIC=308; central scotoma: F(5,50)=55, p<0.001, BIC=236; figure 2) and in all low vision conditions reading speed improved with better contrast (oscillopsia: F(4,44)=30, p<0.001, BIC=249; blur: F(4,44)=65,p<0.001, BIC=238; central scotoma: F(4,40)=16, p<0.001, BIC=213, figure 3). In contrast to subjects experiencing low vision, the reading speed in subjects with normal vision did neither change with contrast (F(4,12)=0.9, p=0.5, BIC=69) nor with magnification (F(5,15) = 2.4, p = 0.1, BIC=77). At the second but lowest magnification and contrast condition reading speed was similar in all three conditions (magnification: F(2,33)=0.1, p=0.9, BIC=167; contrast: F(2,33)=1.0, p=0.4, BIC=154), indicating that all types of low vision led to a comparable impairment of reading at this particular contrast and magnification. We found that the reading speed with the highest magnification was similar to the reading speed measured in normal viewing conditions (F(1,37)=3, p=0.1, BIC=177) and it was significantly better than the reading speed with the highest contrast (F(1,68)=72, p<0.001, BIC=328). The reading speed with the latter was significantly below the reading speed measured in normal viewing conditions (F(3,35)=9.8, p<0.001, BIC=176).

Importantly we found a significant interaction term of magnification and low vision condition (F(2,193)=8.2, p<0.001,BIC=947), showing that the effect of magnifica-tion on reading is significantly different among the three conditions: Little increase in magnitude led to the most improvement of reading speed if low vision was induced by blur, where the benefit plateaued at a font size of 28 pixels. In contrast, low vision induced by a central sco-toma showed a more gradual increase of reading speed with increased magnitude, while the effect of magnifica-tionon low vision induced by text movement was at an intermediate level. Unlike for magnification we found that the interaction term of contrast and low vision condition showed no significance, indicating that the benefit of reading speed from enhanced contrast is not significantly different between the three low vision conditions (F(2,158)=0.9, p=0.39, BIC=897).

We found that artificially induced low vision led to reduced reading speed that could be improved with both magnification and better contrast. We found a near normal reading speed with the highest magnification but not with the best contrast in all types of simulated low vision. Importantly we found that the effect of magnification significantly depended on the type of low vision: The best effect of magnification was found for blur associated low vision, followed by low vision associated with text motion. The least effect was found for subjects with a simulated central scotoma. Contrast enhancement on the other hand, showed benefits that were independent from the type of vision loss.

Even though the reading speed with small text size was not different in the three conditions, the same increase in magnification led a bigger effect for blur associated low vision than for central scotoma and nystagmus. This observation is to our knowledge unprecedented and the reasons for this are unknown. We speculate that the rate limiting factors for reading may be different in the three conditions and these factors may respond differently to magnification. For example, two-point discrimination may be the rate limiting step in the blur condition, which prevents to differentiate the letter “U” from the letter “O” for example. A small increase in size will enable this differentiation and thus facilitate reading. In contrast to blur, the number of obscured letters may be critical for reading with a central scotoma. Possibly the magnification required two recognize more letters of a word is more than the magnification required to improve two-point discrimination. Similar speculations could be made for oscillopsia and reading with nystagmus.

Where data is available, our results correspond well with findings from patients, where too a benefit on reading speed was found with magnification( 18 ). While we found a near normal reading speed with the highest magnification for all low vision conditions, Legge et al. found near normal reading speed only in patients with intact central vision while patients with loss of central vision only reached a median maximal reading speed of 25 words/minute( 18 ). Letters with font size between 12° and 24° were used in the latter condition, while the font size was about 2° in our study. Thus our results may only be applied to patients without loss of central vision as an optimal font size beyond 2° is usually associated with reduced maximal reading speed( 16 ). The finding of near normal reading speed with the highest simulated magnification in all simulated low vision types confirms our previous finding of near normal reading speed in both patients with nystagmus and simulated nystagmus( 5 ). In another study by Barot et. al. about reading speed in 71 patients with infantile nystagmus normal reading speed was found for most patients if texts were provided with optimal font size. Only the group of patients (n=12) with visual acuity below or equal 20/80 showed slightly reduced reading speed( 1 ). Again, all these findings indicate that up on a certain amount of visual impairment reading speed is not limited if text is presented in optimal conditions( 18 ).

Also with regard to contrast our data agree with data obtained from patients. We too found an improved reading speed with increased contrast as others had found in patients( 9, 23 ). Data from patients showed that the effect of contrast enhancement depended on individual contrast sensitivity and that it is independent of the type of low vision( 23 ). These clinical results correspond well to our finding of an absent effect of the type of low vision on contrast enhancement. Giacomelli et al. found that contrast reduction at a given text size led to more pronounced impairment of reading in patients with advanced low vision than in patients with mild low vision( 9 ), thus suggesting that contrast depends on the amount of visual loss. We found that contrast affected reading speed barely if no low vision was present. In a low vision situation, i.e. about 20/100, reduced contrast affected reading significantly however. Both findings indicate that patients with low vision have a smaller bandwidth of contrast where optimal reading is possible than subjects with normal vision.

Simulating visual impairment, especially simulating central field loss with the technique of gaze contingent display has been used by many researchers in the past. Research has been conducted on visual search with central scotoma( 8, 30 ), visual sensitivity in peripheral vision( 13 ) or on reading with central scotoma( 3, 10 ). There is virtually no literature on the effect of magnification and contrast on reading in simulated low vision however. In one study of Fine & Rubin reading speed was measured with different magnifications in three subjects with simulated central scotoma, simulated cataract and both combined( 7 ). They found that if the two low vision conditions were combined, more magnification was needed to reach the same reading speed as when only one low vision condition was simulated. Similar to our results, they also found that simulated cataract had almost no impact on reading performance with large letters.

The simulated low vision as used in the current paper has limitations which make a direct comparison with affected patients difficult. The simulated conditions differ from affected patients in several respects. First, in our conditions subjects are tested in an unadapted state, ie. subjects had no time to adjust to the situation. It is possible that reading speed in conditions of a central scotoma would improve over time, i.e. once subjects develop a new preferred retinal locus or learn to optimally use a pseudofovea (for an overview see for example ( 19 )). Second, we tested in a group of young subjects whereas affected patients are usually older. Third, low vision from ophthalmic disease itself is not directly comparable to the simulated low vision. The shape of the central scotoma is not a circle in real life, a nystagmus has a motor component, which probably interacts with the sensory component of oscillopsia and blur in a patient may not be gaussian but may rather result from higher order optic aberrations. All this prevents a direct and quantitative comparison of our findings with data from patients. And yet we are convinced that the use of a simplified model allows correct qualitative data which also apply to patients and thus our main conclusion that different causes for low vision respond differently to low vision aids is valid. To prove this in a group of patients will be difficult though because different types of low vision are associated with different types of age, genetic background and comorbidities thus making an investigation of the low vision aid effect alone difficult.

Taken together, our data strongly support the use of magnification and contrast enhancement in all types of low vision. Our results suggest that for a given visual acuity and a given visual impairment the benefit of low vision therapy depends on the cause. This advocates measures that are adjusted to the origin of low vision, which today is achieved empirically( 15, 18 ).

This study was supported by the Swiss National Science Foundation. The authors report no conflicts of interest and have no proprietary interest in any of the materials mentioned in this article.