Further Validation of Comfortable Print Size as a Parameter for Clinical Low-Vision Assessment

Abstract

Purpose: Comfortable print size (CfPS) has been proposed as a clinical alternative to deriving critical print size (CPS) in the assessment of reading function of vision-impaired patients. This study aimed to assess the repeatability of CfPS and to compare assessment duration and values to CPS measures and acuity reserves.

Methods: Thirty-four adults with vision impairment had their reading function assessed. Two assessments of CfPS were made by asking, “What is the smallest print size that you would find comfortable using?” Reading parameters including CPS were determined using the MNREAD card chart and MNREAD app.

Results: CfPS was quicker to assess (mean ± SD, 144 ± 77 seconds) than the MNREAD card (231 ± 177 seconds) or app (285 ± 43 seconds). Within-session repeatability of CfPS showed no significant bias or variation across the functional range and limits of agreement (LoA) of ±0.09 logMAR. CfPS values were 0.10 logMAR larger than card CPS values, but no different from app CPS values, with LoA of ±0.43 to 0.45 logMAR. Acuity reserve (comparing CfPS to card reading acuity) was 1.9:1 on average, with a maximum of 5.0:1.

Conclusions: CfPS offers a quick, repeatable, and individualized clinical measure of the print size required for sustained reading that reflects CPS values obtained by more traditional measures.

Translational Relevance: CfPS is an appropriate clinical measure of reading function to use in determining the magnification requirements of vision impaired patients for sustained reading tasks.

Introduction

Globally, it was estimated that in 2020 there were 596 million people with vision impairment (presenting visual acuity worse than 6/12 or visual field of less than 10°) and a further 510 million with uncorrected presbyopia.¹ The most important and difficult task for people with vision impairment is often reading,² and providing magnifying low-vision aids to bridge the gap between current and desired reading function is a common outcome of a low-vision assessment. Effective and efficient clinical assessment of reading function is therefore fundamental to low-vision assessment. 

Functional reading cannot be achieved at the reading acuity threshold.³ Typically, a print size of 1.3 times the size of threshold reading acuity is recommended for “spot” or “survival” reading tasks of a few words (e.g., reading a medicine label or instructions on a packet).⁴ For “fluent” or “sustained” reading for longer periods, an acuity reserve of two to three^4–8 times threshold reading acuity is recommended. Because individual acuity reserves vary considerably,^7,8 determination of the critical print size (CPS), or the smallest print size that supports the maximum reading speed,^9,10 provides an individualized measure of the print size required for optimum reading speed by a person with vision impairment.^4,11 

CPS is not straightforward to assess, however. It requires either recording of reading time across a range of print sizes using a card reading chart¹² or the use of an app to semiautomate the process.¹³ Derivation of CPS, then requires interpretation using one of a variety of methods, including inspection of plotted curves of reading speed as a function of print size,¹⁴ selecting the size reflecting some proportion of maximum reading speed,¹⁵ or curve fitting to the reading plot.^13,16 CPS has consistently lower repeatability than other measures of reading function such as reading acuity (RA) or maximum reading speed (MRS).^15,17,18 Coefficients of repeatability for CPS using MNREAD charts with people with vision impairment have varied from a best case of ±0.20 logMAR¹⁷ to ±0.67 logMAR.^15,18 The poor repeatability and potential subjective examiner interpretation involved in CPS has led authors to suggest that CPS is not a pure measurement¹⁹ and should be considered more of a clinical measure.²⁰ The complexity of CPS assessment is therefore a barrier to its use in clinical practice. Nonetheless, CPS as a measure of reading function has high face validity, as it indicates the smallest print size at which maximum or near-maximum reading speed can be achieved and therefore the print size that is likely to be optimum for sustained reading for a person with vision impairment.¹¹ 

Rather than using examiner interpretation of what constitutes the CPS, comfortable print size (CfPS) uses patient interpretation of what print size a person would consider suitable for sustained reading.²¹ Asking people with vision impairment to identify the print size that they would find comfortable to read provides an individualized subjective measure that relates well to CPS.²¹ 

Because it has been shown that patient identification of CfPS shows promise as a clinical measure, the purpose of this study is to present further parameters of CfPS that would be relevant to its adoption in clinical practice. These include repeatability and duration of assessment, the reliability of CfPS when compared to assessments of CPS other than those previously examined,²¹ and the acuity reserve associated with CfPS judgments. 

Methods

Participants

Participants 18 years of age or older with a vision impairment that they considered affected their daily life were recruited to the study from the Low Vision service at Anglia Ruskin University Eye Clinic (Cambridge, UK) and from the volunteer organization CamSight. All participants gave informed consent when the nature and possible consequences of the study had been explained. Ethical approval for the study was received from the Anglia Ruskin University Faculty of Science and Engineering Ethics Panel (reference 0422-01). Participants were excluded if they did not successfully complete the Brief Interview for Mental Status.^22,23 

Demographic details of age, sex, ethnicity,²⁴ visual impairment registration status,²⁵ and duration that vision loss has been affecting everyday life were requested. In the United Kingdom, people with a full visual field can be registered as sight impaired with best-corrected visual acuity of 1.0 to 1.3 logMAR, or as severely sight impaired if worse than 1.3 logMAR. Registration in both categories is also available to those with better acuity and poorer visual field status.²⁵ Self-reported primary cause of visual loss was assigned to categories utilized on the Certificate of Visual Impairment.²⁵ Habitual refractive correction was determined using focimetry. 

Visual Function Assessment

Distance high-contrast visual acuity was assessed binocularly with habitual distance refractive correction using a 3-meter back-lit Early Treatment of Diabetic Retinopathy Study (ETDRS) chart in a well-lit clinical room. Assessment was terminated when four mistakes were made on a line, with logMAR acuity calculated on a letter-by-letter basis. Contrast sensitivity was assessed binocularly with habitual distance refractive correction using a Pelli-Robson chart at 1 meter in a well-lit clinical room. Assessment was terminated when two mistakes were made within a triplet, with logCS calculated on a letter-by-letter basis. 

To assess reading function, participants were provided with refractive correction equivalent to their habitual distance vision correction with an appropriate reading addition (default +2.50 diopters [D] but increased if necessary for shorter working distances). Correction was provided either by the participant's own spectacles or in a trial frame as necessary. 

MNREAD charts of standard polarity (black text on white) were used throughout to assess reading parameters from the card and app versions and for judgments of comfortable print size. Different MNREAD charts and sentences (English chart variants 1–5) were used in each case. Chart use was randomized among participants, but the same order of assessment was followed for each participant (first comfortable print judgment, card assessment, app assessment, second comfortable print judgment). 

For a subset of participants where an additional researcher was available, the time taken to complete each reading assessment was recorded. Duration was recorded as the clinical time between starting to explain the procedure and completion of the test. Note that, for the card assessment, the time taken does not include later offline analysis to plot reading times for each sentence and derive CPS and MRS values. 

To assess comfortable print size (logMAR), participants were presented with an uncovered MNREAD card on a reading easel at a working distance of 40 cm and asked, “What is the smallest print size that you would find comfortable using?” No specific instructions were provided on how to make this choice. This procedure was followed twice to assess within-subject repeatability: once at the beginning of the reading assessment and once at the end. 

Participants were then presented with a different MNREAD card on an easel at 40 cm. If fewer than three sentences could be read, working distance was reduced to 20 cm and reading addition increased to +5.00 D. Participants were audio recorded reading the chart. In offline analysis, the time taken to read each sentence was determined from the audio recordings using a stopwatch, and the times were plotted for each print size on standard MNREAD recording sheets. RA (logMAR) was determined on a word-by-word basis.¹² CPS (logMAR) was determined by identifying the plotted point falling furthest from a reference line with a slope of 1.0 on log–log coordinates.²⁶ MRS (words per minute) was derived as the average reading speed for all print sizes at and above the CPS. 

The MNREAD app¹³ (version 1.8) was used on a sixth-generation iPad (Apple, Cupertino, CA). An appropriate working distance of 20, 40, or 60 cm was selected, with reading addition adjusted as appropriate. The parameters of RA, CPS, MRS, and reading accessibility index (ACC) were recorded within the app. Parameters derived by the app were reviewed, in keeping with manufacturer’s instructions²⁷ and amended if required. 

Statistical Analysis

Data were analyzed using SPSS Statistics 28.0 (IBM, Chicago, IL). Bland–Altman analysis was used to assess reliability by comparing the difference between two measures with their mean. The statistical significance of the mean difference, or bias, was assessed with repeated-measures t-tests. Cohen's d effect sizes are provided for comparisons of significant difference, with values greater than 0.20, 0.50, or 0.80 indicating small, medium, or large effect sizes, respectively. Limits of agreement around the mean difference were calculated as 1.96 times the standard deviation (SD) of the mean difference. Outer and inner 95% confidence intervals (CIs) to the limits of agreement (LoA) are also given.²⁸ To assess the variation of bias across the functional range, linear regressions were fitted to the Bland–Altman analyses with the significance of the slope value provided. One-way, repeated-measures ANOVA was used to compare the durations of assessment for CfPS and the MNREAD card and app. Where Mauchly's test indicated sphericity was violated, Greenhouse–Geisser corrections to degrees of freedom were applied.²⁹ Simple contrasts referenced to CfPS were used to compare individual methods. 

Results

Demographics

Thirty-four people took part in the study, of whom 15 were male and 19 female, with mean age ± SD of 73 ± 16 years (range, 18–91). Thirty-one participants were white, one was Indian, and two were Arab. The mean duration of vision loss was 14 ± 20 years (range, 0.75–84). Twenty-two people were not registered as vision impaired, seven were sight impaired, and five were severely sight impaired. Thirty participants had retinal causes of loss, including 20 with age-related macular degeneration (wet, dry, or unspecified), one with diabetic maculopathy, two with hereditary retinal dystrophies (retinitis pigmentosa cone dystrophy), three with retinal vascular occlusions, and three with other retinal causes (epiretinal membrane, toxoplasmosis, unknown retinal degeneration). Two people had degenerative myopia, one had glaucoma, and one had cerebrovascular disease. 

Visual Function

Descriptive statistics of the visual function parameters assessed are shown in the Table. With the MNREAD app, CPS values were amended in four cases (11%). In all instances, this was due to a participant slowing down on one sentence (e.g., stumbling on a word) and then speeding up again. The app placed the CPS at the size larger than the slowed sentence, whereas the smallest print size that supported the MRS was smaller. The proportion of results amended is similar to previous work¹³ where 9% of results from low-vision participants were amended following inspection. 

Table.

 

Descriptive Statistics of the Visual Function Parameters Assessed

 

Assessment Duration

The duration that each method of assessment took was recorded for a subset of 20 participants. Duration differed significantly between methods (F_{1.3, 24.5} = 8.94, P < 0.01), with contrasts indicating that CfPS (144 ± 77 seconds; range, 30–334) was quicker to assess than either MNREAD card assessment (231 ± 176 seconds; range, 120–942; F₁ = 4.9, P < 0.05) or app assessment (285 ± 43 seconds; range, 224–376; F₁ = 68, P < 0.01). 

Within-Session Repeatability of CfPS

The first judgment of CfPS obtained prior to full MNREAD assessment (0.71 ± 0.30 logMAR) did not differ significantly (mean difference, −0.02 ± 0.06; t = −1.64, 33 degrees of freedom [df]; P > 0.05) from the repeated judgment of CfPS (0.73 ± 0.30 logMAR) obtained after varied measures of MNREAD had been assessed. Figure 1 shows the associated Bland–Altman analysis, with limits of agreement on the repeated CfPS measure of ±0.09 logMAR. A linear regression fitted to the data, where difference = (0.01 × mean) − 0.03, had a slope that was not significantly different from zero (t = 0.32, P > 0.05), indicating no variation in bias across the functional range. 

Figure 1.

 

Bland–Altman analysis of within-session repeated measures of CfPS. The mean difference (bias) was −0.02 ± 0.06 logMAR. The LoA were −0.14 to +0.11 (±0.09) logMAR, with outer 95% CIs to the LoA of 0.15 to −0.18 logMAR, and inner 95% CIs of 0.08 to −0.12 logMAR.

There was no difference between the two CfPS judgments for 23 participants (68% of the sample). The difference between judgments was a maximum of 0.1 logMAR (one line) for 33 participants (97% of sample), and the maximum difference between estimates was two lines (n = 1). There was a slight preference for the second CfPS judgment to be smaller than the first, with eight people selecting a smaller print size on the second judgment as compared to three selecting a bigger print on the second judgment. 

Reliability

The first judgment of CfPS was compared to CPS as assessed with the MNREAD card or app to assess how closely subjective CfPS judgments compared to the more objectively assessed CPS parameter. 

CfPS Compared to Card CPS

CfPS was statistically significantly larger (mean difference, 0.10 ± 0.22 logMAR; t = 2.7, 33 df; P < 0.05) than CPS derived from MNREAD card assessment (0.61 ± 0.32 logMAR), as shown in Figure 2. The effect size of the difference is small (Cohen's d = 0.32), and the mean difference of 0.1 logMAR represents one unit of difference, as CfPS and CPS can only be scored to the nearest line. A linear regression fitted to the data, where difference = (−0.08 × mean) + 0.15, had a slope that was not significantly different from zero (t = −0.60; P > 0.05), indicating no variation in bias across the functional range. 

Figure 2.

 

Bland–Altman analysis of CfPS and CPS assessed with the MNREAD card chart. The mean difference (bias) was 0.10 ± 0.22 logMAR. The LoA were −0.33 to +0.53 (±0.43) logMAR, with outer 95% CIs to the LoA of 0.68 to −0.48 logMAR, and inner 95% CIs of 0.45 to −0.25 logMAR.

CfPS Compared to App CPS

CfPS was not significantly different (mean difference, 0.02 ± 0.23 logMAR; t = 0.53, 33 df; P > 0.05) from CPS derived from the MNREAD app assessment (0.69 ± 0.29 logMAR), as shown in Figure 3. A linear regression fitted to the data, where difference = (0.06 × mean) − 0.02, had a slope that was not significantly different from zero (t = 0.40; P > 0.05), indicating no variation in bias across the functional range. 

Figure 3.

 

Bland–Altman analysis of CfPS and CPS assessed with the MNREAD app. The mean difference (bias) was 0.02 ± 0.23 logMAR. The LoA were −0.43 to +0.47 (±0.45) logMAR, with outer 95% CIs to the LoA of 0.62 to −0.58 logMAR, and inner 95% CIs of 0.39 to −0.35 logMAR.

Acuity Reserve

CfPS acuity reserve was assessed by considering the difference between CfPS and the reading acuity assessed with the MNREAD card chart. Because both parameters are assessed with the card chart, this reflects the most likely scenario in clinical practice. The mean difference between CfPS and reading acuity was 0.28 ± 0.21 logMAR, which represents a mean acuity reserve of 1.9:1. The distribution of differences for individuals (range, −0.21 to 0.70 logMAR; maximum acuity reserve, 5.0:1) is shown in Figure 4. For comparison, the average difference between card CPS and RA was 0.21 ± 0.16 logMAR, or a mean acuity reserve of 1.62:1. The difference between the app CPS and RA was 0.18 ± 0.13 logMAR, or a mean acuity reserve of 1.51:1. 

Figure 4.

 

Frequency distribution of the difference between CfPS and RA (as assessed with the MNREAD card chart), in logMAR. A difference of 0.30 logMAR represents a 2:1 acuity reserve, and 0.60 logMAR is a 4:1 acuity reserve.

Discussion

Rather than using examiner interpretation of what constitutes the CPS, CfPS uses patient interpretation of what print size a person would consider suitable for sustained reading.²¹ CfPS is quick to assess as a single parameter: On average, it took 2 minutes, 24 seconds (SD, 1 minute, 7 seconds) to explain, assess, and record the value. This compares favorably with the card chart, which took on average 3 minutes, 51 seconds (SD, 2 minutes, 56 seconds) to assess plus further time to plot and interpret the data, or the app, which took on average 4 minutes, 45 seconds (SD, 43 seconds). Solely assessing CfPS will provide a size to aim for with magnification for sustained reading. However, this alone will not provide further parameters that may be relevant to sustained reading, such as reading speed, fluency, or feasible reading duration. 

One benefit of utilizing patient judgment seems to be improved consistency of values, with CfPS demonstrating excellent within-session repeatability of ±0.09 logMAR. This coefficient of repeatability is smaller than that seen for any study of CPS with vision-impaired observers, where values between ±0.20 and ±0.67 have been obtained using MNREAD charts,^15,17,18,30 and between ±0.16 and ±0.33 logMAR using Radner charts.^31–33 

A limitation of the present study is that repeatability has been derived within a single session rather than between sessions separated by several weeks. Further reading tests were undertaken between the two CfPS measures, so by the second measure the participant may have had increased familiarity with the test involved but also greater fatigue. Note that different versions of charts were used for all assessments, so participants were not selecting sentences based on recall of previous choices, although it is feasible that the position of the text on the chart that had previously been selected could be recalled. 

In previous work,²¹ we assessed CfPS after conducting full MNREAD assessment. It was not clear whether the lack of bias between CfPS and CPS was contributed to by participants having had experience with CPS assessments. In the present study, we show that CfPS judgments pre- and post-MNREAD assessment are not significantly different (mean difference, –0.02 ± 0.06 logMAR). CfPS appears to be invariant of whether it is assessed before or after other reading assessments. 

Although CfPS appears to be a consistent measure, it is not useful unless it is measuring what it is intended to measure. Examining differences between CfPS and CPS measures is therefore important. It has previously been shown²¹ that CfPS and CPS derived from the MNREAD card were no different on average. That study also used a mixed cohort of people with vision loss but with slightly poorer function (average distance VA of 0.71 logMAR compared to 0.40 logMAR in the present study). In the present study, CfPS was no different from CPS derived from the MNREAD app (P > 0.05) and was slightly larger on average than CPS derived from the MNREAD card (0.10 logMAR, P < 0.05, small effect size). Although this difference of one line of logMAR was statistically significant it is a difference of one unit and would make a difference of 1.26× to the predicted magnification, making it of marginal clinical significance. No indication of variation in bias across the functional range has been observed in either study. Both studies therefore indicate little difference between CfPS and CPS, and that the quick clinical assessment of CfPS does provide a valid alternative measurement that reflects CPS. 

In addition to mean difference, LoA are important to indicate the range of differences between CfPS and CPS measures. The LoA for the comparison between CfPS and CPS derived by the card or app are relatively wide (±0.43 to 0.45 logMAR), as compared to ±0.35 logMAR previously.²¹ However, the LoA are similarly wide when comparing different methods for deriving CPS: ±0.37 logMAR when comparing two methods of deriving CPS from MNREAD charts,¹⁴ and ±0.24 to ±0.42 logMAR when comparing CPS derived in the same way from three different charts.³¹ 

CfPS provides an individualized measure of how large print must be for sustained reading. An alternative approach is to provide a standard increase in print size beyond reading acuity for sustained tasks, typically an acuity reserve value of 2:1.^4–7 The mean difference between CfPS and reading acuity with the MNREAD card in this study was 0.28 ± 0.21 logMAR, indicating an average acuity reserve of 1.9:1. This figure is a little higher than the 1.7:1 reserve (mean difference, 0.24 logMAR) previously seen with CfPS²¹ and approaches that of the suggested standard of 2:1. 

However, a mean acuity reserve figure masks the variation seen between individuals.^7,8 Individual differences between CfPS and RA values varied up to 0.70 logMAR (Fig. 4) or acuity reserves of up to 5:1. A standard acuity reserve of 2:1 (0.3 logMAR) would only be accurate to within 0.1 logMAR (i.e., 0.2–0.4 logMAR) for 50% of this sample. Spending additional time to identify an individual print size goal for sustained reading is therefore a potentially valuable approach compared to doubling the RA, especially when CfPS is a more repeatable measure than CPS and quick to obtain. 

One limitation of a self-reported value is highlighted by the negative acuity reserve value seen for one participant (Fig. 4). This individual selected an initial CfPS value 0.2 logMAR smaller than their subsequently measured reading acuity, which might suggest that they had not understood what was being asked of them in identifying a comfortable print size. This participant's second judgment of CfPS was larger than their reading acuity, suggesting that they then understood the task. It was also noted that this person's vision had markedly reduced in the previous year, and they were having difficulty adapting to their change in vision, which may have influenced their first choice of print size. Beyond this individual, the relatively short time it took to explain and assess CfPS, as well as its repeatability, suggests that most people found it straightforward to provide a CfPS judgment. 

In this study we deliberately chose to include people with a range of visual function who considered that they had visual impairment that impacts on their daily life, and not all were visually impaired to the point of sight loss registration. However, visual difficulty begins at much milder levels of visual loss than would be required for registration, and we wished to evaluate CfPS with a range of people consistent with those who would be met in a low-vision clinical setting. Although some participants had relatively normal distance visual acuity or contrast sensitivity (Table), all had reduced access to print, as indicated by the mean ACC value of 0.38 ± 0.19 and maximum ACC value of 0.76. In comparison, the ACC value expected of younger people with normal vision is 1.0, and in an older group has been shown to be 0.76 (±0.19).³⁴ 

Assessments in the present study were made considering angular print sizes (logMAR) with both print size and working distance specified, making an implicit assumption that CfPS will be consistent in angular terms. In other words, it is expected that, if a shorter working distance was used (perhaps with provision of a higher reading addition), then an appropriately smaller physical print size (M or N point) would be selected as the CfPS. It will be of value to investigate this assumption and determine whether CfPS is more consistent in terms of angular or physical print sizes if assessed at different working distances. Further, CfPS has only been assessed using MNREAD charts, and its invariance in other formats has not yet been assessed. In particular, given that these judgments regarding comfort of reading are of greatest relevance to sustained reading, they should also be evaluated with longer continuous text presentations. 

In conclusion, the quicker, individualized, and more consistent patient-centered measure of CfPS may be considered preferable to CPS for clinical use in low-vision assessments. 

Acknowledgments

Supported by an ARU Rachel Cook Summer Scholarship (to ES). 

Disclosure: K. Latham, None; H. Subhi, None; E. Shaw, None