1040-5488/15/9205-0559/0 VOL. 92, NO. 5, PP. 559Y565 OPTOMETRY AND VISION SCIENCE Copyright * 2015 American Academy of Optometry
ORIGINAL ARTICLE
Determinants and Standardization of Mesopic Visual Acuity Rachelle J. Lin*, Jason S. Ng†, and Andrew L. Nguyen‡
ABSTRACT Purpose. It is well established that visual acuity (VA) decreases with luminance but the specific factors that are responsible remain unclear. The purpose of this study was to quantify the contributions of accommodative error, pupil size, and higherorder aberrations to the decrease in VA when transitioning from photopic to mesopic light levels. Additionally, repeatability of VA at photopic and mesopic levels was measured to derive a luminance recommendation for mesopic VA testing, which can provide the standardization needed for future translational clinical studies and the widespread adoption of mesopic VA testing. Methods. Monocular VAs were assessed at one photopic and three mesopic light levels: 94, 3, 0.75, and 0.38 cd/m2, with an E-ETDRS testing system in 43 normal subjects. Accommodative error, pupil size, and higher-order aberrations were obtained. Twenty subjects were retested at another visit to assess VA repeatability. Results. The mean (TSD) logMAR (logarithm of the minimum angle of resolution) VA was j0.08 (T0.06) at 94 cd/m2, 0.05 (T0.07) at 3 cd/m2, 0.16 (T0.06) at 0.75 cd/m2, and 0.27 (T0.09) at 0.38 cd/m2. Light level and accommodative error were significantly associated with VA, and light level explained 75% of the variance. The mean differences in VAs between two visits were not significantly different from zero (p 9 0.05). The coefficients of repeatability for 94, 3, 0.75, and 0.38 cd/m2 were 0.08, 0.11, 0.14, and 0.14 logMAR, respectively. Conclusions. Light level, among all other factors studied, contributes the most to the reduction in VA tested under mesopic conditions. Testing mesopic VA at 0.75 cd/m2, or about 2.0 log units less than photopic testing, provides a significant and repeatable decrease in VA similar to standardized low-contrast VA testing, and therefore this level is recommended. (Optom Vis Sci 2015;92:559Y565) Key Words: visual acuity, visual acuity testing, mesopic, repeatability
A
lthough photopic visual acuity (VA) is the most commonly used determinant of visual function, studies have shown that measures of mesopic VA can provide additional insight into functional vision loss.1Y3 The factors that may contribute to decreasing visual function at mesopic light levels, which has been associated with ‘‘night myopia,’’ include spherical aberration, accommodative error, pupil size, photon noise, and neural effects.4Y6 Studies have shown an increase in accommodative error with decreasing light levels. As retinal illuminance decreases, the accommodative posture departs from the positions indicated by the standard photopic accommodative stimulus-response curve and moves in the direction of the subject’s dark focus position.7Y9
*OD, MS, FAAO † OD, PhD, FAAO ‡ PhD Southern California College of Optometry, Marshall B. Ketchum University, Fullerton, California (all authors); and Department of Mathematics, California State University, Fullerton, California (ALN).
Pupil size also changes with light levels. Smaller pupils decrease retinal illumination and increase depth of focus. Larger pupils, resulting from lower light levels, change the aberrations that are introduced into the optical system of the eye.4 The wavefront error varies with pupil diameter.10 Spherical aberration in particular has been speculated to be a cause of ‘‘night myopia,’’ as it increases as pupil size increases.4 Other higher-order aberrations would also induce optical blur. The primary focus of this study was to assess the relative contributions of light level, accommodative error, higher-order aberrations, and pupil size to the decrease in VA at mesopic light levels. It is well accepted that decreased retinal illumination causes decreased spatial resolution.11 However, the quantification of the relative contributions of the multiple factors has not been well studied. As there is no current standardized procedure for assessing mesopic VA,12 the secondary goal of this study was to collect normative VA values at three mesopic light levels in an attempt to identify a particular mesopic light level upon which to build future studies (i.e., proposing to ‘‘standardize’’ a mesopic light level so
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
560 Mesopic Visual AcuityVLin et al.
that future clinical research can be more cohesive). Although various studies involving mesopic VA exist, there is no standard luminance level across these studies, making it difficult to translate the various studies into clinical practice and make progress in systematically understanding how mesopic vision function may be affected in the case of ocular pathology. It should be noted that a level of 3 cd/m2 has been used as a standard for mesopic VA and contrast sensitivity testing for Food and Drug Administration trials evaluating refractive and intraocular lens surgery outcomes,13,14 and for this reason, we not only evaluated this particular luminance level but also tested at two additional lower mesopic levels (0.75 and 0.38 cd/m2), which can be easily achieved with commercial filters. Additionally, we evaluated the repeatability of VA at these light levels to aid in determining a possible recommended clinical standard.
METHODS Subjects Forty-six subjects were recruited from the general campus community and the surrounding general population. A 0.15 logMAR (logarithm of the minimum angle of resolution) difference between photopic VA and mesopic VA at 3 cd/m2 was anticipated, with further decreases in VA at lower mesopic light levels.15 Based on an assumed difference of at least 0.15 logMAR between photopic and mesopic light levels and a conservative SD of 0.3, power analysis using a paired t test indicated that a sample size of at least 34 subjects would be needed to demonstrate a significant difference at the 5% level with a power of 80%. Additional subjects were recruited to account for subject loss or data collection insufficiency. Using intraclass correlation (ICC) power analysis, sample sizes of 20 (or 25) subjects were calculated to achieve 78% (or 86%) power to detect an ICC of 0.70 under the alternative hypothesis when the ICC under the null hypothesis is 0.30 using an F test with a significance level of 0.05. From these analyses, a total of 21 subjects were asked to return for repeat testing. Informed consent was obtained from all subjects. The exclusion criteria were subjects with ocular conditions or prior surgeries known to have a significant adverse effect on media clarity or who had best-corrected distance VA of 20/25 or worse. The procedures complied with the Declaration of Helsinki and were approved by the university institutional review board.
Procedures All subjects were tested monocularly and were fully corrected, wearing their habitual spectacle correction, or no correction if emmetropic, during the tests. Subjects had a mean (TSD) calculated equivalent spherical refraction of j3.71 (T2.63) diopters (D) and a range of 0.00 to j10.12 D. Subjects viewed the VA tests while seated at the chin and headrest of an open-field autorefractor. In this way, it was possible to measure VA and accommodative response under the same viewing conditions. Visual acuity was assessed with a validated computerized VA test (M&S Study, M&S Technologies, Niles, IL), which uses the
E-ETDRS algorithm16 at a test distance of 3 m. The stimulus display (Dell E198FP, 19µ, 1280 1024) subtended 7.2 degrees horizontally and 5.8 degrees vertically. The E-ETDRS protocol uses a single-letter presentation with crowding bars. Letters are presented at random from a limited letter set. Subsequent stimuli were not presented until the subject gave a response. Threshold was determined using a modified staircase technique. Visual acuity was recorded as the output ‘‘letter score’’ by the software, which was converted to the equivalent logMAR value for data analysis.16 Photopic VA was assessed at 94 cd/m2.13 Mesopic VA was assessed at three light levels: 3, 0.75, and 0.38 cd/m2. The 3 cd/m2 light level was achieved by reducing the monitor luminance. The 0.75 cd/m2 light level was achieved by the addition of a 0.6 neutral density filter (#210; Rosco Laboratories Inc, Stamford, CT) over the entire monitor screen when at a luminance of 3 cd/m2. An additional 0.3 neutral density filter (#209) lowered the total luminance to 0.38 cd/m2. The letter contrast (Weber fraction) was always greater than 90% at all light levels. Luminance levels were measured and calibrated with a Minolta LS-110 spot photometer (Konica Minolta, Ramsey, NJ). The open-field autorefractor uses a large beam-splitter, which caused a 13% decrease in light transmission. Therefore, all letter luminances and light levels were measured through the beam-splitter, in the same manner the subjects viewed the stimulus. The light level at which each subject began VA testing was randomized to avoid the potential confounding effect of letter set memorization.17 The subject was allowed 5 minutes of adaptation time before the first VA test. In between subsequent trials, the subject was given adaptation time while the acuity program was reset. Similar adaptation times were given, regardless of light level. Accommodative posture and the corresponding pupil size were measured with an open-field autorefractor (Grand Seiko WAM5500, Shigiya Machinery Works Ltd, Japan). The autorefractor was positioned directly in line with the VA testing monitor. Letters that were generally twice the size of the threshold VA at each light level were used to measure accommodation. Three measurements of accommodative posture were taken at each light level and averaged for data analysis. The difference between the nominal accommodative stimulus and accommodative response was calculated as the accommodative error. The absolute value of the accommodative error, thereby eliminating the numerical difference between equal amounts of accommodative lead and accommodative lag, was used as a variable in the regression model. Aberrometry measurements were collected with a Marco OPDScan III refractive power/corneal analyzer (Marco, Jacksonville, FL), operating in the ‘‘OPD/CT Measurement Mode.’’ The Marco OPD-Scan III provides pupil sizes and aberrometry readings for two preset light levels: ‘‘photopic’’ and ‘‘mesopic.’’ The luminances used for these tests are not specified by the manufacturer. Therefore, the mesopic testing procedure was not varied between subjects. The total higher-order (third through sixth order) root-mean-square (RMS) values on uncorrected subjects as determined at the instrument’s mesopic pupil diameter size as well as for constant wavefront diameters of 4 and 6 mm were collected. To assess repeatability of VA, all subjects were informed and given the option to return for a second visit until the first 21 subjects who volunteered were tested. One subject was not included in the final analysis because the subject’s initial visit had
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al.
561
0.24 logMAR at 0.75 cd/m2, and 0.35 logMAR at 0.38 cd/m2 (Table 1). The mean VA at each light level differed significantly (t Q 11.92, p G 0.001). The measured decrease from photopic VA to mesopic VA was also statistically significant for all three mesopic light levels (t Q 14.20, p G 0.001).
been excluded because of insufficient RMS data. Therefore, repeatability data were collected from a total of 20 subjects.
Data Analysis Anonymized data from the subjects were compiled in Excel and statistical analyses were performed in R (R Foundation, Vienna, Austria) and Stata (StataCorp, College Station, TX). The primary outcome measure, VA (logMAR), was tested for normality using the Kolmogorov-Smirnov test to assess the validity of analysis using parametric statistics. Mean VAs, pupil areas obtained from the autorefractor, and accommodative errors were examined using analysis of variance tests with post hoc paired t tests or Tukey 95% simultaneous confidence intervals when appropriate.18 A linear mixed-effects multiple regression model, adjusted for age, using a stepwise backward method was used to examine the main effects of light level, accommodative errors, and pupil area on VA.18 The repeatability data were analyzed by deriving ICC coefficients and using Bland-Altman analysis. Multiple measures of repeatability were derived to allow others to more easily compare repeatability data from other studies. The ICC statistic was selected to assess the repeatability of the measurements taken from the first visit and the second visit. Bland-Altman plots were generated, and the 95% limits of agreement (LOAs) were derived.19 Coefficients of repeatability (CORs), a measure of the LOAs, were calculated by multiplying the SD of the within-subject differences between the two visits by 1.96.
Accommodation On average, an accommodative lag was found at all four light levels (Table 1). Only 14 subjects had accommodative lead at any of the four tested luminance levels. Accommodative error was significantly different across light levels (p = 0.001) and was significantly different between all the mesopic light levels and the photopic light level (t e j3.16, p e 0.01), but no significant differences were found between the mesopic light levels (p Q 0.89). Absolute accommodative error was also significantly different across light levels (p = 0.04) and was found to be significantly different between the photopic level and 0.75 cd/m2 (p = 0.02), but no significant differences were found between other pairs of light levels (p Q 0.41).
Pupil Size The mean (TSD) pupil areas as calculated from the measured pupil diameters are shown in Table 1. Pupil size was significantly different across light levels (p G 0.001). The increase in pupil size at lower light levels was expected. Pupil area was found to be different for each light level compared with one another (t Q 3.78, p e 0.001), except for the pupil areas at the light levels of 3 and 0.75 cd/m2 (p = 0.07).
RESULTS Of the 46 subjects enrolled in the study, data from three of the subjects were not included in the final analysis because of insufficient data. The aberrometer did not record the higher-order RMS measurements from two subjects, and accommodation could not be consistently measured through the third subject’s high myopic spectacle correction. The resulting 43 subjects (14 men and 29 women) had a mean (TSD) age of 24.7 (T3.0) years with a range of 14.5 to 31.8 years.
Aberrations The mean (TSD) total higher-order RMS for the analyzed mesopic zone pupil diameter is shown in Table 1. The data (not shown) were also tabulated using constant wavefront diameters of 4 mm (n = 41) and 6 mm (n =33).
Visual Acuity
Linear Regression Model
No VA data were found to differ significantly from the normal distribution (p 9 0.05). The mean decrease in VA, as compared with the photopic condition, was 0.13 logMAR at 3 cd/m2,
The initial model examined age, higher-order aberrations, absolute accommodative error, pupil area, and light level. The higher-order aberration term was analyzed for the constant
TABLE 1.
Mean values for the 43 subjects Photopic 2
VA, mean T SD, logMAR Accommodative error, mean T SD, D Absolute accommodative error, mean T SD, D Pupil diameter, mean T SD, mm Pupil area, mean T SD, mm2 Mesopic zone diameter, mean T SD, mm Higher-order RMS T SD, Km
Mesopic 2
94 cd/m
3 cd/m
0.75 cd/m2
0.38 cd/m2
j0.08 T 0.06 +0.34 T 0.27 0.36 T 0.25 4.6 T 0.8 17.5 T 6.2
0.05 T 0.07 +0.25 T 0.31 0.32 T 0.23 6.3 T 0.9 31.4 T 8.3
0.16 T 0.06 +0.23 T 0.26 0.28 T 0.20 6.5 T 0.9 33.9 T 8.6 6.54 T 0.85 0.464 T 0.248
0.27 T 0.09 +0.25 T 0.32 0.32 T 0.25 6.9 T 0.6 37.5 T 7.1
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
562 Mesopic Visual AcuityVLin et al.
wavefront diameters and the aforementioned mesopic zone pupil diameter. Light level was coded within the model using three indicators, with the photopic light level being the reference. The model showed that age, pupil area, and aberrations were not significant variables. Thus, backward stepwise regression resulted in the most parsimonious model consisting of only two variables: light level (p G 0.001) and accommodative error (p = 0.003). A plot of VA as a function of absolute accommodative error is shown in Fig. 1. Bestfit lines for all light levels were determined, and it was found that all of the slopes did not significantly differ from each other (t e 2.65, p Q 0.01, Bonferroni-adjusted > level of 0.008 for each test). All slopes for the mesopic light levels significantly differed from zero (R2 Q 0.17, t Q 2.90, p e 0.006). The photopic line’s slope did not differ significantly from zero (p = 0.18). Subjects with larger accommodative errors had additional losses of about 0.1 to 0.2 logMAR under the mesopic viewing conditions. However, the expected influence of accommodative error on VA varies with pupil size.20,21 Other than light level, only accommodative error was significantly (p = 0.003) associated with logMAR VA. Holding light level constant (at one of the tested lighting levels) resulted in an average logMAR increase (worsening) of 0.076 for every 1 D of absolute accommodative error. Additionally, light level and absolute accommodative error were examined in a mixed-effects analysis of variance model that used a single variable for all light levels to determine which of the two factors had a bigger contribution to the observed changes in VA. Light level (F3,125 = 523.24, p G 0.0001) was significantly associated with VA. Examination of the sum of squares term showed that light level explained 75% of the variance in VA. It was also determined that if retinal illuminance (luminance pupil area) was used in the model, instead of luminance, much less variance (43%) could be explained.
Repeatability Although 21 subjects returned for the second visit (~3 weeks later), one subject was not included in the final analysis because of insufficient RMS data. The resulting 20 subjects (8 men and 12 women) had a mean (TSD) age of 25.1 (T3.6) years with a range of 14.5 to 31.8 years. The mean ICCs (95% confidence interval) for 94, 3, 0.75, and 0.38 cd/m2 were 0.83 (0.63 to 0.93), 0.75 (0.48 to 0.89), 0.55 (0.16 to 0.79), and 0.67 (0.34 to 0.86), respectively. The mean differences in VAs between visit 1 and visit 2 for subjects were not significantly different from zero (p 9 0.17), and thus calculations of the CORs were valid using the formula described by Bland and Altman.19 The CORs for 94, 3, 0.75, and 0.38 cd/m2 were 0.08, 0.11, 0.14, and 0.14, respectively. Repeatability improved with higher light levels, but the difference between mesopic light levels by the COR measure was only 0.03 logMAR. The mean logMAR difference and T95% LOA (1.96 SD) was j0.004 T 0.076 at 94 cd/m2, j0.017 T 0.106 at 3 cd/m2, j0.002 T 0.139 at 0.75 cd/m2, and j0.021 T 0.139 at 0.38 cd/m2. BlandAltman plots were generated (Fig. 2) showing the repeatability results. The 95% LOAs show that VA under the various mesopic light levels used in this study had similar repeatability.
DISCUSSION In this study, a statistically significant decrease was found between the mean VA at the photopic light level and each mesopic light level. Several factors were assessed as possible contributing factors to the decrease in VA at mesopic light levels, including accommodative error, pupil size, and higher-order aberrations.
FIGURE 1. Visual acuity as a function of absolute accommodative error by light level. The slopes of the best-fit lines were not significantly different from each other. Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al.
563
FIGURE 2. Bland-Altman plots of logMAR VA repeatability at the four luminance levels used in this study. Bold lines represent the mean difference and the upper and lower 95% LOAs.
Analyses using a linear mixed-effects multiple linear regression model, which was adjusted for age, demonstrated that only light level and accommodative error were significant variables in the determination of VA, whereas pupil area and higher-order aberrations were not. The light level itself was found to explain the vast majority of the variance. The 0.13 logMAR decrease that occurred between 94 and 3 cd/m2 is similar to the 0.15 logMAR difference found in a previous study that used a filter over a standard ETDRS retro-illuminated cabinet.22 In the classic study by Shlaer,11 the average VAs of two subjects using Landolt Cs were about j0.05 logMAR at 4 cd/m2 and 0.33 logMAR at 0.43 cd/m2, which were somewhat better than those found in this study likely because of the more pure resolution task. In general, comparisons of acuity findings across the studies are difficult given differences in exact light level or stimuli. The 0.21 log MAR difference between high- and low-contrast VA for normal subjects as reported in earlier studies23 is similar to the 0.24 logMAR change between the photopic and 0.75 cd/m2 light level found in this study. The 95% LOAs are also similar, as
compared with published low-contrast VA LOAs of T0.129.23 Therefore, we recommend a luminance of 0.75 cd/m2, or about 2.0 log units less luminance than photopic testing levels, for mesopic VA testing. The well-established clinical norms for differences and changes in VA would be very similar between mesopic VA testing at 0.75 cd/m2 and photopic low-contrast VA testing that clinicians are already using. Finally, although we did not enroll subjects with very small pupils or dense media opacities, it should be noted that, for such subjects, a very low mesopic light level might result in a near scotopic level of retinal illumination. In this study, calculations of the mean accommodative response to the 3-m accommodative stimulus distance showed a consistent lag of accommodation at all four light levels. As expected from previous studies on the changes in average accommodative posture in the dark, the subjects in this study showed a small, but statistically significant, increase in accommodation as the light level decreased from photopic to mesopic. Because the subjects on average had a lag of accommodation, the overall accommodative error actually decreased as the light level decreased. The slightly
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
564 Mesopic Visual AcuityVLin et al.
higher accommodative errors at the photopic light level may have also been attributed to a smaller pupil size, allowing larger depth of focus. Our study did not demonstrate statistically significant differences in accommodative error between the three mesopic light levels. Nevertheless, based on Fig. 1, accommodative error may contribute more to VA decrease at lower light levels and may play a role in night myopia. It is possible that the current study did not have enough power to detect differences in accommodative error between the mesopic light levels though. A larger range of mesopic and scotopic light levels, or a larger sample size, may have been able to demonstrate a change in accommodative posture with decreasing luminance. Our study found an average logMAR increase (worsening) of 0.076 for every 1 D of absolute accommodative error. This change in logMAR VA is less than that predicted from previous studies that approximate a one-line decrease in VA for each 0.25 D of defocus.24 In this study, only the higher-order aberrations were considered because the lower-order aberrations should largely be corrected by each subject’s best correction. Ideally, a higher-order RMS value would have been obtained for each pupil size as determined during the VA and accommodation testing procedures. Then, with a separate higher-order RMS value corresponding to the pupil size at each light level, individual changes in RMS with pupil size and any correlation with mesopic VA could have been assessed more specifically. However, the maximum pupil diameter zone on the OPD-Scan III, which was obtained during the mesopic measurements, was often smaller than those obtained during mesopic VA and accommodation testing. Because RMS values for larger pupil sizes were limited by the OPD-Scan III pupil sizes (and the measurements that did have adequately sized pupil diameters would be additionally limited to 1-mm increments in the RMS output), the single mesopic higher-order RMS value was used in the statistical analysis in this study as it provided aberration assessment for the largest pupil area. Because the RMS value as determined in this study was not found to be a significant variable for VA, it is questionable whether aberrations play a significant role in ‘‘night myopia,’’ as has been previously proposed.4 It is also unknown if the results of the aberrations portion of this study would be significantly different if larger pupil zones could have been obtained with the OPD-Scan III. Pupil area was also not found to be correlated with VA in this study. The sample size calculation performed before data collection did not explicitly aim for detecting pupil area differences, and thus, it is possible that the study was underpowered to detect such a difference. Differences in pupil area between subjects at any fixed light level are directly related to individual differences in retinal illuminance under those conditions. Indirectly reexamining pupil area by way of analyzing retinal illuminance instead of luminance did not aid in explaining any more of the variance. A common clinical criterion for significant VA change is oneline difference. The 0.13 logMAR decrease in VA from 94 to 3 cd/m2, thus, may be too small to be used clinically.23,25 However, these are group data and individual differences are likely much less, which, once established for a particular patient, like Brown and Lovie-Kitchin26 have done for photopic VA, could be useful when monitoring patients. Among the mesopic light
levels, the CORs were not substantially different and thus repeatability alone is an insufficient guide to recommending a particular light level for mesopic VA testing. Because the subjects in this study were within a younger age range, further studies to establish normative mesopic data in a group of older subjects could be important as means to screen or monitor various ocular conditions. This will be particularly applicable for patients with pathology that affects mesopic vision, such as rod-cone dystrophy and retinitis pigmentosa.1 Others have suggested that mesopic testing may also be useful for assessing patients at risk for age-related macular degeneration.27 The quantitative measurements may be especially useful for eye conditions that do not exhibit large decreases in high-contrast VA in early stages. Testing mesopic VA at 0.75 or 0.38 cd/m2 provides significant and repeatable decreases in VA, and thus are better alternatives to testing at the level of 3 cd/m2. Based on the similarities between the mean difference and repeatability of mesopic VA testing at 0.75 cd/m2 and photopic low-contrast VA testing, we propose that mesopic VA testing be standardized using a light level of 0.75 cd/m2. Using a neutral density filter to reduce photopic high-contrast electronic VA testing to a luminance of 0.75 cd/m2 may be an effective and repeatable method to clinically assess mesopic VA. Nevertheless, the standard recommended from the findings of this study will allow for more focused research on applying mesopic vision function findings to patient care.
ACKNOWLEDGMENTS The work in this article represents the majority of the first author’s efforts to fulfill the requirements of her MS in vision science degree. We thank the MS thesis committee members Drs. Lawrence R. Stark and James E. Bailey for their contributions to this study. Received July 10, 2015; accepted February 7, 2015.
REFERENCES 1. Petzold A, Plant GT. Clinical disorders affecting mesopic vision. Ophthalmic Physiol Opt 2006;26:326Y41. 2. Puell MC, Barrio AR, Palomo-Alvarez C, Gomez-Sanz FJ, ClementCorral A, Perez-Carrasco MJ. Impaired mesopic visual acuity in eyes with early age-related macular degeneration. Invest Ophthalmol Vis Sci 2012;53:7310Y4. 3. Tang W, Heng WJ, Lee HM, Fam HB, Lai NS. Efficacy of measuring visual performance of LASIK patients under photopic and mesopic conditions. Ann Acad Med Singapore 2006;35:541Y6. 4. Koomen M, Scolnik R, Tousey R. A study of night myopia. J Opt Soc Am 1951;41:80Y90. 5. Charman N. Night myopia and the dark focus of the eye. Optician 1991;202:13Y8. 6. Banks MS, Geisler WS, Bennett PJ. The physical limits of grating visibility. Vision Res 1987;27:1915Y24. 7. Johnson CA. Effects of luminance and stimulus distance on accommodation and visual resolution. J Opt Soc Am 1976;66:138Y42. 8. Wald G, Griffin DR. The change in refractive power of the human eye in dim and bright light. J Opt Soc Am 1947;37:321Y36. 9. Leibowitz HW, Owens DA. New evidence for the intermediate position of relaxed accommodation. Doc Ophthalmol 1978;46:133Y47.
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al. 10. Ravikumar A, Sarver EJ, Applegate RA. Change in visual acuity is highly correlated with change in six image quality metrics independent of wavefront error and/or pupil diameter. J Vis 2012;12:11. 11. Shlaer S. The relation between visual acuity and illumination. J Gen Physiol 1937;21:165Y88. 12. Wood JM, Owens DA. Standard measures of visual acuity do not predict drivers’ recognition performance under day or night conditions. Optom Vis Sci 2005;82:698Y705. 13. American National Standards Institute. American National Standard for Phakic Intraocular Lenses. ANSI Z80.13-2007 (R2012). Arlington, VA: American National Standards Institute; 2007. 14. Evans DW. FDA update: contrast sensitivity testing standards. Ophthamol Manage February 2005. Available at: http://www. ophthalmologymanagement.com/articleviewer.aspx?articleid=86281. Accessed July 7, 2014. 15. Rabin J. Luminance effects on visual acuity and small letter contrast sensitivity. Optom Vis Sci 1994;71:685Y8. 16. Beck RW, Moke PS, Turpin AH, Ferris FL, SanGiovanni JP, Johnson CA, Birch EE, Chandler DL, Cox TA, Blair RC, Kraker RT. A computerized method of visual acuity testing: adaptation of the early treatment of diabetic retinopathy study testing protocol. Am J Ophthalmol 2003;135:194Y205. 17. McMonnies CW. Control of chart memory for retesting acuity. Clin Exp Optom 2001;84:78Y84. 18. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN 3-900051-07-0 2012. 19. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307Y10.
565
20. Smith G. Relation between spherical refractive error and visual acuity. Optom Vis Sci 1991;68:591Y8. 21. Smith G, Jacobs RJ, Chan CD. Effect of defocus on visual acuity as measured by source and observer methods. Optom Vis Sci 1989;66: 430Y5. 22. Lin R, Ng JS, Li J. Normative data on standardized mesopic visual acuity. Optom Vis Sci 2012;90:E-Abstract 125613. 23. Lovie-Kitchin JE, Brown B. Repeatability and intercorrelations of standard vision tests as a function of age. Optom Vis Sci 2000;77: 412Y20. 24. Rabin J. Optical defocus: differential effects on size and contrast letter recognition thresholds. Invest Ophthalmol Vis Sci 1994;35:646Y8. 25. Manny RE, Hussein M, Gwiazda J, Marsh-Tootle W, Group CS. Repeatability of ETDRS visual acuity in children. Invest Ophthalmol Vis Sci 2003;44:3294Y300. 26. Brown B, Lovie-Kitchin J. Repeated visual acuity measurement: establishing the patient’s own criterion for change. Optom Vis Sci 1993;70:45Y53. 27. Lovie-Kitchin J, Feigl B. Assessment of age-related maculopathy using subjective vision tests. Clin Exp Optom 2005;88:292Y303.
Jason S. Ng Southern California College of Optometry Marshall B. Ketchum University 2575 Yorba Linda Blvd Fullerton, CA 92831 e-mail: [email protected]
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
ORIGINAL ARTICLE
Determinants and Standardization of Mesopic Visual Acuity Rachelle J. Lin*, Jason S. Ng†, and Andrew L. Nguyen‡
ABSTRACT Purpose. It is well established that visual acuity (VA) decreases with luminance but the specific factors that are responsible remain unclear. The purpose of this study was to quantify the contributions of accommodative error, pupil size, and higherorder aberrations to the decrease in VA when transitioning from photopic to mesopic light levels. Additionally, repeatability of VA at photopic and mesopic levels was measured to derive a luminance recommendation for mesopic VA testing, which can provide the standardization needed for future translational clinical studies and the widespread adoption of mesopic VA testing. Methods. Monocular VAs were assessed at one photopic and three mesopic light levels: 94, 3, 0.75, and 0.38 cd/m2, with an E-ETDRS testing system in 43 normal subjects. Accommodative error, pupil size, and higher-order aberrations were obtained. Twenty subjects were retested at another visit to assess VA repeatability. Results. The mean (TSD) logMAR (logarithm of the minimum angle of resolution) VA was j0.08 (T0.06) at 94 cd/m2, 0.05 (T0.07) at 3 cd/m2, 0.16 (T0.06) at 0.75 cd/m2, and 0.27 (T0.09) at 0.38 cd/m2. Light level and accommodative error were significantly associated with VA, and light level explained 75% of the variance. The mean differences in VAs between two visits were not significantly different from zero (p 9 0.05). The coefficients of repeatability for 94, 3, 0.75, and 0.38 cd/m2 were 0.08, 0.11, 0.14, and 0.14 logMAR, respectively. Conclusions. Light level, among all other factors studied, contributes the most to the reduction in VA tested under mesopic conditions. Testing mesopic VA at 0.75 cd/m2, or about 2.0 log units less than photopic testing, provides a significant and repeatable decrease in VA similar to standardized low-contrast VA testing, and therefore this level is recommended. (Optom Vis Sci 2015;92:559Y565) Key Words: visual acuity, visual acuity testing, mesopic, repeatability
A
lthough photopic visual acuity (VA) is the most commonly used determinant of visual function, studies have shown that measures of mesopic VA can provide additional insight into functional vision loss.1Y3 The factors that may contribute to decreasing visual function at mesopic light levels, which has been associated with ‘‘night myopia,’’ include spherical aberration, accommodative error, pupil size, photon noise, and neural effects.4Y6 Studies have shown an increase in accommodative error with decreasing light levels. As retinal illuminance decreases, the accommodative posture departs from the positions indicated by the standard photopic accommodative stimulus-response curve and moves in the direction of the subject’s dark focus position.7Y9
*OD, MS, FAAO † OD, PhD, FAAO ‡ PhD Southern California College of Optometry, Marshall B. Ketchum University, Fullerton, California (all authors); and Department of Mathematics, California State University, Fullerton, California (ALN).
Pupil size also changes with light levels. Smaller pupils decrease retinal illumination and increase depth of focus. Larger pupils, resulting from lower light levels, change the aberrations that are introduced into the optical system of the eye.4 The wavefront error varies with pupil diameter.10 Spherical aberration in particular has been speculated to be a cause of ‘‘night myopia,’’ as it increases as pupil size increases.4 Other higher-order aberrations would also induce optical blur. The primary focus of this study was to assess the relative contributions of light level, accommodative error, higher-order aberrations, and pupil size to the decrease in VA at mesopic light levels. It is well accepted that decreased retinal illumination causes decreased spatial resolution.11 However, the quantification of the relative contributions of the multiple factors has not been well studied. As there is no current standardized procedure for assessing mesopic VA,12 the secondary goal of this study was to collect normative VA values at three mesopic light levels in an attempt to identify a particular mesopic light level upon which to build future studies (i.e., proposing to ‘‘standardize’’ a mesopic light level so
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
560 Mesopic Visual AcuityVLin et al.
that future clinical research can be more cohesive). Although various studies involving mesopic VA exist, there is no standard luminance level across these studies, making it difficult to translate the various studies into clinical practice and make progress in systematically understanding how mesopic vision function may be affected in the case of ocular pathology. It should be noted that a level of 3 cd/m2 has been used as a standard for mesopic VA and contrast sensitivity testing for Food and Drug Administration trials evaluating refractive and intraocular lens surgery outcomes,13,14 and for this reason, we not only evaluated this particular luminance level but also tested at two additional lower mesopic levels (0.75 and 0.38 cd/m2), which can be easily achieved with commercial filters. Additionally, we evaluated the repeatability of VA at these light levels to aid in determining a possible recommended clinical standard.
METHODS Subjects Forty-six subjects were recruited from the general campus community and the surrounding general population. A 0.15 logMAR (logarithm of the minimum angle of resolution) difference between photopic VA and mesopic VA at 3 cd/m2 was anticipated, with further decreases in VA at lower mesopic light levels.15 Based on an assumed difference of at least 0.15 logMAR between photopic and mesopic light levels and a conservative SD of 0.3, power analysis using a paired t test indicated that a sample size of at least 34 subjects would be needed to demonstrate a significant difference at the 5% level with a power of 80%. Additional subjects were recruited to account for subject loss or data collection insufficiency. Using intraclass correlation (ICC) power analysis, sample sizes of 20 (or 25) subjects were calculated to achieve 78% (or 86%) power to detect an ICC of 0.70 under the alternative hypothesis when the ICC under the null hypothesis is 0.30 using an F test with a significance level of 0.05. From these analyses, a total of 21 subjects were asked to return for repeat testing. Informed consent was obtained from all subjects. The exclusion criteria were subjects with ocular conditions or prior surgeries known to have a significant adverse effect on media clarity or who had best-corrected distance VA of 20/25 or worse. The procedures complied with the Declaration of Helsinki and were approved by the university institutional review board.
Procedures All subjects were tested monocularly and were fully corrected, wearing their habitual spectacle correction, or no correction if emmetropic, during the tests. Subjects had a mean (TSD) calculated equivalent spherical refraction of j3.71 (T2.63) diopters (D) and a range of 0.00 to j10.12 D. Subjects viewed the VA tests while seated at the chin and headrest of an open-field autorefractor. In this way, it was possible to measure VA and accommodative response under the same viewing conditions. Visual acuity was assessed with a validated computerized VA test (M&S Study, M&S Technologies, Niles, IL), which uses the
E-ETDRS algorithm16 at a test distance of 3 m. The stimulus display (Dell E198FP, 19µ, 1280 1024) subtended 7.2 degrees horizontally and 5.8 degrees vertically. The E-ETDRS protocol uses a single-letter presentation with crowding bars. Letters are presented at random from a limited letter set. Subsequent stimuli were not presented until the subject gave a response. Threshold was determined using a modified staircase technique. Visual acuity was recorded as the output ‘‘letter score’’ by the software, which was converted to the equivalent logMAR value for data analysis.16 Photopic VA was assessed at 94 cd/m2.13 Mesopic VA was assessed at three light levels: 3, 0.75, and 0.38 cd/m2. The 3 cd/m2 light level was achieved by reducing the monitor luminance. The 0.75 cd/m2 light level was achieved by the addition of a 0.6 neutral density filter (#210; Rosco Laboratories Inc, Stamford, CT) over the entire monitor screen when at a luminance of 3 cd/m2. An additional 0.3 neutral density filter (#209) lowered the total luminance to 0.38 cd/m2. The letter contrast (Weber fraction) was always greater than 90% at all light levels. Luminance levels were measured and calibrated with a Minolta LS-110 spot photometer (Konica Minolta, Ramsey, NJ). The open-field autorefractor uses a large beam-splitter, which caused a 13% decrease in light transmission. Therefore, all letter luminances and light levels were measured through the beam-splitter, in the same manner the subjects viewed the stimulus. The light level at which each subject began VA testing was randomized to avoid the potential confounding effect of letter set memorization.17 The subject was allowed 5 minutes of adaptation time before the first VA test. In between subsequent trials, the subject was given adaptation time while the acuity program was reset. Similar adaptation times were given, regardless of light level. Accommodative posture and the corresponding pupil size were measured with an open-field autorefractor (Grand Seiko WAM5500, Shigiya Machinery Works Ltd, Japan). The autorefractor was positioned directly in line with the VA testing monitor. Letters that were generally twice the size of the threshold VA at each light level were used to measure accommodation. Three measurements of accommodative posture were taken at each light level and averaged for data analysis. The difference between the nominal accommodative stimulus and accommodative response was calculated as the accommodative error. The absolute value of the accommodative error, thereby eliminating the numerical difference between equal amounts of accommodative lead and accommodative lag, was used as a variable in the regression model. Aberrometry measurements were collected with a Marco OPDScan III refractive power/corneal analyzer (Marco, Jacksonville, FL), operating in the ‘‘OPD/CT Measurement Mode.’’ The Marco OPD-Scan III provides pupil sizes and aberrometry readings for two preset light levels: ‘‘photopic’’ and ‘‘mesopic.’’ The luminances used for these tests are not specified by the manufacturer. Therefore, the mesopic testing procedure was not varied between subjects. The total higher-order (third through sixth order) root-mean-square (RMS) values on uncorrected subjects as determined at the instrument’s mesopic pupil diameter size as well as for constant wavefront diameters of 4 and 6 mm were collected. To assess repeatability of VA, all subjects were informed and given the option to return for a second visit until the first 21 subjects who volunteered were tested. One subject was not included in the final analysis because the subject’s initial visit had
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al.
561
0.24 logMAR at 0.75 cd/m2, and 0.35 logMAR at 0.38 cd/m2 (Table 1). The mean VA at each light level differed significantly (t Q 11.92, p G 0.001). The measured decrease from photopic VA to mesopic VA was also statistically significant for all three mesopic light levels (t Q 14.20, p G 0.001).
been excluded because of insufficient RMS data. Therefore, repeatability data were collected from a total of 20 subjects.
Data Analysis Anonymized data from the subjects were compiled in Excel and statistical analyses were performed in R (R Foundation, Vienna, Austria) and Stata (StataCorp, College Station, TX). The primary outcome measure, VA (logMAR), was tested for normality using the Kolmogorov-Smirnov test to assess the validity of analysis using parametric statistics. Mean VAs, pupil areas obtained from the autorefractor, and accommodative errors were examined using analysis of variance tests with post hoc paired t tests or Tukey 95% simultaneous confidence intervals when appropriate.18 A linear mixed-effects multiple regression model, adjusted for age, using a stepwise backward method was used to examine the main effects of light level, accommodative errors, and pupil area on VA.18 The repeatability data were analyzed by deriving ICC coefficients and using Bland-Altman analysis. Multiple measures of repeatability were derived to allow others to more easily compare repeatability data from other studies. The ICC statistic was selected to assess the repeatability of the measurements taken from the first visit and the second visit. Bland-Altman plots were generated, and the 95% limits of agreement (LOAs) were derived.19 Coefficients of repeatability (CORs), a measure of the LOAs, were calculated by multiplying the SD of the within-subject differences between the two visits by 1.96.
Accommodation On average, an accommodative lag was found at all four light levels (Table 1). Only 14 subjects had accommodative lead at any of the four tested luminance levels. Accommodative error was significantly different across light levels (p = 0.001) and was significantly different between all the mesopic light levels and the photopic light level (t e j3.16, p e 0.01), but no significant differences were found between the mesopic light levels (p Q 0.89). Absolute accommodative error was also significantly different across light levels (p = 0.04) and was found to be significantly different between the photopic level and 0.75 cd/m2 (p = 0.02), but no significant differences were found between other pairs of light levels (p Q 0.41).
Pupil Size The mean (TSD) pupil areas as calculated from the measured pupil diameters are shown in Table 1. Pupil size was significantly different across light levels (p G 0.001). The increase in pupil size at lower light levels was expected. Pupil area was found to be different for each light level compared with one another (t Q 3.78, p e 0.001), except for the pupil areas at the light levels of 3 and 0.75 cd/m2 (p = 0.07).
RESULTS Of the 46 subjects enrolled in the study, data from three of the subjects were not included in the final analysis because of insufficient data. The aberrometer did not record the higher-order RMS measurements from two subjects, and accommodation could not be consistently measured through the third subject’s high myopic spectacle correction. The resulting 43 subjects (14 men and 29 women) had a mean (TSD) age of 24.7 (T3.0) years with a range of 14.5 to 31.8 years.
Aberrations The mean (TSD) total higher-order RMS for the analyzed mesopic zone pupil diameter is shown in Table 1. The data (not shown) were also tabulated using constant wavefront diameters of 4 mm (n = 41) and 6 mm (n =33).
Visual Acuity
Linear Regression Model
No VA data were found to differ significantly from the normal distribution (p 9 0.05). The mean decrease in VA, as compared with the photopic condition, was 0.13 logMAR at 3 cd/m2,
The initial model examined age, higher-order aberrations, absolute accommodative error, pupil area, and light level. The higher-order aberration term was analyzed for the constant
TABLE 1.
Mean values for the 43 subjects Photopic 2
VA, mean T SD, logMAR Accommodative error, mean T SD, D Absolute accommodative error, mean T SD, D Pupil diameter, mean T SD, mm Pupil area, mean T SD, mm2 Mesopic zone diameter, mean T SD, mm Higher-order RMS T SD, Km
Mesopic 2
94 cd/m
3 cd/m
0.75 cd/m2
0.38 cd/m2
j0.08 T 0.06 +0.34 T 0.27 0.36 T 0.25 4.6 T 0.8 17.5 T 6.2
0.05 T 0.07 +0.25 T 0.31 0.32 T 0.23 6.3 T 0.9 31.4 T 8.3
0.16 T 0.06 +0.23 T 0.26 0.28 T 0.20 6.5 T 0.9 33.9 T 8.6 6.54 T 0.85 0.464 T 0.248
0.27 T 0.09 +0.25 T 0.32 0.32 T 0.25 6.9 T 0.6 37.5 T 7.1
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
562 Mesopic Visual AcuityVLin et al.
wavefront diameters and the aforementioned mesopic zone pupil diameter. Light level was coded within the model using three indicators, with the photopic light level being the reference. The model showed that age, pupil area, and aberrations were not significant variables. Thus, backward stepwise regression resulted in the most parsimonious model consisting of only two variables: light level (p G 0.001) and accommodative error (p = 0.003). A plot of VA as a function of absolute accommodative error is shown in Fig. 1. Bestfit lines for all light levels were determined, and it was found that all of the slopes did not significantly differ from each other (t e 2.65, p Q 0.01, Bonferroni-adjusted > level of 0.008 for each test). All slopes for the mesopic light levels significantly differed from zero (R2 Q 0.17, t Q 2.90, p e 0.006). The photopic line’s slope did not differ significantly from zero (p = 0.18). Subjects with larger accommodative errors had additional losses of about 0.1 to 0.2 logMAR under the mesopic viewing conditions. However, the expected influence of accommodative error on VA varies with pupil size.20,21 Other than light level, only accommodative error was significantly (p = 0.003) associated with logMAR VA. Holding light level constant (at one of the tested lighting levels) resulted in an average logMAR increase (worsening) of 0.076 for every 1 D of absolute accommodative error. Additionally, light level and absolute accommodative error were examined in a mixed-effects analysis of variance model that used a single variable for all light levels to determine which of the two factors had a bigger contribution to the observed changes in VA. Light level (F3,125 = 523.24, p G 0.0001) was significantly associated with VA. Examination of the sum of squares term showed that light level explained 75% of the variance in VA. It was also determined that if retinal illuminance (luminance pupil area) was used in the model, instead of luminance, much less variance (43%) could be explained.
Repeatability Although 21 subjects returned for the second visit (~3 weeks later), one subject was not included in the final analysis because of insufficient RMS data. The resulting 20 subjects (8 men and 12 women) had a mean (TSD) age of 25.1 (T3.6) years with a range of 14.5 to 31.8 years. The mean ICCs (95% confidence interval) for 94, 3, 0.75, and 0.38 cd/m2 were 0.83 (0.63 to 0.93), 0.75 (0.48 to 0.89), 0.55 (0.16 to 0.79), and 0.67 (0.34 to 0.86), respectively. The mean differences in VAs between visit 1 and visit 2 for subjects were not significantly different from zero (p 9 0.17), and thus calculations of the CORs were valid using the formula described by Bland and Altman.19 The CORs for 94, 3, 0.75, and 0.38 cd/m2 were 0.08, 0.11, 0.14, and 0.14, respectively. Repeatability improved with higher light levels, but the difference between mesopic light levels by the COR measure was only 0.03 logMAR. The mean logMAR difference and T95% LOA (1.96 SD) was j0.004 T 0.076 at 94 cd/m2, j0.017 T 0.106 at 3 cd/m2, j0.002 T 0.139 at 0.75 cd/m2, and j0.021 T 0.139 at 0.38 cd/m2. BlandAltman plots were generated (Fig. 2) showing the repeatability results. The 95% LOAs show that VA under the various mesopic light levels used in this study had similar repeatability.
DISCUSSION In this study, a statistically significant decrease was found between the mean VA at the photopic light level and each mesopic light level. Several factors were assessed as possible contributing factors to the decrease in VA at mesopic light levels, including accommodative error, pupil size, and higher-order aberrations.
FIGURE 1. Visual acuity as a function of absolute accommodative error by light level. The slopes of the best-fit lines were not significantly different from each other. Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al.
563
FIGURE 2. Bland-Altman plots of logMAR VA repeatability at the four luminance levels used in this study. Bold lines represent the mean difference and the upper and lower 95% LOAs.
Analyses using a linear mixed-effects multiple linear regression model, which was adjusted for age, demonstrated that only light level and accommodative error were significant variables in the determination of VA, whereas pupil area and higher-order aberrations were not. The light level itself was found to explain the vast majority of the variance. The 0.13 logMAR decrease that occurred between 94 and 3 cd/m2 is similar to the 0.15 logMAR difference found in a previous study that used a filter over a standard ETDRS retro-illuminated cabinet.22 In the classic study by Shlaer,11 the average VAs of two subjects using Landolt Cs were about j0.05 logMAR at 4 cd/m2 and 0.33 logMAR at 0.43 cd/m2, which were somewhat better than those found in this study likely because of the more pure resolution task. In general, comparisons of acuity findings across the studies are difficult given differences in exact light level or stimuli. The 0.21 log MAR difference between high- and low-contrast VA for normal subjects as reported in earlier studies23 is similar to the 0.24 logMAR change between the photopic and 0.75 cd/m2 light level found in this study. The 95% LOAs are also similar, as
compared with published low-contrast VA LOAs of T0.129.23 Therefore, we recommend a luminance of 0.75 cd/m2, or about 2.0 log units less luminance than photopic testing levels, for mesopic VA testing. The well-established clinical norms for differences and changes in VA would be very similar between mesopic VA testing at 0.75 cd/m2 and photopic low-contrast VA testing that clinicians are already using. Finally, although we did not enroll subjects with very small pupils or dense media opacities, it should be noted that, for such subjects, a very low mesopic light level might result in a near scotopic level of retinal illumination. In this study, calculations of the mean accommodative response to the 3-m accommodative stimulus distance showed a consistent lag of accommodation at all four light levels. As expected from previous studies on the changes in average accommodative posture in the dark, the subjects in this study showed a small, but statistically significant, increase in accommodation as the light level decreased from photopic to mesopic. Because the subjects on average had a lag of accommodation, the overall accommodative error actually decreased as the light level decreased. The slightly
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
564 Mesopic Visual AcuityVLin et al.
higher accommodative errors at the photopic light level may have also been attributed to a smaller pupil size, allowing larger depth of focus. Our study did not demonstrate statistically significant differences in accommodative error between the three mesopic light levels. Nevertheless, based on Fig. 1, accommodative error may contribute more to VA decrease at lower light levels and may play a role in night myopia. It is possible that the current study did not have enough power to detect differences in accommodative error between the mesopic light levels though. A larger range of mesopic and scotopic light levels, or a larger sample size, may have been able to demonstrate a change in accommodative posture with decreasing luminance. Our study found an average logMAR increase (worsening) of 0.076 for every 1 D of absolute accommodative error. This change in logMAR VA is less than that predicted from previous studies that approximate a one-line decrease in VA for each 0.25 D of defocus.24 In this study, only the higher-order aberrations were considered because the lower-order aberrations should largely be corrected by each subject’s best correction. Ideally, a higher-order RMS value would have been obtained for each pupil size as determined during the VA and accommodation testing procedures. Then, with a separate higher-order RMS value corresponding to the pupil size at each light level, individual changes in RMS with pupil size and any correlation with mesopic VA could have been assessed more specifically. However, the maximum pupil diameter zone on the OPD-Scan III, which was obtained during the mesopic measurements, was often smaller than those obtained during mesopic VA and accommodation testing. Because RMS values for larger pupil sizes were limited by the OPD-Scan III pupil sizes (and the measurements that did have adequately sized pupil diameters would be additionally limited to 1-mm increments in the RMS output), the single mesopic higher-order RMS value was used in the statistical analysis in this study as it provided aberration assessment for the largest pupil area. Because the RMS value as determined in this study was not found to be a significant variable for VA, it is questionable whether aberrations play a significant role in ‘‘night myopia,’’ as has been previously proposed.4 It is also unknown if the results of the aberrations portion of this study would be significantly different if larger pupil zones could have been obtained with the OPD-Scan III. Pupil area was also not found to be correlated with VA in this study. The sample size calculation performed before data collection did not explicitly aim for detecting pupil area differences, and thus, it is possible that the study was underpowered to detect such a difference. Differences in pupil area between subjects at any fixed light level are directly related to individual differences in retinal illuminance under those conditions. Indirectly reexamining pupil area by way of analyzing retinal illuminance instead of luminance did not aid in explaining any more of the variance. A common clinical criterion for significant VA change is oneline difference. The 0.13 logMAR decrease in VA from 94 to 3 cd/m2, thus, may be too small to be used clinically.23,25 However, these are group data and individual differences are likely much less, which, once established for a particular patient, like Brown and Lovie-Kitchin26 have done for photopic VA, could be useful when monitoring patients. Among the mesopic light
levels, the CORs were not substantially different and thus repeatability alone is an insufficient guide to recommending a particular light level for mesopic VA testing. Because the subjects in this study were within a younger age range, further studies to establish normative mesopic data in a group of older subjects could be important as means to screen or monitor various ocular conditions. This will be particularly applicable for patients with pathology that affects mesopic vision, such as rod-cone dystrophy and retinitis pigmentosa.1 Others have suggested that mesopic testing may also be useful for assessing patients at risk for age-related macular degeneration.27 The quantitative measurements may be especially useful for eye conditions that do not exhibit large decreases in high-contrast VA in early stages. Testing mesopic VA at 0.75 or 0.38 cd/m2 provides significant and repeatable decreases in VA, and thus are better alternatives to testing at the level of 3 cd/m2. Based on the similarities between the mean difference and repeatability of mesopic VA testing at 0.75 cd/m2 and photopic low-contrast VA testing, we propose that mesopic VA testing be standardized using a light level of 0.75 cd/m2. Using a neutral density filter to reduce photopic high-contrast electronic VA testing to a luminance of 0.75 cd/m2 may be an effective and repeatable method to clinically assess mesopic VA. Nevertheless, the standard recommended from the findings of this study will allow for more focused research on applying mesopic vision function findings to patient care.
ACKNOWLEDGMENTS The work in this article represents the majority of the first author’s efforts to fulfill the requirements of her MS in vision science degree. We thank the MS thesis committee members Drs. Lawrence R. Stark and James E. Bailey for their contributions to this study. Received July 10, 2015; accepted February 7, 2015.
REFERENCES 1. Petzold A, Plant GT. Clinical disorders affecting mesopic vision. Ophthalmic Physiol Opt 2006;26:326Y41. 2. Puell MC, Barrio AR, Palomo-Alvarez C, Gomez-Sanz FJ, ClementCorral A, Perez-Carrasco MJ. Impaired mesopic visual acuity in eyes with early age-related macular degeneration. Invest Ophthalmol Vis Sci 2012;53:7310Y4. 3. Tang W, Heng WJ, Lee HM, Fam HB, Lai NS. Efficacy of measuring visual performance of LASIK patients under photopic and mesopic conditions. Ann Acad Med Singapore 2006;35:541Y6. 4. Koomen M, Scolnik R, Tousey R. A study of night myopia. J Opt Soc Am 1951;41:80Y90. 5. Charman N. Night myopia and the dark focus of the eye. Optician 1991;202:13Y8. 6. Banks MS, Geisler WS, Bennett PJ. The physical limits of grating visibility. Vision Res 1987;27:1915Y24. 7. Johnson CA. Effects of luminance and stimulus distance on accommodation and visual resolution. J Opt Soc Am 1976;66:138Y42. 8. Wald G, Griffin DR. The change in refractive power of the human eye in dim and bright light. J Opt Soc Am 1947;37:321Y36. 9. Leibowitz HW, Owens DA. New evidence for the intermediate position of relaxed accommodation. Doc Ophthalmol 1978;46:133Y47.
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Mesopic Visual AcuityVLin et al. 10. Ravikumar A, Sarver EJ, Applegate RA. Change in visual acuity is highly correlated with change in six image quality metrics independent of wavefront error and/or pupil diameter. J Vis 2012;12:11. 11. Shlaer S. The relation between visual acuity and illumination. J Gen Physiol 1937;21:165Y88. 12. Wood JM, Owens DA. Standard measures of visual acuity do not predict drivers’ recognition performance under day or night conditions. Optom Vis Sci 2005;82:698Y705. 13. American National Standards Institute. American National Standard for Phakic Intraocular Lenses. ANSI Z80.13-2007 (R2012). Arlington, VA: American National Standards Institute; 2007. 14. Evans DW. FDA update: contrast sensitivity testing standards. Ophthamol Manage February 2005. Available at: http://www. ophthalmologymanagement.com/articleviewer.aspx?articleid=86281. Accessed July 7, 2014. 15. Rabin J. Luminance effects on visual acuity and small letter contrast sensitivity. Optom Vis Sci 1994;71:685Y8. 16. Beck RW, Moke PS, Turpin AH, Ferris FL, SanGiovanni JP, Johnson CA, Birch EE, Chandler DL, Cox TA, Blair RC, Kraker RT. A computerized method of visual acuity testing: adaptation of the early treatment of diabetic retinopathy study testing protocol. Am J Ophthalmol 2003;135:194Y205. 17. McMonnies CW. Control of chart memory for retesting acuity. Clin Exp Optom 2001;84:78Y84. 18. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN 3-900051-07-0 2012. 19. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307Y10.
565
20. Smith G. Relation between spherical refractive error and visual acuity. Optom Vis Sci 1991;68:591Y8. 21. Smith G, Jacobs RJ, Chan CD. Effect of defocus on visual acuity as measured by source and observer methods. Optom Vis Sci 1989;66: 430Y5. 22. Lin R, Ng JS, Li J. Normative data on standardized mesopic visual acuity. Optom Vis Sci 2012;90:E-Abstract 125613. 23. Lovie-Kitchin JE, Brown B. Repeatability and intercorrelations of standard vision tests as a function of age. Optom Vis Sci 2000;77: 412Y20. 24. Rabin J. Optical defocus: differential effects on size and contrast letter recognition thresholds. Invest Ophthalmol Vis Sci 1994;35:646Y8. 25. Manny RE, Hussein M, Gwiazda J, Marsh-Tootle W, Group CS. Repeatability of ETDRS visual acuity in children. Invest Ophthalmol Vis Sci 2003;44:3294Y300. 26. Brown B, Lovie-Kitchin J. Repeated visual acuity measurement: establishing the patient’s own criterion for change. Optom Vis Sci 1993;70:45Y53. 27. Lovie-Kitchin J, Feigl B. Assessment of age-related maculopathy using subjective vision tests. Clin Exp Optom 2005;88:292Y303.
Jason S. Ng Southern California College of Optometry Marshall B. Ketchum University 2575 Yorba Linda Blvd Fullerton, CA 92831 e-mail: [email protected]
Optometry and Vision Science, Vol. 92, No. 5, May 2015
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
