Effects of HEV-Filtering in a Soft Contact Lens on Contrast Sensitivity With and Without Veiling Glare

Abstract

Purpose:

The contrast sensitivity function (CSF) can be used to assess even subtle changes in visual function. These changes are likely amplified in the presence of a stressor such as glare. In this study, we tested mesopic and photopic CSF in the absence/presence of an ecologically valid glare source. CSF was tested within subjects, comparing a high energy visible (HEV)-filtering contact lens against a clear comparator lens of similar material and design.

Methods:

This was a randomized, controlled, and double-masked study. Forty participants (19–65 years) provided photopic (n=20) or mesopic (n=20) CSF (1.6, 3.2, 8, 16, and 24 cpd) both with and without veiling glare. CSF was measured using a customized optical system. Two separate light channels were combined with a beam splitter that allowed contrast modulation using sine-wave gratings on glass. Glare was interposed between the grating and the subject's eye using an ancillary xenon-based optics channel.

Results:

The mean area under the CSF curve derived for both photopic and mesopic CSF (with and without glare) was significantly improved by use of the HEV-absorbing lens compared with the control. With a glare source, the least-squares mean differences (95% confidence interval) were 0.3 (0.24–0.36) and 0.14 (0.10–0.18) under photopic and mesopic conditions, respectively. In the absence of a glare source, the least-squares mean differences (95% confidence interval) were 0.26 (0.21–0.31) and 0.28 (0.20–0.37) for photopic and mesopic conditions, respectively.

Conclusion:

The HEV-absorbing lens significantly improved CSF when tested under mesopic and photopic conditions and both with and without veiling glare.

Contrast sensitivity is one of the more widely used assessments in vision science. Changes in the contrast sensitivity function (CSF) have been used to evaluate issues ranging from normal loss with aging¹ to visual,² and even neural disease.³ CSF has been used in refining machine vision,⁴ evaluating the visibility of road hazards,⁵ and to assess the effectiveness of clinical procedures.⁶ Perhaps because there are so many applications, there is an equally wide array of methods used to measure the CSF.⁷ These methods range from presentations using wall-charts, digital devices, and computer displays to relatively simple and complex optical systems.^8,9 The usefulness of CSF depends, of course, on the reliability and validity of the specific method used. Validity is often assessed¹⁰ by evaluating the general shape of the curve against the standard curve with its characteristic bandpass shape. Reliability is determined by the repeatability of the curve over time, optimally with several days between measurements.

When both reliability and validity are optimized,⁹ even small changes to the CSF are elucidative. For example, changing the adaptive state of the individual (scotopic, mesopic, photopic), and light levels used for testing, predictably shifts the peak of the curve lower and to higher frequencies. Some specific conditions can create notch changes to the curve.¹¹ For example, notch loss is an early indicator of even normal tension glaucoma.¹² The optics of the eye, specifically how the cornea and crystalline lens form retinal images, also affect the CSF, as they determine how spatial information is transmitted to the retina. This optical transfer is often described using the modulation transfer function, which quantifies how contrast at different spatial frequencies is preserved by the optical system. The neural representation of this information is reflected in the CSF.¹³ Optical factors typically affect the vertical position of the CSF, meaning they change the overall sensitivity (contrast threshold) across spatial frequencies, rather than shifting the peak spatial frequency itself. For example, individuals with diabetes (without retinopathy) show a uniform loss in sensitivity in both mesopic and photopic conditions.¹⁴ Atmospheric haze likely causes a uniform reduction in the CSF.¹⁵

In this study, we assessed the effects of adding a high energy visible (HEV) light filter¹⁶ to a soft contact lens on the photopic and mesopic CSF. We used an optical method that, in a recent study,⁹ we found to be valid and highly reliable (r=0.85) when tested over different days. The CSF has long been considered a valid measure of spatial vision because the characteristic bandpass shape of the function reflects well-established optical and neural constraints of the visual system. Deviations from this canonical shape are known to occur with optical blur, retinal disease, and neural deficits, making the CSF a widely accepted clinical and research metric of visual performance. The optics of the system allowed us to introduce an ancillary channel that provided a light source that emulated glare from natural sunlight (one of the putative benefits of HEV-filtering lenses). This veiling glare, interposed between the grating stimulus and the eye, was designed to fit the ANSI standard (ANSI Z80.12 – 2007) for disability glare. Disability glare refers to the reduction in visual performance caused by a bright light source within an observer's field of view. It does not necessarily cause discomfort (like discomfort glare) but interferes with the ability to see details or objects.

HEV-filtering devices, such as HEV-filtering ophthalmic lenses, are widely available and frequently promoted for visual comfort and ocular protection, but relatively few controlled studies have evaluated their effects on objective measures of spatial vision under conditions that approximate real-world visual challenges such as veiling glare. Most prior work has emphasized subjective symptoms or visual comfort rather than quantitative measurements of visual performance. HEV-filtering contact lenses have been even less commonly studied than other HEV-filtering devices, such as ophthalmic lenses. The present study, therefore, examined whether an HEV-filtering soft contact lens alters contrast sensitivity under both photopic and mesopic luminance levels and in the presence or absence of veiling glare. These data provide clinically relevant information regarding potential functional benefits of spectral filtering in contact lenses.

METHODS

This was a nondispensing, controlled, double-masked study with two parallel arms consisting of randomized photopic and mesopic conditions. Participants were also randomized contralaterally to the test and control lenses within each arm. The HEV-filtering test lens (ACUVUE OASYS MAX 1-Day; AOMAX) absorbed approximately 60% of the light between the wavelengths of 380 to 450 nm, whereas the “clear” control contact lens (ACUVUE OASYS 1-Day; AO1D) filtered only about 8% of the light within the same range (see the spectra of the lenses in Hammond et al¹⁷). The HEV-filtering lens attenuates short-wavelength light and falls within the spectral band classifications described by the ANSI Z80 Spectral Bands Task Force, which defines standardized terminology for ophthalmic lenses that selectively filter portions of the visible spectrum. This study occurred at a single university site (University of Georgia). The study lenses were worn in a daily wear modality for 1 to 2 hr.

Forty healthy adult participants (85% female; 85% White, 98% non-Hispanic or Latino) participated, ranging in age from 19 to 65 years (see Table 1). Participants were habitual daily wear spherical silicone hydrogel contact lens wearers, had contact lens powers within the range of −1.00 D through −6.00 D, were required to have ≤1.00 diopter (D) of refractive astigmatism in each eye, and were free of any ocular or systemic condition known to interfere with daily contact lens wear. Participants with uncorrected astigmatism greater than 1.00 D were excluded to minimize potential degradation of spatial vision that could interact with the effects of veiling glare. The study protocol was approved by an independent ethics committee and Institutional Review Board (Sterling IRB; Atlanta, GA), with local context provided by the University of Georgia Institutional Review Board. The experimental procedures adhered to the tenets of the Declaration of Helsinki. Patients were informed about the aims and methods of the study, and all patients signed a statement of informed consent. No adverse events were reported. This study was registered under ClinicalTrials.gov (NCT05582304). Participant accountability is shown in Figure 1.

TABLE 1.

Subject Demographics

Characteristic	Photopic (N=20), %	Mesopic (N=20), %	Overall (N=40), %
Gender
Female	16 (80)	18 (90)	34 (85)
Male	4 (20)	2 (10)	6 (15)
Race
Asian	1 (5)	2 (10)	3 (7.5)
Black/African American	1 (5)	2 (10)	3 (7.5)
White	18 (90)	16 (80)	34 (85)
Hispanic/Latino	0 (0)	1 (5)	1 (2.5)
Not Hispanic/Latino	20 (100)	19 (95)	39 (97.5)
Age (years)
Mean (SD)	37.6 (14.1)	37.1 (14.0)	37.4 (13.9)
Median	38.5	39.5	39.5
Range (Min–Max)	19–65	19–61	19–65

FIG. 1.

Consort diagram.

For details of the apparatus, see Hammond et al.⁹ The CSF device was built around two integrating spheres that were constructed from aluminum hemispheres (15.24 cm) and painted with nonluminescent and highly reflective (98%) white paint (Labsphere, North Sutton, NH). Each hemisphere had input and exit ports (three-inch diameter). The light source was two highly stable 520 nm lasers (fine adjustments to the power output were made possible by two small neutral density wedges, not shown in the schematics; Model DS20X90–520-120L; Apinex; apinex.com).

For photopic and mesopic testing, the mean luminance (measured using a PR-650 Spectrascan colorimeter; jadaktech.com) for each beam was set at 65 and 3 cd/m², respectively (light levels based on FDA standards). Participants were not formally light- or dark-adapted before testing. The terms photopic and mesopic refer to the luminance of the test stimulus rather than the adaptation state of the participants. Combining light (made Lambertian by internal reflection within the spheres) from the two spheres, in conjunction with a circular neutral density wedge, allowed for constant variation in the modulation of the sine waves. A dedicated photodetector (PS; United Detector Technology UDT S370 with a photometric lens; Artisan Technology; artisantg.com) was consistently used to monitor the luminance output (half of the combined beam going to the detector and half going to the eye of the subject). The circular (3.2°) gratings (1.6, 3.2, 8, 16, and 24 cpd) were carefully calibrated by the manufacturer (Applied image; appliedimage.com). The modulation accuracy within the system was also checked by moving a tiny pinhole aperture (∼0.5 mm) and photodetector (ILT 950, International Light Technologies; internationallight.com) across and very close to the surface of the grating (just the lowest frequencies were checked) while changing the position of the wedges. As shown in Figure 2, an ancillary channel was used to impose disability glare at a constant intensity (0.39 μW) over the grating stimulus (the stimulus was shuttered with 1-second exposures separated by 1-second delays). A xenon (150-W) light was used in conjunction with an achromatic lens and a diffuser to create a homogeneous broadband (CCT=7,750) veil that was imposed between the eye of the subject and the grating stimulus (through a thin-glass beam splitter). The subject's head position (∼89 cm back; i.e., the viewing distance between the stimulus and the plane of the eye) was stabilized with a chinrest and forehead assembly. Optical baffling was used throughout.

FIG. 2.

The optical system used to measure CSF (for a more detailed diagram see Hammond et al., 2023a). Light output was precisely measured using a dedicated photocell (PC). A xenon system provided the light for the glare source that was imposed over the grating stimulus using a thin-glass beam splitter (BS); L, achromatic lens, FH, filter holder and optical baffling.

On the test day, participants arrived at the test site and completed the informed consent process. After written consent was obtained, participants were given an eye examination by the licensed study optometrist, and eligibility was confirmed. The optometrist contralaterally placed one test lens and one control lens on each participant, according to the randomization scheme, masking participants to the identity of the lenses. All subjects were fit to a 0.00 D spherical over-refraction OD and OS. Eye examinations were conducted in a dedicated examination lane in a different part of the clinical site from the test room, so that experimenters conducting the CSF measurements were also masked to which eye contained the test lens, and which eye contained the control lens.

Once the study optometrist confirmed optimal lens power and fit with the experimental lenses, participants were escorted to the test room, where the apparatus was set to mesopic or photopic CSF measurement conditions, depending on the randomization scheme. Participants were seated in front of the optical assembly, and OS was patched (the test lens was randomized and could have been OS or OD). The right eye was tested first for all participants. Although right eye was always measured before left eye, treatment assignment (test vs. control) was randomized between eyes. Therefore, any potential intereye order effect was balanced across treatments by randomization. All participants were tested on all gratings (to yield contrast sensitivity at 1.6, 3.2, 8, 16, and 24 cpd) in a randomized order. Participants were presented with the first grating and confirmed that they were able to resolve the grating clearly. The experimenter then adjusted the contrast until the participant was no longer able to resolve the grating (using an ascending/descending Method of Limits, participants indicated yes they could see the vertical grating or no they could not). Three trials were completed at each spatial frequency, with rest periods between each grating, until all gratings were tested.

After the nonglare photopic or mesopic CSF was measured, the first grating was then placed back into the apparatus, and the glare stimulus was added. Participants again confirmed that they could resolve the grating clearly. The procedure was repeated until contrast sensitivity was measured at all spatial frequencies, with rest periods between each frequency. The patch was then moved to OD, and OS was aligned with the first test grating. The procedure was then repeated to measure OS with and without glare.

The statistical analyses were performed using the Statistical Analysis System software Version 9.4 (SAS Institute, Cary, NC). The area under the log contrast sensitivity function (AULCSF) was calculated using the trapezoid method with both the spatial frequency and the contrast sensitivity in logarithmic (base 10) values. The AULCSF value for each subject eye was analyzed separately for the photopic and mesopic conditions with and without a broadband glare source using linear mixed models with repeated measures. The mixed models with repeated measures model included lens type, eye dominance, and the lens type by eye dominance interaction term as fixed effects, adjusting for age and gender. The covariance of residuals for within-subject repeated measures (R-side) was modeled using an unstructured covariance structure. The least-squares mean (LSM) for each lens and the LSM difference (test – control) were estimated with the associated 95% confidence intervals in each model. The estimated mean AULCSF values of study lenses were compared under various lighting conditions without alpha adjustment for multiplicity.

RESULTS

The photopic (and mesopic) results for AULSCF are shown in Table 2, displayed as both mean (±SD) and LSM (±SE). In test versus control, the photopic AULCSF with the test lens was significantly higher than the control lens by 0.3 (P<0.0001) and 0.26 (P<0.0001) for the with and without glare source conditions, respectively. Figure 3 shows the photopic CSF for all participants, both test and control lenses, with and without glare, collapsed across age.

TABLE 2.

The Area Under the log CSF (AULCSF) Results Under Photopic and Mesopic Conditions With and Without Glare

	Mean (SD)	LS Mean (SE)	Estimated 95% Confidence Interval	P
AULCSF—photopic with glare
Control	1.62 (0.158)	1.67 (0.057)	(1.55–1.78)
Test	1.92 (0.220)	1.96 (0.057)	(1.85–2.08)
Test – control		0.30 (0.031)	(0.24–0.36)	<0.0001
AULCSF—mesopic with glare
Control	0.43 (0.140)	0.51 (0.061)	(0.38–0.63)
Test	0.57 (0.189)	0.64 (0.063)	(0.51–0.77)
Test – control		0.14 (0.019)	(0.10–0.18)	<0.0001
AULCSF—photopic without glare
Control	2.22 (0.239)	2.28 (0.062)	(2.14–2.41)
Test	2.49 (0.211)	2.54 (0.063)	(2.41–2.67)
Test – control		0.26 (0.024)	(0.21–0.31)	<0.0001
AULCSF—mesopic without glare
Control	1.99 (0.234)	1.97 (0.088)	(1.78–2.15)
Test	2.27 (0.188)	2.25 (0.087)	(2.07–2.43)
Test – control		0.28 (0.040)	(0.20–0.37)	<0.0001

FIG. 3.

Photopic contrast sensitivity curve (all subjects). Contrast sensitivity was calculated according to the Michelson contrast formula.

In with-glare versus no-glare, introducing the glare stimulus reduced sensitivity across all frequencies tested. For example, the glare source reduced contrast sensitivity by 0.49 log units for the control lens and 0.42 log units for the test lens at eight cpd. In addition, Figure 3 illustrates the eye with the test lens had improved sensitivity with and without glare, across all frequencies.

Forty participants enrolled in and completed the study. Twenty participants were randomly assigned to the photopic condition, and 20 participants were randomly assigned to the mesopic arm of the study. Figure 4 shows the area under the curve for the photopic CSF with and without glare separated for the younger (19–39 yrs) and older participants (40–65 yrs). The contrast sensitivity measured at each spatial frequency with the test lens was numerically higher than the control lens within each age group under the photopic condition with and without glare source.

FIG. 4.

Photopic contrast sensitivity curves separated into two equal age groups. Contrast sensitivity was calculated according to the Michelson contrast formula.

Figure 5 shows the mesopic CSF for all participants, both test and control lenses, with and without glare, collapsed across age. As expected, introducing the glare stimulus reduced sensitivity, across all frequencies tested.

FIG. 5.

Mesopic contrast sensitivity curves (all subjects). Contrast sensitivity was calculated according to the Michelson contrast formula.

Participants were significantly more sensitive in the eye fitted with the test lens with and without the added glare stimulus. The estimated LSM difference was 0.14 (P<0.0001) and 0.28 (P<0.0001), respectively, for the with and without glare conditions (Table 2).

Figure 6 shows the area under the curve for the mesopic CSF with and without glare for the younger (19–39 yrs) and older participants (40–65 yrs). Within each age group, the contrast sensitivity with the test lens was numerically higher than the control lens across all frequencies tested under the with and without glare source conditions.

FIG. 6.

Mesopic contrast sensitivity curves separated into two equal age groups. Contrast sensitivity was calculated according to the Michelson contrast formula.

DISCUSSION

HEV-filtering contact lenses improved contrast sensitivity both in the presence and absence of veiling glare. The broadband glare source (emulated sunlight) significantly and uniformly reduced contrast sensitivity in both photopic and mesopic conditions. ANSI Z80.12 – 2007 (R2017) states that for a light source to be considered a significant source of glare it should reduce mean contrast sensitivity by about 0.1 log unit at six cpd (as applied, for example, to intraocular lenses [IOLs]¹⁸). In this study, contrast sensitivity was assessed at 1.6, 3.2, 8, 16, and 24 cpd. At the 8 cpd condition, CSF was reduced under photopic conditions by 0.42 log units for the test lens and 0.49 log units for the control lens; under mesopic conditions, the glare source reduced CSF by 1.56 log units for the test lens and 1.38 log units for the control lens. Despite this reduction, the absolute contrast sensitivity remained higher with the HEV-filtering lens across all spatial frequencies.

Under glare conditions, based on the estimated AULCSF values (LS means in Table 2), the HEV-absorbing lens improved CSF by 18.0% and 27.5% for the photopic and mesopic conditions, respectively, as compared to the control (Table 3). Korde et al¹⁹ studied the clinical significance of the CSF and noted that a 3% to 5% improvement in AULCSF was considered clinically meaningful.

TABLE 3.

Percent Improvement in AULCSF Based on Least Squares Mean Estimates

	Improvement in AULCSF (Test – Control)/Control
	Photopic, %	Mesopic, %
No-glare	11.4	14.2
With-glare	18.0	27.5

It is perhaps not surprising that anterior filtering of highly energetic light, which is especially prone to scattering, would reduce some of the deleterious effects of that light. It has, for example, been widely observed that violet-blue filtering by the macular pigments reduces the negative effects of veiling glare on the CSF.²⁰ A similar effect has been found for violet-absorbing IOLs.²¹ The data on violet-blue light filtering contact lenses, however, are less consistent. Data from Sánchez-González et al²² found a significant benefit of a blue-light filtering (BLF) contact lens compared with a clear lens on the CSF. In contrast, Tavazzi et al²³ found, using somewhat different stimulus conditions, that BLF contact lenses only improved CSF for subjects whose contrast sensitivity was low when tested using a clear contact lens. Directly comparing results across different studies that use widely different filters (macular pigment, different IOL and contact lens types) and testing scenarios is difficult. Interpretation, however, is relatively straightforward. To the extent that glare degrades the CSF, attenuating that glare should improve performance, particularly if shorter-wavelength light contributes disproportionately to intraocular scatter.²⁴ To this end, it is obviously important for the stressor, glare in this case, to be realistic. For example, Tavazzi et al²³ concluded that it might be useful to measure CSF that mimicked more natural “solar lighting,” as we did with our glare source.

When comparing the area under the logged curves with no glare imposed and based on the LS means presented in Table 2, the CSF of the test lens was improved by 11.4% under photopic conditions and by 14.2% under mesopic conditions (Table 3). This is noteworthy because the CSF was improved in the absence of HEV light (no-glare condition), indicating the improvement must be attributed to characteristics of the test lens other than short-wave filtering. Because our method of measuring CSF minimizes higher-order aberrations and absorbance by preretinal chromophores, it likely served as a reasonable proxy for optical quality (analogous to measures such as the modulation transfer function). Although the test and control lenses were similar in base material, power, and general optical design, commercially manufactured contact lenses may differ in additional characteristics such as polymer distribution or manufacturing processes. These differences cannot be entirely excluded and may have contributed to the differences in measured optical quality between the lenses.

Sánchez-González et al, for example, argued that adding BLF to a contact lens increases tear stability.²⁵ The test lens in the current study was manufactured with a new technology that optimizes the distribution of the wetting agent polyvinylpyrrolidone throughout the lens and at the surface. Prior work has shown that the test lens has approximately a 60% greater likelihood of maintaining a stable tear film beyond 10 sec compared with the control lens.²⁴ However, the present study did not directly measure tear stability or other ocular surface parameters, and therefore, no conclusions regarding ocular surface improvements can be drawn.

This study used a contralateral design to reduce confounds and minimize intersubject variability between lens comparisons. Both test and control lenses were tested in the same subjects and under the same photopic, mesopic, no-glare, and with-glare conditions. The test and control lenses were similar in optical design (senofilcon A base material), and both were fit to a 0.00 D spherical over-refraction OD and OS. As shown in Table 3, the test lens showed an average of 11.4% AULCSF improvement compared with the control lens when under photopic no-glare conditions (no HEV present). When the same subjects and lenses are evaluated under photopic with-glare conditions (HEV light present), however, the percent improvement increases to 18.0%. Given the similarity of testing variables, it is plausible that this additional 6.6% increase in photopic AULCSF reflects the influence of the HEV filter in the test lens when evaluated in an HEV-rich environment. A similar finding is seen in the mesopic conditions. Here, there is a 14.2% improvement in AULCSF under no-glare conditions compared with a 27.5% improvement under with-glare conditions, suggesting a 13.3% contribution of the HEV filter in the HEV environment.

It should be noted that the HEV-filtering lens also has slightly lower overall spectral transmittance than the clear comparator lens. Therefore, the present data cannot definitively isolate selective short-wavelength attenuation as the sole mechanism responsible for the observed improvement in contrast sensitivity. Nevertheless, shorter wavelengths are known to contribute disproportionately to intraocular scatter (based on aberrations, Rayleigh effects, etc.). Selective attenuation of this portion of the spectrum, therefore, provides a plausible mechanism for reducing veiling glare and improving spatial vision. It should also be noted that although the visible spectral transmittance of the lenses differs by approximately 10%, the luminous transmittance for the test and control lens are similar.²⁶ An ideal experimental comparison would involve a neutral density control lens with equivalent overall transmittance but without spectral selectivity. Such a design would help isolate the specific contribution of short-wavelength filtering independent of overall luminance attenuation. Neutral density control contact lenses of this type were not available within commercially manufactured designs at the time this study was conducted.

The benefit of the HEV-filtering test lens is further demonstrated in Figures 3 and 4 where under photopic conditions and at high spatial frequencies, the test/glare condition provided the same contrast sensitivity threshold as the control/no-glare condition. In other words, participants were able to resolve the same high spatial frequencies under glare conditions with the test lens as with the control lens under no-glare conditions.

Several limitations should be acknowledged. First, the lenses compared in this study were commercially available products rather than lenses differing only in spectral transmission characteristics. Second, a neutral density control condition was not available to isolate the effects of spectral filtering from overall transmission differences. Third, the sample size was relatively modest because the study was designed primarily as a feasibility investigation. Future studies with larger samples and lenses designed to isolate spectral effects may further clarify the mechanisms underlying the improvements observed here.

In summary, high intraocular scatter has been shown to increase glare and reduce contrast sensitivity.²⁷ Short-wave filters (e.g., like macular pigment) have been shown to decrease glare and increase contrast sensitivity.²⁸ In this double-masked, contralateral, randomized controlled trial, we tested CSF under photopic and mesopic conditions while in the presence and absence of emulated sunlight. We found that an HEV-filtering contact lens improved photopic and mesopic contrast sensitivity both in the presence and absence of glare. This study was designed as a feasibility study with a relatively small sample size. Improvements of this magnitude in AULCSF are within the range previously suggested to be clinically meaningful for visual performance tasks. Despite the limited sample size, the observed effects were large enough to reach statistical significance, suggesting a potentially robust effect that merits further investigation.