Excellent in JAMA by Hwang et al1 and neat follow-up commentary by Massof. Take home message that Massof makes: “yellow night vision glasses do not live up to their product claims, but more work is needed to better understand and ultimately ameliorate a ubiquitous problem that is growing with the aging of the population.”
Over the years, two things have happened that make night driving increasingly difficult for me: I have grown old and automobile headlights have grown whiter and brighter. With the transition from kerosene carriage lamps to sealed beam incandescent and, more recently, high-intensity discharge and light-emitting diode (LED) headlamps, the color and brightness of headlights have gradually evolved from a pleasant and romantic yellow candlelight to a light as harsh, glaring, and painful as the sun is to me. So, if I want to return to the good old days, what better solution could there be then to wear yellow-lensed glasses, which should make all headlights appear to be a nostalgic 2400°K color temperature.
In this issue of JAMA Ophthalmology in a study using a driving simulator that would be coveted by players of Grand Theft Auto (Rockstar Games), Hwang et al1 address the question of whether yellow glasses promoted as night-driving aids actually help drivers, young and old, deal with headlight glare. The computer-generated view through the windows of the simulated vehicle is panoramically displayed on five 42-in flat panel video screens arranged in a 225° arc that is 37° high. A real steering wheel, accelerator and brake pedals, shift stick, rear-view mirror, and instrument panel that interact with the computer, as well as a driver’s seat, complete the virtual driving experience. The liquid crystal display video screens have a limited range of luminances that can be displayed, so Hwang et al invented a headlight glare simulator that optically superimposes dynamic images of oncoming headlights onto the video images.2 The headlight images are generated by LEDs and are the correct luminance, color, and angular distribution of the light synchronized by the computer, with the virtual approaching vehicle producing realistic headlight glare sources. Night-driving conditions were simulated with oncoming cars providing the headlight glare. The behavior measured in the study was the driver’s ability to detect a pedestrian along the side of the road, or carelessly attempting to cross the road, with and without headlight glare. Unlike in Grand Theft Auto, in which colliding with the pedestrian would be considered a minor infraction, in the Hwang et al study,1 the driver more humanely honked the horn to signal his or her detection of the pedestrian. In some cases, the pedestrian was wearing dark clothing (blue jeans and navy blue shirt) and in other cases the pedestrian was wearing an orange Hawaiian shirt instead (undoubtedly deferring to the well-known fashion preferences of one of the authors [E.P.]).
The headlight glare had an association with the amount of time it took young participants to detect the pedestrians wearing dark clothing; the association was in the same direction, but not necessarily as strong, for young participants detecting pedestrians wearing the orange shirt. The older participants were tested only with the pedestrians wearing orange shirts and the association of glare with detection times was 6 times greater than observed for the younger participants. Compared with clear lenses, none of the 3 yellow night vision glasses improved detection times with headlight glare. If anything, the yellow glasses appeared to slow older participants’ detection times for the nonglare condition (see the article’s Figure 2C1).
The authors correctly conclude that the yellow night vision glasses that were tested were not associated with improving driver performance. What this study is unable to indicate is why the yellow glasses were ineffective. The article’s Figure 1 illustrates the spectral emission function of the simulated LED headlights and the spectral absorption functions of the 3 yellow lenses being tested.1 Although the absorption of light by the yellow lenses in the blue end of the spectrum looks impressive, the scale is linear. To fully appreciate the association with vision, we should look at the amount of light reaching the retina on a log scale. I digitized the curves in the article’s Figure 1, calculated the relative retinal irradiance for the headlight at each wavelength for the clear lens (no filtering) and for the 3 yellow lenses (Figure). Those estimated spectral irradiance functions are plotted on a logarithmic scale (the raggedness of the curves is a consequence of the digitizing resolution and an unsteady hand). Note that on a log scale, the filtering effects of the yellow lenses are negligible at wavelengths above 575 nm and steadily increase as wavelengths become shorter. But spectral irradiance is only part of the story; to have an association with vision, the light must be absorbed by the photopigments in the retina.
To estimate the association of yellow lenses with rod vision, which is important under night-driving conditions, I multiplied the spectral irradiance by the International Commission on Illumination scotopic luminosity function and integrated the products across all wavelengths. When I did that using the published LED emission and yellow filter absorption spectra, I estimated that compared with the clear lens the glare effects on rods were reduced by 0.22 log unit for the Nite Lite lens (Eagle Eyes Optics), 0.36 log unit for the high-definition Night Vision lens (Idea Village Co), and 0.37 log unit for the Knight Visor lens (Blupond Inc). To estimate the association of yellow lenses with cone vision, I multiplied spectral irradiance by the International Commission on Illumination photopic luminosity function and integrated it across all wavelengths. Using that method, I estimated compared with the clear lens the glare effects on cones were reduced by 0.07 log unit for the Nite Lite lens, 0.12 log unit for the HD Night Vision lens, and 0.11 log unit for the Knight Visor. The maximum glare reduction, 0.37 log unit for scotopic vision, would be equivalent to reducing glare from a 10 000 cd/m2 headlamp to that from a 4200 cd/m2 headlamp; there probably is at least that much variability between vehicles. For photopic vision, the maximum glare reduction would be from 10 000 cd/m2 to 7500 cd/m2. From these estimates, I concluded that for all 3 yellow lenses, the filtering is too weak to have a meaningful association with visual performance. So, even if yellow lenses theoretically held promise for meaningful glare reduction, the yellow lenses being promoted for night driving are not yellow enough to have a measurable association with visual performance.
Yellow lenses always have been popular with shooters, aviators, and skiers. That is understandable because they potentially can increase contrast by reducing the adverse effects of scattered short wavelength light (Rayleigh scattering) that makes the sky appear blue, majestic mountains appear purple, and moguls and ruts disappear in flat light on the snow. The most interesting finding of Hwang et al1 is the magnitude of the adverse association of headlight glare with older people, which I can confirm from personal experience. Not only are the retinal rod responses saturated by headlights pumping out excess energy in the rods’ preferred part of the spectrum, but rods, cones, and melanopsin in ganglion cells drive pupil constriction,3 which when combined with nuclear or posterior subcapsular cataract, can further exacerbate the disabling glare and discomfort.4 Hwang et al1 provide evidence that yellow night vision glasses do not live up to their product claims, but more work is needed to better understand and ultimately ameliorate a ubiquitous problem that is growing with the aging of the population.