The AMA’s Misguided Report on Human and Environmental Effects of LED Lighting

The American Medical Association (AMA) recently issued a report and recommendations regarding the use of light emitting diode (LED) lighting in outdoor installations [AMA 2016]. The high-level goals of the report are sound and well intended. The report promotes the conversion of outdoor lighting to LED in order to reduce energy consumption and decrease the use of fossil fuels. It also provides recommendations that are intended to reduce glare and minimize the detrimental effects of light on humans and animals. Few would disagree that these considerations are central to high-quality outdoor lighting.

Regrettably, AMA’s recommendations are vague and misguided and reflect fundamental misunderstandings of illuminating engineering. The most harmful error is their recommendation to use only sources that have a correlated color temperature (CCT) of 3000 K or lower for outdoor lighting installations. Their recommendation is based on several misunderstandings and false or unsubstantiated assertions.

First, there is a frustrating lack of specificity within the AMA report that can only leave the reader to guess what might have been intended. For example, blue light is implicated throughout the document (for example, “excessive blue wavelength” [p. 1], “excessive blue spectrum” [p. 1], “intense blue” [p. 1, 2], “very blue” [p. 2], “blue-rich” [p. 3, 4, 5], “too blue” [p. 5]), yet an operational definition for blue light is not offered. Based on the AMA report, one might reasonably conclude that violet and ultraviolet radiation are of no concern, yet the use of optical radiation shorter than those typically considered “blue” would not be compatible with their goals [see, for example, Barghini and de Medeiros 2012]. What wavelength range is of greatest concern to the AMA?

Second, the AMA implies that when a city street is retrofitted with LEDs, a CCT of 4000 K or higher will lead to dissatisfaction, whereas a CCT of 3000 K or lower will be better received. Yet, oddly, even the examples cited by AMA fail to support their claim [CBS Sacramento 2014; Chaban 2015; Sciagliano 2013]. CCT is just one component of lighting quality. An appropriate before-and-after comparison requires measurement of the photometric conditions, including illuminance or luminance of relevant surfaces, spatial uniformity, luminaire luminance, and some characterization of shielding and glare control. Using a source with a CCT of 3000 K will not mitigate complaints associated with poor optics, illuminance levels that are too high or too low, or problems with uniformity. The AMA report argues that discomfort glare is exacerbated by short-wavelength optical radiation that is likely to be more prevalent in high-CCT sources and indeed there is some evidence for that argument. However, that one factor is unlikely to dominate other aspects of photometry and vision that contribute to discomfort glare, including source luminance, size, and position in the field of view. In short, it is not correct to implicate CCT as the sole or primary reason for dissatisfaction.

Third, CCT is an inadequate proxy for the photobiological potency of a light source’s spectral power distribution (SPD) [for example, Rea and others 2006; Esposito and Houser 2016]. As illustrated in Fig. 1, it is incorrect to simplify SPD with CCT when the goal is to minimize circadian disruption.

Fig. 1 A simulation was performed with MATLAB to compute 30 million SPDs from linear combinations of five LED channels (red, green, blue, warm white, cool white). SPDs within ±0.01 D_uv and ±10 K of six practical CCTs were retained. For the retained SPDs, circadian stimulus (CS) [Rea and others 2010] was calculated. The maximum achieved CS is shown for each CCT. Though the maximum CS value increases with CCT, the possible CS values in each CCT bin have a range of zero to each CCT’s maximum. Adapted from Esposito and Houser [2016].

Fourth, the AMA report is silent on the potential benefits of higher CCT lighting at the mesopic light levels typical of nighttime outdoor lighting. Higher CCT lighting is often associated with both a higher ratio of scotopic to photopic lumens (that is, higher S/P ratio) and greater sensitivity of intrinsically photosensitive retinal ganglion cells. SPDs with proportionally more short-wavelength optical radiation improve visual performance [for example, Li and others 2012] and brightness perception [for example, Bullough 2015]. Further, no mention is made that both IES and CIE permit illuminance design targets to be adjusted using the S/P ratio of a light source [CIE 2010; DiLaura and others 2011], though IES disallows using the S/P ratio to adjust light levels on roadways [IES 2014]. At an adaptation of about 0.65 cd/m² (typical of a parking deck at night), changing from 2000 K high-pressure sodium with an S/P of 0.60 to a 4000 K LED with an S/P ratio of 1.8 will allow a reduction in the illuminance design target from 6.4 to 5.4 lx, corresponding to an illuminance reduction of about 20% and associated energy savings, cost savings, and reduction in the use of fossil fuels. Not balancing the pros and cons of CCT as a design tool is a failure of omission within the AMA report.

The AMA is an influential organization and their policies carry weight. It is lamentable that their recommendations, however well intended, are too poorly formed to encourage the types of innovations that would more effectively support their goals. It could have been helpful if their report made recommendations that encouraged the development of technological advancements and design solutions that employ LEDs with thoughtfully engineered spectra, optics, and controls. Absent that, unfortunately, the AMA report offers more confusion and noise than substance.