"Light works as if it’s a drug, except it’s not a drug at all."

-George C. Brainard, PhD, director of the Light Research Program at  Thomas Jefferson University in The New York Times (2011)

LED Lighting, Screens and Health

Flicker 100 Hz: 

Scientific Literature

Is flickering light unhealthy?

Visible and invisible flicker  100 Hz: Biological effects of light flicker in the scientific literature

Flicker ≥100 Hz is found in most modern lighting sources, with the exception of a small subset of LED lights that are completely flicker-free (see Background: LED Lights). The research directly relevant to the ≥100 Hz flicker of LED lighting is discussed in this section, whether or not it is noticeably visible. There are also multiple potential sources of flicker ≥100 Hz from LED screens (see Background: LED Screens).

In addition to the studies described below that attempt to assess the effects of light flicker or screen use on humans, there are also clinical observations from medical professionals in the scientific literature that artificial light, especially non-incandescent artificial light, or screen use are particularly triggering to at least a subset of patients experiencing photophobia, while natural sunlight causes "less trouble." (reviewed in Katz & Digre, 2016; also see Survey: Discussion).

Katz, B.J. & Digre, K.B. Diagnosis, pathophysiology, and treatment of photophobia.  Survey of Ophthalmology. 61, 466-477 (2016). https://www.sciencedirect.com/science/article/abs/pii/S0039625715300072

Flicker of at least 800 Hz may be visible as flashing if the field isn't uniform and gaze is unrestricted

Davis et al. (2015) in a study of 10 normal volunteers, showed that they can consciously perceive at least 200 Hz flicker, and for some 800 Hz flicker, as visible flashing if a flickered image includes one spatial edge, a frequency much higher than the historic estimates of critical fusion frequency (CFF). Their study uses projectors to create the images, but they mention that they have also observed the effect on every computer monitor they have tried. They mention that the effect was even greater if the image contained more than one edge. They provide evidence that the definition of critical flicker fusion frequency as less than 100 Hz only holds true if the flicker is a uniform field of light. 

It would be interesting to repeat this study with more people, including those who are sensitive to flicker.

Davis, J., Hsieh, YH. & Lee, HC. Humans perceive flicker artifacts at 500 Hz. Sci Rep 5, 7861 (2015). https://doi.org/10.1038/srep07861

Flicker of at least 10,000 Hz can be visible through the stroboscopic effect

The stroboscopic effect is the phenomenon where an object moving rapidly across the field of view appears to move in choppy jumps due to its repeated, transient illumination by flickering ambient light. This may create potentially dangerous optical illusions. For example, when observed in flickering light, rotating machinery might appear to be rotating forwards, backwards, or even appear to be stationary, depending on the frequency of rotation relative to the flicker frequency. The stroboscopic effect can also be observed by rapidly waving a wand or one's fingers in flickering light and seeing multiple discrete images rather than the blur of continuous motion.

In a very small study of ten average healthy people, 20% of people were able to detect 25% flicker and 100% flicker at 10,000 Hz indirectly via the stroboscopic effect (Bullough et al., 2012 ). The study did not test faster flicker frequencies, so it is unclear how fast the flicker needs to be in order to be undetectable. Subjects rated flicker slower than 3000 Hz as “unacceptable.” Migraine patients or other sensitive individuals were not included in the study.

The study by Bullough et al., (2012) is the evidence for a prevailing opinion among lighting manufacturers that humans aren’t bothered by flicker ≥3000 Hz, since 10 healthy volunteers reported that the stroboscopic effects of 3000 Hz flicker were “acceptable” to them when viewed for less than a minute. This argument has several flaws, including that sensitive individuals were not included in the study, the flicker exposure times were so brief that even sensitive individuals may not have started to experience health effects, and the sample size was small. I think that instead, a key take-away from this study should be the fact that 20% of healthy volunteers were still able to detect 10,000 Hz flicker, the highest frequency tested. There were also several caveats discussed in the paper about study design that might have led to under-reporting. 

Bullough et al. Detection and acceptability of stroboscopic effects from flicker. Lighting Research & Technology 44 (2012) 477-483; first published online 2011. http://citeseerx.ist.psu.edu/viewdoc/download?doi=

Flicker of at least 11,000 Hz can be visible as a phantom array 

If a person's eyes and a flickering light move relative to each other, the light may be consciously seen as a series of discrete images that create a pattern across the person's field of view, a phenomenon called the phantom array effect. For example,during a rapid gaze shift, a car's LED tail light may appear as multiple red lights in a pattern arranged from one side of a person's field of view to the other, rather than appearing as a single red blur.

In a study of 14 normal volunteers without any history of neurological disorders, a point source of light either had 60,000 Hz flicker or flicker at a value between  3,000 Hz and 13,000 Hz. The volunteers were asked to shift their gaze rapidly back and forth over the light and indicate in which of a pair of trials they could observe a phantom array, guessing if necessary.  The authors assumed that 60,000 Hz flicker should be too fast to be observed, based on calculations that phantom arrays might be visible with up to 21,000 Hz flicker. In each pair, one trial had 60,000 Hz flicker and the other trial had one of the lower frequencies.  Each pair of trials was repeated 15 times and the threshold at which an individual could detect a phantom array was set at the flicker frequency for which they could successfully identify the lights 75% of the time. The group averaged 75% correct at 6,600 Hz and the highest individual 75% threshold was 12,400 Hz (Brown et al., 2019). 

The above study was repeated, but in a scenario where 35 volunteers with corrected-to-normal vision were only allowed to shift their gaze over the light once per trial. This made the test more difficult.  In each pair of trials, one of the trials used 60,000Hz light and the other trial used lights with a flicker frequency between 1,000 Hz and 13,000 Hz. Each pair of trials was repeated 10 times. The analysis focuses on 24 of the volunteers who were "good observers" because they correctly identified the 1,000 Hz light in at least 9/10 tests, while the others had trouble discriminating between lights at any flicker level. The "good observer" group averaged 75% correct at 5,800 Hz and the highest individual 75% threshold was 12,100 Hz. Additionally, one individual correctly identified the lights in every trial through 11,000 Hz, but didn't score better than random chance at 13,000 Hz (Brown et al., 2019). 

These studies provide evidence that people can see flicker up to at least 11,000 Hz. The studies were small and at least the first study explicitly excluded those with neurological disorders who might be particularly sensitive, so the upper limit on visible flicker isn't yet known.

Brown, E., Foulsham, T., Lee, C-s., Wilkins A.. Visibility of temporal light artefact from flicker at 11kHz. Lighting Research and Technology, 52, 371-376 (2019).  https://www1.essex.ac.uk/psychology/overlays/2019-249.pdf

Summary: Visibility of flicker >90 Hz and its relation to IEEE recommendations for flicker limits

The IEEE 1789 recommendations for flicker limits are based on the data shown above for the visibility of flicker from Bullough et al. (2012), Roberts & Wilkins (2012), and Perz et al. (2015). Thus, recommendations for limits on flicker that might protect human health are based on these measurements of flicker visibility along with ratings of the acceptability of visible flicker by 10 normal individuals in Bullough et al (2012). In the IEEE recommendations, there isn't even the (basesless) assumption that if people can't see flicker, then it won't harm human health. Here, the even more baseless assumption is: if less than half of normal control people can see the flicker, then the flicker won't harm human health. There isn't any scientific basis for this assumption. 

The IEEE report emphasizes the need for further research, but no mechanism was created to define the parameters of such research, to fund the research, or to provide a way to collect reports of adverse health effects of LED lights from the public.

Further research extended the range of flicker visibility (Brown et al., 2019), but there has been very little research to assess the health effects of flicker ≥100 Hz. The IEEE recommendations have largely fallen out of use by the lighting industry and replaced by less stringent metrics (see Background: LED Lights).

Brown, E., Foulsham, T., Lee, C-s., Wilkins A.. Visibility of temporal light artefact from flicker at 11kHz. Lighting Research and Technology, 52, 371-376 (2019).  https://www1.essex.ac.uk/psychology/overlays/2019-249.pdf

Bullough et al. Detection and acceptability of stroboscopic effects from flicker. Lighting Research & Technology 44 (2012) 477-483; first published online 2011. http://citeseerx.ist.psu.edu/viewdoc/download?doi=

The Institute of Electrical and Electronics Engineers, Inc. IEEE Std 1789™-2015: IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. 2015. http://www.bio-licht.org/02_resources/info_ieee_2015_standards-1789.pdf

Perz, M. et al. Modeling the visibility of the stroboscopic effect occurring in temporally modulated light systems. Lighting Research & Technology. 47, 281-300 (2015). https://journals.sagepub.com/doi/10.1177/1477153514534945

Roberts, J.E. & Wilkins, A.J. Flicker can be perceived during saccades at frequencies in excess of 1 kHz. Lighting Research & Technology. 45, 124-132 (2012). https://journals.sagepub.com/doi/abs/10.1177/1477153512436367?journalCode=lrtd

Better detection of the phantom array effect weakly correlates with more visual discomfort when reading

The above study by Brown et al. (2019), includes a survey of discomfort or difficulties when reading. The Survey of Visual Discomfort (Conlon, et al., 1999) focuses mainly on visual discomfort or visual disturbance when reading (20 of 23 questions), although one question asks about eyestrain when viewing patterns, one question asks about eye strain with fluorescent lights, and one about headache with fluorescent lights. There is a weak correlation between being able to detect higher frequency flicker through a phantom array and having a higher visual discomfort score in both versions of the study (with multiple gaze shifts or only one gaze shift), although there is significant variability in survey responses. Because the survey is mostly focused on reading problems, there isn't enough data to know whether there is any correlation between an ability to see the phantom array and other potential health conditions like dry eye, or headache. It would be interesting to repeat the study with more people, including potentially sensitive individuals, and to administer a more comprehensive health survey.

Brown, E., Foulsham, T., Lee, C-s., Wilkins A.. Visibility of temporal light artefact from flicker at 11kHz. Lighting Research and Technology, 52, 371-376 (2019).  https://www1.essex.ac.uk/psychology/overlays/2019-249.pdf

Elizabeth G. Conlon, William J. Lovegrove, Eugene Chekaluk & Philippa E. Pattison (1999) Measuring Visual Discomfort, Visual Cognition, 6:6, 637-663, DOI: 10.1080/135062899394885

Greater flicker of commercial LED light bulbs correlates with a higher rate of detection of the stroboscopic effect and with higher rates of annoyance

Veitch & Martinsons (2020) in 2018 used a commercial LED light bulb to illuminate a rotating disc with a white dot in an otherwise dark room and asked 85 university students (18-32 years old)  to indicate whether they could see the stroboscopic effect and to rate the light in terms of acceptability and annoyance. The 10 repetitions of tests for each of 5 bulbs took about 50 minutes in total. Light bulbs were operated either at 120 Hz (Canada) or 100 Hz (France), percent flicker varied from 2.1% to 91.5%, flicker index varied from 0.0043 to 0.2999, and SVM varied from close to zero for the lowest flicker bulbs (0.00 or 0.04) to about 3 for the highest flicker bulbs. None of the bulbs in the study were completely flicker-free. The stroboscopic effect could be detected by at least some of the study participants for all of the tested bulbs, with rates of detection increasing as the flicker of the bulbs increased. Overall, ratings of annoyance were higher for the higher flicker bulbs than for the lower flicker bulbs. Additionally, all of the study participants were administered a test for sensitivity when viewing a repetitive pattern and the 30% of participants with greatest pattern glare sensitivity did not have differences in their ability to observe the stroboscopic effect, but gave higher annoyance ratings for the higher flicker bulbs than the other 70% of participants.

While it is commendable that this study tested more individuals than other studies and attempted to distinguish between possible sensitivity levels of study participants, it isn't possible to infer anything about the health effects of LED lights from this study given the study parameters, including the relatively short periods of exposure to the restricted field of light from the LED bulbs. Additionally, care should be taken when designing studies and interpreting results in terms of assuming that sensitive individuals have been included in the study. When study participants are recruited from a university or from a particular workplace, the lighting conditions or requirements for screen use in the institution can have selected against individuals sensitive to LED lights studying or working at the institution. Individuals who know they can be sensitive to lights, such as migraine patients, may be less likely than others to volunteer for such a study, creating an additional form of bias in the participant pool.

The assumption by Veitch & Martinsons that the Brundrett (1974) observation of greatest flicker visibility and headache incidence for young people exposed to magnetically ballasted fluorescent lights with significant 50 Hz flicker would also hold true for people exposed to 100 Hz/120 Hz LED lights presents some questions. First, as discussed in Flicker below 100 Hz, the absence of gender demographics and perhaps information about relative amounts of sunlight exposure raise the possibility that other factors besides age may explain some of the Brundrett data. In a response following the article, Professor R.G. Hopkinson who conducted many flicker investigations 20 years prior to Brundrett (1974), noted that while there is consensus around higher flicker visibility at younger ages when measuring CFF, they had observed most complaints about distress from flicker coming from the 35-45 age group, especially from women. The idea that flicker visibility decreases with age is based on lower measures of CFF at < 50 Hz with age, which is attributed to lesser illumination of the retina with age. It is unclear that such data can be extrapolated to faster LED flicker. Additionally, there was not a prior consensus in the literature that flicker visibility correlates with distress caused by the flicker, and Hopkinson suggests that some data shows that there may not be a correlation. The lights in the Brundrett (1974) study have fairly severe flicker, with 45% of people reporting headache and 40% reporting eye strain, suggesting that the 50 Hz component of the flicker was probably quite significant. It is not known that the trends in this data will also apply to higher LED flicker frequecies, even if the reported trends are valid.

It may be significant that the Veitch & Martinsons study shows that some people can detect 2.1%  flicker at 100 Hz through the stroboscopic effect, the lowest % flicker tested in the study, however, the highest rate of detecting that flicker was only 25% of the time for at least one of the 27 participants in the portion of the study conducted in France. Repetition would be helpful to verify this result.

Veitch, J.A. & Martinsons, C. Detection of the stroboscopic effect by young adults varying in sensitivity. Lighting Research & Technology. 52, 790-810 (2020). https://journals.sagepub.com/doi/full/10.1177/1477153519898718

Flicker of fluorescent lights: headaches, eyestrain, fatigue, and reduced visual performance

Fluorescent tube lights using magnetic ballasts flicker at 120 Hz in the Americas and 100 Hz in Europe, have a percent flicker modulation depth (% flicker) of about 40% and are associated with multiple adverse health effects. These lights are especially problematic as they age because they also begin to flicker at the mains frequency of 60 Hz in the Americas or 50 Hz in Europe, in the obviously visible range. Newer fluorescent tube lights with high frequency electronic ballasts flicker at more than 20,000 Hz and reduce, but do not eliminate, the modulation depth (% flicker) of the 120 Hz/100 Hz component of the flicker (see graphs of fluorescent light flicker with electronic ballasts in Poplawski et al., 2011). Compact fluorescent lights (CFLs) have 120 Hz/100 Hz flicker and those using electronic ballasts have reduced flicker modulation depths at 120 Hz/100 Hz compared to those using magnetic ballasts (Poplawski and Miller, 2011).

Fluorescent tube lights with magnetic ballasts have been linked to more health effects (such as in Brundrett 1974, described in Flicker below 100 Hz) than fluorescent tube lights with electronic ballasts. In a double-blind study of fluorescent lights by Wilkins et al. (1989) administered in to office workers in a UK government legal department in 1986-1987, magnetically ballasted fluorescent lights with 100 Hz flicker that had 43-50% flicker were associated with more headaches and eyestrain than 32,000 Hz electronically ballasted fluorescent lights with 64,000 Hz flicker that had less than 7% flicker of the residual 100 Hz component of the flicker (see Figure 1 in Wilkins et al. (1989) for a graph of the flicker showing the residual 7% flicker at 100 Hz). Participants completed a survey asking whether headache or eyestrain were experienced each day. The original building lighting was magnetic ballast lights with 43-50% flicker and the electronic ballast lights with 7% flicker were introduced in some, but not all, offices during the course of this study. The lights were changed partway through the study. Survey data included in the study were collected from 159 individuals overall, 42% were women. 42 participants experienced both lighting conditions (43-50% flicker and 7% flicker), with 20 experiencing the 7% flicker in the first part of the survey period and 22 in the second part of the survey period. Notably, only 5 of 159 study participants used computer terminals, so screen use was not a significant factor in interpreting results in this study.

More survey respondents did not report any headache or eye strain days with electronic ballast lighting compared to magnetic ballast lighting (Wilkins et al.,1989). In the overall data, there was a weak correlation of headache with magnetic ballast lighting compared to electronic ballast lighting when comparing the group averages of the weekly headache incidences  (p = 0.059). There was a stronger correlation between eyestrain and magnetic ballast lighting in the averaged group data (p=0.037). However, considering the distribution of individuals' average weekly headache or eyestrain incidence showed a clearer pattern and revealed that most of the correlation between headache or eyestrain and lighting type was due to a small subset of sensitive people. A small subset of survey respondents (~8%) averaged more than 2 headache days per week under magnetic ballast lighting. Similarly, a small subset of survey respondents (~8%) averaged more than 2 eyestrain days per week under magnetic ballast lighting.  No survey respondents reported more than 2 headache days per week and only one survey respondent reported more than 2 eyestrain days per week under electronic ballast lighting. Six out of seven people who averaged more than 2 headache days per week with magnetic ballast lighting also reported eyestrain ~2 or more days per week, while 1 person with frequent headache did not report any eyestrain and 1 person with frequent eyestrain averaged less than 1 headache day per week.Thus, there seems to be a small subset of individuals who are particularly sensitive to the flicker of magnetically-ballasted fluorescent lights in terms of its ability to cause very frequent headaches and/or eyestrain, with most, but not all, reporting both. 

Headache did not strongly correlate with eyestrain in individuals reporting these symptoms on about 1 day per week or less - those data are scattered among the possible combinations (Wilkins et al.,1989)

None of the study participants indicated that they thought the lighting was responsible for causing their headaches or eyestrain, suggesting that people are unlikely to report lighting as a cause of their health concerns when they aren't aware that the lighting flickers. This led the authors to speculate that lighting flicker might be a factor in causing "building sickness" (the phenomenon where something unidentified about the environment in a workplace seems to make some workers sick) more generally beyond their study (Wilkins et al.,1989)

Of note, Wilkins (1995) mentions in a different context having met multiple patients who developed an aversion to fluorescent light following significant screen (visual display terminal) use, to the point that these individuals avoided leaving their homes because of the difficulty of avoiding fluorescent light outside of the home. While it is not possible to draw definitive conclusions from this report, it is quite possible that exposure to screen flicker might have sensitized these individuals to the flicker of fluorescent light.

Table 1 of the IEEE report (IEEE std 1789, 2015) summarizes data on the health effects of 100-120 Hz flicker from fluorescent lights, including headaches, eyestrain, and reduced visual performance. The IEEE report also reviews studies where ambient light with a very high flicker frequency of 100,000 Hz allowed greater clerical job performance and 1000 Hz light allowed less fatigue for factory workers than fluorescent light with 100 Hz flicker.

The IEEE report reviews anecdotal reports and limited scientific data suggesting that fluorescent light flicker might increase repetitive behaviors in autistic children.

Note that the flicker of the fluorescent lights in the above studies does not mimic the flicker characteristics of LED lights in terms of the percent flicker or shape of the waveform. Additionally, the flicker frequency of LED lights can be driven at rates greater than 120 Hz/100 Hz and previous studies of fluorescent lights are of lighting with a 120 Hz/100 Hz flicker component. It is not possible to extrapolate directly from fluorescent light studies to determine the potential health effects of LED flicker. 

The Institute of Electrical and Electronics Engineers, Inc. IEEE Std 1789™-2015: IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. 2015. http://www.bio-licht.org/02_resources/info_ieee_2015_standards-1789.pdf

Poplawski, Michael E. and Miller, Naomi J. Exploring Flicker in Solid State Lighting: What you Might Find, and How to Deal With It. United States: N. (2011). http://www.e3tnw.org/Documents/2011%20IES%20flicker%20paper%20poplawski-miller-FINAL.pdf

Wilkins et al. Fluorescent lighting, headaches and eyestrain. Lighting Research and Technology, 21, 11-18 (1989). https://www.researchgate.net/publication/258168086_Fluorescent_lighting_headaches_and_eyestrain

Wilkins, A.J. (1995) Visual Stress. Oxford University Press. 194 pp. http://www1.essex.ac.uk/psychology/overlays/book1.pdf Warning: This book contains multiple patterns that may bother sensitive individuals. The book warns those with epilepsy and migraine not to look at the frontispiece. Patterns in the book triggered my "LED" symptoms when I ignored the warning.

Ambient sunlight reduces health effects of fluorescent light flicker

Wilkins et al. (1989) show that more ambient sunlight in offices decreases the incidence of headache associated with 100 Hz fluorescent light flicker for office workers. This might be because the addition of continuous sunlight reduces the brain's sensing of the flicker of the fluorescent lights. It could also be because fluorescent lights were switched on less frequently in offices that received more sunlight. Further study would be needed to evaluate these hypotheses. 

Wilkins et al. Fluorescent lighting, headaches and eyestrain. Lighting Research and Technology, 21 (1989) 11-18. https://www.researchgate.net/publication/258168086_Fluorescent_lighting_headaches_and_eyestrain

Health effects of flicker from LED lights: Very little research

Some LED lights do not flicker at all. Some LED lights have much greater flicker than incandescent or fluorescent lights. Some LED lights have intermediate flicker  (Poplawski and Miller, 2011; IEEE std 1789, 2015). The amount of LED flicker depends on the engineering of the circuitry controlling the LED light (see Background: LED Lights). 

Prior to about 2019, no research had been done to attempt to directly evaluate the health effects of the flicker of LED lights, either in normal controls or in potentially sensitive populations. In the absence of research, lighting manufacturers abandoned the IEEE flicker recommendations and began to focus mostly on less stringent metrics that only characterize visible flicker (Pst and SVM, see Background: LED Lights).

No large-scale study has yet been done to determine how frequently people experience health effects from LED lights, to determine the nature of the health effects, or to evaluate whether the degree of flicker or some other characteristic of LED lights correlates with symptom prevalence or severity.

Two recent studies (Zhao et al., 2020, and Sekulovski et al., 2020) of a small number of normal individuals compare the effects of different LED flicker conditions on various human responses. Neither of these studies included the key control of assessing flicker-free lights, so it is impossible to know the expected normal baseline response in either study. Also, neither study included any potentially sensitive individuals, so it is unclear whether any health effects might have been expected in the tested populations.

Zhao et al. (2020) made EEG measurements and collected subjective survey data on health effects, the visibility of flicker, and ease of completing visual tasks for 10 normal individuals (with no history of headache, eye disease, or any other neurological or mental health condition) exposed to 100 Hz, 400 Hz, and 1500 Hz flicker with a flicker percent of 10%, 30%, or 70% in a windowless room in a controlled study lasting about 5 hours. The 100 Hz conditions were rated as having more visible flicker and less acceptable lighting than the other frequencies. There were differences in brain waves with 1500 Hz flicker compared to lower frequencies. There was evidence of attenuation of the ability to observe flicker, especially 70% flicker, over the course of the study based on difference in CFF measurements at two different time points. The stroboscopic effect was more visible with lower flicker frequency or higher percent flicker. Inconsistent with other trends, a task of drawing a Chinese character was rated as most difficult at the middle flicker frequency, 400 Hz. Differences in headache or eye fatigue were not reported among the tested conditions in this study. 

A study by Signify Research, a subsidiary of the parent company of Philips Lighting (Sekulovski et al., 2020), switched the modulation depth (flicker %) of the 100 Hz sinusoidal flicker of ambient LED lighting between two values, 12.5% and 36.4%, in two work spaces occupied by 42 male and 4 female Philips Lighting employees over the course of 13 weeks from January to April. Both spaces had windows with southern exposure, so also received sunlight throughout the day, with sunlight being highest at noon and sunlight increasing overall during the weeks of the experiment. One of the spaces had electronics workbenches and the other was an open office with laptop computers. Employees were surveyed at the beginning and end of each day to assess problems with eyes, problems with vision, pain in head or upper body, fatigue, and dizziness or nausea. Mood was also surveyed. This study did not show any difference in health effects between the two levels of flicker for these workers. On average, 216 surveys were completed each week, with more than 20 surveys completed overall by 35 employees at the beginning of the day and by 24 employees at the end of the day. 

It would have been interesting to measure health effects on individuals likely to be sensitive to flicker rather than on a population of individuals likely to have been normal controls. No information was provided on whether any of the employees has migraines, tension-type headaches, or dry eye. The inclusion of only 4 women in the study reduces the chance that any of the employees gets migraines. The normal 12.5% modulation depth of LED flicker in the work spaces could have, prior to the beginning of the study, selected against any sensitive employees being able to work in the space without significant impacts on health and job performance. It would also have been very interesting to compare the effects of flickering LEDs to completely flicker-free LEDs, rather than only comparing two fairly similar levels of flicker.

Veitch et al. (2021) review the current state of research on the health effects of LED lights, discuss the limitations of the design of recent studies, and suggest future research directions. In addition to the points mentioned by Veitch et al., which include using adequate sample sizes and assessing effects in potentially sensitive populations, it is also important that studies include the extremely important negative control of flicker-free LED light, since no LED lighting flicker conditions have yet been shown to be equivalent to flicker-free LED light in terms of health effects of flicker. Study designs should include women as half of the assessed group and break down results on the basis of gender, especially since women are generally more likely to have headaches than men. Studies should also collect information from study participants about headaches (including migraine and tension type headache), eyestrain, dry eye, concussion history, and whether they experience pain, disorientation, nausea, eyestrain, headache, or any other symptoms when using or following the use of LED lights or screens.

The Institute of Electrical and Electronics Engineers, Inc. IEEE Std 1789™-2015: IEEE Recommended Practices for Modulating Current in High-Brightness LEDs for Mitigating Health Risks to Viewers. 2015. http://www.bio-licht.org/02_resources/info_ieee_2015_standards-1789.pdf

Poplawski, Michael E. and Miller, Naomi J. Exploring Flicker in Solid State Lighting: What you Might Find, and How to Deal With It. United States: N. (2011). http://www.e3tnw.org/Documents/2011%20IES%20flicker%20paper%20poplawski-miller-FINAL.pdf

Sekulovski, D., Poort, S., Perz, M., and Waumans, L. Effects of long-term exposure to stroboscopic effect from moderate-level modulated light. Lighting Research & Technology, 52, 775-789 (2020). https://journals.sagepub.com/doi/full/10.1177/1477153519881473

Veitch, J.A., et al. Correspondence: On the state of knowledge concerning the effects of temporal light modulation. Lighting Research & Technology. 53, 89-92 (2021). https://journals.sagepub.com/doi/full/10.1177/1477153520959182

Zhao et al. The effect of stroboscopic effect on human health indicators. Lighting Research & Technology. 52, 389-406 (2020). https://journals.sagepub.com/doi/abs/10.1177/1477153519871688

Public health research: Increased headache incidence with LED screen use and an unusually high incidence of headache with "photophobia" in IT workers in 2018

Increased rates of migraine without aura have been linked to screen use in the years following the introduction of LED screens (Li et al., 2020 and others, including Montagni, et al., 2015). Li et al., 2020 also observed a high incidence of tension-type headache among IT workers. 

Migraine and tension-type headache (TTH) are two of the most common classes of headache (described in the International Classification of Headache Disorders, ICHD). Migraine is an often unilateral, pulsatile, moderate to severe headache lasting 4-72 hours, often with photophobia (sensitivity to light), phonophobia (sensitivity to sound), nausea, vomiting, and symptoms aggravated by movement. The photophobia experienced during a migraine is sensitivity to light brightness. Light of normal brightness can cause pain or cause the worsening of migraine symptoms so that being in the dark tends to make the patient more comfortable during a migraine. TTH feels like a band is pressing around the head. It is usually a bilateral, mild to moderate headache lasting 30 minutes to 7 days that does not intensify with exercise, does not include nausea or vomiting, and may include either photophobia or phonophobia, but not both. 

In a large-scale study of headache among information technology (IT) workers (n=2012) in 2018 in Beijing, Li et al., 2020 found an elevated headache incidence overall, as well as a somewhat elevated incidence of migraine in females and a much more significant increase in TTH in both males and females. Their data was compared to an even larger study (n=5041) that their research group conducted of headache incidence in the general population in China in 2009 (Yu et al, 2012). Data from these two studies are summarized below. 

Survey of Beijing IT workers in 2018 (Li et al., 2020):

Survey of the general population throughout China in 2009 (Yu et al, 2012)

Li et al., 2020 show that excessive computer use is a risk factor for TTH. 

Interestingly, Li et al. further report an unusually high percentage of people they classified as having tension-type headache (TTH) also reporting photophobia, as well as an unusually high rate of photophobia reported by respondents with any type of headache (75.8%). This abnormally high rate of photophobia was an unexpected anomaly in the data that they couldn’t explain (Li et al., 2020; see section V.3). They assumed that there must have been a mistake in how the description of photophobia was communicated to participants that led to over-reporting of the symptom. They decided to exclude photophobia from their diagnostic criteria because “the prevalence estimate for migraine would otherwise have been much higher” (Li et al., 2020). Thus by excluding photophobia from their diagnostic criteria, they artificially increased the proportion of TTH diagnoses and decreased the proportion of migraine diagnoses among the IT workers with headache. 

Note that I cannot quite figure out myself how their reported 75.8% photophobia incidence (this would be 469 individuals among all respondents with any type of headache, n=619), fits with their report of 30% photophobia in TTH and 31.8% photophobia in migraine. Much more than 30% of the 440 people with TTH and much more than 31.8% of the 152 people with migraine would need to have photophobia to account for the overall rate of 75.8% of people with any type of headache having photophobia. I must be missing something about the calculations. Typically, the reported incidence of photophobia in migraine is around 80% of those with migraine (Noseda & Burstein, 2011). Since Li et al. indicate that photophobia had no diagnostic value in their study, it's likely that closer to 75.8% than 30% of those reporting either migraine or TTH reported photophobia.

Regardless of the details of the other calculations, the 75.8% incidence of photophobia for people with any type of headache is quite high, as noted by the authors. It’s possible that the unusually high rate of photophobia wasn’t a mistake, but instead evidence of a novel class of headache with a unique collection of symptoms. It's quite possible that LED screens and ambient LED lights could cause headaches associated with increased LED light and/or screen sensitivity that led respondents to indicate that they had "photophobia." If someone were bothered by LED lights or the light of LED screens, they may have answered light sensitivity questions in the affirmative, even if they were not sensitive to sunlight or experiencing the light brightness photophobia that is typical of common migraine. The association of longer computer use with a higher incidence of TTH further supports the hypothesis that LED screens may be contributing to a novel class of headache with light sensitivity. It's possible that a previously uncharacterized class of headache may have been uncovered in this study that does not neatly match either the migraine or TTH diagnostic criteria in the ICHD. The 2018 date of the Li et al. study indicates that LED screen use would have been common and ambient LED light may have also been present. The studied population was enriched for those with high screen use, so this may be one of the first studies to have assessed headaches in people using LED screens. This study may not have even fully revealed the scope of the potential issue because headache was more common in women than men, but the IT workers in the study were 77% male. The women in the study were also skewed toward younger workers, while migraine was most enriched in the 41-50 group and TTH in the 31-40 group of women. 

Importantly, the earlier study by the same research group of headache in the general population throughout China in 2009, did not uncover a similar issue with photophobia complicating diagnosis (Yu et al, 2012), perhaps because LED use was not widespread in 2009. It may only be in the past several years that people have begun to experience LED sensitivity symptoms at a measurable frequency due to the relatively recent introduction of LED lights and screens into the environment. It would be particularly interesting to repeat the population-wide headache study now and include an assessment of screen time. It would also be interesting to study ambient LED light flicker exposure for both the general population and IT workers, but this would be experimentally difficult.

IHS Classification Subcommittee. The International Classification of Headache Disorders: 3rd edition https://ichd-3.org/classification-outline/

Li, C., et al. Prevalence of primary headache disorders among information technology staff in China: the negative effects of computer use and other correlative factors. BMC public health, 20 (2020) 443. https://doi.org/10.1186/s12889-020-08497-9.

Montagni, I. et al. Screen time exposure and reporting of headaches in young adults: a cross-sectional study. Cephalalgia 36 (2016) 1020-1027. http://dx.doi.org/10.1177/0333102415620286

Noseda R. & Burstein R. Advances in understanding the mechanisms of migraine-type photophobia. Current Opinion in Neurology, 24, 197-202 (2011). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4502959/

Yu, S., et al. (2012). The prevalence and burden of primary headaches in China: a population-based door-to-door survey. Headache, 52(4), 582–591. https://doi.org/10.1111/j.1526-4610.2011.02061.x  https://www.researchgate.net/publication/224967808_The_Prevalence_and_Burden_of_Primary_Headaches_in_China_A_Population-Based_Door-to-Door_Survey

Artificial lights and screens exacerbate post concussion headache - could patients be sensitive to flicker?

Following a concussion, 10-20% of patients develop chronic symptoms called post-concussive syndrome (PCS). During PCS, the most common symptoms are headache and photophobia and these symptoms tend to occur together (reviewed in Mares et al., 2019). Headaches can be similar to either migraines or tension type headaches (TTH). The photophobia tends to worsen the headache symptoms regardless of whether the headache is more similar to a migraine or TTH. Photophobia during PCS is usually described as sensitivity to artificial light and to using screens. Mares et al. discuss that photophobia in PCS may be more of an issue than is currently recognized because there isn't a standard way to diagnose or quantify it. The reported sensitivity to artificial light and screen use is consistent with the possibility that there may be enhanced flicker sensitivity in concussion. This hypothesis is supported by the observation of higher CFF frequency in the subset of traumatic brain injury patients experiencing light sensitivity or motion sensitivity (Chang et al., 2007), suggesting that there is enhanced flicker sensitivty in these patients that could possibly be a key factor in their photophobia. Also, unlike in common migraine, the photophobia in PCS tends to be a chronic symptom, suggesting it may not be exactly the same as migraine light brightness photophobia. However, it's also possible that there could be migraine and associated light brightness photophobia in some PCS cases that could be initiated through migraine photophobia pathways (see Survey: Discussion).

Chang, T.T. et al. Critical flicker frequency and related symptoms in traumatic brain injury: Brain Injury. 21, 1055-1062 (2007). https://www.tandfonline.com/doi/abs/10.1080/02699050701591437?journalCode=ibij20

Mares, C. et al. Narrative review of the pathophysiology of headaches and photosensitivity in mild traumatic brain injury and concussion.  The Canadian Journal of Neurological Sciences. 46, 14-22 (2019). https://www.cambridge.org/core/journals/canadian-journal-of-neurological-sciences/article/narrative-review-of-the-pathophysiology-of-headaches-and-photosensitivity-in-mild-traumatic-brain-injury-and-concussion/DF6C27F3D6ED12636D4ED1C0A24A5E72

Student test scores have fallen with more smart phone use - could health effects of screen flicker and LED light flicker have a role?

There is evidence that student test scores have been falling since 2012 and poor performance is correlated with screen time (evidence in the PISA report and elsewhere discussed by PISA director Schleicher and reviewed by Thompson in The Atlantic). As an educator, I have witnessed the danger of screens being distracting and leading to adverse social interactions and social anxiety that is described in this article. However, I also think LED and screen flicker could very likely be interfering with student learning through (1) impaired ability to concentrate when in light/screens flicker (through impairment of vision/tracking), (2) physiological impairment of health/brain function caused by flicker, and (3) anxiety/depression triggered through physiological changes in the brain triggered by flicker.

Schleicher, Andreas. PISA 2022: Insights and Interpretations. Programme for International Student Assessment. https://www.oecd.org/pisa/PISA%202022%20Insights%20and%20Interpretations.pdf

Thompson, Derek. It Sure Looks Like Phones Are Making Students Dumber: Test scores have been falling for years—even before the pandemic.The Atlantic, December 19, 2023. https://www.theatlantic.com/ideas/archive/2023/12/cell-phones-student-test-scores-dropping/676889/