Blue-Light Filter from Sunset

What is behind it

The light channel to the internal clock runs via dedicated light-sensitive ganglion cells of the retina, whose photopigment has its maximum in the short-wavelength range. The action spectrum of nocturnal melatonin suppression lies at 446–477 nm, determined in 627 nocturnal suppression tests in 72 healthy people.

This is exactly the range that displays and LED light sources preferentially emit, so a filter acts at the right point. What matters, however, is the dose at the eye, and that depends not only on the spectrum but also on brightness. That is the reason why software night modes address only half of the mechanism.

What screen light in the evening actually does

The cleanest study on dose is an inpatient randomized crossover protocol with 12 healthy young adults who, on five evenings, read for about 4 hours each either on a light-emitting e-reader at 31.73 photopic lux or in a printed book at 0.91 lux.

Every single endpoint came out against the display: sleep onset latency 25.65 ± 18.78 min versus 15.75 ± 13.09 min, melatonin suppression of 55.12 ± 20.12%, melatonin onset delayed by more than 1.5 hours, REM sleep 109.04 ± 26.25 min versus 120.86 ± 25.32 min. The study shows that the problem is real — but it does not test any countermeasure.

Why the filter studies diverge

The core problem is blinding: an orange pair of glasses is immediately recognizable as such. In the Cochrane review, 59% of studies were considered at high risk of bias because participants and personnel were not blinded, and 65% because outcome assessors were not blinded.

This fits the pattern in the meta-analysis of tinted lenses: objectively, sleep efficiency comes in at Hedges’ g = 0.31 with a confidence interval of −0.05 to 0.66, that is, including zero; self-reported, the Pittsburgh Sleep Quality Index shows a clear effect at g = −1.25. Large subjective and small objective effects are a classic sign of expectation effects.

What the studies show

Screen reading in the evening, inpatient crossover

14-day inpatient randomized crossover protocol, 12 healthy young adults, mean age 24.92 ± 2.87 years, five evenings each of about 4 hours of reading on an e-reader versus a printed book. Sleep onset latency 25.65 ± 18.78 min versus 15.75 ± 13.09 min (p = 0.009), melatonin suppression 55.12 ± 20.12%, melatonin onset delayed by more than 1.5 hours, REM sleep about twelve minutes shorter (p = 0.03).

Cochrane review of blue-light filtering lenses

The methodologically strictest assessment, with 17 randomized controlled trials, sample sizes of 5 to 156 people and follow-up periods from less than one day to five weeks. Not a single study had a low risk of bias in all seven Cochrane domains, and a meta-analysis was not possible for any endpoint. On sleep quality there were six studies with 148 participants: three found an improvement, three no difference.

Only double-blind studies with objective measurement

This meta-analysis narrows things down to the methodologically best studies and finds nothing: it included exclusively double-blind crossover studies with actigraphically measured endpoints, of which three exist with a total of 49 people. Sleep onset latency mean difference −4.86 min (p = 0.54), total sleep time +8.75 min (p = 0.70), sleep efficiency −0.61 (p = 0.86), wake after sleep onset −1.47 (p = 0.83), heterogeneity zero. The point estimates point in the hoped-for direction, but the confidence intervals are too wide for a conclusion.

The software night mode in everyday life

167 young adults between 18 and 24 years were randomized to three conditions, which they followed for seven nights in the hour before bedtime: phone use with Night Shift, without Night Shift, no phone use. Measured actigraphically, there were no significant differences across the whole sample. In a post hoc stratification at more than 6.8 hours of sleep, an effect on sleep efficiency (p = 0.014) and wake after sleep onset (p = 0.013) emerged — in favor of the no-phone condition.

What the studies do not deliver

The highest level of evidence reaches no result. The Cochrane review could not calculate a meta-analysis for any endpoint and found a tie across six studies with 148 participants. If you restrict yourself to double-blind crossover studies with objective measurement, three studies with a total of 49 people remain, and all four sleep endpoints are not significant. The body of studies is small and short: samples of 5 to 156 people, follow-up from less than one day to five weeks. On serum melatonin under filter lenses, the included publications provided no data at all.

Two points are important for expectations. The substantial evidence explicitly concerns people with sleep disorders, jet lag or rotating shifts, not healthy normal sleepers. And for the widespread advice to wear the glasses 2 to 3 hours before sleeping, there is no tested protocol.

How to do it

The problem studied is about 4 hours of screen reading before bedtime at 31.73 photopic lux on five consecutive evenings — that is the dose for which the effects are supported. There is no uniform protocol for the countermeasure. Where the effect is most likely to occur: in sleep disorders, jet lag or rotating shifts, and as a shortening of sleep onset latency. The software night mode alone did not work, neither in the laboratory nor over seven nights in everyday life; in the only analysis with a signal, the no-phone condition performed best. The lever that night modes do not operate is brightness.

Safety

Blue-blocker glasses and night modes are low-risk, but not without effects. The Cochrane review reports adverse effects from nine studies with 333 participants at low certainty of evidence; those mentioned were increased depressive symptoms, headache, discomfort from wearing them and lower mood. Of practical relevance above all is color perception: orange lenses alter signal colors, which makes driving, working with machinery and climbing stairs in poorly lit rooms more difficult. With impaired vision, cataracts, night blindness or an increased risk of falling, the additional darkening is a risk of its own. Anyone with depression or bipolar disorder should not use blue blockers on their own initiative for weeks: there they are listed as a treatment approach. Blocking blue light during the day is not a meaningful transfer.

Sources

• Brainard et al., J Neurosci 2001
• Chang et al., Proc Natl Acad Sci U S A 2015
• Singh et al., Cochrane Database Syst Rev 2023
• Hester et al., Chronobiol Int 2021
• Shechter et al., Sleep Adv 2020
• Luna-Rangel et al., Front Neurol 2025
• Nagare et al., Lighting Res Technol 2019
• Duraccio et al., Sleep Health 2021