Do Bright Light and Dark Glasses Help Night Workers Adapt?

The outcome was circadian phase, measured as dim light melatonin onset before and after five consecutive simulated night shifts running 2300 to 0700. Bright light was intermittent, approximately 5000 lux, 20 minutes on and 40 minutes off, four to five pulses a night, delivered in a delaying pattern whose block ended at 0500 on the first night shift and moved an hour later on each shift after that. The comparison condition was constant room light of approximately 150 lux (Crowley et al., 2003).

Among the 42 participants whose baseline temperature minimum fell at or before 0700, the average phase delay rose across conditions: 2.5 hours (SD 2.5) with normal sunglasses at 15% transmission, 5.4 hours (SD 2.6, p < 0.05) with dark lenses at 2% transmission, 6.0 hours (SD 1.0, p < 0.01) when melatonin was added to those dark lenses, and 7.1 hours (SD 2.1, p < 0.001) in the pooled bright light groups. The univariate main effect for phase delay shift was F(3,38) = 9.02, p < 0.001 (Crowley et al., 2003).

The categorical result is easier to act on. Counting people rather than percentages, 11 of the 12 bright light participants re-entrained completely, against 1 of 10 in the normal sunglasses group, where 7 of 10 did not re-entrain at all, chi-square(3, n = 42) = 21.58, p < 0.001. Adding 1.8 mg sustained release melatonin to the dark sunglasses changed nothing: the authors report no statistically significant differences between those two groups by any analysis. The melatonin comparison was the one double-blind element of the design, since light and sunglasses conditions cannot be blinded (Crowley et al., 2003).

Every group, including the weakest one, slept in bedrooms the investigators had blacked out, on an enforced schedule from 0830 to 1530. That regular early dark period appears in the authors' own recommendation alongside intermittent bright light during the night shift and sunglasses as dark as possible on the commute home, so it is a condition of the result rather than an incidental detail of the method (Crowley et al., 2003).

The headline contrast holds in only one half of the sample. Among the 23 participants whose baseline temperature minimum fell after 0700, every person re-entrained completely no matter which group they were in, including those wearing ordinary sunglasses with no bright light at all. Baseline circadian phase, rather than the assigned intervention, predicted who adapted better than anything else measured (Crowley et al., 2003).

That earlier versus later split was made after the researchers saw the baseline data, which makes it a post hoc stratification rather than an assigned factor. The paper also never describes random assignment to the six groups, and it is indexed as a controlled clinical trial rather than a randomized one. The tabulated analysis rests on cells of 10, 10, 10 and 12 people, and one group in the full design held 7, so read the percentages as counts and treat the precision as low.

A Cochrane review searched to August 2015 and included 17 relevant trials with 556 review-relevant participants across bright light, napping, and other person-directed interventions. Most comparisons yielded low to very low quality evidence, only one comparison provided moderate quality evidence, and the authors describe the included studies' results as inconclusive (Slanger et al., 2016).

The one bright light meta-analysis in that review pooled 2 studies with 184 participants and found a mean reduction of 0.83 score points of sleepiness on the Stanford Sleepiness Scale (95% CI -1.3 to -0.36), rated very low quality evidence. The protocol closest to the laboratory combination, bright light plus sunglasses at dawn, did not significantly influence sleepiness in the one trial that tested it, with 17 participants assessed by reaction time and again rated very low quality (Slanger et al., 2016).

The disagreement is between settings and designs rather than between endpoints. Crowley et al. (2003) reported where the body clock sat; Crowley et al. (2004) reported performance, sleepiness and mood from the same 67 people on the same simulated night shifts, and all three were better in the groups that re-entrained than in the group that did not, with no significant difference between partial and complete re-entrainment. Those groups were formed after the fact by degree of re-entrainment rather than by assigned intervention, so the result links how far a clock moved to how people felt and performed inside a laboratory, and it does not test what a lighting program does to a workforce.

Slanger et al. (2016) sits on the other side of that line, pooling randomized trials in shift workers who also work nights and asking whether sleepiness on shift, sleep length and sleep quality can be improved, then concluding that this cannot be determined. Crowley et al. is not among the 17 included trials: it randomized nothing and its participants were volunteers on simulated shifts rather than shift workers, so it fails the review's design and population criteria. What remains unsettled is therefore what a lighting program does for people who actually work nights, on real shifts with assigned conditions, rather than whether alertness was ever measured alongside circadian phase.

The 67 people who completed the protocol, and who appear in both reports, were healthy volunteers aged 18 to 43, mean age 23.9 years (SD 6.2) with a median of 22, most of whom had never worked a night shift. Two of them sit outside the phase comparison, because the 42 earlier and 23 later baseline strata count only completers with a usable dim light melatonin onset at both assessments. Anyone with a medical, psychiatric or sleep disorder was excluded, as was anyone on prescription medication other than oral contraceptives, and anyone over 105 kg. These are not working nurses, officers or plant operators, and the finding should not be extended to older workers, to people with shift work disorder, or to anyone on medication (Crowley et al., 2003).

Neither report followed a safety, error or health outcome, and both describe simulated night shifts from 2300 to 0700 with a scheduled home sleep period from 0830 to 1530 rather than a live workplace (Crowley et al., 2004). The work also predates current lighting and eyewear products. The two papers were published in 2003 and 2004 and the review searched only to August 2015, so none of them speaks to tunable LED workplace lighting or consumer blue-blocking glasses. The only hardware the evidence covers is approximately 5000 lux light boxes and lenses at 15% and 2% transmission. Both the light boxes and the melatonin were donated by commercial suppliers of those products, and the study was funded by NIOSH grant R01 OH003954 (Crowley et al., 2003).