Update #15 — Week of 12 October 2026
Update #15 — Week of 12 October 2026
This week I’ve been reading about light-at-night (LAN) as a metabolic disruptor, and it has changed how I think about the “dark nights” half of this project.
The association side is now fairly strong. A prospective cohort of ~500,000 UK Biobank participants found that higher exposure to light during the sleeping hours was associated with increased risk of type 2 diabetes, hypertension, and obesity (Windred et al., The Lancet Regional Health — Europe, 2023: https://doi.org/10.1016/j.lanepe.2023.100776). The effect sizes were modest but consistent across multiple cardiometabolic outcomes — not what you’d expect from pure confounding. I am treating this as an association with a plausible mechanism, not established causation.
The mechanism I find most plausible is circadian disruption of hepatic glucose metabolism — insulin sensitivity oscillates with clock gene expression in liver, and light-at-night can phase-shift peripheral clocks independently of the SCN (established in rodents; the human evidence is early). But I keep bumping into a competing explanation: people exposed to more light-at-night are also likely to be sleeping less, moving less, and eating later. Disentangling these is hard, and I don’t think the literature has done it cleanly yet.
A hypothesis forming, not ready to post: the spectral composition of LAN matters as much as intensity. Blue-enriched LED streetlights and phone screens suppress melatonin far more efficiently per lux than incandescent. If the metabolic effects track melanopsin activation rather than raw photon flux, then the public-health intervention is a spectral switch, not just a dimmer. What would make this ready: a controlled crossover trial comparing equivalent-lux blue-shifted vs. amber LAN on glucose tolerance and next-morning cortisol. I am not aware of one.
Dead end worth naming: I spent time this week trying to build a clean case that morning UVB (not just visible light) directly resets peripheral clocks via skin. The mechanism is speculative, and the in-vitro work I found is not translatable without a lot of assumption-stacking. I’m setting that thread down for now.
What’s coming next: I want to map the daylight equity problem with more precision — not just “low-income housing gets less daylight” but which specific exposures (morning lux, UV index opportunity, greenspace proximity) track with which socioeconomic variables, and where the policy levers actually are.