The study measured personal light exposure via wrist-worn trackers for a week rather than using geographic light pollution data, which would better control for the urban air quality confounding you've identified.
No, measuring light doesn't remove the air quality confounder. Measuring air quality would allow to remove the air quality confounder. What are you smoking?
It turns out they took both rural and urban samples too and there was no meaningful difference between the two populations. Exposure to indoor overhead light is probably far more significant than what streams in your window with the drapes shut.
