Not all light exposure measures are equal: a framework for exposure assessment and causal inference in light epidemiology

Abstract Background Light exposure is increasingly recognised as a modifiable environmental determinant of human health. Large-scale studies using satellite-derived outdoor light, wrist-worn sensors, bedroom illuminance meters and other scalable exposure measures have reported associations between daily light-dark patterns and psychiatric, metabolic, cardiovascular, sleep-related and mortality outcomes. These findings have increased the visibility of light exposure within population health research and have provided evidence at scales that cannot be achieved in laboratory or small field studies. At the same time, they raise a foundational exposure-science question: what light exposure construct is being measured and how closely does that construct correspond to the biologically relevant stimulus at the eye and retina? Main body In this Debate , we argue that the next phase of light epidemiology requires a construct-centred framework for exposure assessment. Satellite-derived night-time radiance, wrist-worn illuminance, bedroom light, corneal-plane light, retinal exposure after optical and pupillary modulation and modelled circadian stimulus are related but non-equivalent constructs. Each may be appropriate for particular scientific questions, but each supports different forms of inference. The central challenge is therefore not whether scalable proxies should be used, but how their meaning, uncertainty and relationship to ocular exposure can be specified and empirically quantified. We propose a forward-looking framework for population light exposure research based on explicit definition of exposure constructs, validation across measurement sites and spatial scales and pre-specified causal models linking behavioural determinants, 24-hour light exposure, biological pathways, intermediate phenotypes and clinical outcomes. Conclusions This framework is intended to support the integration of high-throughput epidemiology with physiologically interpretable measurement, enabling the field to move from broad associations towards mechanistically informed and translationally useful evidence.