Red-light therapy

The primary photoacceptor is cytochrome c oxidase, complex IV of the electron transport chain, which has genuine absorption peaks in the red (~660 to 680 nm) and near-infrared (~810 to 830 nm). The leading hypothesis is that light photodissociates inhibitory nitric oxide from the enzyme's binuclear centre, briefly restoring electron flow, raising membrane potential and ATP output, and producing a small burst of reactive oxygen species that acts as a signal rather than as damage, activating NF-kB and Nrf2 dependent transcription. Light-gated TRP channels and changes in interfacial water have also been proposed. The cytochrome mechanism has real biochemical support behind it.

The optical window is why those wavelengths and not others: between roughly 650 and 1000 nm, haemoglobin absorption has fallen away and water absorption has not yet risen. But penetrating best is not penetrating far. Attenuation is exponential, and useful depths are millimetres to a couple of centimetres depending on tissue. Treating skin is a fundamentally different problem from delivering a dose to the middle of a hamstring.

The crux of the whole audit is the biphasic dose response. Too little light does nothing; too much is inhibitory. That means an error in either direction produces a null result, and a field whose parameter reporting is inconsistent cannot distinguish a failed mechanism from a badly aimed one.

So the questions to ask a device are specific: wavelength, irradiance in mW/cm2 measured at the treated surface, treatment time, resulting fluence in J/cm2, spot or panel area, distance, and whether the beam is pulsed. Consumer marketing typically supplies a single wattage figure that describes what the unit draws from the wall, which is not any of those things.

Meta-analyses do find effects. Vanin et al. 2018 reported improved muscular performance and reduced fatigue markers with pre-exercise photobiomodulation. But effect sizes vary enormously across trials, heterogeneity is high, and positive findings cluster in a small number of research groups, several with device-manufacturer relationships. That clustering is a pattern worth noticing anywhere it appears.

Zein et al. 2018 reviewed parameter reporting across the field and found it inconsistent enough to impede replication. When a literature cannot tell you which dose was delivered, it cannot tell you which dose worked, and it cannot be graded above underpowered no matter how many positive trials it contains.

There is a hard translation problem between research and retail. A laser delivering 20 J/cm2 to a 1 cm2 spot is not a panel radiating a similar total energy across an entire torso. Irradiance and area both matter, and most consumer devices are far below the surface irradiance used in the trials they cite.

The strongest clinical evidence for photobiomodulation is elsewhere entirely: oral mucositis in cancer care, where it carries genuine guideline support, and some wound-healing applications. That is meaningful, because it shows the mechanism does something real in tissue. Athletic recovery is a much weaker literature borrowing the credibility of a stronger one.