Does using red light for reducing wrinkles work?

I was watching a Mrwhosetheboss Youtube video in which he showed off some Korean tech, including a red-light scalp massager. While watching I was getting a bit annoyed, as he pretty much just parroted the sales pitch of the red-light craze. It goes a little like this: shining certain wavelengths of light at your skin will stimulate cells in the dermis to make more collagen, which reduces wrinkles and promotes "anti-aging". I'm generally skeptical of these kinds of mechanisms and sales pitches, so let's dive into some of the evidence and counterpoints in this article.

Dermatology, to be fair, has used light-based treatments for a long time. Phototherapy is older the current wave of home masks and influencer routines by decades. What we're seeing now is different, though: devices moved from clinics into homes, got cheaper, got more wearable, and turned into something you can sell as skincare tech in a TikTok.

In this article I want to take a skeptical view at some of the evidence. Let's take a look at a classical paper supporting red-light treatment for wrinkles, think about if it makes sense, and then look at a review that describes the broader literature.

The start of the red-light craze

One of the early studies that started the current hype around the red-light anti-aging trend is Lee et al. (2007), published in the Journal of Photochemistry and Photobiology B. This study at first sounds relatively high-quality - it is a randomized, double-blinded, split-faced trial with instrument measurements and photos.

Participants were assigned to one of four groups: 1) 633 nm light, 2) 830 nm near-infrared light, 3) a combination of both, or 4) sham light (stand-by mode of the light). They did this twice weekly for four weeks, with follow-up about three months after the treatment period. The measurements they did were photography, profilometry (surface texture measurements), elasticity of the skin, and melanin levels. In short, the paper says they found an improved wrinkle metric over time compared to baseline and increased elasticity.

What are my problems with these kinds of studies? I will use Lee as an example, but these things are generally true:

• Samples sizes (Lee uses n = 112, which is only 28 patients per group, with a 36 person dropout overvall)
• Good blinding (I doubt anyone would not realize they are not actually shining light on your face, or absence of warmth and sound. Nowadays, they also use lower light strengths as better shams)
• No pre-registration: they measure a lot of things in one study, a few of which turn out positive. This is akin to p-hacking
• Funding conflicts of interest (Lee was paid by Photo Therapeutics Ltd.)
• Many "early-phase" studies with small samples sizes, but never a large high quality trial to break out of the early stages. This also suffers from publication bias, where negative results are less likely to be published. Case-in-point: there have been only 6 high quality RCTs between 2005 and 2022!

This often trickles down into reviews or meta-analyses, where the only available studies are the ones that have these shortcomings. Many weaker papers don't make a strong point, though.

While researching for this article, I ran into a short article by Grimes in PLOS One where some of the methodological problems in LED therapy are highlighted. In the paper, he talks about how most studies report different wavelengths, irradiances, treatment schedules, and measured outcomes. This amount of heterogeneity makes comparison and synthesis of results hard. He goes on to compile 27 studies mentioned in systematic reviews, and checks them for among other wrinkle reduction. Using these studies, he calculates a patient received dose, and converts this to an equivalent solar exposure time as a unifying metric.

It turns out the studies have three-orders of magnitude difference in energy supplied per area, ranging from 0.1 to 126 J/cm². The wavelengths range from 405 to 660 nm (and since LEDs are not monochromatic, we're already simplifying the spectral distribution. None of the studies actually report independent validation of dose/irradiance given by the devices tested. Outside of physical differences, 37% of the papers were sponsored by the device manufacturers and 11% were even conducted by a commercial practice. Many of the studies were unblinded, too.

Biological reasoning for red-light skincare mechanisms is relatively thin and poorly understood (see here), and has barely any link to why certain parameters should be used. Papers can be poorly compared to each other due to different settings. Many studies have conflicts of interest. I'm not ready to say there's nothing here that works, but I don't think the technology is ready for the limelight yet. First, we should have at least a few large (let's say n = 250 per group for the effect sizes we're seeing), pre-registered RCTs with standardized physical settings, without industry being the main sponsor. Yet somehow we already have full-on marketing campaigns selling these kinds of products accompanied with very strong claims.