Does red light work for pain? Why 412 reviews disagree

The short answer

There are 412 systematic reviews and meta-analyses on light for pain in humans. You would think that after 412 goes we would know. This is the pain chapter of our wider read of the red and infrared light literature.

We do not. For knee osteoarthritis, one of the best studied conditions in this field, two reviews reach opposite conclusions. Not slightly different: one says it works at the right dose, the other says the best available evidence does not support efficacy.

This article is about why that happens, and what you should do with it as a buyer.

Does red light work for knee pain?

Same condition, broadly the same treatment, and a flatly opposite headline conclusion. The difference lies in which trials were included and how doses were pooled or split. The practical upshot for you is that the answer depends on which review you happen to meet first.

That is not a one-off. An older, much-cited meta-analysis of 23 studies concluded, once restricted to properly double-blind trials, that light has "no effect on pain in musculoskeletal syndromes" ^[5]. For neck pain, a meta-analysis of 16 studies found an effect that persisted up to 22 weeks after treatment ^[6].

Does red light work for low back pain?

For one condition the strictest evidence body in medicine has ruled, and the ruling is not gentle.

A Cochrane review of 22 studies with 1,153 participants found very low quality evidence and concluded that there are "no data to support any clinical effect of LLLT in the treatment of carpal tunnel syndrome" ^[1]. That is Cochrane's own wording.

A separate network meta-analysis landed in the same place: there was a statistically significant pain reduction versus sham, but the size was not clinically significant, on which the authors said laser therapy is not recommended as an addition to standard splinting ^[2].

What do the abstracts leave out about the dose?

The newest review in this field is on patellofemoral pain, and it shows exactly why you have to read the full text rather than the abstract.

That last point is not a detail. It means the measured benefit could just as well come from the exercise, and we do not know, because nobody has studied it separately. That sentence is not in the abstract.

Why do 412 reviews still disagree?

Somebody studied the quality of the research, and it may be the most useful finding of the lot.

An umbrella review appraised 20 systematic reviews on high-intensity laser therapy for musculoskeletal pain using AMSTAR-2, the standard instrument for it. The average score was 12.9, and several reviews were rated low or critically low ^[7].

In other words, in this field the label "systematic review and meta-analysis" guarantees nothing. That applies to the reviews we cite here too, which is why we write down per source how certain the authors themselves are.

Does red light help recovery after training?

The picture is not uniformly bleak, and that belongs here just as much.

For athletic recovery, a meta-analysis of 24 studies found that light applied before exercise improved leg muscle performance at 24, 48 and 96 hours, alongside less soreness and lower creatine kinase and IL-6, markers of muscle damage ^[10].

For orofacial pain, a GRADE-appraised meta-analysis found genuine benefit over placebo, with certainty rated low to moderate ^[9].

And when high-intensity and low-level laser were compared directly in knee osteoarthritis, both improved pain and function over placebo, with the high-intensity version significantly better at eight weeks ^[11]. So intensity matters, not only wavelength.

What we think

If you buy a lamp for pain, buy it with the right expectation.

The evidence is strongest for pain, weakest for function, and for carpal tunnel syndrome the best evidence says it does not help. Where an effect is found it is usually modest, and in the newest review it came out of trials where people were also exercising.

That is not a reason not to try it. It is a reason not to expect a lamp to replace the exercise. In the research it never did, because it has never been tested apart from it.