Complementary medicine: what we actually know

“Does complementary medicine work?” is not a question with an answer, because complementary medicine is not one thing. The useful question is narrower, and the research can usually answer it.

By Tristan Carter · August 2026

Traditional, complementary and integrative medicine — TCIM — covers acupuncture, herbal medicine, vitamin and mineral supplements, massage, chiropractic, naturopathy, yoga, meditation and homeopathy, among others. Asking whether all of that works is like asking whether drugs work. Some of it has good trial evidence for particular problems, some of it has evidence that it does nothing, and a great deal of it has never been tested properly at all.

Prevalence estimates vary widely depending on how “use” is defined and which therapies are counted, but every national survey finds the same broad pattern: a large share of the population uses something, and they use it alongside conventional care rather than instead of it.^1,2 That pattern is what makes the honest questions practical rather than ideological.

Ask about one therapy, for one problem

Evidence in this field is specific. A therapy can have reasonable support for one condition and none for anything else, and the marketing rarely respects the distinction. Two examples of where the evidence is comparatively strong:

Neither entry says “acupuncture works” or “St John’s wort works”. They say something much more limited, which is what a good evidence summary always does. Apply the same discipline to a claim on a label and most of them stop being claims at all.

Natural does not mean inert

A substance strong enough to help is strong enough to interact. St John’s wort is the textbook case: it induces the liver enzymes that clear a long list of prescription medicines, which can reduce their effect — including some contraceptives, anticoagulants and cancer treatments. It is not an exotic risk, and it is well documented.⁵

The broader point is that harm in this field is usually indirect rather than toxic. The recurring patterns are a delayed presentation, a conventional treatment quietly stopped, or an interaction nobody assessed because nobody knew about the exposure.⁶

In Australia, “listed” is not “approved”

Most complementary medicines sold here are listed rather than registered, and carry an AUST L number instead of an AUST R one. Listed medicines may only contain ingredients the regulator has already assessed as low risk, and the sponsor certifies that they hold evidence for the claims they make — but the regulator does not evaluate that evidence before the product goes on sale. Registered medicines are assessed for efficacy up front.

So an AUST L number on a box is a statement about ingredient safety and manufacturing quality. It is not a statement that the product does what the front of the box says. This is one of the most consequential things a consumer can know about the shelf they are standing in front of, and it is almost never explained on the shelf.

The risk nobody records

Pooled evidence indicates that a large proportion of people using complementary medicine do not tell their treating clinician, and non-disclosure is patterned rather than random — people who expect a dismissive response are less likely to mention it.⁷ That silence is the part of the problem the health system can actually fix. A clinician who does not know what someone is taking cannot check for interactions, and a patient who expects to be judged will not volunteer.

What good evidence looks like here

Three things are worth checking before believing a positive result in this literature:

• What was the comparison? Against nothing at all, almost anything looks effective, because attention and expectation do real work. A sham or active comparator is what separates the therapy from the ritual around it.
• How big was the trial, and who ran it? Small single-centre trials, often from a research group with a stake in the result, dominate parts of this field. They are a reason to keep looking, not a reason to conclude.
• Is the review systematic? A list of supportive studies is not evidence; it is a selection. A systematic review looks for the unsupportive ones too.

There is a structural problem underneath all of this. Research funding for TCIM is modest and unevenly distributed relative to how widely these therapies are used, and concentrated in a small number of countries and modalities.⁸ That produces a circularity worth naming: these therapies are dismissed for lack of good evidence, and good evidence is hard to produce without funding proportionate to use.