About half the general population in developed countries uses complementary and alternative medicine (CAM).1 Yet many conventional healthcare professionals refuse to take CAM seriously — one often-voiced argument is “there is no research in CAM”.2 Certainly, for some modalities there is no compelling evidence base,3 and some of the research into CAM has methodological flaws and biases.4,5 On the other hand, many doctors and medical educators are uninformed about the quality evidence that does exist.6 In this article, I discuss some of the obstacles to developing an evidence base for CAM.

In most countries, CAM research funding is on a very small scale. For instance, only 0.08% of the British National Health Service research budget goes towards CAM research.7 Even though recent initiatives in the United Kingdom, United States and Australia have specifically freed up funds for CAM research, these amounts are minute compared with funding in other areas of medicine. It is likely that lack of plausibility of many CAM therapies deters scientific review committees from defining CAM as a priority. A vicious circle may ensue: little plausibility means no funds, therefore no preliminary research, therefore little plausibility.

Clinical trials of CAM can be even more expensive than those of conventional medicine. CAM treatments are often therapist-led, effect sizes are often small (requiring large sample sizes), and therapeutic effects may appear only after long treatment periods, all of which mean greater expense. For most CAM modalities, intellectual property cannot be protected; thus commercial investments are rarely forthcoming.

This shortage of CAM research funds has three important consequences:

• it prevents relevant projects from happening;

• it hinders the development of a research infrastructure similar to that of conventional medicine; and

• it keeps well-trained career scientists from entering into the field.

Many CAM therapies (eg, massage therapy) are physical by nature, which creates methodological challenges. What, for instance, is an acceptable “placebo” control for a trial of massage treatments? Like several other areas of conventional medicine (eg, physiotherapy, surgery, psychotherapy), blinding patients in clinical trials can be difficult or even impossible. Thus the highest level of scientific rigour can be barred to trials of CAM. Many CAM researchers also believe that their holistic approach can not be readily put into the “straitjacket” of a randomised controlled trial (RCT).8 This argument is demonstrably wrong, and its persistence in CAM circles continues to impede efficacy research. One can, of course, conduct an RCT comparing a complex, individualised, “holistic” treatment package to the standard care for that condition. This may require some innovative adaptations to the standard design, but, in principle, RCTs are usually feasible.9

For many people, CAM is an emotive subject. As a result, patients may not want to take a chance with randomisation, and many CAM practitioners may oppose scientific evaluation of their treatments, further hindering clinical trials.

There is more to most CAM interventions than meets the eye. For instance, some are based on theories that fly in the face of science. Researchers might conduct a clinical trial of traditional acupuncture, spiritual healing or homoeopathy and see this as a relatively straightforward exercise. Proponents of these therapies may, however, view it as a test of some ancient theories of life forces, spiritual energies or ultramolecular phenomena. Such discrepancies can (and usually do) create unforeseeable methodological problems, as well as obstacles for research and interpretation of results.

It is an important ethical requirement for randomised clinical trials that the investigators be in the state of equipoise (ie, they must believe that the test intervention is at least as good as the control intervention or placebo). If this is not the case (as for many CAM researchers), it is, strictly speaking, unethical for investigators to conduct the study. “Randomisation is only ethical if there is substantial uncertainty about the best treatment for that patient”.10

A further important ethical requirement for clinical research is informed consent from patients or healthy volunteers.11 As mentioned above, this may be difficult or impossible to obtain in an environment where patients’ enthusiasm often is strongly in favour of CAM and against receiving a control (placebo or non-CAM) treatment. Such problems constitute further impediments to good CAM research.

Although high quality CAM research does exist,3 many projects have, in the past, been less rigorous than they could have been. Before we condemn CAM for this situation, we should ask what the obstacles to CAM research are. Removing these obstacles will require dedicating adequate funds to CAM research, attracting career scientists into the field, adequately addressing the complexity of CAM and minimising bias with carefully designed studies.