45

I was surprised by the following claim from NaturalNews:

Most clinical research conducted on homeopathic medicines that has been published in peer-review journals have shown positive clinical results

This goes against what I heard "It's always been found equivalent to placebo", so what's going on here?

Twinkles
  • 1,591
  • 15
  • 24
  • 26
    Clinical trials show that placebos show slightly positive results. Therefore is hoeopathic medicines were equal to placebos then they would also show slightly positive results. A neutral result would have shown hoeopathic medicines to be worse than placebos. – Rincewind42 Nov 05 '11 at 01:05
  • 5
    "positive clinical results" doesn't necessarily mean "statistically significant results." – Flimzy Nov 05 '11 at 02:17
  • 4
    This document is relevant (_includes 17 references_) -- **Homeopathy: The Ultimate Fake**: http://www.quackwatch.org/01QuackeryRelatedTopics/homeo.html – Randolf Richardson Nov 05 '11 at 02:21
  • 6
    Re: `Whats going on here` - NaturalNews.com is **lying**. – Fake Name Nov 06 '11 at 09:38
  • I would have guessed that [publication bias](http://en.wikipedia.org/wiki/Publication_bias) could explain naturalnews.com's claim. – Andrew Grimm Mar 30 '12 at 23:37
  • 2
    Related question: http://skeptics.stackexchange.com/questions/1915/has-there-ever-been-a-double-blind-study-validating-homeopathy – Oddthinking Aug 21 '12 at 01:02

2 Answers2

61

Short answer, No.

Explanation of why you hear different best summed up by Ben Goldacre here:

So how come you keep hearing homeopaths saying that there are trials where homeopathy does do better than placebo? This is where it gets properly interesting. This is where we start to see homeopaths, and indeed all alternative therapists more than ever, playing the same sophisticated tricks that big pharma still sometimes uses to pull the wool over the eyes of doctors.

Yes, there are some individual trials where homeopathy does better, first because there are a lot of trials that are simply not “fair tests”. For example – and I’m giving you the most basic examples here – there are many trials in alternative therapy journals where the patients were not “blinded”: that is, the patients knew whether they were getting the real treatment or the placebo. These are much more likely to be positive in favour of your therapy, for obvious reasons. There is no point in doing a trial if it is not a fair test: it ceases to be a trial, and simply becomes a marketing ritual.

There are also trials where it seems patients were not randomly allocated to the “homeopathy” or “sugar pill” groups: these are even sneakier. You should randomise patients by sealed envelopes with random numbers in them, opened only after the patient is fully registered into the trial. Let’s say that you are “randomly allocating” patients by, um, well, the first patient gets homeopathy, then the next patient gets the sugar pills, and so on. If you do that, then you already know, as the person seeing the patient, which treatment they are going to get, before you decide whether or not they are suitable to be recruited into your trial. So a homeopath sitting in a clinic would be able – let’s say unconsciously – to put more sick patients into the sugar pill group, and healthier patients into the homeopathy group, thus massaging the results. This, again, is not a fair test.

Actual summaries to the proper reviews in the medical literature referenced here (original lancet article; full text on ben Goldacre's blog) with this conclusion:

Five large meta-analyses of homoeopathy trials have been done. All have had the same result: after excluding methodologically inadequate trials and accounting for publication bias, homoeopathy produced no statistically significant benefit over placebo.

Update

Eduard Ernst summarises a recent (april 2014), and fairly comprehensive, Australian review (pdf) of the subject thus:

Not for a single health conditions was there reliable evidence that homeopathy was effective. No rigorous studies reported either that homeopathy caused greater health improvements than a placebo, or that homeopathy caused health improvements equal to those of another treatment.

Community
  • 1
matt_black
  • 56,186
  • 16
  • 175
  • 373
  • Wouldn't "accounting for publication bias" completely eliminate your "No" answer since you are discounting methodologically adequate articles? It's not clear how the "accounting" is done. If 100 studies were done with negative result and 3 with positive result, and only 2 were published, that shows publication bias. It does NOT however allow one to claim that "NO" studies showed results since 2+ studies (presumably with proper methodology, since otherwise publication bias wouldn't need to be invoked) did show a result. – user5341 Nov 04 '11 at 19:21
  • 2
    @DVK I think the difference is between most studies *published* and most studies that get *mentioned* by pro-homeopathy groups. Hence the meta-anlyses that show no effect yet the claim that most published results are positive. In reality pro-homeopathy groups mean *most studies we will ever tell you about* are positive. – matt_black Nov 04 '11 at 20:06
  • Thanks very much for your reply. You cited a big analysis of lots of studies that came to the desired conclusion.... but so did Dana: http://www.ncbi.nlm.nih.gov/pubmed/9310601 ("The results of our meta-analysis are not compatible with the hypothesis that the clinical effects of homeopathy are completely due to placebo"). To be intellectually honest (rather than just picking the study that proves the conclusion I wanted to prove) I think I need something more to break the symmetry. –  Nov 04 '11 at 20:18
  • 3
    @QED The authors of the meta-review you mention are not claiming that homeopathy works: "Our study has no major implications for clinical practice because we found little evidence of effectiveness of any single homoeopathic approach on any single clinical condition". Their main conclusion is that we need a more systematic approach to settle the issue, they write that the quality of most homeopathy studies is mostly poor. – Mad Scientist Nov 04 '11 at 20:36
  • 4
    @Fabian - I wonder if any of those studies ever dealt with homeopathy the way it's **supposed** to be practiced (assuming it works). Anything I saw so far was either attempting to pin down whether specific mechanisms work or not (since the proposed meachanisms are mostly rooted in 18th century, DUH!), OR on mechanial "does the remedy X relieve symptom Y". The problem is that the latter is decidedly NOT how homeopathy is supposed to be prescribed (you don't treat **symptoms**), so such a study doesn't really address anything useful. – user5341 Nov 05 '11 at 00:24
  • 1
    @Fabian - ... The problem is that homeopathy - when done "correctly" - is a lot fuzzier than a typical study can account for since there are too many variables. I wonder if there was any study done on a less-variable stuff (e.g. bruises - where I have empirically observed a noticeable effect from Arnica) - there's a Skeptics question begging to be asked but I don't have the drive to research it to make into a good question. – user5341 Nov 05 '11 at 00:27
  • 9
    @DVK: *"You don't treat **symptoms**"*. ???! The whole point of homeopathy is the Law of Similars -- homeopathy is *entirely* predicated on treating symptoms with symptomatic protagonists, and ignores root causes. It's the naturopaths who claim to ignore symptoms. – Stan Rogers Nov 05 '11 at 04:00
  • @QED One explanation for some of the experimental observations that suggest homeopathy has an effect is that placebos have *significant* effects and those effects depend on the nature of the placebo. See http://www.bmj.com/content/336/7651/999.full.pdf and http://www.bmj.com/content/332/7538/391.full.pdf?sid=c46335de-02e1-44b5-9038-0cbe880d74ab and the discussion on this question: http://skeptics.stackexchange.com/questions/6673/do-placebos-have-an-effect-on-ulcers – matt_black Nov 05 '11 at 13:41
  • Indeed, it's a frequent claim of homeopathy that "we treat the root cause and not just the symptoms", which would seem to imply that if you hit your head and walk into Emergency complaining of a headache and fluid leaking from your ears, they will give you (a) a tylenol and (b) a ball of cotton, and send you on your way, rather than look for the root cause. Whereas, homeopathic practice is essentially based on symptoms without regard for actual cause. "Part of you is red and swollen? Bee venom for you". The claim, I think, was more valid 200 years ago, and never reevaluated. – greggo Aug 04 '14 at 01:10
  • And of course it is a lot fuzzier than that, there are many ifs and buts and special cases. The effect of all these is to fragment the patient set into a large number of different cases, each of small sample size, to the point where no individual homeopath is likely to see a pattern which clearly indicates that the remedy is not actually what's causing the healing. I don't claim that this was set up deliberately to deceive, but rather evolved into the practice. When you can't see what's wrong, and you think it actually works, you're going to keep doing what you're doing. – greggo Aug 04 '14 at 01:20
6

I think the claim is technically true but misleading.

In 1997 Klaus Linde et al did a meta analysis to answer that question titled "Are the clinical effects of homoeopathy placebo effects? A meta-analysis of placebo-controlled trials".

In their literature review they find:

The combined odds ratio for the 89 studies entered into the main meta-analysis was 2·45 (95% CI 2·05, 2·93) in favour of homoeopathy. The odds ratio for the 26 good-quality studies was 1·66 (1·33, 2·08), and that corrected for publication bias was 1·78 (1·03, 3·10). Four studies on the effects of a single remedy on seasonal allergies had a pooled odds ratio for ocular symptoms at 4 weeks of 2·03 (1·51, 2·74). Five studies on postoperative ileus had a pooled mean effect-size-difference of −0·22 standard deviations (95% Cl −0·36, −0·09) for flatus, and −0·18 SDs (−0·33, −0·03) for stool (both p<0·05).

The results of our meta-analysis are not compatible with the hypothesis that the clinical effects of homoeopathy are completely due to placebo.

What's the problem? They looked at all homeopathy studies. Most studies that are conductive are not high quality studies. Low quality studies frequently find homeopathy to be effective but the effect gets smaller as the quality of the study increases.

Because of this subsequent reviews of homeopathy focus generally on high quality studies instead of trying to analyse all studies. Focusing on high quality studies seems to be the only way to get the answer: "Homeopathy doesn't work."

The lesson of the affair that quite easy to get any result you want if you settle for low quality studies and run enough studies.

Community
  • 1
Christian
  • 33,271
  • 15
  • 112
  • 266
  • 2
    I would have written a quite similar answer, but it might help to state explicitly and unequivocally: the Linde study, while well-intentioned, was methodologically **flawed**, hence the misleading conclusion. – Konrad Rudolph Apr 24 '14 at 22:18
  • 2
    @KonradRudolph : I don't believe in the usefulness of unequivocal statements. They don't bring people to think for themselves and investigate evidence for themselves. – Christian Apr 24 '14 at 23:05