TABLE Ill-Bleeding after two "identical" oral surgical interventions in a crossover trial in 24 patients treated with placebo and homoeopathy
drugs in this clinical model, which have all shown significant differences.23 ,_. The power of the present trial was, however, reduced by the lower pain scores of
No of patients reporting bleeding
Mean scores for bleeding*
these patients. As seen from the confidence intervals, a relative difference of about 30% to 40% would have been needed for an effect to become significant.
Generally, small sample sizes will affect power-that is, the probability of detecting small effects-and is a particular problem with negative findings. Power calculations indicate that detection of a 5% relative difference would have required a crossover study in about 1000 patients (that is, 11 000 visits by the
*Scores: 0-none, I =taste of blood, 2-some blood in mouth, 3-much blood in mouth.
tNon-significant increase in bleeding with homoeopathy; estimated strength 0•02 on the ordinal scale from 0 to 3
(95%confidence interval -0•26 to 0•29; P-0•86).
patients (fig 3). There was a non-significant preference for placebo: 0•2 mm (95% confidence intc:rval
-15•6 mm to 16•0 mm; P-0•98). Eleven of the 24
patients correctly indicated or guessed after which operation they had received homoeopathic treatment. The high confidence of the patients in homoeopathy remained essentially unchanged: mean score 79 mm (range 50 to 100) before and 82 mm (range 37 to 100) after the trial.
patients to the surgeon and the homoeopaths). The lack of any trend in figures 2 and 3 indicating a possible effect of homoeopathy suggests that even a large scale trial with many patients and accordingly greater power would probably not change the validity of our negative findings. Practically, attainment of clinical proof for the non-existence of a homoeopathic effect is impossible.
As the effects of treatment are claimed to last for long periods, crossover studies on homoeopathy should be examined for carry over effects.' " Our analyses found no significant carry over effect, but the estimate is rather insensitive, being based on comparison between
Maximum preference for placebo treatment
No
difference
Maximum preference for homoeopathic treatment
patients.
Homoeopathic treatment is selected on an individual basis. The few individualising symptoms in the patients (for example, low pain) made it difficult to choose the most appropriate homoeopathic medicine. Not surprisingly, 21 of the 24 patients ended up with arnica. Arnica is recognised by homoeopaths as the• remedy par excellence for wounds and injuries,•• not least to reduce the effects of dental surgery.',_" The present potency (D30) represents a medium potency (dilution 1: 10'0) often used in homoeopathy. Perhaps
a higher potency could have been more appropriate as
FIG 3-0verall preference of patients for postoperative course assessed after second operation. Abscissa: assessment by 100 mm visual analogue scale. (Maximum preference for homoeopathy= -.50 mm. No difference=O mm. Maximum preference for placebo=50 mm). Ordinate: each patient ranked according to preference
Period and cany over effects-Statistical analyses revealed significant period effects in pain and trismus before treatment but no significant carry over effect. A period effect is a general change in response from the first to the second period, whereas a carry over effect means that the effect of a treatment in the first period lasts and changes the response in the second period. Pain before treatment after the second operation was
40% lower compared with the first operation (95%
confidence interval 9% to 61%; P=0•02). During treatment, the assessment of pain showed a weaker period effect, which became insignificant when adjusted for values before treatment (P=0•32). The mean ability to open the mouth was virtually the same before the two operations (48•4 mm v 48•6 mm). After the operations, however, there was a significant period effect, with less trismus after the second operation. Difference on day 3 was 5•9 mm (2•5 mm to
9•2 mm; p..Q•001) and on day 7 was 3•3 mm (0•8 mm to 5•9 mm; P 0•01).
Discussion
The review of clinical trials in homoeopathy by Kleijnen et a/ refers to 20 studies on trauma and pain.' In 18 of these, the outcome was interpreted as positive by the authors. Our results, however, indicated no significant effect of homoeopathy on pain and other events after an acute soft tissue and bone injury. In fact, the postoperative course with placebo was remarkably similar to that with homoeopathy (figs 2 and 3). This is in contrast with previous placebo controlled trials on analgesic and anti-inflammatory
BMJ VOLUME 310 3JUNE 1995
several authors have recommended higher potencies of arnica for treating trauma and pain.'o-"'o-22 In two pilot trials on arnica as a remedy for injuries, Campbell found arnica ineffective in a dilution ofC30 (1:10•0) but apparently effective in a dilution of 10 M (1:10'0 000).•• This would agree with homoeopathic theory, accord ing to which a higher potency is expected to be more effective than a lower one.
Buckman, in a recent discussion in the BMJ on "What does homoeopathy do-and how?" pointed out that, besides evidence and mechanism, the important issue is whether patients feel better when they take homoeopathic remedies." The pain scores of our patients were remarkably low after both operations. Despite inability to identify correctly the postoperative course with homoeopathic treatment, their strong confidence in homoeopathy remained unchallenged. Conventional physicians and devoted homoeopaths cooperated without problems in this study. The physicians believe that the low pain scores and satisfaction of the patients may at least partly reflect the clinical skill of the two homoeopaths.
Key messages
• There is growing use of homoeopathy but equivocal evidence of efficacy
• This crossover trial studied the effects of homoeopathy on pain and other events of removal of impacted wisdom teeth
• Remarkably similar assessments of pain were observed with homoeopathy and placebo
• No positive evidence could be found for homoeopathic effects