I wondered about the number of pills given for each treatment as I am aware of a couple of papers that suggest the number of pills is related to the strength of the placebo effect. I find this paper interesting:

placebo injections elicit greater placebo responses than placebo pills, and larger pills are associated with stronger placebo responses than smaller pills; the number of pills taken is also directly related to the magnitude of the placebo response

I also blogged about the increased placebo effect from an increasing number of pills in relation to this diclofenac/arnica study.

Diclofenac is known to reduce the pain, at the cost of increasing the inflammation and healing time, of a wound. In this comparison of patients treated with diclofenac and arnica 4D, those treated with arnica 4D had wounds which were more painful but showed less irritation and healed faster than those of patients treated with diclofenac. This is what we would expect if arnica 4D had no effect what so ever. We can’t compare this with a true null because there was no placebo group.

And an exoteric explanation of my their conclusion isn’t valid even if their experiment was, which it isn’t:

A high p-value indicates insufficient information to reject the null hypothesis not evidence that the null hypothesis is correct. If you know the power (a measure of the sensitivity of the study to differences between groups of a certain magnatude) you can say that the odds of having detected a difference of a particular size with the data collected were x and hence the failure to detect a difference means that any difference that might exist is likely to be less than y. If they haven’t done so then they can’t justify any claim of near equivalence between the two treatments.

On the ethics of not having a placebo group:

There seems no reasonable justification given that the actual evidence that arnica is only as effective as placebo there is no reason to think members of the placebo group would suffer greater harm than the arnica group. Indeed if it was unethical to include a placebo group it was unethical to do the study at all.

One interesting thought I had. The study shows the arnica group suffered more pain but were more mobile than the diclofenac group. This raises the question; Is mobility always good? Perhaps the arnica group patients were in too much pain to sleep so went wandering round the ward looking for distraction. I might be wrong, but it is always worth thinking a bit deeper about how the results of a study fit together.

Diclofenac is sometimes prescribed as a postoperative pain killer – and a useful one it is too. That’s all it does – relieve pain.

It does not improve wound healing, and in fact published studies show it actually delays wound healing. So the fact that arnica might be better at wound healing than diclofenac should come as a surprise to no-one. Taking absolutely nothing would also be better at wound healing than diclofenac. But then with arnica, that’s in effect what they were doing anyway.