Question:
in asthma patients. In fact, the study's authors noted that: "Anindependent criteria-based review of over 100 published controlled trialsof homeopathic treatments noted that 77% show a positive effect."The conclusion ot that article is that there are no scientificproof that homeopathy works. The sentences your have selectedare out of context.

Answer: I have read that article over an over again. Meant is here thestudy of Kleijnen et al. published in BMJ (vol 302, 1991, p. 316-323).

Kleijen et al. examined 96 papers with 107 investigations of homeopathy(clinical trials).

They assigned grades for quality to these papers. They were verymild. There were 16 investigations with a "passing grade".This shows the general quality of research in this field.One of the papers examined was an abstract of that asthma study.It received 35 points on a scale of 100 (>59 is passing).

Among these "passing grades" there was one paper in which the authorapparently was unaware of the meaning of randomisation. One paperdid not have a placebo group, only compared 4 different homeopathicpreparations.

Kleijnen et al. were also extremely mild in judging a paper asyielding a "positive result". For example a paper by Maiwald et al.gives no significant result: 18 out of 62 patients in the verum groupwere rated as "succesful", and 12 out of 53 in the placebo group.In this case the placebo was aspirin, so the patients were not blind,but the treating doctors were blind according to the paper.For a secondary outcome measure the results were similar.This paper was rated "positive" bij Kleijnen et al.

Eight investigations with together 1080 patients yielded a positiveresult, and six with together 2496 patients yielded "no diference"between homeopathy and placebo.

Among these eight there is the famous headache investigation ofBrigo which has not been duplicated in two different trials,there is Reilly's hayfever investigation which has a dubiousanalysis (if the patient's subjective scores are simplified tobetter, same, worse, then no significant difference remains; moreoverit had nothing to do with homeopathy).Very much weight carries the investigation of Ferley et al. (478patients) of the effect of potentized rotted duck liver on flu.This showed no overall effect. Only by examining the 63patients that got better within 48 hours, a "significant effect"was found (p=4.5% in my computation). I.e. if there had beenone single placebo patient more that was reported to be cured in48 hours, then even this partial result would have been non-significant.The 48 hour criterium is the only one that makes this happen.

One of the coauthors of Kleijnen et al. (his Ph.D. advisor) haspublicly stated his view that publication bias is the mainexplication for this seemingly positive effect.

If you restrict your attention to trials that are(1) executed in cooperation with skeptics(2) properly randomised and double blind(3) at least 100 patients randomised (and no excessive fall out)