Differences between homoeopathic immunotherapy trials by Lewith et al and Reilly et al

	Lewith et al¹	Reilly et al²	Possible implications
Research context and stage of inquiry	1st pilot of new design for HIT in asthma	One of a series of 4 trials using HIT and allergy as a model	Lewith et al inquiry needs a second phase
Aims	Evaluate efficacy of HIT in asthma	Evaluate placebo hypothesis of homoeopathy	Different questions
Clinical context	Primary care patients (38 practices)	Specialist care patients (one centre for investigation of respiratory disease)	Different patient group, different illness severity, different context effects
Design bias	Effectiveness�long clinical observation in the real world of primary care application	Efficacy�highly focused, intense, short observation to ensure internal validity	Lewith et al showed effectiveness but could not show efficacy; Reilly et al is the opposite
Conventional diagnosis of asthma	Clinic criteria, symptoms, and simple respiratory tests	As for Lewith et al�s trial plus laboratory confirmation with histamine provocation testing	Lewith et al�s trial OK for an effectiveness study but Reilly et al�s trial had more accurate selection for an efficacy study
Basis of the homoeopathic diagnosis for the HIT prescription	Conventional practitioners� decisions	Homoeopathic expert with 15 years� experience of using HIT. Case conference for every case; respiratory and homoeopathic physicians able to bar entry to study	Reilly et al�s trial had more accurate prescribing; expert decided in cases of ambiguity�for example, when skin reaction weal for house dust mite is equal to a competing allergen
Treatment regimen and observation period	Three doses over 24 hours; no repetition over next 16 weeks	Three doses over 24 hours, observed for 4 weeks. In optional 4 week extension period patients given repeat doses at 4 weeks (that is, usual clinical practice)	Likely inadequate treatment regimen in Lewith et al�s trial. Data at 4 weeks after randomisation should be explored
Baseline data	Off placebo	On placebo	Radically different context effects and baseline
Data collection	Discontinuous data	Continuous data	Unknown impact; patient liable to "reset" to recent baseline after a break ("poor memory of pain")
Seasonal effects in exposure to house dust mite	Recruitment timing begins 5 months before Reilly et al�s trial and lasts for 30-32 weeks. Wide geographical spread?	After February and restricted to 4 weeks. Patients from restricted geographic zone	Reilly et al�s trial has more even exposure; avoids winter peak of house dust mites as HIT will be less effective when seasonal challenge is maximal. May have avoided aggravations
Outcome measures	"Wider"; discontinuous; visual analogue scale labelled about asthma, different from that in Reilly et al�s study	"Focused; continuous; visual analogue scale worded differently from Lewith et al�s study ("Overall today I felt�")	Different, so comparison problematic. All 4 HIT trials by Reilly and Taylor et al kept the wording steady
Blinding	A/B labelling, with research nurse and patient recording this	Double coded individualised numbering	Possibility in Lewith et al�s trial of researchers developing and transmitting conscious and unconscious attitudes
Study power and risk of false negative result (type II error)	Pilot study with best guess power calculation. Underpowered. High placebo response so high risk of a false negative result. No post hoc power calculation	Based on previous trial in the series. Underpowered yet shows a positive result, so no risk of a false negative result	Lewith et al�s trial may have false negative result
Outcome	Good result in both groups. Shows effectiveness, not efficacy, against large non-specific placebo clinical gain. Groups showed different response patterns�possible aggravations mixed with improvements	Good result only with active treatment. Apparently poorer placebo response	Placebo enhanced in Lewith et al�s trial. Placebo run-in qualification period in Reilly et al�s trial screened out and corrected for initial placebo effects, increasing power of the study