How should homeopathy be assessed? Still applying conventional science Lex Rutten, the Netherlands
Belief or science Belief: conventional medicine works; homeopathy does not.
Homeopathy cannot succeed in Randomised Controlled Trial
Scientific mirror Homeopathy - conventional medicine: same results in RCT
Hypothesis: selection bias, low quality, heterogeneity Shang, the Lancet aug. 2005
RCT works only for conventional medicine (????)
‘It works’ doesn’t mean it works • Allen Roses (GSK, December 2003): 90% of conventional medicines work in 30-50% of all patients • Pharmacogenetics: not only the disease, but the whole person: genotype • Homeopathy: same, but phenotype
Why not 100% success? RCT
side-effects
diagnosis
therapy-result co-morbidity
circumstances
Unreal certainty
From complaint towards result diagnostic research
RCT
symptom A symptom B
diagnosis
result
test X test Y
probability of diagnosis
chances placebo < 0.05
Homeopathic ‘diagnosis’ symptom A symptom B
Diagnostic/prognostic research Diagnosis
Result
symptom C symptom D Diagnosis
Probability of result
Effect modification comorbidity age > 65 sex social status suicidal?
Diagnostic/prognostic research Result Probability of result
Bayes method • • • • •
Knowledge from experience Direct results in practice Handling of complex clinical symptoms Probability instead of certainty Step-by-step increasing certainty by adding data
Bayes and homeopathy • • • •
More symptoms more certainty Peculiar symptoms are more important Vagueness is no problem Disadvantage: difficult calculations (use a calculator)
Likelihood Ratio (LR) LR+ =
Occurrence in target population Occurrence in rest-population
Odds = chance / (1-chance) Chance = odds / (1+odds) Bayes’ rule:
posterior odds = LR x prior odds
Prior to posterior chance (LR+=5) Prior chance • 1% • 10% • 30% • 50% • 80%
Posterior chance 4.8% 35.7% 68.1% 83.3% 95.2%
Homeopathic diagnosis stepwise Lachesis in menopausal complaints symptom LR success (%) 1. menopause
10%?
2. loquacity
5
35%
3. left-sided complaints
3?
62%
4. clothing <
3?
83%
Repertory with LR Partly hypothetical rubric ‘Fear of death’: FEAR - Death, of: in general, more than expected considering circumstances; occurring almost daily. Prevalence 5% (±0.5) Acon. (4), act-sp., agn. (2.5) , all-s, ..., Anac.(13) , apis (3) , arg-n. (3) , Ars. (6) , asaf., aur. (1) , ... Calc. (2) , lach.(3.5) , Lat-m. (6), led., ...
Practical homeopathic research • Relation between symptom and success • No conflict with daily practice • Takes a few seconds during each consultation • Outcome: a reliable repertory
Prospective research • Check the presence of 6 symptoms in each new patient • Keep record of medicines and results • a=occurrence of symptom in Lachesispopulation • b=occurrence of symptom in restpopulation • Likelihood Ratio = a / b
A few seconds of each consultation
Results fear of death medicine
fear death, n=
Anacardium Arsenicum album Calcarea carbonica
2 2 2
medicine population, n= 4 8 27
Fear of death in whole population: 3.8% Does this lead us to a more reliable repertory?
Repertory: Anac., Ars., Calc.
LR+ 13.36 6.66 1.95
Results after 15 months, n=1634
medicine Diarrhea anticip. arg-n.
Grinding teeth
LR+ 15
calc.
2.2
ph-ac
10
bell.
6.1
carc.
2.6
merc.
5
tub.
7.3
Sens. injustice
Loquacity
medicine
LR+
bell.
3.4
caust.
6
hyos.
7.8
lach.
4.7
Lachesis in menopausal complaints • LR+ loquacity - Lachesis = 5 (1,8-12,3) • succes by Lachesis in menopausal complaints with loquacity goes from 10%(?) to 35%. Dear GP, Treat your most loquacious patients with menopausal complaints with Lachesis
Conclusions • Homeopathy (and conventional medicine) is bayesian science • LR research is easy, cheap and rewarding • Effectiveness of homeopathy can be much improved by LR research
Vagueness Herpes lip
Loquacity loquacity
herpeslips 14,00%
30,00%
12,00%
25,00%
10,00%
20,00%
8,00%
herpes=1
6,00%
herpes=2
loquax=1
15,00%
loquax=2
10,00%
4,00%
5,00%
2,00%
0,00%
0,00% 1
2
3
4
5 doctor
6
7
8
9
1
2
3
4
5 doctor
6
7
8
9