Part one of this series analysed the dirty local politics of commercialised combination / indication over-the-counter so-called homoeopathic remedies; identified how these seriously conflicted with traditional homoeopathic philosophy and evidence-based complementary medicine; and summarised the results of recent scientific appraisal of existing clinical trials which set out to prove that, or to determine whether homoeopathic remedies are efficacious. Complex homoeopathic OTC's are marketed as medicines for serious indications without verification. This analysis deals only with such products and in no way denigrates the classicist/wholist.

After evaluating all scientific reviews to date, as principle author, even though the remedy 'appeared' in many cases to perform better than placebo, I concluded that in the final analysis, placebo actually works better than the remedy, based logically on the indisputable (measurable and reproduceable) scientific existence of placebo (1),(2),(3),(4),(5),(6)&(7), plus conservative elimination of considerable confounding trial factors comprising serious methodological flaws, reducing any supposed evidence in favour of the remedy to mere artifacts. Placebo beat the remedy in the sense that David as underdog beats Goliath, even if battle ends in a truce.

Personal opinion aside, I shall substantiate my taking care to choose only publications and authors known to be objective in the evaluation of complementary medicine. These searches included all published reports of controlled clinical trials, including journals, books and conference proceedings, as well as homoeopathic researchers, institutions and manufacturers, spanning 13 countries and covering all homoeopathic types and potencies (8) & (9). The 'quality' of the scientific 'evidence' for 'efficacy' of homeopathic 'remedies' follows:

Investigation started with Scofield (cited by Kleijnen), whose comprehensive 1984 review article concluded that, "because of bad design, execution, reporting and failure to repeat experimental work, there is limited evidence to 'suggest' that homoeopathy is effective" (8). In the Hill citation, I could have added that "out of 40 randomised trials, all but three had major design flaws and that only one had reported a positive result" (10). An item of local interest is a trial by Prof. Hofmeyer (Univ. Witwatersrand) which obtained the virtually unique quality score of 100 % in Linde et als' 1997 meta-analysis. The results, definitively negative over placebo (9).

Prof. Jos Kleijnen, first author of the 1991 BMJ review (8) (CRD, NHS, York Univ. UK) has published positively on eg. vitamin E, garlic, placebo, the colour of drugs, and Ginkgo (11) (incl. BMJ, BJ Clin. Pharm., & Lancet) and wrote a textbook on the effectiveness of alternative medicine (12). His 1991 BMJ comments:

"The results of all studies may be seriously biased because of several methodological shortcomings. In 42 of 107 trials, there was insufficient data to check the author's often overoptimistic interpretation of the outcome(s). Overall, the quality was disappointing. Sometimes only some of several interventions, measurements of outcome, or data presentations met the criteria. Only 23 scored greater, and 84 less than 55 for the maximum of 100 for quality. With limited participants (often not mentioned) (less than half had over 25 patients per group), one cannot be confident that randomisation will equally divide the known and unknown confounders" (8).

"Publication bias is an important problem. Only 17 described the method of randomisation. Whilst 75 were double blind trials, placebo was 'described' as indistinguishable in only 31. Patients have many ways to break the code, which might explain any differences in favour of homoeopathy. Double blinding was not checked in any trial of homoeopathy. The process of producing preparations and their composition, especially herbs, differs greatly among manufacturers and hence preparations may still have pharmacological effects since it is sometimes difficult to demarcate phytotherapy (Prob.>1C/2D-2C/4D)(ST) from modern homoeopathy" (8).

 "A trial of very high quality by the Groupe de Recherches et d' Essais Cliniques en Home'opathie initiated by the French Ministry to retest (apparently positive) earlier results in a new rigorous trial, found no positive evidence for homoeopathy (13). Will more such trials refute the existing 'evidence' ?", asked Prof. Kleijnen (8). Dr. Jean-Pierre Boissel (Dir. French Cochrane Centre), principle author of the 1996 Homoeopathic Medicine Research Group report to the European Communities Commission, after examining 184 reports of controlled trials, considered only 17 to be worth considering and concluded that the number of participants was too small to draw any conclusions about the effectiveness of homoeopathic remedies for any specific condition (14).

Dr. Klaus Linde, principle author of the 1997 Lancet meta-analysis (9) (Centre for Complementary Medicine Research, Munich, FRG), authored the BMJ review of research on St. John's Wort for depression (15). The final author was Dr. Wayne Jonas (Dir. Off. Alt. Med., NIH, USA). Funding included the pro-homoeopathic Carl and Veronia Carstens Foundation, Essen, FRG. (9) and for Linde's additional negative homoeopathic review (16), also the pro-homoeopathic OAM, Blackie Foundation Trust and Homoeopathic Trust (9). They also acknowledged the contributions of the documentational centres of the laboratories of Boiron, Dolisos and Heel. To placate sponsors, they resorted to interpreting the results as "not compatible with the hypothesis that the clinical effects of homoeopathy are 'completely' due to placebo", with an honest bottom line: "We found insufficient evidence (185 trials) that homoeopathy is clearly efficacious for any single clinical condition" (9).

What Linde et al found and why. "The combined odds ratio for the 89 studies entered into the main meta-analysis was 2.45 in favour of homoeopathy, (reduced to) 1.66 for the 26 best-quality studies. The ratios were computed such that a result greater than 1 indicated greater effectiveness of homoeopathy. A combination of publication bias and poor-quality trials and/or other factors still unaccounted for might have led to erroneous results. The evidence in our overall analysis would be more compelling if there were independently replicated, large-scale rigorous trials of defined homoeopathic approaches in at least a few specific disorders" (9).

To put this into perspective, a review in the journal Bandolier: Evidence-based health care (CRD, NHS), which provided the comparative quantative analysis of the clinical categories used in part 1 (of this series) and favourably reviewed Kleijnen' s ginkgo and Linde's St John's wort papers (17) & (18), described the results thus: "This will be interpreted by some as signifying that homoeopathy works, but in 60% of trials, homoeopathy could not be shown to have any benefit over placebo. If this were a new treatment, we would look at it with a very cold and fishy eye. A skeptic might say, if this is the best they can do, why bother ?" (19).

"Publication bias is a significant problem and occurs when the chance that a trial is reported depends to some extent on the outcome of the trial. We cannot completely rule out bias as an explanation for positive results. Funnel plot of log odds ratios versus their standard errors, has been widely used to detect potential publication bias. The asymmetry indicates missing negative trials. The general non-parametric selection model applied to the 89 studies confirmed that there was statistically significant publication bias and suggested the bias was primarily due to under-reporting of studies with statistically insignificant effects and with negative effects" (9).

"Quality of evidence is a major problem, the mean quality score being 52%. About 2/3 were poor, 1/10 good. Many trials by advocates with high enthusiasm risks incomplete and selective reporting. Major shortcomings were evident on the clinical level. Inadequate peer-review allows other undetected 'fatal flaws'. Overall quality-assessments can mix and obscure confounding, eg. unequal distribution of prognostic factors might explain positive results; knowledge and expectations about receiving 'active' treatment can bias judgements during reporting or measurement of outcomes; dropouts, withdrawals, or inadequate follow-up can result in unequal distribution of results between groups not due to treatment effect; and multiple outcome-measures or post-hoc selection of outcomes can lead to reporting false-positives. No trials met our criteria for reproducibility" (9). Of the only three qualifying industry inclusions, the combined quality scores were 48.5, 31.5 and 24 out of 100 (9).

"Patients, physicians, and purchasers need valid and reliable information (unencumbered by opinion) on which to make decisions. Whilst randomised placebo-controlled trials hold an important place in such decisions, it is likely that higher quality trials in homoeopathy will show less significant results. We found little evidence of effectiveness for a single homoeopathic approach on any single clinical condition. In the end homoeopathy may be found to have no value" (9). In subsequent correspondence, Linde and Jonas respond to 3 letters to the editor enthusing over the data: "We do not share the enthusiasm. The evidence is not overwhelming" (20). Responding to information of this nature, a London health authority recently stopped paying for homoeopathic purchases, after a decision to support only evidence based medicine led to a review of recent research, including that by the Royal Homoeopathic Hospital, which produced no evidence of clinical benefit (21).

In the Lancet, Prof. M Langman, (Univ. Birmingham) commented: " Only 34 trials showed adequate evidence of concealment of treatment allocation and 28 sufficient handling of drop-outs". Prof. J Vandenbroucke (Univ. Leiden) commented: "A randomised trial of 'solvent only' versus 'infinite dilutions' is a game of chance between two placebos. The authors used a funnel plot to look at the results. If there is publication bias, there should be a gap on the negative side of the plot. Linde et al find a bunch of outliers among the positives" (9).

In this regard, Vandenbrouke in the BMJ petitioned for experts' views, pointing out that "Egger et al's funnel plot test predicts that their might be a problem because the funnel plot is asymmetrical and that the cause of the asymmetry can be anything from publication bias, willingness to please during data collection, data massage in the analysis, downright fraud or a mix of these". Matthias Egger (Univ. Berne, Switzerland) responded: "Results of meta-analysis will depend on how many small or large studies are included (more positive results in smaller trials). Vandenbroucke could have benefited from a formal analysis of funnel plot asymmetry when he discussed a recent meta-analysis on homoeopathy (9), since the significant funnel plot asymmetry lent support to his assertion that bias had produced a body of false positive evidence" (figure provided) (22).

In subsequent Lancet correspondence, Dr. M Francis-Kahn (Me'decin de l'Hospital Bichat, Paris, Fr.) wrote: "One can challenge results obtained with dilutions retaining some active molecules and high dilutions in which no active molecule is present and results presented by a homoeopathic drug company. A report rated negative by Kleijnen is in Linde's meta-analysis rated as positive. Andrade's overall conclusion is a negative. The report by Fisher (London Homoeopathic Hospital) was so poor that a critical study was published in the Lancet showing the inappropriate use of statistics. With respect to the negative best controlled study by French health authorities to confirm or contradict two previous quite poor reports, it is unfair to write that the pooled effect was in favour of homoeopathy"(21). Dr. A Koch, (Univ. Heidelberg) wrote: "Where there is no concealment, two placebos might well differ with respect to efficacy if there is one in which one can belief more" (20).

Prof. E Ernst, holds the world's first permanent Chair in Complementary Medicine, (Dept. Compl. Med. Univ. Exeter, UK). Prof. Ernst has published positively in medical journals on eg. garlic, St John's wort and yohimbe; extensively on placebo and on safety and efficacy of complementary medicines, and has authored textbooks on complementary medicine (23), garlic (24) and homoeopathy (25). In the Lancet he responded as follows: "We compiled data from trials of homoeopathy published after Linde and collegues' searches were completed. Linde mentions two, both of which were negative. We found four further reports and the only common factor is that none of them show any superiority of homoeopathy over placebo. Furthermore, a recent systematic review of seven controlled trials of homoeopathy for a condition judged non-clinical by Linde, included three randomised controlled trials, all of which reported negative results for homoeopathy. The picture painted by Linde may well be slightly more positive for homoeopathy than recent published evidence implies" (20).

All the above-mentioned researchers have in common an interest in complementary medicine taking its rightful place in health care, which is only possible if evidence-based. They are recognised authorities in their respective fields, eg. Ernst and Linde were invited speakers at the September OMEOMED '97 (Univ. Urbino, Italy), first World Congress- "New Frontiers in Medicine: From Molecular Paradigm to Biophysical Paradigm".

Ernst, Berman, Kleijnen and Linde and their institutions were collaborators in the October 1997, 3^rd Scientific Placebo Symposium of Einsiedeln, Switzerland, organised by the Foundation for Innovative Critical Appraisal and Practice in Medicine. Jonas and Kleijnen are to be keynote speakers at the 5^th Annual Symposium on Complementary Healthcare in December 1998, in Exeter, UK. All are key members of the Cochrane Complementary Medicine Field, of which Dr. Brian Berman, Division of Complementary Medicine, Univ. Maryland USA, is the Field Coordinator. Prof. Jean-Pierre Boissel directs the French Cochrane Centre.

Cochrane Centres world-wide are evaluating both paradigms according to the available evidence. Dr. Ian Chalmers, Director of the UK Cochrane Centre, a vociferous proponent of systematic reviews, illustrated their objectivity when he told delegates at a conference on integrated medicine in London recently that "Critics of complementary medicine often seem to operate a double standard" and that "the aim should not be to indulge in data-free arguments, but to assess the effectiveness and safety of any healthcare intervention, be it orthodox or complementary" (26). Dr. Jimmy Volmink, Director of the South African Cochrane Centre at the MRC in Parow, Cape Town, is quoted as saying that he believes that "particular attention should be paid to reviewing complementary medicine" (27). Volmink opinioned that though only 60-80% of routine orthodox clinical intervention is evidence-based, "the highest standards should apply equally". An offer to assist my research and an invitation that I collaborate with the Centre was extended (28), of which the present report is the first stage, based in part on information exchange with Dr. Patrice Matchaba, and with further collaboration intended.

1. The Placebo Response: Biology and Belief. Conference, Univ. Westminster, Lond. Nov. 1996;
2. OAM Placebo and Nocebo Conference. Office of Alternative Medicine, NIH, Dec. 1996;
3. The Placebo Effect: An Interdisciplinerary Exploration. A Harrington, Ed. Harvard Univ. Press. 1997;
4. Brown W. The Placebo Effect. Scientific American. Jan, 1997;
5. Davidson J H. Mind, Medicine and the Placebo Effect. Positive Health. Feb. 1998;
6. Shapiro A & E. The Powerful Placebo. John Hopkins Univ. Press. 1998;
7. Vincent C, Furnham A. Complementary Medicine: A Research Perspective. J Wiley & Sons. 1998;
8. Kleijnen J et al. Br. Med. Jour. 1991, Feb. 9; 302(6772);
9. Linde K, et al. Lancet. 1997; Sept. 20; 350(9081);
10. Hill C, Doyon F. Rev. Epidemiol. Sante Publique. 1990; 38(2);
11. Kleijnen, J. Knipschild, P. Lancet 1992, Nov.7; 340(8828);
12. Kleijnen, J. et al. Effectiviteit van alternatieve geneeswijzen. Rijkuniversiteit Limburg 1993;
13. GRECHO. Presse Med. 1989; 18;
14. Boissel, J. et al. HMRG. Rep. to the Europ. Comm., Brussels. 1996;
15. Linde, K. et al. BMJ. 1996; 313;
16. Linde, K. Jobst, K. Cochrane Rev. Cochr. Libr., Issue 2, Nov.1997;
17. Bandolier No. 18, Aug. 1995;
18. Bandolier No. 32, Oct. 1996;
19. Bandolier No. 45, Nov. 1997;
20. Lancet, 1998; Jan. 31; 351(9099);
21. BMJ, 1997; May 31; 314 (1569);
22. BMJ, 1998; Feb. 7; 316(469);
23. Ernst, E. (ed.) Complementary Medicine: An objective appraisal. Lond. Butterworth Heinemann, 1996;
24. Ernst, E. How Garlic protects your heart. Surrey. Amberwood Publ. 1996;
25. Ernst, E. Hahn, E. Homoeopathy, a critical appraisal. Lond. Butterworth Heinemann, 1998;
26. BMJ, 1998; June 6; 316(1694);
27. Judith Soal, Cape Times "Insight", 1998; May 15;
28. Personal communication, 1998; July 30.