Skip to content

Does homeopathy have any role in the treatment of sepsis?

March 19, 2016

Ok, this could be just a single word answer – no. However, let’s look at this with a critical eye and examine the available evidence. A PubMed search using the term “homeopathy sepsis” yields just 14 results. Ok, not a great start for the homeopaths and their acolytes. A review of the results reveals that just one of those 14 results is any sort of clinical trial of homeopathy in sepsis. That paper is entitled “Adjunctive homeopathic treatment in patients with severe sepsis: a randomized, double-blind, placebo-controlled trial in an intensive care unit.” It was published in the journal “Homeopathy” in April 2005, a journal published by Elsevier.

My attention has been drawn to this paper on several occasions by homeopaths and homeopathy “fans” who claim that it provides incontrovertible evidence that homeopathy is of benefit in sepsis. This includes the likes of Dana Ullman (aka @HomeopathicDana on Twitter), John Benneth (@JBennethJournal), Steve Scrutton (@stevescrutton) and Sandra Hermann-Courtney (@BrownBagPantry). Interestingly, Ullman and Benneth both feature on the blog site “Encyclopedia of American Loons“. Dana Ullman is here and John Benneth is here. Hermann-Courtney has a second blog site where she talks about how she blocks homeopathy skeptics – the internet equivalent of being an ostrich putting her head in the sand as the cognitive dissonance is too much for her to bear. Meanwhile, Scrutton has fallen foul of the UK’s Advertising Standards Authority here and here. Many people have written about his antics and indeed elsewhere on this blog site you can read about my own “interesting” encounter with him here.

So I decided to critically appraise the paper to see if the claims made stand up to scrutiny.

The authors of this paper are listed as “M Frass, M Linkesch, S Banyai, G Resch, C Dielacher, T Lobl, C Endler, M Haidvogl, I Muchitsch and E Schuster”. The lead author is no less than Professor Michael Frass. In this paper, his place of work is listed as “Ludwig Boltzmann Institute for Homeopathy, Graz, Austria”.  He held the post of Director of this institute from 2002 until 2005. Subsequently, he became president of the Institute for Homeopathic Research. However, he’s also known as a Professor at the Medical University of Vienna, Austria. Frass has written papers on various subjects related to atrial natriuretic peptide in ventilated patients and he is the inventor of an emergency airway device called the Combitube, which was popular for a while in situations where endotracheal intubation was difficult. (Subsequently the laryngeal mask airway and variants such as the iGel have become better known).

Given his clear links with institutes that support and promote homeopathy, it’s no surprise that Frass might be involved with clinical trials of homeopathy. It’s less clear as to why a clinical trial would be conducted in an ICU setting as he doesn’t seem to have any particular connection to the specialty of Intensive Care Medicine. He would seem to be the director of a homeopathy outpatient clinic and to work in the Division of Oncology at the Department of Medicine I in Vienna. Hmmm! Ok, so his credentials for conducting a clinical trial in an ICU seems somewhat underwhelming, but it may be that he has a reasonable background in clinical research that might be applicable. So, I won’t write him off as not being suitable for conducting research in the field of Intensive Care Medicine (ICM), particularly as it seems he did write papers on atrial natriuretic peptide as mentioned earlier. A PubMed search for “Frass M” yields 199 hits. How many of these are clinical trials and how many are related to ICM? Trawling through the results finds, as expected, a fair number of results relating to airway management devices but just a few articles relating to sepsis, but no clinical trials in sepsis. Changing the search term to “Frass M sepsis” narrows the search to just 17 articles. Apart from the trial I will critique in due course, the only other publication which relates to a sepsis clinical trial is a pilot study of a Protein C concentrate, which was published in 2006 (Frass was 5th author). I think, therefore, it would be fair to describe his research background relevant to sepsis as “limited”.

Now, let’s move on to the paper itself. I’m going to compare it against the CONSORT criteria for reporting a clinical randomised controlled trial (RCT). These criteria were described in 2010 while the paper was published in 2005, but they nevertheless make a good way to assess a paper reporting the results of an RCT.

The CONSORT criteria ask a series of questions so let’s go through them in regard to this paper:-

Title and abstract:-

1a Does the title identify the study as being a randomised trial? Yes, the title clearly states this.

1b Is the abstract a structured summary of trial design, methods, results, and conclusions? Yes, this conforms to the CONSORT standards

Introduction :-

2a Does the introduction describe the study background and objectives? Is there a description of the scientific background and explanation of rationale? No. The abstract gives a brief summary of (severe) sepsis incidence and mortality and of some therapies that have been unsuccessfully trialled. It then claims that homeopathy has been shown to be superior to placebo in studies (ignoring the quality and published meta-analyses) and that it has an effect in “high dilutions, even beyond Avogadro’s number”. Here we have a problem – the evidence to support these claims doesn’t bear up to critical appraisal and the concept of a substance exerting an effect when serially diluted (a fundamental tenet of homeopathy – the “law” of infinitesimals). We have, therefore an issue with “prior plausibility“.

2b Is there a description of the specific objectives or hypotheses? Yes, the study objective is stated – to evaluate the effect on outcome of homeopathy on patients with severe sepsis at 30 and 180 days. Here we have another issue with this study. While a 30 day mortality is a standard end-point for clinical trials in Intensive Care and sepsis, no sepsis trials consider 180 day mortality as a reasonable end-point, as there are too many confounding factors that can influence mortality the further out from the time of the trial intervention. We see similar issues when we look at registries of outcomes in other areas of healthcare, for instance the MBRRACE-UK (formerly CEMACH) series of audits of maternal deaths, where any death for up to one year after childbirth is recorded as a “maternal death”, even if (for example) the death was the result of a car crash. Of course, the key is in the interpretation of the data. Recently, 3 large multi-centre trials (ProMISe, ARISE and ProCESS) on early goal-directed therapy in sepsis reported mortality at either 60 or 90 days.

We’ll look at the data in due course.


3a Is there an adequate description of the trial design (such as parallel, factorial) including allocation ratio? Yes, there is a very reasonable description of the trial design. It clearly states that it received the necessary ethical approval and consent process. It describes the eligibility criteria and the randomisation process, which both seem appropriate. It describes the data to be collected and the statistical analysis process.

So far, so good!

That is, until we consider the sample size for the trial. To compare treatments in a clinical trial, you need to know what the baseline outcome frequency is and you also need to assess how big a difference in outcome would be regarded as a worthwhile improvement to consider the new tested therapy as providing a meaningful benefit. For example, if you decide to study the effect of a new treatment on a disease which is always fatal, then if the new treatment results in any survival, you might say that it’s a worthwhile benefit. On the other hand, if the disease typically results in a 50% mortality, you may consider that a 10% (i.e a 45% mortality rate) or a 20% (40% mortality rate) improvement in survival rates is acceptable. While the authors give generalised mortality figures, the range quoted is from 40% to 90% for their target patient group. This, however, makes no reference to their own local outcome data. Hmm. With this in mind, how did they decide what an appropriate sample size would be in order to show an acceptable clinically and statistically significant improvement in mortality rate for the treatment arm patients?

They didn’t.

Oh dear!

Let’s try to help here. In the UK, mortality rates from severe sepsis and septic shock vary across the country, but recent data from the HSCIC and Welsh Government gives a mortality rate of 30% in England and 24% in Wales. (Yes, Jeremy Hunt and David Cameron, just one example of how NHS Wales isn’t a second class service as you so frequently lie about). We’ll, perhaps generously, assume that the 30 day mortality rate of the study control arm represents a true mortality rate for patients with severe sepsis in their ICU. We’ll ignore the 180 day mortality figures for the reasons explained above. The paper gives us a control arm mortality rate of 32.3% –  fairly reasonable for the time at which this study was conducted. With no attempt by the authors to consider what would represent an acceptable clinically and statistically significant improvement in mortality rate for the treatment arm patients, I’ll have to make some assumptions to try to work out how big a sample size is needed to see if such an improvement is both clinically and statistically significant. As a rough guide, in the PROWESS trial for activated Protein C in severe sepsis (now debunked and the product withdrawn following the PROWESS SHOCK trial), the treatment arm mortality rate was 24.7% as compared to 30.8% in the control placebo arm. This was approximately a 20% relative risk reduction and a 6.1 % absolute risk reduction. Let’s therefore calculate, on that basis in the Frass trial, what size patient population would be needed if a 20% relative risk reduction was to be achieved with a reasonable degree of certainty. In other words, we need to perform a power calculation to see whether the required change in outcome could be shown in the trial with a reasonable certainty that the trial result is both likely to be genuine and statistically significant. Without getting too deep into this and the statistics behind sampling errors and statistical significance (you can read an article from the North Bristol NHS Trust’s Research and Innovation Department here), we are fortunate in that we can readily obtain the sample sizes required at the levels required using calculators freely available on the Internet. Given that we’re looking at a baseline mortality in this paper – which isn’t far off that seen in the PROWESS study – let’s work on the basis that a 20% reduction in mortality would have been regarded as an acceptable, clinically significant end-point. In other words, we’re looking for the “treatment” to improve mortality from it’s baseline of 32.3% to 25.8%. We can now use an online calculator to work out a sample size for the study which would have had adequate power to detect that 20% relative risk reduction. We’re looking at what is described as a “binary outcome” – the patient is either alive or dead at the end of the study period. Now the appropriate way to design a clinical study is to aim to compare the “null hypothesis” (i.e. the treatment has no effect) against the “alternative hypothesis” (i.e. the treatment works) with sufficient rigour to reject the null hypothesis. Most clinical trials accept a 5% statistical significance level to eliminate a false positive result (otherwise called a Type I error), which is usually quoted in the results section of scientific papers as being “p ≤ 0.05″. The p-value gives us the statistical significance with that cut-off figure of 0.05. There are problems with accepting this value, as it doesn’t tell the whole story of the validity of the results of a trial, as explained here by Professor David Colquhoun, but we’ll use this figure as being one that’s commonly used. We now need to have a sample size which is capable of accurately rejecting the null hypothesis when the null hypothesis really is wrong. This is often described as having sufficient power to not make a Type II error, i.e it won’t produce a false negative result. The greater the power of a study, the less likely it is to produce a false negative result. In clinical trials, a minimum acceptable power is at least 80% (See this paper on sample size calculation in clinical trials). So, for the purposes of the study in question, I’ll work on a power of 80% as being appropriate for this trial. Putting these figures into an online sample size calculator will allow generation of a sample size necessary for the trial to yield a reasonably meaningful result:-

For this trial to show an improvement from the baseline mortality of 32.3% to 25.8% with a power of 80% at a significance level of 5%, the study would need 1526 patients!!!!! This assumes equal allocation of patients to each treatment arm. By comparison the PROWESS study enrolled 1690 patients.

This trial enrolled a grand total of just 70 patients. The study is so grossly underpowered that it had no chance of producing any meaningful result whatsoever.

The paper does describe the process by which the homeopathic globules (the “remedies”) would be selected and chosen. Now – here we have a MASSIVE fundamental flaw. To quote “The homeopathic doctors (my italics) were free to decide which homeopathic medicine should be applied”. Seriously? The homeopaths were free to decide what remedy to give the patients? On what basis were the different remedies to be chosen?

In homeopathy, a claimed crucial part of the process is the consultation in which the homeopath elicits a symptom history from the patient, in order to choose a remedy. A fundamental tenet of homeopathy is that “like cures like” – usually referred to as the “Law of Similars“. It is the very concept that led Samuel Hahnemann to invent homeopathy. Patients with septic shock are frequently incapable of giving a meaningful history as they are too unwell. How then is it possible for the homeopath to determine with any degree of accuracy (as far as any homeopathic prescription could be described as accurate!) which remedy or remedies to prescribe for a particular patient?

What about the (theoretical) risk of cross-contamination from combinations of therapies? I say theoretical as, no matter what any homeopath might claim, there is no difference in any so-called homeopathic remedy which has been diluted beyond the 12C dilution and no practical difference in any diluted beyond 6C. For an excellent explanation of homeopathic dilutions, I recommend reading “Homeopathy: The Ultimate Fake” and “Discover Homeopathy – Science“. The “remedies” to be used in the studies were to be at the 200C “potency”. Ummm……ok. What this means is that 1 ml of the “mother tincture” – the starting material of the, err, remedy – is serially diluted in 100 ml of water 200 times! This gives a theoretical final concentration of 1 in 100 raised to the power of 200. To put it simply, this is a 1 with 400 (!) zeros after it. This is vastly greater than the estimated number of molecules in the universe (about one googol, which is a 1 followed by 100 zeroes). It’s fair to say that there’s no realistic prospect of finding a single atom or molecule of the “mother tincture” substance in the remedy. That’s before we even consider the problem of how the process of dripping the “remedy” on to sugar (typically lactose) pills can somehow transmit the “water memory” of that mother tincture to those pills.

At this point, there’s really no point in continuing to assess the paper against the CONSORT criteria as I’ve already shown that it has zero chance of reliably produce any sort of meaningful result. At best, it could be regarded as a pilot or feasibility study, but it has been presented as a clinical trial of an adjuvant therapy for sepsis, in much the same way as activated Protein C was studied as an adjuvant therapy in the PROWESS study. At this point I should remind the reader that subsequent trials of activated Protein C led to a re-evaluation of its efficacy, prompting the manufacturer, Eli Lilly, to withdraw it from the market. This shows that even when a study fulfils the criteria for being a “good” study and produces a positive result, it is still subjected to rigorous critical appraisal and its findings challenged by this process and by further data collection from other clinical trials and post-marketing surveillance.

Given the number of meta-analyses and reviews of homeopathy (Linde et al 1997 and 1999, Shang et al, The House of Commons Science and Technology Committee report, Cochrane Reviews, and Australia’s NHMRC) that have shown it to have no benefit for any disease or condition, it beggars belief that homeopathy is still marketed and practiced as a system of healthcare. Putting it bluntly, homeopathy is quackery. If a “Big Pharma” company were to behave in this manner (and it’s pretty certain that many of them have), there would, quite rightly, be an outcry as such behaviour is immoral, unethical and quite possibly illegal (though I would defer to legal experts on the question of legality). In recent years, a number of successful campaigns have been organised which are having the effect of drastically reducing the use of homeopathy on the NHS (yes, it is used in some parts of the UK – unbelievable!) and have demonstrated to the public that homeopathy is just simply useless. I refer the reader to the Nightingale Collaboration, Sense About Science, the Good Thinking Society and the 10:23 campaigns for further information on this.

To summarise this paper:-

A poorly structured study with no clue as to what they were aiming to show. This led to the study being grossly unpowered by a factor of more than 20. A variety of “remedies” used with no apparent control mechanism which had even the remotest possibility of detecting any effect from a particular remedy or combination of remedies.

It’s an utter car-crash!

To be fair to the authors, they conclude only that “Our data suggest that homeopathic treatment has a beneficial effect on the long-term survival of patients with severe sepsis, further research is required before making firm recommendations”. Sadly several homeopaths think that this paper is incontrovertible evidence that homeopathy is of benefit in septic shock. This begs the question – why, when this paper was published in 2005, has no-one published any further papers of clinical trials of homeopathy in sepsis?

To anyone with any reasonable knowledge of what homeopathy really is, it will not surprise them that there are no further publications on the subject. Homeopathy is an ineffective treatment of nothing more than sugar pills which have – at best – a placebo action. Homeopaths may twist and turn and use special pleading and downright lies and deceptions to promote their wares, but they cannot escape the facts of the matter. I’m disappointed that any Research Ethics Committee gave permission for a trial of useless sugar pill quackery in patients with a real, serious, life-threatening illness. I’d certainly be surprised if any Research Ethics Committee gave permission for any future trials of homeopathy in sepsis.

From → Health

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: