After yesterday’s post about a naturopathy (a.k.a., an ND, or “Not-A-Doctor”) named Anke Zimmermann who treated a four-year-old boy with a homeopathic concoction claimed to be made from the usual dilution to nothing of saliva from a rabid dog in order to cure him of aggressive behavior, pretending to be a dog, and fear of werewolves (I kid you not), I thought more about homeopathy and just how ridiculous it is. Given that naturopathy is a form of pseudo-medicine based on prescientific vitalism and, essentially, principles of magic, it’s not surprising that naturopaths embrace homeopathy. Indeed, as I’ve said so many times before, you can’t have naturopathy without homeopathy because homeopathy is an integral part of naturopathic education and practice. However, it is surprising that a number of real doctors (i.e., MDs) actually believe in homeopathy and prescribe homeopathic remedies.
Having been born, raised, and trained in medicine in the US, I note that homeopathy isn’t as big a deal here as it appears to be in the UK and Europe, particularly Germany and France. It’s more accepted there in ways that it isn’t in the US, where seldom will you find an actual physician prescribing homeopathy, although, of course, there are a number of over-the-counter homeopathic remedies that one can buy at major pharmacy chains like CVS, Walgreens, and the like. In the US, however, most people are unaware of the magical principles of homeopathy, such as the law of similars (i.e., “like cures like,” in which to relieve a symptom a homeopath will use something that causes the symptom) and the law of infinitesimals (which claims that diluting a remedy makes it stronger, with the very highest dilutions being many orders of magnitude greater than Avogadro’s number, meaning that the remedy is in essence diluted away to nothing, sometimes by factors greater than the estimated number of atoms in the known universe). Here, most people think that homeopathy is just herbal medicine.
An interesting question arises when one considers physicians who embrace The One Quackery To Rule Them All, along with real medicine. This is not a question limited to homeopathy, but homeopathy is one of the best modalities to consider for this because it is so obviously utter bullshit. The reason is that it is an extreme version of “integrative medicine,” which “integrates” alternative medicine with real, science-based medicine. Advocates of integrative medicine claim that it represents the “best of both worlds,” while critics (like myself) point out that combining pseudoscience and quackery with science-based medicine doesn’t make medicine better, but rather dilutes the scientific basis of medicine, thus making it worse. But how can we test the hypothesis that integrating quackery with medicine? Ben Goldacre did just that in a recent study.
Here’s the basic idea. Goldacre and colleagues examined practices in the UK that prescribed homeopathic remedies and measured associations with four prescribing and two practice quality indicators. The basic question is simple: Do practices that use homeopathy do worse in standardized quality indicators. AFter noting in the introduction that homeopathy is not only incredibly implausible from a scientific standpoint but that the clinical evidence for it is also negative, Goldacre describes the rationale:
Despite the lack of evidence for homeopathy, and its lack of a plausible mechanism, some NHS doctors still prescribe it. However, there is limited evidence on clinician factors associated with choosing to use hom- eopathy, mostly based on surveys. German medical students taking elective modules in homeopathy scored lower in ‘science orientation’, but higher in ‘care orientation’ and were less motivated by ‘status’ compared to their peers. A commonly cited advantage of homeopathy is safety, and in a small survey of healthcare staff from general practice in London, 70% thought homeopathy could reduce costs for some conditions; however, 55% thought it could increase costs for others. Globally, personal use of homeopathy in doctors has a strong association with prescribing of homeopathy. There is also some evidence on patient factors associated with choosing homeopathy, again based on surveys: patients are most likely to be female, better educated, have healthier lifestyles and report lower tendency to seek medical help when their child is ill.
Using publicly available data, practices may be measured on their prescribing quality through assess- ment of the cost-effectiveness, efficacy and safety of medicines prescribed, based on national guidelines. Practices may also be judged by their quality out- comes framework score and patient recommendation rates. We hypothesised that practices that prescribe any homeopathy might differ in their prescribing in other measurable ways. We therefore set out to explore whether general practices prescribing homeo- pathic remedies also behave differently on these other measures of general practitioner behaviour.
To investigate this question, Goldacre used a retrospective cross-sectional design incorporating UK general practices that have prescribed any homeopathic remedies in the last six months versus those that did not prescribe any in the same period of time. They used both univariable and multivariable logistic regression to investigate correlation between ever using homeopathy with several measures of practice prescribing quality and behavior. The data source used was the OpenPrescribing.net project, which imports prescribing data from the monthly prescribing data files published by NHS Digital. And then:
These contain data on cost and volume prescribed for each drug, dose and preparation, for each month, for each English general practice. We extracted the most recent six months of data available (December 2016 to May 2017 inclusive). This allowed us to determine practices where homeopathy is prescribed and generate com- posite prescribing measures for practices using the various standard measures of prescribing quality already in use on the OpenPrescribing project (sum- marised below). We also merged the prescribing data with publicly available data on practices from Public Health England.15 This allowed us to adjust for sus- pected confounders at the practice level. All standard English practices labelled within the data as a ‘general practice’ were included within the analysis; this excluded prescribing in non-standard settings such as prisons. Additionally, in order to further exclude practices that are no longer active, those without a 2015/2016 quality outcomes framework score were excluded. Using inclusive criteria such as this reduced the likelihood of obtaining a biased sample.
Homeopathic prescribing practices were defined as those with at least one prescription for homeopathy within the most recent six months, and several measures were examined, including a composite measure score, aggregate potential savings from measures, composite low-value prescribing; and aggregate price-per-unit potential savings. A number of factors were considered to try to control for potential confounders and systematic differences between practices and practice populations that could influence homeopathy use.
Goldacre found 8,184 practices in the dataset labeled as “general practice,” of which 566 were excluded because they did not have a quality outcomes framework score, leaving 7,618 practices. Of these, 644 (8.5%) met the criteria of having prescribed homeopathy, although 363 of these practices only prescribed one homeopathic remedies and 38 prescribed more than 10 prescriptions and three over 100, with the maximum number of homeopathic prescriptions in any of the practices being 252. Goldacre’s findings are summarized thusly:
Each of the four prescribing quality measurements was strongly associated with prescribing any hom- eopathy. Furthermore, all of the four prescribing quality measurements exhibited a strong, significant trend where practices in each category of worsening prescribing were more likely to prescribe any hom- eopathy (Table 2). Adjustment for various prespeci- fied practice factors such as demographics and index of multiple deprivation score slightly reduced the size but not significance of associations, with adjustment for prescribing volume accounting for almost all of the change in odds ratio.
In the composite measures score, those in the worst score category (>51.4) were 2.1 times more likely to prescribe any homeopathy than those in the best category (<40.3) (multivariable odds ratio: 2.1, 95% confidence interval: 1.6–2.8). The aggregate savings from the measures where a cost saving could be calculated also showed a strong, significant trend: those in categories with more available savings were more likely to have prescribed any homeopathy than those in categories with fewer available savings. Prescribing of other ‘low value’ items, as identified by NHS England, was also strongly associated with prescribing any homeopathy: only 3.9% of practices in the best category prescribed any homeopathy, compared to 12.8% in the worst category (multivari- able odds ratio: 2.6, confidence interval: 1.9–3.6). Aggregate price-per-unit savings were also strongly associated: in the category where the fewest savings are available where prescribing value has therefore been better optimised, only 3.5% of practices prescribed any homeopathy within the last six months, while for those where the most savings were available, 12.7% of practices prescribed homeopathy (multivariable odds ratio: 2.7, confidence interval: 1.9–3.9).
There was a dose-response relationship found as well:
We found that prescribing any homeopathy is asso- ciated with poorer performance at practice level on a range of standard prescribing measures. We also found a dose–response relationship, with increasing odds of prescribing any homeopathy associated with worsening categories of performance on each prescrib- ing measure: the worse a practice’s performance was on our standard prescribing measures, the more likely they were to have ever prescribed homeopathy. This finding was robust to inclusion of data into the model on a range of plausible confounders. Lower quality outcomes framework scores were not associated with increased odds of prescribing homeopathy nor were patient recommendation scores. We used a highly inclusive criteria for homeopathy prescribing (?1 pre- scription over six months), and most practices that did prescribe homeopathy did so in small volumes. Given the low level of homeopathy prescribing, it is therefore remarkable that any difference was found.
I’m surprised that Goldacre could resist making the obvious joke about homeopathy, namely how dilution makes it stronger. Of course, this is a real medical journal; so humor is rarely permitted. Still, I’d have tried to sneak a joke like that into the manuscript in a way that got past the editors and peer reviewers. It’s true, though. It is rather amazing that any difference was seen at all in this sample. It’s also true that this is an association. It doesn’t demonstrate causation. However, it does suggest that there’s something about practices that use homeopathy that also leads to their using poorer prescribing practices. These could be, as Goldacre speculates, deeper underlying features of these practices, such as respect for evidence-based practice or the quality of teamwork in optimizing treatment and managing cost.
Another interesting thing about this study is that it does not show what integrative medicine advocates claim for integrative medicine, namely decreased costs. Indeed, it found the opposite. Again, the cost of quackery doesn’t generally replace the cost of evidence-based medicine (which might save money by substituting ineffective treatment). More commonly, it’s additional expense added to evidence-based treatments.
Yes, this is just one study, and it’s a retrospective study. It’s possible that there were confounders that weren’t adequately controlled for. On the other hand, doctors who prescribe homeopathy, The One Quackery To Rule Them All, have gone further down the rabbit hole of pseudoscience in having decided that science-based practice is optional than doctors who might think that the occasional herbal remedy is acceptable or that there might be something to acupuncture because it involves sticking needles into the body. It’s not unreasonable to wonder whether they do more poorly in other areas of medical prescribing. This study suggests that they do. It’s also a cautionary tale that the effects of integrating quackery into medicine could well be more insidious and deeper than even I’ve warned. Combining nonsense like homeopathy (or naturopathy or energy medicine or traditional Chinese medicine) infects medicine with pseudoscience in ways that are not always immediately apparent. There are powerful and wealthy forces promoting the integration of pseudoscience, including homeopathy, into medicine that must be opposed.