As I have covered in the pages of this magazine previously, the official regulation of alternative medicine in the UK is far from straightforward. The General Chiropractic Council (GCC) and General Osteopathic Council (GOsC) are the only statutory regulators of pseudomedical disciplines, while other therapists are left only to voluntary regulation via an ‘Accredited Register’. More broadly, all advertising (including website and social media content) in the UK is regulated by the Advertising Standards Authority, and consumer protection law is upheld by Trading Standards – the latter of whose budgets have been under severe strain along with all Local Authority spending, rendering it practically impossible to receive even a response to a complaint, let alone action.
Beyond those organisations sits a small network of skeptical activists, whose interests in consumer protection sees them engaging with the regulatory framework by seeking out misleading health claims and reporting them to the relevant regulator or register. For some of the most fortunate of that small band, doing so has become a full-time job.
This voluntary role of external watchdog can feel like an uphill struggle and a thankless task, though success is possible: for example, despite the GOsC’s initial (and farcical) insistence that no complaints would be investigated until they’d first been looked at by the Advertising Standards Authority, reporting misleading claims by osteopaths seemed to make an appreciable difference. A review in 2015 by my colleague at Good Thinking found that 33% of osteopaths were in clear breach of advertising standards; by 2017, after submitting hundreds of complaints to the GOsC, that figure had dropped to 16%. It is currently unclear whether those gains have stood the test of the last seven years.
As for the chiropractic industry, in 2021 I conducted an audit which found that 56% of fifty randomly chosen chiropractors were making claims that were not compliant with the advertising code. After several discussions with the GCC, I decided to contact those chiropractors directly, to let them know of my concerns, to direct them to which parts of their advertising I felt were non-compliant, and to invite them to update their messaging rather than begin the lengthy process of making a formal complaint. Of the 28 chiropractors I’d identified as making erroneous health claims, 21 updated their advertising as a result of my email approach, with many thanking me for directing them to the problematic content. It was an unexpected outcome – and one whose lasting impact is yet to be assessed – but one that highlighted that many practitioners are genuinely willing to try to stay within the guidelines, if they’re helped to do so. Clearly, there is a role for a regulator to take a proactive approach, rather than to wait for complaints to come in (or, worse, to wait for other less specific regulators to have investigated and made a ruling).
While osteopaths and chiropractors are the only alternative medicine practitioners who are subject to statutory regulation, other alternative healthcare practitioners are subject only to voluntary regulation, via the Professional Standards Authority’s Accredited Register (AR) programme. Inclusion in the AR programme is not a mark of efficacy, but it is meant to be a mark of good governance, and of trustworthiness – essentially, a sign that unprofessional behaviour or misleading health claims will not be tolerated. For many of the organisations who have signed up to the AR scheme, the risk of the latter is likely minimal – it’s less likely that practitioners in the fields of play therapy, sports rehab or counselling are comprehensively filled with false health claims. However, in the case of other accredited registers, the risk of being seen to endorse pseudoscientific modalities is much higher – such as in the case of the Complementary and Natural Healthcare Council (CNHC).
The CNHC was set up in 2008 as a voluntary register, and currently oversees more than 6,000 registrants, including from the fields of aromatherapy, colon hydrotherapy, craniosacral therapy, kinesiology, naturopathy, reflexology and Reiki. It is by some margin the largest register in the PSA’s Accredited Register scheme, and the breadth of pseudoscience it regulates no doubt makes it’s remit tricky to manage. At least when the Society of Homeopaths were part of the PSA’s AR scheme, it’s 1,400 members all broadly believed in the same core principles (the SoH, it’s worth noting, withdrew from the AR scheme in 2021 after members were discovered to be promoting homeopathic vaccines and cures for autism); the same cannot be said for modalities like Reiki, aromatherapy, or naturopathy, whose fundamental principles share little in common, and in many cases are fundamentally incompatible with one another.
To date, the CNHC’s approach to regulation has been a manual one, with members of the CNHC’s small team searching their database periodically to peruse member websites, looking to spot claims that don’t have a solid foundation. The limitations of this labour-intensive form of regulation ought to be readily apparent: even if it took as little as 15 minutes to scour and fully evaluate the claims on a given website, the CNHC would struggle to review 25 websites per day – barely scratching the surface of their 6,000+ membership. In reality, with so many different members to review, and such a wide range of therapies, based on such a divergence of theories, the chances of a misleading health claim being spotted by the regulator is very low – meaning the chances that a false health claim will find its way to a vulnerable member of the public is incredibly high.
The scale of the health misinformation problem among CNHC registrants was today highlighted in a new paper, published in the online journal Royal Society Open Science, in which researchers trained an AI tool to scan and evaluate health claims from over 11,000 web pages, representing 725 of the CNHC’s registrants. The AI tool, developed by independent IT scientist Simon Perry, also produced a rationale for why each claim was judged false.
According to the study, 97% of the websites in question contained health misinformation, including some claims related to the treatment of cancer – claims which could therefore be in breach of the Cancer Act (1939). To ensure the tool was accurate, results were compared to the independent findings of a group of experienced skeptics tasked with appraising a sample of the CNHC registrants’ websites, who found a similar prevalence of misleading health claims.
Among the results, the paper analysed claims from 240 reflexologists, 236 of whom were found to be making claims that could not be substantiated, while 146 of the 149 Reiki practitioners assessed were making problematic claims. Elsewhere, all of the 20 “Healing” practitioners, 18 Kinesiologists, and 26 Shiatsu massage therapists were found to be making misleading health claims. Examples of claims which were flagged by the AI tool as erroneous include an acupuncturist who claimed to be able to resolve a chronic knee problem within three sessions, a detox therapy which proponents claim is effective in curing hangovers and relieving allergy symptoms, and an aromatherapist who claimed to be able to prevent shingles, pneumonia and chicken pox.
“It was believed by CNHC and PSA that only 30% of the websites regulated by CNHC contained false claims”, explained Simon Perry. “While we did not look at all therapy types, we demonstrated that in the therapy types we did look at, the problem is much more serious – 97%”.
“We have also demonstrated that, using Large Language Models, it is easy and cheap to analyse thousands of web pages and check for false claims with the same accuracy as an expert,” he continued. “That low cost is particularly significant. We estimated it would cost $150,000 to do this manually, or around $500 with ChatGPT4. But OpenAI halved their prices by the time we started, and it cost $250. It has again halved twice since then, bringing the cost to just $60. That’s to check through 12,500 web pages – but CNHC would only need to do this once. After that, it’s completely feasible to check all the websites every day, scanning for changes, and picking up false claims as soon as they come online. The CNHC and other regulators of pseudoscientific health practices can now choose to largely automate the compliance process”.
While the paper has only just been published, I have been aware of its findings for some time, as I’ve been working with the research team all year to put the findings to the CNHC and to find ways in which to help the CNHC take effective action to protect the public from misleading claims. As a result, following on from my experiences with chiropractors, I randomly selected 100 therapists from the study’s results, and contacted them to bring their attention to the issues the AI had identified.
Once again, the response was interesting. Discounting the 21 practitioners for whom no contact details were listed, or whose email address bounced, I was left with 79 registrants that I was able to successfully contact. Of those, half never responded, but of the responses I received, 4 had already made edits to the site by the time I got in touch, 5 indicated that they’d be speaking to their professional organisation for guidance, and 15 indicated that they were very happy to make changes, and thanked me for my help in bringing their attention to the issues. Only 7 responded to refuse to review their content at all, and instead asked that I make a formal complaint so that the ASA could make an assessment – which, naturally, I happily did.
Many of the registrants I contacted responded to say that they do strive to stay within the guidance, but that they struggle to understand what they can and can’t say; most seemed unaware that the ASA provides a free service to check advertising copy for compliance, so that any issues can be spotted before being shared with the public.
It seems to be the case that there is an appetite among many CNHC registrants to understand the rules, and to stay within them, and what is needed is proactive engagement to help them identify and correct misleading claims. Clearly, the CNHC’s current approach of manually finding, scrolling, reading and checking websites, page by page, is not sufficient, and as a result thousands of false health claims are finding their way to the public, via websites whose pages carry the stamp of legitimacy of an accredited register. This latest research has demonstrated that there is a smarter, quicker and far more effective way of spotting claims that go far beyond the evidence.
Equally, the fact is that half of those registrants I reached out to never even responded, and a fifth were uncontactable in the first place. Of those I did hear from, 10% were unwilling to listen to my concerns – which illustrates how important it is that the proactive engagement and enforcement comes from the register themselves, whose carrot of cooperation is backed up by the stick of potential removal from the register.
What does this mean for the future of alternative medicine regulation? According to Perry, the next step could be to replicate the study with other regulators, such as the GCC, GOsC or Society of Homeopaths, while the same tool that highlighted the scale of this problem can also be used to monitor the effectiveness – or otherwise – of any corrective action the CNHC take.
The tools are out there to make regulation less random, and more strategic and targeted – this latest research is a perfect illustration of that. As I’ve also discovered, the will is there among some therapists to ensure they’re not misleading the public; and where there isn’t the will to clean up their act, regulatory action has to be taken. This applies not just to the CNHC, but to all healthcare regulators and registers. Ultimately, ineffective regulation only harms the end user: the patient or consumer who puts their trust and their health into a treatment that cannot possibly help them.
The post Could AI help fix the issues of ineffective alternative medicine regulation? appeared first on The Skeptic.
A new study uses AI to identify misleading health claims, showing that with the right motivation and technology, proper regulation of alternative medicine is possible
The post Could AI help fix the issues of ineffective alternative medicine regulation? appeared first on The Skeptic.