Ironing out kinks in the evidence base

Pity poor Terry Hamblin. In 1981 in The BMJ the haematologist from Bournemouth set out to debunk claims about the iron content of spinach. Years later he realised his work contained a major inaccuracy, and he spent the rest of his life trying to kill the myth. It was too late: an academic urban legend had been born.

Nicholas Peoples and colleagues focus on the scourge of citations that are inappropriate, misleading, missing, or just plain wrong (doi:10.1136/bmj-2023-076441).1 Such citations can send you down a rabbit hole of references and, at worst, allow “incorrect ideas to masquerade as facts.” Far worse than denting Popeye’s image, misrepresentation in the medical literature can help create conditions for tragedy, they argue.

For example, the North American opioid crisis was aided by a narrative that alleviated prescribers’ concerns about long term use and addiction. That crisis continues: harm reduction experts in the UK are urging the government and the drugs sector to act, after an alarming rise in overdoses and deaths associated with synthetic opioids, specifically nitazenes (doi:10.1136/bmj.p2421).2

Peoples and colleagues estimate that erroneous citation has been rampant for nearly a century and warn that the emerging powers of artificial intelligence will enable this trend to flourish “with an efficiency and scale not previously possible.” Citations in biomedical literature build an evidence base that helps direct care through, for example, clinical guidelines. But, as the world gets to grips with the AI revolution, how does medicine evaluate new AI tools and ensure that the evidence base they build is of benefit to patients and doctors? Of particular concern is the “black box” nature of the technology, with some AI researchers unsure how it comes to its findings. AI has the potential to help doctors make sense of complex medical data and broaden or refine diagnostic thinking, says Chris Stokel-Walker (10.1136/bmj.p2362 ), but it is important to develop frameworks to check AI’s homework.3 The time to act is clearly now—and the UK is ahead of many, with its careful approach requiring its regulator to approve the use of any system before it can be used to treat patients.

Elsewhere, the timing seems right for constructive discussions on pay (doi:10.1136/bmj.p2601),4 and also for the newly expanded vaccine programme for herpes zoster, which now includes people from 50 years of age with severely weakened immune systems. Research from the US shows that live herpes zoster vaccination can save many people from severe infection related outcomes (doi:10.1136/bmj-2023-076321).5 With a growing population of older people, the prevention of herpes zoster and its complications is an important public health issue (doi:10.1136/bmj.p2497).6

Climate change is a public health emergency, and The BMJ has launched a new series focused on the practical actions clinicians can take to support reaching net zero (doi:10.1136/bmj.p2461).7 Each article suggests one action or project that frontline doctors can implement, whether that be switching to dry powdered inhalers for patients with asthma or finding alternatives to single use facemasks.

The series includes two new articles: how to take steps to reduce waste and loss of nitrous oxide, a potent greenhouse gas, in healthcare settings (doi:10.1136/bmj-2023-076649), and guidance on reducing iron dosing in people with iron deficiency (doi:10.1136/bmj-2023-075741).89 It aims to tap into the power of individuals, channelling the wisdom of UN secretary general António Guterres, who warns that the climate crisis requires us to do “everything, everywhere, all at once” (doi:10.1136/bmj.p2461).7