Statistics from Altmetric.com
When we decided to publish Clarkson and Choonara’s paper we realised that the media might misinterpret their findings, whether unwittingly or mischievously. Consequently children and their families might reasonably be alarmed or unreasonably misled. They might even stop taking their prescribed medicine if it was one mentioned as associated with an adverse reaction. We had seen, and deplored, damage done to the public health by poorly informed publicity over alleged risks of MMR vaccine after the Lancet published a contentious paper.1
We worked closely with the authors to make sure the language used was unambiguous and the conclusions borne out by the data. Nevertheless we and they realise that some doctors find it difficult enough to distinguish between a statistical association and cause-and-effect—so surely news journalists are even less likely to spot the difference.
Clarkson and Choonara provide numerous caveats. They are careful to point out that they did not “formally assess” causality, predominantly because the source of their data—the Yellow Card Scheme of the UK Committee on Safety of Medicines—was not designed for this purpose. They point out that the adverse drug reactions they identify are suspected rather than proven. They state that for all of the drugs listed, benefits are likely to be far greater than risks.
We hope that responsible journalists who decide to report on the study bear these comments in mind before deciding on the message they wish to convey. For example, we hope they will not read too much into two infants dying suddenly, who happened to be taking an antibiotic at the time and whose doctors thought this worth reporting; the odds are they were also consuming proprietary brands of baby food and their parents may have owned a mobile phone. We hope no parent of a child with epilepsy will be provoked by inaccurate news reports into stopping treatment.
But we also hope that paediatricians will learn from the prime message of this paper, namely that they should make themselves aware of guidelines which recommend avoiding certain medications in high risk groups. The authors single out propofol in the critically ill and valproate in those under 3 with developmental delay or being treated with more than one anticonvulsant. We echo their call for more risk-benefit analyses of medicines used by children, particularly newer anticonvulsants.
Above all we hope that UK based colleagues will respond with their usual diligence to the current British Paediatric Surveillance Unit enquiry into suspected adverse drug reactions in children, funded by The Medicines Control Agency, UK. The Yellow Card scheme is too blunt an instrument to tell us anything useful about causation or about balancing risks and benefits.