- Luis Gabriel Cuervo, clinical editor BMJ Knowledge (, )
- Jeffrey K Aronson, reader in clinical pharmacology ()
“There are some patients that we cannot help; there are none whom we cannot harm.”
Attributed to Arthur L Bloomfield
Before you next get into your car or walk out into the street, reflect that the number of deaths from road crashes in the United Kingdom is about 3400 a year.1 Elsewhere in the world the rates are up to six times higher.2 Now compare that with the extrapolated figure for the United Kingdom of more than twice that number of deaths each year from adverse drug reactions, estimated from the results of a prospective study published in this theme issue of the BMJ.3 Now try to estimate the benefit to harm balance of a road trip and the benefit to harm balance of a drug that you have recently used.
But drugs are far from being the only interventions we use in health care; even when drugs are prescribed they are often only part of an overall plan of care. Other interventions include lifestyle changes, surgery, physiotherapy, radiotherapy, rest, and a panoply of other interventions that may be recommended by medical and paramedical staff, relatives and friends, practitioners of complementary and alternative medicine, and the latest issue of Cosmopolitan magazine.
Even with drug treatments we often lack the necessary information to estimate the balance between benefits and harms. Randomised trials are primarily concerned with addressing uncertainties about potential benefits and seldom focus on safety issues. Moreover, even when information about safety is collected, the data may be sparse, may not be collected systematically, and may not be comparable from study to study. This makes the benefit to harm balance difficult to assess.
Consider how much more challenging it is to assess the benefit to harm balance for non-drug interventions. Often such interventions are implemented without reliable evidence of their beneficial effects, let alone their harmful ones. For example, most evaluations of different prostheses in total hip replacement are uncontrolled, and a systematic review found that the primary evidence was too weak to draw valid conclusions.4 Cryotherapy is widely used to treat warts, but evidence of its efficacy and safety rests on a handful of randomised controlled trials and is reportedly both limited and contradictory.5 Wax softeners are used by the gallon, but we don't know about their relative effectiveness, let alone their safety.6 Children with bronchiolitis are segregated in many hospitals to reduce the risk of nosocomial infection, but we do not know if this is effective.7 And we are still unable to answer with certainty the age old question of which position women should adopt when delivering their babies.8
In some cases not only is there no evidence of benefit, but we may remain oblivious to the possibility of harm. How many times have you heard someone say that even if an untested treatment is not going to do any good it is unlikely to be harmful? But there are several examples of interventions that were confidently expected to be both beneficial and completely safe, expectations that were not fulfilled. For example, high concentrations of oxygen for premature babies, which caused retrolental fibroplasia9; bed rest for low back pain, which can delay recovery10; albumin for resuscitating critically ill patients with hypovolaemia, burns, or hypoalbuminaemia, which probably worsens outcomes11; and class I antiarrhythmic drugs, which unexpectedly increased mortality after myocardial infarction.12 We banned relatives from accompanying women during delivery in many hospitals, only to learn decades later that this is far from ideal in most cases,13 and we have implemented some guidelines on how to approach deliberate self harm without realising that they were often wasteful of resources if not actually harmful.14
In all areas of health care—in preventive medicine in both individuals and the population, in screening and diagnosis, in therapeutic interventions (whether drugs are used or not), and in monitoring treatment—we need information about all the effects, both beneficial and harmful, of the relevant interventions, effective ways of gathering that information, proper reporting and effective indexing in databases, and analysis and integration of all this information. And we need people from different disciplines to work together to obtain it. Armed in this way, healthcare professionals should be able to offer their patients a more balanced view of therapeutic benefits and harms and more realistic explanations of what treatments have to offer, generating more active and trusting relationships; and governments and healthcare authorities should be able to make better decisions about choosing strategies to improve public health.
People trust healthcare professionals to make balanced decisions on healthcare issues at the individual and community levels. We are expected to harness reliable evidence, competent expertise, and sensible judgments to inform our decisions, but evidence may be biased or unavailable, and many decisions rely on pious hopes that things will turn out well. Some of the problems and the ways forward are discussed in this theme issue. Have a nice trip.
Competing interests LGC has been reimbursed by BMJ Knowledge, WHO, and MediLegis for participating in meetings and conferences related to evidence based medicine; he is a contributor to the Cochrane Collaboration, and an employee of BMJ Knowledge, a division of BMJ Publishing Group dedicated to evidence based publications.