Dare to know: risk illiteracy and shared decision makingBMJ 2011; 342 doi: https://doi.org/10.1136/bmj.d2075 (Published 06 April 2011) Cite this as: BMJ 2011;342:d2075
- Iona Heath, president, Royal College of General Practitioners
The health of nations is undermined by inadequate health literacy, and particularly by failures of statistical literacy, among doctors, patients, journalists, and policy makers. This conviction led to the convening of a meeting of the Ernst Strüngman Forum to discuss to find ways to combat this problem. The meeting was convened in Frankfurt am Main in October 2009, and this book summarises those deliberations and represents with remarkable accuracy all the strengths, but also perhaps the weaknesses, of such a process.
The best part of the book presents a devastating dissection of the statistical illiteracy of doctors. To give just one example, 160 German gynaecologists were provided with the information needed to calculate the chances that a woman with a positive screening mammogram actually has breast cancer: a prevalence of 1%, a sensitivity of 90%, and a false-positive rate of 9%. The doctors were asked, “What would you tell a woman who tested positive that her chances were of having breast cancer?” The best answer is that about one out of 10 women who test positive have cancer. Yet 60% of the gynaecologists thought that eight or nine women out of the 10 have cancer. The emotional toll on the women given such misleading information must be immense. Such studies have been replicated in many different healthcare systems and for many different diagnoses and this book rightly deplores the situation. Yet the teaching of medical statistics is represented in only a tiny proportion of the curriculums of most medical schools, and the science of uncertainty and the challenges of risk communication are almost completely absent.
It is no wonder that doctors struggle, and their attempts are further undermined by the misleading presentation of data. “Every health statistic can be reported in a transparent or misleading way. For instance, absolute risks, mortality rates, and natural frequencies are transparent, whereas relative risks, five year survival rates, and conditional probabilities (eg, sensitivities) tend to mislead physicians and patients alike. At issue is not lying, but the art of saying something correctly in a way that most listeners will understand wrongly.”
For this reader, there is a rich irony in the presentation of David Cameron as a model of good practice. He is quoted as announcing in 2010 that, “the Government intends to bring about an NHS information revolution, to correct the imbalances in who knows what. Our aim is to give people access to comprehensive, trustworthy, and easy to understand information.” Yet, he has been supporting the logic of Andrew Lansley’s so called reforms of the NHS by claiming that UK survival rates for cancer and heart disease are the worst in Europe. In so doing he seems to pretend that survival rates equate with mortality rates—which they do not—and, as is carefully explained elsewhere in the book, he succumbs to both lead time bias (“screening results in early detection and thus increases five year survival rates by setting the time of diagnosis earlier”) and overdiagnosis bias (“screening increases survival rates by including people with non-progressive cancers, which by definition do not lead to mortality”).
Instead of providing information that is trustworthy and comprehensive he models the art of saying something correctly in a way that he knows his listeners will understand incorrectly. He should read at least some chapters of this book, but the chances of that seem vanishingly small, and are not helped by the weaknesses of the book itself.
Books with multiple authors are always a challenge, and the inherent problems of repetition soon become clear and those of contradiction not much later. The book speaks of the dream of statistical literacy, which is seen to embody Immanuel Kant’s ideal: sapere aude, or dare to know. The more courageous authors acknowledge that the uncertainties inherent in healthcare mean that often there is no single right answer for patient or doctor, and that both must actively share the decision as to which of several options to pursue. This does not accord well with the contributor who tells us that the causes of “patients failing to receive optimal care” include “the unwillingness of patients to accept evidence” and “failure to adhere to treatments.” Such attitudes would seem to do nothing to encourage patients to dare to know.
There is a constantly recurring tension between the desires to inform and to persuade. Some of the persuasion, perhaps best exemplified by leaflets that support screening programmes, is motivated by good intentions abetted by wishful thinking, but persuasion can also be more malevolent and more deliberately deceptive. Good intentions seem too often to blind their owners to the need to trust in patients’ ability to make their own decisions.
The multiple authors of the book make the point that more care is often worse than less. If only they had applied the same judgment to their own efforts. At 100 pages this book could have been riveting and invaluable. Sadly, at almost 400 pages, it becomes repetitive, contradictory, and ultimately tedious.
Cite this as: BMJ 2011;342:d2075
Better Doctors, Better Patients, Better Decisions: Envisioning Health Care 2020
A book edited by Gerd Gigerenzer and JA Muir Gray
MIT Press, 2011, 404 pp
Competing interests: None declared.
Provenance and peer review: Commissioned; not externally peer reviewed.
IH, though president of the Royal College of General Practitioners, writes in the BMJ in her personal capacity.
bmj.com/podcasts Listen to a round table discussion on shared decision making with the authors of this book (http://podcasts.bmj.com/bmj/2011/03/31/bmj-round-table-shared-decision-making-full-version/) and an accompanying feature (BMJ 2011;342:d2057, doi:10.1136/bmj.d2057); bmj.com/archive Read Iona Heath’s columns on bmj.com, including her latest, The politics of fear (BMJ 2011;342:d1273, doi:10.1136/bmj.d1273)