Overdiagnosis and overtreatment
Overdiagnosis and overtreatment
Beware of guidelines with expanded disease definitions
Abstract and Keywords
Along with other aspirations, shared decision making (SDM) has the laudable aim of bringing more scrutiny of evidence into our health care choices, including assessing the risks and benefits of all treatment options. But as evidence about the problem of ‘overdiagnosis’ grows, we may need to bring more attention to the risks and benefits of a diagnostic label, even before any scrutiny of the pros and cons of possible treatments. This chapter explores how heavily conflicted guideline panels are expanding disease definitions, labelling those with milder symptoms or at lower risks, for whom a diagnosis and treatment may do more harm than good. It explores several examples where there is controversy over recently widened definitions. The chapter concludes that if we’re to better inform people about the risks of overdiagnosis and related overtreatment, and ultimately assist in reducing them, genuine SDM will include information about controversies over disease definitions.
Introduction to overdiagnosis and overtreatment
Along with other aspirations, shared decision making (SDM) has the laudable aim of bringing more scrutiny of evidence into our health care choices. More valuable still, the approach reminds us to examine risks and benefits of all treatments options, including where appropriate, doing nothing, or watching and waiting (Hoffmann et al., 2014). In the era of too much medicine (Glasziou et al., 2013), such an approach is prudent. But as evidence about the problem of overdiagnosis grows, we may need to bring more attention to the risks and benefits of a diagnostic label, even before any scrutiny of the pros and cons of possible treatments.
Changes to disease definitions, often arising from clinical guidelines from heavily conflicted expert panels, are lowering diagnostic thresholds and expanding the pool of those labelled as sick, medicalizing more and more people with milder symptoms or at lower risk (Moynihan et al., 2013). Such changes may benefit some, but for others among the newly diagnosed, the deleterious impacts of a label and subsequent treatment may do more harm than good. Former head of the DSM IV task force Allen Frances argues conflicts of interest, both financial and intellectual, mean all guidelines which define disease thresholds are suspect, and former president of the UK Royal College of General Practitioners Iona Heath suggests provocatively that if guidelines drive overdiagnosis or overtreatment, it’s our responsibility ‘not to follow the rules.’ (Heath, 2014) As explored in this chapter, genuine controversy exists within the scientific literature over many recently widened disease definitions, including, for example, gestational diabetes, chronic kidney disease, and attention deficit hyperactivity disorder (ADHD).
As we embark on genuine SDM, uncertainty and controversy over diagnostic thresholds, and the resulting risk of a person receiving an unnecessary and potentially harmful diagnosis, may need to become part of our clinical conversations far more often. There is as yet no established evidence base to help recommend the most effective strategies for such communication about overdiagnosis, but this research is beginning and more is needed. Bringing such complex and sometimes counter-intuitive concepts into our (p.130) decision making will be confusing and challenging, but crucial, if we are to help properly inform people about the risks of overdiagnosis and related overtreatment, and ultimately assist in reducing them.
What is overdiagnosis?
The ‘modern epidemic’ of overdiagnosis is increasingly recognized as a significant source of harm and waste within health care systems (Hoffman and Cooper, 2012; Welch et al., 2011). Overdiagnosis happens when people are given a diagnosis for a disease that will not harm them, and it can sometimes lead to unnecessary tests and treatments, taking resources away from those who could most benefit. The problem has many technological, commercial, and cultural drivers (Moynihan et al., 2012). One important driver is sophisticated diagnostic technology like CT and MRI scans which enable the detection of ever smaller ‘abnormalities’, many of which will never cause harm. An example is the increased diagnosis of pulmonary embolism, blood clots in the arteries of the lungs, where there are now strong concerns that new technology is driving some degree of overdiagnosis and overtreatment (Wiener et al., 2012).
Screening programmes targeted at the healthy, along with their benefits, can also cause harms, including overdiagnosis. While estimates of the magnitude of screening-related overdiagnosis vary, an expert meeting convened by the United States National Cancer Institute observed, ‘overdiagnosis is common’ and occurs ‘frequently more with cancer screening’ (Esserman, 2013, p. 797). For example, after assessing all available evidence, an independent panel in 2012 estimated that 19% of breast cancers detected during mammography screening may be overdiagnosed: defined as detection of cancers that do not progress to be symptomatic and ‘would never have been found were it not for the screening test’ (Marmot et al., 2012, p. 1782). To aid SDM in this field, work is already underway to develop effective methods for communicating about overdiagnosis risk and mammography (Hersch et al., 2014; Waller et al., 2014). For example, Hersch and colleagues (2014) are currently conducting a randomized trial among women in Australia approaching the age of invitation to breast screening. In the trial, one group will receive information about overdetection of inconsequential cancers, in addition to standard information received by women in the control group. Similarly, Waller and colleagues (2014) found that brief written information about overdiagnosis and mammography, of the sort being sent to women in the UK, was not well understood and as a result may not facilitate informed choice.
In this chapter, the focus is on a less investigated pathway to overdiagnosis: the way disease definitions are expanding, often caused by small changes to diagnostic criteria made by specialist expert panels in clinical guidelines. Such tiny changes can have large implications, sometimes automatically medicalizing millions of people more than before.
How are guidelines causing expanded disease definitions?
Concerns about collateral damage from medicine’s expanding empire have deep historical roots, dating from old desires to avoid iatrogenic harm to Illich’s prophetic work on (p.131) over-medicalization (Illich, 1976). In 2012, I led a cross-sectional study of the influential expert guideline panels in the USA setting which had made recent changes to the definitions of 14 common costly conditions, including hypertension, dementia, ADHD, and asthma (Moynihan et al., 2013a). Of 16 publications from those expert panels, ten proposed changes deemed to lead to a potential expansion of the numbers of people labelled, one publication narrowed the definition, and for five it was unclear. The expansion fell into three categories: creating pre-diseases like pre-dementia, lowering diagnostic thresholds, as happened with high cholesterol and ADHD, and earlier or different diagnostic methods, as occurred with changes for rheumatoid arthritis and myocardial infarction. No guideline panel publication included a rigorous assessment of the risk of overdiagnosis, and references to potential harms of the proposals to change definitions were cursory.
Among those guideline panels which disclosed conflicts of interest, 75% of members had extensive financial ties to a median of seven pharmaceutical companies, many of which sold products for the relevant conditions. For example, among the 11 members of a panel which created the large new diagnostic category called ‘pre-hypertension’, nine, including the chair, had financial ties (e.g. payments for speaking and/or consulting and/or research), to a median of 12 pharmaceutical companies, including those selling hypertension medications. Importantly our study made no finding of causation between financial ties and definition changes. In response to this evidence, and wider calls for reform of guideline panel constitution (Guyatt et al., 2010), representatives from an international group of organizations, including the Guidelines International Network, are planning new guidance for panels which review disease definitions. While the 2012 study of expert guideline panels did not analyse the merits of specific disease expansions, there are many examples, including the three below, where there is heated debate within the literature about where the line is being drawn between the healthy and the sick.
Conflict over new definition which triples prevalence of gestational diabetes
In 2010, the International Association of Diabetes and Pregnancy Study Groups (IADPSG) proposed new criteria and assessment approaches for gestational diabetes which would expand the proportion of all pregnant women diagnosed from around 6% to an estimated 18%. While adopted by successive nations, and based on the hope that it would help prevent harm to mother and child, the expanded definition is also the subject of sustained criticism from some diabetes researchers and clinicians. For example, Cundy and colleagues have argued in the BMJ (2014) that the new criteria mean ‘a large number of hitherto healthy pregnant women will become labelled as diseased’ and that the changes seem ‘a striking example of overdiagnosis’ (p. 3).
In 2013, a National Institutes of Health (NIH) Consensus Development Conference was convened in the USA to investigate the controversy, using a three step process—an independent panel of professional and public representatives was assembled free of (p.132) reputational and financial conflicts of interest, a systematic review was undertaken, and testimony was sought from researchers and clinicians active in the area. The report of the independent 2013 NIH panel expressed several serious concerns about the expanded 2010 definition: concern over whether newly diagnosed women would benefit from treatment and if so by how much, concern about direct and indirect costs to the health system, concern about unintended consequences of labelling, including an increase in caesarean births, life disruptions, and patient costs. It recommended staying with the old criteria. Notwithstanding the challenge in communicating such controversy to pregnant women, genuinely sharing decision making surely requires it.
Controversial expansion of ‘chronic kidney disease’ labels one in two seniors
A new broad framework created in 2002 for defining and classifying ‘chronic kidney disease’ (CKD) labels around one in eight adults, and one in two people over 70, with CKD. The diagnostic threshold is based in part on a measure of kidney function, unadjusted for age, and while its aim is early detection and prevention of genuine illness, the definition also labels people who have normal aging of their kidneys (Moynihan et al., 2013b). Despite criticism from some kidney researchers, and concern among general practitioners about medicalizing normal aging processes, an expert panel largely reaffirmed the expanded definition in 2012. The original 2002 guideline was sponsored by a pharmaceutical company, over half of the 2012 panel disclosed financial ties to device or drug makers, and the body which develops the guidelines takes funding from a consortia of device and drug companies. A recent analysis of the controversy (Moynihan et al., 2013b), estimated the current definition may ‘misclassify’ at least 30% of the elderly, with those classified as having stage 3A CKD at the ‘highest risk of overdiagnosis’ (p. 3). The analysis suggests clinicians should ‘share uncertainty’ with patients about the appropriateness of the CKD diagnostic thresholds, and it calls for an independent panel to review the entire CKD framework.
Lowering thresholds increases concerns about risk of attention deficit hyperactivity disorder
Thomas and colleagues (2013) have argued that the recent lowering of diagnostic thresholds for ADHD, ‘increases the risk of confusing ADHD with normal development processes’ and are ‘likely to increase what is already a significant concern about overdiagnosis’ (p. 3). Despite professed attempts by the American Psychiatric Association to minimize conflicts, over half of the members of the panel which effectively expanded the number of people who will be labelled ADHD disclosed ties to pharmaceutical companies which sell ADHD medications (Moynihan et al., 2013a). For mild and moderate cases, which constitute the bulk of diagnoses, Thomas and colleagues (2013) propose a conservative approach to diagnosis, ‘to reduce unnecessary diagnoses without risking undertreatment of those who really need psychiatric help’ (p. 3).
First steps in preventing overdiagnosis and associated overtreatment should include both informing and reforming. Information about expanded disease definitions should be shared more often with those most at risk of harm from them. To that end it is encouraging that a series in The BMJ on overdiagnosis (Glasziou et al., 2013) is currently being translated into lay language by the influential US group Consumer Reports, though much more innovation and evaluation is needed to optimize such counter-intuitive communication strategies. At a basic level, communication about controversy around disease definitions could become more of a routine part of clinical conversations. At a more complex level, clinical interactions could more routinely involve linking different diagnostic thresholds within specific conditions to different chances of benefits and risks of a diagnosis and/or treatment, including the risk of overdiagnosis and overtreatment. However, in any attempts at innovative communication strategies, much lateral and cautious thinking is required, to make sure our efforts to wind back harms do not create any more of them.
A key challenge is identifying reliable evidence about the appropriateness of different diagnostic thresholds, when the quasi-official definitions embedded in conflicted clinical guidelines can be so problematic. In my view, addressing that challenge requires clear reform of disease definition processes, involving fresh new panels, free of financial and intellectual conflicts, much more broadly representative, and focused on the potential harms, as well as potential benefits of their endeavours. It is unclear how such reforms will emerge and who will drive them, given the power and influence of established vested interests, both professional and commercial. Yet, whatever the outcome of such reforms should they occur, uncertainty over changing thresholds for diagnosis may become a permanent and certain reality. Enabling people to make more informed decisions about whether or not to accept controversial labels may well emerge as a key aim of the SDM approach.
Cundy, T.Ackerman, E., Ryan, E. 2014. Gestational diabetes: new criteria may triple the prevalence but effect on outcomes is unclear. BMJ, 348: g1567.
Esserman, L., Thompson, I., Reid, B. 2013. Overdiagnosis and overtreatment in cancer: an opportunity for improvement. JAMA, 10: 797–8.
Frances, A. 2013. Too much medicine is bad for our health. Huffington Post 13/9/2013.
Glasziou, P., Moynihan, R., Richards, T., Godlee, F. 2013 Too much medicine; too little care. BMJ, 346: f4247.
Guyatt, G., Akl, E.A., Hirsh, J., et al. 2010. The vexing problem of guidelines and conflict of interest: a potential solution. Annals of Internal Medicine, 152: 738–41.
Heath I, 2014. Quoted in Overdiagnosis: Findings and Action Plan, Quebec Medical Association Report, released August 18, 2014 Available at https://www.amq.ca/images/stories/documents/m%C3%A9moires/surdiagnostic-plan-action-en.pdf [Accessed 31 October 2014].
Hersch, J., Barratt, A., Jansen, J., et al. 2014. The effect of information about overdetection of breast cancer on women’s decision-making about mammography screening: study protocol for a randomised controlled trial. BMJ Open, 4: e004990.
Hoffmann, T.C., Légaré, F., Simmons, M.B., et al. 2014. Shared decision making: what do clinicians need to know and why should they bother? Medical Journal of Australia, 201: 35–9.
Illich, I. 1976. Limits to Medicine. London: Marion Boyars.
Independent UK Panel on Breast Cancer Screening. 2012. The benefits and harms of breast cancer screening: an independent review. Lancet, 380: 1778–86.
International Association of Diabetes and Pregnancy Study Groups. 2010. International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care, 33: 676–82.
Moynihan, R., Doust, J., Henry, D. 2012. Preventing overdiagnosis: how to stop harming the healthy. BMJ, 344: e3502.
Moynihan, R.N., Cooke, G.P., Doust, J.A., et al. 2013a. Expanding disease definitions in guidelines and expert panel ties to industry: a cross-sectional study of common conditions in the United States. PLoS Medicine, 10: e1001500.
Moynihan, R., Glassock, R., Doust, J. 2013b. Chronic kidney disease controversy: how expanding definitions are unnecessarily labelling many people as diseased. BMJ, 347: f4298.
NIH Consensus Development Conference on Diagnosing Gestational Diabetes Mellitus. 2013. NIH Consensus Development Conference Statements Volume 29, Number 1 March 4–6, Available at http://prevention.nih.gov/cdp/conferences/2013/gdm/files/Gestational_Diabetes_Mellitus508.pdf
Thomas, R., Mitchell, G., Batstra, L. 2013. Attention-deficit/hyperactivity disorder: are we helping or harming? BMJ, 347: f6172.
Waller, J., Whitaker, K.L., Winstanley, K., Power, E., Wardle, J. 2014. A survey study of women’s responses to information about overdiagnosis in breast cancer screening in Britain. British Journal of Cancer, 111: 1831–5.
Wiener, R.S., Schwartz L.M., Woloshin, S. 2013. When a test is too good: how CT pulmonary angiograms find pulmonary emboli that do not need to be found. BMJ, 346: f3368.
Welch, G., Schwartz, L., Woloshin, S. 2011. Overdiagnosed: Making People Sick in the Pursuit of Health. Boston: Beacon.