Mapping the drivers of overdiagnosis to potential solutionsBMJ 2017; 358 doi: https://doi.org/10.1136/bmj.j3879 (Published 16 August 2017) Cite this as: BMJ 2017;358:j3879
- Thanya Pathirana, PhD scholar,
- Justin Clark, senior information specialist,
- Ray Moynihan, senior research fellow
- Correspondence to: R Moynihan
Interest is growing in tackling the problems of overdiagnosis and overtreatment
Possible drivers and potential solutions arise across five inter-related domains: culture, the health system, industry and technology, healthcare professionals, and patients and the public
More work is needed to develop and evaluate interventions aimed at preventing overdiagnosis
Raising public awareness of overdiagnosis is a priority
In our collective enthusiasm to diagnose and treat disease, a growing body of evidence indicates that we may often be doing too much of a good thing.12345 “Overdiagnosis” is now widely recognised to occur when people are labelled with or treated for a disease that would never cause them harm—often as a result of undergoing screening—and it can lead to the overuse of further tests and treatments.26 One example is thyroid cancer, with estimates that over 500 000 people may have received overdiagnoses across 12 countries in the past two decades, leading to unnecessary surgery and lifelong medication for many.7
Overdiagnosis is a challenge to the sustainability of human health and health systems. Its causes—including the best of intentions—are as complex and multifaceted as the potential solutions.8910111213 As part of the preparation for a possible national action plan in Australia, we searched the literature for causes of and responses to overdiagnosis. Here we provide the first comprehensive analysis of the possible drivers of overdiagnosis and related overuse, mapped to potential solutions.
Searching the literature
Our approach to the initial and updated PubMed searches of the literature is outlined in the data supplement (see bmj.com) and yielded a total of 36 articles, to which we added a further five (fig 1⇓). We included articles that explicitly discussed possible drivers and potential responses or solutions to the problem of overdiagnosis. We included original research as well as opinion, commentary, and analysis articles. Reflecting the relatively recent and growing interest in overdiagnosis, the vast majority were published since 2013, with generally increasing numbers each year (rising from three before 2013 to nine in 2016 and four in 2017 up until July).
Given the limitations of the literature to date, we couldn’t assess the quality of evidence behind each claim in each article, so this is not a systematic review. The vast majority of included articles are analyses or commentaries, with a small number of important exceptions, including a systematic review. Although discussion of drivers and solutions in many of the included analysis pieces were informed directly by empirical evidence, including systematic reviews the original studies are not included here, as they did not explicitly discuss drivers or solutions in ways captured by our search strategy. The body of empirical evidence highlighting the problem of overdiagnosis is growing, but a systematic review is beyond the scope of this article. Similarly, although analysing the complex inter-relation between overdiagnosis and the overuse it drives14 is vitally important, it is outside the reach of this article.
The map arising from our analysis is broad but not definitive—potential causes or solutions might not yet have been identified in the literature, and breadth might come at the cost of depth. In addition, our search was based in medicine, and a wider analysis might identify important sociological investigations of medicalisation15 resulting in different conceptions of the problem, drivers, and solutions. Importantly no strict or established criteria for what defines a driver or a solution exist, so our decisions about inclusion and mapping are open to discussion. Moreover, individual items could in some cases map to more than one domain. We have made no attempt to integrate this map with the extensive science of behaviour change16 or to specify potential actors to work on solutions, but hope that others will pursue this work.
What’s driving overdiagnosis?
Possible drivers of overdiagnosis span five domains: culture, the health system, industry, professionals, and patients and the public (table 1⇓; fig 2⇓). In this section, we offer a narrative summary of some important drivers that appear most commonly in the literature.
Popular deep seated beliefs that in healthcare “more is better” and “new is better” are often cited as drivers of unnecessary testing and overdiagnosis.34891718192021222324 Related to this is a strong collective faith in the benefits of screening the healthy and making an early diagnosis, arising in part from our fears of a serious disease being missed or a diagnosis made too late.3913192425272829 As Welch, Schwartz, and Woloshin argue in their 2011 book Overdiagnosed, which draws on a wealth of empirical evidence, “early diagnosis is a double edged sword,” with the potential to help but also hidden danger: “the detection of abnormalities that are not destined to ever bother us.”2 Fears of uncertainty, ageing, death, and disease also collectively contribute to this culture of too much medicine.8913171920222324252829303147
Health system drivers
Expanding disease definitions, which identify more previously healthy people as “sick,” are commonly cited as a driver of overdiagnosis.38911171920222728323347 Health professionals and hospitals frequently have financial incentives to perform more investigations or treatments for their patients, favouring increased and sometimes unnecessary care. Moreover, a system based on fee for service may lead to time restraints during consultations with inadequate time available for shared decision making or the complex explanation of the counterintuitive problem of overdiagnosis.489101112131718212223283032 As Malvinder Parmar said, “the current fee-for-service system does not compensate for a comprehensive explanation to the patient why a test is not required.”9 Current quality measures in health systems may lack emphasis on preventing overdiagnosis or overuse and instead may indirectly promote these problems.3491721
Industry and technology drivers
The most important driver in this domain is the use and promotion (to clinicians and the public) of increasingly sensitive tests, leading to detection—often incidentally—of minor “abnormalities,” which may be of uncertain clinical significance and can cause overdiagnosis.3891113171920222324272829323347 Industry promotion can also include the funding of patient and advocacy groups.8172328 As Eric Coon and colleagues point out in their well reasoned and evidence based exploration of potential drivers of overdiagnosis among children, “Advertisements capitalize on our fear of undiagnosed disease and urge us to see our doctor for testing . . . Once considered unbiased, third party advocacy groups are often used to deliver the same message.”17 Commercial imperatives and conflicts of interest, including financial or reputational conflicts of interests of those involved in guideline panels that expand disease definition, are also cited as a concern.3891317192125273247
Many authors argue that health professionals are driven to practise defensive medicine owing to their fear of litigation arising from a purported omission.34891113171920212223272829303241 Closely related is the doctor’s fear of missing a diagnosis, also commonly cited as a potential driver of overdiagnosis.34911121922232829 Health professionals’ unease with dealing with an uncertain diagnosis may lead them towards overtesting and overdiagnosis. This lack of professional confidence and knowledge of harms,891317192330 as well as the tendency to routinely diagnose or “do something” may arise from flaws in medical training,9132123 with underemphasis on patient preferences and overemphasis on diagnosis.202347
Patient and public drivers
While important, the results of our analysis indicate that this domain has received less attention in the literature, although it clearly overlaps with the culture domain. A number of authors point to a perception that many people have a lack of knowledge about the limits to, and harms of, medicine82347 and suggest that patients tend to over-rely on tests, including as a means of reassurance.8911131719202328 Others identify patient expectations that clinicians will “do something” as a potential driver.91321
What are the potential solutions?
Many of the potential solutions commonly identified in the literature map closely to explicit drivers, with some important exceptions (table 1⇑). For some drivers, such as the increasing complexity and fragmentation of care, specific relevant solutions were not identified. Other drivers showed considerable overlap, both within and across domains. The health system domain, for example, overlaps with the industry and technology domain, where enhanced government regulation of commercial promotion or health technology evaluation clearly falls primarily to policy makers. We made every attempt to link drivers to potential solutions.
Public awareness and education campaigns are needed to challenge beliefs that in healthcare “more is better” 912131722242526 and to promote a more healthy scepticism about the benefits and potential harms of early diagnosis.2361322 Arguably, initiatives like The BMJ’s Too Much Medicine campaign,48 Choosing Wisely,49JAMA Internal Medicine’s “Less is more,”50 and Health News Review51 are moving in this direction. Given the powerful role that media can play in shaping public beliefs, strategies to improve media reporting on overdiagnosis are needed.21328
Health system solutions
Reforming incentives for professionals and healthcare organisations to reward the quality rather than quantity of care is commonly cited as a key way to tackle the problem of too much medicine.3891317212230 Some authors also cite the need for new evidence informed frameworks to be used when disease definitions are changed,52 with calls for changes to disease terminology and new expert panels that are more widely representative and have reduced or minimal conflicts of interests.3112627283234353637 An influential group convened by the US National Cancer Institute is among those advocating changes to disease terminology for indolent lesions.26 Quality indicators and guidelines are also targeted for reform, to tackle any incentives for medical excess, as well as include new measures of overdiagnosis and overuse.492238 More targeted screening programmes that might, for example, limit some screening to well defined high risk populations9112635363739 and mandated strategies to inform patients of the benefits and harms of screening3111336 are among potential solutions for minimising the risks of overdiagnosis associated with screening. The 2016 systematic review of studies aimed at reducing low value care and underuse across different parts of the health system found that interventions using multiple strategies and targeting the roles of both clinicians and consumers had the greatest potential.10
Industry and technology solutions
More rigorous evaluation of the effects of both new and existing diagnostic technology on health outcomes is commonly recommended31217232427303439 as a key solution to the problem of increasingly sensitive tests that detect “abnormalities” of uncertain clinical significance. Drawing from the field of ecological economics to frame overdiagnosis as overconsumption , Hensher and colleagues call for “a more rigorous application of the precautionary principle” in technology assessment to avoid giving “potentially harmful overuse the benefit of the doubt.”12 Other potential solutions include stronger regulation of the advertising of new tests and treatments to the public and health professionals30 and paying greater attention to managing and reducing conflicts of interest with industry.47
The need to tackle the medicolegal concerns regarding missing or delaying a diagnosis was one of the key solutions discussed in the literature.39132247 Another recommended solution is updating current medical curriculums and continuing medical education to include overdiagnosis and overuse, for both students and practitioners.39101317212527304247 As future practitioners, students must be taught to “look always for the possibility of harm alongside that of benefit.”25 This is vital because, according to Eric Coon and colleagues, “if physicians are not aware of the potential harms of overdiagnosis, patients and families cannot be expected to appreciate them either.”17
Patients and the public solutions
Widespread awareness campaigns to inform and educate patients and the public on harms as well as benefits of screening and treatment options are commonly cited as essential to tackling overdiagnosis,389101321242728303638424647 echoing and overlapping with solutions we have classified in the cultural domain. Another frequently recommended solution was promoting shared decision making as a response to several key drivers in this domain.131721222328304647 In addition, several authors proposed the need for prioritising treatment options such as watchful waiting or active surveillance, where appropriate.31121262835
Where to from here?
We have attempted to retrieve, analyse, and summarise the existing literature on drivers and responses to overdiagnosis and related overuse. Many authors have identified a wide range of potential solutions across five inter-related domains of culture, the health system, industry and technology, professionals, and patients, with the ultimate shared aims to prevent harm, reduce waste, and redirect resources to treating and preventing unmet need in healthcare. The results of this analysis emphasise the need for more evidence about the problem, increased evaluation of potential solutions, and enhanced education across all sectors, to help wind back the harms of too much medicine effectively, safely, and fairly.
As part of multiple level strategies, in our view the most urgent need is to generate accessible evidence based information and educational materials about overdiagnosis for the public, professionals, and decision makers—both general information and condition specific. Tackling the gamut of financial incentives that drive unnecessary diagnoses and strengthening regulatory processes to enhance evaluation of new and existing diagnostic technology are two more solutions, as difficult as they are desirable. Reforming inappropriately widened disease definitions is arguably the most challenging but most important solution.
We strongly encourage critical responses to this analysis, offering alternative interpretations or missed drivers or solutions. Indeed optimising societies’ responses to medical excess will require broader thinking and analysis outside medicine—from such places as economics and sociology.
As the evidence base around this problem continues to grow, so do attempts to translate that evidence into action. In Norway, a position paper from the College of General Practice calls for action53; in Australia, major influential professional and consumer organisations are launching a national plan; and in Canada, the Quebec Medical Association is already implementing a province-wide strategy. Building on existing initiatives, we hope this analysis will help offer a suite of possible solutions to those seeking to reduce iatrogenic harm and enhance health system sustainability.
Contributors and sources: TP is a medical doctor and PhD candidate, undertaking her doctorate about overdiagnosis of prostate cancer at the Centre for Research in Evidence-Based Practice (CREBP) at Bond University’s Faculty of Health Sciences and Medicine, Australia. JC is a senior research information specialist with extensive experience working on systematic searches including for systematic reviews, based at CREBP. JC is also information specialist with the acute respiratory infections group within the Cochrane Collaboration. RM is senior research fellow, and National Health and Medical Research Council early career fellow, based at CREBP, researching overdiagnosis. RM has a background in medical reporting, including as a columnist with The BMJ, and gained his PhD titled Preventing Overdiagnosis in 2015. The sources of information for the paper, and the accompanying tables, are all listed as references and the methods used are explicitly listed in the supplementary file. RM is guarantor for the paper.
Competing interests: RM is co-chair of the scientific committee for the Preventing Overdiagnosis conference, which is supported by the BMJ. TP and JC declare no competing interests.
Provenance and peer review: Commissioned; externally peer reviewed.