Clear articulation of the problem, and the factors driving it

Overdiagnosis and overtreatment are acknowledged to be complex and not necessarily easy to define.1 Many likely contributing factors have been identified, including: increasingly sensitive tests that identify indolent, non-progressive or regressive abnormalities; expanded disease definitions and lowered thresholds for intervention; creation of ‘pseudo diseases’; public enthusiasm for screening or testing and the desire for reassurance15; clinicians’ fears of missing a diagnosis or of being held to account for decisions not to act16; and financial incentives.17 In an effort to develop a comprehensive overview, recent work from Australia has sought to map these possible drivers and link them to their potential solutions.18 This mapping suggests drivers of overdiagnosis and overtreatment fall into one of five key domains: culture (eg, an intolerance of uncertainty, or a belief that prevention is better than cure); health system (eg, quality measures that encourage doing more, or financial incentives for tests, treatments and diagnoses); industry and technology (eg, industry promotion of tests and treatments to both professionals and public, or diagnostic tests with increased sensitivity); professional (eg, fear of litigation and associated defensive medicine, or lack of knowledge of potential harms); and patient and public (eg, reliance on tests for reassurance, or lack of understanding of potential harms). Of note though, is that the vast majority of articles included in this mapping exercise were analyses or commentaries rather than empirical studies (although the authors state that many of the analysis pieces were informed by empirical work). Therefore, while there is no doubt much merit in the drivers proposed, further empirical work to really understand whether and importantly how these work in practice, and how they may be interconnected and mutually reinforcing, is still needed. Overdiagnosis and overtreatment are terms that tend to be used rather imprecisely, meaning different things to different people, and in different contexts.1 Given this complexity, it is unlikely that the same drivers will be relevant in all cases or that they will function in the same ways even if they are.

Careful, theory-informed design of interventions

There is growing recognition within healthcare improvement research of the gains to be had from ensuing that the design and execution of improvement efforts are appropriately theoretically informed, and that while a more intuitive approach to solution design may be effective in some cases, this is often biased, distorted and limited in scope.19 Articulating a clear theory of change, an account of how and why the planned activities are credible as a means through which the desired outcomes can be achieved, is an important early step in improvement work. A good example of this is Leviton et al’s thoughtful approach to increasing the use of antenatal corticosteroids for fetal maturation.20 Here, an ‘active, focused dissemination effort’ on the benefits of using corticosteroids drew directly on research about adoption of medical practices and significantly increased the odds of corticosteroid use. Engaging in this form of interrogation is important for, among other things, making the necessary shift from implicit to explicit thinking about how interventions are supposed to work and, relatedly, surfacing any weakness or incoherence in the proposed causal links.19

Achieving clarity about and shared understanding of the outcomes sought is an important part of intervention design. In the case of tackling overdiagnosis and overtreatment the outcomes sought may vary. One objective may be to seek to reduce the use of particular tests or procedures which are regarded as in some way unnecessary in at least some cases. The number of tests or procedures carried out is likely to be quite straightforward to measure, but likely to be less clear is whether and how it is possible to discern which are truly not necessary. Another objective may be to reduce the rates of harms such as invasive treatment of anomalies that were unlikely to cause problems, but again being able to identify to which cases this applies is not straightforward. Another objective may be to facilitate more informed decision-making on the part of patients about whether they wish to participate in interventions which may bring harms as well as possible benefits, such as participation in screening or taking preventive medication such as statins, or to achieve more meaningful use of approaches such as shared decision-making in these types of cases. The challenges here are possible differences of opinion about what constitutes a harm or a benefit, and how to most appropriately trade these off against each other.1

Numerous possible approaches to tackling the problems of overdiagnosis and overtreatment are starting to emerge, including the development of decision aids for people invited to take part in population-based screening programmes,21 shared decision-making as a means through which to ensure that decisions about both diagnosis and any potential treatment are person centred,22 awareness raising campaigns and educational programmes,23 and changes to financial and other incentives that are embedded within healthcare systems and which influence healthcare professionals’ behaviour.18 Some initiatives encompass several of these strategies and may be understood as a form of complex intervention. The Choosing Wisely initiative is one such example, combining the development of lists of tests and procedures that providers and patients should question, materials to help patients engage with healthcare professionals and ask questions,24 and educational resources for healthcare professionals to facilitate their engaging more effectively with patients.22 While appealing in principle, efforts to drive change through the facilitation of such ‘bottom up’ approaches can be deceptively difficult to achieve in practice.25

Ensure robust evaluation of efforts to tackle the problem

The importance of evidence-based improvement is becoming ever clearer, with healthcare improvement research all too often showing that efforts to bring about change are either not effective at all or that those that appear effective in one place either cannot be sustained or fail to reproduce the same impact elsewhere. While overdiagnosis and overtreatment may be worthy problems to seek to address, robust evaluation of the impact of having done so is important. One reason for this is that the reward may turn out to be smaller than anticipated,26 with some studies of the likely gains to be made through targeting so-called ‘low value’ care raising questions about the magnitude of any likely impact.27 28 A second reason that robust evaluation is needed is that these issues may be trickier to gain traction on than might be anticipated. For example, Kullgren et al’s recent study of the impact of clinicians’ precommitting to follow Choosing Wisely recommendations with decision supports showed only a small, unsustained decrease in potentially low-value activity for one of three targeted conditions and a small increase in alternative orders.29 This is an interesting example as seeking to draw on the behavioural economics strategy of precommitment (committing one’s future self to a course of action) makes intuitive sense given its apparent success in changing other behaviours. Kullgren et al’s study provides sobering evidence of the apparent lack of effect in practice, however.

While efforts to develop integrated frameworks for evaluating the impact of some kinds of interventions are starting to be developed,30 these tend to focus on quantitative measurement of things like provider attitudes, patient experience and outcomes, and rates of use of different tests or procedures. What tends to figure less prominently, but which healthcare improvement research increasingly shows to be vital, is process evaluation using qualitative methods such as observations, interviews and documentary analysis.31 Such an approach allows the depth of exploration and analysis necessary in order to ensure a comprehensive understanding of how implementation has happened in practice, why and how processes give rise to particular consequences and why the same intervention may play out differently in different contexts.

Using such an approach to critically reflect on the underpinning theory of change for any intervention, and how this played out in practice, can be particularly useful. This is likely to be especially valuable for interventions seeking to achieve deimplementation rather than implementation,32 as professionals’ practice is often deeply influenced by historical, social, cultural and economic forces, requiring complex interventions that take account of context. Even with seemingly adequate theorisation of change mechanisms, efforts to improve care can still fall prey to the sometimes bewildering complexity of healthcare systems and their challenging contextual features.33 Returning to the example of precommitting to follow Choosing Wisely recommendations, while Kullgren et al’s study is valuable it does little to help us understand why the approach taken did not work in the ways hoped, what the flaws in the underpinning theory of change and/or the operationalisation of this in practice may have been and what strategies may be more successful.