Foregoing low‐value care: how much evidence is needed to change beliefs?

It's not always easy to let go of time-honoured practices which are shown to be ineffective or even harmful. A considerable literature has grown up around how to encourage clinicians to align their routine practice with robust scientific evidence.1 Much of this ‘implementation’ or ‘translation’ research has focused on explaining and overcoming underuse of effective interventions.2 The landmark CareTrack study of the quality of healthcare in Australia3 estimated that just over 40% of eligible patients do not receive evidence-based care. On the flip side, widespread use of ineffective interventions, which may consume up to 30% of healthcare budgets,4 has received much less attention at a time when healthcare systems are required, by fiscal constraints, to maximise health benefit for every dollar spent.5 Clinicians are often reluctant to relinquish established clinical practices despite seemingly compelling evidence to do so. For example, US studies showed no decrease in the use of percutaneous coronary intervention (PCI) to reopen totally occluded infarct-related arteries beyond 24 h after acute myocardial infarction despite publication of a large trial showing no benefit and release of revised guidelines.6 Similar experience has been noted in regards to PCI use in patients with stable, non-critical coronary artery disease.7 In many hospitals, protocols decree that batteries of preoperative tests, routine replacement of intravenous cannulae every few days and daily chest X-rays on all intensive care patients be performed, irrespective of clinical need – despite considerable evidence that more selective, less costly manoeuvres are equally safe and effective.8 Futile, aggressive care delivered to patients with advanced end-stage disease or terminal illness diminishes dignity and quality of life while consuming resources. Potential waste and possible harm also occur when an intervention proven to be effective within a very specific population is inappropriately applied to a wider spectrum of patients for whom benefit has never been demonstrated (indication creep). More than one in five implantable cardioverter-defibrillators in the US are inserted for off-label indications for which benefit has not been evaluated.9 To healthcare funders and policymakers, clinician resistance to foregoing interventions that robust evidence reveals are of no benefit, or even harmful, is perplexing, to say the least. The question they may well ask is: how much evidence do you need to change your mind? Clinicians will rightly argue that evidence has to be compelling before they disown practices that have stood the test of time and, in their personal experience at least, appear safe and to have done some good. However, what constitutes ‘compelling’ evidence is very much in the eye of the beholder, although two themes predominate: epistemic factors around the trustworthiness of the research (objectivity, consistency, clinical plausibility and non-selective reporting of outcomes) and the likely benefit for individual patients (effect size, applicability of ‘average study patient’ treatment effects to specific subgroups or unselected real-world populations, time to effect, and value to patients of the outcomes measured).10 Still, the threshold of ‘compellingness’ that may serve as a ‘tipping point’ for a collective change in beliefs remains uncertain, at least in the medical world, and will likely differ from one clinical scenario to the next. Moreover, research evidence is not the only determinant of clinical decision-making. Various cognitive and non-cognitive factors may explain the tendency towards unnecessary care,6, 10-14 with many reflecting clinician desire to avoid potential injustice to individuals from withholding interventions that may possibly bestow some benefit (Table 1). However, while these factors are no doubt important, there are two more fundamental factors that may underpin resistance to change. Cognitive dissonance is the inability to reconcile new evidence with highly ingrained prior beliefs that both determine and are reinforced by routinised practice16 – one believes so one does, and as one does, so one believes. Such beliefs are highly personal and internalised based on individual experience, interpretation of past research, exposure to the views of respected peers, and socialisation into the norms and traditions (or culture) of one's chosen specialty. What we believe (and want to believe) is tightly bound to the central human need to belong to and seek comfort within, a group that shares similar values and outlook. Challenging current professional paradigms runs the risk of being cast out and ostracised from the group. Pioneering medical thinkers from Ignaz Semmelweis (septic technique to prevent puerperal sepsis) to Barry Marshall (eradication of Helicobacter pylori to prevent recurrent peptic ulcer) have had to confront and overcome collective cognitive dissonance. When research supports strongly held beliefs, clinicians more readily accept the conclusions – despite major methodological flaws – and use them to reinforce current practice17 or in some cases add to current practice, even if the evidence is far from definitive.18 In contrast, when research runs counter to strongly held beliefs, even multiple studies involving patients representative of everyday practice may not prove persuasive. All sorts of reasons, including flaws in study designs (both real and imagined) and limited applicability ‘to my patients’ may be cited to discredit the results.19 There are notable exceptions, chiefly reports that unequivocally show serious harm from commonly performed interventions, more so if patients get to know about it, raising the risk of medicolegal liability. There was an immediate reduction in the use of hormone replacement therapy in post-menopausal women20 and of prophylactic anti-arrhythmic agents following acute myocardial infarction,21 following trials showing increased risk of serious adverse events, information that was quickly disseminated in the lay press. Clinicians value their autonomy and being perceived as sound arbiters of medical knowledge. Having to accept evidence that runs contrary to one's beliefs and refutes what had been regarded as effective interventions can threaten one's sense of competence, professionalism and freedom to choose. This may incite a state of psychological reactance, a tendency to resist perceived attempts by others – especially those outside one's professional network – to control behaviour.22 Individuals can react in a way that affirms their ability to choose and often become more entrenched in their original beliefs. Once aroused, reactance may heighten sensitisation to additional threats to freedom of choice that further constrains the capacity for dispassionate debate. To date, psychological reactance of clinicians has been little studied in the medical literature with most reports pertaining to patient non-adherence with medical advice.23 Putting these cognitive biases to one side, there is no place for continued support of ineffective or harmful practices among clinical professions that base their practice and authority on good science. Recent investigations have disclosed more than 150 high-volume clinical services on the Medicare Benefits Schedule that are potentially of low-value, with almost half being ineffective or harmful on the basis of multiple trials and systematic reviews.8 Past experience in encouraging clinicians to align practice with best evidence suggest educational or awareness-raising strategies, clinical audits and feedback, academic detailing, and other professionally mediated interventions have limited impact in curtailing inappropriate care.1 Funders and policymakers increasingly view such activities as too slow and incremental. Consequently, they place greater faith in ‘forced function’ manoeuvres of which the first and foremost are financial incentives (or increasingly disincentives) in the belief that clinicians will no longer do things if they are not being paid, or paid as much, to do them. Value-based funding programmes and pay for performance schemes in the US and UK are seeking, rightly or wrongly, to reduce funding for publicly subsidised services deemed to be of no value and reallocate saved resources to higher value healthcare that confers greater benefit. So-called ‘bundled payment’ models, for example, provide clear disincentives for including low-value services in the bundle that attracts a set price for an entire care package. Other strategies include public reporting of quality standards for hospitals (and in the future possibly individual units and consultants) that includeoveruse as well as underuse of care, restricting use of certain interventions to narrowly defined populations of eligible patients and highly competent providers in order to maximise benefit, and increased use of blocking systems that disallow the ordering or prescribing of selected tests or treatments. As stewards of tax-payer funds, Australian authorities will likely consider these approaches in the near future. While interventions that are clearly of low value represent low-hanging fruit in any programme aimed at reducing waste, more challenging are interventions ‘at the margin’ where genuine uncertainty exists as to who will and will not benefit from them. It is in no one's interest to reduce indiscriminately utilisation such that patients who clearly need specific interventions are denied them. If as clinicians, we wish to avoid being marginalised in debates around minimising use of ineffective or harmful interventions and reduce the risk of being governed by policy decree with its inherent clumsiness and potential for unintended consequences, we will have no choice but to define low-value care and answer the question ‘how much evidence is enough to change beliefs?’