Healthcare across the world pushes at the limits of its affordability. Driven by the development of newer and usually more expensive treatments, financial incentives that tend to reward providers for doing more, and consumerism and entitlement, and lower value care have grown alongside essential services. Market-based health systems, such as in the USA, have grown fastest while most state-funded systems have required above-inflation uplifts to keep pace with demand. The National Health System (NHS) in England has been warned that sustainability cannot be achieved through traditional methods of funding and management: services must be provided in new ways.1 Otherwise, austerity means cuts, planned or unplanned.

In this context, the notion that 20%–30% of healthcare is unnecessary and/or harmful2 offers an attractive and intuitively simple solution. Stop providing low-value care and important care will be more affordable. Picking this low-hanging fruit is simple in concept and morally defensible, but it will not be easy to achieve. It will require an unprecedented level of change.

The complexity of any large-scale change can be quantified in three dimensions:3 pervasiveness (how much of the system does it affect?), depth (how different is the new model to current ways of thinking and doing?) and size (how widely spread is the change across geographical boundaries, organisations or distinct groups of people).

The examples below illustrate that complexity in practice.

Pervasiveness can be deceptive

A study of low-value tests and surgical procedures among US Medicare patients4 identified low-value care for 42% of beneficiaries. However, this attractively large proportion accounted for just 2.7% of overall spending. Moreover, these estimates reflected ‘sensitive measures’ of overuse (ie, ones unlikely to miss overuse, but likely to include some instances of appropriate use). With more specific measures, the portion of patients receiving low-value care fell to 25%, accounting for only 0.6% of total spending.

A separate study of Medicare patients looked specifically for evidence of overuse of cardiac tests.5 The authors found that each year from 2006 to 2011, about 13% of low-risk Medicare beneficiaries received cardiac tests that were probably not indicated (because these were low-risk patients). But, the majority of non-indicated tests consisted of ECG, the least expensive of cardiac tests. In other words, this careful analysis of national US data did not identify a target of clearly unindicated use of more expensive tests such as echocardiograms and more advanced cardiac imaging.

Low-value care still constitutes a pervasive problem in total. But, for any given example (various investigations and procedures in the above examples), the clearly inappropriate cases may account for a small portion of spending.

Depth: apparently easy targets can require substantial culture change

It is often said that junior doctors overorder tests. They do so when they admit patients, when they receive minimal support from seniors, more expert staff and when they anticipate a ward round, or clinic led, by a supervisor who may want various test results. They learn that they receive praise for considering all possible diagnoses and castigation for the opposite. Successfully changing this behaviour requires addressing the culture that gives rise to it. But that is no easy task—culture change rarely is—especially when it runs counter to perceived positive values, such as demonstrating breadth of knowledge.

Another example of a deceptively deep target for reducing low-value care is screening for prostate cancer using the prostate specific antigen (PSA) test. The American Academy of Family Physicians sensibly recommends that PSA testing should not be offered unless the physician is willing to support the patient in a discussion of benefits and risks.6 Because of the unfavourable risk–benefit trade-off, the American Society of Clinical Oncology specifically recommends against PSA testing in men with no symptoms and a life expectancy of less than 10 years.7 We do not know how many tests are actually ordered in these circumstances. But, let us assume that it represents a worthwhile change in effort. Suppose a man comes to his family doctor after seeing campaign material promoting PSA testing or because friends or family have recommended it.

The ‘depth’ of the change relates to what is now required in that conversation. How does the doctor broach the subject of life expectancy when the man just wanted a simple PSA test? Probably, those men who already know they have a limited life expectancy are not the ones asking for the test. So, responding to this request for a simple test turns into a conversation with a not-particularly-ill 75-year-old, in which the physician informs him he is unlikely to make it to 85 on the basis of his various chronic illnesses. This may well be an important conversation to have with the patient—‘Your heart failure has reached the point that roughly 50% of patients in your situation will not be alive in one year’. But, it is hardly the easy conversation that either patient or doctor would expect to have as a result of a request for PSA testing.

Size: the complexity of implementation at scale

The Medicare study of cardiac test usage5 also illustrated regional variation of overuse (6.6%–23.6%). These differences suggest that the relative effect of drivers like doctors’ practice and patient demand are determined locally and that change initiatives designed at scale will not succeed.

Other barriers to change

In addition to considerations about pervasiveness, depth and size, calls for reducing low-value care often ignore other barriers to change. First, we lack data. Healthcare activity is not generally expressed in terms of benefit per cost; so, it is not easy to be sure where value-based priorities lie. This also creates problems in gauging the scope of any proposed change. As the PSA example above suggests, we often do not have the data to understand potential impact. Is the effort worthwhile, and what results should we expect? We also need to recognise that people—patients—may not want to forego low-value care. Healthcare is traditionally an entitlement or a privilege. More care is better. Government-designed targets, especially in the UK, have driven this perception through their preoccupation with access. Arguments about low value will need to apply to ‘my’ caesarean, ‘my’ PSA test in ‘my’ setting. Shared decision-making tools show promise, but their use will need to become commonplace.8

Finally, healthcare professionals will need to drive change. The admirable ‘Choosing Wisely Campaign’ 9 has driven discussion and consensus among clinicians and, in the USA, resulted in agreed lists of low-value care, although perhaps predictably, some professionals are better at identifying other people's opportunities for change rather than affecting their own practice or income. (eg, the previously mentioned example of avoiding PSA tests in patients expected to live less than 10 more years comes from oncologists, but screening for prostate cancer is usually conducted by primary care physicians.) Owen Dyer has reviewed Choosing Wisely from the healthcare professional's view: perverse incentives currently drive overprovision, including targets for clinicians and what might be called inertia.10 How many clinicians will drive a process that sees their expertise less valued, their workload reduced and their livelihoods threatened? Who in an orthopaedic pathway, for example, would currently make the case for non-surgical intervention? The answer may be dependent on context. Some surgeons may be quite happy to drop unnecessary procedures where they already have a high workload and waiting lists for services, or when they anticipate poor patient outcomes, but these factors likely vary geographically.

None of these problems is insurmountable, but they illustrate the complexity of applying a simple principle in practice. Success will require planned and coordinated change right across the system. Changing incentives, information systems and the shape of services will all be needed. So, will major cultural shifts. If low-value care is to be eliminated then value rather than cost, profit, affordability or entitlement must be the accepted currency in which healthcare is judged by its providers and users.

There are some examples of success

Nuka Healthcare in Alaska offers an interesting case example11 that is often cited as a success story to emulate efforts to reduce low-value care.12 In a relatively short time, Nuka has achieved radical reductions in hospital lengths of stay and use of emergency care, healthier lifestyles and better outcomes. A cultural change, placing the value of care at its centre, has been enabled by a shift to holistic and preventative measures, reskilling the workforce and excellent data. What Nuka has achieved for its 60 000 population is indeed remarkable. But, replicating it will require achieving changes in attitudes towards the purposes of healthcare (achieving greater health, not necessarily specific services), which is a worthwhile goal, but hardly the low-hanging fruit that many imagine when proposing to eliminate unnecessary care.

A recent well-conducted, randomised study of a multifaceted intervention to reduce caesarean section rates in Quebec13 provides a type of success in a more familiar healthcare setting, namely hospitals. The study produced an overall reduction from 22.5% to 21.8% in the intervention group with a statistically significant fall among women with low-risk pregnancies and no difference in the high-risk group. The intervention included education, training, audit and feedback. It seems fairly feasible to replicate. But, 0.7% is a very small decrease. Granted, a small reduction in caesarean sections will compound over time, since previous section is the biggest determinant of future surgery. But, for now, we will fall far short of the WHO's call for a target caesarean section rate of 10%–15%14 and fail to reduce avoidable risk.15

An earlier North American uncontrolled study16 looked at a home delivery model for 5418 low-risk births, and showed much lower rates of caesarean section. Less than 4% of deliveries involved caesarean delivery, and birth outcomes were comparable with births in a hospital setting. But the same study also showed that change requires concerted effort to provide advice, facilities and reassurance, which shift the outcome of these individual encounters. The authors concluded that while birth at home with midwives is associated with a lower intervention rate, this is a complex target to achieve. That does not mean we should avoid pursuing this goal. It just means that eliminating low-value care requires substantial investments in multifaceted change strategies that include addressing deeply held beliefs and attitudes towards healthcare. It would be a major feat to say the least to achieve widespread acceptance of home birth in developed countries.

Health versus healthcare

We know that adoption of healthy lifestyles can postpone death and disease,17 but invest increasing amounts in drug and other treatment for their risk factors. We have also contributed to what Atul Gawande recently described as overkill18: overdiagnosis, overordering diagnostic tests and then generating further testing and excessive treatment for mild conditions.

Healthcare should maximise benefit and minimise avoidable harm. Decisions about how best to do this are best made by patients and carers together to reflect their individual circumstances. This is why campaigns like Choosing Wisely, targeted at patients and physicians, represent a welcome advance over previous unilateral efforts by payers to limit care. But, even with greater engagement of patients and healthcare providers, achieving real change will require all parties—providers, payers, patients—recognising the complexity of reducing ‘low-value care’. Success will require re-envisioning the goals of healthcare and new ways of working together to achieve those goals. The problem with eliminating low-value healthcare is not, therefore, that it is not worth doing. The problem is that it will be very hard, and success will likely come in small-to-modest increments due to continuous, effortful quality improvement rather, and not the easy picking of low-hanging fruit.