Looking for value in health care

Ineffective or harmful clinical interventions and services are of no value to patients and waste limited health care resources. About 30% of United States health care expenditure reflects health services of uncertain value.1 More than half of established practice standards, when formally evaluated for effectiveness compared with alternatives, are found to be either less effective or of questionable benefit.2 Even practices deemed effective in trials conducted decades ago have been shown to be ineffective in more recent studies.3 Public and private payers, professional bodies and government agencies are now engaged in identifying “low-value” practices that should no longer be funded.4 The challenge in this is to develop evidence-based, transparent methods for undertaking this process, that are sensitive to the needs and concerns of patients and clinicians.

The report by Elshaug and colleagues in this issue of the Journal5 describes the initial results of an Australian project tasked to identify potential low-value non-drug items among the roughly 6000 listed on the Medicare Benefits Schedule (MBS), very few of which have been formally assessed against contemporary evidence of safety and effectiveness. The authors used a three-pronged approach to gather evidence in identifying such items: a broad peer-reviewed literature search of PubMed articles published between 2000 and 2010, using a detailed search strategy and selection criteria; analysis of databases from the Cochrane Library and organisations in the US, United Kingdom and Canada heavily involved in health technology assessment; and opportunistic sampling based on the authors’ own past experience, and that of clinical and non-clinical stakeholder groups in Australia and overseas. After applying various filters and choosing the most recent publications (because of time and resource constraints), the authors selected 177 articles from which they identified 156 low-value items. About half of these were treatments (many were operative procedures), and the rest were diagnostic, screening or monitoring investigations. While some items would be instantly recognised by seasoned clinicians as being obsolete (eg, adenoid removal in children with recurrent otitis media with effusion; hysterectomy as first-line treatment for heavy menorrhagia), opinions on the value of others will be divided (eg, exercise electrocardiography in patients with suspected angina; endovascular repair of infrarenal abdominal aortic aneurysms in medically fit patients).

One way of looking at the results is to divide the low-value items identified by Elshaug et al5 into three groups, as summarised below (full version in the Appendix).

Group A: Items associated with robust evidence of no benefit or of harm compared with “usual care” or an alternative intervention (comprising 50% of items). Examples include arthroscopic lavage or debridement for knee osteoarthritis, vertebroplasty for painful osteoporotic vertebral fractures, and upper airway surgery for obstructive sleep apnoea.

Group B: Items associated with equivocal evidence of benefit, requiring more research (comprising 42% of items). Examples include lumbar discectomy for disc herniation, intracavity lavage in potentially contaminated surgery and cystoscopy in uncomplicated lower urinary tract symptoms.

Group C: Items associated with both benefits and potential harms, such that their use depends on patient preferences (comprising 4% of items; the remaining 4% of items were unable to be classified into any group). Examples include transurethral resection for symptomatic benign prostatic obstruction and rubber band ligation for grade 3 haemorrhoids.

In prioritising items that should not be MBS funded, group A would go first, followed by group B. In either case, the removal of MBS funding can be reversed if and when unequivocal evidence of benefit and no serious harm becomes available. Group C could continue to be subsidised while risk–benefit trade-offs become better quantified in specific patient subgroups. Elshaug and colleagues invite more discussion with key stakeholders around the nuanced contexts in which specific interventions may be appropriate, before an all-embracing label of low value is applied.

Several possible criticisms of the study deserve mention. First, only 4% (1200 of 31 853) of retrieved systematic reviews and clinical studies were selected for appraisal, although the most recent (and likely most relevant) articles were included, and reports with Level 1 evidence6 were oversampled. Second, almost two-thirds of the articles selected comprised assessments from overseas agencies, and these may not be relevant in Australia (although this is unlikely). Third, the evidence cited for a few items, as of August 2010 when the search was performed, is already out of date. For example, more recent studies suggest carotid endarterectomy is not of low value compared with carotid artery stenting in preventing stroke (indeed the evidence trends in the opposite direction), even in high-risk patients.7 This inaccuracy emphasises the importance of continually reviewing and updating the status of items as new evidence emerges within rapidly developing areas of practice.

So, where to from here? First, the identification of low-value interventions and tests must involve Australian specialist colleges, as is happening in the US Choosing Wisely campaign.8 Concerted efforts are needed to understand and reverse continued use of unequivocally low-value interventions, such as arthroscopic debridement and lavage in knee osteoarthritis.9 Second, the Australian Government Department of Health and Ageing, in collaboration with the National Health and Medical Research Council (NHMRC), should resource and support groups of content experts and methodologists to investigate, in an ongoing fashion, the effectiveness of high-volume, high-cost MBS items. Evidence and decisions around such items need to transfer rapidly into regularly updated, NHMRC-approved clinical guidelines and into low-value items being delisted from the MBS. Finally, all stakeholders, including patient advocacy groups, must come to appreciate the harms and opportunity costs of ineffective interventions and be prepared to let go of them. The aim is not to reduce current health care expenditure or impose ad hoc rationing of services, but instead to reallocate limited resources to interventions of proven value.