The value of low-value lists.

AN INTERNATIONAL GROUNDSWELL OF ACTIVITY IS seeking to identify and reduce the use of health care services that provide little or no benefit— whether through overuse or misuse. There are strong imperatives for identifying such waste: (1) an ethical imperative to ensure patient safety and thus avoid tests and treatments that cause harm directly or indirectly without providing commensurate benefit; (2) a quality imperative to measure and reward best practices; and (3) an economic imperative to reduce spending and enhance the diffusion of cost-effective innovations. England’s National Institute for Health and Clinical Excellence (NICE) commenced a formal agenda in this area in 2005. The most recent initiative garnering attention is Choosing Wisely, a US campaign led by the ABIM Foundation. Other countries are implementing similar approaches. A major challenge faced by these initiatives has been how to identify and prioritize candidate services for consideration in a reasoned and transparent manner. Today, several lists compiled by prominent organizations have identified numerous services as potentially low value in certain clinical circumstances (eTable, available at http://www.jama.com). The challenge facing payers and health care service providers such as physicians and hospitals is to develop and implement strategies to reduce the use of services that are identified in these lists, many of which are discretionary, if not potentially harmful. The intent of the evidence-informed lists is to provide sets of specific services used in defined clinical scenarios that payers and health care professionals can target directly in rewarding value and limiting inappropriate care. As suggested by the lists, services that are ineffective, unsafe, or both for all patients andindicationsarerare.Typically,aservicedemonstratessafety and effectiveness profiles that depend on the characteristics of the population to whom it is provided. In essence, a service that is low value in some clinical circumstances might be high value in others. This clinical heterogeneity makes it difficult to develop simple approaches for identifying low-value services. For instance, although routine stress testing in asymptomatic patients is clearly of low value, stress tests can be very high value in those presenting with symptoms of ischemic heart disease. The main challenge is that interventions proven to be effective for specific clinical populations are often inappropriately applied to patients for whom benefit has never been demonstrated (indication or scope creep). In the United States in particular, extrapolation of evidence is encouraged by financial incentives embedded in physician payment systemsandcoveragedesignswith limitedcostsharingforpatients. Just as the development of low-value lists is beset with clinical complexity, so too is their implementation. Although evidence-based assessments of individual health services often focus on use in specific populations and indications, the presumption of detailed clinical data is often at odds with the nature of existing data sources such as administrative claims. For instance, imaging for acute back pain usually is considered of low value, but this may not be the case in certain cases such as trauma or evidence of neurologic compromise. Although this information might be present in medical records, it typically is not captured well in claims. These data deficiencies present fundamental obstacles to translating comparative effectiveness research into effective policies because often the lack of detailed information on the clinical context (ie, indications) limits the usefulness of claims data for identifying and measuring the use of these low-value services. One strategy to reduce waste is to deny coverage for wasteful services. Yet because of the aforementioned clinical heterogeneity that cannot always or easily be observed with current claims systems, the effectiveness of using coverage design to discourage use of low-value services is likely to be limited in scope. Similarly, value-based insurance design and related supply-side strategies (eg, not paying for never events) are fraught with measurement and data issues when applied to services of heterogeneous value. For example, developing benefit-based co-payments for automatic implantable cardiac defibrillators or for coronary revascularization procedures (higher co-payments for lower-value uses) would require the incorporation of complex and evolving guide-