Low-value care is the approximately 30 percent of all diagnostic and therapeutic health care that provides no demonstrable benefit or where the risks of known harms outweigh the expected benefits. One key aspect of addressing low-value care entails de-implementation—the reduction or elimination of a health care service that may be ineffective or inappropriate. For example, in the past 20 years, research evidence and clinical guidelines have radically changed related to hormone replacement therapy for prevention of cardiovascular events for women entering menopause, percutaneous coronary intervention for treatment of ischemic heart disease, slow-release opioids for chronic pain, and supplemental oxygen for patients with chronic obstructive pulmonary disease.
We argue that de-implementation research entails unique challenges from implementation, and that how we approach and communicate about those challenges matter. In this blog series, we make three arguments. First, as a research community we should communicate about de-implementation as a routine, expected process in a learning health care system, not an anomaly, and we need to study how best to do that. Second, de-implementation can occur at and have different drivers at different levels—from the patient, to the provider, to health care organizations. Finally, de-implementation entails the potential for unintended consequences, and as researchers we need to anticipate and measure unintended consequences. These latter two points deserve consideration together, as the drivers of low-value care and levers of de-implementation entail different potential unintended consequences at different levels.
De-implementation needs to be a routine, expected process
In some instances, we know low-value care results from clinical tradition that is never questioned or tested, from faulty research that is eventually corrected, and other times from sensational research findings that receive inappropriate attention before being properly vetted and corroborated. But low-value care, at least in part, can be understood as the inevitable result of the evolution of science. Scientific “facts” display a very predictable half-life in which a certain proportion of findings or accepted truths expire over time. Some of what we think to be true today, we will understand to be incorrect tomorrow. While we cannot predict which clinical practices will ultimately be overturned or supplanted, we can be certain that some clinical practices will require de-implementation or abandonment. Consequently, even if we were to eliminate all low-value care today, given enough time we will inherit a new set of low-value practices. This means de-implementation is not a discreet, temporary process, but is something that will always be necessary.
This has several implications for how health services research approach studies of overuse and de-implementation, and for how we communicate our work and findings. It is useful for us as a research community to use the label of de-implementation to identify this work to each other and help advance this body of knowledge. At the same time, the term de-implementation is not useful when we communicate our work to our clinical and community stakeholders, and in fact is harmful because at best it is confusing and at worst perpetuates a paradigm of medical overuse that conflates it with patient safety and quality assurance when instead medical overuse is equally the result of the natural evolution of scientific knowledge. For purposes of communicating with the community, we will be better served by framing de-implementation research as part of the effort to become a learning health care system that is constantly evaluating its performance against the evolving body of knowledge about how to achieve the best possible care. But how best to do that is still unclear—which settings, media and language are most effective in communicating with patients, practicing clinicians and policy makers. This is an important subject of research even as we develop and test interventions and policies to reduce low-value care.
In the next part of this blog series, we examine how low-value care may look different—and de-implementation efforts may have different effects—for different stakeholders at different levels. This series will explore the implications and challenges of de-implementation at the patient, provider, and systems levels respectively, including anticipating and measuring unintended consequences.
This four-part blog series is a product of a thematic working group on de-implementation convened by AcademyHealth and the ABIM Foundation as part of the Research Community on Low-Value Care. A special acknowledgement goes to Cheryl Boyce, Branch Chief at the National Heart, Lung, and Blood Institute for her participation in the thematic working group and for her contribution to this work.