A conceptual framework for developing de-implementation strategies based on un-learning and substitution

Au DH, Zeliadt SB, Winchester D. A conceptual framework for developing de-implementation strategies based on un-learning and substitution. Poster session presented at: AcademyHealth Annual Interest Groups Meeting; 2015 Dec 14; Washington, DC.

Background The goal of de-implementing harmful or ineffective clinical practices has become an increasingly prominent focus of health services research, as evidenced by the Choosing Wisely campaign. While de-implementation is not a completely new idea, relatively little is known about how best to accomplish this goal. De-implementation may differ in important ways from implementation, and likely entails greater potential for unintended consequences. The harmful practices we wish to de-implement often have deep historical, economic, political and social roots. As a result, de-implementation efforts might be perceived by patients and especially clinicians as an unwelcome intrusion on their prerogative and a loss of freedom, and might provoke strong resistance. Methods We propose a planned action model in which de-implementation of ineffective clinical practices results from two distinct, potentially synergistic processes: 1) a process of unlearning; and 2) a process of substitution. Findings We define de-implementation as abandoning an existing clinical practice because evidence emerges that the practice is ineffective or harmful, even in the absence of a specific superior alternative. Unlearning is an active process in which clinicians consciously change their knowledge, beliefs and intentions about the ineffective practice, and alter their behavior accordingly. There is a spectrum from simple unlearning, in which abandoning the ineffective practice fits within existing mental models, to deep unlearning, which requires adopting new mental models. Substitution involves the promotion of one or more alternatives to the ineffective practice, in which the substitute practice either precludes the ineffective practice (e.g., watchful waiting for prostate cancer), or makes it less likely to occur (e.g., referring back pain to physical therapy rather than a surgeon). We expect to find that each approach can work independently, and that in cases they may work synergistically. We also expect that attributes of the practice targeted for de-implementation, such as the clinical setting, will help determine which approach will work best. Implications for D and I Research This planned-action model proposes two distinct de-implementation processes, unlearning and substitution, each potentially effective for different clinical practices and settings. De-implementation strategies based on unlearning and substitution can be empirically tested and compared to increase our knowledge about which strategies are most effective in which contexts.