Tipping the scale of resources : de-implementation of low-value care from an operant perspective

Background: The overarching goal within the field of implementation science is to generate knowledge that can contribute to bridging the gap between research and practice. Most studies focus on how to implement research findings, with the aim of using evidence-based interventions. However, it has increasingly been recognized that working in accordance with evidence not only requires implementation of research findings but also the “opposite,” i.e., de-implementation of so called low-value care (LVC). LVC makes up between 11 and 30 percent of all care provided, depending on type of LVC and study population. To address the issue of LVC, several guidelines have been published – but this does not seem to be sufficient to influence its use. Like implementation, deimplementation of LVC requires behavior changes among professionals within health care. The difference is that whereas implementation most often involves increasing certain behaviors, de-implementation involves both decreasing and increasing behaviors. In implementation, strategies are designed by identifying factors influencing behavior, identifying theoretically or empirically validated change methods to address those factors, and developing or choosing strategies that use those methods. However, it is not known what factors influence use and de-implementation of LVC. It is also not known if the same theories, models, and frameworks are relevant for de-implementation as for implementation or what de-implementation strategies are effective. Applied behavior analysis (ABA) is the only theory within psychology and sociology that discriminates between processes for increasing and decreasing behavior suggesting that this could be a valuable theory to use to understand factors influencing the use of LVC and to design de-implementation strategies. The overarching aim of the thesis was to generate new knowledge and insights concerning use and de-implementation of LVC. To achieve this aim, the four studies of the thesis have had the following objectives: -To identify factors that influence use and de-implementation of LVC (Study I). -To understand why physicians in primary care use LVC (Study II). -To understand which management strategies are being used to de-implement LVC and possible mechanisms for those strategies using concepts from ABA (Study III). -To demonstrate how ABA can be used to understand contingencies related to use of LVC and how de-implementation strategies can be developed by arranging alternative contingencies (Study IV). Method: Four studies were conducted: one scoping review, one qualitative study with physicians within primary care, using a grounded theory approach, one qualitative study with managers and key stakeholders within primary care on management strategies for deimplementation, and one intervention study where two strategies for de-implementation were developed based on applied behavior analysis to reduce use of unnecessary X-ray examinations for knee arthrosis. Results: The scoping review showed factors influencing the use and de-implementation of LVC related to both the outer and the inner context, the professionals, the LVC itself, the process of de-implementation, and the patients and their relatives. The qualitative study showed three factors that influenced use of LVC: uncertainty and disagreement about what not to do, perceived pressure from others, and a desire to do something for the patients. The qualitative study on management strategies showed eight different management strategies: financial systems, scorecards, quality assurance systems, guidelines, lectures, local process strategies, discussions about guidelines, and local lectures. The intervention study provided an analysis of factors influencing the unnecessary use of X-ray examinations for knee arthrosis: a rule stating that X-ray examinations are beneficial for diagnosing arthrosis and patients expressing expectations of being referred to an X-ray examination and showing appreciation for being referred for one. Two strategies were developed: A lecture aiming at introducing a new rule stating that X-ray examinations are not beneficial for diagnosing arthrosis and feedback meetings providing consequences encouraging diagnosis of arthrosis without the use of an X-ray examination. The strategies were perceived as helpful by the physicians who participated in the study. Conclusion: This thesis has provided knowledge about factors that influenced use of LVC as well as an understanding of how strategies for de-implementation could be developed. Factors external to the health care organizations seem to create a demand for LVC, mostly inadvertently through financial conditions that provide payment or reduce costs in relation to LVC. Factors within health care organizations, such as lack of continuity and standard ordering sets for laboratory tests, can also influence use of LVC, as can factors in the immediate environment of the individual health care professionals, such as problems with guidelines, pressure from others, and a desire to do something for patients. Thus far, de-implementation strategies developed at a local level seem to have the greatest potential to influence use of LVC. These strategies can be better adapted to local contextual factors. One way of doing so is by using ABA to understand local contextual factors or in ABA terms – contingencies. Strategies that influence processes, such as improved continuity or removing unnecessary laboratory tests from standard ordering sets, also have the potential to reduce use of LVC. Lastly, there is a lack of strategies involving factors external to the health care organizations, even though these factors influence use of LVC