Using theories and frameworks to understand how to reduce low-value healthcare: a scoping review

The reduction of low-value care is necessary to ensure patient safety, reduce costs, and promote a sustainable healthcare system [1, 2]. Patients can be harmed by low-value care, whether directly, by downstream effects, or by over-testing or over treatment [2, 3]. In addition to harm, it is estimated that 30% of current healthcare dollars, both in Canada and in the USA, are spent on harmful or wasteful practices [4, 5]. Low-value care can be tests, treatments, medications, or procedures in any healthcare setting (e.g., hospital, primary care, long-term care, or public health) which have been deemed, through evidence, to be ineffective, harmful, or unnecessary. Awareness has increased regarding the prevalence of low-value care in the healthcare system and the high frequency with which these practices are used.

In recent years, a number of initiatives have been created, such as the Choosing Wisely campaigns, which seek to identify, raise awareness of, and address the prevalence of low-value care. International Choosing Wisely campaigns, currently in 25 countries worldwide including Brazil, Wales, and Korea, have gained considerable traction with healthcare organizations and providers and have also been the subject of an increasing amount of health services research [6, 7]. The campaigns have identified prevalent practices, such as imaging for low back pain, benzodiazepines for insomnia in older adults, and pre-operative testing for low-risk surgical procedures, which provide evidence and guidance for healthcare providers and organizations endeavoring to reduce low-value care [8]. Researchers have noted that, as a result of the Choosing Wisely campaigns, efforts to identify and prioritize low-value care have increased exponentially but these efforts far exceed research to evaluate the efficacy of clinical and policy initiatives and interventions to reduce these practices [9, 10].

De-implementation—reducing or stopping the use of a health service or practice provided to patients by healthcare practitioners and healthcare delivery systems [11]—is an emerging area of study in healthcare. Numerous researchers have called for increased attention to understanding the process of de-implementation and determining the most effective and efficient strategies to create and sustain this type of practice change [9‐16]. Efforts to reduce low-value care can be hindered by practice patterns and habits that can drive the overuse of procedures, medications, and medical devices; overuse is often a consequence of routines or protocols that are embedded in systems which can be difficult to change or alter [17‐20]. The acknowledgement of these complexities has caused many to argue for the use of theory to understand de-implementation and to develop effective strategies to de-implement low-value care [10, 12, 18, 21, 22].

While de-implementation research, intervention design, and evaluation has just begun to use theory, more mature areas, such as implementation research, have recognized the value in using theory to design behavior change interventions [15, 23‐25]. Implementation and de-implementation research both endeavor to identify what interventions work, for whom and under what circumstances [10]. Grimshaw et al. state that implementation science theories and frameworks can support de-implementation as, they argue, the effectiveness of these theories is a function of an effective process rather than unidentified factors that may reduce intervention effectiveness [10]. Implementation science knowledge, theories, and frameworks can provide support and guidance for de-implementation efforts.

Research with underused theoretical perspectives makes it difficult to understand and explain how and why interventions succeed or fail, “thus restraining opportunities to identify factors that predict the likelihood of implementation success and develop better strategies to achieve more successful implementation” [[26]: 1]. Theory-based research should be used to develop a more complete understanding of factors which influence healthcare professional’s behavior, with the goal to better inform the design of behavior change interventions [27‐29]. Specifically, individuals’ decisions, primarily physicians’ decisions, are key to adopting a new practice; therefore, understanding the cognitive mechanisms underlying behaviors is critical to developing effective interventions [29].

Currently, the amount of empirical evidence on the use of theory in de-implementation is low, which limits knowledge about which specific factors influence de-implementation, the relevant barriers and facilitators, and which interventions are successful at reducing low-value care [22]. Therefore, there is a need for a systematic examination of the literature to identify and characterize the current body of research on the use of theory and frameworks in de-implementation. Through our review, we aimed to describe and synthesize the theories and frameworks, their use and application, and identify gaps in research. A recent review did much work to contribute to this literature, providing foundational analysis on the use of theories, models, and frameworks for the reduction of low-value care [15]. Our review builds on this work by identifying an additional 41 articles in which theory is used to understand the phenomenon of de-implementation; with the majority published in the last 2 years. As such, our scoping review adds breadth to the existing knowledge base, including a broad spectrum of theories and frameworks, providing comprehensive analysis and insight to this critical area of practice change.

We selected a scoping review as the most appropriate approach as de-implementation research is relatively new, and little is known about the use of theory and frameworks in this area. We used Arksey and O’Malley’s [30] methodology for scoping reviews with consideration of recent perspectives on the methodology [31] and the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Extension for Scoping Reviews (PRISMA-ScR) to guide the conduct and reporting of this review [32].

The database searches identified 1998 articles (after duplicates were removed) for which the titles and abstracts were screened for inclusion. Of these, 232 articles were selected for full-text screening and 48 articles [22, 34‐80] were included for this review. See Fig. 1 for the PRISMA diagram representing the complete article selection process.

These findings provide important insights into the emerging research area of de-implementation, and how theoretically based work is approaching this critical practice change. This review affords an up-to-date and comprehensive collection of theories and frameworks, and an analysis of how these theories and frameworks are used. This section offers a detailed discussion of the key findings, application, and gaps.

While the number of theory informed articles on this topic has steadily increased—over 60% of all included studies were published in the last 2.5 years—these numbers are only a small proportion of the studies published about reducing low-value care. This observation corroborates Niven et al.’s [12] finding that the majority of de-implementation efforts are likely atheoretical. Despite this low number, the included articles provide valuable insight into the maturing use of theory and frameworks in de-implementation and the dominant perspectives being used.

The TDF was the most commonly used framework in the included studies. Nilsen et al. noted that the use of TDF and TICD reinforces the importance of both individual and contextual determinants to de-implementation efforts [15] and results of the included studies align with this assertion. The majority of studies that used TDF identified individual—knowledge and beliefs about consequences—and contextual—environmental context and resources, and social influences determinants as key factors impacting de-implementation. The popularity of the TDF may be explained by the fact that it is useful for researchers in light of the myriad options available to them [10].

Our results also show a prevalence of psychological theories, particularly the Theory of Planned Behavior; this aligns with implementation science research where social cognitive theories have been the dominant theories used to understand clinical behavior change [26]. These classic theories, which typically only describe mechanisms and change but do not prescribe how to bring about change [15], are well established and have an abundance of empirical support in primary research. The prevalence of these theories also corresponds with the notion that it is the healthcare providers who ultimately prescribe or perform the low-value care and therefore the responsibility lies with the individual [21, 95]. In 2008, Godin et al. [29] published a systematic review that investigated the application of different social cognitive theories used to study healthcare professional’s behavior. Similar to our results, their review found that the TPB was the most commonly used theory, and the included studies reported between 25% and 34% of the variance in behavior was explained by the theory [29].

Nilsen et al. [96] detailed a novel perspective in their 2012 paper Creatures of habit: accounting for the role of habit in implementation research on clinical behaviour change. While this paper is not explicitly about de-implementation, the authors raise key questions about the utility of psychological theories, based on intention, to study behavior change, particularly in the context of stopping an existing practice. The authors acknowledged that social cognitive theories offer insight into analytical processes, but do not address repeated behaviors that do not require decision-making processes and noted that much of a physician’s daily work and interactions occur in unvaried settings and this behavior is guided more by habit than intention. These arguments also align with an observation made by Godin et al. [29] based on the findings of their systematic review of the use of social cognitive theories: that habit is a missing component to understanding healthcare professional’s behavior. Nilsen et al. [96] draw attention to dual processing models which detail two distinct modes—automatic and intentional—for reasoning and information processing. The lack of incorporation of automatic processing—fast, intuitive, and often heuristic-based—into behavior change interventions could be key to increasing uptake of evidence-based practice [96]. The authors stated that interventions that impact habits—discontinuation of harmful or ineffective habits or the development of beneficial habits—could be key to effecting behavior change [96]. This critique of the use of psychological theories in behavior change research may be reflected in our findings as increasingly determinant frameworks are being used in de-implementation research.

Our results align with Birken et al.’s [33] findings regarding the use of theory in implementation science. The implementation scientists surveyed in their study reported that they used theory primarily to identify facilitators and barriers which aligns with our findings. Seventy-seven percent of the implementation scientists in Birken et al.’s study reported that they used theory for data collection which is a significantly higher rate than shown in our study (46%). Sixty-five percent of our included studies used theory or frameworks to inform data analysis which also aligns with the 60% of participants in Birken et al.’s survey.

Similar to Nilsen et al.’s scoping review of the use of theory in de-implementation research [15], our results show that few studies examine the impact of patients on efforts to de-implement low-value care. Twelve of the 48 included studies considered patient impacts, with only five of these studies focusing on patient behavior alone. While the number of studies that include patients is low, analyzing these studies on the timeline shows that 11 of these 12 studies which examined patient factors were published in the last 2 ½ years (Nilsen et al.’s [15] inclusion period concluded mid-2018). Prusaczyk et al. argue that the impact of patients (community members and consumers, together “stakeholders”) is a unique and significant factor for de-implementation [17]. Therefore, while the number of studies is low, the trend implies that de-implementation research is increasingly including patient factors to understand and increase the effectiveness of efforts to reduce low-value care.

Many of the included studies not only explicitly report that a lack of knowledge or education is a key barrier to reducing low-value care but also acknowledge that knowledge is necessary, but not sufficient to effect change. This finding aligns with recent work on the effect of passive versus active change interventions which suggest that passive interventions, such as education sessions or the publication of guidelines or “do not do” lists alone, do not appear to be effective to change healthcare provider behavior to reduce low-value care [10, 27, 97‐101]. A 2015 study of early trends for seven Choosing Wisely recommendations concluded that the recommendations alone were not enough to produce significant changes in practice and that active interventions, such as financial incentives, data feedback, and systems-level interventions increase the effectiveness of practice change [100].

Three studies cited the Choosing Wisely campaigns as the motivation for the study, with an additional 3 studies referencing the campaigns. Other studies have identified a significant amount of research published in recent years addressing the Choosing Wisely campaigns and efforts to reduce practices identified on their “do not do” lists. This suggests to us that research specific to addressing Choosing Wisely recommendations is largely atheoretical at this point. Choosing Wisely campaigns are premised on the idea that the responsibility to reduce unnecessary practices is to be shared between physicians and patients with physicians as the stewards of healthcare resources, charged with ensuring that care is appropriate and sustainable [102, 103]. This individual-level focus aligns with all of the studies included in this review and the type of theories used could benefit Choosing Wisely inspired efforts to reduce low-value care. Grimshaw et al. developed a Choosing Wisely De-implementation Framework (CWDIF) in an effort to address this gap [10]. Models and frameworks developed specifically to support de-implementation is the focus of our subsequent scoping review, which will complement and expand the findings of this review.

This study has several limitations. As Niven et al. [12] and other researchers have noted there are many different terms used to refer to the process of reducing or eliminating low-value care; this study may have omitted some terms from the search criteria and may therefore have excluded relevant studies. In addition, articles published in languages other than English were not included due to a lack of resources. As is typical with scoping reviews, we did not assess the quality of the included articles. In addition, this review used implementation science as a basis for data collection and analysis. While this approach is debated in the literature [2, 10, 15, 16], we argue that this approach offers a solid foundation from which to study emerging research in de-implementation. This review focused on the use of existing theories and frameworks in de-implementation research. There is a growing body of literature that has produced frameworks and models to understand or explain efforts to reduce low-value care. We acknowledge the importance of this emerging literature, and this area of research is the focus of our subsequent scoping review.

This review adds value to de-implementation research through our use of enhanced rigor and a broad and comprehensive search. We systematically examined and characterized the theories and frameworks, their use and application, and gaps in the literature. These insights into how and why theories and frameworks are being used to understand and explain factors that influence efforts to reduce low-value care can inform research priorities, a systematic review of the literature and the development of targeted strategies and interventions. As healthcare organizations, hospitals, and healthcare providers continue to address the reduction of low-value care, more interventions and studies will use theory and more data on this topic will be available to facilitate more comprehensive and in-depth studies.