Incorporating new and expensive medications into practice is a persistent challenge for healthcare systems. Systems can embrace new medications enthusiastically; regulate them as a last resort, or gradations in between. In a sense, all VA sites can refer to Criteria for Use (CFUs) generated by Pharmacy Benefits Management (PBM) on direct-acting oral anticoagulants (DOACs) and other new medications. CFUs are designed to act as guidelines across the VA.
However, some VA sites are adopting DOACs more enthusiastically than others. By FY 2013, there were over 8,000 unique patients receiving dabigatran within VA (pilot data). At some VA sites, almost no patients received dabigatran (about 1 patient per 10,000 outpatients), raising the possibility of an inordinately restrictive site-level policy. At other sites, dabigatran was given to as many as 70 patients per 10,000. Such a 70-fold difference cannot be explained by clinical factors, and therefore represents an ideal opportunity to explore the causes of local variation in prescribing practice. These causes of variation go well beyond formal CFU documents, and even beyond written site-level policies. The true causes of variation reside in the informal policies and opinion climate present at each site. Such factors are not usually well-captured with qualitative inquiry.
We studied the expanding use of DOACs, a convenient but more expensive replacement for the older medication warfarin, within the VHA system as a way to examine how decisions are made about the relative advantages and disadvantages of new medications that lead to disparate patterns of utilization.
The objectives of this study were to:
1) Using DOACs as a case study, we examined how new medications are adopted and disseminated by identifying factors that lead to variation in prescribing. Our aim was to elicit data about context within which prescribing practices unfold, including who makes decisions, what is the role of evidence in these decisions, and what role is played by local formal and informal policies and practices in DOAC prescribing variation.
2) Identify how Pharmacy Benefits Management (PBM) Criteria for Use (CFUs) are used to guide decision-making at the site and/or medical center level.
We conducted a qualitative study involving semi-structured interviews with key informants from 5 of the sites with the lowest rates of prescribing (range: 1.31-4.76 per 10,000) and from 5 of the sites with the highest rates of prescribing (range: 30.84-69.78). Key informants (N=30) include people whose position enables them to influence local policies and practice patterns in this area; such as, chiefs of pharmacy, cardiologists, members of Pharmacy & Therapeutics committees, and frontline anticoagulation pharmacists. Participants were asked to reflect on the evidence for use of DOACS, their knowledge of PBM CFUs, local policies and procedures for approving DOAC prescriptions, the sources of those policies, and informal personal networks that may contribute to local practice patterns. Data were analyzed using theory-driven analysis, based on Rogers' Theory of Diffusion, with emergent thematic analysis. Using Roger's meant that particular attention was paid to differences between high and low prescribing sites. Coding was performed by multiple researchers with extensive qualitative training, who met frequently to assess the coding categories and ensuring coding agreement. This enabled us to construct portraits of DOAC utilization at the 5 high and 5 low prescribing sites.
We then used these portraits to explain variation in prescribing for a novel agent.
We found that while sites had practices and policies based on the same clinical studies or guidelines (CFUs), variation hinged on how sites interpreted "evidence." We identified 4 key domains of evidence perception that influenced DOAC uptake. Here we list these evidence perception domains and summary statements on what separated high prescribing/high adopting sites from low prescribing/low adopting sites:
1) Sites varied on what constituted evidence.
- Variation existed in what was cited as evidence and the relative weighting of various types of evidence.
a)High sites felt more comfortable with recent studies supporting DOACS.
b)Lower sites were less comfortable with recent studies, and preferred established medications.
2)Sites varied on how evidence was interpreted.
-Practice patterns are linked to how clinical trials and guidelines were interpreted.
a)High sites reported proactively reviewing the guidelines and literature to think critically about how to adapt evidence into practice.
b)Lower sites put more emphasis on a "wait and see" approach and continued to use older medications as first-line agent.
3)Sites varied on how national clinical practice guidelines are adapted into practice.
-When new guidance emerged, sites varied on how this guidance was adapted to daily practice.
a)High sites quickly worked to change approval processes to adopt new guidelines.
b)Lower sites were cautious and practiced under the limits of new guidelines, often maintaining extra local barriers to use of DOACS.
4)Sites varied on how they negotiated the "edge" of evidence.
-Sites different in how they manage uncertainty when guidelines were not specific or when a patient case was not adequately reflected by evidence from clinical trials.
a)In cases of limited clinical guidelines, high sites left decision-making to providers and patients.
b)Low sites perceived guidance as lacking for their particular patient population and preferred to establish organizational layers for each DOAC decision.
Anticipated Impacts on Veterans' Healthcare: VHA has an imperative to continue to provide high-quality care within a fixed budget. This case study provides lessons for how healthcare systems adopt and utilize new medications. Interpretation of what constitutes evidence impacts variation in adoption. Implementation science frameworks emphasize the importance of evidence, yet our work emphasizes the need to further examine how evidence is conceptualized and used to shape how sites choose to incorporate new medications into clinical practice. This study was a necessary one for our operational partners, PBM, who also wanted to better understand how CFUs were received and put into practice. The results will be useful for our operational partners in VA Pharmacy Benefits Management in planning for the release of other new medications in the future.
External Links for this Project
Grant Number: I21HX001719-01A1
None at this time.