Lead/Presenter: Linda Takamine,
COIN - Ann Arbor
All Authors: Takamine LH (COIN - Ann Arbor), Forman, JH (COIN - Ann Arbor), Damschroder, LJ (COIN - Ann Arbor) Youles, BW (COIN - Ann Arbor) Sussman, JB (COIN - Ann Arbor)
With the introduction of Big Data, risk prediction is playing a rapidly-growing role in clinical medicine. This is especially true in cardiovascular disease (CVD) prevention, where a patient's risk of developing a heart attack or stroke is now a core determinant of treatment for every class of drugs. However, the influence of these risk prediction tools on clinical decisions has been has been limited. We sought to understand providers' facilitators and barriers to incorporating risk prediction into their regular clinical practice.
Between June and November 2018, we conducted 36 semi-structured interviews with primary care providers at nine VA sites. Facilities were primarily chosen based on size and geography. Interviews (30 to 60 minutes) contained brief clinical scenarios and introduced a risk-prediction approach to CVD treatment. We used qualitative content analysis and matrix analysis to generate findings.
We identified three groups of providers with respect to their use of and attitudes toward risk prediction tools. Those who: (1) were enthusiastic about adoption; (2) saw potential value, but had concerns; and (3) rejected the approach. Enthusiastic providers had prior exposure mostly through training and had already integrated prediction tools into their treatment approach. The second and most common group accepted risk prediction conceptually, saw value primarily for patient communication and reduction in cognitive burden, but were concerned about integration into workflow (e.g., accessibility of tool during clinical encounters) and questioned how risk was calculated. The last group rejected risk prediction conceptually, saw no added value versus a traditional approach, and instead wanted to retain targets for individual risk factors (e.g., blood pressure, LDL). They were not trained to use risk prediction, and were concerned with workflow and cognitive burden.
While providers generally accept the concept of guiding care using risk prediction, its introduction into regular clinical practice faces many practical and psychosocial barriers.
Introduction of risk prediction will require complex interventions to improve care.