Purpose: To identify facilitators and barriers to implementing shared decision making (SDM) in primary care.
Method: We conducted 23 semi-structured interviews with leaders and clinicians from nine primary care practice sites participating in a current SDM implementation demonstration. Using a guide developed with input from demonstration conveners, interviewers queried respondents about their sites’ processes for integrating decision aids (DAs) into ongoing clinical operations, focusing on facilitators and barriers to operational tasks such as engaging clinicians, distributing DAs, and tracking patients through subsequent steps of SDM. Researchers inductively analyzed interview responses for recurrent themes.
Result: Facilitators. All respondents reported that SDM was consistent with their sites’ professional cultures, and most identified “champions” who engaged other clinicians in DA use. To facilitate DA distribution, some sites developed protocols that empowered non-physician staff: “The most successful sites…developed workflows that take the physician out of making the decision [about DA distribution].” To identify DA-eligible patients, these sites leveraged existing data (e.g., patient demographic characteristics, for screening decisions) and clinical processes (e.g., specialist referrals, for surgical decisions). When identifying DA-eligible patients required case-by-case physician judgment, single-click DA order entry and DA viewing by physicians facilitated greater distribution. Barriers. Physicians’ lack of prior SDM training was a barrier to participation: “Physicians felt that they were already doing shared decision making [before introducing DAs].” Physician DA ordering, though sometimes necessary for patient identification, limited distribution in multiple sites: “As long as you have the physicians in the middle of [DA ordering] they have too many other things on their plate to reliably ensure this would happen every time…in a 10-15 minute appointment.” Medical record systems (paper or electronic) posed significant barriers to tracking patients through the SDM process. For example, nearly all sites’ records lacked indicators for which patients had received DAs, mechanisms for communicating patient-reported values and preferences, and registry functions to follow patients' progress towards their decisions (e.g., to determine whether patients had timely post-DA decision making conversations with providers).
Conclusion: Even among highly motivated demonstration sites, there are significant educational, operational, and informatics challenges to implementing SDM in primary care. Empowering non-physicians may enhance distribution reliability for some DAs. However, improving post-DA follow-through may require better mechanisms for tracking patients and facilitating information exchange between patients and clinicians.