Mixed Methods: Capturing Complexity in Family Medicine Research

Abstract

98 In 2004, Borkan advocated for qualitative and mixed methods as important research strategies to address key challenges facing family medicine as a scientific discipline.1 These study designs help researchers capture the complexity inherent to the practice of family medicine with clarity and rigor. More recently, experts advocated for the use of qualitative and mixed methods to untangle the causal pathways in health disparities research.2 Qualitative data can corroborate and expand on quantitative research results, but only if data collection is carefully designed to capture the complexity. One design strategy is to develop a vision of how quantitative and qualitative results would ideally be jointly displayed and then work backward to design data collection strategies to get there.3 In this issue, a mixed methods analysis led by Solberg et al4 sought to unpack in a sequential design how leaders at highand low-performing clinics in Minnesota thought they were managing diabetes. Their data displays stratified high and low clinics by quantitative performance metrics (Solberg, Table 1) and by qualitative code and comment frequency (Solberg, Table 2). They adjusted for socioeconomic status before stratifying clinic performance to select their interview targets from among clinical leaders. That adjusted sample frame meant authors could be confident when the qualitative analysis detected no relationship between clinic performance and patient mix. Initially, all the care strategies reported in interviews seemed to hang together, however, although the number of comment/ clinic reflecting proactive approaches did not differ, as quotes show, the content was dramatically different by clinic performance. These differences led authors to distinguish between a traditional visit-based model “with individual patient responsibility for attendance and adherence” and a proactive model where the clinic teams used panel reports to drive systematic outreach (mail, e-mail, telephone) while also taking “advantage of visits for any reason to reinforce those suggestions....” The paper’s analytic meta-framework shows no single step from one-on-one doctor-patient relationship but more of an evolution that leveraged “reminder systems, checklists, data audit and feedback, and patient education” to drive proactive clinic behaviors. John Frey has warned that continuity cannot simply be systems managing the disease without the patient.5 These multisite data help us see a way forward for care monitoring systems to empower the broader clinic team and reinvigorate patient engagement. Wolk et al6 describes a staged implementation of the Collaborative Care Model (CoCM) to integrate mental health treatments in primary care, following the RE-AIM framework (ie, reach, effectiveness, adoption, implementation, and maintenance). The first year saw over 6,000 patients referred, nearly 7% of empaneled patients! Penn Integrated Care featured a centralized mental health resource center for intake, triage, and referral if indicated. Authors noted, “effectiveness often declines when programs move from efficacy trials to real world implementation” but in this study, length and frequency of treatment suggest high fidelity for CoCM services; rates of enrollment, symptom reduction, and remission are consistent with results from randomized controlled trials. From a service perspective, the multilevel triage met patient demand while allowing mental health clinicians to focus on treatment delivery. Mixed-method results can help bridge the gap for decision makers seeking to disseminate programs.7 Fortin et al8 report on a pragmatic randomized controlled trial of an integrated care management pathway to change how multimorbidity care is delivered. When the regional health authority adapted the intervention for regional dissemination, it altered the intervention design increasing the pragmatic dimension of the trial. At 4 months, the intervention did not improve the EDITORIAL