Association of Team Structure and Chronic Disease Outcomes: Biomarker Evidence from U.S. Primary Care Clinics

Abstract

Research ObjectiveThe quality of chronic disease management is highly variable and low on average. As chronic care is increasingly delivered by teams instead of solo providers, an important yet unresolved question concerns the structure of teams that tend to perform best. Optimal team structure likely depends on the mechanisms by which team‐based care may improve outcomes. Mechanisms that may favor larger, more diverse teams include more frequent patient touchpoints, complementarities between team member skill, or a higher chance of developing a strong patient‐provider relationship. Conversely, coordination costs or diffusion of responsibility may cap the optimal team size. We examine the relationship between team structure and chronic disease outcomes.Study DesignAnalysis of patient outcomes by team structure (i.e., number and types of providers seen) using rich, de‐identified EHR data. We classify teams by constructing a measure analogous to the Herfindahl–Hirschman Index (HHI) to assess provider concentration across a patient's primary care visits. Our chronic disease outcomes include biomarkers that often determine clinical decision‐making, including hemoglobin A1c, blood pressure, and LDL cholesterol, as well as process measures. First, we compare outcomes for patients who receive care from solo providers (HHI 10,000), concentrated teams (5000 < HHI < 10,000), balanced teams (HHI 5000), and diffuse teams (HHI < 5000), adjusting for organization fixed effects and baseline disease biomarkers. Second, we identify the effect of team‐based care by studying practices that switch from solo to team‐based care, using a difference‐in‐differences design.Population Studied>10 million primary care visits of >1.2 million patients cared for by ~2000 providers at >250 practices part of the athenahealth network from 2013–2018.Principal FindingsTaking diabetes as an example, we find that 75.7% of patients with new‐onset disease receive care from a solo provider, 12.3% from a concentrated team, 8.5% from a balanced team, and 3.5% from a diffuse team. We find that process measures differ by team structure, with solo providers ordering fewer A1c tests and prescribing fewer anti‐diabetics, compared to concentrated and diffuse teams, but not compared to balanced teams. Finally, we find that solo providers are less likely to bring their patients' diabetes under control compared to diffuse teams, but not compared to balanced or concentrated teams.ConclusionsPerformance differed by team concentration, suggesting that optimal team structure may trade off team size and frequency of visits per team member. Our study is the first, to our knowledge, to investigate the effect of different team structures on individual biomarker‐level outcomes.Implications for Policy or PracticeOur results may provide insights about the most effective team structures for increasingly consolidated primary care practices (e.g., PCMHs and ACOs). For practices seeking to adopt a team‐based care model, team structure is a potentially important factor to consider when deciding on staffing models. For payers, our results suggest that novel care coordination incentives, such as the CMS Chronic Care Management Services CPT code introduced in 2015, can be designed and further improved upon to encourage team structures that optimize the value of team‐based care.