Exhibit 3. Adjusted lab monitoring rates for patients with new onset of type 2 diabetes or hyperlipidemia by team-based care and provider type, 2013–2018.

Source: Authors’ analysis of de-identified electronic health record data from 2013 to 2018 for adults with a new chronic disease as identified by an abnormal biomarker (N = 22,874 for type 2 diabetes, and N = 24,510 for hyperlipidemia).

Notes: Colored bars indicate the predicted values from a regression of an indicator for a patient having at least one disease-appropriate lab result within one year (i.e., hemoglobin A1c for type 2 diabetes and LDL cholesterol for hyperlipidemia) on provider, including organization fixed effects as well as patient and disease severity covariates as described in the text. Error bars represent 95% confidence intervals. p-values for team total vs. solo total were as follows: diabetes, <0.001; hyperlipidemia, <0.001. Non-physician providers are physician assistants and nurse practitioners. A mixed team is composed of both physician and non-physician providers. This outcome was not examined for hypertension since blood pressures are routinely taken during visits and thus may not be informative of active hypertension monitoring.