Research exploring the Primary Care Behavioral Health (PCBH) model has been increasing in the literature. However, the body of work on this specific model of integrated care is still being built, and many unanswered questions remain. There is consensus in the field that more study and research is needed to expand our understanding of the effectiveness, efficacy and systemic implications of PCBH work. Gaining a more precise understanding of these issues will have policy implications on our health care system as well as on funding resources. Research demonstrating effectiveness could assist in increasing implementation of this model of integrated care into the systems best suited for it, and expand funding resources to support the work. Additionally, studies exploring the systemic implications for the PCBH model will increase our understanding of the complex, inter-related variables that impact access to behavioral health care and the effectiveness of this care.
One recent study that has scratched the surface of the systemic implications of PCBH work was conducted in Madison, WI (Serrano et al, 2018). This study utilized a data set that included primary care utilization data for an FQHC with 3 sites that utilized the PCBH model and for a primary care clinic that did not utilize PCBH. Data was also collected on ED encounters for a subset of patients of the same four primary care clinic who were with a mental health diagnosis. The study was pre-post in nature, exploring ED utilization before and after PCBH implementation at the FQHC clinics. The primary care clinic (which did not utilize PCBH served as a control arm. We found that one FQHC site did show a decrease (11.3%) in the ratio of ED visits to primary care encounters, and the other 2 FQHC sites did not. We also discovered, not surprisingly, that research on health care systems is highly complicated, with many layers of intersecting variables that can make interpretation of data challenging.
One specific challenge was finding a control site that controls for all relevant variables. There are several reasons for this. Finding similar clinics with and without PCBH, that have the same demographics, geographic area, same insurance mix in the same time-period would be extremely challenging if not impossible. These are simply too many variables to control in a “real-world” situation, particularly considering health care disparities that exist both nationally and regionally. The lack of adequate controls can make it difficult to interpret cause and effect in data sets.
Ideally a controlled study could be conducted to help move the field forward and deepen our understanding of the model. However, this would be extremely challenging to conduct. For a formal randomized clinical trial, there would need separate clinics in the same geographic region with the same demographic mixes and insurance mixes in the same time period, one that has implemented PCBH and one that has not. They would need to have similar services, provider types, access to care. Of course, a formal randomized controlled trial, patients would need to be randomly assigned to treatment conditions (in this case clinics), taking away patient choice and likely not realistic. Instead, it seems that more naturalistic studies will be more realistic, where clinics are compared but patients themselves are choosing the clinics. It is recommended that as many factors as possible be controlled for in the analyses, including patient demographics, diagnoses, contextual factors, payor mix, geography and so on. The more that is controlled for, the more specific the results can be, with less room for multiple interpretations.
Meghan Fondow PhD is the Primary Care Behavioral Health Manager and Clinical Training Director at Access Community Health Centers in Madison, WI. Meghan has worked as a BHC at Access for over 10 years, and also holds an adjunct clinical faculty appointment with the Department of Family Medicine and Community Health at the UW-Madison School of Medicine and Public Health.