PCR011: Association of psychiatric consultant characteristics with clinical outcomes of collaborative care for depression

Evan Chen; Juan Carlos Bazo-Alvarez; Ian Bennett, MD, PhD; Davis Patterson, PhD

Abstract

Context: Adult depression treatment in primary care is improved by integrated behavioral health such as the Collaborative Care Model (CoCM) but outcomes vary across health centers. Objective: Identify CoCM team factors associated with variation in clinical outcomes. Study Design: Correlative study of survey with linked clinical data from routine care. Setting or Dataset: Primary care health centers in Washington state participating in two CoCM implementation and sustainment initiatives (MHIP and BHIP). Clinical data from the Care Management Tracking System (CMTS) used as part of routine clinical care in all sites. Population studied: CoCM clinical teams and adults diagnosed with depression and receiving care from 31 distinct clinic sites throughout Washington state. Instrument: A survey for CoCM Care Managers was developed based on qualitative work exploring CoCM team function in eight domains: overall team function (OTF), psychiatric consultant (PsyC), relationship with primary care physicians (RPCP), role in the practice (RP), training experience (TE), thoughts on CoCM (TOC), leadership support (LS), and overall experience with CoCM (EXP). Outcome Measures: Depression symptoms (PHQ-9) measured as part of regular clinical care. Clinically significant improvements were defined as ≥50% improvement, last PHQ-9 measured <10, and a last PHQ-9 measured <5 (remission). Results: Data from 59 Care Manager surveys and 2509 patients from 31 clinics showed psychometric evidence for the 8 survey domains. Unadjusted multilevel mixed-effects logistic regression found the PsyC domain was associated (p<0.05) with >50% change in PHQ-9 (OR=1.32, p=0.046), last PHQ-9 <10 (OR=1.34, p=0.009), and last PHQ-9 <5 (OR=1.25, p=0.029). Adjusting simultaneously for both clinic-level and patient-level variables, the PsyC domain was significantly associated with likelihood of patient remission (last PHQ-9 <5; OR=1.22, p=0.031). Conclusions: An instrument to assess perceived functioning of the CoCM team had psychometric support. Perceived characteristics of psychiatric consultants was associated with likelihood of depression remission. Characteristics included interest in teaching, commitment to the site, making efforts to have weekly meetings, and willingness to assist care managers on adjustment of treatment strategies. This study is the first to quantify variation in CoCM team functioning with patient outcomes and can be used to inform training and the use of the CoCM.
Leave a Comment
Jack Westfall
jwestfall@aafp.org 11/19/2021

Great poster and abstract. very interested in this topic and your findings. the Robert Graham Center is doing some research in this area and would be interested in learning more. please email me. thanks.

Ian Bennett
ibennett@uw.edu 11/22/2021

Thanks Jack - will do - great to hear about the work of the Graham Center and I would love to hear about it.

William R. Phillips
wphllps@uw.edu 11/21/2021

Very important Fascinating research question. I would love to see a map of the clinical sites. Does the model/study include other behavioral health professionals - who are more broadly available and more commonly integrated into family medicine teams - besides consulting psychiatrists? Thanks. - Bill Phillips

Ian Bennett
ibennett@uw.edu 11/22/2021

Hi Bill, we were looking at collaborative care sites that follow the CMS defined model and so all had psychiatric consultants - mostly consulting remotely. Sites could have other mental health specialists but are not a major focus of this particular analysis.

Gillian Bartlett
gillian.bartlett@health.missouri.edu 11/22/2021

Such an important area - would love to know more about what patient-level characteristics you included in you multi-level modelling. Was obesity a co-morbidity you considered?

Ian Bennett
ibennett@uw.edu 11/22/2021

Hi Gillian, we did not include obesity - our variables were gender, age, and ethnicity

Ian Bennett
ibennett@uw.edu 11/22/2021

individual level variables that is.

Gillian Bartlett
gillian.bartlett@health.missouri.edu 11/22/2021

Any way to consider this? although age, gender and ethnicity are poxies for a LOT of things.

Gillian Bartlett
gillian.bartlett@health.missouri.edu 11/22/2021

That was supposed to be "proxies" - I wasn't try to say "a pox on you" :)

Andy Pasternak
avpiv711@sbcglobal.net 11/26/2021

Excellent work- I'd love to see a lot more of this model as a doc in private practice

Social Media

Address

NAPCRG
11400 Tomahawk Creek Parkway
Leawood, KS 66211
800.274.7928
Email: napcrgoffice@napcrg.org