Implementation of an Interdisciplinary, Team-Based Complex Care Support healthcare Model at an Academic Medical Center: Impact on healthcare Utilization and Quality of Life

Ritchie C, Andersen R, Eng J, Garrigues SK, Intinarelli G, Kao H, Kawahara S, Patel K, Sapiro L, Thibault A, Tunick E, Barnes DE
Source: PLoS One
Publication Year: 2016
Patient Need Addressed: Behavioral health, Chronic Conditions, Long-term services and supports
Population Focus: Complex care, Low income
Demographic Group: Adult, Urban
Intervention Type: Best practices, Service redesign, Staff design and care management
Study Design: Mixed-Methods
Type of Literature: White

The Geriatric Resources for the Assessment and Care of Elders (GRACE) program has been shown to decrease acute care utilization and increase patient self-rated health in low-income seniors at community-based health centers.

To describe adaptation of the GRACE model to include adults of all ages (named Care Support) and to evaluate the process and impact of Care Support implementation at an urban academic medical center.

152 high-risk patients (≥5 ED visits or ≥2 hospitalizations in the past 12 months) enrolled from four medical clinics from 4/29/2013 to 5/31/2014.

Patients received a comprehensive in-home assessment by a nurse practitioner/social worker (NP/SW) team, who then met with a larger interdisciplinary team to develop an individualized care plan. In consultation with the primary care team, standardized care protocols were activated to address relevant key issues as needed.

A process evaluation based on the Consolidated Framework for Implementation Research identified key adaptations of the original model, which included streamlining of standardized protocols, augmenting mental health interventions and performing some assessments in the clinic. A summative evaluation found a significant decline in the median number of ED visits (5.5 to 0, p = 0.015) and hospitalizations (5.5 to 0, p<0.001) 6 months before enrollment in Care Support compared to 6 months after enrollment. In addition, the % of patients reporting better self-rated health increased from 31% at enrollment to 64% at 9 months (p = 0.002). Semi-structured interviews with Care Support team members identified patients with multiple, complex conditions; little community support; and mild anxiety as those who appeared to benefit the most from the program. DISCUSSION: It was feasible to implement GRACE/Care Support at an academic medical center by making adaptations based on local needs. Care Support patients experienced significant reductions in acute care utilization and significant improvements in self-rated health.

Insights Results

Overview of article

  • The Geriatric Resources for the Assessment and Care of Elders (GRACE) program is a health care delivery model that was designed to serve as a support system between patients/caregivers and the primary care provider (PCP). The model includes a nurse practitioner/social worker (NP/SW) team that performs comprehensive, structured assessments in patients’ homes and then meets as part of a larger interdisciplinary team that includes a geriatrician, mental health liaison and pharmacist. The GRACE Team works alongside the patient’s primary care team to implement the care plan and modify it as needed over time
  • The primary objective was to evaluate the adaptation and implementation of GRACE at an urban academic medical center. One of the key adaptations was to include adult patients 18 years and older who met enrollment criteria; therefore, the program was renamed Care Support to reflect this more inclusive age range
  • The setting for this implementation study was 4 primary care medical clinics at a large urban academic medical center. Within each clinic, ‘high-risk’ patients—defined as patients with ≥5 emergency department (ED) visits or ≥2 inpatient hospitalizations in the past 12 months—were identified based on lists of recent admissions and direct referrals. These high-utilizing patients were then vetted by PCPs for appropriateness of enrollment taking into consideration the patients’ needs, current resources and potential for engagement
  • Patients who were approved by their PCPs were placed on the Care Support Registry. The Care Support program includes an initial in-home assessment for a comprehensive evaluation of the patient’s needs and available resources. In some cases, the initial assessment was performed in the clinic or by phone if the patient declined the in-home assessment. The initial assessment was then discussed with the larger interdisciplinary team that included a geriatrician, mental health liaison and pharmacist. An individualized care plan was created for each patient that included activation of specific care protocols, which were then reviewed and modified as needed by the primary care physician. Follow-up visits were typically conducted by phone
  • The study evaluation was informed by the Consolidated Framework for Implementation Research (CFIR) and was performed as a partnership among clinical and research teams


  • In general, Care Support adhered to the process goals of GRACE with high fidelity. An average of 6 standardized protocols were activated, the most common of which were chronic condition self-management (95%), social service coordination (87%) and advanced care planning (83%). Mental health protocols were activated more often in younger than older patients (70% vs. 48%, p = 0.007). The Care Support team interacted with patients an average of once in person and three times by phone during the first 30 days of enrollment
  • The median number of ED visits/1000 observation days declined significantly from 5.5 (range: 0–54.9) before Care Support to 0 (range: 0–87.0) after Care Support enrollment (p = 0.015). This difference was primarily attributable to a significant increase in the proportion of patients with 0 ED visits before and after enrollment (40% vs. 54%, p = 0.015). The total number of ED visits in these patients was 227 before Care Support and 203 after Care Support
  • Similarly, the median number of hospitalizations/1000 observation days declined significantly from 5.5 (range: 0–33.0) to 0 (range: 0–43.0) before and after Care Support enrollment (p<0.001). This difference also was primarily attributable to the proportion of patients with 0 hospitalizations, which nearly doubled from 33% before Care Support to 60% after Care Support (p<0.001, Fig 3). The total number of hospitalizations in these patients was 186 before Care Support and 128 after Care Support
  • Barriers to implementation: 1) Changes in the enrollment criteria as the program evolved, which made it more difficult for PCPs to identify appropriate patients for enrollment; 2) Limited initial access to mental health services, which made it difficult to support patients with more severe mental illness; and 3) Being spread out across multiple sites, which made meeting and communication more difficult; 4) and the size of the patient panel, which sometimes limited the amount of time available to address the needs of each patient
  • Facilitators to implementation: 1) The GRACE protocols, which provided structured and adaptable templates and enabled the team to more efficiently manage patients with complex care needs; 2) The home visit, which provided the team with key insights into the real-world issues and day-to-day needs of each patient in their own environment; 3) The comprehensive assessment, which provided a complete picture of all of the patient’s needs and the interdisciplinary team, which enabled incorporation of a wide range of perspectives; and 4) Colocation with PCPs enabled the Care Support team to build greater rapport with PCPs and to meet patients at their clinic visits, thereby increasing the frequency of in-person “touches” and the ability to implement interventions quickly
  • Findings: 1) Care Support team members felt that the patients who appeared to benefit the most were those with complex medical needs, little community support, and mild levels of anxiety, which are known drivers of high health care utilization; and 2) Those who benefitted the least were those with more severe mental health issues and lack of interest in engaging with the health care system; 3) In patients who enrolled in Care Support, health care utilization declined significantly for both ED visits and hospitalizations when comparing utilization 6 months before versus 6 months after enrollment

Key takeaways/implications

  • Overall, the study suggests that it is feasible to implement the GRACE/Care Support model at an academic medical center by making adaptations based on local needs, and that patients who participated in Care Support experienced significant reductions in acute care utilization and improvements in self-rated health