Contra Costa Health Services Whole Person Care (CommunityConnect) Program Evaluation

The Effect of Administrative Enrollment Into Case Management and Linkage Services on Health Behaviors, Utilization, and Outcomes in a High Risk Population

Contra Costa Health System's WPC Pilot Program, titled Community Connect (CMCT), delivers case management and linkage services to high-risk Medi-Cal members in Contra Costa County, California. This program is funded under the CMS/DHCS 1115 Waiver Whole Person Care (WPC) Pilot Program through 2020. High-risk individuals from the population of Contra Costa County full-scope Medi-Cal enrollees are connected with a case manager who provides linkage services to address their social determinants of health.

Program capacity is below the eligible population, so a tiered randomization strategy is used to identify enrollees and similarly risky controls (who are eligible for enrollment at later intervals). Health behaviors of enrollees and controls are tracked via electronic health records, billing claims, and other social service administrative databases to create a detailed record of post-randomization health behavior. The primary outcome of interest is avoidable utilization of emergency room and in-patient services.

Program Description:

The goal of this proposal is to evaluate the effectiveness of a social needs case management program targeting a population of high-utilizers of multiple systems in Contra Costa County, primarily its health delivery system. The program to be evaluated through this study is funded through 2020 under the CMS/DHCS 1115 Waiver Whole Person Care (WPC) Pilot Program. Contra Costa Health System's WPC Pilot Program, titled Community Connect (CMCT), is implemented administratively for enrollment. High-risk individuals, 18 years of age or older, are identified from Contra Costa County full-scope Medi-Cal enrollees as preliminarily eligible for enrollment. A predictive-modeling algorithm is applied to this population to identify the members of the population at high risk for future avoidable emergency room (ED) visits and inpatient (IP) admissions. These highest risk individuals are then randomized to an administrative enrollment or a control groups. The enrolled program participants are then contacted by program staff for consent to participate in the program for services. They can decline services at that contact and then would remain in the intended to treat population that declined services at initiation. The primary hypothesis is that delivery of a social needs case management intervention will reduce avoidable ED and IP utilization among the target population.

High-risk individuals are identified for enrollment using a population health predictive-risk model. This model integrates variables from a number of Contra Costa Health Services databases, including the electronic health record used by all in-network providers and claims detailing all out-of-network utilization. This predictive-risk model has been continuously updated throughout the study to include more relevant features and change the model form. Diagnosis-based algorithms are used to determine the 'avoidability' of a given emergency room or in-patient visit.

Individuals enrolled in this program are cared for within the context of a multi-disciplinary care team that frequently consult with each other and seek consultation around specific expertise within context of social needs case management. After enrollment, patients are assigned to either a face-to-face visit case manager (Tier I) or to a telephonic case manager (Tier II) depending on the severity of their risk within the predictive risk model and other variables including program capacity. Tier enrollment counts have varied across time; currently the program is designed to enroll 5,977 patients simultaneously in Tier I care and 6,605 patients in Tier II care. Program design began in 2016, the first patients were enrolled in April 2017, and the program is projected to reached full staffing capacity in August 2018. New patients are added to the program rolls on a monthly basis as staffing capacity increases and to replace program dis-enrollments. Starting in September 2017, enrolled patients were matched with controls identified from the population with similar risk profiles. Controls were initially recruited at a 1:1 ratio with cases; this ratio was increased to 2:1 as of January 2018.

Currently a pool of the top 12,000 high-risk individuals in the population (as defined by the risk model) are eligible for Tier I care while the pool of those ranked 12,001 to 25,000 are eligible for Tier II care. Each month, open spaces in each Tier are filled with randomly chosen patients from the eligible pools, and twice as many matching controls from the same pool are identified simultaneously. All non-enrolled pool members, regardless of control status, are then returned to the eligible pool for enrollment the next month. The CMCT program lasts one year, providing that the patient is amenable to the case management relationship. At initial enrollment, every program participant is asked to answer a quality of life questionnaire and a social needs questionnaire. The quality of life questionnaire is repeated on an annual basis. The social needs questionnaire is used by the case manager to identify the client's social and health care goals. During the course of program participation, the case manager works with her/his clients using motivational interviewing techniques and change readiness assessments to identify and prioritize the client's care goals. Every case manager provides a group of core services that include navigation support and linkages to resources to help clients reach their goals. At the end of the year, if a client risk profile still makes them eligible for program inclusion, they are automatically re-enrolled for another year. Otherwise, they return to the client pool, but are eligible for a second enrollment if their risk increases.

Specific Project Aims:

1. To evaluate the effect of a social needs case management program on the rate of avoidable health care utilization within the Medi-Cal enrollees utilizing services within CCHS population.
2. To identify the effect of a social needs case management on the rate of avoidable health care utilization among the populations who a) receive high-touch social needs case management/Face-to-face visit case, and/or b) lower-touch telephonic social needs case management: and/or c) receive any of these services in the context of an interdisciplinary team model of service delivery for social needs case management.
3. As secondary analyses, to evaluate the effect of the case management program in the above populations on a number of secondary outcomes, including, but not limited to: specialty care utilization, no show rate, primary care physician engagement, health insurance maintenance, incarceration rates, social service program enrollment, housing status and self-reported quality-of-life survey.

Analytic Plan:

All outcomes will be compared using both an 'intention-to-treat' and an 'effect of treatment among the treated' approach. For the 'intention-to-treat' analysis, rates of events will be calculated using all person-time from enrollment to end of follow-up. Rates for the enrollees will be compared to weighted rates for the controls. The control weights (or survival weights) will be inverses of the probability that a control remains unenrolled, adjusting for the fact that controls are eligible for later follow-up and this enrollment necessitates and end to the control period.

For the 'effect of treatment among the treated' effect, rates will be calculated including only person-time subsequent to an enrollee being matched with a case manager and documenting a goal to address a social need. This allows the analysis to focus on the time in which the program could presumably be expected to impact a patient's health and behaviors. As a portion of enrollees decline services, or otherwise never have a documented goal, the comparison population will be adjusted to account for this possibility. A statistical model will be created to predict the probability of an enrollee ever documenting a goal and the control population will be re-weighted by the inverse predictions from this model (IPT weights). These IPT weights will be combined with the survival weights to create a comparable control population.

Comparisons between populations will be performed with targeted minimum loss based estimation, targeting the mean effect of enrollment on the outcome within the relevant population. A more traditional approach will also be performed, estimating the size of the difference in outcomes between the enrolled and control populations.

Inclusion Criteria:

• Full Scope MediCal members administered by Contra Costa Health Services and in Contra Costa Regional Medical Centers network.
• Having sufficiently high estimated risk of future avoidable utilization to rank within the top 25,000 of the potentially eligible population

Exclusion Criteria:

• Not case managed under other locally administered plans
• Not living outside Contra Costa County
• Not in detention for the past month
• Not hospitalized for the past month
• Not previously enrolled in the program (some disenrollment reasons allow for subsequent re-eligibility)