Context
Informed by a pre-implementation assessment of the current transitions of care processes(32), and supported by existing evidence-based practices, we developed ACC to address a fragmented care coordination process. Patient perspectives on care coordination needs were paramount to the development and implementation of ACC. To ensure patient needs were addressed effectively we met with the VREB to obtain guidance. The VREB is comprised of eight Veterans, three VA employees who serve as liaisons to the board and represent two VA research centers, and one non-VA employee to facilitate meetings. Established in 2014, this board provides a forum for Veterans to meet monthly to review and share their unique perspectives with researchers on proposals/interventions. Prior to meeting with the VREB we provided information on our intervention, our goals for the meeting, the program aims, and relevant education materials (Additional file 1). We reviewed our intervention ideas with the board. They advised us on ways to further develop and improve our intervention and program materials to better meet dual-use Veterans’ needs and have a patient-centered approach. For example, they requested we educate Veterans on how to access their Patient Aligned Care Team (PACT) social worker as most Veterans do not know this is an available resource to them. This education was integrated into the clinical intervention component of ACC. Additionally, they revised the wording and format of our Veteran Care Card (see Intervention for details of this card) to make it more user friendly for Veterans.
Intervention
We developed and implemented ACC (Aim 1), a social worker-led care coordination intervention that provided Veterans who accessed non-VA EDs with longitudinal case management addressing SDOH up to 90-days post ED discharge to home. Prior to ACC there was no standardized care coordination process between VA and non-VA EDs. Veterans who accessed non-VA EDs were often not linked to necessary follow-up care upon discharge to home; thus, ACC was developed and implemented to address gaps in care for dual-use Veterans and to standardize care coordination between VA and non-VA EDs. We had one full-time and one part-time social worker at ECHCS and one full-time social worker at Nebraska-Western Iowa Health Care System (NWIHCS). These social workers were hired for this role and had to be willing to conduct home and community visits. The ACC social workers had ongoing collaboration with non-VA hospital staff. Based on information from our pre-implementation assessment(32), we learned coordinating care with the VA is frustrating and challenging. Non-VA hospital staff needed to have streamlined processes with the VA to coordinate Veterans care; thus, they were motivated to partner with us. Non-VA EDs are often overburdened with the number of Veterans accessing their ED and experience challenges in contacting the VA for reimbursements and coordinating follow-up VA care for Veterans. Regular in-services with non-VA EDs enhanced stakeholder buy-in and provided education on how ACC addressed SDOH and coordinated care for Veterans with the intention of decreasing frequent utilization of non-VA EDs. Non-VA ED staff were enthusiastic to partner with us as we provided a direct contact at the VA to assist with notifying appropriate VA departments to cover Veterans’ ED visit costs, helped with coordinating VA follow-up care, and assisted with developing the Veterans’ discharge plan.
Non-VA ED staff were asked to notify the ACC social workers when a Veteran visited their ED to ensure smooth care transitions post-discharge. Early notification was crucial for timely care coordination. Following this initial notification, the ACC social workers reviewed charts to determine program eligibility. The ACC social workers had access to VA charts through the VA’s electronic health record system and to non-VA charts through Joint Legacy Viewer which is an established electronic health record sharing system between hospitals. No releases of information were required to access these records as Health Insurance Portability and Accountability Act (HIPAA) regulations state hospitals can disclose protected health information without patient consent or authorization for the purposes of care coordination(33).
The ACC social workers called eligible Veterans within 24 hours post non-VA ED discharge to complete the social work comprehensive assessment (Additional file 2) to determine SDOH needs and develop a patient-centered care plan. This assessment took 30-60 minutes to complete and consisted of 21 questions pertaining to reason for referral, demographics, medical and mental health, social supports, living arrangement, education and employment, income and finances, current mental status, and psychosocial problems. Based upon this assessment, the ACC social workers utilized clinical judgement to determine the Veteran’s acuity level, with level 1 needing less case management support and level 4 needing the most. Veterans with acuity levels 1-2 were enrolled 1-4 weeks. Veterans with acuity levels 3-4 were enrolled up to 90-days post-ED discharge. Case management through phone calls was provided to all acuity levels. Home visits were completed for acuity levels 3-4 and community visits were utilized for Veterans experiencing homelessness.
Following the assessment, the ACC social workers provided individualized clinical interventions through phone calls and home/community visits. The ACC social workers continuously assessed Veterans for SDOH needs and linked them to appropriate VA and non-VA resources. All ACC participants had SDOH needs. SDOH were addressed by assisting with benefits acquisition (e.g. completing applications and/or placing referrals for financial and housing assistance and Medicaid, enrolling Veterans into VA services, etc.), providing education on accessing health care (e.g. mental health and substance use treatment, primary care appointments, etc.), scheduling health care appointments and placing treatment referrals, and addressing financial barriers (e.g. linkage to transportation resources including VA, Medicaid, and Medicare transportation, etc.). Enrolled Veterans preferences and needs informed clinical decisions and care coordination. The ACC social workers employed Motivational Interviewing techniques and teach-back methodology throughout the intervention. We developed a Veteran Care Card with information about ACC and the Veteran’s VA primary care physician (PCP) for Veterans to show to non-VA ED staff when they accessed their services to enhance care coordination between the VA and non-VA hospitals. Veterans were mailed this card either when they completed their participation in ACC (acuity levels 1-2) or during the first week of enrollment (acuity levels 3-4). The ACC social workers utilized the VA Office of Community Care (VA OCC) Care Coordination guidelines to inform the intervention(34).
When the Veteran reached the 90-day point or was no longer in need of ACC care coordination, the ACC social workers completed a warm hand-off through closed loop electronic communication to the Veterans’ VA primary care team. Veterans who needed case management after 90-days were connected to their assigned VA PACT social worker. PACT social workers collaborate with the Veteran’s VA PCP to enhance care coordination and patient-centered care. Throughout ACC implementation the ACC social workers documented data in our program database.
Setting and Participants
ACC was initially implemented in ECHCS and then disseminated to NWIHCS (Aim 2). We selected partner non-VA hospitals in Denver, Colorado and Omaha, Nebraska based on the high volume of Veterans served by these facilities. They were informed about ACC prior to launch. Veterans discharged home from non-VA EDs were referred to ACC between April 2018 to April 2020. Veterans already receiving case management in the VA were excluded from ACC to not duplicate services.
Implementation/Evaluation Team
Our multidisciplinary team consisted of social workers, nurses, a national training educator, clinical intervention specialists and consultants, experts in qualitative and quantitative research, statistics, data management, implementation science, and health economics. Our team developed and disseminated a toolkit outlining ACC’s core components, note templates, resource guides, care coordination processes, and VA and non-VA staff and provider training materials (Aim 3). We standardized training to ensure ACC was implemented with fidelity at ECHCS and NWIHCS.
Study of the Intervention
We evaluated the effectiveness of our intervention by comparing outcomes between ACC and a control group. Following the standard process for propensity matching, we performed propensity matching for Veterans who completed all four core components between 4/10/2018 – 4/1/2020 (N-=161) before outcome comparison due to differences in patient conditions and sample sizes between ACC and control group. Control group were pulled from Veterans who had non-VA ED visits and discharged home using both Corporate Data Warehouse Fee Basis Claims System and community care Program Integrity Tools System.
To ensure standardized program delivery, the ACC social workers were trained using evidence-based training curriculum developed as part of program implementation. Fidelity to the intervention was assessed using multiple methods. First, virtual learning collaborative meetings were conducted to discuss program progress, assess enrollment goals, and set benchmarks to improve outcomes. Second, the program database was designed to collect data on completion of program core components and to flag incomplete Veteran entries. Data quality reports were pulled weekly and any discrepancies and data issues were discussed during weekly check-ins with the implementation team and the ACC social workers. Finally, we conducted site visits during mid-implementation to assess program delivery process in real-time. Each site visit included meetings to obtain feedback about the program progress from various stakeholders and real-time observations of the program staff.
Measures
The primary outcomes were 30-day ED visits, 30-day hospital readmissions, and 30-day VA PCP visits following ED discharge. The secondary outcomes were 14-day PCP visits, 90-day ED visits and 90-day hospital readmissions. Additionally, we utilized our program database to collect data and understand what resources enrolled Veterans were linked to addressing SDOH. This database was designed for ACC, with built-in visual dashboards that enabled the ACC social workers and implementation team to track multiple points of health information of each Veteran and ACC programmatic information in real time.
Analysis
Due to huge differences in patient conditions and sample sizes between the ACC and control groups, we performed propensity matching prior to outcome comparison. The control group were matched to ACC intervention group with exact matching on site, discharge time, race, Urban, Rural or Highly Rural, Elixhauser variables (Coagulopathy and Pulmonary Circulation Disorder) (e.g. a group of control subjects were selected as a match to the intervention subject based on having the same values on exact matching variables), and nearest neighbor matching on age, gender, all other Elixhauser comorbidity variables, and number of hospitalizations/ED visits/PCP visits in the past year. To reduce the impact of COVID-19 on our outcome comparison, we matched the control and ACC groups on the time of discharge, by quarter for patient discharged before 2020 and by month for patients discharged in 2020. We matched with a ratio of 3 control to 1 ACC patient due to the large control group sample size. The matched cohort was checked by assessing the propensity score balance between control and ACC groups, as well as the standardized differences of the matched variables. Standardized differences less than 0.1(10%) between control and treatment groups are commonly considered as negligible imbalance. After matching, all standardized differences between ACC and control groups were below 0.08 for predictive covariates, indicating appropriate covariate balance. The histogram plot of the propensity score distribution also showed well balance between two groups after matching.
To account the correlations between 30-day hospital readmissions and 30-day ED visits, we fitted these two outcomes with a joint survival model, using Markov Chain Monte Carlo technique, with 10000 iterations and burn-in of 500. The survival models were assumed to have proportional hazard with baseline risk function of Weibull distribution. All 90-day outcomes and PCP visit outcomes were fitted with Cox proportional hazard model. The covariates included in the survival model were Elixhauser score and number of ED/hospitalization/PCP visits in the past year. Based on Kaplan Meier curves, the effect of the ACC intervention on 30-day PCP visits changed over time. We included interaction term of intervention and time in the Cox model for this outcome. To compare changes in outcomes before and after intervention, between the control and ACC intervention groups, we performed Difference-In-Difference (DID) analyses on number of ED visits, admissions, and PCP visits 120-day pre/post discharge. The DID were implemented as an interaction term between time and intervention group in a regression model for each outcome.
Ethical Considerations
ACC is a Department of Veterans Affairs grant funded QI program (see Ethics approval and consent to participate). We were exempt from the Institutional Review Board. Appropriate regulatory approvals were obtained to implement this program.