Setting
Denver Health (DH), a large, integrated, public safety-net healthcare system in Colorado [28], provides emergency medical services, inpatient treatment in a 500-bed hospital, outpatient primary and specialty care across 10 federally-qualified health centers (FQHCs) and 19 school-based clinics, and federally funded substance treatment services. DH serves approximately 250,000 patients annually across its facilities. In 2017, while there was no comprehensive registry for SUD, a cross-sectional continuum of care analysis [16] identified 3,300 individuals with OUD at DH. Using diagnosis codes and additional markers of OUD identified in the 2017 continuum of care analysis [16], an internal analysis estimated that nearly 11% of patients served by DH in 2022 had some kind of SUD (unpublished observations, ART).
Since 2016, DH has utilized a certified EHR (Epic, Verona, WI, USA) to document patient care. Parallel to the EHR, a separate set of spreadsheets and databases have been developed and maintained over decades to comply with 42 CFR part 2. These legacy and standalone systems required special access and authorization. In 2020, an updated DH consent for the use and disclosure of substance use and treatment information was revised to say that specialty SUD treatment data could be shared across the system with any DH provider involved in the patient’s care, designating all of DH specialty SUD treatment programs as one entity under 42 CFR part 2.
As informed by SUD treatment care team members anecdotally and through an internal needs assessment (which included thematic analysis of structured key informant interviews and focus groups with staff), tracking and/or communicating with other care team members about a patient were challenging tasks. To see a comprehensive patient story required labor intensive review of provider notes across multiple data sources (e.g., programs), and time (e.g., episodes [29]). Unstructured provider documentation precluded efficient outcome evaluation for individuals or populations. Addiction therapists redocumented information into a separate spreadsheet to track tasks and outcomes. Rather than an EHR-mediated referral, messaging started through email or a telephone call with an addiction therapist, who maintained their own spreadsheet for tracking. Caseload information was stored in a separate database which hampered ready access by supervisors for monitoring. A standard, cross-program definition for SUD treatment episodes was needed.
In 2019, DH established the Center for Addiction Medicine (CAM) to provide the infrastructure to integrate SUD care employing a hub and spoke model (see Fig. 1), different from traditional models that span multiple health systems and are state-wide (e.g. Vermont [30], Washington [14]). A hub resides in Outpatient Behavioral Health Services (OBHS), which provides specialty outpatient addiction services, including methadone treatment and care for priority populations including pregnant women and adolescents. Spokes include referral sources to the hub including an inpatient addiction consult service; emergency services, with 24/7 opioid agonist induction and linkage to care; and FQHCs providing integrated primary care with co-located SUD addiction therapists and behavioral health professionals. Outpatient SUD treatment occurs at FQHCs and several OBHS programs: (1) a general office-based addictions clinic, (2) an OUD dispensary clinic, (3) an adolescent SUD clinic, and (4) specialty women and family SUD services. Since its inception, CAM has provided the infrastructure, resources, and human capital to develop workflows to move patients from spokes to hubs. All of these locations utilize a single EHR instance which is managed and deployed by the DH information technology department. Areas still targeted for future data integration (stippled circles in Fig. 1) represent the withdrawal management service, the transitional residential treatment program, calls to the community line, and correctional care. The DH CAM hub and spoke model also includes strong partnerships with local organizations that provide residential treatment and other levels of care that DH does not offer internally.
Hub and Spoke Model diagram for integrating substance use disorder engagement, treatment, and recovery care from multiple entry points, Denver Health, Denver CO, 2023
EHR data integration enhancements
To establish standardized and streamlined SUD processes for referral and treatment episode tracking, two categories of EHR enhancements were implemented. The first was a treatment referral order that replaced phone calls, emails, and spreadsheets previously employed for making and tracking referrals from inpatient and ED locations. The new SUD treatment referral order, a single customized EHR referral order, collected patient-specific data from inpatient and ED SUD counselors to ensure the necessary information was available when the patient presented for treatment at the location receiving the referral. The second set of enhancements was a standardized treatment episode definition and episode-related tools that replaced program-specific definitions and data capture external to the EHR. Episodes of care needed a uniform definition for data sharing between programs (hub and spokes). Clinical and CAM leaders came together and decided that the new standardized treatment episode definition would start when an addiction therapist completes a comprehensive (1–1.5 h) biopsychosocial intake (which occurs in OBHS, FQHCs, ED, and inpatient setting) and ending on the discharge date from outpatient services. The new EHR-defined SUD treatment episode allowed standard information to reside within the patient’s chart, eliminating redundant documentation and expanding standardized documentation to additional existing programs. Episode tools customized to the new SUD treatment episode allowed users to easily manage standard episode-level information such as substance(s) of use and primary addiction therapist. Tools were centralized in an EHR-based dashboard where users were able to monitor SUD treatment checklists and oversee caseloads, among other capabilities, instead of in an external Microsoft Access database as had been done prior to enhancement.
Implementation of these EHR data integration enhancements began in earnest in July 2021. At this time, DH engaged an EHR consultant to efficiently design and develop data integration and maintenance enhancements. The EHR consultant began with weekly planning and feedback meetings with CAM team leadership (i.e., epidemiologist and public health planner) and clinical leadership. Clinical leadership contributed to the development of the system and defined integration requirements. The SUD treatment referral order was piloted in November 2021 and the SUD treatment episode definition and tool were piloted in a single FQHC clinic in Summer 2021. All tools were fully launched in February 2022. Data from external systems were imported to establish episodes for active patients right before the launch. A one-month, post-launch period allowed for training and uptake before post-implementation evaluation data collection. All day support was available for end-users via a communications tool (Webex) for the first week of operations. Ongoing weekly meetings were conducted with the EHR consultant, clinical and CAM leadership to assess enhancement use, identify issues, and provide feedback to the development team. By the third quarter of 2022, periodicity of meetings decreased to quarterly. Figure 2 represents a timeline of integration activities. Before integration (between March 1, 2021 and January 31, 2022), SUD treatment data were compiled monthly from systems external to the EHR from only one of four OBHS programs as it was the only program collecting SUD treatment initiation data and referral data from ED and inpatient areas. After integration (between March 1, 2022 and January 31, 2023), all previously existing OBHS and FQHC programs contributed standardized data into the common EHR, along with referral orders from ED and inpatient areas.
Timeline for electronic health record data integration, March 2021 – January 2023. Denver Health, Denver CO.
Evaluating the impact of integration
We evaluated the impact of EHR integration by comparing multiple factors before and after integration. Factors included (1) SUD treatment referral and episode volume; (2) patient characteristics; (3) treatment episode characteristics; (4) linked referrals; (5) treatment episode retention; and (6) tool satisfaction.
SUD treatment referral and episode volume: The number of referrals from ED and inpatient areas as well as SUD treatment episodes started within the before and after integration periods were collected and compared.
Patient characteristics: Patients undergoing SUD treatment before and after integration were compared on sociodemographic factors including sex, age, race/ethnicity, insurance class, and housing status, which were collected from the EHR. A patient was considered unhoused if there was evidence in the EHR that the patient was unhoused at any point in the observation period.
Treatment episode characteristics: The department where treatment was initiated and the primary substance of SUD treatment episodes were also compared before and after integration to understand whether types of SUD treatment episodes changed after integration. Patients could fall out of care and re-enter the process multiple times and in either period, each of which was considered a distinct episode of care.
Linked referrals: Referral linkage was tracked before and after integration. Integrating data in the EHR meant creating one common, operational definition of linkage. Therefore, linkage definitions were different before and after integration. Before integration, linkage was defined for those with OUD as a medication for opioid use disorder (MOUD) dose in the outpatient setting. Linkage was defined for those with a SUD (other than OUD) as an outpatient counseling visit after a referral from a hospital stay or ED visit. After integration, linkage was defined as whether a patient referred to SUD treatment from the hospital or ED had a completed appointment at an outpatient clinic (OBHS or FQHC) within 30 days of discharge. Before integration, referral data were collected from an external system and matched to the EHR to capture outpatient treatment within 30 days of the treatment referral. After integration, linkage had the same definition for all SUDs and data on referral and completed outpatient appointment after referral were collected entirely from the EHR.
Treatment Episode retention: 90-day retention in SUD treatment episodes was defined as whether a patient was retained in SUD treatment for 90 days or more. Similar to linkage, the definition of retention was standardized as part of integration. For both periods, episode start dates were the date the biopsychosocial intake was completed by an addiction therapist. Before integration, patient discharge information was not reliably recorded, which meant that the only reliable definition of retention in care was based on continuation of MOUD (for patients with OUD) or continuation of counseling services (for patients with a SUD other than OUD) without gaps in treatment. After integration, per the new SUD treatment episode definition, episode end dates were outpatient treatment program discharge dates.
Tool satisfaction: To ensure the system was developed as intended, we conducted a 17-question survey (supplemental information) among addiction therapists, developed from existing standardized surveys [31, 32], focused on whether the data integration efforts improved: continuity of care, efficiency of placing referrals, quality of care, patient monitoring, sharing of information, personal efficiency, and patient safety. The survey also asked about ease of analyzing outcomes of care, communicating with colleagues to coordinate care, and facilitating care documentation. The survey used a combination of Likert scale responses and open-ended questions. Requests to complete online surveys were sent to addiction therapists across departments (i.e., OBHS, FQHCs, ED, and Inpatient) in May 2022, three months after EHR integration tools were deployed. Respondents completed the survey online and data were collected and managed using REDCap electronic data capture tools [33]. Survey response rate was stratified by department. Questions were tabulated overall and by department. Open ended questions were grouped by department and coded for common themes.
Descriptive analyses of quantitative metrics and chi-square tests of categorical patient characteristics were carried out using SAS Enterprise Guide software, Version 8.3 of the SAS System for Windows (SAS Institute Inc., Cary, NC, USA). Survey results were analyzed in Tableau Desktop (Tableau, Seattle, WA, USA).
Add Comment