Tri-County Health Department (TCHD) is located in the Denver Metro area, serving approximately 1.6 million people in Adams, Arapahoe, and Douglas Counties.  TCHD serves over 25% of Colorado’s state population and its jurisdiction includes ~3,000 square miles, 26 distinct municipalities and 15 school districts.

In spring 2020, TCHD was faced with the need to complete COVID-19 case investigation and contract tracing (CI/CT) for 30-50% of all of the COVID-19 cases in the State of Colorado.  The state health department decided to contract with a vendor to develop a statewide system, but that system would not be ready for implementation by most local health departments until September 2020.  TCHD had a need to manage case investigation and contact tracing for thousands of COVID-19 cases and their contacts sooner, and we were facing limitations with common data management tools used in outbreak management.

TCHD leveraged a multidisciplinary team and existing geographic information system (GIS) infrastructure to develop a COVID-19 surveillance system for CI/CT activities, allowing for 200 CI/CT staff to simultaneously complete case/contact interviews and data entry. The system also included interfaces for community referral processes and CI/CT performance management. The goals and objectives were to complete COVID-19 case investigation and contact tracing efficiently and effectively; to utilize COVID-19 surveillance data to inform decisions around interventions and policy measures; to enable TCHD partners and the public to access real-time COVID-19 data, and to capture performance management data on CI/CT operations.

We implemented our application in a phased approach using best practices in information technology (IT) project management with case investigation implemented first, contact tracing second, and community resource and performance management solutions implemented in later phases. TCHD has used our system for all COVID-19 CI/CT activities since June 2020 to carry out critical public health activities in a more coordinated way while waiting on the development of the statewide system.  TCHD’s success relied on a highly skilled, multi-disciplinary team of internal staff including epidemiologists, informatics specialists, information technology staff, IT project manager, public health nurses, environmental health staff and public health preparedness staff.  Our success was also the result of existing investments in advanced informatics and IT infrastructure at TCHD.

TCHD’s software system allows for real-time data analyses to describe the impact of the COVID-19 pandemic across populations based on racial, ethnic, socioeconomic, occupational, and geographic descriptors.  Using our neighborhood- and census tract- level analyses, we were able to provide recommendations for place-based public health messaging to our counties and municipalities.  The development of this system has allowed TCHD to create 15 distinct, interactive public-facing COVID-19 data dashboards (https://data.tchd.org/covid19_resources/) to inform the public and decision-makers.

In sum, TCHD piloted using existing GIS technology for COVID-19 CI/CT activities.  TCHD has demonstrated the utility of this application to numerous public health agencies and to industries outside of public health via webinars and conference presentations.  Based on TCHD’s prototype, ESRI has marketed a similar system to health departments all over the United States and several local public health agencies have implemented these systems.

TCHD serves a population of approximately 1.6 million residents in Adams, Arapahoe, and Douglas Counties in the Denver Metro area.  In spring 2020, TCHD was tasked with conducting CI/CT on COVID-19 cases and contacts residing within its jurisdiction, which comprised 30-50% of all COVID-19 cases in the State of Colorado (up to 1,944 newly reported cases per day).  At the time, a statewide software solution for CI/CT was not available, and TCHD was quickly outgrowing data management solutions for COVID-19 surveillance activities.  The department was quickly scaling up its CI/CT teams and needed a technological solution for tracking all newly reported COVID-19 cases, allowing for completion of case interviews and contact tracing, and for monitoring surveillance and response metrics. 

Prior to the COVID-19 pandemic, TCHD had intentionally invested in informatics capacity, including conducting an agency reorganization that specifically called out informatics as a core service, conducting informatics-focused strategic planning, strengthening internal data governance, hiring and training a multidisciplinary team of informatics specialists and epidemiologists, and building a technological environment that readily facilitates public health informatics.  Previous infrastructure-related projects included implementation of a suite of ESRI products, including an internal GIS server, Survey123, and ArcGIS online, among other technology such as Tableau server and FME.  While this particular innovation is not evidence-based, the innovation answers well-documented recommendations for building informatics capacity in local health departments, particularly around innovative approaches to garner reliable data and utilization of GIS to fight COVID-19 (Khurshid, et al. 2020).

In May 2020, TCHD examined several options for COVID-19 CI/CT surveillance and data management, looking at off-the-shelf software products, common freeware used for outbreak management, and exploring options for customized software with and without vendor support.  We determined that utilizing existing infrastructure and leveraging existing staff expertise within the department was the best option until a statewide solution was available.  We utilized a project team of staff including infectious disease epidemiologists, population health epidemiologists, an informatics project manager, information technology staff, environmental health staff, and emergency preparedness staff to manage development of the software system.  Ultimately, the software system, which we fondly named CARL, was a complex collection of interdependent data workflows built upon an ArcGIS and Survey123 backbone (Attachment 1).  We also highly utilized the data integration platform called FME by Safe Software to facilitate various workflows within the software system.  Each workflow represented a unique set of tasks necessary for managing COVID-19 surveillance data.  These tasks included review and assignment of newly reported cases to various teams of CI/CT staff, review and assignment of newly reported contacts for contract tracing, case or contact re-assignment, referral of cases or contacts to our community referral staff, and the onerous task of review and resolution of duplicates.  Deduplication of COVID-19 data required several unique data work flows, and included the task of comparing newly-reported cases against existing cases, comparing newly-reported cases against known contacts, comparing newly-reported contacts against existing cases, and comparing newly-reported contacts against existing contacts.  Our staff also had to deduplicate electronic laboratory reporting data.  These deduplication tasks were necessary in order to reduce redundant work among CI/CT staff and to ensure that each individual received the most appropriate and timely isolation or quarantine guidance.  We used formal project management to ensure delivery on all defined business requirements.  We utilized a formal user acceptance-testing plan for each unique workflow and user interface.

The innovative development of our homegrown COVID-19 software system allowed for real-time data visualizations through the ESRI products.  As case interviews and contact tracing was conducted, information was updated in a host of publicly available data dashboards.  At one point, TCHD had over 30 publicly facing dashboards, including municipality-based, neighborhood-level, and school district-level dashboards (Attachment 2).  Our dashboards were used extensively internally and also by community partners such as municipal governments and school district superintendents.   The system also facilitated real-time data sharing through a series of representational state transfer application programming interface (REST API) where we shared select, de-identified data with critical governmental partners.  Our ability to depict the local COVID-19 pandemic geospatially and to share data in real-time with governmental decision makers allowed for us to implement and evaluate place-based interventions.  Our public information staff extensively used heat maps and other geospatial analyses to target COVID-19 prevention messaging.  These interventions included targeted and culturally relevant COVID-19 prevention messaging up to municipal and county-level policies such as mask mandates and curfews.

Early in the pandemic, we identified higher case incidence rates among our Hispanic and Black residents and higher incidence rates among our Adams County residents.  Among our three Counties in the TCHD jurisdiction, Adams County has the higher proportions of Hispanic/LatinX residents (Hispanic/LatinX 41%, Black 3%) compared to Arapahoe County (Hispanic/LatinX 20%, Black 10%) and Douglas County (Hispanic/LatinX 8%, Black 2%).  Within our jurisdiction, COVID-19 crude incidence rates by race/ethnicity vary greatly.  Hispanic/LatinX residents account for the highest crude incidence rate at 8,404 per 100,000 followed by White/Non-Hispanic at 3,020 per 100,000, Black at 1,907 per 100,000, Asian at 1,718 per 100,000, and American Indian/Alaska Native at 477 per 100,000.  We found that our Hispanic/LatinX and Black residents with COVID-19 experienced inequities in hospitalization and death rates.  Our homegrown COVID-19 software system allowed real-time data visualization to facilitate communication about these health inequities, allowing for crucial dialogue and problem solving around public health messaging and population-level interventions. 

Our innovation facilitated numerous activities aimed at reducing health inequities identified in our COVID-19 response.  Through this innovation, we created a dashboard for community testing metrics that looked at both incidence rates and testing rates together to examine areas where residents may have been lacking adequate COVID-19 testing resources (Attachment 3).  Additionally, TCHD holds four separate weekly or biweekly webinars for the business and faith-based communities in our jurisdiction.  Our data prompted the development of a weekly webinar for business leaders in Spanish-speaking communities within our jurisdiction with a community partner called Las Chingonas.  Our weekly business webinar in Spanish is our most well attended business webinar with approximately 1,000 views per week.

We had multiple goals of this innovation, including:

· To conduct surveillance on COVID-19 cases and contacts and to provide descriptive epidemiologic measures on these cases/contacts;

· To examine the occurrence of COVID-19 across demographic and geographic descriptors;

·  To conduct CI/CT activities within the software system with a large team (200+ staff) doing concurrent interviews and data entry;

·  To provide select, real-time, de-identified COVID-19 incidence and testing data to critical governmental partners via API;

·  To display critical epidemiologic information to the public and to community partners via public-facing data dashboards, stratified by various relevant geographic areas, including county-level, municipal-level, neighborhood-level, and school district level boundaries;

·  To inform population-level interventions including mask mandates, curfews, and business and school capacity decisions. 

We designed our COVID-19 software system with timely data reporting and data transparency in mind from the beginning.  Early in the pandemic, county and municipal governmental partners requested access to real-time data on both incidence data and on testing data via REST API.  At various points in software development, we reached out to critical partners including county and municipal governments, healthcare partners, emergency management partners, school districts, and testing partners to identify COVID-19 data needs and to determine how our innovation might answer these data needs and other critical questions.  Because we utilized ESRI products for data collection and data management, we were able to put together data dashboards that include geospatial elements to depict COVID-19 incidence, testing metrics, testing sites against common geographic boundaries.  TCHD actively provided data across many different geographic boundaries that helped to inform policies across various levels of government.  Many of our governmental partners created their own data dashboards using raw data from the REST APIs.  For example, the Cherry Creek School District was able to create a district-focused data dashboard for staff, parents, students, and the public to capture what all was known about COVID-19 incidence, testing, school exclusions, and school closures using raw data from TCHD in addition to data that the school district captured itself.

TCHD received feedback from community partners in various settings including weekly meetings with elected officials and county and municipal government partners, webinars with school district officials, and webinars with business and faith-based leaders.  Policy concerns evolved over the course of the pandemic, and the public’s and community partners’ interest in various topics changed with these policy concerns.  Our innovation was agile, and allowed us to make system changes around data collection, data analyses, and data visualization to best answer the evolving questions around the pandemic.

Our partnerships and collaboration with our vendors, ESRI and Safe Software, were critical to our success.  ESRI provided technical assistance to TCHD beyond their contractual obligations and helped to assure that our business requirements were met.  ESRI has provided several platforms for TCHD to share our successes with others in the GIS and public health communities.  Our collaboration resulted in the development of a prototype of a fully functional COVID-19 surveillance system that included functions for CI/CT activities, community referrals, and response performance management.  ESRI has been able to market to and to implement this product in other health departments across the United States.  We believe that our innovation has been helpful to other LHDs as the COVID-19 response continues to evolve.

This innovation enabled quality improvement in a number of different ways.  We reallocated internal resources, including staff and information technology infrastructure for the development of this software.  Our project team consisted of a multidisciplinary team of staff who often had other normal roles within the organization.  As the project progressed, many staff were able to develop new professional and technical skills, particularly in database management, data analyses and data visualization, and project management.  In order to maintain the innovation, we purchased ESRI field licenses for all of our CI/CT staff; even with the cost of these licenses, the homegrown software system was still much more cost effective than purchasing off-the-shelf or customized software directly from a vendor.  TCHD hired a handful of temporary staff to support the software system in fall 2020, including three COVID-specific data analysts.  Through this innovation, we learned that we are capable of solving complex data management problems with existing staff, tools, and resources and that investment in informatics capacity had major benefits to our department in terms of our ability to manage a multifaceted pandemic response.  Our innovation demonstrates to other LHDs that investment in informatics capacity and informatics skill development among staff is critical to handling future large-scale events.

As mentioned above, TCHD was able to adapt swiftly to adapt the software system to capture fields of high interest as the pandemic evolved.  In late September, we added fields to better categorize community exposures among cases and added additional questions regarding community resources needed among cases and contacts.  The software system enabled us to better address the social and economic needs through our community referral programs and facilitated instantaneous information sharing from the CI/CT teams to the community support staff.  Furthermore, an Equity Officer position within our Public Health Incident Management Team was developed to better address ongoing health inequities observed through the data captured in our software system.

Additionally, we used our homegrown COVID-19 software system for performance management throughout the response.  We defined performance metrics for our CI/CT teams that included: the number of newly assigned cases and contacts per day, the number of completed interviews per day, the number of investigations left in the queue for the day, the number of investigations needing reassignment per day, average length of time in queue, average number of contacts identified per case household, the number of investigations completed per day by each staff member, the number of cases and/or contacts who refused interview per day, and the number of CI/CT staff working per day.  We displayed these metrics on internal data dashboards for unit leaders within our Public Health Incident Management Team (Attachment 4).  We routinely used these data in improving upon our training modules for CI/CT, improving our written standard operating procedures, and implementing occasional team restructuring.  In sum, we developed a single, comprehensive COVID-19 software system that enabled numerous types of public health work through the innovative use of department staff, established informatics infrastructure, and implementation of best practices.