![[ Visit Client Website ]](images/banner.jpg)
BACKGROUND: Chronic disease surveillance has typically been limited to self-reported prevalence in national surveys or ad hoc analyses from administrative or clinical sources. Tracking changes in chronic disease indicators is challenging, with limited capacity to show temporal trends or analyze sub-county geographic units. Clinicians routinely capture data in their electronic health records (EHR) that present an opportunity for targeted and localized continuous chronic disease surveillance. Several federal agencies (e.g., FDA, AHRQ) and national institutes (e.g., PCORI) have demonstrated the feasibility of national federated query networks. Local public health data query networks to access, aggregate, transform, and visualize chronic disease related EHR data are needed to leverage and maximize recent health information technology investments.
METHODS: Colorado Health Observations Regional Data Service (CHORDS), a distributed surveillance network based on federated network technology, was established to share EHR data between healthcare providers and public health entities. Using an integrated registry approach, CHORDS combines: 1) a common data model, 2) a federated query service, 3) a regional data sharing governance model based on Data Use Agreements, and 4) customized data visualization tools. Each provider builds a local datamart, installs federated query client software, and links those components to receive and respond to data queries. Public health surveillance sites currently execute queries focused on tobacco use, BMI, and cardiovascular disease risk. Tabular results are stratified by geographic and demographic factors to support visualization through chronic disease scorecards, maps and dashboards.
RESULTS: Domain-specific query modules for tobacco and BMI surveillance were developed, deployed and tested across multiple sites representing more than 800,000 registered individuals. Each partner mapped, loaded and quality assured local EHR data to the normalized data model. Aggregated results were successfully returned from participating partner sites despite widely varying IT skills, resources, and infrastructure. Data obtained via the CHORDS network was used to develop and implement chronic disease dashboards that display critical chronic disease indicators and identify disparities using stratified analysis and GIS functionalities.
CONCLUSIONS: Cross-provider aggregation of EHR data allows for monitoring chronic disease prevalence and intervention assessment. Future plans seek to increase the number of sites contributing data, expand governance model, develop methods to deduplicate patients across sites, and add dashboard functionality and capacity for longitudinal assessment. This local monitoring effort leverages existing national models for distributed federated queries to promote data-driven decision-making for implementing and evaluating interventions.