TL;DR
The COVID-19 pandemic highlighted critical gaps in public health data sharing infrastructure, hindering rapid response. This paper outlines lessons learned and a framework for building a "learning digital public health data architecture" to enable efficient, multi-scale data sharing and collaborative analytics for current and future public health crises.
What Problem Does This Paper Address?
The COVID-19 pandemic underscored the urgent need for robust public health data infrastructure capable of supporting rapid, multi-scale responses. Despite past public health crises, consistent data sharing agreements and transactional standards are often lacking between institutions. This fragmentation hampers the creation of a foundational infrastructure for effective data sharing and integration, which is crucial for advancing public health. The paper addresses the challenges of achieving this, particularly in the context of a rapidly evolving pandemic, where timely, data-driven insights are paramount.
Methods
The authors describe their experience in addressing data sharing challenges during the COVID-19 pandemic, focusing on practical issues encountered when seeking to integrate data across local, regional, and national levels. Their approach involved:
- Identifying and filling technology "gaps": Repurposing existing technologies and platforms (like EHRs and REDCap) to enable rapid data sharing and collection, followed by iterative improvements.
- Collaborative design of data sharing requirements and transmission mechanisms: Engaging with public health departments to identify shared data elements and actionable transfer formats to simplify and systematize requests.
- Facilitating cross-domain discussions: Involving legal and research compliance professionals to navigate regulatory hurdles and establish pathways for data sharing.
- Establishing or participating in multi-institutional convening or coordinating activities: Creating regional data "commons" and exploring statewide data aggregation efforts.
Datasets
Not discussed in the paper. The paper focuses on the process and framework for data sharing rather than specific datasets used for training or evaluation.
Key Insights and Ideas
- The "real world" lag: The CDC's strategic plan for surveillance, emphasizing data standards and reduced reporting burden, has not been fully realized, necessitating a focus on learning from the current epidemic.
- Fragmentation is a major hurdle: Geopolitical fragmentation (e.g., across multiple counties and states) creates significant challenges for data infrastructure and consistent data flow.
- "Honest broker" model: Establishing a trusted intermediate entity (like Washington University in their case) can significantly streamline data sharing agreements by reducing the number of pairwise agreements required.
- Minimum data sets are crucial: Defining a "minimum data set" for data sharing efforts can reduce the burden on healthcare provider organizations, especially during crises when they are inundated with requests.
- Repurposing is key for speed: In urgent situations, repurposing existing technologies and platforms is more effective than developing entirely new systems.
- Harmonization is unglamorous but essential: The "unglamorous work" of assessing, adjudicating, and cleaning data elements is critical for ensuring data quality and enabling reliable analytics.
- Standards evolve: The emergence of new health threats like COVID-19 can highlight a lack of established data standards (e.g., LOINC, ICD-10), requiring significant effort to harmonize data from various sources.
- Building a "learning digital public health data architecture": This conceptual model, deployable across multiple scales, is presented as a foundation for iterative data collection, integration, analysis, and action.
Key Results
The paper doesn't present quantitative results in terms of specific performance metrics. Instead, it highlights qualitative outcomes and lessons learned:
- Successful establishment of a regional COVID-19 data "commons": This facilitated operational and research use of data at local, regional, national, and international levels.
- Streamlined data sharing agreements: Through the "honest broker" model and templated documents, the complexity and burden of data sharing agreements were reduced.
- Identification of key challenges: Inadequate technology/data infrastructure, duplication of data requests, inconsistent technical requirements, and insufficient coordination were identified as major obstacles.
- Development of a conceptual framework: The paper proposes a "learning digital public health data architecture" as a scalable and sustainable approach for future public health responses.
Limitations and Future Work
- Not a comprehensive evaluation: The authors acknowledge that their experiences, while likely universal, are not a complete evaluation of all data sharing issues.
- Scalability of statewide efforts: While they investigated statewide data aggregation in Missouri, they found that not all Health Information Exchanges (HIEs) had the capability to perform the necessary data extraction and aggregation, indicating a need for scalable solutions.
- Need for higher levels of funding and coordination: The authors emphasize that addressing national-scale data sharing challenges requires significantly more funding, coordination, and attention from policymakers, industry, healthcare providers, and researchers.
- Future work: The paper calls for a robust community-wide dialogue to determine the steps needed to ensure regional and national healthcare systems can learn rapidly and respond effectively to emergent public health crises. They suggest sharing best practices and resources across entities and potentially establishing centralized hubs for data management and analytics.
Summary updated