Informing Pandemic Preparedness Through a Digital Global Health Security Library

To assemble this database of research, oversight, guidelines, and plans, we first undertook an evaluation of preexisting digital databases relevant to the field. Based on expert recommendations, we identified a set of digital collections from which the research team gathered records. These collections included the World Health Organization Institutional Repository for Information Sharing (IRIS)⁷; the United Nations Digital Library⁸; United Nations Security Council meeting records⁹; the World Organisation for Animal Health Documentation Centre¹⁰; the US Government Accountability Office Action Tracker and Recommendations databases¹¹; the Congressional Record¹²; the Congressional Research Service¹³; and the National Academies of Science, Engineering, and Medicine Research Center (under the Biology and Life Sciences and the Health and Medicine topics)¹⁴; among others. To extract relevant records from these databases, we developed a standardized set of search terms to identify documents relating to global and national awareness of emerging infectious disease or other naturally occurring pandemic threats and the risk from those threats. These 26 search terms (Box) encompassed key areas of interest within the field, including critical activities related to preparedness (eg, “biosurveillance,” “medical countermeasures,” “medical supply chain security”), commonly used terms (eg, “emerging infectious disease,” “health security,” “pandemic,” “biodefense”), and select pathogens or diseases with pandemic potential (eg, “Ebola,” “coronavirus,” “influenza,” “MERS,” “SARS”). As we reviewed the records, we excluded items focused only on endemic diseases or those typically addressed by the global health (as opposed to global health security) community, such as HIV, tuberculosis, malaria, and noncommunicable diseases, as well as fact sheets on diseases. Resources related to bioterrorism and biosafety or biological accidents were only included if, in expert judgment, the item made a substantial contribution to understanding the global health security field or a specific pandemic threat.

Limitations to this systematic approach included the sheer quantity of results produced and sites that restricted batch exports or lacked text matching. For instance, IRIS hosts more than 200,000 global health records dating from the 1800s and is the most specific source available for global health documents.⁷ Only a subset of these, however, helps tell the story of global health security and readiness for outbreaks with pandemic potential. Furthermore, these documents are distributed throughout the expansive database and not readily accessible via the given IRIS filters. Searches using the predefined terms yielded thousands of results, although IRIS only has the capability to export 500 documents at a time. Given time and resource constraints, researchers were required to identify relevant Medical Subject Headings (MeSH)¹⁵ subject categories in IRIS, and manually review only the documents within those identified categories. Many of the same search challenges with IRIS persisted across other organizations' websites, reflecting the difficulty stakeholders may face when trying to work across and within these libraries to assess preparedness. Similar limitations for other sites necessitated the use of a subset of the keyword search terms to consult gray literature, including within United Nations organization websites (ie, United Nations General Assembly documentation,⁸ United Nations Security Council meeting records,⁹ Food and Agriculture Organization of the United Nations conference documents,¹⁶ World Health Assembly¹⁷) and the World Organisation for Animal Health Documentation Centre.¹⁰ To systematize the search for third-party reports (eg, nongovernmental organization, academic, private sector), we consulted the top 10 think tanks from the University of Pennsylvania Think Tanks and Civil Societies Program Global Go To Think Tank Index reports.¹⁸ Additional reports were captured through other methods, such as through Google searches for simulations and exercises, and review of reference lists within captured US Government Accountability Office reports. This step led to the capture of documents from organizations such as the American Hospital Association,¹⁹ the Bill & Melinda Gates Foundation,²⁰ the Johns Hopkins Center for Health Security,²¹ and the Transatlantic Biosecurity Network.²² We conducted a limited search of the peer-reviewed literature in Ovid MEDLINE,²³ parameterized to capture papers about coronavirus risk. Each record gathered includes a hyperlink to the source and, where permissible by the publisher and/or licensing permissions, a PDF of the publication. Detailed methods are available on the Health Security Net documentation page.²⁴

To ensure the documents gathered through the literature review are easily accessible and fully searchable for stakeholders, the research team developed a custom data taxonomy and data dictionary to define key metadata and organize the dataset. A total of 20 metadata elements were defined for each document, including information about each document's publication, funding, geographic scope, and applicability to global health security. A coder manually characterized each included document based on a detailed review of its content, including an original written description and hand-selected topic areas and tags. A second coder reviewed each document for internal consistency. If coders disagreed about the categorization of a document, a third researcher adjudicated the decision.

To aid end users in identifying broad areas of interest across the preparedness spectrum, topic areas were chosen that stem from the core elements of health emergency preparedness to prevent, detect, and respond to infectious disease threats (eg, strategic planning, threat/risk awareness, disease surveillance/detection, medical preparedness/emergency response, international aid/collaboration). Each record was categorized into a single topic area reflecting the predominant topic of that record; if multiple topic areas were applicable, the coder was instructed to choose the most relevant, and the decision was confirmed by a second coder. When no topic areas applied or the record was broad, coders chose the “Other” category. Combinations of topic areas were not supported in the interest of straightforward resource categorization, visualization, and exploration. However, the inclusion of tags and a search bar in the online tool help address search limitations imposed by the single-select topic area. Tags, or subtopics, were defined based on recurring themes identified in the literature review and support sort and filter of all documents for end users to refine the results. Multiple tags were applied to single resources to best characterize its contents. Finally, a search bar was added to Health Security Net to give users full flexibility in their search. Users can enter any term into the text field to search for that keyword in resource titles, descriptions, and the first 1,000 characters of an attached PDF if one is available. Topic areas, tags, and the search bar all exist independently of one another; researchers using Health Security Net could use any combination of these features to filter based on the level of specificity of their search. The full taxonomy is described in detail in the site's documentation and glossary of terms.

Finally, we developed an ontology to link public health events to documents with content about that event. We included all public health emergencies of international concern, excluding polio, and events not declared or predating the public health emergencies of international concern declaration process that we assessed to meet the definition of an event per International Health Regulations Annex 2. Documents were categorized by event based on relevance to the topic independent of date of publication. This structure enables researchers to review the activities taken during different stages of a public health emergency in the context of lessons learned and after-action reviews.