This is the third and final deliverable from WorldFAIR WP07 on Population Health. Its primary aim is to provide a broad set of practical recommendations for health data producers in low- and middle-income countries who are seeking to make their data FAIR. This includes recommendations for: data documentation methods for those new to the process; harmonisation of data structures through use of a Common Data Model; standards for the publication of rich machine-readable metadata; methods for developing study packages to conduct federated analyses within and across domains; and the publication of FAIR Implementation Profiles.
This document is a synthesis of the discovery and work described in the two previous WP07 deliverables. As a reflection of the environment in which WP07 works, the methods proposed are open-source, freely available and frequently used. Some of them, such as the OHDSI/OMOP Common Data Model and the schema.org metadata standard, have grown in popularity in high-income settings and are underpinned by active user networks and freely available teaching resources. Their promotion here supports the argument for a coherent global standardisation of FAIR data methods, and this consistency promotes capacity building of the required data skills in lower-income settings that are integral to the adoption and success of FAIR data aspirations world-wide.
As well as reinforcing compatibility within the health data domain, WP07 has also chosen standards and methodologies that are compatible with other domains that comprise the WorldFAIR project. This approach is expanded in detail within WorldFAIR’s WP02 Cross-Domain Interoperability Framework (CDIF) which recognises that true interoperability of data – the ability to combine them in a scientifically rigorous way within and across domains – can dramatically enhance their value.
The intersections between population health, climate change and humanitarian crises are already acknowledged and within WorldFAIR itself: WP07 has clear links with urban health (WP08), social surveys (WP06), and disaster risk reduction (WP12). Such commonality in FAIR approaches across domains lies at the heart of what WorldFAIR seeks to achieve.
These recommendations are written for all health data producers in lower- and middle-income countries, regardless of their stage on the FAIR journey. Some may have little or no experience of data documentation; others may be ready to adopt a Common Data Model or publish machine-readable metadata; whilst others may be ready to conduct federated analyses. This document suggests recommendations for all, and these recommendations are consistent with those that would be made in higher-income settings.
The closing part of this document explores the burgeoning field of FAIR metrics – tools for measuring the extent to which FAIR has been achieved. This document does not add to this body of work because it takes the view that aiming for the FAIR principles by a data producer is such an individual and iterative process that following these recommendations, rather than applying an externally-generated metric, is more likely to achieve the goal of making health data FAIR.
The full report is available on Zenodo.
