Article type
Year
Abstract
Background: Malnutrition, in all its forms, affects every country, wealthy people, poor people, and most of the world’s population at some point from infancy to old age. Given the exponential growth in the number of systematic reviews about nutrition and the complexity of skills needed to identify them, new approaches are required. Aligned with the Epistemonikos Foundation’s vision, a dedicated database of all nutrition-relevant synthesized evidence would help bring evidence closer to those who need it. This initiative aims to support access to and timely production of synthesized evidence to enable rapid-learning and decision-making to reduce malnutrition.
Objectives: To share our approach, progress and lessons learned in planning, curating and building the nutrition database
Methods: The collaboration was coordinated and established by the Epistemonikos Foundation, with researchers, information technologists and nutrition experts from four countries contributing. For the database, we developed draft review eligibility criteria. The potentially eligible articles being screened come from two sources: traditional search in the Epistemonikos database and reviews in the Living OVerview of Evidence (LOVE) platform. LOVE is a platform that retrieves all evidence hosted in the main Epistemonikos database and classifies the information using artificial intelligence algorithms. For the traditional search, we developed a sensitive and complex boolean search strategy to identify potentially eligible reviews. For calibration across multiple collaborators, screening was piloted in a sample of retrieved records, and eligibility criteria further enhanced, particularly in areas of likely uncertainty. Two screeners independently assess the retrieved records, consulting a third if needed. For the second source, we are developing a LOVE for each relevant nutrition topic.
Results: The comprehensive final search yielded >60 000 records. Pilot screening was done on a sample of records and final detailed eligibility criteria compiled, particularly . Screening is ongoing, coordinated through Red de Nutrición Basada en la Evidencia (RED-NuBE), the Centre for Evidence-based Health Care, Stellenbosch University and Epistemonikos, and database architecture and design is in progress. Screening for a pilot nutrition LOVE showed high sensitivity. Simultaneous screening via the creation of relevant nutrition LOVES is ongoing. Records included in LOVES are automatically removed from the database search yield.
Conclusions: We are working to establish an up-to-date, interconnected ‘one-stop shop’ of synthesized nutrition evidence and LOVES to enable timely data- and evidence-informed changes at all levels of health systems to address the universal malnutrition burden. The systematic simultaneous screening approach to building the nutrition database and LOVES strives to be efficient, while upholding high methodological standards.
Patient or healthcare consumer involvement: No direct patient or healthcare consumer involvement.
Objectives: To share our approach, progress and lessons learned in planning, curating and building the nutrition database
Methods: The collaboration was coordinated and established by the Epistemonikos Foundation, with researchers, information technologists and nutrition experts from four countries contributing. For the database, we developed draft review eligibility criteria. The potentially eligible articles being screened come from two sources: traditional search in the Epistemonikos database and reviews in the Living OVerview of Evidence (LOVE) platform. LOVE is a platform that retrieves all evidence hosted in the main Epistemonikos database and classifies the information using artificial intelligence algorithms. For the traditional search, we developed a sensitive and complex boolean search strategy to identify potentially eligible reviews. For calibration across multiple collaborators, screening was piloted in a sample of retrieved records, and eligibility criteria further enhanced, particularly in areas of likely uncertainty. Two screeners independently assess the retrieved records, consulting a third if needed. For the second source, we are developing a LOVE for each relevant nutrition topic.
Results: The comprehensive final search yielded >60 000 records. Pilot screening was done on a sample of records and final detailed eligibility criteria compiled, particularly . Screening is ongoing, coordinated through Red de Nutrición Basada en la Evidencia (RED-NuBE), the Centre for Evidence-based Health Care, Stellenbosch University and Epistemonikos, and database architecture and design is in progress. Screening for a pilot nutrition LOVE showed high sensitivity. Simultaneous screening via the creation of relevant nutrition LOVES is ongoing. Records included in LOVES are automatically removed from the database search yield.
Conclusions: We are working to establish an up-to-date, interconnected ‘one-stop shop’ of synthesized nutrition evidence and LOVES to enable timely data- and evidence-informed changes at all levels of health systems to address the universal malnutrition burden. The systematic simultaneous screening approach to building the nutrition database and LOVES strives to be efficient, while upholding high methodological standards.
Patient or healthcare consumer involvement: No direct patient or healthcare consumer involvement.