# Data reuse in global health: perspectives from actors in policy, funding and research

## TL;DR

This paper explores how clinical research data is reused in global health, finding that data sharing alone is not enough to ensure impact, especially in low- and middle-income countries.

## Contribution

The study provides new insights into the practical and ethical challenges of data reuse and offers policy recommendations to improve data utility and trust.

## Key findings

- Secondary data analyses have influenced clinical guidelines and policy in low- and middle-income countries in a few documented cases.
- Shared data are often not ready for analysis due to incomplete metadata and under-resourced curation.
- Mistrust among data contributors limits data reuse and risks selection bias in analyses.

## Abstract

Data-sharing mandates from funders and journals have increased in recent years, but little is known about how shared data are used. Existing research has focused on access frameworks, with less attention to conditions that enable or hinder subsequent analyses and their impact on science and policy.

We conducted semi-structured interviews with 22 key informants with experience using clinical research data. Participants included researchers, policy makers and senior staff from funding and pharmaceutical organisations. Interviews explored motivations, ethical and practical challenges, and enabling conditions for reuse. Data were analysed thematically using a combination of deductive and inductive coding. Reporting follows the Consolidated criteria for Reporting Qualitative research framework.

Secondary data analyses have, in a few documented cases, shaped clinical guidelines and policy in low- and middle-income countries (LMICs). Individual participant data meta-analyses informed WHO recommendations for maternal and child health interventions, and analyses of COVID-19 data guided decisions at national and subnational levels in several countries. However, such cases remain uncommon. Secondary data users reported that shared data were seldom ready for analysis owing to incomplete metadata and under-resourced data curation. In academia, secondary analyses were driven by the potential for publication rather than health impact. Mistrust, particularly where data contributors feared reputational harm or exploitation, resulted in underutilisation of valuable data as analysts relied on a limited set of well-known or easily accessible datasets. This risks selection bias and limits the evidence base, especially for under-represented groups.

Mandating data sharing alone is insufficient to deliver impact in LMICs. Policies must be coupled with resourcing for data curation, efforts to avail machine-actionable metadata and incentives for impact-driven analyses. Equally critical is trust, built through recognition of contributors and equitable, transparent benefit-sharing between analysts and data generators.