Mandatory accreditation is necessary, but not sufficient, for trusted research environments

AI Summary

The article argues that trusted research environments need mandatory accreditation, strong output checks, and technical controls to prevent data disclosure. It cautions that excessive restrictions can block socially valuable health research, obscure geographic inequalities, and cause harm through underuse of health data.

Jefferson and Smith are correct that the UK Biobank case demonstrates the need for stronger safeguards around trusted research environments (TREs).1 Mandatory accreditation, robust output checking, and effective technical controls should reduce the risk of uncontrolled data egress. But accreditation must assess whether TREs enable socially valuable research, not only whether they prevent disclosure.The Five Safes framework was designed to manage, not eliminate, risk.2 In practice, institutional risk aversion can mean that legitimate public interest projects are delayed or made unviable. For research linking health records to granular information about housing, neighbourhoods, and environmental exposures,34 requirements to aggregate place based data into coarse categories may reduce disclosure concerns, while obscuring inequalities and modifiable determinants. Such caution carries its own risks as health systems confront rising multimorbidity.5Non-use and under-use of health data also cause harm by leaving major public benefit unrealised.6 A technically secure TRE that trusted researchers cannot access affordably...