OpenSAFELY

OpenSAFELY performs secure analysis of electronic health records (EHR) to enable large-scale epidemiological studies using NHS England patient-level data.

Key Features:

• Data Integration and Security: Links primary care EHR with external datasets including death statistics from the Office for National Statistics and hospitalization data from the NHS Secondary Uses Service while maintaining stringent data privacy standards.
• Scalability and Efficiency: Processes data from millions of patients, leveraging patient-level data from over 17 million individuals and supporting analyses reported on datasets covering over 58 million records during the COVID-19 response.
• Cohort Study Support: Enables cohort study designs to assess health outcomes such as hospitalization and mortality across defined populations, including adults and children in England.
• Comprehensive Subgroup Analysis: Supports detailed subgroup analyses for conditions such as Down syndrome and cerebral palsy and for individuals on general practice learning disability registers (LDR), with adjustment for confounders.
• Outcome Measurement: Facilitates examination of primary outcomes (COVID-19-related deaths and hospitalizations) and secondary outcomes (non-COVID-19-related mortality).

Scientific Applications:

• Assessment of COVID-19 risk in people with learning disabilities: Quantified increased risks of COVID-19-related hospitalization and mortality among individuals with learning disabilities, including estimation of hazard ratios and higher risks for severe or profound disabilities.
• Identification of high-risk subgroups: Characterized differential risks across subgroups such as those with Down syndrome, cerebral palsy, and individuals listed on learning disability registers to inform prioritization in public health studies.

Methodology:

Links primary care EHR to Office for National Statistics death data and NHS Secondary Uses Service hospitalization data; implements cohort study designs with subgroup analyses and adjusts for confounders including age, sex, ethnicity, and geographical location to evaluate outcomes such as COVID-19-related deaths, hospitalizations, and non-COVID mortality.