select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
Scientists comment on the expansion of the OpenSAFELY data platform.
Prof Andrew Morris, Director of HDR UK, said:
“OpenSAFELY is an excellent example of what is possible when we get health data right with the confidence of patients, the public and health professionals. Access to comprehensive GP data across all of England is a great step forward for safe and approved research. GP data offers greater breadth and depth than hospital data, providing a detailed picture of people’s health over time. Many common conditions, like arthritis, depression and back pain are mostly managed by GPs, so this data is vital for research that can improve care for millions.
“The OpenSAFELY platform is one that proved its worth during the pandemic, giving us much needed knowledge about COVID-19. It permits researchers to work with the information the data provides – while preventing them from accessing the data itself. Now by moving beyond COVID-19, researchers will be able to uncover groundbreaking insights that can improve the health and well-being of countless individuals. Significant challenges remain – the system is still evolving, with much work still to be done. But as OpenSAFELY and other initiatives show, the UK has both the skills and the will to make it work.
“The UK has long been a global leader in health data research. But to stay ahead, we must make coordinated investments in secure data infrastructure if data driven research is to power improvements in patient care, public health, NHS efficiency, clinical trials and enable medical discovery. This includes secure data sharing with flagship programmes such as Our Future Health, UK Biobank and Genomics England.”
Professor Sir Rory Collins, Principal Investigator and CEO of UK Biobank, said:
“The expansion of OpenSAFELY should be welcomed as it enhances an innovative and useful tool for health researchers working on GP data. However, the most significant leaps in scientific discovery will come from comparing many different types of data simultaneously, and at scale. For example, the 20,000 researchers who use UK Biobank can analyse over 10,000 variables on many of our 500,000 volunteers, with whole genome sequencing being just one of those.
“It is this ability to study the genetic, imaging, lifestyle, secondary and – soon – primary care data in combination that is so vital for research. That’s why we’ve seen over 14,000 peer-reviewed papers published using UK Biobank data, including developments that should lead to better diagnostics and treatments for conditions such as diabetes, dementia and heart disease.
“GP data is a critical national asset, and both researchers and patients will benefit from this expansion. The next step is adding consented GP data to larger datasets, and we at UK Biobank are delighted to be working with NHS England to add the de-identified primary care data of our 500,000 volunteers.”
Prof Sheila Bird, Honorary Professor, University of Edinburgh’s College of Medicine and Veterinary Medicine; and Visiting Senior Fellow at the MRC Biostatistics Unit, University of Cambridge, University of Cambridge, said:
“Dr. (now Professor) Ben Goldacre, a physician by profession, was first to receive the Royal Statistical Society’s Award for Statistical Excellence in Journalism for his Bad Science column in the Guardian.
“Professor Goldacre, who authored the Goldacre Review in 2022 [1] is against Bad Science. But he is staunchly for properly-approved record-linkages which respect patient confidentiality: and his team at OpenSafely have worked, during SARS-CoV-2 and since, to deliver just that. The delivery is a work in progress, as the excellent video about OpenSafely makes clear. Hence, my comment is about elements of enhanced delivery.
“First, as the Royal Statistical Society has argued for since swine-flu in 2009/10, the public – and OpenSafely – need legislation to end the late registration of fact-of-death in England, Wales and Northern Ireland. Only in Scotland, in our dis-United Kingdom, is fact-of-death registered, by law, within 8 days of death having been ascertained. OpenSafely for E&W urgently needs prompt and proper registration of fact-of-death which – for inquest deaths – is delayed by months or years [2].
“Second, since one of five deaths aged 5-44 years in E&W is not registered for at least 6 months [2], ending the late registration of deaths is essential if we are to learn by OpenSafely’s research how to prevent or reduce premature mortality such as deaths due to suicide or addictions.
“Third, analysts – including biostatisticians such as I – need to know in more detail about the random generators that OpenSafely uses for creating its pseudo-data, on which, as a biostatistician, I would develop and test my analysis routines. In particular, real data are often more complex in structure than statistical approximations to them in terms of their distribution (eg lognormal distribution assumed but the actual ln-data are not normally-distributed) or correlation structure. Analysts typically need to check assumptions on real data but may be writing checking-code based on approximations. For the checking-code to be incisive enough, analysts may need to understand in some detail the “random generation” processes.
“Fourthly, enhancements to OpenSafely may lead to important evolution in how some data are recorded by general practitioners. For example, when Gao et al. used record-linkage within Scotland’s safe-haven to analyse the methadone-specific death-rate and other opioid-related deaths in Scotland’s Methadone Client Cohort (2009-2015)[4], we found that the available data were quantity of methadone prescribed (not daily-dose) and reimbursement date (not prescription end-date) because those quantities were the data needed to audit the reimbursement of pharmacists[5]. By contrast, guidelines on safe prescribing of methadone are written in terms of daily-dose!
“Finally, the precautions built-into OpenSafely may mean that patients who registered objection to the use of their GP-data by care.data or the subsequent attempted grab during SARS-CoV-2 (which also failed) may wish to re-consider their objection. How does one do so?
This was announced at an SMC Press Briefing, and was accompanied by a funding announcement from Wellcome. The embargo lifted at 11:30am on Wednesday 19th February.
Declared interests:
Prof Andrew Morris “Andrew Morris is Director of Health Data Research UK, the national institute for health data science; is Professor of Medicine and Vice Principal at the University of Edinburgh; is President of the Academy of Medical Sciences, has minority (<1.5%) shareholding in Aridhia Informatics and a small number of shares in GSK (<£5,000). He receives renumeration in his roles as a member of the Novartis industrial council and as a member of the Clinical Research Data Sharing Alliance (CRDSA).”
Prof Sir Rory Collins “I am CEO and PI of UK Biobank, which is a Charitable Company established as a Joint Venture by the MRC and Wellcome. I have been in that role since September 2005, seconded 60%FTE from the University of Oxford where I am Head of the Nuffield Department of Population Health (which, along with other research organisations globally, benefits from using the UK Biobank – without any preferential access – for health-related research that is in the public interest).”
Prof Sheila Bird “has 30-years of experience of confidential record-linkage; & leads for Royal Statistical Society on need for legislation to end late registration of fact-of-death in E&W and Northern Ireland.”
This Roundup was accompanied by an SMC Briefing.