This is a news article that UPD first published on 25th May 2018. 

On the 25 May 2018, the new national data opt-out was made available. It gives everyone in England a choice about whether their confidential patient information (CPI) is used beyond their individual care. But if we all start opting out what will happen? What types of health research and NHS planning can no longer lead to conclusive results? And why is it important that research and planning are based on large and diverse datasets? 

If I opt-out, what will happen? 

If you opt-out, your patient data will still be shared for your individual care, or where you have specifically given consent. 

But your CPI will no longer be used for improving health, care and services through research and planning. 

Effectively, you won’t be represented in some types of planning activities to help make the NHS run better, such as audits and shaping local services. Nor will you be represented in some types of research studies working to find new treatments or monitor patient safety. 

Why does this matter? 

It is important that data used for research and planning is as representative of the population as possible. It leads to more accurate studies that have enough statistical clout to draw useful conclusions. 

If lots of people opt-out, when NHS commissioners or researchers investigate the population’s health, there will be times when certain groups of people or demographics might not be accounted for. In addition, many conditions and drug side effects are rare. Therefore, it’s important to have comprehensive data to firstly spot rare occurrences, and then be able do something about it. 

Data also needs to represent diversity: data that represents people from different locations, age groups, ethnicities and socio-economic backgrounds. For example, Jewish women are much more likely to carry the BRCA mutation (the most common genetic cause of breast cancer). If more women of Jewish ethnicity opt out, it would be harder to determine whether opportunities for early diagnosis were being missed, as the data from this group would be sparse. 

There is worrying evidence that certain groups are more likely to opt out. Previous opt-outs were highly ‘clustered’ with particular groups opting out often by age, geography or ethnicity. For example, the UK’s national heart attack audit is used to make sure operations for heart attack are used consistently, but if lots of people opt out from one area, the NHS couldn’t determine whether everyone receives the same quality of care. 

Those that opt out may have different characteristics, different health needs and different health outcomes than the overall population. This means that data risks being biased, as some groups aren’t represented in the data. Sometimes, even knowing who has opted out isn’t enough to make the results conclusive, and it’s often difficult to correct biases by using statistical techniques alone. Research then becomes less useful as it doesn’t reflect the true make-up of a population. 

Research that would be affected 

High opt-out rates would affect these examples:   

  • Understanding disease: for example, better understanding blood cancer. CPI from every person with blood cancer in Yorkshire has been used by doctors, researchers and health service planners to gain better insights into the disease. The results have allowed doctors to better tailor treatments as they know who will, and will not, respond. Improved symptom guidelines were produced, helping GPs decide whether or not to refer someone to a specialist.    

  • Improving treatments and prevention: for example, improving the treatment of bacterial meningitis. Babies who contract bacterial meningitis very young have a one in ten chance of dying. A studying using CPI showed that infection rates has not fallen for twenty-five years, despite measures being put in place to prevent it. Clearly indicating a stronger prevention strategy is needed. 

  • Ensuring patient safety: for example, using GP record data to demonstrate the safety of the whooping cough vaccination in pregnant women. The more people that opt out, the less reliable the data is for checking the safety vaccines.   

  • Evaluating policy: for example, updating the way the NHS screens for childhood visual impairments. This was a direct result of a study using CPI showing risk for children depends on their ethnicity, birthweight, levels of deprivation and other early life factors. Doctors also use the results to plan services for children. 

  • Planning NHS services: for example, planning hip fractures. The National Hip Fracture Database allows hospitals to assess their performance and plan where they may need to improve. Leading to better patient experience and cost savings for the NHS. 

Why can’t anonymised data always be used?   

The vast majority of research and planning uses anonymised patient data and would therefore be outside of the opt-out. However, there are some types of research that rely on CPI to: 

  • link data: For example, CPI was needed to link hospital and GP records in a study revealing that women from more privileged socio-economic backgrounds have a higher incidence of breast cancer, but those from less privileged socio-economic backgrounds have poorer survival outcomes. 

  • identify and recruit participants: For example, the UK Collaborative Trial of Ovarian Cancer Screening was set up to investigate different ovarian cancer screening methods. Of the 1.2 million women invited to take part, 32 women raised any concerns about being contacted. 

In all these cases, and those above, the use of CPI enabled the NHS and researchers to understand more fully what factors contribute to disease or health outcomes than would be possible with only anonymised data.   

What happens next? 

The more people that opt out, the less reliable the data. If the opt-out level remains low, the types of examples described above are still possible. But if the percentage of people opting out steadily increases, there would come a point where these studies would be compromised or could not happen at all. 

Find out more about the national data opt-out

Read about why the preferred approach is an opt-out, rather than opt-in or consent.