Peter Bull and James Mucklow
18 August 2014
A colleague of ours visited his out of hours doctor service because of a painfully swollen knee, a condition he’d had once before. While there, he asked whether they could see any useful information from a previous GP visit for the same condition. They could not access the information.
The reason was not technical: GP systems store vast quantities of detail about your health electronically. It is because the information governance hurdles that GPs have to leap over to share the data between services are high and few have the patience or motivation to try.
If it is too hard to share your records among the various people and organisations who need to treat you, care will be poorly coordinated and more expensive because of wasteful duplication. It may also be unsafe because important facts will not be accessible, such as an allergy to a specific drug. Most patients assume that data relevant to their healthcare will be shared across the NHS and are frustrated and surprised when it is not.
Of course, confidentiality is absolutely vital, and should always be protected with strong rules and systems, but it seems unreasonable to let fear of risk jeopardise our ability to coordinate and improve care.
We should remember that central databases have a strong historic track record of using data appropriately. And, while there has been much public concern about care.data – the national programme to share NHS patient data – the data is usually pseudonymised to protect patient confidentiality.
Our work with hospital episode statistics (HES) data, which covers hospital activity, has shown that technology can make finding patterns in large health datasets both fast and easy. This offers a whole range of opportunities to drive improvements in the NHS and lead to better care.
One of the particular issues we examined is whether over-prescribing certain drugs could have serious side effects. Although clinical trials should tell us this, they are only done on small numbers of patients and pharma firms do not always release all the data.
This means many side effects don’t show up until a large number of people are taking the drug and some new blockbuster drugs have caused problems. For example, Vioxx was launched as a major breakthrough in pain relief but was eventually withdrawn because of concerns about a possible link with heart disease.
We looked at prescribing a class of anti-ulcer drugs called PPIs. We can tell from public prescribing information that some GPs use this a lot more than others. This happens because the drugs are now used for a variety of mild digestive conditions, but is this a good thing? Small studies had indicated that the drugs compromised the immune system when overused and led to patients contracting serious conditions like pneumonia. By using available NHS data we were then able to look at the pattern in the 250,000 per year emergency admissions from pneumonia for each GP practice and compare it to their prescribing rates, and that work appears to highlight a potentially serious side effect of over-prescribing.
This potential link could be investigated further if the full care.data programme was up and running. At the moment we can only show an association between GPs with high PPI prescribing and hospital admissions. If we had the GP prescribing records we would know whether these admissions were individuals actually taking PPIs. The analysis would be easier, faster and the results more robust.
If care.data were available now, we could build a system to monitor the effectiveness and the side effects of all new drugs. We could pick up problems like Vioxx in weeks, not years, and save many lives. We could learn whether long standing drugs really work and we could resolve which of the many choices GPs make about treatments are effective.
A further problem is that the current system underuses the data it currently has. We routinely find that major hospital reconfigurations plans are six months into the process without having looked at the national HES data about what is currently happening to patients. We see government spending £500m to resolve the A&E performance crisis in ways that do not address what is actually happening in emergency departments. We see enormous differences in the prescribing patterns in GP practices but almost no use of the national data to help them identify where their prescribing could be improved.
We need strong, transparent controls on confidentiality that protect individuals. Furthermore, if anyone’s fears about confidentiality are not assuaged by these controls, we need to make it easy for them to opt out. But the risks to confidentiality must be balanced against the potential gains to our health. Trepidation should not prevent us from achieving those gains.
For our thinking on data-driven insight click here or contact us now.