By Dr Tom Foley.
Kingsley Manning is Chair of the Health and Social Care Information Centre. He was previously Founder and Managing Director of Newchurch Limited, Executive Chairman of Tribal Group’s health business and Senior Adviser at McKinsey & Company.
HSCIC is the UK national provider of high-quality information, data and IT systems for health and social care. (http://www.hscic.gov.uk/about-us)
Data for operational management, such as for simple rotas, logistics, production management, flow management, etc., in the NHS could be a lot better, but there appears to be little interest in these areas. “Why are we always surprised when the day after New Year is the busiest in A&E?” In these cases routinely collected data could make a tremendous difference to our planning. Routine data would also enable the use of research methodologies for understanding comparative performance and causal relationships for different outcomes in different contexts. This would challenge the pseudoscientific notion of much of what is currently done in health management and health economics.
“This could challenge the current view, held by many researchers and clinicians, that we can only trust answers if we have a Randomised Controlled Trial and three decimal points.” “In real life, we say, shop A is doing well and shop B is not doing well, so what is shop A doing that shop B isn’t.” We can then try different approaches and try different hypothesis in real time. This would have enormous potential for improving quality and safety. Poor care causes significant harm and better routine data could allow a much quicker management response.
Changing role of the clinician
The concept of the independent practitioner who relies solely on their own judgement is outdated. For example, in the airline industry, every aspect of the aeroplane is monitored in real time and systems are designed to support the pilot.
There is now less focus on the individual skill of the clinician. With the increasing data available to clinicians, including tests, phenotyping etc. the skills required for diagnosis are changing. The clinician can now logically follow through the data pathway to come to an appropriate outcome. We have seen that without this data that there is significant variation in diagnosis. For example if you give 25 GPs a patient with chest pain you will end up with significant variation in the diagnosis.
Priorities may need to change. The current system encourages clinicians to focus on complex cases whilst from a health economist point of view it would be better to focus on the common problems which represent the greatest burden on the system.
Historically the brightest people have been recruited into medicine on the basis that it is as much an art as it is science, however with the increasing knowledge and clinical pathways this it is becoming less of an art. “If we just did what it says in the book, most of the time, the outcomes would be better.” This often results in doctors becoming bored. Evidence suggests that physician assistants are both effective and safe and can deliver 70-80% of what a GP does. They just need to recognise situations that are not routine, a then to seek a second opinion.
This occurs in most other professions. “Technology changes how we work. Nobody is assumed to know everything anymore.“ For example, the current workforce largely qualified before the emergence of the smart phone and mobile technologies, but now the penetration of smart phones is over 60%. The availability of this data changes the role of the clinician towards someone who interprets and uses a rules based approach to diagnostics rather than someone who relies on a judgemental approach.
Medical education should change. It should become postgraduate, shorter, more modular, with an emphasis on lifetime learning. “We should do away with the division between doctors and nurses. We should have a spectrum of ongoing training with much more flexibility to the changing needs of the service.”
Routine data is often of poor quality and the types of data that are collected also vary significantly. It evolved for other uses, public health, then payment. We collect financial and prescribing data from primary care but clinical data is not aggregated, for secondary care there is less prescribing data and A&E data is modest. Community data is largely overlooked and social care data is non-existent. Incorporating personal data into this would be a huge undertaking. There is also the concern of how to improve this data without additional burden to the provider.
This gives reasonably good data on admissions, however it is poor for outpatients and A&E attendance. The quality of what is recorded is also very mixed. For example, the error rate in HES data is around 20-30%. Currently around 70% of hospital records are still handwritten, so does HES data accurately represent what happened? Data is normally not up to date and is often received weeks or months after an event. There should be an aim to speed up this process.
HSCIC has a statutory obligation to consider the burden that it places on providers. HSCIC accounts for 25-30% of data flows within the system currently. There are many local adhoc data flows and the rest of “the centre” account for a proportion of specific data transfers. Even in the US, most of the data flows through a small number of intermediaries. HSCIC along with NHS England recognise the need to consolidate those data flows and is in the process of creating a national information service called the Data Service Commissioning. This will attempt to consolidate all of these data flows and extend the research community. Data should be collected once and then distributed appropriately. This system will be enforced through contract and statutory powers.
Data flow to HSCIC is standardised so for example HES data is in the same format irrespective of the provider or system that it came from. Providing the output is in a standardised format and structure this is not a problem.
The coding structures used by providers can create major challenges. For example, there are problems linking primary and secondary care data with different coding structures. You might receive an intervention code from secondary care and a diagnostic code from primary care but there is nothing to link the two together, or an outpatient event that has no coding at all. These definitional issues make connectivity of data very challenging. There are major concerns about the degree of interoperability from this point of view.
There is a role for HSCIC, to try to develop consistency, as the standards body for the industry.
HSCIC cannot tell hospitals what to do within their own system but can specify the format, if they want to exchange data outside of their own system.
HSCIC is establishing standards for data exchange and reporting and this will be based on the Academy of Medical Royal Colleges documentation for interchanging data.
In the future wearable technologies may provide huge volumes of data. There are already vast quantities of data that no one ever looks at. “I am always surprised by how little use is made of our data.” We have scanning data going back 10 years and it is rarely accessed. We publish 10 million lines of prescribing data every month and much more use could be made of it.
Wearables raise very interesting issues, for example the apple watch might measure blood pressure, oxygen levels, blood sugars etc. Apple claim that they will not have any access to this data, but that is unclear. I don’t think that we could describe this type of data as routine – it needs a different terminology. “Then the question becomes what you do with a year’s blood pressure data for a middle aged man who takes his BP every half hour, thousands of data points. Perhaps it could be very meaningful but it could also be meaningless.”
Crunching all of these numbers may produce associations that previously we were not able to find. In the past, we focussed on population level data, but this would move towards cohorts. HSCIC could potentially provide the context, the mechanics and the safe haven where this work could be done.
There will be an abundance of data available and we need to think about the validity of this data, how it is used and who owns it. For example, a GP is unlikely to be able to make sense of an electronic health diary in the short time that they have for their consultation, unless it is presented to them in a meaningful way. There is an interesting issue about risk and reward with this. For example, if somebody with a known heart condition suddenly develops an arrhythmia, does the system automatically call an ambulance or does it simply continue to collect the data for review later. This is about appropriate response or usage of this type of data. Another example is the use of foetal monitoring in the US, where insurance companies provide mothers who are prone to premature delivery with foetal monitoring systems. If they can prevent a premature birth, then there is a better outcome for the baby and in turn, a better financial outcome for the insurance company.
Privacy and confidentiality
The European attitude towards privacy is even stronger than here. The US is very different due to the nature of their healthcare system, where it is highly fragmented and distributed and there are commercial relationships in many cases. For example, you may need to share your information with the insurer in order to gain insurance. That type of relationship would be illegal here under data protection legislation that makes it illegal to coerce people into providing data.
In the UK, there has been a loss of confidence in established authorities, with the level of confidence in the NHS to steward their data appropriately falling as low as 40 or 50%. The public no longer trusts the NHS, on this issue, as it did historically. People are now much more aware of the capacity for this data to be used and moved around.
The nature of our society is increasingly autonomous, with a move towards individualisation. The Secretary of State has said that the individual owns their own data. This can be contrasted with how easy or difficult it is to get your own medical records from a hospital or GP.
There are increasing numbers of commercial organisations offering to host medical records of individual patients. Some members of the public are very interested in this and others are appalled. It is unclear at present what the appropriate response to this should be. There are major cultural and political issues around the commercial interests in healthcare. There is a very powerful lobby that thinks that it is inappropriate to profit from healthcare. This is often the same group that is very vocal with regard to the data issue.
There is significant sensitivity around the linking of health data to that of employment or benefits. Researchers often become frustrated with the lack of access to data but they often do not understand the intricacies of data protection.
It is often not made clear what benefit the individual would gain from sharing their information. It is sometimes assumed that government wants this information so that they can spend less money on healthcare.
There are also huge issues surrounding consent and privacy. The individual is going to have to have much greater control over their own data. Potentially, there will be a lot of people who will share their data, particularly by those who take a calculated risk to share their information on the basis that it may aid a diagnosis or improve treatment. There may also be a recognition by organisations that they have to take their privacy responsibilities very seriously as any failure in this duty would likely impact very negatively on their business.
The European parliament position is currently that we should be moving towards an opt-in consent process for every stage of every consultation. This would require regular consent to ensure it is still valid. Ministers will need to consider the implications of this.
There is also a need to demonstrate value from using this data. The Health and Social Care Act has meant that the rules on dissemination of data have been tightened up. Researchers now have to demonstrate what value their work will add. The HSCIC receives a lot of requests for data, but some of these have to be declined because it is difficult to see the value or benefit in the research.