The precision of genome sequencing (e.g. the 1000 genome project) is extraordinary, and it is now also remarkably ‘cheap’. Randomised clinical trials of pharmaceuticals are rigorous. However, what use is genomic data without phenotypic data to understand its meaning? How can you evaluate the true benefit of pharmaceuticals in ‘real life’ without using data from populations who are different in profile from original trials and who will generally be taking many different medications? Hence the rush for big data using data aggregated from electronic health records in the NHS.
In a criminal court, the jury must decide whether or not a defendant is guilty. The prosecution and defence present evidence to prove their points. They argue about whether evidence means what it appears to mean. The forensic teams of police departments go to great care to ensure that evidence is true, valid and is what it appears to be. The judge rules on what is permissible and what is not.
In health care and clinical research, the evaluation of evidence is determined by statistical analysis, acceptable only if the likelihood of it being due to chance is less than 1,000 for example. Curiously the same scrutiny is not always applied to whether the source data actually means what it appears to mean – that it is accurate, reliable and valid. For big data, the need for this scrutiny appears to be almost completely forgotten.
Semantic interoperability is the ability of computer systems to exchange data with unambiguous, shared meaning – critical in health care where the meaning of a piece of clinical data is highly dependent on the context in relation to the patient. It requires technical standards, clinical terminology standards, and also record structure standards. Record structure standards ensure that the context of clinical data is preserved, and that it means with clarity what the clinician intends it to mean.
Patients and the clinicians to whom they turn for health care probably have by far the greatest interest in ensuring that information they use to make their decisions on appropriate treatment and determine the effectiveness of the treatment is accurate. They are the most likely to ensure that what is recorded in the notes is correct, reliable and can be reviewed with confidence.
There is now a substantial body of clinical professional and patient led work developing evidence and consensus based standards for the structure and content of electronic patient record records. The Professional Record Standards Body, which represents patients, and all health and social care professions, is taking the work forward. It is building relationships with policy makers, computer scientists, system vendors and the pharma and biotech industry. The objective is to ensure that electronic health records record high quality data at the point of care, and that it can be safely aggregated and transmitted in big data.
If we are to benefit from genomics and improve effectiveness and efficiency of health care, we have to ensure that the evidence from big data really means what it appears to mean. We must not forget to act as judge as well as jury.
Contributed by Prof Iain Carpenter MD FRCP, Chair, Professional Records Standards Body, Associate Director, Health Informatics Unit, Royal College of Physicians
The One Nucleus blog is written by individuals and is not necessarily a reflection of the views held by One Nucleus.