Big Data

As you read this, the seasonal festivities of 2017 will be behind us. Yet since this is the first IJPP issue of 2018, it behoves me once again to wish all our authors, readers and reviewers a prosperous New Year. I hope it will be a productive one for us all as we seek to secure funding to conduct rigourous research that will generate evidence to inform better patient care. Alternatively, you may of course be focusing on methodological developments relevant to improving the health services research we undertake or looking for new approaches to answer your research question. In turn, of course here at IJPP, we hope that this research will translate into strong papers that you will submit to IJPP for consideration! Looking forward to 2018, a topic of current relevance to researchers, policymakers and even politicians is big data. Can we use big data to answer previously unanswerable questions or to answer questions more quickly? So what do we mean by big data? Data have been described by the three Vs of volume (the scale and size of data), velocity (the speed at which is generated) and variety (the varying forms of data such as both images and text). Conversely, it has been said that there is in fact no universal agreement on a single definition of big data although the RCPE has posited that, traditionally, it refers to the large volume of data generated by the electronic devices the majority of us use in daily life but it also consists of the increasing volumes of data generated when we attend for health or social care. In the UK, for example, every time you visit a family doctor, receive a prescription or attend hospital, a little bit more digital information is recorded. The potential to link these data to other data about you such as your shopping habits, your travels or your lifestyle suggests infinite possibilities for understanding both the development and management of disease. Understanding how these data can be used in research requires consideration of not only its potential but also its limitations and the governance arrangements needed to ensure its appropriate use. Data security is a big concern, and it needs only a quick scan of the news’ archives to find reports of data sets being hacked and individual details being accessed by a third party. These events are increasing each year as the return for the criminal hacker gets ever greater. Nonetheless, because the convenience of the digital world is now a feature of most people’s life, both personal and professional, there is no suggestion that our enthusiasm for taking advantage of the digital world is decreasing. What does this mean for health and health services research? Firstly, we must remember that every piece of data we deal with comes from an individual who may also be a patient and we need to treat that data with confidentiality and respect. Even anonymised data can sometimes be liked to an individual because of unusual combinations of individual data items – for example, a person in a particular age group with a rare disease from an identifiable geographical area. There is much debate about who actually owns the data and who is responsible for its use. In health, there is an argument that the patient themselves owns the data, some believe it is the healthcare professional, for example the general practitioner, who collected the data, and some think it is the NHS/Secretary of State for Health. It is also unclear if the owner of the data is accountable for its use. So in the previous examples, as the data set builds, there are multiple owners at different levels. For example, whilst the patient might own their own dataline, they do not own the whole data set whether this be the pharmacist’s record of dispensed prescriptions or the GP practice electronic patient record system. Clearly, the pharmacist and GP, respectively, have a responsibility for ensuring these collated data sets are securely stored and access is limited to those with a justified need. Likewise, once the pharmacy and GP and other records may be collated at national levels, ownership or custodianship passes on. A systematic review suggested that healthcare professionals were in general positive towards data sharing for public health purposes. Barriers identified were concerns about costs, governance and interference with the professional–patient relationship. These were built on in subsequent empirical work which showed conditional support for data linkage for enhanced pharmacovigilance contingent on adequate anonymisation