The data laboratory: predicting future health care needs

Blog post

Published: 25/03/2011

Most interactions you might have with any organisation leave behind a trail of data. In health care, these datasets have many uses and the Nuffield Trust is interested in ways that we can exploit them to inform health policy.

We are particularly interested in how linkage between data sets can reveal a fuller picture of what is happening to people as they use services.  We refer to this as a “data laboratory”, and use this in evaluation and in developing models that predict the likelihood of future events. 

Session three of the Nuffield Trust’s 2011 Health Strategy Summit run by Dr Martin Bardsley, our Head of Research, was about aspects of the “data lab”.

Part of Martin’s presentation was a short exchange on whether the reforms proposed in the White Paper: Equity and Excellence: Liberating the NHS (Department of Health, 2010) are revolution (as indicated by the title of the Department of Health’s consultation document on information) or evolution (as indicated by the Prime Minister’s comments) – thereby rediscovering the same evolution/revolution argument which arose in response to New Labour’s White Paper back in 1997.

Certainly, anyone reading An Information Revolution (Department of Health, 2011) will find much that feels familiar – its aspirations are similar to those set out in earlier strategies.  Yet in information we might have a special case.  Driven by ever increasing sophistication, connectivity and ubiquity, the potential uses of new information streams grows exponentially.  Each new advance reveals previously unthought-of possibilities.  The uses of information in health care could be said to be in a state of almost perpetual revolution, so much so that revolutionary ideas quickly start to look evolutionary.

Following the presentation there was a choice of three breakout sessions featuring external experts. All were related to information research projects in progress at the Nuffield Trust. David Behan on the interface of health and social care; Stephen Johnson on telehealth and the management of chronic disease; and Professor Peter Smith on allocating budgets to general practices. 

The aim of the Person Based Resource Allocation project is to use pseudonymised person level data to calculate GP practice budgets based on an individual’s needs.  These were originally used for setting suggested practice-based commissioning (PBC) budgets, but the shift to GP led consortia could make these budgets increasingly important.  This is a highly sophisticated technique – world leading according to Professor Smith – which can predict future need with about 75 per cent accuracy at GP practice level.

One point that came across very clearly from the presentation and discussion was the need for good communication around how this complex method really works.  This is especially important as the accuracy of predictions will always be inherently limited. 

Some of the audience members presented contrasting views on the perceived strengths and weaknesses of these models.  Such debates will become increasingly impassioned if the tool is used to set definitive budgets. It is vital that contributors have a sound understanding of the mechanism that drives it.

Suggested citation

Blunt I (2011) ‘The data laboratory: predicting future health care needs’. Nuffield Trust comment, 25 March 2011.