Predicting social care costs

Blog post

Published: 24/02/2011

The latest Nuffield Trust report has been a long time in gestation.  It is an important piece of work for us as it signalled our first attempts to link together health and social care information at a person level.

The idea behind this work was very simple.  Can we identify people who are at high risk of needing high cost social care in the coming year?  If we can, then we stand a better chance of targeting prevention strategies to help people now.  The result is better quality of care in the short term and lower costs so everybody is happy. In the NHS there are a number of tools that do this type of prediction for people at risk of admissions (or readmission) to hospital.  Our challenge was to see if we could build comparable tools for social care.

The hardest part was (and still is) getting hold of the data to do this modelling.  We needed structured information for whole populations. Whilst the NHS has huge computerised data sets that cover care for the whole country, these are not perfect, although they at least they exist and have certain recording standards. The equivalent information on social care resides in hundreds of local data sets, all slightly different. Within these lies the information we want about who used social care services (at least council funded ones) and something about their needs.

Thanks to the help and advice of five local authorities we were able to get hold of the data we needed.  We had to extract information at person level anonymously but in a way that allows you to link to health records.  This meant we could see what happened to a person – but couldn’t tell who that person was.  We were able to construct this linkage for around 1.8 million people, of which we focussed on the 180,000 aged over 75 years.  We were able to build models but the test now is to see how they work out in practice.

But the bit that excited me, sad as I am, is the way you can use information to see what happened to individuals as they collide with health and social care systems.  So for one person we can see trips to hospital, perhaps a social care assessment or a period of home care.  We often talk about pathways of care yet hardly anybody has information to see them in this very simple way, and all this is derived from data we are currently sitting on that just needs to be stitched together in the right way. 

I think the ability to look in this way should be fundamental.  It’s important for the care professionals who want to understand the history of care received (or not). For commissioners and planners it can show how different services interact and where gaps and overlaps might be.  So, for example, we looked at social care at end of life.  For care users (or their carers), these views help answer the question ‘what has happened to people similar to me’ and ‘what can I expect for my future care?’

Why isn’t everybody else doing this?  Well, there are examples that this is happening more often – I hear exciting things in Torbay and I’m sure there are others.   I think it’s something that we will see increasingly over the next few years.

Suggested citation

Bardsley M (2011) ‘Predicting social care costs’. Nuffield Trust comment, 24 February 2011.