What do you do about tracking data relating to an organisation or area over time, if the boundaries keep changing?

Recently, I’ve been doing more with English NHS data. This is a new one as I’ve mostly been dealing with Welsh NHS data! Which it turns out is relatively straightforward in comparison.

One of the problems is that over, even a short period of time, hospital trusts are created, merged, and cease to exist. This means if you’re looking to collate data over a longer period, you need to find a way to combine data to reflect the most recent incarnation of the trust.

It isn’t too difficult to sort things out if what’s happened is a straight merge and you have the raw numbers. So, for example, Bart’s. Healthcare is made of Barts and the Royal London, Newham, and Whipps Cross. By adding together the number of attendances at A&E, for example, you should get an reasonably accurate idea of what’s been going on at all the hospitals in the trust, even when they used to belong to other trusts.

However, as of October 1, South London NHS Healthcare Trust has ceased to exist. But it has straightforwardly merged with another trust. Instead the three hospitals that were managed by the trust have been taken over by different trusts – Princess Royal University Hospital, Bromley, became part of Kings College London Foundation Trust, Queen Elizabeth Hospital became part of the Lewisham Healthcare Trust (now Lewisham and Greenwich) and Queen Mary’s, Sidcup, was taken over by Oxleas NHS Foundation Trust (which wasn’t even an acute trust before this (see, I said this was complicated).

So you end up with data for three hospital trusts that you can’t compare to previous years.

One way around this might be to break everything done to the smallest unit, in this case hospitals, as they’re less likely to be subject to change, as they have a physical location, and it would be easier to keep track of how the data has changed over time.