Skip to 0 minutes and 4 seconds The NHS delayed discharge project was a really interesting challenge that was brought to us by a number of different people. The real challenge behind delayed discharge is how do we get people out of hospital at the right time? There’s all sorts of stories in the news, and we have to deal with the data and the information behind that. There’s all sorts of challenges of getting people out of hospital where they don’t need to be. So people come in to get treatment and for whatever reason they get stuck. And the challenge that we put to ourselves was actually two fold. One was, can we predict if someone will be delayed in hospital before they even arrive?
Skip to 0 minutes and 43 seconds And the second challenge we set ourselves was where in that whole hospital journey are they getting delayed and can we predict that in advance of that situation occurring? There’s a lot of focus on this area, a lot of money in this area. The NHS spends hundreds of millions of pounds a year treating people in hospital that don’t need to be there, so there’s a real incentive to use the data, use the information that we have to change that, to break that pattern from happening. The main area that we did focus on was supported by the Data Lab, supported by Scottish government, was to look at can we predict before someone arrives in hospital.
Skip to 1 minute and 20 seconds So my organization, which is information services division, which is part of NHS National Services Scotland, actually hold data on all patients for Scotland from cradle to grave. So even before you’re born, we’re collecting some data from you, right through primary care, secondary care, hospital, into the community, right through to end of life, we have data and information about you. What we decided to do was take all of those different sources of data, put them together in a model and just simply try, just simply experiment to see can we actually do that prediction? It was a fascinating journey. We threw all kinds of different pieces of information together and worked very closely, very collaboratively to find out that yes, you can.
Skip to 2 minutes and 8 seconds You can actually predict someone before they arrive in hospital, they’re almost certainly likely to be a delay in hospital. So the output from the tool and the output from the project was to create a model that risk stratifies patients. So basically it tells patient’s high risk or low risk, they’ll still be a delay in hospital or not. And that has significant benefits and potential opportunities to improve care for patients by just knowing that one piece of information. I like to think that ultimately we’ve prevented or at least changed the outcomes for patients by not being delayed in hospital.
Skip to 2 minutes and 39 seconds The big advantage we’ve had with the model that we’ve developed is because we can now identify which patients could be delayed, it gives staff, and in some cases the patients themselves more time to plan. So we have what’s called an anticipated care plan, where we try and think ahead to see if this person gets taken to hospital, what should happen to them or what shouldn’t happen to them. So sometimes it’s about avoiding particular treatments or particular care being carried out that may have been done before or aren’t necessary for that patient’s condition. So we’d like to think that we’ve had some success in stopping all of that from happening.
Skip to 3 minutes and 23 seconds But it’s a complex picture and I think that it’s important to stress the model on its own isn’t the absolute silver bullet to everything, and it’s important we’re trying to build that in with other processes and other treatment patterns that are ongoing within a very complex environment that is a hospital.
NHS Case Study
I asked Jonathan Cameron, Head of Service (Strategic Development) at National Services Scotland NSS (the shared services arm of the NHS in Scotland) if he could give us an overview of the delayed discharge project, specifically the value to patients.
At the time of this project, around 1,200 patients were delayed from leaving hospital each month. A delayed discharge is when a patient has completed their medical treatment but is unable to leave the hospital because they cannot look after themselves at home.
Scottish Government and NHS/NSS worked with The Data Lab to perform a proof of concept that resulted in an assessment of the risk of delay when a patient presents at a hospital.
In the early stages of the project, around 125 data points were identified as candidates for determining whether a patient would become a delayed discharge or not. Through the hard work of cleaning and understanding the data, this was reduced to just six key data points. They were then used in an algorithm to predict, with 97% accuracy, the likelihood of becoming delayed discharge.
Initially I saw this as a decision-making case study as they were asking who is likely to be a delayed discharge? But it became clear that there were operating model improvements possible when they moved on to asking what do I need to do now to avoid this becoming a delayed discharge?
Improved decision making and operations (Click to expand)
The first point I would like to draw your attention to is the fact that having a lot of data is not necessarily a good thing. Having the right data, the right small amount of data, can be much more effective. It can reduce the amount of time and effort required to clean the data and subsequently reduce the overall project duration, resulting in delivering value quicker.
The second point is that the team identified that these six key data points were captured early on in the process. This was really valuable because if you can predict which patients are likely to be a delayed discharge early on, you can intervene while they are still getting treatment and put care packages into place to allow them to be discharged when they are able to do so.
I think the NSS team achieved great things on the delayed discharge project in a short amount of time and managed to take patients, employees and stakeholders on the data journey with them.
Have a look in the “See Also” section below for more information on this project and some addition interview content.
© The Data Lab & Seek And See Ltd