Data projects in the public sector, particularly within health organisations, have got an extremely chequered past. Only recently the National Health Service (NHS) in England was forced to backtrack and reassess its controversial care.data programme, which aims to collect patient information en-masse, store it and analyse it for trends that could improve service provision and disease research.
North of the border, however, NHS Scotland appears to be having some success with a data project of its own. Whilst different in scope to care.data, the nature of the project had all the same warning signals of a project that could go awry. And that's because when it comes to data, and comparing data on a peer-to-peer level to drive efficiencies and improve service, there can be a lot of cultural barriers in play.
But NHS Scotland anticipated this from the get-go and has attempted to avoid the pitfalls through an approach of cooperation and co-creation across the structures of the organisation, so that it brings all the different functions and divisions along with it as it heads towards making better decisions using visual data and analytics.
I got to sit down with Michael Muirhead, lead on the project, and his colleague Jamie Gray at Tableau's user conference in London this week, where the pair explained to me how the health organisation needed to move on from its troublesome incumbent systems and into a new phase of data delivery.
For those unaware, NHS Scotland delivers its services via 14 territorial health boards, each of which is responsible for service provision in their local areas. There are also several specialist health boards, which provide central support and national services to each of these territorial boards. Muirhead and Gray work for one of these central boards – National Services Scotland, which is responsible for all things data.
It collects 1.6 million records each year relating to patients being admitted to hospital, 4.5 million records for patients going to out-patient clinics and 100 million records for people going to pharmacies each year. The type of data being collected includes things pertaining to patient age, gender, location, GP information, types of visit, length of visit etc.
Muirhead explains that in the past the analysis of this data has historically remained in local silos. But going forward the aim is to allow for more collaboration and cross comparison, through the use of a new Tableau platform – called Discovery. He says:
Because the information is person-centred, we can link all this data and that allows us to do a lot of added value work. We are trying to use the information in a more preventative approach.
We hold this data on behalf of the NHS, it's not our data, we hold it. What we can bring to local health boards is comparisons and peer group comparisons. If you look at a particular health board, they know their data and we know their data, but they don't know how they compare with other health boards.
For a long time we have been packaging data and putting it back to them in what is called a benchmark purpose – but it's not for us to say you are good or bad or whatever, it's for them to use it because it's a complex environment out there. We want to make sure that what we are doing is complementing them locally, complementing their intelligence.
From old to newNHS Scotland's plan is to turn off its old data analytics systems – including the likes of Xcelsius and Business Objects. Muirhead says that these tools were created ten years ago and are now “jaded, clunky and tired”. He says that they were no longer keeping pace with the best products available in the market and towards the end of 2013 there was demand from senior stakeholders within the NHS to provide a more visual, interactive product to a wider range of people.
Muirhead adds that his team was keen to continue providing this service in-house if at all possible.
There was a desire to do this in-house if we could. Could we do this within the NHS family for the NHS family? Because if we can do it more efficiently and more effectively than private sector companies, the advantage is that we can work with our customers, we can shape the content and functionality and direction of travel. Rather than getting what the private sector deems of value.
But it wasn't a dogma, we could only do it if we could prove it was of use.
Muirhead and Gray proceeded to conduct a six month evaluation of the best products on the market, which Tableau now describes as being “put through the mill”, according to the pair. They said that the length of the selection period may be put down to public sector conservatism, but add that it was a “big, big investment” for the NHS.
The evaluation period included a lengthy questionnaire and a week-long on-site assessment. Muirhead says:
We didn't want to be taken in by any flash-Harry sales person. We've all been stung by pre-sales talks. That caused a bit of frustration back at the organisation about lack of pace, but we wanted to make sure that we got it right.
And Gray explains that out of the final four vendors that were brought in for the on-site assessment, Tableau won out because of its agility.
The visualisation options from all the vendors were pretty good. It was the process afterwards by which where we asked them to craft things on top – 'what if we wanted this?'. We got them on the fly to respond, to see how quickly they could make amendments, and the guys at Tableau were the most proficient at that.
Co-creation rather than mandate
NHS Scotland officially launched the Discovery product two months ago and is now up to
200 users, with the aim of getting to 400 users by the end of the year and with a final target of 1500 users. Both Muirhead and Gray state that the reaction from the health boards has been extremely positive, which they put down to the collaborative design approach they adopted.
For Discovery we have worked in a co-creation, co-production approach, with about 50 representatives from those boards. So each board gave the time of about three or four people. They have been involved in every step of the way.
I think that the reason it has been so well received is because in the past we have been so turgid with data, we were giving them print outs that were 30 pages long. We are now giving people their own information back to them in a way that is truly interactive to them. It's got a whole new dynamism about it and people are reengaging with their own information and turning it into action. It's not for us to turn that information into action, it's for the end user.
The Discovery tool is designed with a three tier approach – where the top level shows aggregate data, mostly for use by executives. These dashboards largely provide information in a green, amber and red rating, so that busy directors can get a quick view of activity across the organisation. The next level down is intended to be used by BI/analyst staff, which provides a more detailed analysis of what's going on. And then the final level provides richer information about patients that is intended to be used by clinical staff.
Level 1 example:
Level 2 example:
Our spectrum of users is now stretching into non-traditional analytical folks, like clinical staff. We have got a lot more work to do to create the right dashboards to get them on that hook and keep them on that hook, but we have got some work lined up this year to do that for those stakeholders.
It's a no for cloud
Despite a big push from Whitehall and other areas of government towards cloud technologies, and whilst NHS Scotland did consider this as a delivery method, ultimately Muirhead and Gray decided that there was too much concern surrounding security and governance. Gray says:
One of the things we looked at from the outset was cloud technologies, but there was a lot of uncertainty at that level about security and accreditation. Even if you secure something in Europe, the data could still get fired around the world. We are really impressed by those technologies, but we just couldn't settle the nerves of all the information/security governance colleagues around us.
They will rest easier in their beds at night. It doesn't mean we won't explore them when we get the proper security and guarantees around it, but that was why we went for the server solution. It's not that the technology doesn't exist, it's cultural.
However, the pair do say that the introduction of Discovery has created a stronger working relationship between them (the business) and their IT colleagues. Muirhead adds:
There wouldn't have been the confidence in the community at that point, we can't fight that battle and fight that challenge.
With the old systems we would have had to spec what we wanted, send an internal mail to our IT department, they would build it and give it back to us, then it wouldn't quite work. They don't understand the business side, so they might miss something that would have seemed obvious. Then you have to get into changes – where it can take months to change a name and title. It's not agile, it's ridiculous.
This new approach has fostered a much more healthy relationship with our IT colleagues. It's much more of a partnership and we are working together really collaboratively. I think both sides of the organisation are playing to their respective strengths. IT are doing all the heavy lifting and data processing, but then passing that processed data over to the business side and we are building all the visualisations.
That really makes sense. When we come across an error we can now fix it in a day, which is a real step-change in the way we work.