Valtech @ My NHS - “How we improved NHS decision-making & transparency through open data in 2 weeks”

Valtech @ My NHS - “How we improved NHS decision-making & transparency through open data in 2 weeks”

Recently we have written extensively about our work with the My NHS team. Here we give a more detailed account of the project – why it’s important, how we worked, what we delivered, and what is set to come next.

The Why

Over the past 2 years there has been an unprecedented drive to ensure that the NHS is the most open and transparent healthcare system in the world.

As part of this, the NHS have launched My NHS, a powerful new digital service. Drawing on over 7,000 data points, it enables comparison between NHS organisations on 132 measures (at present) that matter to commissioners, planners, clinicians, Trust budget-holders and numerous other stakeholders, to both better equip them to make management decisions, and foster a strong culture of learning and improvement within the NHS.

It is acknowledged that the current data aggregated through My NHS is one-dimensional, meaning the quality of insights and resulting decision-making power is somewhat limited. As such, the My NHS team is currently progressing towards taking the site to the next stage, incorporating multi-dimensional data, open APIs, and improving data visualization on the site. This will enable users to more easily interpret, compare, and act upon service performance across the health and care sector. To this end Valtech were engaged in a 2-week Discovery phase, to research and validate the user need and demonstrate the art of the possible for the My NHS site moving forward.

The How

Our method is broadly illustrated in the following diagram:

Our joint My NHS and Valtech team used a Hypotheses Driven Development (HDD) approach to quickly and collaboratively understand and validate user needs. We ran two 1-week iterations, or “Sprints”, focusing our delivery cadence for each on testing and research with real users, across first mental health and then hospital care, with hypotheses specific to each forming the activities for the respective sprint. We used a technique called Assumption Mapping to help understand areas of uncertainty and urgency, enabling us to develop the most appropriate prototypes and visualisations to understand user needs whilst keeping within strict time constraints.

What?

At the end of the 2-week Discovery phase we were able to demonstrate the power of visualisation and multidimensional, timely data. Describing these in a narrative is a lot easier than the current static tabular data, enabling more powerful insights through which to make decisions more effectively.

The discovery outcomes were:

- A proven user need for an enhanced My NHS service - throughout the Discovery the main driver has been whether there is a user need for the proposed solution. Although it was hard work to get the right user research subjects in the right numbers, we were able to achieve this and thus validate the user need for a more intuitive, data-rich NHS performance comparison service.

- Dashboards – using the open source tools R Shiny (which we have also used on a similar project at Smart DCC – deploying smart energy meters across the country), our team of a Project Manager (Tawheed Alam) Data Business Analyst (James Lefas), Data Scientist (Mikayel Mirzoyan), UX specialist (Matt Crean), developed working dashboards for Mental Health and Hospital data. Utilising real, open NHS data we were quickly able to show early insights as to what was possible for the future of My NHS, and how putting data at the centre of the NHS & social care could bring wide-ranging benefits to the system.

For example, we looked at various correlations and non-correlations, the most interesting being the non-correlation between deprivation index and quality of care.

-Non-correlation-

We found that, based on the data available, quality of NHS care stayed the same regardless of the deprivation score of an area, thus going some way to busting the myth of a “postcode lottery” within the NHS. As you can see from the following two diagrams, the Spearman scores show no correlation between Deprivation Index / Average House Price and A&E Performance (namely how many patients are seen within the NHS target of 4 hours).

-Correlation-

When it came to a correlation, we found that comparing Mental Heath and Acute Health Trusts yielded some interesting areas for learning improvement. By mapping Safe Staffing (the target being 100%) and Financial Performance (the target being £0), we were able to show a trend for both, as below:

Thus, we hypothesised that Acute Trusts could learn a thing or two from their counterparts in Mental Health when it came to Financial Performance, and vice versa when it came to safe staffing – at the very least enough to start a conversation.

- Static prototypes – we created static prototypes for both Mental Health and Hospital care, which were refined after a round of user research to demonstrate what would be possible with certain additional data points. As well as validating the user needs for both, by identifying missing data points we were able to show the clear benefits of expanding the data the NHS & social care system currently captures, effectively writing the business case for further investment in the service from both user experience and data science points of view.

User needs will drive the future development of My NHS and where gaps in data have been identified the relevant commissioners can feed into the appropriate National Information Board (NIB) workstream where data gaps are currently being addressed.

What next?

Now that we have quickly gained an understanding of the user needs sufficiently to continue with the project, we move to planning for the Alpha stage. The objective for this will be to demonstrate that we are able to valuably improve the digital service in question. For that we will require the following:

  • A prioritised list of user needs that will form the beginning of the backlog
  • User journeys representing a scenarios in which a user might interact with the service
  • Working prototypes
  • Static visualisations
  • MyNHS site Stakeholder map
  • Resource plan/costs/business case
  • Technology options

This was a fantastically interesting project to be involved in, and we believe that together we have made real progress in both establishing some momentum for the next iteration of the My NHS service, and cementing the vital role of data in driving wide-ranging NHS improvement. We would like to thank Julie Fidler, Rob Sinclair, Sean Craig and all of the My NHS team of stakeholders for their support, and look forward to the next phase.

We also presented much of the above at the NHS Expo in Manchester on the 8th September, click here to watch the presentation.