Ignore all the distracting noise and proliferation of choices and focus on this strategy that delivers results, says Max Jones

As the health and social care sector grapples with the demands of efficiency, quality and improved outcomes, there is an increasing need to invest in solutions that support effective decision making.

Faced with rapidly evolving strategic needs and surrounded by an abundance of technology choices, healthcare executives often struggle to implement an effective approach to creating a data and analytics framework for their organisations.

Rather than adding one more product to their portfolio, healthcare organisations should strive to create a sustainable business intelligence capability that supports an effective action-oriented decision-making culture.

When building a data and analytics framework for your organisation, ask yourself the five key questions below. This approach should ensure that you are able to successfully deploy a data and analytics approach that supports the strategic objectives of your organisation.

1. What are the key problems you wish to solve?

To establish an analytics framework that drives decision making and action within the organisation, you should ensure that you start with the end in mind. The benefit of a clear set of problem statements cannot be under-estimated. Are you clear about the key transformation initiatives for your organisation and how data could support their implementation? What demands will be placed on your BI function to allow it to function effectively in the local STP landscape? What operational decisions could be enhanced by data being available at the right time and place? How could data be used to drive up patient satisfaction? Succinctly outlining end goals guides what problems need to be solved and helps define the analytics framework that will assist with the decision making.

2. What type of data is required at what frequency?

Transformational goals and metrics often require harmonised data across the health and social care continuum. Typical data includes EPR data, emergency department data, bed utilisation data, staff scheduling data, and patient outcomes and mortality data. However, with an eye to the future, you should also now be considering other data sources – for example, what might your services be able to achieve with at-home monitoring data, social media, or location data such as local weather forecasts and pollen count, if they were successfully integrated with your existing data sources?

Linking back to the problems which you defined in step 1), outline which data types are required to be updated in real time to support predictive analytics (clinical data elements such as diagnosis, problems, medications, etc) and which ones can be brought in a batch format (staffing, scheduling etc). This will assist you in establishing an integration architecture with various source systems across the network. Developing integration architecture also offers an opportunity to more accurately estimate how much the infrastructure will cost and how long it will take to implement.

Build a data integration roadmap that outlines the sequence in which your data domains will be aggregated into the analytics platform. If you started with clear problems statements, then this step is significantly simplified. Start with high value problems and build momentum – reusing BI architectural components as you move to the next problem.

3. How good is your data?

The relative quality of data from varied locations across a continuum of an integrated care network can influence your data acquisition strategy. Key considerations include:

  • Relevance. How relevant is the data to your decision making process - what problems does it help to solve?
  • Accuracy. How close is the recorded data to the true values?
  • Punctuality and Timeliness. Is the data available with low lag and to a reliable frequency?
  • Accessibility. Can you get the data and process it in a cost effective manner which others will recognise as valid data?
  • Interpretability. How easily will your data consumers understand what the data is telling them?
  • Coherence. Is the data based on definitions which are relatively static over time?

This assessment of the quality of your various data sources forces you to consciously focus on the limitations of the data. In the context of the agreed problems statements you can then seek ways to address any significant data deficiencies or at the very least make limitations known to the consumers of your analytics.

4. How will your data be consumed?

Even the best information is useless if it is not made available at the right time to the right person. Your data and analytics framework needs to explore the many and varied routes that your data can be consumed. This consumption needs to target the key decision making steps in the processes in question. It should be grounded in the processes that support the key problems identified in 1) and be accessible in an intuitive form. The data tools required to support a clinician in deciding how to refer a patient or to support an acute hospital divisional manager in deciding how many surgical beds are needed next week is both very different in content, layout and access device.

Your data and analytics framework also needs to cater for the known unknowns – your super-users. They don’t yet know the questions they will ask but they know their processes, their patients and their data. What can you do to liberate the data so they the can do great things with it? Self-Service BI is still an unmet opportunity for much of healthcare – are you ready?

5. What organisational capabilities need to be developed to support the future state?

It is important to identify key consumers of analytics within your organisation early on. Understanding key information needs of the users, level of data literacy (ability to understand and interpret data) and the ability to exploit information offered via an analytics platform determines the pace at which your organisation can adopt a knowledge-based decision making approach. Consider setting up a multidisciplinary data governance council that aims to provide an operational framework that allows for maximising data exploitation for the organisation’s benefit.

Making an honest assessment of your organisation’s in-house capability for designing, developing and operating an effective BI function is equally important. How do you plan to develop your BI teams to increasingly become a key part of how the organisation makes decisions? How are they led? How do they work with clinical teams and others to develop new solutions balancing agility with safety? How are they identifying opportunities to innovate and drive the adoption of new approaches?

Conclusion

Organisations can cut through the complexity of business intelligence by asking a few key questions on their journey to a meaningful data and analytics strategy. Through close-coupling BI to your business problems, focusing on supporting your decision makers with great data and building momentum around successful BI deliveries your BI function will be recognised as a defining feature of your organisation’s future.

Max Jones is a director at GE Healthcare Finnamore, specialising in technology enabled change underpinned by data, analytics, information governance and programme delivery. He can be contacted at max.jones@ge.com.