Primary care holds some of the richest data in health and care which can tell us a great deal about local populations and their needs. However, faced with so much data, it can be difficult to know what to look at and what we can do with it to drive improvements in health and care of our population. David Sgorbati, chief analyst at the Health Economics Unit, Midlands and Lancashire Commissioning Support Unit offers his insights in the first of our data series.
Data is incredibly powerful when it comes to helping us make decisions about changes to care pathways or the allocation of resources and it really is true that primary care networks (PCNs) hold the Crown Jewels of NHS data. As the ‘front door’ of the NHS, PCNs have an unrivalled insight into their populations, covering everything from consultation data to prescription information and a whole host of secondary care markers. What’s more, they also have a very deep and profound understanding of the community and its needs. For this reason, it’s vital that they are involved in shaping any collection and analysis of local data.
Familiarising with uncertainty in your data
Firstly, it’s important to get familiar with the limitations or weaknesses in your data. For example, there is always bias in data due to either the way in which the dataset is defined or in how the data are recorded. Health data typically represents certain groups more than others and this is something that must be considered when developing data driven insight and planning onwards action.
Another common issue with health data is that there isn’t really a standard approach to how we record information. The way each GP practice in a network records certain data may vary. It may also vary between GPs within a practice, and even individual GPs might use different approaches from one day to the next. For this reason, it can be beneficial to lean on those who understand how the dataset was created as well as those who understand your patient population when embarking on a project.
What’s the right question?
One of the most crucial elements to get right when starting any data analysis project is choosing the right question. That might sound obvious, but it can be challenging to ensure the question we’re asking will give us the answers we are looking for. For example, we might say we want to understand how many patients are attending A&E each month. However, the answer we are looking for here might really be what proportion of patients attend A&E, or what the characteristics are of those patients who attend A&E, or even why patients are attending A&E, so we can understand if there are any trends or unexpected variations in the type of patients we find.
This is where combining your rich data and local expertise with the know-how of an experienced and professional analyst can help you frame the question correctly from the start. This is known as problem structuring and is an absolutely critical step in any successful analytics project. Without spending time on clarifying and applying some critical thinking to the question in hand, you will often later find that you’ve wasted a lot of time analysing the wrong pieces of information. If you do this in partnership with an analyst, it can also help you to understand and appreciate what analysts are truly capable of and what they need to be able to deliver on your request.
The role of the analyst and how to empower them
Data analysis is powerful, but it’s also challenging and complex and is best attempted by people who truly understand data and the many ways in which it can be interpreted and analysed. The analyst profession is a varied one, and not all analysts are the same. For example, within our team at the Heath Economics Unit we have data engineers, analysts, data scientists, health economists and econometricians – all with specific skillsets and areas of focus. Very few people can do all of this (and do it well) and, especially when dealing with people’s health, it pays to draw on the right expertise.
For example, it’s important to use the right tools and algorithms when embarking on an analysis project, as some will be completely ineffective against certain types of questions and datasets. A professional analyst will know what approach to use and when to get the best result. This is why it’s crucial to employ the services of the right kind of analyst with the appropriate skills and experience. Often the problem you are trying to solve will be complex and it may require the inputs of several different kinds of analysts.
In some parts of the health system, most notably for Civil Service roles, individuals are expected to align with competency frameworks, and it is seen to be beneficial to do so. In other areas, specifically the wider NHS, this is less well-developed and there can be a lack of direction for analysts in terms of understanding which framework they should align to. The benefits of adopting a common framework for analytical disciplines are numerous for all parts of the system, including patients, employers and individual analysts.
To support the upskilling of analysts, the Health Economics Unit and the AnalsytX community have been working in close partnership to develop the Population Health Data and Analytics Centre of Excellence. We have collaborated with NHS England, the Association of Professional Healthcare Analysts (AphA), local analytical teams and the healthcare data industry, to build a curriculum with pathways for training. We have also worked to identify existing resources and best practices to upskill analysts and improve every interaction with data in the population health management cycle. The curriculum helps guide people towards the training that is relevant to them at each point in their career.
If you do already have an analyst role embedded in your PCNs or wider system, there is plenty of opportunity out there for professional development and upskilling. Over recent years, and particularly since the start of the Covid-19 pandemic, a buoyant community has developed through networks such as AphA and the NHS Python and Analyst X communities. These offer many options for analysts to share best practice, offer peer-to-peer support and access professional training and development opportunities.
Intelligence-led improvements in health and care: Population health management
Of course, data analysis is important, but how can you take the insight you gain and use it to support you in making decisions about how you care for your community? There are many different insights you can glean from your data which can help you target interventions or resources most effectively. However, it can be challenging to do this successfully without also considering the wider determinants of health, which are central to the concept of population health management. The worsening cost-of-living crisis we are living through makes this more important than ever, as an increasing number of people begin to fall into extreme poverty, or even destitution.
For example, studies suggest that homeless people are dying 30 years earlier on average than the general population. This doesn’t just apply to those living on the streets, but also those in sheltered accommodation where alcohol and drug misuse are common. This demonstrates why we must start to look more at identifying the people who are seemingly well today but who may become ill in the near future, rather than focussing solely on treating those who are already unwell.
There are several places to find population health data, including the Office for Health Improvement and Disparities’ (OHID) fingertips tool, which gives information about the population at a granular level. It is not linked to individuals but can provide a good overview of the wider determinants of health. Local authority, Integrated Care Systems or Integrated Care Boards may have have a population health management team who could help.
In different projects we have also worked with GPs to provide insights at a high (multi-patient) level, supporting them to better undertake quality improvement case-finding activities in their own patient data systems and enabling them to drill down and identify those patients with suboptimal care who require additional attention.
At the Health Economics Unit, we do a lot of work on population health management to help our clients make changes to the care they provide. This includes a range of services such as:
- population segmentation to help them better understand their local community,
- risk stratification to identify those most at risk of an ‘event’,
- impactibility analysis, which is about working out which segments of the population are likely to be most amenable to change,
- and allocative efficiency, which helps us understand where the most value lies.
We are currently supporting five ICSs working together with primary care to make decisions around resource allocation for COPD. Here, we are using the STAR (socio-technical allocation of resources) approach to allocative efficiency to identify the most valuable use of resources across the entire local system. STAR is made of equal parts social and technical elements. This means there is equal importance attributed to the perspectives and experiences of key stakeholders across the pathway, and the data.
First, we collect data on interventions along the pathway, each is analysed for value, in terms of quality-of-life gains, costs and the number of people who benefit. Then we get all the people involved in the pathway, including patients, into a room together. This is known as ‘decision conferencing’. Their role is to discuss the value of the current pathway using their expertise and collated local data. Finally, with vast collective knowledge of the current pathway and the population, the group then identifies key interventions for investment. This is ultimately to ensure we are doing the right amount of the right interventions along the pathway to maximise our population’s health with the limited resources we have.
Help to get started
Of course, health data is extremely complex (and not always up to date or fit for purpose) and I appreciate that not every area is fortunate enough to have access to a team highlighted in the case study. This can make attempting any kind of data analysis seem daunting. That said, just because you can’t do everything you might want to do, it doesn’t mean you should do nothing. Knowledge really is power and there is no better knowledge than that which is backed up by solid data analysis. Organisations such as the Health Economics Unit, the Strategy Unit or AphA boast a wealth of expertise and a deep understanding of the different analytical communities and various approaches you can take and would be more than happy to help you understand your requirements.
By David Sgorbati, chief analyst at the Health Economics Unit, Midlands and Lancashire Commissioning Support Unit first published on sister title Pulse PCN