NHS RightCare uses population data to identify opportunities for improvement, helping local systems plan and develop primary care services and reduce unnecessary referrals to hospitals. This type of population health analysis is very useful, but can rely on assumptions about certain groups of patients.
Dr Foster recently partnered with Bradford CCG to support its Out of Hospital (OOH) programme, which aims to encourage people living with long-term conditions (LTCs) to better manage their own care more effectively. It required a better understanding of the patient cohorts using the most primary and secondary care resources in order to identify who would benefit from the OOH services.
We linked primary and secondary care data, conducting risk stratification and patient segmentation analyses across the CCG’s population to determine the patients who were using the most emergency care resource. Further patient segmentation analyses then established key themes and eligibility criteria for accessing OOH services within the patient group identified. We used a range of indicators, including frailty, LTCs, and primary diagnosis of hospital admission, and applied indicative cost estimates per patient based on national tariff data.
We found that, while the assumptions Bradford CCG had made about its population were largely accurate, we were able to identify further important trends. One of the most significant findings from our work was that patients were often presenting in secondary care with UTIs and skin infections. These can be treated in primary care, so by targeting patients before the conditions escalated, Bradford CCG can divert them from costly secondary care services. Also, in recently updated primary care diagnostic guides for the management of UTIs, the risk factors for this type of infection are identified as: a previous UTI, urinary catheterisation, hospitalisation, antibiotics in the previous month, and older age. Knowing these risk factors can enable GPs to prevent UTIs from occurring in the first place.
Dr Foster replicated the work carried out with Bradford CCG on smaller areas of its population, grouped into community partnerships. The analysis followed the previous methodology, first looking at the difference across groups and then combining multiple segmentation analyses, delving into the relationships between patient characteristics. The analyses provided valuable insights, uncovering patterns and revealing the patient groups in the community incurring the greatest costs in an emergency secondary care setting.
Our work aims to remove the bias from population health analysis by removing human assumptions, instead using machine learning and clustering techniques to reveal patterns in the data that might otherwise go unnoticed. We use bespoke data visualisations to help the people we work with understand what the data is telling them. Our analysis generates graphs, turns reports around quickly and efficiently, and we will soon be providing interactive clustering of population segments, so clients can refocus the analysis around a particular demographic with the click of a mouse. Each community has a unique set of patients with different needs and therefore has different issues to deal with. Our work offers commissioners and health professionals a clearer, more detailed picture of the populations they serve, giving them the power to deliver better, more proactive care and cultivate healthier communities.
You can find out more about our population health analysis work here