Don’t leave pop health to the data scientists
7 May 2019
This is a watershed moment for population health management in the UK. The NHS Long Term Plan, the expanding LCHR programme and the proposed countrywide network of Integrated Care Systems (ICSs) create the financial, organisational and policy framework to take pop health mainstream.
Things are moving fast. It’s only recently that we have started to realise the potential for using NHS, social care and patient-recorded information safely and intelligently to understand the health needs of a population.
But population health is much more than a data playground for data scientists. It promises fantastic operational benefits, in terms of:
- Prevention - delivering targeted interventions for improved public health
- Planning and designing new service delivery and capacity models
- Measuring the impact of interventions
- Identifying variations in care delivery down to individual levels for specific disease groups
- Putting system-wide data into the hands of operational managers to improve patient flow across the whole care domain
It also comes with significant new challenges – such as not clogging the entire health system in a morass of data and information.
Three essentials for the ICS leadership team committed to change
So, what might all this mean for an ICS leadership team?
- We must use this opportunity to impact key challenges in the health care system at real scale by demonstrating value quickly
- We must be ready to act on the insights that are provided and drive change as a result
- Clinical managers need to get involved to provide real practical insight to the process – not leave the ‘data people’ to see what they can find out
The focus now should be to identify programmes of work to drive improvements.
Any population health management solution worth its salt should deliver value in three key areas:
Longitudinal maps of the health and care system and the patient flow through it can help us diagnose where individual patients are not receiving optimal care (wrong combination of drugs, missed tests etc) and also where there are systemic failures. With cohort mapping, we can look at parallel pathways and analyse where changes could improve outcomes or reduce costs.
A linked longitudinal dataset allows for forward-facing as well as retrospective analysis. So it is possible to establish the metrics by which you will judge a service change, identifying cohorts to monitor against and embedding the metrics in dashboards. Commissioners can study the downstream impact of a new community service for COPD patients, for example, based on looking at what happened before and comparing the new pathways of care.
Much population health discussion has tended to focus on predicting an individual’s behaviour and its impact on health outcomes. Our approach at Graphnet, developed in partnership with our customers, involves predicting impact on the healthcare system and patient groups:
- Predicting disease onset – applying proven risk factors to determine groups most likely to develop long-term conditions and then targeting existing and new interventions where they will have most impact. Similarly, avoiding unnecessary interventions on those at lowest risk.
- Predicting demand – both at an operational level in terms of service capacity, and also predicting cohorts with high resource needs so multi-disciplinary capabilities can be scaled or focused in specific geographical hotspots.
All this requires partnership – between suppliers, researchers, clinicians and managers. Most of all, though, it requires strong operational leadership - focused individuals, committed to driving change, and with the ability to act on the insights population health provides.