This site is intended for health professionals only
19 October 2017
Share this article
From fire-fighting to forward planning – why data analysis should be part of your practice DNA, writes Helen Seth
When it comes to providing more efficient and effective services, the quick wins have already been banked. So how do GP practices continue to improve services and save money, while grappling with an ageing population, an exponential rise in long-term conditions, tighter budgets, increasing vacancies and a heavier workload?
While data analysis may not be the answer to all these challenges, using risk stratification does give clinicians a greater understanding of the likely behaviour and vulnerability of their patients and enables them to plan more effective services. As more practices work together across localities or under new arrangements such as federations, the opportunities to use data to improve preventive care increase.
The use of risk stratification is increasingly widespread, but the focus continues to be on analysing top-costing patients and those most at risk of unplanned hospital admissions. While this work is vital in addressing immediate pressures on emergency services and primary care budgets, further gains are to be had when risk stratification is used more creatively by practices to anticipate needs and support preventive care.
Risk stratification in practice
Risk stratification tools are used to identify and predict outcomes for particular cohorts of patients based on a range of algorithms. The Arden & GEMCSU risk stratification tool uses the highly regarded population profiling and risk markers developed by the Johns Hopkins adjusted clinical group (ACG) system. Rather than focusing on specific diseases or episodes, this system encourages a holistic view of the patient, including co-morbidities that could affect commissioning and care management decisions.
Using primary and secondary care data on admissions, GPs and clinical commissioning groups (CCGs) can use the tool to build registers of high-risk patients and drill down to view individual care pathways. The tool also shows patient attendances at A&E, hospital admissions and outpatient appointments. This gives primary care professionals essential insight into how they might consider modifying care plans and pathways to improve self-care, reduce GP appointments and hospital admissions, and support patients’ needs in the most cost-effective way.
Unlike some legacy algorithms, the Johns Hopkins ACG system is updated monthly with both clinical and secondary care data. This enables us to create reports that offer insight into specific issues or treatment pathways, as well as identifying population trends to support future commissioning. Using filters such as age band, condition and frailty, cohorts of patients can be used to create bespoke lists for case management programmes. This approach is empowering health and social care organisations to deliver targeted care and health promotion strategies where their work will have the best impact.
The Lakeside Surgery in Corby uses risk stratification to improve its proactive monitoring and management of patients with complex conditions. Lakeside GP Dr Tony Penney commented: ‘My practice reviewed the list of highest risk stratified patients against our own list of patients at risk of unplanned admissions. We found 27 patients who had a risk of 75% or greater of a hospital admission in the next few months who were effectively off our radar. This included patients having chemotherapy that we were unaware of, and patients on multiple medications with poor control of long-term conditions. We have now added these patients to our proactive care list and fully anticipate being able to manage their care better and more efficiently.’
Some GPs are going one step further by adopting a partnership approach and sharing relevant patient data with selected community service providers, with appropriate information governance controls in place. In one area, GP practices have given the integrated locality team (ILT) access to their clinical data so they can look at the top 3% of patients at risk of unplanned hospital admissions.
Combining the tool’s algorithms and local knowledge, the ILT can speak to GPs and patients to assess who is appropriate for further intervention versus those who have had a one-off emergency requirement that caused their risk score to spike.
Similarly, another CCG has used a bespoke risk stratification report to support work with its multi-disciplinary teams. Using the filters embedded in the report allows their practices to create shortlists of potential candidates who might be suitable for assessment or intervention from a range of community teams, practice nurses, administrative staff or healthcare assistants.
Examples of criteria used include:
Using combinations of these parameters enables the CCG and its practice to select cohorts of patients that would benefit from early intervention from a heart failure nurse specialist, diabetes service, medication review or falls prevention work.
Enhancing preventive care
Naturally, the temptation is to focus on using risk stratification to tackle the most immediate high-cost issues. This work is certainly important and there is evidence that this approach is helping to reduce A&E attendances. But this is just the tip of the iceberg – and it’s also the toughest area to deliver sustainable results. Tackling top-costing patients is always going to be challenging as the likelihood is they will have a complex combination of conditions that may already be reasonably advanced. While some proactive intervention may reduce A&E admissions, it may not be possible to make sustainable changes that deliver significant benefits.
As organisations become more familiar with risk stratification, we are starting to see exciting work being done to radically improve patient care before needs become too great. Where practices and other organisations work together, either through an alliance, across a CCF or across community services, risk stratification can be used more effectively to identify common conditions and lifestyle factors, understand the impact of different care pathways, and support future commissioning decisions.
For example, Leicester CCG developed care programmes designed to reduce costs and improve patient experience through medication reviews, patient and carer education and improved communication and
co-ordination among those providing care. Using the Arden & GEM risk stratification tool, patients were assessed for eligibility based on the following criteria:
Eligible patients are enrolled in the care programme and reviewed by a clinical care coordinator. The initial feedback from patients and doctors has been good and results are showing reductions in cost and improvements in quality.
Another CCG we work with has used the tool to identify that their locality has a high proportion of adults aged 20-49 with multiple comorbidities. This cohort is already making higher-thanaverage use of secondary care services.
Stratifying this cohort demonstrates that as they get older, their use of health and social care services will increase and they are more likely to be at risk of unplanned emergency admissions. The tool has enabled the
CCG to consider commissioning services that not only anticipate growth in demand as these patients age, but also provide early interventions that could reduce their risk score in the future.
Working in partnership with business intelligence analysts
One of the common concerns about using a tool like risk stratification is time. Against the backdrop of a fast-growing workload, how can you possibly justify practice time spent analysing data? Much depends, of course, on how you work with your analysts.
In our experience, the biggest gains to be had are when we work in partnership with customers to develop bespoke reports that provide the data needed to tackle specific priorities. For example, the Five-Year Forward View highlights frailty as a key priority. NHS England has tasked GPs with moving from opportunistic to systematic population-based identification of frailty to help reduce inequalities, improve access to care and enable the needs of individuals to be met though early, proactive targeted and appropriate interventions. But as NHS England acknowledges, not all older people are frail, and not all people living with frailty are elderly – identifying who is most at risk from frailty related falls and medication side-effects is essential.
We are working closely with clinicians to incorporate the right indicators into our risk stratification tool to support this new area of work. This type of partnership working between expert analysts and customers not only helps practices with their immediate priorities but also enables business intelligence teams to identify new areas to investigate and to share best practice across the NHS.
Risk stratification is improving all the time as clinicians and analysts become more confident and conversant in how it can be used to improve patient care. By looking at data across comparable localities, variations in performance and approach between practices can be identified. This has already led to learning being shared across localities, helping to fast-track improvements and deliver much-needed efficiencies.
Data analysis may not immediately appear to go hand in hand with improving patient care, but together we can develop more creative and innovative ways to stratify patients and understand the triggers for health deterioration. Analysts, clinicians and commissioners can work together to identify new and sustainable opportunities to improve quality and outcomes.
Helen Seth is associate director of business intelligence at Arden & GEM Commissioning Support Unit in Leicester