When deciding how to group the
population, it is necessary to have a primary organising characteristic. Some of the different possible
primary organising characteristics, and the benefits and limitations of using them,
are discussed overleaf:
of condition and age
- Social and demographic factors
- Utilisation risk (risk stratification)
1. Type of condition and age
Condition and broad age group are two factors that are easy to define, intuitively understandable and that will stay relatively constant over periods of time. For example, the majority of people diagnosed with a long-term condition will need to manage that condition consistently for many years. In addition, there is also already data collected about these factors across the population in North West London. This approach has been used across many integrated care systems internationally, including the programmes in Torbay and Tower Hamlets. The limitations of using this approach are that it is in some ways similar to the approach that we already use in the system of splitting people by disease pathway. This issue is addressed later in the document.
2. Social and demographic factors
Social and demographic factors such as ethnicity, age-related frailty, social connectedness, behaviour and levels of economic well-being have all been shown to have significant effects on health outcomes. These factors should therefore be included in any holistic understanding of the population; however, when forming mutually exclusive, intuitive and easy-to-use groups, these categories can become problematic. Many of these are fluid and subjective categories, with limited consistency of definition and people constantly moving in and out. This makes it difficult to design consistent models of care around these categories and to base a payment model on them. It is difficult to intuitively understand who will fit into any given group based on these categories. For example, quantifying the level of social engagement that a person has is a difficult task to implement across a large population. However, these themes are likely to have considerable effect on people’s capacity to manage their own care and their need to rely on statutory services. An effective personalised approach will therefore need to take such factors into account for each individual.
3. Utilisation risk (risk stratification)
Many providers across North West London are already using a risk stratification approach to understand their populations. This approach is currently applied within the North West London Integrated Care Programme, for example. Tools currently available to measure utilisation risk do so only on the basis of an unplanned emergency admission in the next year. Advantages of this approach include that it is simple, that it is already in use across much of the patch, and that it has a fairly high ability to predict whether people with long-term conditions will be admitted to hospital. There are three major limitations to only using this approach, however. First, focusing on the top 5 or 10 percent of people most at risk is not the most effective way of organising models of care, as those people do not necessarily have similar needs. Second, it fails to capture many of the possible improvements in care and outcomes that could result from interventions that will not directly affect acute care. Finally, grouping by risk is done year by year, and is therefore less effective in enabling long-term planning of support for people who will need ongoing, consistent management of conditions.
However, the risk stratification approach to understanding groups can be complementary to grouping by other primary organising characteristics. For example, grouping the population around people with similar needs according to age and condition is complementary to a risk stratification approach. Grouping by condition type and age helps us understand where in the population the types of needs are similar, while risk stratification helps us understand where in the population the magnitude of needs are similar. By using these approaches together, we can gain a much more nuanced and useful view of where to target services; they provide us with an understanding of not only who will have the greatest needs, but also who will have similar needs.
One final way that we could group is based on individual service user behaviour. This includes things like attitude towards the health system, daily habits, mobility and spending habits. These are often closely linked to social and demographic characteristics. For example, someone who has limited mobility is more likely to be lonely and feel isolated. There are many examples of programmes that have based their groupings on individual behaviour. One example can be found in many smoking cessation programmes, which segment based on the person’s attitude towards quitting. By targeting different interventions to people who have different attitudes, the programmes are able to be more personalised and more effective. The pros of this approach are that it allows personalisation at a level that may not be possible with the other methods. The cons include: 1) it is difficult to establish a baseline for a large population, as it would need to include large-scale surveys or interviews, and 2) using behaviour as the primary organiser will mean that many other characteristics that may make people more similar are missed.