Blog Post

Successful high-risk care models must exclude patients. Here's how the best programmes decide to do it.

June 20, 2019

    Health systems around the world have built complex care management models to better care for the small group of patients who represent an outsized amount of health care spending. The most successful care management programmes activate patients to better self-manage in the community, thereby reducing their over-reliance on the hospital.

    Learn more: How to design a dedicated high-risk patient clinic

    But a care management programme is only as strong as its foundation—and in this case, that means starting with the right patients. A model that enrols the wrong patients is destined to fail. Yet while it's uncomfortable to think about leaving some patients out of these programmes, it's worth taking a look at how the best models think about who to include to maximise participants' chance of success.

    Differentiating between high-cost and high-risk

    Most care management models identify patients for inclusion based on utilisation and cost patterns. These factors are the right place to start, but, in our experience, these data tell only part of the story and neglects a key element: patient risk.

    We define a patient at risk as someone who will likely use the system unnecessarily in the future. The word 'risk' connotes something hasn't yet happened, but might occur soon.

    With that definition, there are a number of high-cost patients who are in fact not high-risk—patients with surgical complications, major trauma, end-of-life needs, or incurable cancers. These patients will certainly cost the system a lot, but that's because they actually need to access care frequently. In other words, while all high-risk patients are high-cost, not all high-cost patients are high-risk.

    Our analysis of dozens of models found that purely high-cost patients are unlikely to benefit from a high-risk care management programme whose end goal is self-management and less utilisation. However, while this approach makes sense in theory, how can systems in practice differentiate between high-cost and high-risk patients to make sure they're enrolling the right people?

    Using exclusion criteria to filter past 'high-cost'

    The most successful programmes start with cost or utilisation data, then layer in exclusion criteria to narrow down to the patients most likely to benefit from their models. For example, programmes might exclude patients better suited for a specialised treatment programme, such as cancer patients.

    One organisation that's done this is Stanford Health Care, based in California in the United States. Stanford Coordinated Care (SCC), Stanford's high-intensity primary care practice for high-risk patients, first identifies a large pool of patients based on common inclusion criteria, such as the number of chronic conditions or medications.

    Then, SCC migrates patients who have conditions for which there are already existing and specifically tailored care models out of the participant pool and into those care models. For example, if patients have HIV, SCC connects him or her to Stanford's HIV GP clinic.

    It's important to note that the 'excluded' patients aren't simply left on their own; rather, they're redirected to programmes better suited to meet their needs.

    SCC's criteria are not written in stone. There is still room for professional opinion, and SCC will enroll otherwise ineligible patients if a clinician believes it would benefit them. However, SCC only actively recruits patients who meet both the inclusion and exclusion criteria.

    Exclusion criteria not one size fits all

    It's also important to note that the right exclusion criteria will depend on the model you're trying to build, as well as your local resources and context. We've seen high-risk exclusion criteria adapted to markets all over the world.

    Take HealthLinks: Chronic Care programme in Victoria, Australia, for instance, which offers hospitals capitated funding to care for enrolled high-risk patients. Eligible patients are identified using an algorithm based on characteristics such as age, number of unplanned admissions, and chronic conditions.

    The model then filters out patients with conditions already served through comprehensive state-wide programmes, such as HIV and cystic fibrosis. HealthLinks: Chronic Care also excludes patients requiring unavoidable inpatient care (such as active cancer treatment), as well as those who are not reimbursed through the public state department's funding (such as patients in private hospitals).

    To identify the right exclusion criteria for your organisation, there are some guiding principles to keep in mind:

    1. Consider the goal of the model you're trying to build (is it aimed at self-management or a different end goal?); and
    2. Assess your community's resources (are there other programmes that can better serve certain groups of patients?).

    Have a Question?


    Ask our experts a question on any topic in health care by visiting our member portal, AskAdvisory.