Risk Stratification for Population Health Management Teams
Risk stratification is the first of the Three Pillars of Population Health and its applied cousin, value-based care. Resources are finite, so discerning risks to prioritize their mitigation is a basic requirement of managing care. When we answer the question, “Where do our efforts need to be focused the most to gain the greatest positive effect?” we’re not rationing care but rather being good stewards of our healthcare resources. For that reason, as noted in the first installment of this series, risk stratification is represented by “cost” in the Value Equation.
Stratifying risk occurs in many forms, and tools exist to support each. Disease-specific and acute condition risks are often assessed through instruments like the GOLD Score for COPD, the ASA score for preoperative patients, the HEART Score for chest pain patients in the ED, and the SOFA Score for sepsis. All are designed to inform medical care of a specific patient at a point in time. Chronic disease risk tools can be useful in population health management, but those for acute conditions are less helpful in that context.
To best inform pop health management and value creation, we need assessment of behavioral health issues, risk of readmission, social risks, and the holy grail of risk stratification: rising risk.
Unfortunately, behavioral health and social risks tend to get overlooked in evaluating a patient. Behavioral/mental health conditions are often the silent obstacles in an otherwise sound plan of care. The comorbidity of depression and anxiety with chronic conditions exponentially increases a patient’s risk of adverse events and poor outcomes, so our failure to screen and treat these behavioral/mental health conditions sets a patient up for failure.
Population Health Management Must Understand the Social Factors in Patient Health
Social risks are the Social Determinants of Health (SDoH) that negatively affect a patient’s health and well-being. The World Health Organization defines SDoH as the conditions in which people are born, grow, live, work and age. Keep in mind, these conditions can be positive or negative determinants.
When one or more SDoH has a negative effect on a person’s ability to maintain health (like lack of transportation, inadequate housing, food insecurity, social isolation, etc.), we define that as a social risk. Our job is to identify the social risks negatively impacting a patient, so we can adequately inform their plan of care.
Mental and social forces can exert far more influence on a patient’s overall health than the medical care we prescribe. For that reason, behavioral health and social risk screenings need to be as commonplace in the practice of medicine as the evaluation of physical vital signs like pulse, blood pressure and oxygenation. They may be normal and reassuring, or they might signal a problem that, if left unchecked, could seriously compromise the health of the patient.
When we recognize individual-level mental health and social risk factors, we can devise interventions to address them specifically, as well as collect cohort data to understand where community-level interventions might improve a population's health. These insights can guide providers and health systems in devising specific strategies at the patient or community levels, as well as direct where providers and health systems need to act as expert resources instead, letting others lead through community programs or policy and legislation.
Identifying Readmission Risks
The risk of readmission is another gage used by health systems. Advances in predictive analytics have afforded us the opportunity to take discrete fields of data — medical, social, geographic, etc.—and combine them in algorithms to determine a patient’s risk of an unplanned return to the hospital. These same predictive stratification principles can be applied to a patient before a hospitalization event to determine whether an emerging, or rising, risk exists.
Top-of-the-line stratification means proactively finding the rising risk patient before they have an event, rather than reacting to high cost episodes, which is much less effective. Most research shows that high cost patients either have conditions that don’t lend themselves to management alterations (chronic kidney failure, late-stage cancer, etc.), or have issues that, when left to their own devices, allow their healthcare needs and spending to regress to the mean. In other words, to be most effective, we need to find patients and intervene before their risks and conditions cause them to land in the hospital.
Consider the example of two 68-year-old, male patients with diabetes and hypertension, both with documented “well control” of their diabetes per a Hgb A-1c blood test (7.0 for both).
Mr. A is in a committed relationship, sees his PCP regularly, fills his prescriptions on time, gets his bloodwork done, has access to healthy food and reliable shelter, and is active in his social circles.
Mr. B’s wife, on the other hand, died six months ago, he hasn’t been seen by his PCP in nine months, has been late filling his prescriptions three out of the past six months, failed to get the last ordered Hgb A-1c, lives in a food desert, rarely leaves his house, and does not have access to a car.
It doesn’t take a genius or a complex predictive analytics platform to discern that without some interventions, Mr. B is far more at risk for an adverse event like an Emergency Department visit or a hospitalization. Information is key, so the more we know about patients and then apply that knowledge to our assessment of them, the better understanding we will have of the risks and challenges they face.
The Best Health Risk Stratification Leads to Action
Numerous stratification engines are available on the market. AI and predictive learning can be helpful, but sometimes it just comes down to asking. Additionally, the best risk stratification tool with the highest C-score (a measure of a tool’s ability to predict risk accurately) matters very little if no meaningful action is taken on the insights yielded in the risk stratification process.
Analysis and the creation of actionable insights are useless without action or the means by which to act. More on this next time as we discuss the second pillar of population health management – plan of care.
For now, just remember you can’t address what you don’t assess, so uncover the risks and stratify the needs.
5 Tips for Successful Risk Stratification
· Decide what risks you want to assess in your cohort, and don’t forget social and mental health issues
· Find the people who would most benefit from your help and interventions, BEFORE they have problems preferably
· Assess social risks with the same frequency and diligence as checking physical vital signs
· The more information you can gather and analyze, the better. Use predictive analytics tools when possible
· Be sure your risk stratification points you toward actionable steps you are prepared to take
