Risk Stratification: A Two-Step Process for Identifying Your Sickest Patients
A proposal in two steps:
Step one involves sorting patients into one of three risk groups (high, medium, and low) based on objective data, which we take from claims or our electronic health record (EHR). We make our determinations based on the presence or absence of such factors as chronic conditions, advanced age, multiple comorbidities, physical limitations, substance abuse, a lack of health insurance, low health literacy, frequent hospitalizations or emergency department (ED) visits, recent major surgery or brain trauma, polypharmacy, or difficulty following a treatment plan. Some EHRs will calculate a risk score automatically based on this data. In either case, it is important to adjust the score based on additional, subjective considerations, which are the focus of step two.
In step two we assign each patient to one of six risk levels based on how physicians and staff answer the following questions:
- Is the patient healthy with no medical problems? If so, are his or her biometrics in or out of range?
- Does the patient have chronic conditions but he or she is doing well?
- Does the patient have chronic conditions that are out of control but without complications?
- Does the patient have complications of chronic disease or high-risk social determinants of health? (If you or your care team are unsure how to assess or address a patient's social determinants of health, the AAFP's EveryONE Project1 includes tools and resources.)
- Is the patient potentially in danger of dying or being institutionalized within the next year?
I missed any reference to morbidity and episode measurement. My impression is that objective measure is the first mandatory step through groupers, and risk stratification improves with subjective sources of information. And levels can be allocated through probabilistic fuzzy systems.
