Risk stratification forms the foundation of effective population health management, providing a powerful framework for strategic resource allocation. Despite how fundamental this capability is for identifying which patients would benefit from additional support, there are a number of common missteps and questions we hear from leaders across the country.
Keep reading for answers to some of the more common ones.
What risk stratification tools do most organizations use for discharge planning?
We've seen a range of tools used, with the LACE Index being the most frequently cited. Others will use disease-specific tools like the Krumholz Model (for Medicare heart failure patients).
For maximum impact, best practice suggests running a regression analysis to determine specifically what is predictive of readmission within a given population. Mayo Clinic found that the four factors statistically significant in predicting readmission risk in its population were: level of disability, whether the patient lives alone, age, and self-reported walking limitation.
Using these factors, Mayo developed an "early screen for discharge planning" algorithm that dictates whether a patient will receive basic or specialized assessments and discharge planning to guide prioritization of effort.
Mount Sinai Medical Center in New York ran a similar type of analysis and found that alcohol abuse, depression, dual-eligible status, and race were highly predictive of readmission for its area. Notably, the researchers also found that hospitalization history alone was a reasonable proxy for their multivariable regression model in predicting patient risk for 30-day readmission for their population.
Which non-clinical risk factors are most predictive of readmission risk?
Few risk assessments validated in academic studies incorporate broad range of both clinical and psychosocial risk factors, which makes it difficult to provide a definitive answer. However, in a review of the literature on commonly used non-clinical risk factors for 30-day readmissions, several risk factors were deemed predictive across multiple evaluations:
- Payer status (particularly dual eligibility)
- History or active diagnosis of mental health conditions (particularly an active depression diagnosis)
- Level of social support (marital status serves as a good measure)
We also looked into the predictive strength of demographic characteristics, financial status, and education level, but did not find these factors to be significant in the evaluations reviewed. This is not to say these types of measures are not important, but accessibility, accuracy, and consistency are data challenges that need to be overcome first.
When should I use non-clinical risk factor data?
Though the value of non-clinical risk factors in predicting patient readmissions remains inconclusive, psychosocial characteristics can still offer meaningful guidance in implementing patient care management interventions.
For organizations that risk stratify patients based on clinical indicators, layering an understanding of non-clinical risk factors enables care managers to more effectively customize patient care. This type of data collection supports targeted, evidence-based care planning, as well as a more holistic understanding of a patient's health.
We've seen a range of multidimensional risk-assessments used, from 5-minute iPad questionnaires administered in waiting rooms to hour-long bedside assessments with a nurse or social worker. These assessments can be customized to meet the needs of your population and map to the supports you're equipped to provide, but common tool areas include housing, food, transportation, and literacy levels.
Next post on risk assessment
3 pitfalls to avoid when assessing patient risk
Do you have other questions on risk stratification or deployment of finite care management resources? Contact me at TyrrellR@advisory.com.