Condition Cohorts – Risk Stratification Analytics Part 4
It is common practice for those looking to improve the health of a population to first start by categorizing patients into different condition categories. One of the more common approaches is to segment a population into condition cohorts, especially to identify those conditions that are of high prevalence, higher cost and commonly linked to care management programs. Creating condition cohorts helps:
- Characterize the prevalence of a specific disease within a population
- Compare condition distributions across different populations
- Equip clinicians and teams for care management activities
Condition Prevalence in Context
When evaluating how common a condition is in a population, it is helpful to understand which conditions are more or less common than in the average population. Standardized Morbidity Ratio (SMR) outputs help you to gain a better understanding about what condition frequencies are abnormal.
Is the number of cardiovascular conditions I am seeing in this population out of the norm, or what one would expect? To make this distinction, the frequency of cardiovascular conditions in your population is compared against an expected frequency of cardiovascular conditions in a ratio, the Standardized Morbidity Ratio (SMR). These expected values are derived from a variety of built-in reference populations, helping you make a more accurate apples-to-apples comparison between the populations based on the sex and age of the populations.
You may find that the difference isn’t really a difference at all and something that might occur naturally. Or, you may observe a discernable difference and gain insights into the magnitude of the variation.
A SMR of 1.0 indicates an exact ratio match between what you would expect naturally and what is being witnessed within your population of analysis. A value less than 1.0 indicates that the frequency of that condition in your population is lower than you might expect, whereas a value greater than 1.0 indicates a greater prevalence. If the difference is statistically significant, more or less, there will be a (+) or (-) flag indicating such.
In this above example, you see can see that the number of patients that have cardiovascular conditions are lower than what you’d expect, whereas the number of patients with respiratory conditions are higher, even statistically so, and may be a factor driving higher levels of population risk.
With all the data in our health care system, there is a strong need for flexible analytic platforms to support population health under a variety of strategic aims – whether that be from the perspectives of specific conditions like we detail here, overall morbidity or a customized set of risk variables. There is no “one size fits all” methodology to managing and improving the health of populations but a flexible platform that aligns to the unique data needs will foster care delivery improvements for better health outcomes.
About the Author
Director of Quality Programs at KPI Ninja, Inc.
Renee provides operational leadership of quality initiatives at KPI Ninja. Towne has a background in occupational therapy, education and experience in operational excellence across a variety of healthcare domains. Based on prior experience as a clinician that drove outcomes patient by patient, she is leaving a larger footprint by improving health care more comprehensively, population by population.
About KPI Ninja
KPI Ninja is a data analytics company that helps healthcare organizations accelerate their quality, safety, and financial goals with a unique combination of software and service. We are differentiated by our signature mix of technology, performance management consulting and healthcare expertise. We don’t merely offer software solutions but work shoulder to shoulder with clients to help them draw on the power of analytics and continuous improvement methodologies to become more efficient. In harmony with our data-centered ethos, we truly believe that our success is strongly co-related with yours.