Predictive Analytics Strengthen Readmission Risk Models by Defining Socio-demographic and Behavioral Attributes of Patient Populations

Targeting a defined patent population for reducing readmissions is all well and good but effective and meaningful readmission risk models require analytic data that provides insight into fact-based decision making for scope, management, outcome measurement and learning. According to a 2015 US Hospital and Health System Analytics Survey directed by Deloitte, the investment of predictive analytics for the purpose of successfully managing populations will be seen as a growing priority over the next three years.

Reducing readmissions has long been a thorn in the sides of many hospitals since the inception of the Accountable Care Act. Over recent years, a number of readmission risk assessment models have been developed, but yet readmission rates continue to be a sore spot for hospitals. At Connance, we recognize the value of identifying risk for readmission and have developed predictive analytic solutions that strengthen those models and enables more meaningful, timely action.

Traditional readmission risk models, are limited in their capabilities of providing a holistic understanding of who a patient is. The ability to segment populations to uncover the finer details of a single patient within a population cannot simply be accomplished through extracting historical data from EHR, claims and outdated information on patient populations. Disease-specific models also commonly used by hospitals provide an extremely narrow view of a patient population as it predicts readmission risk based on disease (CHF, COPD, AMI) only leaving out many other contributing factors such as other chronic conditions and comorbidities. Other ineffective methods of assessing readmission risk have been through collecting data at discharge which is neither timely nor meaningful for applying proactive preventative measures by clinicians. Often these routine practices of gathering information to make determinations on readmission risk counteract with an efficient and effective workflow resulting in frustrated clinicians.    

Predictive analytics can enhance the numerous readmission risk tools being used by delivering actionable insight into the hands of clinicians responsible for mitigating post-discharge failures. Connance, through its “Whole Patient Insight,” is driving healthcare organizations towards greater success in reducing readmissions. Connance enables healthcare providers to easily recognize socio-demographic and behavioral attributes of homogeneous populations that impact an organization’s efforts in improving resource allocation that affect cost and quality measures.

Connance is able to accurately determine which patients are at greatest risk for readmission and strongly advise where clinicians should be focusing their time and attention on. We’ve illustrated this in the below diagram depicting two scenarios on how “Whole Patient Insight” can drive healthcare organizations in their mission to reduce readmission risk.

sepsis

The top of each tier above shows how two patients can look very similar when assessing for readmission risk through traditional measures. At first glance you’ll notice the average readmission rate for our two 69 yea-old septicemia patients with a 4-day length of stay is at 19%. As we move down the decision tree Connance applies “Whole Patient Insight,” which is our predictive analysis of socio-demographic and behavioral attributes of our two septicemic patients. Through our analysis of the patients household size, home market value to income, car ownership, marital status, food assess, community, and more,  we are able to drill down to the finest points of the whole patient to determine who is truly at risk for readmission. Based on these numerous socio-demographic and behavioral attributes Connance is able to provide the true risk for readmission.

Patient A now has a risk of 6%, opposed to Patient B, with a risk of 60% for readmission. This valuable insight drives clinicians proactively towards applying time and resources to patient B. Resources are most effective, and cost effective, when applied to those patients most at risk, and not inefficiently allocated on others with no/low risk. Ultimately, not only has Connance defined in great detail the difference of two patients that are in the same patient population, but has also provided more accurate readmission scores from that which was presented at the very top of the diagram, that’s the power of “Whole Patient Insights.”