OBJECTIVES: Health plans expend effort to try and predict the less than 1% of their members who will need extra support in order to maintain their health status by avoiding hospitalizations and emergency department visits. In addition to identifying high risk members, there must also be interventions available such as disease management programs and member participation in order to impact behavior and mitigate risk. This research pilots the development of a new metric termed “impactability” which is a score that evaluates a high risk members likelihood of benefitting from an intervention.

METHODS: Blue Cross impactability score was developed using machine learning algorithms with over 8,000 data elements such as patient demographics, credit score, family size, social support system, ZIP code, job title, location of workplace, and medical history. Data were combined with risk scores, social determinants, program evaluation results, and clinical feedback to create the impactability score housed in an enterprise wide data warehouse and made available to healthcare providers and Blue Cross case managers via a dashboard.

RESULTS: Preliminary results show that providers are increasing their engagement which is leading to more evidence based decision making. For example, a member that is over the age of 80 with end stage renal disease in hospice care is a high risk member; but his or her health status is not really impactable; whereas, a patient newly diagnosed with cancer is also a high risk patient, their impactability factor is high. There are care management programs that can be offered to mitigate this patients risk while reducing costs.

CONCLUSIONS: The impactability score allows Blue Cross the ability to evaluate a patient’s likelihood of benefitting from a particular intervention which in turn directs Blue Cross and providers to the right member, at the right time, through the best intervention approach maximizing care and reducing costs.