We know that data analytics is being used to influence a wide range of things such as the pair of shoes one might want to buy or what news is “trending” on Facebook. Similar tools are being applied to employer-sponsored group health plans. According to a recent HealthcareITnews article, vendors such as Advanced Plan for Health (APH) are using predictive modeling functionality to support population health management. The ability to better anticipate and manage plan costs while shaping plan design to meet the needs of plan participants likely will be very appealing to plan sponsors, but employers should think through implementation carefully.
According to the article, these products (APH calls its product “Poindexter”) can make predictions about when certain health events are likely to occur (such as an ER visit), or forecast the nature of the services to be provided (such as the length of the participant’s hospital stay). We will leave to the data scientists to describe how this sausage is actually made, but here is how it is summarized in the article:
Currently, the Poindexter engine calculates care gaps and predicts the likelihood of hospital admissions, as well as readmissions, 6 to 12 months in advance for any given patient population — typically covered lives in a self-insured employer’s health plan. The tool also examines data from claims, pharmacy and clinical sources, benchmarking against real-world health data adjusted for comparable demographics, geography and industry of the employer.
Poindexter assigns risk scores to individuals within that population – identifying people whose health profile suggests elevated risk. With this information, case managers can improve outcomes and lower costs when they help patients avoid catastrophic events by improving their health through timely interventions.
One thing seems clear about this process – there’s a lot of data, a lot of very sensitive data, involved that is coming from a number of different sources. Certainly, data privacy and security compliance, yes this means HIPAA, must be taken into account by employers when considering whether and how to apply these analytical tools to their group health plans. Employer-sponsored wellness programs have raised similar issues as participants often must tender personal health information about themselves to take advantage of incentives under those programs.
Speaking of wellness programs, if analytics can predict and help employers better design their health plans, couldn’t the technology also be used to help prevent or put off more adverse and expensive health events. That is, in the course of “population health management,” would it be unreasonable to expect that a health plan that can reasonably anticipate or predict a significant health event would take some steps to try to prevent it from happening? Coupling analytics with traditional wellness programs, incentives perhaps could be more targeted to better steer participants toward healthier behaviors or to get care sooner and less expensively.
In the course of administering benefit plans with features like these, keeping protected health information anonymous may be easier said than done. Additionally, providing inducements can raise issues under HIPAA, the ACA, and the Equal Employment Opportunity Commission’s ADA and GINA regulations, which also have confidentiality protections. So, as technologies like analytics emerge to power employee benefit plans, particularly health plans, they need to be run through the array of law and regulations that apply to those plans.