For self-insured employers, one purpose of predictive analytics software is to tell a narrative – to expound on the history of an employee population’s health. Lab data, claims and biometrics data (to name just a few data-sets) are analyzed and structured in a cohesive, refined and chronological order to help predict risk and avoid substantial health plan costs.
A data analysts’ job is to understand that narrative and find achievable ways that plan members can be made to work in unison to optimize population health, and hence the health of a company’s bottom line.
However, not all data is equal.
Some data is less comprehensive, resulting in holes (costly risks) throughout it’s narrative. Individual behavior and socioeconomic factors are often overlooked for lack of experience and wherewithal, or quite simply, for lack of technological resources, and thus employers often fail to avert avoidable risks.
In the words of sage financial icon Warren Buffett, “risk comes from not knowing what you’re doing.”
Translation: risk is not taking that extra step to understand all of the factors that are affecting your financial outcome.
The “Typical” Approach to Identifying Risk
Traditionally, self-insured companies identify risk by evaluating the most available factors and data: existing health conditions, age, population health trends, how much conditions historically cost them. And there’s nothing incorrect about this approach. In fact, it’s well-known industry-wide that typical population health management solutions aim to reach the 5 – 7 percent of high-cost members in a population to try to effect change in outcomes and curb costs.
What this approach doesn’t take into consideration, however, is each member’s behavior.
An Example: Company X
Take Company X, which was experiencing significant increases in its premiums using a traditional approach to managing its self-insured population. Company X had shopped the payer, narrowed the network, increased the deductible and co-pays, implemented a disease management program and still the costs were increasing.
Using a more focused analysis of each individual in the health plan with a 2-year retrospective analysis, it was discovered that:
There were a significant number of plan members who were seeing 12 or more providers in a 12-month period. In addition:
7% of the group had 7 or more providers writing prescriptions; and
almost 11% of the group used 3 or more pharmacies when filling their prescriptions.
The typical analysis of the population would have evaluated the diagnosis and financial elements of the plan, but would not have included some important characteristics that result in high risk to the individual and the employer.
The Missing Link: Behavior
What we’ve found is that so-called “big data” isn’t much use to a company unless that data can be pared down into actionable intelligence – specifically, what can an organization do that isn’t an overly broad, sweeping change or an employee-wide health initiative that relies on a vague hope that people will participate and health costs will decrease?
To effect lasting, meaningful change, you need to do more. You need to know what behaviors are actually causing the rise in costs and be able to predict future risk and potential spikes in healthcare spending. That’s the promise of behavioral analytics.