Pinpoint what’s driving 60% of ER overuse costs by targeting only 20% of members for improved ROI.

It was a year ago that Linda took over as GM of the health insurance company’s regional business unit and six months since she had signed off on the multi-million dollar analytics package. In the intervening six months, the analytics team worked hard — launching dozens of programs to reduce medical costs through various consumer engagement tactics and channels. The team had a number of projects in progress to reduce emergency room overuse because it was such a visible and costly use case. They targeted everyone they could — from rural populations to expectant mothers and urban families.

However, after six months of hard work, Linda still didn’t have solid answers to two simple, yet crucial questions:

  1. Which plan members are driving ER overuse costs?
  2. Where should the team focus their consumer engagement efforts for the highest ROI?

Despite launching dozens of consumer engagement programs, the team’s “spray and pray” approach to containing medical costs wasn’t working.

But what if Linda’s team could pinpoint 60% of ER overuse costs by targeting only 20% of members? What if they knew which members were likely to drive up costs? Her team could both save money on program spend and reduce ER overuse costs at the same time.

Getting More from Your Data

While Linda’s story is an example, it mirrors what many healthplan executives are facing every day. To put a spotlight on how healthplans can arrive at new results by approaching their analytics differently, NextHealth analyzed data from a leading regional insurer with $25 million in quarterly ER costs. This case study illustrates how large healthcare insurers can drive new insights from data they already have to reduce medical costs at scale.

Here’s how we did it. NextHealth analyzed first- and third-party data from 232,750 members using advanced algorithms. The analysis looked at all members including those who previously used the ER and those who had no historic ER use. The analysis was designed to provide clarity on which members were likely to overuse the ER and which members were likely to drive ER overuse costs. The analysis was designed to predict both ER use and ER costs.

Two types of data were analyzed: first-party member medical and pharmacy claims data from the healthplan; and third-party demographic, location-based and behavioral data. The NextHealth team organized and structured the data to make it ready for analysis and insight, turning otherwise static and ‘dumb’ big data into dynamic and ‘smart’ big data. “Combining first- and third-party data this way created rich member profiles that were more predictive of future behavior,” said Dr. Doug Popken, Senior Vice President of Analytics for NextHealth Technologies.

“An insurer that targets just the top 20% of members according to NextHealth’s predicted risk would actually target a much higher number of people who use the ER than when deploying a random approach,” said Jeremy Schendel, Senior Data Scientist at NextHealth Technologies.

By using NextHealth’s predictive model, an insurer can capture 60% of the ER cost in the population by targeting only 20% of the population. This leads to a much higher ROI for each dollar spent on outreach to members when trying to lower ER costs.