As health care consulting and cost analysis become more and more complicated, brokers can help clients reduce costs by using data mining - a field that is constantly evolving with new technology and tools.

"Employers want to know what's driving their trends and their costs. [They] want to go beyond the high-level costs such as hospitalization versus prescription drugs," says Edward Kaplan, SVP and national health practice leader at Segal Co. "It used to be that 10 to 15 years ago they would say, 'We are paying $10,000 a year for a family of four and how much of that is hospitalization?' ... Now, the clients are demanding" to know what drugs are the driver of the prescription costs, he explains.

"They are demanding more detailed inspections about what are the root causes of their costs," he adds, and data mining typically lets an adviser find that root cause to satisfy client needs.


What is data mining?

Data mining is a set of techniques that allow users to identify patterns and extract significant facts from potentially large volumes of data, explains James Taylor, chief executive officer of San Francisco-area Decision Management Solutions, a data mining company.

A broker typically will offer the tools as part of a service offering, keeping the data analysis themselves, but also providing clients direct access to the information, Kaplan says. Some brokerages have their own software, but many often contract with some of the large data mining players.

Brokers and advisers can then use the information output to recognize trends not seen before and build a model that allows them to predict future behavior and often reduce future cost, adds Stuart Rose, global insurance marketing manager at Cary, N.C.-based SAS, a business analytics software company.

It is important then to have a clear objective in place about how the change can happen. "At the end of the day, the only value to data mining is in its ability to change behavior," Taylor says. "Those using data mining should be clear which decision they are hoping to change, who makes that decision and what it will take to change it in a positive direction."


Where is it going?

In the last few years, the field has dramatically changed. The amount of data that insurance companies have been able to analyze has increased, as has the sophistication of the analytics - allowing companies to get more from the same amount of data, Rose says.

Brokers, insurers and employers have also begun working even more closely to analyze the data, often shipping all the information to one company to mine the aggregated numbers.

"By having the PBM and having the lab all give the data to one data mining company, the clients get all the data and put it in one data warehouse updated every quarter," says Kaplan. " ... The tools are getting sophisticated."

As more and more companies, including smaller ones, began to offer data mining services, the price is decreasing, so much so that a "small employer can pay under $1 a month to have really great analytical data reporting," Kaplan says.

Further, the field is becoming ever "ubiquitous, with companies of every type realizing that the data they have about their customers and their productions is a source of potential competitive advantage," adds Taylor, the author of Decision Management Systems: A practical guide to using business rules and predictive analytics. "More automation and better user interfaces also mean that data mining is becoming more accessible to organizations that can't afford to employ a specialist data mining."


What can it do?

Just 10 years ago, looking at health and wellness, it was similar to throwing spaghetti on the wall and seeing what sticks, says Dr. Ian Chuang, SVP and medical director at Kansas City, Mo.-based Lockton, which has more than 400 client groups in its data mining warehouse, and more than 1.3 million members.

Even the smallest numbers can make a huge difference in predicting future behavior. At smaller employers, a 2% difference in prevalence and norms can account for a large difference in outcome, according to Lockton.

Through data mining, "we've become smarter about what's necessary to move the meter," Chuang adds.

Data mining can also help with future sales. Taylor explains that by segmenting customers based on their purchase behavior, brokers can find groups of customers that are similar, even if they did not appear to be so at first.

"You can find out which characteristics of a customer might make them a good cross-sell target for a product, a retention risk, or similar," he says.


The numbers work

There are numerous real-world examples of when data mining helped reduce cost. Among them are:

Segal's Kaplan points to a client who had been paying for 100% of fitness costs for all employees. The client wanted to know if the gym program was saving them money in health care claims.

Segal took three years of data from people who use the gym sometimes, never and often and found that of those who had steadily used the gym, there was a 17% reduction in health care cost, versus health care costs rising 8% to 10% annually for other employees. Yet, only 3% of employees used the gym eight or more times a month, and in the end the company was paying 10 times more on gym memberships than they were receiving in overall health care savings.

Lockton points to a client who looked at what was driving hospital admissions. Through education over a three-year period, the employer was able to reduce hospital admission by 21%, and emergency room utilization by 19%. Additionally, the company's annual cost trend has been 3.5%, versus that national average of 10% to 12%.


Legal implications

There are often two main legal concerns raised when mining data: Data privacy and data integrity, says Jim Kunick, principal at Chicago-based law firm Much Shelist.

With data integrity, "you have no idea of knowing how accurate it is. You may uncover 100 smokers in a 1,000 [person policy] and people may have lied," he says. "You can't do anything about it and you take the risk."

But with data privacy, it is a lot more "complicated," says Kunick. He points out that states have ever-evolving data privacy regulations, and many are moving toward data privacy laws that are similar to the more stringent ones found in the European Union and Canada.

Taylor agrees that privacy is the key legal concern with data mining, but also its biggest pro.

"Data mining mostly does not care about personally identifiable information - it's looking for patterns, not individuals," he says. "This means that personally sensitive data can be removed or obfuscated and the rest of the data can still give powerful results."

While Kunick does not recall any data mining legal cases recently, he advises those who data mine to take precautions.

"As a company that does data mining, you have to make sure you have a privacy policy and that your data mining does not run afoul of it," he says. "Be careful with the information you do collect, in terms of understanding what you are asking and putting the right agreements [in place] with anybody to who you disclosed that information."

For brokers who may farm the data mining services out, he advises, as with any business deal, to undertake due diligence, use references and see what experiences competitors have had.

Most important of all, Kunick says, is having an indemnity agreement. Meaning, as a broker, "if I get data from you and you provide it without someone's consent and a third party brings a claim against me, you will stand in my shoes and defend me."

Register or login for access to this item and much more

All Employee Benefit Adviser content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access