Add to Technorati Favorites

 Subscribe in a reader

Trends and Outliers

TIBCO Spotfire's Business Intelligence Blog


Prescribing BI and Analytics For The U.S. Health Care Debate

j0409702 150x150 Prescribing BI and Analytics For The U.S. Health Care DebateOne sure thing in the cloud of uncertainty regarding changes to the U.S. health care system: analytics and data verification will be critical to reviewing thousands of pages of documents and the charts, graphs and statistics used by various industries in the debate.  When Congress convenes to examine H.R. 3962 there will be a lot to analyze.  Entire global industries that have grown by sharing broad, unknowable risks– insurance and especially health insurance – now face the 21st century realty of predictable outcomes and specific forecasts.  Tom Davenport, co-author of Competing on Analytics: The New Science of Winning, writes about the future prospects of change with clear-eyed realism.

When a company or organization seeks health insurance for a large group of employees, the particular attributes of those employees aren’t usually assessed in detail.  “Increasingly, however, life and property & casualty insurers have attempted to increase their profits by predicting just how much risk a particular customer represents, and pricing the risk accordingly,” he writes at the Harvard Business School Press website. “Pooling the risk, it seems, is no longer an attractive proposition for life and property insurers.

Yes, some elements of your personal medical history are protected INSIDE of doctor offices and hospitals by HIPAA and other laws, there are plenty of other unsecured and correlated data outside those offices that can link to greater risk of early death, behavioral, financial or other risks.

“All you need to know is how much someone weighs, what kind of food he or she eats, how much exercise they get, and so forth.  Much of that information is publicly available, can be bought, or can be legally requested in insurance applications.  One health insurance actuary told me that such “lifestyle” data is a much better predictor than age of who is going to contract, say, diabetes.  Among 45-year olds, for example, there is an eightfold difference in annual medical spending between the highest-risk lifestyle group and the lowest,” Davenport writes.

Lifestyle indicators are already gaining popularity in health insurance firms, who enroll certain customers in “disease management” programs. Some can reduce risks of certain conditions.  But Davenport notes that data can also be used to refuse coverage, or to price coverage at a much higher level.  Sometimes the details unearthed by advanced analytics can have unexpected or negative results.

The prospect of universal coverage in the United States could move more people to shop for insurance as individuals (and trackable) instead of as employees who might be harder to classify as group.  It will be possible for insurance firms to identify which customers will be profitable, and which will cost too much money to insure.  And some insurers will be better at predicting risk than others.  “This will lead to dramatic differences in performance between the more and less analytical health insurers.

Some will go out of business, creating disruption for the entire industry and its consumers,” Davenport predicts.  “If there is a “public option” that takes consumers no one else wants, it will undoubtedly get the citizens who are most likely to acquire expensive diseases.  Taxpayers will foot the bill, while the private health insurers who are good at prediction will become much richer.”

Auto insurers, for instance, have used “risk adjustment” in many states, sharing losses when they arise and preventing companies from skimming only the most desirable clients.  Another view comes from the public radio program “Marketplace” which found healthcare outcomes showed worse care or repeat hospitalization for minority patients compared with native English speakers.  A Nov. 23 report explored ways hospitals and insurers are looking to better data for improving predictive, responsive information.  For example, cultural details identifying someone more specifically as Cambodian, Indian or Japanese instead of more generically as “Asian” can make a difference when communicating care or prescription drug use.

David Wallace
Spotfire Blogging Team

Image Credit: Microsoft Office Clip Art

Print post

Is your data hiding something?
Download Your Free 30-Day Evaluation of TIBCO Spotfire®

Post a Comment

Your email is never shared. Required fields are marked