Predictive Modeling: The Second Most Important Ingredient for Provider Accountability

Jaan Sidorov

First posted 10/04/11 on The Disease Management Care Blog

“Accountability.”  Everyone wants it, right?

While it’s one thing for health care providers to be “accountable” for costs, it’s another for them to actually make money at it.  The Disease Management Care Blog is continuously amazed at how many physicians and administrators believe that dollops of “primary care,” “prevention” and “wellness” will empty hospital beds and cause insurance money to appear like the morning dew on a windshield of a physician specialist’s BMW.

Believe that and the DMCB has an ORD-SFO United Airlines upgrade “departure management card” it would like to sell you.

If the DMCB could do only one thing to reduce costs within one fiscal year for a health care system , it wouldn’t be PCPs, wellness or prevention. Instead, it would place one vastly overpaid case manager in every emergency room in the network.  His or her job would be to find alternative levels of care for persons that don’t really need to be in the hospital and begin discharge planning for those who do.

The second thing the DMCB would do is implement predictive modeling.  That’s why it’s immodestly tooting its horn and suggesting research like this is so important.  In this web first publication, Chris Hollenbeak, Mark Chirumbole, Benjamin Novinger, Frank Din and your humble DMCB examined the baseline demographic data and medical conditions that can predict the likelihood of a future high cost hospitalization within a Medicaid population.  Armed with knowing which of its patients are at risk ahead of time is a critical level of intelligence for any newly “accountable” hospital, clinic, IPA, PHO or IDN.  By reaching out to high risk individuals prior to any crisis, physicians and administrators can engage these patients in outpatient settings and keep them away from the emergency room.

The good news is that the art and science of predictive modeling is within reach of most data bases, desk-top computers, statistical software packages and trained analysts.  The bad news is that interpreting and executing on that information remains a daunting challenge.

Health systems like HealthCare Partners is a good example of how it can work well.  As this “accountability” movement in health care evolves, the DMCB will undoubtedly learn about other good examples.

Health systems that rely on vague notions about primary care will form the bad examples. They will fail to understand the role of case management and ignore the business case for predictive modeling.

About Brian Klepper

Brian Klepper is a health care analyst, commentator and a Principal in Health Value Direct.
This entry was posted in Analytics, Medical Management, Physicians, Quality and tagged , , , . Bookmark the permalink.

2 Responses to Predictive Modeling: The Second Most Important Ingredient for Provider Accountability

  1. Brad F says:

    Jaan
    Identifying high risk folks is one thing. Engaging them to change prior to those events occurring is much more difficult.

    I have done some work examining predictive factors for readmits. Greater than 65 yo, more than one chronic condition, non-English speaking/low SES, etc are senstive, but not specific (AKA, “tell me something I dont know).

    The solutions go way beyond an ER CM’er, and predictive scores. Its one piece, but not close to a home run; only one cog in the bigger machine that needs to coalesce to get patients home and healthy.

    Brad

  2. Brad is absolutely correct. In retrospect, my description of the “ingredients” for this cake is only limited to two items. There are other resources and interventions that are needed before you can bake up some real accountability.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s