A wide range of models in the marketplace can identify missed charity-eligible accounts. If a hospital is considering implementing predictive analytics, the following elements should be discussed during the evaluation process.

Local calibration. Poverty is heavily weighted to local economic circumstances and socioeconomic attributes. Better predictive analytics will be calibrated during implementation to the hospital's specific community.

How the model handles households without bank accounts and credit files. Many of the poor in this country live in the financial shadows. However, some of those who do not have bank accounts live in wealthy households. Predictive analytics deal with these "unbanked" patients differently. For a model to simply assume that "unbanked" equals "poor" has been proven to be incorrect. Similarly, assuming an average for all the unbanked is also incorrect. Hospitals need to understand how a model specifically handles the unbanked.

Information required. Some models require a current address and guarantor social security numbers for scoring. Understanding differences in data requirements is important as it can have significant impact on patient access.

Portion of accounts a model cannot evaluate. Better models will have broader coverage (e.g., fewer accounts that are not able to be predicted or assessed). Some models cannot evaluate as high as 30 percent of their accounts, while others can keep this group as low as 1 to 2 percent. The greater the coverage, the better.

Difference between credit scores and poverty. A household can have an excellent credit score and still be poor. A solid credit score means only that the household lives within its means. Similarly, low credit scores imply only that the household lives beyond its means. Poverty is not a function of spending, but a sociodemographic reality.

Socioeconomic factors. In addition to income, a number of other factors are known to correlate with living in poverty. Predictive analytics can incorporate socioeconomic attributes that improve a hospital's ability to consistently deliver community benefits to those parts of its community with the highest levels of need.

Sliding-scale calibration. Models differ in the extent to which they can be tuned to a hospital's sliding scale.

Value-added information. Some solutions, along with evaluating for charity, will also check addresses and contact information, look for fraud, or scan for bankruptcies or death. In this way, the model analyzes the portfolio to facilitate financial counseling or future collection efforts.

Acceptance by IRS, regulators, and other organizations. With many vendors offering models, hospitals should understand the extent to which specific models have been used in previous filings or been recommended as an effective solution.

Shari Bailey is vice president, Verité Healthcare Consulting, LLC, Washington, D.C. (shari.bailey@veriteconsulting.com)

David Franklin is chief development officer, Connance, Inc., Waltham, Mass. (dfranklin@connance.com).

Keith Hearle is president, Verité Healthcare Consulting, LLC, Washington, D.C. (keith.hearle@veriteconsulting.com).

The information provided herein does not represent tax advice or advice regarding how states define and value charity care for reporting and/or disproportionate care revenue purposes. Organizations should consult their tax advisers in preparing any tax forms. 

Publication Date: Thursday, April 01, 2010

Login Required

If you are an existing member, please log in below. Username and password are required.



Forgot User Name?
Forgot Password?

If you are not an HFMA member and would like to access portions of our content for 30 days, please fill out the following.

First Name:

Last Name:


   Become an HFMA member instead