Article by Tia Goss Sawhney

Abstract:
This article will provide readers with a primer on the mechanics of the algorithms underlying today’s prescription drug risk assessment tools. By pointing to specific scenarios where the tools produce illogical results, the author will then discuss some of the most obvious shortcomings of these tools. Finally, the author will describe how underwriters, medical directors, and statisticians could work together to improve the tools.

Background:
Prescription drug risk assessment tools assign risk scores to insureds (or potential insureds) based on their prescription histories. These tools are increasingly used by actuaries and others within health insurance organizations to assess future health risk. Prescription drug and diagnosis-based risk assessment tools are used in areas as diverse as: new business and renewal underwriting for small and medium size groups (and less frequently large groups), utilization and disease management, trend analysis, comparisons between blocks of business, and provider compensation.

Prescription drug risk assessment tools have been built by statisticians and have confirmed statistical validity. Yet when potential early adopter underwriters and medical directors have looked at combining risk assessment tools with the prescription histories available from prescription history profile services, they more often than not have concluded that the risk assessment scores do not adequately reflect their applicants’ potential risk even with respect to producing a relative ranking of applicants.

As actuaries who buy and use prescription drug risk assessment tools, we should be very concerned about the unfavorable evaluation of these tools by underwriters and medical directors. If risk is not being well assessed at the individual level, then it is not being well assessed at the group or block of business level. We should pressure the tool vendors to work with underwriters and medical directors to improve their products.

Published by:
This article will appear in a January 2009 of Contingencies. For more information prior to publication contact Tia Goss Sawhney at tia@tsstrategic.com.