New work by HCP researchers demonstrates that prescription drug utilization for medical conditions is predictive of enrollee unprofitability. This information could be used to remove additional vulnerable enrollees from needed insurance and access to health care.
The drug formulary is one dimension on which insurers (in conjunction with pharmacy benefit managers) can distort their plan benefits, as other dimensions of health plan design, such as pre-existing conditions, are currently highly regulated in The Patient Protection and Affordable Care Act. Thus, insurers are effectively free to use the drug formulary to raise or lower the out of pocket cost of different drugs, thereby making their health plans more or less attractive to profitable or unprofitable groups.
In fact, a recent lawsuit asserts that major insurers in the Health Insurance Marketplaces are designing their coverage so that individuals with HIV/AIDS will be less likely to enroll.
While risk adjustment to control for enrollee health is reasonably successful at weakening the relationship between drug utilization and unprofitability, this new research shows that some drug classes remain predictive of insurer losses. The enrollees whose prescription drugs fall in these classes may need special protection from regulators in health insurance market design.
The paper, “Computational Health Economics for Identification of Unprofitable Health Care Enrollees,” was published in Biostatistics and authored by Sherri Rose and Tim Layton, along with health policy PhD student Savannah Bergquist. The authors deployed ensemble machine learning techniques with an application-specific variable selection tool designed for this research question, and have made the code for their analyses available on GitHub.