Monday, December 9, 2019

U.S. Senators Target UnitedHealth, Humana, Aetna, Cigna, BCBSA with Algorithm Bias Inquiry; Asks for FTC Probe


By Greg Shulas December 4, 2019
The recent public scrutiny of racially biased algorithms at a UnitedHealth affiliate has been escalated on Capitol Hill as two U.S. senators have launched an inquiry into the matter. Their effort involves sending out letters to five health insurance organizations — as well as the CMS — that include explicit questions about how these payers are using algorithms in their operations.
Moreover, the lawmakers have asked the Federal Trade Commission if it has any plans to undertake “an investigation” into algorithms that “unfairly discriminate against members of protected classes”— essentially saying that the FTC should be probing this matter, if it has not already.
The five health insurance companies to receive letters from U.S. Senators Cory A. Booker (D-N.J.) and Ron Wyden (D-Ore.) are: UnitedHealth GroupHumana, AetnaCigna and the Blue Cross Blue Shield Association, or BCBSA. No other payers so far have received letters from the lawmakers about this topic.
The review follows a report in the journal Science that concluded an algorithm created by UnitedHealth affiliate OptumHealth — which helps refer patients to critical care programs — inadvertently contained a bias against African-American patients, as reported on by Health Payer Specialist. The October Science article has already sparked separate probes by the New York Department of Financial Services and the state’s Department of Public Health.
At issue is the algorithm’s apparent use of healthcare costs as a proxy for a patient’s true healthcare needs. Such an approach would fail to take into account a minority patient's lack of equal access to healthcare facilities in communities, as well as “low levels of trust” in local health systems, the lawmakers noted. Experts have told Health Payer Specialist that the OptumHealth matter is far from the only case of less-than-neutral algorithms in the burgeoning world of healthcare analytics.
Questions that the lawmakers posed to the payers, among others, include:
·         What specific decisions are these algorithms making?
·         How many people do these algorithms impact?
·         What is [the payer] doing to ensure that these algorithms are free from bias?
·         How often are audits done to detect bias?
·         Do you compare performance before and after algorithm implementation to ensure the algorithm actually works and does not maintain or increase bias?
·         Do any of the algorithms that [the payer] is currently utilizing include a bias that would negatively affect certain patients relative to others?
·         Will [the payer] commit to immediately halting the use of these algorithms where possible, until such bias is eliminated? In cases where the algorithm cannot easily be halted, will [the payer] commit to providing adequate resources to investigate and fix the problem?
The senators acknowledge in their letter to the FTC that UnitedHealth is seeking to actively address the issues exposed in the algorithm. In a special Health Payer Specialist report Wednesday about the algorithm matter, a UnitedHealth spokesperson told the publication that the organization appreciated the work done by the Science researchers, and noted that algorithm “was highly predictive of cost, which is what it was designed to do.”
Both Booker and Wyden did note that algorithm can serve an important role in healthcare delivery by removing human flaws and prejudices from clinical processes. However, the senators stressed how the data sets that the algorithms rely on are influenced by “historical and human biases” and thus can’t be entirely trusted, according to their letter to Cigna’s CEO.
“Technology holds great promise in addressing these issues and improving population health. However, if it is not applied thoughtfully and with acknowledgment of the risk for biases, it can also exacerbate them,” the senators wrote to Cigna Chief Executive David Cordani.

No comments:

Post a Comment