Berwick: ACOs incentivize 'coding games'

September 11, 2012

The financial incentive of shared savings may lead to "coding biases" whereby practices list more diagnoses per patient visit than they otherwise would, according to a recent study.

A new study of a federal program that created the precursor to accountable care organizations (ACOs) suggests that ACOs can incentivize “coding games.”

The study, which was undertaken by the researchers from the Dartmouth Institute for Health Policy and Clinical Practice and published in the Journal of the American Medical Association, examined data from 2001 through 2009 from 10 organizations participating in the Medicare Physician Group Practice Demonstration (PGPD) project.

In the PGPD program, participating physician groups were eligible for up to 80% of any savings they generated if they were also able to demonstrate improvement on 32 quality measures, including the adequacy of preventive care and the effectiveness of chronic disease management.

It’s that financial incentive of shared savings that may have led to what the study’s authors call “coding biases” in which practices participating in the PGPD project listed more diagnoses per patient visit than practices at control sites did. The more diagnoses a patient has, the more severe and costly that patient’s expected condition and thus the more potential savings a practice is able to generate from that patient.

In an editorial accompanying the study, Donald Berwick, MD, former director of the Centers for Medicare and Medicaid Services (CMS), said, “The study also offers clear evidence of coding games - lawful but games nonetheless - as PGPD sites tended to list more diagnoses at visits than control sites did.”

Berwick said the “coding games” show the need for regulatory oversight of the ACO program, which was a major component of the Affordable Care Act.

“Neither patients nor the nation are well-served when administrative manipulations masquerade as changes in care,” Berwick said. “What is needed is better care, not better coding.”

Overall, the study found “modest” savings associated with the PGPD project - $114 annually per Medicare beneficiary - but that savings varied widely across organizations.

For example, one group reported savings of $866 per beneficiary, whereas another reported an increase in spending of $749 per beneficiary, according to the study.

The study’s authors said their research shows that ACOs have the potential to slow spending growth, but the vastly different results achieved by different participants in the PGPD project underscore the need for better understanding into how payment reforms affect healthcare practices.

Berwick said the study shows that ACOs offer potential, but they’re just a small piece of the overall puzzle when it comes to improving healthcare while lowering costs.

“ACOs deserve energy, investment, discipline, and good faith; they can help,” he said. “But, whether encouraged or discouraged by the PGPD experience, a lot more innovations than ACOs alone will be needed to emerge successfully from this fraught time.”