In President Obama’s June 2009 speech to the American Medical Association (now infamous for containing the most emphatic articulation of his promise that people who liked their health care plan could keep it), he described how his legislation would reform the way medical providers are paid.

“We need to bundle payments so you aren't paid for every single treatment you offer a patient with a chronic condition like diabetes, but instead paid well for how you treat the overall disease,” Obama told the crowd of physicians.

Obama was articulating what would become one of the key payment reforms in his health care law — a proposal aimed at giving incentives to providers to control costs by rewarding them for providing less expensive care.

But a study published in the journal Health Affairs looked at an ambitious three-year pilot program of bundled payments in California that was funded by a $2.9 million grant from Obama’s 2009 economic stimulus package — and found that the program was such a massive failure, it could hardly get off the ground.

“In spite of a high level of enthusiasm and effort, the pilot did not succeed in its goal to implement bundled payment for orthopedic procedures across multiple payers and hospital-physician partners,” the study reads. “An evaluation of the pilot documented a number of barriers, such as administrative burden, state regulatory uncertainty, and disagreements about bundle definition and assumption of risk. Ultimately, few contracts were signed, which resulted in insufficient volume to test hypotheses about the impact of bundled payment on quality and costs.”

The hope of the bundled payments approach is that by moving toward compensating providers for the full treatment of a medical condition, those providers have less incentive to order more tests and procedures and prescribe the most expensive drugs, because instead of getting paid for each service, they only get to keep whatever is left over after the full treatment is complete.

In addition to controlling costs, advocates of this approach argue that it would improve quality of treatment by creating performance incentives.

But the study notes that despite the furor within the health care policy community for this type of payment reform, “Evidence is lacking on the effectiveness of bundled payment in terms of improving the quality of care, reducing its cost, or both. Existing evidence about bundled payment programs mostly comes from bundled payment designs with more limited scope that have little generalizability to current programs.”

In Sept. 2010, a division of the Department of Health and Human Services known as the Agency for Healthcare Research and Quality awarded a $2.9 million grant from economic stimulus funds to a three-year project aimed at testing bundled payments in the real world. The Integrated Healthcare Association and the RAND Corporation oversaw the project.

The study in Health Affairs, which was also funded from the stimulus grant, was written by Susan Ridgely, David de Vries and Peter Hussey of RAND and Kevin Bozic of the University of California, San Francisco.

Initially, six health plans and eight hospitals signed up for the pilot program, but eventually only three insurance plans and two hospitals participated. The study was supposed to measure costs of treating orthopedic patients under age 65 through bundled payments, but there were just 35 such cases over the three-year period — far too few to draw any conclusions about cost controls.

The program ran into a number of complications, according to the study. One problem was defining the length of time or types of services that would be covered under the bundled payment arrangement, and eventually providers excluded obese patients and rehab services. Hospitals and insurers got into their usual fights about payments — insurers thought they should be paying less because payments were being bundled, but hospitals sought more money because they argued they were absorbing more risk.

Other technical problems arose, the authors write. For instance, insurance company computer systems were set up to make payments the traditional way, rather than in bundles, and there wasn’t enough volume to justify purchasing new software. In addition, the program ran into the problem of regulatory uncertainty given the complex set of rules in place in California.

Ultimately, the study reports, “little experience was gained in actually providing orthopedic services under bundled payment arrangements and learning what the requirements might be for ‘scaling’ bundled payment approaches nationally.

The authors write, “There remains a strong conceptual basis to support a role for bundled payment in bending the cost curve in the health care system. Still, there is limited evidence that bundled payment will achieve its promise.”

This one study doesn’t necessarily disprove that bundled payments could work under different circumstances, and the authors write that the study resulted in “hard-won lessons” that could be useful in future pilot programs.

But its also worth noting that its failure comes on top of other issues raised about the bundled payments approach.

For instance, a May 2012 study reported in the New York Times found that bundled payments were having dangerous consequences for dialysis payments, because clinics were prescribing fewer prescription drugs, and blood transfusions spiked. Scott Gottlieb of the American Enterprise Institute has argued that this incentive for doctors to prescribe fewer drugs could be particularly harmful for cancer patients who rely on expensive medications.