ACC: With model, fewer fears of bumbling through bundling

SAN FRANCISCO—Hospitals may be able to stay one step ahead as healthcare marches toward bundled payments if a piloted model unveiled March 9 proves robust. The pilot study was presented in a poster at the American College of Cardiology (ACC) scientific session.

Matthew Bunte, MD, of the Cleveland Clinic, and his colleagues set out to model costs using Medicare administrative claims data for PCI patients at the Cleveland Clinic for an episode of care, which they defined as 30 days before and 90 days after PCI. “In order to do that you need to bundle the PCI care, and exclude those things that aren’t related and include those things that are related to those 120 days of care,” Bunte explained.

The researchers collaborated with a Cleveland Clinic innovations spinoff, Explorys, which has expertise in linking administrative claims data with de-identified patient data to design models that predict costs. Researchers obtained data on 300 Medicare patients undergoing PCI at the Cleveland Clinic between 2008 and 2009 and clinical data from the clinic’s EHR, matched to National Cardiovascular Data Registry data and administrative claims.

The research team standardized payment to include private as well as public payers and informed the model further by adding hospital costs. They considered preoperative clinical variables from 200 patients in three different models to estimate an episode cost, giving weight to the model that had the best prediction and validated the models using data from 100 more patients.

“Basically, we are trying to put people in bins,” Bunte said, which were low, medium and high price classes. “The hard part is trying to figure out which bin they will fall in.”

Bunte pointed out that costs are concentrated in a small number of patients. With the pilot, they could estimate full episode price and then break that full episode price into acute care costs (in-patient cost) and post-acute care cost (90-day post-procedure costs).  

Bunte proposed that the model, if further developed, could be applied to any procedure or diagnosis. They are negotiating with private insurers to add more patient data to the model.

“I see this being used by hospitals to contract with insurance companies in bundled care,” he said. Hospitals may be able to use the model to negotiate prices that offer favorable profit margins.