In 2012, Dr. Gabrielle Rocque, a medical oncologist who specializes in the care of women with breast cancer, was medical director for a Centers for Medicare and Medicaid Innovation Award project that gave her access to a sizable quantity of claims data.
That experience led her to think about the potential within her own workplace, the University of Alabama at Birmingham (UAB) Health System Cancer Community Network (CCN), to discover where clinicians might improve on the quality of care delivery.
“I saw an opportunity to look within our network to examine concordance rates and identify areas for quality improvement” with the goal of raising awareness about value-based care, said Rocque, Assistant Professor of Medicine at UAB in the Divisions of Hematology and Oncology, and Gerontology, Geriatrics and Palliative Care.
Rocque and her team of researchers conducted a retrospective analysis of Medicare claims data (2012-2015) to examine concordance with nine Choosing Wisely recommendations; six from the American Society of Clinical Oncology and three from the American Society for Radiology Oncology. The recommendations involved imaging, chemotherapy, radiation and supportive medications.
The claims data—including associated costs—came from the 12 cancer centers (Alabama, Georgia, Florida, Mississippi and Tennessee) that comprise UAB’s cancer treatment network. CCN includes academic medical centers and community cancer centers in urban and rural areas.
According to the study, overall concordance ranged from only 39 percent to 94 percent, based on the recommendation. There was significant variability from center to center for all recommendations, with as much as 89 percent difference in concordance rates between cancer centers.
Concordance was lowest for women with early-stage invasive breast cancer receiving a short course of radiation after lumpectomy, and highest for men with early prostate cancer not receiving PET, CT or bone scans within three months before or after diagnosis. The UAB CCN database included 37,686 unique Medicare patients; the study found patient factors such as race, age and comorbidities had modest influence on concordance variability.
“Choosing Wisely is an incredibly powerful tool for setting benchmarks and being a reliable source of information for things we shouldn’t do in clinical practice,” said Rocque, adding her research clearly indicates that differences in concordance were driven, at least in part, by site of care.
She said greater education of clinicians, data-driven learning, and constant quality improvement effort and reinforcement were key ways to improve care. As Director of CCN at that time, she met regularly with clinicians in the region to foster discussion about their clinical practice based on her research and how they can improve care.
Rocque added that there are lots of reason why clinicians shouldn’t follow the recommendations “100 percent of the time,” noting that one drawback of her 2016 study was the reliance on claims-data analysis, which does not allow for discriminating between when lack of concordance is appropriate clinical decision-making or patient preference.
That limitation, and others, led Rocque and colleagues to undertake a more nuanced study, examining adherence to an ASCO guideline that recommends no staging imaging in early stage breast cancer patients with low risk of metastasis and identifying factors contributing to unnecessary staging imaging. Research data was abstracted from chart review of 872 patients, diagnosed from 2013-2015, at a single-site academic medical center. She has led a team to conduct a similar analysis of concordance rates for Choosing Wisely measures within ASCO’s Quality Oncology Practice Initiative.
Rocque views her research as setting an example for how claims data can be used to inform efforts to deliver value-based cancer care. “We are moving quickly into the space where we are partnering with payers, and they are providing use with data to measure, and the measurements are straight forward, good-value targets,” she said. “The information is out there to support good oncology care models.”