- New study examine variability in breast cancer treatment due to patient factors, geographic, and unidentified factors.
- The study analyzed treatment of more than 31,000 patients.
- Findings highlight the many cases where unidentified factors are affecting treatment.
For as long as researchers have known about disparities in cancer treatment, they’ve sought to explain them. Do some patients receive different care than others because of race, age, wealth, or gender? Or does the region where they receive treatment matter most? Only by answering questions like these can researchers know where to focus their efforts to improve care.
A recent study by Dana-Farber researchers suggests investigators need to re-evaluate the lens through which they’ve traditionally viewed treatment differences — and consider types of variations beyond those previously examined.
In breast cancer, research has repeatedly shown that age, race, ethnicity, and socioeconomic status play a role in the treatment patients receive. There’s also evidence that breast cancer care differs by region, with some states and counties providing more extensive treatment than others.
Despite these findings — and numerous efforts to redress these inequities – quality improvement programs have struggled to make a sizable impact on the patient population as a whole. This is partly because of questions left open by previous research, says Dana-Farber’s Michael Hassett, MD, MPH, the first author of the study.
“Previous research hasn’t really identified which factors are most important — which are responsible for the variation in care that we’re seeing?” he remarks. “What data should we be using in deciding where to direct our efforts at quality improvement?”
The new study sought to determine how much of the variability in breast cancer treatment is due to patient factors such as ethnicity or household wealth, to geographic factors such as region of care delivery, or to other, not-yet-identified factors.
Analyzing patient and geographic factors
Using information in the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, investigators analyzed the treatment of 31,571 patients with stage I to III breast cancer between 2007 and 2013. They focused on five measures of care delivery: whether patients were initially treated with chemotherapy, radiation therapy, or endocrine (hormone-blocking) therapy; whether they continued to receive endocrine therapy three to five years after diagnosis; and whether they were diagnosed when their cancer was stage I.
The analysis showed that use of chemotherapy, radiation therapy, and endocrine therapy, and the stage at which breast cancer was diagnosed, varied widely from one region to another — a largely expected finding. More surprising, it found that patient factors accounted for only 1-4% of variations in treatment. Regional factors were considerably more significant, accounting for 24-48% of variations (depending on which of the five measures of care delivery was being assessed). For radiation and endocrine therapy in particular, regional variations were particularly pronounced.
Ultimately, however, both personal and regional factors were dwarfed by the 35-54% of variations that were unexplained.
“For this cohort of patients with breast cancer, our findings show that current categories cannot account for a large portion of variations in breast cancer treatment,” says Hassett, who collaborated with Dana-Farber’s Rinaa Punglia, MD, MPH, and Hajime Uno, PhD, on the study. “This suggests that we’ve been focused on variables like race, age, and geographic location that, while statistically significant and sociologically meaningful, may not actually help us make major improvements in care delivery, broadly speaking.”
Variables that researchers have overlooked may include patient characteristics other than age, race, gender, and wealth, Hassett continues. “Maybe it’s not race, per se, that is significant but some aspect of social support: patients who are married and have strong social support systems tend to do better than those without such assistance. None of our traditional measures are good at picking up that aspect of patients’ lives.” Another possibility is that genetics affect treatment and outcomes in ways that have yet to be identified.
In terms of geographic influences, “perhaps we’re not defining regions in useful terms,” Hassett notes. “Maybe the regions we’re focusing on are too big. Maybe the most important consideration is not the general area you live in, but the specific hospital or clinic where you’re treated.
“As researchers, perhaps instead of looking at the ‘usual suspects,’ we should be creating data sets and tools to capture information that has a bigger impact on treatment variations in this country.”
Hassett and his colleagues are working to do just that. In a follow-up study, they’re unpacking the concept of “region” to determine whether any of 14 specific aspects of a region play a prominent role in treatment disparities. They hope to publish their findings in the next year or two.