WASHINGTON -- Refusal to undergo cancer surgery continues to be a problem in certain sociodemographic groups, a researcher said here.
In particular, patients were more likely to refuse surgery for breast, colon, lung, or prostate cancer if they were older, African American, and unmarried, Joseph Rapp, MPH, of the CUNY School of Public Health and Health Policy, in New York City, said at the AcademyHealth Annual Research Meeting.
"Despite progress in the field, cancer patients who are non-white, lacking health insurance, and unmarried are less likely to receive surgical intervention in cancer," Rapp said. "Recent research has identified refusal of recommended surgery as a potential reason for some of these disparities persisting, but it's relatively understudied."
In particular, "no study specifically exists on early-stage cancers, and this is important, because surgical resection in early stages ... typically confers the best survival benefits and sometimes is curative," he said.
So Rapp and his colleagues decided to assess clinical and demographic predictors of surgery refusal in certain early-stage cancers, as well as survival among those who underwent the surgery versus those who refused it. They used data from the Surveillance, Epidemiology, and End Results (SEER) database, which collects cancer incidence data from population-based cancer registries covering approximately 34.6% of the U.S. population.
They looked at breast, lung, colon, and prostate cancer cases diagnosed from 2007 through 2014, and focused on stage 1 and 2 diagnoses. Exclusion criteria included no recommendation for surgery and cases in which it was not the patient's first cancer diagnosis. After further stratifying the analysis to include only certain types of each kind of cancer, the investigators ended up with a total of 485,000 cases, more than half of which were breast cancer patients.
Possible sociodemographic predictors of surgery refusal included:
- Age at diagnosis
- Sex
- Race/ethnicity
- Marital status
- Insurance status
- Proximity to a metropolitan area
Possible clinical predictors included stage at diagnosis and -- for breast cancer only -- the cancer histology (tubular vs ductal).
In terms of demographics, the average age for each type of cancer patient ranged from 60.2 for breast cancer patients to 68.2 for colon cancer patients. The vast majority of patients in all four groups -- 97% to 99% -- were insured. Most of the patients were married, with a range of 54% for colon cancer patients to 75% for prostate cancer patients.
The majority of patients in each group -- ranging from 70% of colon cancer patients to 79% of lung cancer patients -- were white. Non-Hispanic black was the next most common ethnicity (9% to 13% in each group), followed by Hispanic, Rapp said.
Overall, surgery refusal was quite uncommon, but occurred most often among prostate cancer patients (3.3%), followed by lung cancer patients (2.5%), and then breast and colon cancer patients (0.35% in each group). Refusal predictors for each cancer type were similar but slightly different in their order -- for example, the biggest predictor for refusing breast cancer surgery was being uninsured (adjusted odds ratio 2.74), while for refusing colon cancer surgery, being older was the biggest predictor (adjusted OR 5.26).
In general, patients who refused surgery had worse survival outcomes than those who underwent surgery, Rapp said. Using a propensity-matched analysis, the researchers found that colon cancer patients who refused surgery had the worst outcomes -- they were nearly eight times more likely to die during the study period (adjusted hazard ratio 7.92) than their counterparts who underwent surgery. They were followed by breast cancer patients with an adjusted HR of 6.50, lung cancer patients (2.67), and prostate cancer patients (1.81).
As to why patients refuse surgical intervention, "some studies done about 10 years ago indicate there are possible misconceptions about surgical intervention, in particular within the African-American population," Rapp told ľֱ. "That study doesn't parse out why that might be either. The only speculation I can make is that these are populations that aren't receiving adequate healthcare education."
The study's limitations included the fact that reasons for refusal were unknown, as were what type of insurance patients had and their comorbidities. Strengths included the large sample size, the focus on early-stage cancers, and the ability to use propensity score match analysis to control for selection bias.
"Future studies should probably look to investigate why patients are refusing their surgeries, and include insurance and confounder data, and possibly investigate procedures being refused," he said. "Other studies could possibly investigate whether patients are being recommended surgery at the same rate" across various groups.