Racial Differences Pertaining to a Belief about Lung Cancer Surgery: Results of a Multicenter Survey

Context Some people believe that exposing the lungs to air during surgery for lung cancer causes tumor spread. Contribution Of 626 outpatients seen in 5 pulmonary and lung cancer clinics in different parts of the United States, 38% believed that air exposure causes tumor spread. African Americans espoused the belief more often than did white people. Nineteen percent of African Americans said that it was a reason to decline surgery if they had a tumor. Cautions These findings may help explain some apparent disparities in lung cancer care, but they don't elucidate underlying reasons for differences in beliefs and don't directly link beliefs to actions. The Editors At the Philadelphia Veterans Affairs (VA) Medical Center in Philadelphia, Pennsylvania, we frequently encounter patients undergoing evaluation for lung lesions who have beliefs that might constrain acceptance of surgical options, including the notion that lung cancer spreads if exposed to air during surgery (referred to in this paper as the study belief). Several African-American veterans declined surgical evaluation on this basis and confided that this belief was common in the African-American community. We wondered whether this conviction could be disproportionately common among African-American patients, thereby more frequently undermining the best chance for cure in this population. To assess the origin, distribution, and importance of this belief, we surveyed outpatients attending 5 pulmonary or thoracic surgery practices located in 3 U.S. cities. Methods Patient Recruitment Patients were recruited from the outpatient clinics and medical practices of the Philadelphia and Los Angeles VA Medical Centers, University of Pennsylvania (Philadelphia), and Medical University of South Carolina (Charleston). When checking in at the clinic site, patients were given a 1-page survey and were asked to complete it while awaiting their physician encounter. Participation was voluntary, and health care entitlements were not altered or abridged if a patient declined to participate. The Committee on Studies Involving Human Beings at the University of Pennsylvania approved the study, and the institutional review boards of the study sites waived informed consent. The study sites were chosen to include adequate numbers of African-American and white patients from the Northeast, South, and West Coast regions of the United States, including 2 VA sites; 3 non-VA sites; 3 clinics serving pulmonary outpatients, many at increased risk for lung cancer; and 2 clinics serving patients with lung tumors or cancer. The VA and university clinics comprised general pulmonary clinics with the following estimated distributions of patient conditions: chronic obstructive pulmonary disease, 40%; asthma, 20%; miscellaneous, 20%; sarcoidosis and interstitial disease, 15%; and lung cancer, less than 5%. The University of Pennsylvania surgical practice and Medical University of South Carolina lung cancer clinic comprised 90% of patients with lung cancer who were considered for or had undergone surgical biopsy or resection and 10% miscellaneous other patients. Data Collection A multicenter, cross-sectional survey of consecutive clinic attendees was conducted between 1 July 1999 and 31 December 2000 by using anonymous, self-administered questionnaires distributed by clinic ancillary personnel to each patient at the time of a clinic visit. In 1 clinic, the surveys were distributed directly by the physicians. Clinic personnel at the 5 participating sites included 15 physicians (12 white people, 1 Pacific Islander, 1 Asian American, and 1 African American) and 28 nurses, respiratory therapists, and clerks (14 African Americans, 13 white people, and 1 Pacific Islander). Surveys were completed while the patient was waiting to be seen by a physician. The survey instrument was tested and refined in a pilot study (1). The questionnaire was reviewed by an epidemiologist and a psychometrician and reflects a composite FleschKincaid grade level of 5.9. It contains 6 key questions about the study belief near the beginning of the survey, 1 of which was intentionally phrased negatively to assess potential yea sayer biases among respondents. Statistical Analysis The prevalence of individual beliefs was calculated with 95% CIs. In the univariable analysis, we assessed the association of African-American race with belief. Racial comparisons were made only between African Americans and white people, since the sum of Asian Americans, Hispanics, and others was less than 5% of the cohort. Responses did not include ethnic subgroups within the white category. In addition, we assessed the association of other baseline variables with individual beliefs, including age, sex, education level (less than eighth grade, eighth to twelfth grade, college, or more), household income (<$20 000 per year, $20 000 to $50 000 per year, or >$50 000 per year), religious affiliation, VA or non-VA site, surgical clinic versus general pulmonary clinic, urban versus rural residence, region of the country, marital status, and individual survey site. To test the strength and independence of the association of each risk factor with individual beliefs, we performed a multivariable explanatory analysis. In this analysis, each demographic variable (for example, African-American race) was evaluated after adjustment for other potential confounding variables by fitting all covariables in a logistic regression model. A P value less than 0.05 after adjustment for all confounding variables was considered significant. All statistical analyses were performed by using Stata software, version 7.0 (Stata Corp., College Station, Texas). Role of the Funding Source The funding source had no role in the collection, analysis, or interpretation of the data or in the decision to submit the manuscript for publication. Results We received 626 completed surveys of 652 that were distributed; 96% of respondents answered at least three quarters of the questions. Respondents included 251 VA patients (153 from Philadelphia and 98 from Los Angeles), 151 pulmonary outpatients and 127 thoracic surgery outpatients at the University of Pennsylvania, and 97 chest cancer clinic patients from the Medical University of South Carolina. Four percent of clinic attendees declined to participate. The patients were predominantly middle-aged or elderly (mean age [SD], 60.4 14 years), male, and white and lived in an urban setting, which reflected the geographic distribution of the clinics surveyed. Table 1 lists patient demographic characteristics. Table 1. Patient Demographic Characteristics Regarding the 6 key questions, 45% of all patients had heard of the study belief and 37% believed it to be true. Although 49% thought most people they know believed it, only 12% knew of a case, 10% identified it as a reason to oppose lung surgery, and 9% would not believe their physician's advice that the belief was false. For each key question, highly significant differences were found in the relative proportions of responses from white people versus African Americans (Figure). For example, 29% of white people versus 61% of African Americans thought the study belief was true (P < 0.001), 5% of white people versus 19% of African Americans would oppose surgery on the basis of the belief (P = 0.001), and 5% of white people versus 14% of African Americans (P = 0.001) would not believe their doctor on this issue. Figure. Responses to survey questions by race. The percentage of African-American respondents who thought the study belief was true (61% [86 of 142 patients]) was actually marginally higher than the percentage who had heard of it (57% [89 of 156 patients]). This apparent paradox was related to a smaller number of African-American patients responding to the truth question. No patient reported belief in a statement that he or she had never heard of. Univariable analyses showed that African-American versus white race was a significant factor for each of the 6 key questions. Education (college education versus less than eighth grade), household income greater than $50 000 per year versus $20 000 to $50 000 per year or less than $20 000 per year, women versus men, and VA site versus non-VA site were significant for 3 or 4 of the questions. Urban versus rural residence, region of the United States, and pulmonary versus surgical clinic were significant for 0 or 1 of the questions. Multivariable analysis (Table 2) showed that most of the statistically significant differences by univariable analyses were no longer significant when confounding factors were considered. However, racial differences remained significant (P < 0.006) for all questions despite adjustment for all covariables. Pulmonary clinic versus surgical clinic was the only factor to emerge as significant in the multivariable analysis; pulmonary clinic patients were significantly more likely to believe the study belief was true (odds ratio, 2.34 [95% CI, 1.24 to 4.44]). No significant first-order interactions were detected between race and other clinical variables about the study belief. Table 2. Multivariate Analyses of Demographic Variables for 3 Key Survey Questions about the Study Belief Only 21% of patients responded to the survey question asking where and when the study belief was first encountered; responses were generally vague (don't remember or from the gossip mill) and did not differ appreciably between African-American and non-African-American respondents (data not shown). Discussion Using a questionnaire survey of patients attending 5 U.S. pulmonary or thoracic surgery outpatient clinics, we found widespread belief that exposure of lung cancer to air at the time of surgery promotes cancer spread. At all study sites, this belief was considerably more common among African Americans than white people. Nineteen percent of African-American respondents versus 5% of white respondents expressed opposition to lung cancer surgery on the bas