Recognition of the following four symptoms was assessed
Recognition of the following four symptoms was assessed: per-sistent cough or hoarseness, unexplained bleeding, unexplained lump or swelling, and change in the appearance of a mole. For each symptom, the interviewer asked: Do you think [insert symp-tom] could be a sign of cancer? Responses were categorized dichot-omously (no/don’t know or yes), and refusals were coded as missing.
Data were collected on age (categorized by Medicare [federal health insurance program] eligibility: <65 years or ≥65 years), sex, partner status (single or married/cohabitating), race (white, black, or other), education (no bachelor’s degree or bachelor’s degree or above), cancer experience (none, friend/family member only, self), and self-reported health (very good/good/fair or poor/ very poor). Following the original ABC survey,11 ease of health-care access was also assessed using the following question: How easy, or difficult, is it for you to get to see a doctor if you have a symptom that you think might be serious? Response options included very good, good, fair, poor, or very poor.
Univariate and bivariate analyses were used to describe and explore predictors of anticipated delay and recognition of cancer symptoms. Multivariate logistic regression models were used to test associations between anticipated delay and recognition of the related cancer symptom, adjusting for all covariates. Analysis was modeled after the approach of Quaife et al.,18 who examined care seeking in a sample of UK adults, to compare the context of the U.S. with the UK. All analyses were weighted for nonresponse and survey design and performed using Stata, version 13.1. All statisti-cal tests were two-sided and a p-value of <0.05 was considered statistically significant. Data were analyzed in 2017.
Recognition of early warning signs of cancer was high across all symptoms but greatest for unexplained lump (95.5%), followed by changes in Wortmannin appearance (93.4%), unexplained bleeding (89.6%), and persistent cough (76.9%). Across all four symptoms of cancer, in bivariate models, higher education was significantly associated with symptom recognition (Table 1). African Americans were significantly less likely than other racial groups, in bivariate models, to identify unexplained lump (p<0.01) or change in mole appearance (p<0.01) as potential symptoms of cancer. Women were more likely to recognize unexplained bleeding (p<0.01) or change in mole appearance (p<0.05) than men in bivariate models. No significant relationships were found between self-reported health or healthcare access and recognition of any cancer symptom. Table 1 provides full details of the bivariate correlates of cancer symptom recognition.
The prevalence of anticipated delay (>2 weeks) varied across symptoms and was the highest for persistent cough (41.2%), followed by mole changes (33.1%), breast changes (14.7%), and rectal bleeding (9.1%). For all symptoms except rectal bleeding, there were significant associations in bivariate models between higher educa-tion and greater anticipated delay (Table 2). Adults who reported difficulty accessing a doctor had significantly higher odds of anticipated delay in seeking care for rectal bleeding (p<0.05) and breast changes (p<0.05). These adults also more commonly reported delays in seeking care for mole changes, but this association was not statis-tically significant. Older adults reported fewer antici-pated delays for all symptoms except breast changes (Table 2). There were no significant associations between self-reported health or cancer experience for any symp-toms. Table 2 provides full details of bivariate correlates of anticipated delays.
Figure 1 shows that for each cancer symptom, antici-pating delay (>2 weeks) for care seeking was more com-mon among those adults who did not recognize the symptom as a potential warning sign of cancer. These dif-ferences were statistically significant for rectal bleeding (p<0.05) and mole changes (p<0.01). In multivariate analyses, the likelihood of anticipated delay in care seek-ing was significantly greater among those individuals who did not recognize the symptom as a potential sign of can-cer, after adjusting for potential confounders, for rectal bleeding (AOR=2.65, 95% CI=1.31, 5.36) and mole changes only (AOR=3.30, 95% CI=1.48, 7.33). Across all symptoms except breast changes, African Americans were significantly less likely to delay than other races (Table 3). Adults with a college degree or higher were sig-nificantly more likely to delay care seeking for persistent cough (AOR=1.41, 95% CI=1.04, 1.92) and mole changes (AOR=1.46, 95% CI=1.07, 1.99). For rectal bleeding and breast changes only, those who reported difficulty access-ing a doctor were significantly more likely to delay care seeking than those who reported ease in accessing a doc-tor (rectal bleeding: AOR=2.44, 95% CI=1.09, 5.47; mole changes: AOR=2.97, 95% CI=1.41, 6.25).