|Home | About | Journals | Submit | Contact Us | Français|
We examined the knowledge and prevalence of mouth and throat cancer examinations in a sample drawn from rural populations in north Florida.
Telephone interviews were conducted across rural census tracts throughout north Florida in 2009 and 2010, in a survey that had been adapted for cultural appropriateness using cognitive interviews. The sample consisted of 2526 respondents (1132 men and 1394 women; 1797 Whites and 729 African Americans).
Awareness of mouth and throat cancer examination (46%) and lifetime receipt (46%) were higher than reported in statewide studies performed over the past 15 years. Only 19% of the respondents were aware of their examination, whereas an additional 27% reported having the examination when a description was provided, suggesting a lack of communication between many caregivers and rural patients. Surprisingly, anticipated racial/ethnic differences were diminished when adjustments were made for health literacy and several measures of socioeconomic status.
These findings support the notion that health disparities are multifactorial and include characteristics such as low health literacy, lack of access to care, and poor communication between patient and provider.
Mouth or throat cancers make up approximately 3% of all diagnosed malignancies in the United States annually, with risk increasing with age.1 Alarmingly, a median survival time of 37 months was recently reported, regardless of the tumor stage at diagnosis.2 Considerable disparities exist, with greater mortality for African Americans compared with Whites, particularly among men.1,2 Multiple factors are likely to account for these disparities, including differential stages of diagnosis.3 Diagnosing cancers at a premalignant or early stage is crucial to reducing morbidity because survival varies dramatically by the stage of disease.4,5
The principal screening test for mouth and throat cancer in asymptomatic persons is inspection and palpation of the oral cavity. The American Dental Association suggests an annual mouth and throat cancer examination for adults aged 40 years or older.6 Although a mouth and throat cancer examination is fast and usually painless, many people report never having heard of it, and of those who have, many have not been examined.7,8 National data from the 1992 National Health Interview Study (NHIS) Cancer Control Supplement indicated that only 15% of the US population aged 40 years and older had ever been examined for mouth and throat cancer.8,9 Statewide surveys in Maryland, North Carolina, New York, and Florida have also assessed self-reported mouth and throat cancer examinations, with rates varying from 28% to 38% for lifetime prevalence.10–13
With the exception of levels of educational attainment10–13 or having a regular source of dental or preventive medical care,11,12 psychosocial variables have not been tested for associations with receipt of an examination. The literature for prostate cancer, breast cancer, and cervical cancer provides some support for risk factors such as depression for women14–16 or social support for men17,18 and women.19,20 An often overlooked, but potentially important, risk factor for screening rates is health literacy.21 Although some studies have shown an association between health literacy and knowledge of colorectal cancer screening issues22–25 and self-efficacy to be screened,26 none have shown an association between low literacy and screening rates.
Persons living in rural areas have the additional burden of limited access to health care and, in particular, of being less likely to receive preventive cancer services than urban residents.27–29 These differences may be compounded for racial/ethnic minorities, who are more likely to have less education and live in counties with higher poverty rates and lower insurance rates, all of which are associated with reduced access to health care.30 Existing studies have not reported data on mouth and throat cancer examinations for rural areas. Our primary aim was to test the hypothesis that, among rural north Floridians, African Americans would be less likely to report (1) having heard of a mouth and throat cancer examination, (2) having had an examination in their lifetime, and (3) having had an examination in the past year, and that in addition (4) psychosocial variables of financial status, education, depression, social support, and health literacy would be associated with these mouth and throat cancer examination outcomes.
We adapted the mouth and throat cancer survey questions from items used in the NHIS (supplemented by a literature review). We conducted cognitive interviews with focus groups of rural African Americans to assess the perception, usefulness, and interpretation of each item. Our intent was to adapt the survey to better fit the given culture and demographics of the study areas. We field tested the final instrument using a test-retest model with 93 participants. We observed acceptable reliability for the “having had a mouth or throat cancer exam” question (r=0.79).
We sampled 36 rural census tracts in 2 geographic areas: north central Florida (Union, Alachua, and Bradford counties), and northwestern Florida (Jefferson, Leon, Gadsden counties). We defined sampling strata by census block groups according to percentage of African Americans (≥30%, 20%–29%, <20%). On the basis of these percentages, we over-sampled block groups with higher concentrations of African Americans. We included in the sample only telephones with landlines to maintain stability for recontact, obtain clearer communication signals, and sample the older population (who were more likely to only use landlines).31 To maximize participation of older men and hopefully balance representation by gender, we implemented a within-household respondent selection procedure. To ensure random sampling but maintain a high proportion of older men, we asked for the “oldest male with immediate substitution,” which allowed the person who answered to be interviewed if the oldest male household member was not available.
The University of Florida Bureau of Economic and Business Research Survey Center performed the survey using professional interviewers. Each phone number was dialed 10 times, including at least 1 call during a weekday, weeknight, and weekend. A total of 16000 phone numbers were dialed, resulting in 2605 interviews conducted at 2527 households, with an average interview length of 22.7 minutes (SD=5.3). A $15 Wal-Mart gift card was offered for completion of each survey. Despite oversampling in areas with higher percentages of African Americans, the number of interviews with African American men was below ideal after the first month of baseline survey fieldwork. The race question was therefore used as a screening criterion, with interviews being completed only with households in which the selected respondents were African American. Consequently, more than 3000 calls to non–African American households did not result in an interview. The cooperation rate for eligibility screening was 44.7%, with 26.4% eligible and completing the interviews. The survey began in late November 2009 and was completed in March 2010.
Three dichotomized primary response variables were of interest: (1) whether a participant had ever heard of a mouth or throat cancer examination, (2) whether a participant had ever had an examination, and (3) whether the examination was in the past year. Two secondary response variables consisted of whether respondents recognized they had received a mouth or throat cancer examination after one was described and whether the examination had occurred in a medical or a dental setting.
Predictor variables were psychosocial factors of interest, which included education, financial status, health literacy score (range=0–3, with 3 indicating highest health literacy skills),32 having a doctor of record or dentist of record33 (yes or no for both), Medical Outcomes Social Support score (rescaled to 0–4, with 4 indicating most social support),34 and Center for Epidemiologic Studies Depression Scale (CES-D) short form35 (rescaled to 0–3, with 3 indicating most depression).
We created a continuous financial status scale (range = 0–2, with 2 indicating highest) based on 2 questions. First, participants were asked to describe their financial status as 1 (“I really can’t make ends meet”), 2 (“I manage to get by”), 3 (“I have enough to manage plus some extra”), or 4 (“Money is not a problem; I can buy about whatever I want”). In the second question, the respondents were asked to describe how comfortably they would be able to pay an unexpected $500 medical bill.
We classified participants’ education level into 6 categories: 1 (eighth grade or less), 2 (some high school, but did not graduate), 3 (high school graduate or general equivalency diploma [GED]), 4 (some college or 2-year degree), 5 (4-year college graduate), and 6 (more than 4-year college degree).
We evaluated racial differences in demographic and psychosocial variables using survey-sample-weighted t tests (for continuous variables) and χ2 tests (for categorical variables). We used survey-sample-weighted multiple logistic regression models to assess the association of the 3 response variables (see previous subsection) and the psychosocial predictor variables, adjusting for age, gender, and race. We call these 3 analysis-tested hypotheses models 1 through 3, respectively. We performed model selection following the strategies described in Muller and Fetterman.36 We constructed a maximum model with all of the predictors, their interactions with gender and race, and predictor-gender-race 3-way interactions (Table 1). After minimizing removable collinearity through effect coding, centering, and scaling,36 we evaluated collinearity and found it acceptable for the maximum model (and therefore not a concern for any smaller model). Model selection was preceded by a backward step-down selection starting from the interaction terms. We evaluated the Bonferroni method to control for multiple testing, with the exclusion P value set at .05 divided by 3, since there were 3 response variables of interest. We used the survey procedure PROC SURVEYLOGISTIC of SAS version 9.2 (SAS Institute, Cary, NC) for the analysis. We used the “TOTAL=” option in PROC SURVEYLO-GISTIC to identify stratum population totals in our survey data set. We specified stratification and weights using the STRATA and WEIGHT statement.
The sample consisted of 2526 respondents (1132 men and 1394 women; 1681 Whites and 712 African Americans). The mean age was 56.1 years (SD =14.7, range=25–94). The distribution across education was as follows: 2% (n=55) had completed eighth grade or less, 7% (n=175) had attended some high school, 27% (n=672) had completed high school or received a GED, 30% (n=756) had attended some college, 16% (n=406) were college graduates, 18% (n=449) had attended postgraduate school, and fewer than 1% (n=13) declined to answer. Race and gender by community are presented in Table 2. Descriptive statistics for the sample characteristics by race are presented in Table 3.
Forty-six percent of the sample had heard of an examination for mouth and throat cancer and 19% indicated, without any prompting, that they had received an examination. An additional 27% answered yes to the question, “Has a doctor or dentist ever examined your mouth, by pulling on your tongue, with gauze wrapped around it, and feeling under your tongue and inside your cheeks?” Including this subgroup, a total of 46% of the sample (n = 1168) was scored as having had at least 1 lifetime examination for mouth and throat cancer. Of this group, 56% reported an examination in the past year (26% of the total sample) and 70% reported that the last examination had occurred at a dental office. Prevalence of mouth and throat cancer examination by racial/ethnic group is presented in Table 3.
Results from the logistic regression analysis are summarized in Table 4. Because the race group “others” comprised less than 1% of all participants, we excluded it from the logistic regression models. In all 3 primary regression models, education and regular dentist were significant predictors. Compared with those who did not have a regular dentist, participants who had a regular dentist were more likely to have ever heard of the examination (odds ratios [OR]=1.53; 95% confidence interval [CI] =1.24, 1.88), to have ever had an examination (OR=1.87; 95% CI=1.40, 2.40), and to have had an examination in the past year (OR=4.81; 95% CI=3.34, 6.92). Participants with a higher education level were more likely than those with a lower education level to have ever had an examination (OR=1.21; 95% CI=1.07, 1.35) and to have had an examination in the past year (OR=1.21; 95% CI=1.08, 1.36). However, there was a gender difference in the predictability of education in model 1. Among female participants, those with a higher education level were more likely to have ever heard of the examination (OR= 1.21; 95% CI=1.09, 1.35), whereas among male participants, the odds of hearing about the examination were the same across all education levels. In model 1, higher financial status was also a significant predictor of whether a participant had ever heard of the examination (OR=1.30; 95% CI=1.10, 1.55). Moreover, among male participants, those with high health literacy skills were more likely to have ever heard of the examination (OR=1.45; 95% CI=1.19, 1.77). In model 3, among White participants, those with a regular physician were more likely to have had an examination in the past year (OR=2.31; 95% CI=1.18, 4.51).
As a secondary analysis, we developed logistic regression models for 2 additional response variables: whether respondents recognized they had received a mouth or throat cancer examination after one was described and whether the examination had occurred in a medical or a dental setting. In model 1, Whites (OR=1.72; 95% CI=1.14, 2.59) and people who had a regular dentist (OR=1.77; 95% CI=1.18, 2.67) were more likely to recognize they had received an examination after it was described to them (data not shown). In model 2, people with higher socioeconomic status (SES) (OR= 1.56; 95% CI=1.16, 2.08), people who had a regular dentist (OR=3.38; 95% CI=2.34, 4.89), and White women (OR=2.27; 95% CI=1.34, 3.84) were more likely to have had their examination in a dentist’s office (data not shown).
This study was the first to examine the prevalence of mouth and throat cancer examinations in a sample drawn exclusively from rural populations in the United States. Awareness and receipt of an examination was higher than reported in earlier statewide studies, possibly because of a growth in overall awareness among the general public. However, anticipated racial/ethnic differences were diminished when adjustments were made for several psychosocial variables along with appropriate adjustments for type 1 error. We found that health literacy and several measures of SES, but not depression or social support, were related to the mouth and throat cancer examination outcomes. Another finding was that a substantial number of respondents were unaware they may have received the examination until a description was provided.
These findings, combined with those of several earlier studies,8,10–13 suggest a pattern of increased awareness of and receipt of mouth and throat cancer examinations. The rural north Florida data indicated that 46% of all respondents and 48% of those aged 40 and older had received at least 1 lifetime examination. Initial information collected as part of the 1992 NHIS indicated that only 15% of adults aged 40 years and older reported ever having had a mouth and throat cancer examination, and 7% of the overall respondents indicated the examination was in the past year.8 This original NHIS question includes a description of an examination:
Have you ever had a test for oral cancer in which the doctor or dentist pulls on your tongue, sometimes with gauze wrapped around it, and feels under the tongue and inside the cheeks?
A subsequent study in Maryland by Horowitz et al. (data collected in 1996) reported that 21% of the participants had heard of an examination for oral or mouth cancer and 28% indicated they had had an examination once it was explained.10 Two additional statewide studies from North Carolina11 and Florida12 (both performed in 2002) reported similar percentages of respondents receiving a mouth and throat cancer examination. Data from the 2003 administration of the New York State Behavioral Risk Factor Surveillance System survey indicated that 35% of participants aged 18 years and older and 40% of respondents aged 40 years and older reported having had an examination in their lifetime.13 This trend toward an increase of mouth and throat cancer examinations between 1992 and 2010 is encouraging. Prostate cancer examinations measuring prostate-specific antigen (PSA) levels have also shown a pattern of increase, although over a shorter period,37 which may show a greater general awareness of the importance of cancer screening.
Within the state of Florida, there have been several interventions aimed at increasing awareness of oral cancer screenings that may have contributed to this increased awareness. First, a campaign in Miami and Dade County used 9 billboards to encourage the public to go to the dentist for a mouth and throat cancer screening.38 However, no differences were found between Miami–Dade County and a comparison county on the proportions of individuals aware of the billboard campaign or affected by it. Later, a small intervention study in Jacksonville resulted in a significant increase in awareness of the mouth and throat cancer examination (from 30% to 41%), whereas no significant changes were found for a control city.39
The earlier statewide studies used the original NHIS phrasing that included a description of the examination; these studies found small increases in the lifetime prevalence for mouth and throat cancer examinations since the earlier national survey.10–13 We chose not to use the NHIS question; rather, without prompting them, we asked participants if they knew they had been examined. If they answered no, we provided a description of the examination, adding “This is for an examination of the mouth, not just holding the tongue away for dental procedures.” Our strategy allowed us to determine examination rates among both those who had received a description of the examination from caregivers and those who had not. On the basis of our findings, an alarming number of respondents are not told that an examination was performed, and certain patients may lack understanding of the specifics of medical or dental procedures and avoid asking questions of their clinicians.40–43 These characteristics are common among persons with limited heath literacy.24,44,45
Several definitions for health literacy can be found, each of which emphasizes the availability of skills to obtain, understand, and use health information appropriately.46,47 People with poor health literacy have difficulties with written and oral communication, which may limit their understanding of cancer screening and of the symptoms of these cancers, potentially adversely affecting their stage at diagnosis.24,44,45 Specifically how health literacy is related to stage of diagnosis for mouth and throat cancer is not known; however, our data support the view that lower levels of health literacy are associated with lack of awareness of the examination for mouth and throat cancer among men. Men, in general, tend to be unaware of health problems and are more likely to delay seeking care.47 These findings are consistent with the finding of the 2003 Centers for Disease Control and Prevention (CDC) report on adult health literacy that men scored lower than women on health literacy testing.48 We did not find that health literacy was associated with receipt of an examination for either gender.
A large part of the American public has low overall health literacy skills.48,49 Data presented by Horowitz and others8,10–13 indicate that overall knowledge of mouth and throat cancer—a specific form of health literacy—is low. Lower levels of general health literacy are associated with race, education, SES, and age,49–51 but it should be emphasized that low health literacy is distinct from these factors.52–54 Rather, current models suggest that health literacy may serve as a more proximal mediator of health disparities through an effect on actions to promote health, prevent disease, or comply with diagnosis and treatment.26,55
Our use of a 4-item health literacy screening measure was supported in part by a need to keep the telephone interview a comfortable length for respondents. Whereas many measures assess health literacy directly by testing an individual’s ability to read or recognize health-related terms,56,57 this measure asks respondents to self-report their ability to understand health information (i.e., “How often do you have a problem understanding what is told to you about your medical condition?”).58
The relationship between SES and the use of primary and secondary prevention is well documented, and SES is frequently implicated as a contributor to the disparities in health observed among US populations.27–29,59 Consistent with previous studies of mouth and throat cancer examinations,8,10–13 we found that educational attainment was a significant predictor of all 3 of our primary outcomes. For example, among non–high school graduates, only 26% reported having had the examination compared with 57% for college graduates.
Because changing the SES of rural residents may be difficult to achieve, determining and addressing how levels of education and income might influence the receipt of mouth and throat cancer examinations is warranted. For example, if poor and less-educated people have low rates of examinations through lack of knowledge,10,12 then public health education might lead to a greater receipt of examinations. With lower incomes and lack of insurance, people are likely to delay seeking treatment because of other financial priorities, keeping them from being seen by health care providers.60 This may be a greater issue in rural areas, where there is already a shortage of health care providers, particularly dentists.61,62 These individuals also may worry about lacking the resources to address a positive finding and decline the recommendation for an examination.63 Another possibility is that when people of low SES are seen by dentists and physicians, they are less likely than others to have the examination recommended to them.64
Defining SES is challenging; it is generally measured by indicators of human capital, such as income, education, or occupational prestige, that offer advantages to individuals and families.65 SES affects health through its association with health care, the environment, and health behaviors.66 There is much debate in disparities research on how to measure SES and how to make appropriate models for it. Different measures of SES include current income (unstable with high rates of nonresponse), wealth (ability to meet emergencies and absorb economic shock), education, and occupation. Our measures included education and wealth.
In our study, the strong effect associated with having a caregiver of record is consistent with previous studies indicating that a usual source of care has a positive impact on screenings for both mouth and throat cancer12 and other common cancers.67–70 Some studies, including this one, show that controlling for differences in a number of socioeconomic variables, including provider availability, alleviates or eliminates many racial differences.52,71,72 Certainly, evidence suggests that a dentist’s office is the most common site for a mouth and throat cancer examination; in our study, 69% of respondents reported having been screened in a dentist’s office. On the basis of our supplementary analysis, African Americans were more likely to receive the screening in a nondental setting than Whites (42%–26%).
One potential weakness shared by all studies using self-reported examination questions is unknown validity. Some respondents may have received the examination and could not recall it, and it is known that minorities often over-report having received examinations.73 Another limitation is that only households with current residential telephone service were sampled. It should be noted that north Florida has one of the nation’s highest rates of mouth and throat cancers,74 and this may have contributed to the increased awareness we observed.
We found that in rural north Florida, awareness (46%) and receipt (46%) of a mouth and throat cancer examination were higher than reported in other statewide studies. Only 19% of respondents were initially aware of their examination, whereas an additional 27% reported having had the examination when a description of it was provided, suggesting a lack of communication between many caregivers and patients. Surprisingly, anticipated racial/ethnic differences were diminished when adjustments were made for several psychosocial variables and statistical procedures. These findings support the notion that health disparities are multifactorial and include characteristics such as low health literacy, lack of access to care, and poor communication between patient and provider.
This study was funded with support from the National Institute of Dental and Craniofacial Research (grant 1U54DE019261).
Reprints can be ordered at http://www.ajph.org by clicking the “Reprints/Eprints” link.
Data from the study on which this article is based were presented at the American Association for Dental Research 4th Fall Focused Symposium, “Oral Health Disparities Research and the Future Faces of America”; November 3–4, 2011; Washington, DC.
ContributorsJ.L. Riley III was responsible for the conception and design of the study, conceptualized the analyses, and interpreted the findings. V.J. Dodd, K.E. Muller, and H. L. Logan contributed to the design of the study and assisted in the writing. Y. Guo performed the data analysis and assisted with the writing.
Human Participant Protection
All work for this study was approved by the University of Florida institutional review board.
Joseph L. Riley, III, Department of Community Dentistry, University of Florida, Gainesville.
Virginia J. Dodd, Department of Health Education and Behavior, College of Health and Human Performance, University of Florida, Gainesville.
Keith E. Muller, Department of Health Outcomes and Policy, College of Medicine, University of Florida, Gainesville.
Yi Guo, Department of Health Outcomes and Policy, College of Medicine, University of Florida, Gainesville.
Henrietta L. Logan, Department of Community Dentistry, University of Florida, Gainesville.