Demographic, Social, and Behavioral Determinants of Lung Cancer Perceived Risk and Worries in a National Sample of American Adults; Does Lung Cancer Risk Matter?

Background: Perceived risk and worries of developing cancer are important constructs for cancer prevention. Many studies have investigated the relationship between health behaviors and subjective risk perception. However, factors correlated with lung cancer risk perception and worries in individuals more susceptible to lung cancer have rarely been investigated. Objective: To determine demographic, social, and behavioral determinants of cancer perceived risk and worries and to explore heterogeneities in these associations by the level of lung cancer risk in a nationally representative sample of American adults. Methods: For this cross-sectional study, data came from the Health Information National Trends Survey (HINTS) 2017, which included a 2277 representative sample of American adults. Smoking status, cancer perceived risk, cancer worries, age, gender, race, education, income, and insurance status were measured. We ran structural equation models (SEMs) for data analysis. Results: “Ever smoker” status was associated with higher cancer perceived risk (b = 0.25; 95% CI = 0.05–0.44, p = 0.013) and worries (b = 0.34, 95% CI = 0.18–0.50, p < 0.001), suggesting that “ever smokers” experience higher levels of cancer perceived risk and worries regarding cancer, compared to “never smokers”. Other factors that correlate with cancer perceived risk and worries were race, age, income, and insurance status. Blacks demonstrated less cancer perceived risk and worry (b = −0.98, 95% CI = −1.37–0.60, p < 0.001) in both low and high risk lung cancer groups. However, the effects of social determinants (income and insurance status) and age were observed in low but not high risk group. Conclusions: Determinants of cancer perceived risk and worries vary in individuals depending on the level of lung cancer risk. These differences should be considered in clinical practice and policy makings with the goal of improving participation rates in lung cancer screening programs.


Introduction
With an estimated 234,030 new cases and 154,050 cancer-related deaths in 2018, lung cancer is the leading cause of cancer death for both genders in the United States [1]. A low overall five-year relative survival rates of 24% in women and 17% in men are due to the fact that most lung cancer cases

Design and Setting
This cross-sectional study used data from the 2017 Health Information National Trends Survey . Periodically administered by the National Cancer Institute (NCI) since 2003, HINTS is a nationally representative survey. The purpose of HINTS is to provide a national picture of cancer information among American adults [16]. Data of the HINTS-5-Cycle 1 were collected from January 2017 through May 2017.

Ethical Considerations
The HINTS-5 study protocol was approved by the Westat's Institutional Review Board (IRB) (Westat's Federalwide Assurance (FWA) number is FWA00005551 and Westat's IRB number is 00000695. The project used to have an OMB number (0920-0589). The NIH Office of Human Subjects did exempt the HINTS study from IRB review. All participants provided informed consent.

Sampling
The HINTS target population is non-institutionalized American adults (age ≥18) who reside in the United States. HINTS-5-Cycle 1 used a two-stage sampling design. First stage of the sampling was a stratified sample of addresses that were derived from all residential addresses received from the Marketing Systems Group (MSG). All non-vacant residential addresses were considered eligible for sampling. In the second stage of the sampling, one adult was selected from each sampled household. The sampling frame was grouped into two strata: Stratum #1, areas with a high concentration of minorities, and Stratum #2, areas with a low concentration of minorities. Equal-probability sampling was used to draw addresses from each sampling stratum [16].

Surveys
The surveys were sent to the participants by mail. Monetary incentive was included in the mails to encourage participation. Two toll-free telephone numbers were provided to respondents: one was used for English calls and one was used for Spanish calls. The overall response rate was 32.4 percent [16].

Study Variables
The study variables included race, ethnicity, age, gender, education, income, smoking status, health insurance status, and cancer perceived risk and cancer worries.

Independent Variables
Smoking Status. Smoking status was measured using the following item: "Have you smoked at least 100 cigarettes in your entire life?". Response options for this question were yes and no. "Ever smoker" status was defined as a positive response to this question.
Demographic Factors. Race, ethnicity, age, and gender were measured. Race was a dichotomous variable (0 Whites, 1 Blacks). Ethnicity was Hispanic versus Non-Hispanic. Age was a continuous measure ranging from 18 to 101. Gender was a dichotomous variable (0 female, 1 male).
Socioeconomic Status (SES). SES indicators in this study included education and income. Education attainment was measured as an ordinal variable with the following five categories: (1) Less than high school, (2) high school graduate, (3) some college, (4) bachelor's degree, and (5) post-baccalaureate degree. In this study, education attainment was operationalized as a continuous measure, ranging from 1 to 5, with a higher score reflecting higher educational attainment. Household income was measured using a five level ordinal variable: (1) Less than $20,000, (2) $20,000-34,999, (3) $35,000-49,999, (4) $50,000-74,999, and (5) $75,000 or more. Household income was also treated as a continuous measure, ranging from 1 to 5, with higher scores indicating higher household income.
Health Insurance. Health insurance status was evaluated using the following types of insurance: (1) Insurance purchased directly from an insurance company, (2) Medicare, for people 65 and older, or people with certain disabilities, (3) Medicaid, Medical Assistance, or any kind of government-assistance plan, (4) TRICARE or other military health care, (5) veterans affairs (VA, including those who have ever used or enrolled for VA health care) (6) Indian Health Service and (7) Any other type of health insurance or health coverage plan. Insurance status was treated as a dichotomous variable (0 without insurance, 1 with insurance).

Dependent Variables
Two single items were used to measure cancer perceived risk and cancer worries. Cancer perceived risk was measured using the following item: "How likely are you to get cancer in your lifetime?". Responses included (1) Very unlikely, (2) Unlikely, (3) Neither unlikely nor likely, (4) Likely, and (5) Very likely. Cancer worries were measured using this item: "How worried are you about getting cancer?". Responses included (1) Not at all, (2) Slightly, (3) Somewhat, (4) Moderately, and (5) Extremely. Both variables were operationalized as continuous measures, with a potential range score from 1 to 5.
For both items, a higher score indicated a worse condition (more cancer perceived risk or more cancer worries) [17,18].

Effect Modifier
Lung Cancer Risk. Ages between 55 and 77 years and ever smoking status were used to group participants into the following two risk groups: High risk group, ages between 55 and 77 years and positive history of ever smoking. Low risk group, any other individual. This categorization was based on recommendations of the CMS for identification of high risk beneficiaries for the lung cancer screening program [9]. Since pack-year smoking history was not documented in the HINTS dataset, we could not adjust exactly based on pack-year smoking.

Statistical Analysis
We used Stata 15.0 (Stata Corp., College Station, TX, USA) for univariate, bivariate, and multivariable analyses. For univariate analysis, we reported mean, frequencies, and their standard errors. For bivariate associations, Pearson's correlations tests, independent sample t-tests, and paired t-tests were used. To test demographic, social, and behavioral determinants of perceived risk of cancer and cancer worries, we ran multi-group structural equation modeling (SEM) [19] where groups were defined based on the level of lung cancer risk. Perceived risk of cancer and cancer worries were dependent variables and race, gender, ethnicity, age, education, income, insurance, and smoking status were independent variables. To test the effects of smoking status on cancer perceived risk and worries, we ran models in the pooled sample, as well as based on the level of risk. We reported path coefficients, SE, 95% CI, z-value, and p-value. P < 0.05 was considered significant.
We used maximum likelihood estimates in the presence of missing data [20,21]. Conventional model fit statistics such as the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) were used to evaluate the goodness of fit. A chi-square to degrees of freedom ratio of less than 4, a CFI above 0.95, and a RMSEA value of 0.06 or less were considered as indicators of good fit of the data [22,23].

Descriptive Statistics
The mean age of the participants was 49 years (SE = 0.34), and 52% of the participants were females. Thirteen percent of the participants were Black. About 92% of the participants had insurance. Table 1 provides a summary of the descriptive statistics for the pooled sample and subgroups based on lung cancer risk.
Associates of cancer perceived risk and worries based on lung cancer risk is shown in Table 3 and Figure 2.

Discussion
We found that being Black and of old age were associated with lower cancer perceived risk and worries irrespective of lung cancer risk. However, the effects of income and having insurance on cancer perceived risk and worries were conditional on cancer risk as they were only seen in individuals with low risk of lung cancer. Gender, ethnicity, and education were not correlates of cancer perceived risk or cancer worries irrespective of lung cancer risk level.
Smokers experienced higher levels of cancer perceived risk and worries in our analysis. These results are in line with the results of HINTS 2005 [14] regarding a positive association between smoking and cancer perceived risk. We further analyzed a large cohort of a random US population to identify the determinants of cancer perceived risk and worries in subgroups with high and low lung cancer risk.
Racial minority status, particularly being Black, is shown to be associated to lower cancer perceived risk and cancer worries [24], which may be due to low cancer literacy [25]. This is paradoxical and undesired because Blacks are at an increased risk of many types of cancer [26]. The finding that Blacks have lower cancer perceived risk and cancer worries was persistent in both high and low lung cancer risk groups. There is a need to address racial disparities in cancer perceived risk, as it may be one mechanism explaining racial gap in lung cancer survival rate [27]. It has been shown that despite coverage provided through the Affordable Care Act, Black patients are less likely to qualify for lung cancer screening [28]. In a cross-sectional study performed in the state of Indiana on 438 long-term smokers, racial and geographic disparity has been show in lung cancer screening participation [29]. Being White has also been shown to be an independent associate of high risk perception in a data collected from 630 national lung screening trial participants [30]. Racial disparity in perceived risk and worries of lung cancer among those with high risk for contracting this disease have implications in policy making, with the goal of increasing participation of high risk black individuals in lung cancer screening programs.
We found a negative correlation between age and cancer perceived risk and worries in the whole cohort and low lung cancer risk group. In a cross-sectional study on lung cancer-eligible patients, age was not a significant factor affecting lung cancer screening programs [29]. In a qualitative study assessing attitudes to participation in lung cancer screening, "being too old to benefit from lung cancer screening" was among the causes of declining lung cancer screening participations [31]. Older age has been among the determinants of declining participation in lung cancer screening in the UK [32]. Although age increases the risk of cancer, and many health problems including cancers are age

Discussion
We found that being Black and of old age were associated with lower cancer perceived risk and worries irrespective of lung cancer risk. However, the effects of income and having insurance on cancer perceived risk and worries were conditional on cancer risk as they were only seen in individuals with low risk of lung cancer. Gender, ethnicity, and education were not correlates of cancer perceived risk or cancer worries irrespective of lung cancer risk level.
Smokers experienced higher levels of cancer perceived risk and worries in our analysis. These results are in line with the results of HINTS 2005 [14] regarding a positive association between smoking and cancer perceived risk. We further analyzed a large cohort of a random US population to identify the determinants of cancer perceived risk and worries in subgroups with high and low lung cancer risk.
Racial minority status, particularly being Black, is shown to be associated to lower cancer perceived risk and cancer worries [24], which may be due to low cancer literacy [25]. This is paradoxical and undesired because Blacks are at an increased risk of many types of cancer [26]. The finding that Blacks have lower cancer perceived risk and cancer worries was persistent in both high and low lung cancer risk groups. There is a need to address racial disparities in cancer perceived risk, as it may be one mechanism explaining racial gap in lung cancer survival rate [27]. It has been shown that despite coverage provided through the Affordable Care Act, Black patients are less likely to qualify for lung cancer screening [28]. In a cross-sectional study performed in the state of Indiana on 438 long-term smokers, racial and geographic disparity has been show in lung cancer screening participation [29]. Being White has also been shown to be an independent associate of high risk perception in a data collected from 630 national lung screening trial participants [30]. Racial disparity in perceived risk and worries of lung cancer among those with high risk for contracting this disease have implications in policy making, with the goal of increasing participation of high risk black individuals in lung cancer screening programs.
We found a negative correlation between age and cancer perceived risk and worries in the whole cohort and low lung cancer risk group. In a cross-sectional study on lung cancer-eligible patients, age was not a significant factor affecting lung cancer screening programs [29]. In a qualitative study assessing attitudes to participation in lung cancer screening, "being too old to benefit from lung cancer screening" was among the causes of declining lung cancer screening participations [31]. Older age has been among the determinants of declining participation in lung cancer screening in the UK [32]. Although age increases the risk of cancer, and many health problems including cancers are age related [33,34], aging may be associated with a mental discounting of perceived risk and worries about cancer. A previous meta-analysis has shown in most studies, age is negatively correlated with cancer perceived risk; however, the effect size is small [35]. This is in line with our findings, which show less cancer worry and perceived risk in older individuals. Our further analysis demonstrates that the effect of age on cancer perceived risk and worry disappears in the lung cancer high risk group and persists in the lung cancer low risk group. This is an interesting finding which has not been reported before. Note that what we see in this study as the effect of age may be in fact a cohort effect [36]. It is very difficult to separate age and cohort effect which requires longitudinal data with multiple observations [37].
In our analysis, women showed more cancer worries only in the low risk group. Men and women in the high risk group had similar levels of cancer worry and cancer perceived risk. In a study on long-term smokers eligible for lung cancer screening program, gender was not significantly different in screening versus non-screening groups [29]. Gender is shown to be a salient determinant of perceived risk across domains [38,39] including cancer risk [34]. Not only in cancer related worries but all types of worries are more common in women than men [40]. The same pattern of worries are shown in a wide range of health domains [41] and holds for sub-clinical and clinical levels of anxiety, fear, and worries [42,43]. This may be in part due to gendered socialization and upbringing [44]. The fact that there is no gender disparity in cancer perceived risk and worries in high risk group might suggest that health system appropriately increases lung cancer awareness in this group.
High income was found to be associated with higher cancer perceived risk and worries in the whole cohort. However, further analysis demonstrated that income may not determine cancer perceived risk and worries in the high risk for lung cancer group, but is still a significant determinant for low risk for lung cancer group. It has been shown that low income group are less likely to participate in lung cancer screening programs [29]. It is paradoxical that SES (high income in this analysis) is associated with higher cancer perceived risk and worries [25] while in fact high SES is protective against cancer risk behaviors, such as smoking, being overweight, and low physical activity [45]. Concerns of high SES individuals about cancer might be related to exposure to health literacy, shaped by messages and media, and communication of health care providers, or may be simply due to a tendency to have more concerns about one's own health [46].
To avoid unnecessary and over adjustment [47,48], we decide not to control for quality of life and health behaviors. Quality of life is a broad measure which is correlated with most of the variables and constructs in this study as well as many others. For example, quality of life correlates with age, gender, socioeconomic status, health, cancer risk, and perceived risk of cancer [49]. Health behaviors are also correlated with actual and perceived risk of cancer [50]. While health behaviors and quality of life possibly correlate with our independent and dependent variables, controlling for them would have biased our results toward the null, as they could partially mediate our associations of interest. To reduce the risk of bias, researchers should be cautious about omitted confounders as well as not to control for potential mediators that are involved in underlying mechanisms [47,48]. Causal directed acyclic graphs (DAGs) can be a useful guide for selection of confounders in future research [51][52][53][54].

Limitations
Our study had a few limitations. We did not assess lung cancer screening participation rate or predictors of lung cancer screening participation. We did not include a comprehensive measure of cancer perceived risk. This analysis did not differentiate between absolute and comparative perceived risk [55,56]. Although we could not generate the exact lung cancer high risk population based on the CMS guideline due to a lack of measurement of pack-year smoking in HINTS, we included individuals between 55 and 77 years old with a history of smoking as per CMS guidelines. Considering the fact that the mean age was 49 years in our cohort and most smokers initiate smoking prior to age 26, it is probable that the majority of the smokers in our cohort are long-term smokers [57]. The cross-sectional nature of our data and the use of self-report measures are also among the limitations of our analysis. Despite these limitations, large sample size, national representative sample, and conceptualization of risk as a moderator were among the strengths of this study.

Implications
Although we did not directly assess the effect of socioeconomic factors on lung cancer screening participation, the results of our analysis might have significant implications for practice and policy-making to improve participation of individuals at high risk for lung cancer into a lung cancer screening program. First, fewer determinants are available in high lung cancer risk group. This makes the promotion of cancer screening in high risk group more difficult and probably costlier. Second, older individuals, minorities, and low SES people, although at an increased risk of lung cancer, paradoxically reported low cancer perceived risk. These people may be less willing to undergo screening for something they do not find as a risk or threat. Efforts for education are needed and may require messages by clinicians as well as media campaigns, particularly for high risk groups. Third, as smokers already perceive the risk, they may have higher readiness to participate into lung cancer screening programs.

Conclusions
We found that determinants of cancer perceived risk and worries vary in adults with low versus high lung cancer risk. Income and insurance are correlated with higher cancer perceived risk and worries in low lung cancer risk group but not in the high lung cancer risk group. Being Black was associated with lower cancer perceived risk and worries in both low and high lung cancer risk groups. While cancer perceived risk and worries reduce quality of life and cause distress [50], these constructs can be leveraged to promote lung cancer screening participation. More research is needed on the most efficient strategies to improve lung cancer screening participation in the high risk group.