Abstract
Background
Population-based data from the National Health Interview Survey were examined to provide estimates of a wide range of health behaviors in cancer survivors (ie, physical activity, sun protection, alcohol use, cigarette and e-cigarette use, sleep, and diet) and trends over time.
Methods
Data were collected from 92 257 participants across 3 waves of the National Health Interview Survey. A total of 8050 participants reported having had cancer (2428 in 2005, 2333 in 2010, 3289 in 2015). Weighted and adjusted odds ratios (OR) between cancer survivors and controls were calculated using logistic and multivariable regressions in SPSS, and trend analyses from 2005 to 2015 were conducted. All statistical tests are 2-sided.
Results
After adjusting for demographics (2005-2015), cancer survivors, compared with controls, were more likely to wear sunscreen (OR = 1.41, 95% confidence interval [CI] = 1.32 to 1.51) and protect their skin (P < .001) and were less likely to tan indoors (OR = 0.81, 95% CI = 0.69 to 0.95), but reported less sleep (OR = 0.91, 95% CI = 0.85 to 0.98). In adjusted models, no differences were found for physical activity, sunburns, alcohol use, smoking, e-cigarette use, and diet. Smoking rates for cancer survivors decreased from 2005 to 2015 (P < .001) and physical activity increased (P = .02), but physical activity was not statistically significant after adjusting for multiple comparisons. All other health behavior rates for cancer survivors were unchanged from 2005 to 2015 (P > .14).
Conclusion
After adjusting for covariates, cancer survivors exhibited healthier sun protection, but not sleep behaviors, compared with controls. Cancer survivors (and controls) exhibited decreased smoking rates over time. These results may inform interventions focused on improving cancer control and prevention of other chronic conditions among cancer survivors.
In 2019, there were an estimated 16.9 million cancer survivors in the United States; by 2030, that number is expected to rise to 22.2 million (1). This sharp increase is fueled in part by recent advances in the early detection and treatment of cancer, and as a result many can expect to live for decades after diagnosis (2). Despite the promise of longer term survival, cancer survivors are at an increased risk of recurrence and secondary cancers (3), particularly when they engage in behaviors shown to increase cancer risk. Cancer survivors may also have preexisting comorbidities that are exacerbated by primary cancer treatments (4). Finally, exposure to cancer treatments (particularly radiation and chemotherapy) may result in long-term adverse effects that negatively affect quality of life well beyond the duration of treatment itself (5, 6). For these reasons, cancer survivors represent an important population to consider from a health behavior perspective.
Previous population-level research investigating multiple health behaviors of cancer survivors found they were similar to controls on a number of domains (smoking, alcohol consumption) but more likely to meet physical activity recommendations (7). Younger survivors (ages 18-40 years) were also at a higher risk of smoking than controls (7). However, this research did not include all health behaviors associated with cancer prevention (eg, diet, sun exposure, sleep). Although National Health Interview Survey (NHIS) data have been used to examine health behaviors in cancer survivors, the purpose of the current analysis is to address a wider range of behaviors, including those of more recent interest (eg, e-cigarette use). Additionally, 10-year trends of health behaviors were examined where relevant data were available.
Methods
Source of Data
Nationally representative data were obtained from the 2005, 2010, and 2015 NHIS cross-sectional adult sample and cancer control supplements (CCS). The NHIS is a population-based, cross-sectional survey, primary source of health information of US noninstitutionalized adults conducted through in-person interviews (8). For more information about the sample design and data collection procedures, see http://www.cdc.gov/nchs/nhis/about_nhis.htm.
The basic NHIS module includes 4 major components: household, family, sample adult, and sample child. Demographic information is collected in the household component. The adult component samples 1 adult within each family per household and contains questions pertaining to health behaviors, health status, and health-care utilization. CCS are included to assess current health topics and vary across survey years. The CCS is administered every 5 years and assesses cancer-relevant variables in a variety of health domains, including physical activity, sun protection and exposure, alcohol and tobacco use, dietary behavior, sleep, and cancer screening.
Participants
A total of 92 257 participants were surveyed. Cancer status was assessed with a question examining cancer history (“Have you ever had cancer?”). Across the 3 years of data, 8050 participants were identified as cancer survivors (2428 in 2005, 2333 in 2010, 3289 in 2015). Participants with responses of “refused” and “don’t know” were excluded from analysis, as were those whose responses could not be ascertained (N = 84). Cancer treatment, related symptoms, and current cancer status were not available in NHIS data, but a functional limitations item assessed degree of current symptomatology for cancer survivors and controls. Analyses of these data were exempt from review by the National Cancer Institute institutional review board because the data are publicly available and deidentified.
Measures
Self-reported demographic information was obtained for age, sex, race, income, and education. Additionally, current health status (compared with 12 months before the time of the survey) and functional limitations due to chronic illness were assessed. For those reporting a history of cancer, time since diagnosis was computed by subtracting age at initial (or first) diagnosis from age at time of interview.
Descriptions and derivation of scores for physical activity (adequate amounts of moderate and vigorous physical activity), sun protection and exposure (sunscreen use, sun protection, sunburn, and indoor tanning), alcohol use (current drinking status), smoking behavior (cigarette and e-cigarette current smoking status), dietary behaviors (fruits and vegetables consumed), and sleep (adequate amounts) are reported on in detail in the Supplementary Material (available online). All outcomes were binary (eg, met sleep recommendations or smoked vs did not) except for the sun protection index and dietary behavior, which were continuous.
Statistical Analyses
The NHIS uses a complex, multistage sampling design that includes stratification, clustering, and oversampling of specific population subgroups. With complex samples (like NHIS that includes multistage sampling, stratification, etc), traditional standard error estimates may be inaccurate (9). To account for this, Taylor series linearized variance estimation methods were used for analyses, which is a method recommended in the NHIS Variance Estimation Guidance document for 2006-2015.
Chi-squared analyses were used to compare demographic characteristics of cancer survivors with controls; the tested variables included age, sex, race, education, income, health status, and overall functional limitations. Because of the large sample size, a more stringent P value (at or below the <.001 level) was used to determine demographic differences from χ2 analyses. All demographic variables, however, were included in the multivariable models, and all other analyses used a 95% confidence interval (CI) or P less than .05 threshold with the exception of overall trend analyses (P < .008; 2005-2015) (7). All statistical tests were 2-sided.
Multivariable logistic regressions were used to compare cancer survivors with controls across health behaviors that were binary, and multivariable (nonlogistic) regression analyses were similarly used for health behaviors measured on a continuous scale (eg, sun-protective behavior and dietary behavior). Both types of multivariable regression analyses adjusted for the demographic variables included in the aforementioned χ2 analyses. Because indoor tanning and e-cigarette use are uncommon among middle- and older-aged adults relative to younger adults, for these outcomes only we conducted separate multivariable logistic regression analyses for participants younger than 40 years of age, comparing cancer survivors with controls. For all other analyses, we conducted multivariable (logistic and nonlogistic) regressions based on age (19-39 years vs 65+ years, 40-64 years vs 65+ years) for cancer survivors only to test for differences between age groups. Finally, χ2 analyses were used to compare estimates of health behaviors between the 3 time periods (2005, 2010, and 2015) stratified by cancer survivors between years as well as by controls between years, applying sample weights across all 3 years. The SPSS IBM software subscription using the Complex Samples add-on was used to analyze data.
Results
Characteristics of Cancer Survivors
A total of 92 173 respondents were included in the analyses. Chi-squared analyses showed that cancer survivors were statistically significantly more likely to be older (>65 years, 50.5% vs 14.5%; P < .001), female (56.6% vs 51.3%; P < .001), White (90.8% vs 80.0%; P < .001), and have a lower income (<$35 000, 34.3% vs 31.6%; P < .001). Cancer survivors were also more likely to report their current health status as worse than that of the previous 12 months (16.3% vs 8.0%; P < .001) and to report functional limitations (61.1% vs 32.1%; P < .001). Cancer survivors and controls did not differ on education (P = .05) using the P ≤ .001 threshold. See Table 1 for descriptive statistics.
Table 1.
Characteristics of cancer survivors
| Characteristics | Self-reported history of cancer |
No history of cancer |
||
|---|---|---|---|---|
| Total No. (%)a | 95% CI | Total No. (%)a | 95% CI | |
| Total | 8050 | 84 123 | ||
| Sexb | ||||
| Male | 3246 (43.4) | 42.0 to 44.8 | 37 535 (48.7) | 48.3 to 49.1 |
| Female | 4804 (56.6) | 55.2 to 58.0 | 46 588 (51.3) | 50.9 to 51.7 |
| Ageb | ||||
| 18–39 y | 568 (8.0) | 7.2 to 8.8 | 32 773 (42.1) | 41.5 to 42.7 |
| 40–64 y | 3047 (41.5) | 40.1 to 42.9 | 35 909 (43.5) | 43.1 to 43.9 |
| 65+ y | 4435 (50.5) | 49.1 to 51.9 | 15 441 (14.5) | 14.1 to 14.9 |
| Raceb | ||||
| White | 7028 (90.8) | 90.0 to 91.6 | 64 247 (80.0) | 79.4 to 80.6 |
| Black | 703 (6.0) | 5.4 to 6.6 | 12 948 (12.5) | 12.1 to 12.9 |
| Asian | 157 (1.6) | 1.2 to 2.0 | 4576 (5.2) | 5.0 to 5.4 |
| American Indian or Alaska Native | 48 (0.6) | 0.4 to 0.8 | 783 (0.8) | 0.6 to 1.0 |
| Multiple | 108 (1.0) | 0.8 to 1.2 | 1401 (1.5) | 1.3 to 1.7 |
| Education | ||||
| <High school | 1238 (13.5) | 12.5 to 14.5 | 13 449 (14.1) | 13.7 to 14.5 |
| High school graduate | 2152 (27.2) | 26.0 to 28.4 | 22 282 (27.1) | 26.7 to 27.5 |
| 2- or 4-Year college graduate | 2353 (29.1) | 27.9 to 30.3 | 24 827 (30.2) | 29.8 to 30.6 |
| >Bachelor’s degree | 2245 (30.2) | 28.8 to 31.6 | 22 574 (28.6) | 28.0 to 29.2 |
| Incomeb | ||||
| <$35k | 3178 (34.3) | 32.9 to 35.7 | 31 732 (31.6) | 31.0 to 32.2 |
| $35k–$75k | 2078 (29.8) | 28.4 to 31.2 | 22 653 (30.6) | 30.2 to 31.0 |
| >$75k | 2069 (35.9) | 34.5 to 37.3 | 22 604 (37.8) | 37.2 to 38.4 |
| Health statusb,c | ||||
| Better | 1481 (18.6) | 17.4 to 19.8 | 14 887 (17.8) | 17.4 to 18.2 |
| Worse | 1353 (16.3) | 15.3 to 17.3 | 7307 (8.0) | 7.8 to 8.2 |
| About the same | 5181 (65.2) | 63.8 to 66.6 | 61 809 (74.2) | 73.8 to 74.6 |
| Functional limitationsb | ||||
| Not limited | 2912 (38.9) | 37.3 to 40.5 | 55 130 (67.9) | 67.5 to 68.3 |
| Limited | 5138 (61.1) | 59.5 to 62.7 | 28 993 (32.1) | 31.7 to 32.5 |
All percentages are weighted. Cancer type prevalence across years: breast (16.1%), skin nonmelanoma (15.1%), prostate (9.7%), skin (don't know) (6.5%), cervical (5.6%), melanoma (5.3%), colon (5.1%), 2 or more (11.7%), all others (24.9%). CI = confidence interval.
Chi-squared test of independence (cancer survivors vs controls), P ≤ .001.
Compared with 12 months earlier.
Health Behaviors of Cancer Survivors
After adjusting for all covariates, cancer survivors had higher odds of using sunscreen (OR = 1.41, 95% CI = 1.32 to 1.51, P < .001), were more likely to engage in other sun-protective behaviors (β = .05, P < .001), and had lower odds of engaging in indoor tanning (OR = 0.81, 95% CI = 0.69 to 0.95, P = .007) relative to controls. However, cancer survivors had lower odds of getting adequate sleep (OR = 0.91, 95% CI = 0.85 to 0.98, P = .03). For other health behaviors, there were no statistically significant differences. See Table 2 for inferential statistics and adjusted mean estimates and Figure 1 for a forest plot of odds ratios.
Table 2.
Inferential statistics and adjusted mean health outcomes for cancer survivors and controls
| Health behaviora | Odds ratio (95% CI) | β (95% CI) | Adjusted means, % |
|
|---|---|---|---|---|
| Survivors | Controls | |||
| Physical activity | 1.00 (0.91 to 1.11) | — | 19.1 | 19.1 |
| Sunscreen use | 1.41b (1.32 to 1.51) | — | 41.9 | 34.6 |
| Sun-protective behavior | — | 0.05b (0.04 to 0.06) | 0.43 | 0.37 |
| Sunburn | 0.95 (0.88 to 1.04) | — | 33.7 | 34.6 |
| Indoor tanning (all) | 0.81b (0.69 to 0.95) | — | 6.5 | 7.9 |
| Indoor tanning (18-39 y) | 1.01 (0.74 to 1.38) | — | 11.5 | 11.4 |
| Alcohol use | 0.99 (0.92 to 1.05) | — | 63.3 | 63.7 |
| Smoking | 1.01 (0.92 to 1.10) | — | 18.3 | 18.2 |
| E-cigarettes (all) | 1.06 (0.76 to 1.47) | — | 3.4 | 3.2 |
| E-cigarettes (18-39 y) | 1.44 (0.73 to 2.85) | — | 6.8 | 4.9 |
| Adequate sleep | 0.91b (0.85 to 0.98) | — | 64.3 | 66.3 |
| Dietary behavior | — | −0.01 (−0.02 to .002) | 1.63 | 1.63 |
Physical activity = dichotomous variable (Y/N) where meeting physical activity guidelines (yes) indicates meeting both aerobic and strengthening guidelines; sunscreen use = dichotomous variable (Y/N) where sunscreen use (yes) indicates using sunscreen at least most of the time; sun-protective behavior = continuous variable index of shade seeking, protective clothing use, and hat-wearing behavior (higher scores indicate more sun protection); sunburn = dichotomous variable (Y/N) where a sunburn (yes) indicates 1 or more sunburns in previous 12 months; indoor tanning = dichotomous variable (Y/N) where “yes” indicates 1 or more reported uses of indoor tanning in previous 12 months; alcohol use = dichotomous variable (Y/N) where "yes" indicates current light, moderate, or heavy drinking; smoking = dichotomous variable (Y/N) where “yes” indicates reporting being a current smoker; E-cigarettes (2015 analysis only) = dichotomous variable (Y/N) where “yes” indicates having used an e-cigarette in previous 30 days; adequate sleep = dichotomous variable (Y/N) where “yes” indicates getting 7-9 hours of sleep per night; dietary behavior = continuous variable composite of fruit and vegetable consumption.
A statistically significant difference was found. Those without a history of cancer constitute the reference group for odds ratio. All analyses were adjusted for demographic covariates (age, sex, race, education, income, health status, and overall functional limitations).
Figure 1.
Forest plot of adjusted odds ratios comparing cancer survivors to controls. All analyses were adjusted for demographic covariates (age, sex, race, education, income, health status, and overall functional limitations). Physical activity (PA) = dichotomous variable (Y/N) where meeting PA guidelines (yes) indicates meeting both aerobic and strengthening guidelines; sunscreen use = dichotomous variable (Y/N) where sunscreen use (yes) indicates using sunscreen at least most of the time; sunburn = dichotomous variable (Y/N) where a sunburn (yes) indicates 1 or more sunburns in previous 12 months; indoor tanning = dichotomous variable (Y/N) where “yes” indicates 1 or more reported uses of indoor tanning in previous 12 months; alcohol use = dichotomous variable (Y/N) where “yes” indicates current light, moderate, or heavy drinking; smoking = dichotomous variable (Y/N) where “yes” indicates reporting being a current smoker; e-cigarettes (2015 analysis only) = dichotomous variable (Y/N) where “yes” indicates having used an e-cigarette in previous 30 days; adequate sleep = dichotomous variable (Y/N) where “yes” indicates getting 7-9 hours of sleep per night. CI = confidence interval.
Except for e-cigarette use (due to low prevalence among older adults), health behaviors were analyzed by age (18-39 years, 40-64 years, 65+ years) for cancer survivors. Differences in age were found in physical activity, sunscreen use, sun-protective behavior, sunburn, indoor tanning, alcohol use, smoking, and sleep. There were no statistically significant differences in dietary behavior. In general, older cancer survivors engaged in more healthy behaviors than younger cancer survivors except for physical activity. See Tables 3 and 4 for inferential statistics and adjusted mean estimates for each age group. See Figure 2 for a graphical representation of the results. Unadjusted descriptive statistics are presented in the Supplementary Table 1 (available online).
Table 3.
Inferential statistics and adjusted mean health outcomes within cancer survivors by age for dichotomous variables
| Health behaviora | Odds ratio (95% CI) |
Adjusted means, % |
|||
|---|---|---|---|---|---|
| 18-39 y vs 65+ y | 40-64 y vs 65+ y | 18-39 y | 40-64 y | 65+ y | |
| Physical activity | 1.40b (1.00 to 1.95) | 1.14 (0.95 to 1.38) | 17.6 | 15.1 | 13.6 |
| Sunscreen use | 0.78b (0.61 to 0.99) | 0.94 (0.82 to 1.08) | 42.1 | 46.4 | 47.7 |
| Sunburn | 8.69b (6.79 to 11.12) | 3.39b (2.91 to 3.96) | 56.6 | 34.8 | 14.0 |
| Indoor tanning | 6.74b (4.52 to 10.04) | 2.59b (1.85 to 3.62) | 13.2 | 5.6 | 2.3 |
| Alcohol use | 2.06b (1.59 to 2.67) | 1.54b (1.37 to 1.75) | 68.6 | 62.8 | 53.4 |
| Smoking | 7.92b (6.04 to 10.38) | 4.28b (3.53 to 5.18) | 33.1 | 22.2 | 6.7 |
| Adequate sleep | 0.41b (0.32 to 0.52) | 0.66b (0.58 to 0.74) | 48.9 | 59.7 | 68.7 |
Physical activity = dichotomous variable (Y/N) where meeting physical activity guidelines (yes) indicates meeting both aerobic and strengthening guidelines; sunscreen use = dichotomous variable (Y/N) where sunscreen use (yes) indicates using sunscreen at least most of the time; sunburn = dichotomous variable (Y/N) where a sunburn (yes) indicates 1 or more sunburns in previous 12 months; indoor tanning = dichotomous variable (Y/N) where “yes” indicates 1 or more reported uses of indoor tanning in previous 12 months; alcohol use = dichotomous variable (Y/N) where “yes” indicates current light, moderate, or heavy drinking; smoking = dichotomous variable (Y/N) where “yes” indicates reporting being a current smoker; adequate sleep = dichotomous variable (Y/N) where “yes” indicates getting 7-9 hours of sleep per night.
A statistically significant difference was found. The reference group for odds ratio is the 65+ year-old age group. All analyses were adjusted for demographic covariates (age, sex, race, education, income, health status, and overall functional limitations).
Table 4.
Inferential statistics and adjusted mean health outcomes within cancer survivors by age for continuous variables
| Health behaviora |
β (SE) |
Adjusted means |
|||
|---|---|---|---|---|---|
| 18-39 y vs 65+ y | 40-64 y vs 65+ y | 18-39 y | 40-64 y | 65+ y | |
| Sun-protective behavior | −0.23b (0.01) | −0.13b (0.02) | 0.34 | 0.45 | 0.58 |
| Dietary behavior | −0.02 (0.02) | 0.00 (0.01) | 1.60 | 1.62 | 1.62 |
Sun-protective behavior = continuous variable index of shade seeking, protective clothing use, and hat-wearing behavior (higher scores indicate more sun protection); dietary behavior = continuous variable composite of fruit and vegetable consumption.
A statistically significant difference was found. The reference group for odds ratio is the 65+ year-old age group. All analyses were adjusted for demographic covariates (age, sex, race, education, income, health status, and overall functional limitations).
Figure 2.
Bar graphs comparing age groups within cancer survivors adjusting for covariates. The y-axis represents percentage rates (0%-100%) for individuals within each age group engaging in these behaviors after adjusting for covariates, with the exception of “sun health” and “dietary,” which are adjusted mean rates because these were continuous (not binary) variables. All analyses were adjusted for demographic covariates (age, sex, race, education, income, health status, and overall functional limitations). Functional limitations split by age among individuals with a history of cancer: 18-39 years (39.8%), 40-64 years (57.1%), 65+ years (71.5%). Functional limitations split by age among individuals without a history of cancer: 18-39 years (17.3%), 40-64 years (37.5%), 65+ years (63.7%). Physical activity (PA) = dichotomous variable (Y/N) where meeting PA guidelines (yes) indicates meeting both aerobic and strengthening guidelines; sunscreen use = dichotomous variable (Y/N) where sunscreen use (yes) indicates using sunscreen at least most of the time; sun-protective behavior = continuous variable index of shade seeking, protective clothing use, and hat-wearing behavior (higher scores indicate more sun protection); sunburn = dichotomous variable (Y/N) where a sunburn (yes) indicates one or more sunburns in previous 12 months; indoor tanning = dichotomous variable (Y/N) where “yes” indicates 1 or more reported uses of indoor tanning in previous 12 months; alcohol use = dichotomous variable (Y/N) where “yes” indicates current light, moderate, or heavy drinking; smoking = dichotomous variable (Y/N) where “yes” indicates reporting being a current smoker; adequate sleep = dichotomous variable (Y/N) where “yes” indicates getting 7-9 hours of sleep per night; dietary behavior = continuous variable composite of fruit and vegetable consumption.
10-Year Trend Analysis
To assess the trends in health behaviors between 2005 and 2015, χ2 tests were conducted for categorical variables and linear tests were conducted for continuous variables; all analyses were stratified by cancer survivor status. Due to multiple trend comparisons, we adjusted alpha from less than .05 to less than .008 (ie, [0.05/6]), but only for the overall trend analyses (ie, 2005-2015) within cancer survivors for our 6 primary outcomes (physical activity, sun-protective behavior, drinking status, smoking status, sleep, and dietary behavior). An α of less than .05 was used for all other secondary outcomes (sunscreen use, sunburns, and indoor tanning; see Supplementary Methods available online), between-year comparisons (ie, from 2005 to 2010 or 2010 to 2015; see Table 5 and Supplementary Methods available online), and overall control group trends.
Table 5.
Weighted trends in health behaviors among cancer survivors and controls
| Health behaviora | Self-reported history of cancer |
No history of cancer |
||||
|---|---|---|---|---|---|---|
| Year |
Year |
|||||
| 2005 | 2010 | 2015 | 2005 | 2010 | 2015 | |
| Physical activity, % (SE) | 12.4 (0.8) | 15.5b (1.0) | 15.4 (0.9) | 16.5 (0.3) | 20.6b (0.4) | 21.2 (0.3) |
| Sunscreen use, % (SE) | 45.4 (1.1) | 45.4 (1.2) | 48.9b (1.2) | 32.7 (0.4) | 34.5b (0.5) | 36.9b (0.4) |
| Sunburn, % (SE) | 25.2 (1.0) | 28.3b (1.2) | 24.8b (1.1) | 34.9 (0.4) | 37.9b (0.4) | 35.0b (0.5) |
| Indoor tanning, % (SE) | 9.7 (0.7) | 3.3b (0.5) | 1.7b (0.4) | 14.2 (0.3) | 5.7b (0.2) | 4.1b (0.2) |
| Alcohol use, % (SE) | 56.7 (1.1) | 58.9 (1.2) | 59.0 (1.3) | 61.8 (0.4) | 65.4b (0.5) | 65.7 (0.4) |
| Smoking, % (SE) | 17.1 (0.9) | 16.0 (0.8) | 11.5b (0.7) | 21.2 (0.3) | 19.7b (0.3) | 15.5b (0.3) |
| Adequate sleep, % (SE) | 63.7 (1.1) | 63.7 (1.2) | 63.1 (1.1) | 67.4 (0.4) | 67.1 (0.4) | 64.1b (0.4) |
| Sun-protective behavior, Mean (SE) | 0.51 (.01) | 0.50 (.01) | 0.52 (.01) | 0.36 (.003) | 0.38b (.003) | 0.39 (.003) |
| Dietary behavior, Mean (SE) | 1.61 (.01) | 1.62 (.01) | 1.62 (.01) | 1.64 (.003) | 1.63 (.003) | 1.62b (.002) |
Physical activity = dichotomous variable (Y/N) where meeting physical activity guidelines (yes) indicates meeting both aerobic and strengthening guidelines; sunscreen use = dichotomous variable (Y/N) where sunscreen use (yes) indicates using sunscreen at least most of the time; sunburn = dichotomous variable (Y/N) where a sunburn (yes) indicates 1 or more sunburns in previous 12 months; indoor tanning = dichotomous variable (Y/N) where “yes” indicates 1 or more reported uses of indoor tanning in previous 12 months; alcohol use = dichotomous variable (Y/N) where “yes” indicates current light, moderate, or heavy drinking; smoking = dichotomous variable (Y/N) where “yes” indicates reporting being a current smoker; adequate sleep = dichotomous variable (Y/N) where “yes” indicates getting 7-9 hours of sleep per night; sun protection behavior = continuous variable index of shade seeking, protective clothing use, and hat-wearing behavior (higher scores indicate more sun protection); dietary behavior = continuous variable composite of fruit and vegetable consumption.
Denotes statistically significant change from previous year of measurement at alpha (α) less than .05.
From 2005 to 2015, physical activity increased for cancer survivors (χ2 = 12.41, P = .02) and smoking rates decreased (χ2 = 38.11, P < .001). However, after adjusting for multiple comparisons, the physical activity increase was not statistically significant. For controls, physical activity and sun protection increased and smoking rates decreased from 2005 to 2015 (P < .001), but there were also statistically significant increases in drinking status and decreases in adequate sleep and health dietary behavior (P < .001). Cancer survivors did not exhibit changes in sun protection, alcohol use, adequate sleep, or health dietary behavior from 2005 to 2015 (P > .14). For all other individual year comparisons or secondary outcomes (sunscreen use, sunburns, and indoor tanning), see Table 5 or the Supplementary Methods (available online) for between-year comparisons, secondary variable outcomes, and associated test statistics.
Discussion
These data provide important updates and insight into the health behaviors of cancer survivors. Unadjusted rates of physical activity were lower among cancer survivors; however, when demographic differences—especially age and functional limitations—were controlled for, cancer survivors were no less likely than controls to meet physical activity recommendations. These results may warrant additional consideration given the fact that functional limitations were more prevalent among cancer survivors than controls, and functional limitations also arise shortly after cancer treatment and are predictive of survival (10). Cancer status may be introducing an additional insult into the aging process (ie, accelerated aging), whereby becoming functionally limited creates a barrier to physical activity. It will be important to identify the degree of functional limitations in cancer survivors and account for and address them in surveillance and intervention research.
Cancer survivors generally demonstrated greater sun-safe behaviors than controls—they were more likely than controls to use sunscreen and other sun-protective behaviors, and they were also less likely to have used indoor tanning. However, the incidence of indoor tanning among young-adult cancer survivors is concerning, as has been previously reported, because cancer survivors may be at increased risk for melanoma as a second cancer (11, 12). Despite relatively greater use of sunscreen and other sun safety behaviors, cancer survivors had similar odds of getting sunburn compared with controls when adjusting for demographic characteristics. Cancer treatment could have increased sun sensitivity such that greater engagement in sun safety practices did not convey similar protection from sunburn. Because sun-protective practices were broadly assessed and specific contexts of exposure were unknown, it is possible that cancer survivors’ use of sun-protective practices differed from individuals without a cancer history, in turn affecting sunburn rates. Future research should consider the context of sun-protective behaviors and sunburn, particularly among younger cancer survivors, to identify possible intervention targets.
Cancer survivors were similar to controls on current smoking status and e-cigarette use in adjusted models. Younger and middle-aged cancer survivors, however, had substantially higher rates of smoking than older survivors in adjusted models. Notably, this same finding was observed in earlier estimates of NHIS data (1998-2001, 6); though the proportion of younger survivors who reported smoking has decreased from the previously reported estimate of 42.6% to the current estimate of 33.1%, this may suggest a continued opportunity for behavioral intervention. For e-cigarette use, rates did not differ among cancer survivors and controls, nor did they differ between these groups for the 18- to 39-year-old age group (after adjusting for covariates). However, the sample size among cancer survivors reporting smoking e-cigarettes was small (n = 72).
Unadjusted overall rates of alcohol use were lower among cancer survivors, but adjusted percentages indicated that cancer survivors were similar to controls for current alcohol use. Adjusted models indicated that cancer survivors were getting less sleep than controls, and younger and middle-aged cancer survivors reported lower rates of adequate sleep compared with their older cancer survivor counterparts. Finally, diet quality was similar between cancer survivors and controls.
The 10-year trends of health behaviors observed among cancer survivors and controls indicated that for both groups, physical activity increased from 2005 (but remained steady between 2010 and 2015). For both groups, sunburn rates increased from 2005 to 2010 but then decreased in 2015. Indoor tanning rates decreased overall for both groups from 2005 to 2015. Smoking rates decreased across the 3 time points for both groups, and the proportion of current drinkers increased slightly over time for both groups, although this increase was only statistically significant for controls. Diet quality did not change overall for cancer survivors, but for controls it declined slightly in 2015. Similarly, the proportion of respondents getting adequate sleep did not change over time for cancer survivors, but did decline in 2015 for controls.
Several limitations should be noted. First, the NHIS uses self-report data that may be susceptible to response bias. Additionally, these cross-sectional data do not allow for inferences to be made regarding the temporal nature of these relationships. For example, it is not known whether greater sun-safe behaviors are a function of having been diagnosed with cancer, or if those behaviors preceded the cancer diagnosis. This survey did not assess whether cancer survivors were currently undergoing treatment, nor did it include survivors who were hospitalized or in care facilities (eg, hospice). As such, these estimates may be biased towards a healthier subset of cancer survivors. Another unmeasured assessment that could have provided insight into the temporal nature of health behaviors for cancer survivors was motivation to engage in such behaviors as a result of a cancer diagnosis, and conversely, motivation for engaging in such behaviors (or not) among controls to prevent cancer. Finally, it is unclear if health behavior trends were affected by shifts in prevalence rates of certain types of cancer that have occurred over time.
This study addresses an important consideration for cancer care providers. Because cancer survivors are living longer, consideration of health behaviors of survivors is critical to address prevention of secondary cancers and mitigation of functional limitations attendant to cancer treatment and to lessen morbidity from other chronic diseases and conditions. These data provide insight into potential areas of health risk of cancer survivors (eg, alcohol use) as well as specific subsets of survivors who are more likely to engage in risky health behaviors (eg, smoking in younger survivors) or those survivors who may be at greater risk for second cancers (eg, young survivors incurring sunburns or indoor tanning). Such information is important for clinicians and oncologists recommending behavioral counseling to cancer survivors and for the development and implementation of interventions aimed at managing the adverse effects of cancer to prevent cancer recurrence and other chronic conditions.
Funding
No funding included for this article.
Notes
Disclosures: The authors report no conflicts of interest.
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the National Cancer Institute or the National Institutes of Health.
Author contributions: K.M. and F.P. helped conceive of this study. P.B., M.L., and K.M. cleaned and analyzed data. P.B., M.L., and K.M. wrote the paper with input from all authors.
Supplementary Material
References
- 1. American Cancer Society. Cancer Treatment and Survivorship Facts and Figures 2019-2021. Atlanta, GA: American Cancer Society; 2019. [Google Scholar]
- 2. Bluethmann SM, Mariotto AB, Rowland JH. Anticipating the “Silver Tsunami”: prevalence trajectories and comorbidity burden among older cancer survivors in the United States. Cancer Epidemiol Biomarkers Prev. 2016;25(7):1029–1036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Donin N, Filson C, Drakaki A, et al. Risk of second primary malignancies among cancer survivors in the United States, 1992 through 2008. Cancer. 2016;122(19):3075–3086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Leach CR, Weaver KE, Aziz NM, et al. The complex health profile of long-term cancer survivors: prevalence and predictors of comorbid conditions. J Cancer Surviv. 2015;9(2):239–251. [DOI] [PubMed] [Google Scholar]
- 5. Ganz PA. Late effects of cancer and its treatment. Semin Oncol Nurs. 2001;17(4):241–248. [DOI] [PubMed] [Google Scholar]
- 6. Carver JR, Shapiro CL, Ng A, et al. American Society of Clinical Oncology clinical evidence review on the ongoing care of adult cancer survivors: cardiac and pulmonary late effects. J Clin Oncol. 2007;25(25):3991–4008. [DOI] [PubMed] [Google Scholar]
- 7. Bellizzi KM, Rowland JH, Jeffery DD, McNeel T. Health behaviors of cancer survivors: examining opportunities for cancer control intervention. J Clin Oncol. 2005;23(34):8884–8893. [DOI] [PubMed] [Google Scholar]
- 8. National Center for Health Services, Division of Health Interview Statistics. 2000 National Health Interview Survey (NHIS) Public Use Data Release: NHIS Survey Description. Hyattsville, MD: United States Department of Health and Human Services, Center for Disease Control and Prevention; 2002. [Google Scholar]
- 9. Bergdahl M, Black O, Bowater R, Chambers R. Model Quality Report In Business Statistics. Theory and Methods for Quality Evaluation. Vol. 1 Luxembourg: Eurostat; 1999. [Google Scholar]
- 10. Jones LW, Hornsby WE, Goetzinger A, et al. Prognostic significance of functional capacity and exercise behavior in patients with metastatic non-small cell lung cancer. Lung Cancer (Amsterdam, Netherlands). 2012;76(2):248–252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Wiznia L, Dai F, Chagpar AB. Do non-melanoma skin cancer survivors use tanning beds less often than the general public? Dermatol Online J. 2016;22(8):13030/qt75n111ds. [PubMed] [Google Scholar]
- 12. Carretier J, Boyle H, Duval S, et al. A review of health behaviors in childhood and adolescent cancer survivors: toward prevention of second primary cancer. J Adolesc Young Adult Oncol. 2016;5(2):78–90. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.