Factors associated with physical inactivity in adult breast cancer survivors—A population‐based study

Abstract Background Physical activity has been shown to reduce the risk of breast cancer‐specific mortality. Although factors associated with physical inactivity in breast cancer survivors have been studied, a detailed examination at the population level is still lacking. Methods We addressed this gap in 1236 women with a diagnosis of breast cancer from the 2016 Behavioral Risk Factor Surveillance System Cancer Survivorship module. Physical inactivity was defined as self‐reported absence of leisure time physical activity. Factors examined in the multivariable logistic regression model included sociodemographic, behavioral factors, access to health care, health history, current cancer treatment, and pain from cancer or treatment. Results Overall, older age (≥65 years: OR = 2.63, 95% CI: 1.25‐5.55) and being underweight (BMI <18.5: OR = 6.11, 95% CI: 1.35‐27.66), were identified as significant factors associated with physical inactivity. In models adjusting for sociodemographics (Model 1), and the prior plus behavioral factors (Model 2), pain from cancer or treatment was significantly associated with physical inactivity (Model 2: OR = 2.23, 95% CI: 1.16‐4.28); however, after fully adjusting for all variables (Model 3), there was no longer evidence of a significant association between pain from cancer and physical activity in female survivors with breast cancer. Conclusions We identified demographic (older age) and physical (low BMI and pain) factors to be significantly associated with physical inactivity among breast cancer survivors. Future interventions to promote physical activity in breast cancer survivors could benefit by taking into account these factors to develop tailored recommendations for increasing activity.


| Study population
The BRFSS is a population-based telephone survey conducted annually in all 50 states, Washington, DC, and participating US territories to collect health information including health behaviors, preventive health practices, healthcare access, and chronic conditions among noninstitutionalized US adults ≥18 years of age. 43 BRFSS is a "public-use" deidentified data set that does not require IRB approval for use in research. The median survey response rate in 2016 was 47%, and the median cooperation rate (the percentage of eligible persons contacted who completed the interview) was 70.5%. 42 The 2016 BRFSS Cancer Survivorship module was administered in the following eight states: Idaho, Indiana, Louisiana, Michigan, Missouri, South Dakota, Virgin Islands, and Wisconsin. 42 The data were weighted using poststratification methodology to adjust for the unequal probability of selection, differential nonresponse, and possible deficiencies in the sampling frame. 43 As part of the BRFSS cancer survivorship module, respondents were asked about cancer type; 1236 women who indicated breast cancer diagnosis were included in the study. The BRFSS code book can be accessed from https://www.cdc.gov/brfss/annual_data/2016/pdf/code-book16_llcp.pdf.

| Outcome variable: physical inactivity
Physical inactivity was based on a "no" response to the question "During the past month, other than your regular job, did you participate in any physical activities or exercises such as running, calisthenics, golf, gardening, or walking for exercise?" Those responding "yes" were considered physically active. Those with missing or invalid physical activity values were excluded (n = 4).

| Independent variables
We examined the relationship between sociodemographic factors, binge drinking, healthcare access, health history, and cancer survivorship-related variables, and physical inactivity. We chose these categories based on the previous physical activity studies in various cancer populations. [23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] The following sociodemographic variables were included: age (18-64, and ≥65 years), race/ethnicity (White, Black, and Other), education attainment (<high school and ≥high school), members in the household (1-2, 3-4 and >4), marital status (with partner vs without partner). We calculated the number of members in the household by adding the number of children and adults in a household (320 [25.6%] had missing data). For the marital status category, the "with partner" category included: "married" and "member of an unmarried couple," and the "without partner" included "Divorced," "Widowed," "Separated," and "Never married" survivors. The annual household income was categorized into <$25 000, $25 000 to <$50 000, ≥$50 000 (225 (18.1%) participants had missing income data), and the employment status was classified as employed, unemployed, unable to work, and other. Binge drinking was defined as women having ≥4 alcoholic drinks on one occasion in the past month. Healthcare access variables included: healthcare coverage based on the "Do you have any kind of healthcare coverage?" with responses of "yes or no," and "Could Not See Doctor Because of Cost" with responses of "yes or no.".
We were interested in the following comorbidities: body mass index (BMI), heart attack/coronary heart disease (CHD) or myocardial infarction (MI), and diabetes. BMI was calculated as weight divided by the square of height (kg/m 2 (underweight BMI <18.5 kg/m 2 , normal BMI = 18.5-25 kg/m 2 ), overweight (BMI = 25-29.9 kg/m 2 ), or obese (BMI ≥30 kg/ m 2 ) with 7.4% of the population have missing data (N = 92, Don't know/Refused/Missing). We were also interested in general health and physical and mental health status. General health was categorized as "excellent to good" and "fair to poor." Survey respondents were asked how many days during the past 30 days they experienced poor physical and mental health, with responses in number of days.
Cancer-related variables were based on the following questions: "Are you currently receiving treatment for cancer?" and "Do you currently have physical pain caused by your cancer or cancer treatment?" There was a moderate proportion of missing data (N = 285; 23%) for the "current pain" variable.

| Statistical analysis
After preliminary review of the data for completeness and accuracy, we summarized the characteristics of the population by physical activity "yes" and "no" status using descriptive statistics that included frequencies, percentages or means, and standard deviations, depending upon each variable's scale of measurement and distribution. To assess the bivariate relationships, continuous and categorical variables were analyzed using independent sample t-tests/Wilcoxon ranksum test and Fisher's exact/chi-squared tests, respectively. All significant variables (P < 0.05) in the bivariate analysis and race were included in the multivariable logistic regression models to evaluate the association [odds ratios (OR) and   [44][45][46] we specifically evaluated the association between pain and physical inactivity by using separate multivariable logistic regression models. All analyses were performed in SAS, version 9.3, and all P-values are 2-sided and P < 0.05 was considered statistically significant.

| RESULTS
The study sample included 1236 participants from the BRFSS 2016 cancer survivorship module dataset. A total of 376 (30.4%) breast cancer survivors were physically inactive. We compared the various factors of interest (sociodemographic, binge drinking, healthcare access, comorbidity, and pain) by physical inactivity status (Table 1). Physically inactive survivors were more likely to be older (≥65 years, P = 0.003), reported having ≤high school education (P < 0.0001), were without a partner (P < 0.0001), had lower annual household income (P < 0.0001) and were unable to work (P < 0.0001). Physically inactive survivors were over-represented at both ends of the BMI spectrum; that is, were more likely to be underweight as well as obese (P = 0.005), were more likely to have diabetes (P < 0.0001), CHD or MI (P = 0.005) and depressive disorders (P = 0.0004). Physically inactive survivors were more likely to report fair/poor general health (P < 0.0001), reported a larger number of days when physical health or mental health was not good (P < 0.0001, P = 0.0004, respectively). Finally, physically inactive cancer survivors were more likely to report cancer-related pain (P = 0.02). Factors associated with physical inactivity among breast cancer survivors are presented in Table 2, including both crude and the various models after adjusting for sociodemographic factors (Model 1), sociodemographic factors and binge drinking (Model 2), and sociodemographic factors, binge drinking and comorbidities (Model 3) with their estimated odds ratios and 95% CI. After adjusting for the respective variables in the table and accounting for BRFSS survey weights, we found that those ≥65 years old had 2.6-fold higher odds of being physically inactive compared to 18-64 year-olds (OR = 2.63 95% CI: 1.25-5.55), and underweight survivors had 6.1-fold higher odds of being physically inactive compared to normal weight survivors (OR = 6.11 95% CI: 1.35-27.66) ( Table 2). This association was positive in all the three models. Additionally, we found that lower education (≤HS) was associated with twofold greater odds and inability to work with threefold greater odds of physical inactivity in both Models 1 and 2. However, there was no evidence of these factors being associated with physical inactivity when adjusted for comorbidities in Model 3. There was a marginal association between physical inactivity and "days not feeling good" (OR = 1.04 95% CI: 1.00, 1.08).
To assess the relationship between physical inactivity and cancer-related pain in detail, we again conducted a series of analyses adjusting for sociodemographic factors (Model 1), sociodemographic factors and binge drinking (Model 2), and sociodemographic, binge drinking and comorbidities (Model 3) (

| DISCUSSION
Physical activity is recommended for cancer survivors to enhance their health and quality of life. Regular PA can increase recurrence-free survival rates of cancer patients. [6][7][8][9][10] About two-thirds of cancer survivors do not adhere to ACS exercise guidelines due to multiple sociodemographic, economic, health, and cancer-related factors that could prevent them from regularly engaging in physical activities. 11 In the present study, we examined the relationship between physical and psychosocial factors associated with physical inactivity in adult breast cancer survivors. Overall, we found that older age and being underweight are significant risk factors for physical inactivity in breast cancer survivors.
Our findings are consistent with previously published findings by Kampshoff et al 31 who utilized data from 574 female breast cancer survivors from three different lifestyle intervention studies in Australia and New Zealand, who reported older age as a significant barrier to physical activity. Kampshoff et al 31 also found that higher BMI and presence of comorbidities were associated with physical inactivity. We found associations between BMI and the comorbidities assessed in our crude analyses, but these factors did not remain significant after adjustment. Similar to our study, Kampshoff et al 31 did not show a significant association of PA with marital status and all treatment-related characteristics.
Although we did not find an association with high BMI, we did observe a negative effect of being underweight on physical activity. However, there were only nineteen underweight individuals and of these only five were physically inactive in our dataset. It has been reported that women who are underweight before breast cancer diagnosis are at the greatest risk of all-cause mortality, 47,48 possibly because being underweight could be a sign of poor general health or malnutrition which may affect the woman's ability to exercise due to poor muscle mass index. In the current study, we could not explore whether malnutrition was an issue as no nutrition data were available in the BRFSS dataset. Brunet et al 49 found that fatigue, pain and a lack of energy were major factors that breast cancer survivors perceived as barriers to performing physical activity. Likewise, Blaney et al 30 in their cross-sectional study also reported that pain and fatigue were among the top 10 factors that interfered with participation in physical activity. Although not significant after adjusting for all sociodemographic, behavioral, and health history factors, we found that cancer pain was associated with twofold higher odds of being physically inactive in Models 1 and 2, when adjusted for sociodemographic and binge drinking.
We did not find race/ethnicity to be a significantly associated with physical inactivity. This could potentially be attributed to small sample sizes for Black and Other race categories. Previous studies in which Black or other minority samples were sufficiently large, a significant association between race and physical activity was reported. Lu et al 50 also reported racial difference in physical activity in breast cancer survivors; Asian American women reported the lowest level of recreational physical activity, followed by Latinas. Hair et al 27 also found that Black women were less likely to meet national physical activity guidelines and reported lower levels of pre-diagnosis and post-diagnosis physical activity.
The findings in this report are subject to several limitations. First, BRFSS is a random-digit-dialed telephone survey where the information is reported directly by the respondent, so it may be subject to information based on social desirability, which could lead to inaccurate estimates of physical inactivity as well as information bias around the factors associated with PA. As BRFSS is a survey of community dwelling adults, it would not capture women with breast cancer who are in nursing homes and/or hospice care, whose physical activity may be limited based on their disease. The "Cancer Survivorship" module was included in BRFSS only in 2016 and is limited to thirteen questions. The "cancer" specific questions do not include diseaserelated variables such as cancer stage and type of cancer treatment. Also, years since diagnosis is not reported and could not be calculated, as exact "current age" is not recorded but categorized as 1-24, 25-34 etc years. Smoking behavior could not be included in the analysis due to significant missing values. In addition, the underweight category for the BMI variable had a small sample size and can affect the generalizability of the results. The results are from a cross-sectional study, and due to the fact that both risk factors and outcome(s) are measured simultaneously, causal inferences cannot be determined. However, there are strengths to our study. To our knowledge, this is one of the few studies to evaluate factors associated with physical inactivity in breast cancer survivors by sociodemographic factors, psychosocial factors, comorbidities, and pain from treatment or cancer at the population level. We have also used the most recent BRFSS dataset and cancer survivorship module to evaluate the barriers to physical activity in adult breast cancer survivors.

| CONCLUSIONS
In our population-based study, age (older) and BMI (underweight) are significant risk factors for physical inactivity among adult breast cancer survivors. Additionally, pain from cancer or treatment was significantly associated with physical inactivity. Providers should routinely screen patients for physical inactivity, provide recommendations for increasing activity and implement appropriate lifestyle interventions that could help breast cancer survivors adopt and maintain a healthy behavior to potentially reduce morbidity and mortality. Additional research is needed to understand why older and underweight survivors are at highest risk of inactivity so that appropriate tailored interventions can be developed.