The BRFSS is a continuous telephone interview covering a variety of health-related topics, providing estimates of the civilian noninstitutionalized population. For the 2009 Missouri BRFSS, the Council of American Survey Research Organizations (CASRO) response rate was 57.4%. For the 2010 Missouri BRFSS, the CASRO response rate was 59.5% (13). Most questions were asked in both 2009 and 2010, but some were asked only in 2009 or 2010.

Participants were asked “Have you ever been told by a doctor or other health professional that you had cancer?” Follow-up questions included number of cancers and cancer type(s). Cancer survivors were compared with people who reported no cancer history (controls). Similar to the methods of other studies, we excluded participants who reported nonmelanoma skin cancer and who did not complete the cancer survivorship questions (14,15). The county of residence at the time of the interview was coded according to its location in a metropolitan statistical area (MSA) as defined by the US Office of Management and Budget. Counties were coded as urban if they were located in an MSA; otherwise, they were coded as rural.

Self-rated health was measured using the question, “How would you rate your health — would you say it is excellent, very good, good, fair, or poor?” Self-rated health was dichotomized into “fair or poor” versus “excellent, very good, or good.” Frequency of physical and mental distress was assessed by asking participants to report separately the number of days in the past month when their physical and mental health was “not good,” which was dichotomized into less than 14 days versus at least 14 days. Activity limitations were measured by asking participants to report the number of days in the past month when their poor physical or mental health kept them from doing their usual activities, which was dichotomized into less than 14 days versus at least 14 days. Participants rated their life satisfaction on a scale from very satisfied to very dissatisfied, which was used as a continuous variable. In 2010, a single question assessed the prevalence and frequency of fatigue, asking about how many days the participant felt tired or had little energy in the last 14 days, which was dichotomized into less than 7 days per 2 weeks versus at least 7 days per 2 weeks.

Respondents were asked about their physical activity with the question, “During the past month . . . did you participate in any physical activities such as running, calisthenics, golf, gardening, or walking for exercise?” During 2009, participants were asked more details about their physical activity, and we compared participants who met Centers for Disease Control and Prevention (CDC) physical activity recommendations with those who did not. Smoking status was classified as being a current smoker or a former or never smoker. To measure overuse of alcohol consistent with public health guidelines and previous research, we categorized alcohol intake as 0 to 1 drink per day versus more than 1 drink per day for women and 0 to 2 drinks per day versus more than 2 drinks per day for men (16). We compared participants who ate at least 5 servings of fruits or vegetables per day with those who did not. To assess the number of hours slept per night, the BRFSS asked, “On average, how many hours of sleep do you get in a 24-hour period?” Body mass index (BMI) was calculated based on self-reported height and weight. In 2009, participants also reported on aspirin use every day or every other day (17). In 2010, the Patient Health Questionnaire (PHQ-8) was used to assess the prevalence and severity of depression (18). In 2010, perceived emotional support was obtained from the question “How often do you get the social and emotional support you need?”

During 2010, participants were asked about aspects of a Survivorship Care Plan, including receiving 1) a written summary of the cancer treatments they received, and 2) advice about where they should go or where they should receive routine follow-up cancer checkups from a doctor, nurse, or other health professional. Participants also were asked whether their health insurance paid for all or part of their cancer treatment, denial of health or life insurance because of their cancer history, participation in a clinical trial as part of their cancer treatment, and current physical pain caused by cancer or its treatment.

On the basis of previous research, we selected the following factors to include as covariates in our analysis: 1) sociodemographic factors (age, race/ethnicity, sex, annual household income categories, employment status, marital status), 2) access to medical care (having health insurance, having a checkup within the past year, unable to see a doctor because of cost), and 3) number of chronic conditions (ie, diabetes, heart disease, stroke, asthma) (19–21). We used county-level data from the Area Resource File (number of surgeons per population, number of primary care physicians per population), Bureau of the Census (poverty rate), Department of Agriculture (number of fitness or recreation centers per population, percentage of people with low income and who live more than 1 mile from a supermarket or large grocery store if in an urban area or more than 10 miles from a supermarket or large grocery store if in a rural area), and average monthly dollar amount of the Supplemental Nutrition Assistance Program benefits per population.

We calculated the prevalence of the health status and behavioral risk factor measures by cancer history (cancer survivors vs noncancer controls) and rural–urban location. Model 1 is unadjusted; Model 2 adjusted for sociodemographic characteristics; Model 3 adjusted for sociodemographic characteristics and access to medical care; Model 4 adjusted for sociodemographic characteristics and chronic conditions; and Model 5 adjusted for sociodemographic characteristics and county-level data. We used multivariable logistic regression to compute adjusted proportions of the dependent variables, called predicted marginals, for each of the 4 groups; the predicted marginals can be interpreted as conditional prevalence estimates. An advantage of using predicted marginals over traditional odds ratios is that the former approach does not require the use of a reference group against which all other groups are compared. Pairwise least squares mean differences were also calculated and their significance tested using the z test. Variance estimates were calculated using the Taylor series approximation. The covariates were included as groups of variables in the logistic regression models to examine their effects on the predicted marginals. To obtain representative estimates, all data were weighted by adjusting for the probability of inclusion in the sample. We used SAS version 9.3 (SAS Institute, Inc, Cary, North Carolina) for data analysis.

The prevalence of fair or poor health was 38.5% among rural survivors compared with 27.4% among urban survivors and less than 20% among both control groups (Model 1, each P < .05) (Table 2). After controlling for sociodemographic characteristics (Model 2), prevalence of fair or poor health was similar for rural and urban survivors but was higher compared with rural and urban controls. Adding access to medical care (Model 3) did not affect the differences among the 4 groups. Adding the number of chronic conditions (Model 4) to sociodemographic characteristics (Model 2) reduced differences between rural survivors and rural controls, suggesting that differences in the number of chronic conditions explained the initial differences between both groups. There was no significant difference in frequent physical distress between rural and urban survivors (Model 1), but prevalence was higher when comparing rural survivors (29.1%) with rural controls (13.2%) and when comparing urban survivors (23.5%) with urban controls (9.1%). Only the difference between urban survivors and controls remained significant in fully adjusted models. The prevalence of frequent activity limitations was higher among rural survivors (22.8%) than among urban survivors (15.7%) and among rural controls (8.4%). Prevalence of frequent activity limitations also was higher when comparing urban survivors (15.7%) with urban controls (6.8%). After controlling for sociodemographic characteristics, differences remained when comparing rural and urban survivors with rural and urban controls, respectively. Access to medical care and number of chronic conditions did not further explain any differences. Rural survivors (48.8%) had a higher prevalence of fatigue than rural controls (27.9%), and urban survivors (36.1%) reported more fatigue than urban controls (26.5%). These differences persisted in adjusted models. Overall, life satisfaction was 1.6% on a 4-point scale, with no differences across the 4 groups. Adding the county-level variables (Model 5) to the model with sociodemographic characteristics (Model 2) did not change any of the results.

Although there was no significant difference in physical activity between rural (60.3%) and urban (64.0%) survivors, both groups were less likely to report physical activity compared with their no-cancer controls (Table 3). After controlling for covariates, the only difference in physical activity prevalence that remained significant was between urban survivors and urban controls. Although there were no significant differences between rural and urban survivors in meeting the CDC physical activity recommendation in 2009, rural survivors were less likely to meet these recommendations than were rural controls. This was the only significant difference among the 4 subgroups even when controlling for covariates.

Rural survivors (24.9%) were more likely to smoke than urban survivors (14.8%) but equally likely compared with rural controls. Urban survivors (14.8%) were less likely to smoke compared with urban controls (20.8%), but this difference was due to differences in sociodemographic characteristics (Model 2). Access to medical care and chronic conditions did not further alter the findings. There were no significant differences among the 4 groups for excessive alcohol use, fruit and vegetable consumption, hours of sleep, or overweight or obesity. Although there were no significant differences in aspirin use between rural and urban survivors, both survivor groups were more likely to take aspirin compared with rural and urban controls, respectively. Controlling for access to medical care or chronic conditions made all differences in aspirin use among the groups disappear. Adding the county-level variables (Model 5) to the model with sociodemographic characteristics (Model 2) did not change any of the results.

Although rural survivors had higher levels of depression (20.9%) than both urban survivors (10.5%) and rural controls (10.5%), there were no differences among the 4 groups when controlling for sociodemographic characteristics, access to medical care, or chronic conditions. The percentage who reported getting social and emotional support when needed did not vary significantly across the 4 subgroups.

Of cancer survivors, 26.4% reported receiving a written treatment summary; no significant difference was found between urban (28.9%) and rural (20.6%) participants (P = .48). Rural survivors (61.7%) were less likely to receive follow-up care instructions than urban (78.2%) survivors (P = .02). There were no significant differences in lack of insurance for cancer treatment between urban and rural survivors, denial of health or life insurance coverage because of cancer, clinical trial participation, or cancer-related physical pain.