Design, Sampling and Procedures
We conducted a cross-sectional study with a population-based sample of cancer survivors in Pennsylvania, home to over 700,000 cancer survivors and the fifth largest cancer survivor population in the U.S. [2]. The Pennsylvania Cancer Registry was used to identify eligible and representative cancer survivors who 1) were ≥20 years of age, 2) had received a breast, lung, colorectal, prostate, or gynecologic (cervical, endometrial, ovarian or uterine) cancer diagnosis between January 1, 2015 and December 31, 2016, and 3) were able to read, speak and write in English and respond to mailed surveys. A random sample of 2500 cancer survivors from 28 central Pennsylvania counties, 500 from each of five cancer sites, were mailed surveys. Of those 28 counties, 18 are categorized as urban/metropolitan and 10 as rural/nonmetropolitan using the 2013 Rural/Urban Continuum Codes (RUCC) [17]. Participants in urban and rural counties were sampled relative to population size, and participants whose registry data were missing age, race/ethnicity, or zip code were excluded from this study. A National Death Index review was conducted before each mailing to avoid sending the survey to the homes of deceased cancer survivors. Figure 1 shows the flow of participants throughout the study.
Eligible participants were mailed questionnaires based on questions included in the Behavioral Risk Factor Surveillance System (BRFSS) from April-July 2017. Participants were initially sent a study recruitment letter with a brief description of the study and were instructed to opt-in or opt-out of the study by contacting the study team by phone, mail or email. After two weeks, those who did not opt-out were mailed a consent note and questionnaires with instructions to return in a self-addressed stamped envelope (1st mailing). Four weeks later, those who had not returned completed questionnaires were re-sent materials (2nd mailing). Those who had not returned their questionnaires four weeks after the second mailing were sent an additional, identical third mailing. Participants who opted-out by phone, mail or email were no longer contacted. Participants who opted-in after the initial recruitment letter or returned completed questionnaires provided implied consent to participate in the study. All study procedures and materials were reviewed and approved by the Penn State College of Medicine’s Institutional Review Board, and all participants provided informed implied consent prior to participation.
Measures
Demographics and Rural-Urban Residence
Age, gender, cancer site and county of residence at diagnosis were extracted from the Pennsylvania Cancer Registry. Additional demographic information, including education and annual household income, was collected via self-report.
Rural-urban residence was based on county of residence at diagnosis, to which we applied the 2013 RUCC. The RUCC distinguish counties by their population size, degree of urbanization, and how adjacent they are to a metro area [17]. Each county in the U.S. is assigned one of nine codes, allowing for the assessment of trends related to population density and metropolitan influence. Urban counties are designated by codes 1-3 and include counties in metro areas with populations fewer than 250,000 to over one million. Nonmetro and rural counties are designated by codes 4-9 and include counties in nonmetro areas that may or may not be adjacent to a metro area with populations ranging from less than 2,500 to 20,000 or more [17].
[17]. Physical Activity
Self-reported leisure-time PA was assessed using questions based on the 2015 edition of the BRFSS [18]. Participants were asked if they participated in any PA or exercises during the past month. Those who responded ‘yes’ were asked to specify the type of activity, frequency per week or month, and duration. Participants were given the option to provide information for up to two activities or exercises and were asked to report how many times per week of month they performed muscle-strengthening activities [18]. Activities or exercises were coded and categorized as aerobic and moderate- or vigorous-intensity [19, 20]. Using the BRFSS scoring protocol [20], minutes per week of aerobic PA and times per week of muscle-strengthening activities were calculated. To maintain consistency with the 2008 Physical Activity Guidelines for Americans and the ACSM exercise guidelines for cancer survivors [12, 21], participants were categorized as meeting recommendations for 1) aerobic PA (≥150 minutes of moderate-intensity or ≥75 minutes of vigorous-intensity PA per week), 2) muscle-strengthening activities (≥2 times per week), 3) both aerobic and muscle-strengthening activities, or 4) neither aerobic nor muscle-strengthening activities.
Health Status
Health status was assessed using three items from the 2015 edition of the BRFSS [18]. Participants reported their general health status using a five-point Likert scale ranging from excellent to poor. Those who reported ‘Poor’ or ‘Fair’ were categorized as having poor health compared to those who reported their health as good, or reported their health as ‘Excellent,’ ‘Very Good,’ or ‘Good.’ Additionally, pParticipants were asked to report the number of days within the past 30 days that their physical and mental health were not good and the number of days that their poor physical and/or mental health kept them from their usual activities. Participants who reported <7 days on which they experienced poor physical or mental health or <7 days on which their poor physical/mental health impeded their usual activities (e.g., self-care, work, recreation) were categorized as having good health compared to those who reported ≥7 unhealthy days.
Statistical Analysis
Analyses were conducted in SPSS Version 24.0 (IBM SPSS Statistics, Armonk, NY), and statistical significance was set at p<.05. Independent samples t-tests and chi-square tests were used to assess rural-urban differences in demographics. We conducted a series of logistic regression analyses to assess rural-urban differences in meeting PA recommendations (neither vs. aerobic, muscle-strengthening, or both), adjusting for cancer type, gender, and income. Lastly, logistic regression models were used to explore the associations between meeting PA recommendations and dichotomized poor physical health days (≥7 days vs. <7 days), poor mental health days (≥7 days vs. <7 days), and poor physical or mental health days impeding activities (≥7 days vs. <7 days). Regression analyses included cancer type, gender, and income because they were associated with PA and rural-urban residence. Effect modification by rural-urban differences was assessed.