Data for this analysis were collected from the baseline survey of the ACTS of Wellness Study, a CRC prevention trial conducted in urban African American churches in North Carolina and Michigan. Eligible individuals were African American church members, age 50 or older, who participated in the baseline survey. Churches were recruited from urban areas in two states: Flint, Michigan, and Wake, Durham, and Guilford counties in North Carolina. Churches were eligible to participate in the project if they had at least 50 members who were age 50 or older and had a coordinator who was willing to help recruit members. A total of 19 churches were enrolled in the study: 12 churches in North Carolina and 7 in Michigan. Prior to randomization to the intervention, participants completed a self-administered 100-item paper and pencil survey which included information on demographics, fruit and vegetable intake, cancer screening, physical activity, and health status.

Weight Group. Body Mass Index (BMI) was calculated using self-reported height and weight. Individuals were then categorized into weight groups based on their BMI: normal weight (BMI 18.5–24.94), overweight (BMI 24.95–29.94), obese I (BMI 29.95–34.94) and obese II+(34.95+).

Colorectal Cancer Screening behavior was ascertained using measures proposed and validated by Vernon et al. (10, 11). From this were created a variable indicating whether individuals had met average-risk guidelines for “Any CRC Screening” including stool blood test in the past year, colonoscopy in the past 10 years, flexible sigmoidoscopy in the past 5 years, double contrast barium enema in the past 5 years or virtual colonoscopy in the past 5 years (12). We also looked separately at use of colonoscopy (within the past 10 years) and stool blood test (within the past year).

Other screenings. We measured other cancer screening behaviors for men (PSA and DRE) and women (mammogram and pap smear) using two questions for each test. Test descriptions and questions were taken verbatim from the Healthy Information National Trends Survey (13). A woman was considered up-to-date for a given screening if she had a mammogram within the past year or a pap smear within the past two years. We considered a man with a PSA test or DRE within the past year to be up-to-date for those tests.

Other covariates. Continuous variables included age, co-morbidities, and self-reported fruit and vegetable intake (servings/day). The co-morbidities variable was created by summing all positive responses to the question “Have you been diagnosed with any of the following illnesses: high blood pressure, heart disease, diabetes, arthritis, or cancer?” Fruit and vegetable (F&V) intake was assessed with two validated measures. The first was a 10-item food frequency questionnaire assessing frequency over the last month, adapted from a 35-item measure for US Southern African Americans (14). We excluded the item on French fries/fried potatoes, leaving nine items for analysis. We also used a two-item measure to assess usual F&V intake (separate items for total fruits and total vegetables consumed “each day”) (14). Categorical variables included state (North Carolina or Michigan), gender, marital status (married or living with a partner, never married, divorced/separated, or widowed), education (11th grade or less, high school graduate/GED, trade/beauty/some college, college graduate, more than college), health insurance (yes/no), had polyps removed (yes/no), family history of colon cancer (yes/no), health status (excellent, very good, pretty good, fair/poor), income (<20,000, 20,000–49,999, 50,000–99,999, 100,000+, missing). Physical activity minutes per week were calculated based on seven activities (run/jog, bike, active sports, dance, swim, walk/hike, and aerobics or other moderate/vigorous activity). For each activity, participants indicated frequency (rarely or never, 1–3 times per month, 1–2 times per week, 3–4 times per week or 5 or more times per week) and duration of activity (less than 20 minutes or 20 minutes or more); these numbers were multiplied for each activity and the resulting minutes per week were summed across activities to create a final physical activity score (15). Participants were then dichotomized based on whether they were meeting the physical activity recommendation of 150 minutes per week of moderate or vigorous activity (16).

All analyses were completed using SAS version 9.2 survey procedures to account for randomization by church. We used chi-square and t-tests to assess relationships between weight group and both categorical and continuous variable, respectively. All tests were adjusted for clustering at church level. To justify stratifying the adjusted analyses by gender, we tested whether gender moderated the relationship between each CRC screening variable and weight. We created multivariate logistic regression models for each CRC screening outcome that included BMI group, gender, an interaction term (BMI group*gender) and selected covariates. For each model, we initially included any covariate which was associated with BMI group or screening (p<0.1). The final model retained covariates which were associated with the outcome (p>0.1) and did not change the estimate of the interaction term by 10% or more. If the interaction term remained significantly (p<0.05) associated with a given screening outcome in the final model, we conducted gender-stratified multivariate logistic regression analyses to assess relations between weight group and screening measures separately for men and women. If the interaction term was not significant, the interaction term was removed and the relationship between screening and weight groups was examined for men and women together.

In addition to the primary analyses using previously-defined definitions of colorectal cancer screening, we conducted analyses using alternative definitions of “up-to-date CRC testing.” These analyses accounted for the fact that more frequent screening is recommended for certain high-risk groups, such as those who previously had polyps. First we stratified the sample by polyp status: individuals who responded ‘yes’ to the question “have you ever had polyps or growths removed from your colon?” versus those who answered ‘no’ or ‘don’t know.’ For those without polyps, we created an up-to-date screening variable indicating whether or not they were meeting guidelines for average-risk individuals (as discussed above). For those with polyps, we created an up-to-date surveillance variable using self-reported data on when their doctor told them to return for follow-up. Possible answer choices were: every year, every two years, every five years, every 10 years, other, (s)he didn’t tell me when to return, and don’t know. Individuals were considered up-to-date for surveillance if they reported having a colonoscopy within the time-frame that the physician had indicated. If a person answered ‘other’, ‘(s)he didn’t tell me when to return’, or ‘don’t know’ they were considered up-to-date if they had a colonoscopy within the past 5 years. Lastly, we created a combined variable (up-to-date screening or surveillance) that included all individuals and indicated whether they were meeting the guidelines which applied to them: people without previous polyps were considered up-to-date if they were meeting screening guidelines while people with previous polyps were considered up-to-date if they were meeting surveillance guidelines. We repeated all adjusted and unadjusted analyses using the new variables.

Characteristics for the sample and for each weight group are shown in Table 1. There were statistically significant (p<0.05) differences by weight for gender, age, co-morbidities, health status, and physical activity. Obese individuals were younger, more likely to be female, had more co-morbidities, and were more likely to rate their health as poor/fair. We also found a relationship between weight group and physical activity (p=0.001). Only 25.3% of obese individuals reported engaging in at least 150 minutes of moderate or vigorous recreational physical activity, compared with 40.0% of normal-weight participants.

CRC Screening rates for the entire sample and by weight group are shown in table 1 (men and women combined), table 2 (men only) and table 3 (women only). Overall, women had slightly higher rates of CRC screening than men, but with one exception, these differences were not statistically significant. Among individuals with polyps, women were more likely to be meeting surveillance guidelines than men (p<0.03). Gender moderated the relationship between weight group and being up-to-date with any CRC screening (p=0.04); therefore, analyses were stratified for this variable. Bivariate analyses found no significant relationships between weight group and being up-to-date for CRC screening for men (Table 4) or women (Table 5). After adjusting for potential confounders, we found a significant association between weight and CRC screening for men (p=0.04). Men in the obese I group were significantly more likely to be up-to-date with screening (OR 2..40, 95%CI: 1..05–5.48) than normal-weight men. There were no significant differences in screening for women, but we saw a pattern in which obese women (I and II+) were less likely to be up-to-date with CRC screening than non-obese (normal or overweight) women. Gender did not moderate the relationship between weight group and CRC screening for the other variables (colonoscopy only, stool test only), nor was there a relationship between weight group and those variables.

Prostate cancer screening (PSA and DRE) rates are shown in Table 2, and odds ratios are shown in Table 4. We found a statistically significant relationship between PSA screening and weight group (p=0.05): men in the obese I group were more likely to report an up-to-date PSA test than normal-weight men (75.0% vs. 54.3%). This relationship persisted after adjusting for confounders (p=0.01); men in the obese I group were more likely to be up-to-date with prostate cancer screening than normal weight men (OR 4.25, 95%CI 1.63–11.12). After adjusting for confounders, we found a relationship between DRE screening and weight group (p=0.05); overweight and obese I men had higher rates of up-to-date DRE than normal-weight men.

Mammogram and Pap Smear rates for women are shown in Table 3, and odds ratios are shown in Table 5. After adjusting for confounders, we saw a statistically significant relationship between weight group and up-to-date mammogram (p=0.03). Overweight women were more likely than normal-weight to be up-to-date with mammogram recommendations (OR 1.47, 95%CI 1.00–2.88). A similar pattern was seen for pap smear, but that relationship was not statistically significant.

In order to differentiate between preventive CRC screening and polyp surveillance, we looked separately at individuals who reported ever having had polyps removed from their colon. Bivariate analyses did not find any statistically significant differences by weight group for compliance with physician recommendation for polyp surveillance or CRC screening among men (Table 2) or women (Table 3). Also, there was no statistically significant relationship between weight group and the combined “up-to-date screening or surveillance” variable for men and/or women. While the pattern by weight was similar as was seen for the “any CRC screening variable,” screening rates were lower across the board when individuals with polyps were held to a higher standard. Gender did not moderate the relationship between weight group and the alternative variables; thus we did not conduct gender-stratified adjusted models for these variables. Adjusted analyses with men and women together revealed no statistically significant association between any of the alternative screening variables and weight.