- Email: [email protected]
Relationship Between Body Mass Index and Prostate Cancer Screening in the United States
Oncology: Prostate/Testis/Penis/Urethra Relationship Between Body Mass Index and Prostate Cancer Screening in the United States Charles D. Scales, Jr., Lesley H. Curtis, Regina D. Norris, Kevin A. Schulman, Philipp Dahm and Judd W. Moul* From the Duke Prostate Center, Division of Urology, Department of Surgery (CDS, RDN, PD, JWM), and Department of Medicine and Center for Clinical and Genetic Economics, Duke Clinical Research Institute (LHC, KAS), Duke University Medical Center, Durham, North Carolina
Purpose: Obesity is associated with more advanced disease and worse outcomes in men with prostate cancer. To our knowledge the relationship between obesity and prostate cancer screening behavior in men 40 or older is unknown. Thus, we examined associations between body mass index and prostate cancer screening behavior. Materials and Methods: We used the 2002 Behavioral Risk Factor Surveillance System to study prostate cancer screening in a representative sample of 57,827 men 40 years or older. Primary outcomes were the proportion of men ever screened and the proportion screened in the last year for prostate cancer. Results: Obese men were more likely than normal weight men to have had a prostate specific antigen test (62.1% vs 56.1%, p ⬍0.001) and to have had a prostate specific antigen test in the last year (44.2% vs 38.2%, p ⬍0.001). After controlling for sociodemographic characteristics obese men remained more likely than normal weight men to have had a prostate specific antigen test (OR 1.46, 95% CI 1.33–1.61) and to have had a prostate specific antigen test in the last year (OR 1.42, 95% CI 1.30 –1.55). Respondents reporting an ongoing relationship with a physician (OR 2.88, 95% CI 2.57–3.22) and black nonHispanic men vs white men (OR 1.58, 95% CI 1.38 –1.81) were also more likely to have had a prostate specific antigen test in the last year. Conclusions: Obese men are more likely than normal weight men to be screened for prostate cancer. Associations between advanced stage, worse outcomes and obesity may not be explained by disparities in the screening of obese men for prostate cancer. Key Words: prostate, mass screening; obesity; prostate-specific antigen; prostatic neoplasms
recommend an individualized decision to screen, acknowledging the potential risks and benefits of PSA screening, and individual patient preferences.15 To our knowledge the relationship between obesity and prostate cancer screening behavior has not been explored in men 40 years or older. Thus, we used data from the 2002 BRFSS to examine the relationship between obesity and prostate cancer screening.
besity is a growing health concern in the United States and it influences many other health conditions, including cancer.1,2 Although there is conflicting evidence for an association between obesity and prostate cancer risk, a stronger link was established between obesity and biochemical recurrence, advanced stage and higher Gleason score.3– 8 Obesity may influence prostate cancer biology through increased insulin-like growth factor 1 and estrogenic compounds, and decreased sex hormone-binding globulin.9 –11 Obesity may also affect disease ascertainment in men with prostate cancer. For example, recent evidence suggests that lower PSA in obese and overweight men may mask prostate cancer.12 Alternatively obesity may influence cancer screening behavior, as shown for cervical, breast and colorectal cancers.13,14 Professional societies generally support prostate cancer screening in men 50 years or older who have a life expectancy of at least 10 years or in younger men who may be at increased risk for prostate cancer. They
O
MATERIALS AND METHODS Data were obtained from the 2002 BRFSS, an annual, population based telephone survey done by the Centers for Disease Control and Prevention. BRFSS queries a nationally representative sample of civilian noninstitutionalized adults residing in households in the United States regarding health care use and risk behaviors relating to 1 or more of the 10 leading causes of mortality. Using random digit dialing BRFSS identifies a probability sample of all households with telephones in each state, randomly selects 1 adult per household and interviews respondents about health behaviors.16 A total of 99,262 men completed the BRFSS survey in 2002. We excluded 34,743 men (35.0%) younger than 40 years from analysis because they were not asked questions about prostate cancer screening.16 We also excluded 2,524
Submitted for publication April 14, 2006. Study exempt from approval requirement according to the Duke University Health System institutional review board. * Correspondence: Division of Urology, Department of Surgery, Box 3707, Duke University Medical Center, Durham, North Carolina 27710 (telephone: 919-684-5057; FAX: 919-684-4611; e-mail: [email protected]).
0022-5347/07/1772-0493/0 THE JOURNAL OF UROLOGY® Copyright © 2007 by AMERICAN UROLOGICAL ASSOCIATION
493
Vol. 177, 493-498, February 2007 Printed in U.S.A. DOI:10.1016/j.juro.2006.09.059
494
BODY MASS INDEX AND PROSTATE CANCER SCREENING
participants who reported a history of prostate cancer, and 3,484 and 684 who did not respond to questions about prostate cancer screening, and to questions about height and weight used to determine BMI, respectively. Outcome Measures Prostate cancer screening. Men 40 years or older were asked a series of questions regarding prostate cancer screening (see Appendix). PSA testing was the primary outcome measure for analysis. Although the combination of serum PSA testing and DRE is the preferred screening method, the question structure did not address whether the PSA test and DRE occurred together (see Appendix). Because PSA is the more sensitive test for prostate cancer, we selected it as the primary outcome. Obesity. Respondents were classified by National Institutes of Health standards as normal weight (BMI less than 25.0 kg/m2), overweight (BMI 25.0 to 29.9 kg/m2) and obese (BMI 30 kg/m2 or greater). Obese patients were further divided into subgroups according to National Institutes of Health guidelines as class I—30.0 to 34.9, class II—35.0 to 39.9 and class III— 40.0 kg/m2 or greater. Other covariates. Participants were asked several questions related to sociodemographic characteristics. Participants were also asked about their general health and access to health care, including insurance status and whether they had a personal physician or health care provider. To control for potential frequent interactions with health care providers we included responses to the question “Have you ever been told by a doctor that you have diabetes?” States were grouped into census region according to United States Census Bureau classification. For purposes of analysis we combined some response categories. For education level we grouped responses of never attended, grades 1 to 8 and grades 9 to 11 into less than high school graduate. For marital status we classified men who were married or part of an unmarried couple as married and we classified men who were divorced, separated, never married or widowed as unmarried. We dichotomized the annual household income variable as less than $35,000 and $35,000 or greater. For employment status we classified students, homemakers, retirees and persons unable to work as not in the labor force. Statistical Analysis We used BRFSS sampling weights in all analyses. These weights account for demographic differences and the complex survey structure. The sampling weights also account for nonresponse and lack of coverage in segments of the population.16 Statistical analysis was performed using SAS® 9.1. The institutional review board of the Duke University Health System determined that the study was exempt from the requirement for approval. We compared weighted prostate cancer screening frequencies using the Rao-Scott chi-square test.17 We created a logistic regression model to explore prostate cancer screening behaviors based on categorical BMI (normal weight, overweight and obese) controlling for age, race/ethnicity, education level, marital status, annual household income, employment status, general health, insurance status, whether the respondent had a personal physician and con-
sidering the complex survey design. The incidence of missing responses (do not know/not sure or refused) for each variable was 1% to 7%. We retained respondents with missing data regarding race/ethnicity, income, education level, employment status, marital status, general health, insurance status, personal physician and diabetes mellitus, and we included dummy variables in the model to denote missing data. To examine the regression model sensitivity to the specification of the BMI variable we fit separate models with BMI as a linear variable and as a 4-level categorical variable (normal, overweight, obese [classes I and II] and extremely obese [class III]). Given the possibility of a nonlinear relationship between age and PSA testing, we treated age as a categorical variable (5-year increments) in the final model. We examined the sensitivity of findings to this specification by fitting a separate model with age as a linear variable. Results were considered statistically significant at 2-sided ␣ ⬍0.05.
RESULTS Table 1 shows the characteristics of the 57,827 respondents. Mean age was 55.4 years (95% CI 55.3–55.6) and 61% of respondents were 50 years or older. Of the respondents 48.0% were overweight and 24.9% were obese. Of the respondents 88.6% reported having health insurance and 82.3% reported having at least 1 personal physician. Obese respondents were younger than normal weight respondents (54.2 vs 56.7 years) and there was a sharp decrease in the proportion of obese men beyond age 69 years. Obese respondents were also much more likely than normal weight respondents to report a diagnosis of diabetes mellitus (43.8% vs 17.6%). Overall 59.9% of respondents (95% CI 59.2– 60.7) reported ever having a PSA test and 42.0% (95% CI 41.2– 42.7) reported having a PSA test in the last year. Of the men 74.6% reported ever having DRE (95% CI 74.0 –75.3). The proportion of respondents reporting DRE in the last year (42.7%, 95% CI 42.0 – 43.4) was similar to that for PSA testing in the last year. Table 2 shows screening rates by BMI category. Obese men were more likely than normal weight men to have ever had a PSA test (62.1% vs 56.1%, p ⬍0.001) and to have had a PSA test in the last year (44.2% vs 38.2%, p ⬍0.001). Similarly obese men were more likely than normal weight men to report having DRE. Overweight men were also more likely than normal weight men to have had PSA or DRE. The increased screening propensity in obese men persisted after controlling for sociodemographic characteristics and access to care. The odds for obese men ever having a PSA test were 1.37 times higher than the odds for normal weight men (95% CI 1.24 –1.51, table 3). Obese men were also more likely to have had a PSA test in the last year (OR 1.32, 95% CI 1.21–1.45) and more likely to report having DRE. Black nonHispanic men were more likely to undergo PSA testing, as were men from higher income households. Access to care had a strong association with the likelihood of prostate cancer screening. The odds of men having a PSA test in the last year were almost
BODY MASS INDEX AND PROSTATE CANCER SCREENING
495
TABLE 1. Respondent characteristics* Wtd % (95% CI)* Characteristic
No. Pts
Normal Wt
Mean ⫾ SE age: 40–49 50–59 60–69 70–79 80 or Older BMI Race: White Black Other Multiracial Hispanic No response Annual household income ($): Less than 35,000 35,000 or Greater Refused Do not know Education level: Less than high school graduate High school graduate 1–3 yrs college 4 or Greater yrs college No response Employment status: Employed Self-employed Unemployed Nonlabor force No response Marital status: Married Not married No response General health: Excellent Very good Good Fair Poor No response Health insurance coverage: Yes No No response Personal physician: Yes (at least 1) No No response Diabetes mellitus: No Yes No response Census region: Northeast Midwest South West
57,827 20,123 16,864 11,130 7,159 2,551 57,827
56.7 37.2 25.8 17.4 13.1 6.5 27.1
(0.17) (35.8–38.5) (24.5–27.0) (16.4–18.5) (12.3–14.0) (5.9–7.1) (26.4–27.7)
55.3 37.7 29.3 18.4 11.5 3.2 48.0
(0.11) (36.7–38.7) (28.3–30.2) (17.6–19.2) (10.8–12.1) (2.9–3.5) (47.3–48.7)
54.2 39.9 31.3 18.6 8.9 1.3 24.9
(0.15) (38.5–41.4) (29.9–32.6) (17.6–19.7) (8.1–9.6) (1.0–1.5) (24.3–25.5)
47,037 3,420 2,557 976 3,194 643
75.1 7.3 7.0 1.3 8.1 1.2
(73.6–76.6) (6.5–8.2) (6.0–8.0) (1.0–1.6) (7.0–9.1) (0.9–1.5)
76.7 7.3 4.2 1.3 9.6 0.9
(75.6–77.7) (6.7–7.9) (3.7–4.8) (1.1–1.6) (8.8–10.5) (0.7–1.0)
74.7 9.7 2.9 1.6 10.3 0.8
(73.2–76.2) (8.8–10.6) (2.4–3.4) (1.2–1.9) (9.0–11.5) (0.6–1.0)
19,890 31,751 1,823 4,363
36.3 52.9 3.3 7.5
(34.9–37.6) (51.6–54.4) (2.7–3.9) (6.8–8.1)
29.3 60.7 2.6 7.3
(28.4–30.3) (59.7–61.7) (2.3–2.9) (6.8–7.9)
34.9 56.2 3.0 5.9
(33.5–36.3) (54.8–57.7) (2.1–3.9) (5.2–6.5)
6,581 17,477 13,682 19,994 93
12.3 (11.4–13.3) 27.5 (26.3–28.8) 22.8 (21.6–24.0) 37.2 (35.9–38.5) 0.16 (0.07–0.25)
11.0 (10.3–11.8) 28.4 (27.5–29.2) 23.9 (23.0–24.8) 36.6 (35.6–37.6) 0.14 (0.06–0.21)
13.8 (12.5–15.0) 32.3 (31.0–33.7) 25.1 (23.9–26.4) 28.6 (27.3–29.9) 0.09 (0.03–0.15)
27,534 8,547 2,184 19,495 67
46.4 (45.0–47.8) 14.2 (13.2–15.2) 4.6 (3.9–5.2) 34.7 (33.4–36.0) 0.13 (0.07–0.20)
51.9 (50.9–53.0) 14.2 (13.5–14.9) 4.1 (3.6–4.6) 29.6 (28.7–30.5) 0.11 (0.05–0.17)
51.9 (50.5–53.4) 11.7 (10.8–12.6) 4.8 (4.1–5.4) 31.5 (30.0–32.9) 0.11 (0.05–0.16)
38,758 18,975 94
70.5 (69.3–71.7) 29.3 (28.1–30.5) 0.17 (0.02–0.31)
77.1 (76.3–77.9) 22.7 (21.9–23.4) 0.21 (0.06–0.36)
77.0 (75.8–78.1) 22.9 (21.7–24.1) 0.10 (0.04–0.15)
11,283 17,766 17,738 7,408 3,448 184
23.9 30.4 27.1 11.7 6.6 0.3
21.3 32.4 30.1 11.4 4.4 0.4
11.2 28.3 35.6 17.2 7.4 0.3
51,577 6,182 68
87.2 (86.2–88.2) 12.7 (11.7–13.7) 0.1 (0.03–0.15)
89.3 (88.6–90.0) 10.5 (9.8–11.2) 0.1 (0.07–0.20)
88.8 (87.8–89.8) 11.1 (10.1–12.1) 0.1 (0.03–0.12)
48,211 9,502 114
79.6 (78.4–80.9) 20.1 (18.9–21.4) 0.2 (0.1–0.4)
82.5 (81.6–83.4) 17.3 (16.4–18.2) 0.2 (0.1–0.3)
84.9 (83.7–86.2) 14.8 (13.6–16.0) 0.2 (0.1–0.4)
51,516 6,200 111
92.7 (91.9–93.5) 7.1 (6.3–7.9) 0.2 (0.1–0.3)
91.0 (90.4–91.6) 8.8 (8.2–9.4) 0.1 (0.1–0.2)
80.6 (79.4–81.8) 19.2 (18.1–20.4) 0.1 (0.1–0.2)
12,273 12,960 12,964 13,197
19.6 21.1 24.7 23.4
19.1 23.0 24.1 22.2
17.6 24.2 25.9 19.3
(22.7–25.0) (29.1–21.7) (25.8–28.3) (10.8–12.6) (5.9–7.3) (0.2–0.4)
(18.5–20.8) (20.2–22.1) (23.6–25.8) (22.0–24.9)
Overwt
(20.5–22.2) (31.5–33.4) (29.1–31.0) (10.7–12.1) (3.9–4.8) (0.2–0.6)
(18.3–20.0) (22.2–23.7) (23.2–24.9) (21.1–23.3)
Obese
(10.3–12.1) (26.9–29.7) (34.2–37.0) (16.1–18.4) (6.6–8.1) (0.2–0.3)
(16.5–18.7) (23.1–25.3) (24.7–27.1) (17.9–20.7)
* Male population in the United States (percents may not sum to 100 due to rounding).
3 times higher if the respondent reported having at least 1 personal physician (OR 2.95, 95% CI 2.63–3.32). There was no association between diabetes mellitus and PSA testing. Findings were not sensitive to the manner in which BMI was classified. In the categorical model obese men (OR 1.38, 95% CI 1.25–1.53) and extremely obese men (OR 1.22, 95% CI 0.94 –1.59) were more likely than normal weight men to report ever having a PSA test. Obese men were also more likely to have had a PSA test in the last year (OR 1.34, 95% CI 1.22–1.48). In the linear model each point increase in BMI increased the odds of ever having a PSA test (OR 1.02,
95% CI 1.01–1.03) and having a PSA test in the last year (OR 1.02, 95% CI 1.01–1.02). Findings were similar for DRE. As in the primary regression, influential covariates in these analyses included race/ethnicity, income, education level and access to care. Findings also were not sensitive to the manner in which age was classified. In the linear model obese men (OR 1.46, 95% CI 1.33–1.61) and overweight men (OR 1.28, 95% CI 1.18 –1.38) were more likely to undergo PSA testing than normal weight men after adjusting for all other covariates. Findings were similar in the 2 models when only men 50 years or older were included.
496
BODY MASS INDEX AND PROSTATE CANCER SCREENING TABLE 2. Screening rates by BMI
Screening Test PSA: Last yr Ever DRE: Last yr Ever
Wtd % (95% CI)* Normal Wt
Overwt
Obese
38.2 (36.8–39.5) 56.1 (54.7–57.5)
42.9 (41.9–44.0) 60.9 (59.9–61.9)
44.2 (42.8–45.7) 62.1 (60.7–63.6)
40.2 (38.9–41.5) 72.1 (70.3–73.0)
43.9 (42.9–45.0) 75.9 (75.0–76.9)
43.2 (41.7–44.6) 75.5 (74.1–76.9)
p ⬍0.001. * Male population in the United States.
these gaps have been closing recently and higher screening rates for cervical and prostate cancer have been documented in black patients.19 The higher rates may reflect patient and practitioner knowledge about the risk of prostate cancer in black men. Health insurance coverage and an ongoing rela-
TABLE 3. Odds of undergoing PSA test by respondent characteristic OR PSA Test (95% CI) Characteristic
DISCUSSION Our findings suggest that obese men are more likely than normal weight men to be screened for prostate cancer. Obese and overweight men are more likely than normal weight men to have ever had a PSA test and to have had a PSA test in the last year. These differences were not explained by respondent sociodemographic characteristics or access to care. Findings also were not sensitive to the manner in which BMI was classified since linear and categorical BMI demonstrated the association between higher BMI and prostate cancer screening. Thus, differences in prostate cancer screening behavior may not explain prostate cancer outcomes in obese men. This study extends a previous analysis of older BRFSS data, which showed a positive association between obesity and PSA testing in men older than 50 years.18 Our analysis included men 40 to 49 years old, which includes men at higher risk of prostate cancer who may elect earlier screening, such as black American men, and it provides a real world view of prostate cancer screening. In addition, the associations were not sensitive to different classifications of age or BMI. The model includes important covariates, such as a personal relationship with a physician, marital status and geographic location. Finally, the inclusion of dummy variables for missing responses (do not know or refused) accounts for potential associations between item nonresponse and screening behavior. We found a strong positive association between obesity and prostate cancer screening, although other studies documented negative associations between obesity and cancer screening.13,14 For example, Rosen and Schneider found that BRFSS respondents with BMI more than 35 kg/m2 were less likely than normal weight respondents to undergo screening for colorectal cancer.14 After controlling for sociodemographic variables this difference was entirely attributable to differences in BMI in women. Using data from the National Health Interview Survey Wee et al found that overweight and obese women were less likely than normal weight women to be screened for cervical and breast cancer.13 These differences may be attributable to the simplicity of serum PSA testing and DRE or to public perceptions of the importance of prostate cancer screening. Several other results of our analysis are noteworthy. Certain sociodemographic and access variables were associated with increased odds of prostate cancer screening. For example, higher income and education levels were associated with higher screening rates, as noted for other cancers.19 Black nonHispanic men were more likely to undergo PSA testing. Although racial/ethnic disparities in cancer screening exist,
Age: 40–45 46–50 51–55 56–60 61–65 66–70 71–75 76–80 81–85 Older than 85 BMI: Normal (less than 25.0 kg/m2) Overwt (25.0–29.9 kg/m2) Obese (30.0 kg/m2 or greater) Race: White nonHispanic Black nonHispanic Other nonHispanic Multiracial nonHispanic Hispanic No response Annual household income ($): Less than 35,000 35,000 or Greater Refused Do not know Education level: Less than high school graduate High school graduate Some college College graduate No response Employment status: Employed Self-employed Not employed Not in labor force No response Marital status: Not married Married No response General health: Excellent Very good Good Fair Poor No response Health insurance coverage: No Yes No response Personal physician: No Yes (at least 1) No response Diabetes mellitus: No Yes No response Census region: Northeast Midwest South West United States territory
Ever
Last Yr
1.00 2.20 (1.99–2.43) 4.13 (3.70–4.60) 6.82 (6.02–7.73) 7.63 (6.53–8.91) 9.92 (8.38–11.8) 11.9 (9.79–14.6) 10.6 (8.46–13.3) 7.21 (5.61–9.29) 5.95 (4.29–8.25)
1.00 1.97 (1.76–2.20) 3.53 (3.16–3.95) 4.87 (4.32–5.50) 5.57 (4.84–6.41) 6.87 (5.92–7.98) 8.39 (7.11–9.88) 6.46 (5.40–7.73) 4.45 (3.58–5.52) 3.87 (2.87–5.20)
1.00 1.23 (1.13–1.33) 1.37 (1.24–1.51)
1.00 1.20 (1.12–1.30) 1.32 (1.21–1.45)
1.00 1.77 (1.51–2.07) 0.62 (0.50–0.76) 1.00 (0.75–1.28) 1.19 (1.01–1.41) 0.96 (0.73–1.28)
1.00 1.58 (1.37–1.83) 0.75 (0.61–0.93) 1.02 (0.75–1.40) 1.30 (1.10–1.53) 1.29 (0.98–1.70)
1.00 1.57 (1.43–1.73) 1.50 (1.31–1.72) 0.94 (0.75–1.17)
1.00 1.47 (1.35–1.60) 1.49 (1.31–1.70) 0.92 (0.76–1.11)
1.00 1.51 (1.32–1.72) 1.87 (1.61–2.16) 2.12 (1.84–2.45) 1.32 (0.64–2.74)
1.00 1.38 (1.21–1.57) 1.55 (1.35–1.77) 1.73 (1.52–1.98) 1.05 (0.50–2.20)
1.00 1.07 (0.96–1.18) 1.02 (0.86–1.22) 1.38 (1.22–1.55) 1.40 (0.67–2.95)
1.00 0.98 (0.88–1.08) 0.94 (0.79–1.13) 1.24 (1.12–1.38) 1.18 (0.60–2.30)
1.00 1.18 (1.09–1.28) 1.44 (0.60–3.45)
1.00 1.10 (1.02–1.18) 2.27 (0.90–5.71)
1.00 0.94 (0.86–1.04) 0.98 (0.89–1.09) 1.02 (0.89–1.17) 0.90 (0.75–1.07) 0.65 (0.37–1.15)
1.00 0.93 (0.85–1.02) 1.06 (0.96–1.16) 0.97 (0.86–1.10) 0.91 (0.77–1.07) 0.90 (0.54–1.51)
1.00 1.52 (1.33–1.72) 0.83 (0.36–1.96)
1.00 1.55 (1.35–1.78) 0.77 (0.29–2.06)
1.00 2.60 (2.35–2.89) 0.83 (0.36–1.96)
1.00 2.95 (2.63–3.31) 1.59 (0.63–3.98)
1.00 1.12 (0.98–1.28) 1.20 (0.70–2.05)
1.00 1.07 (0.95–1.20) 0.94 (0.53–1.54)
1.00 0.93 (0.85–1.03) 1.20 (1.09–1.33) 1.00 (0.89–1.13) 1.25 (1.10–1.42)
1.00 0.95 (0.87–1.04) 1.20 (1.10–1.32) 0.95 (0.85–1.07) 1.14 (1.01–1.28)
BODY MASS INDEX AND PROSTATE CANCER SCREENING tionship with a physician were also strongly associated with prostate cancer screening. The reason for an association between obesity and prostate cancer screening is unclear. The analysis controlled for well-known sociodemographic confounders and access to care. Inclusion of a variable for diabetes mellitus was an attempt to control for more frequent interactions with physicians due to chronic health conditions. It may be that overweight and obese men have more interactions with physicians due to conditions other than diabetes and, therefore, they have more opportunities for screening. Alternatively providers may believe that obese men are a higher risk group. However, it is noteworthy that associations between worse outcomes, advanced stage, higher grade and obesity were published after the 2002 BRFSS survey was completed, and so they are unlikely to have influenced our findings. Our study has some limitations. The analysis is based on patient self-report, which may not correspond with actual screening behavior. Given the high prevalence of PSA testing in this survey, ORs may magnify small absolute differences in screening rates. However, even a small absolute increase in the screening rate in obese men supports the conclusion that differences in prostate cancer outcomes may not be explained by differential screening. The PSA test may not be exclusively used for prostate cancer screening, so that PSA testing rates may overestimate actual cancer screening rates. It is unclear how this source of error would affect our findings. However, respondents were asked about PSA testing in the context of prostate cancer screening, so that we anticipate small effects, if any, from this limitation. Finally, clinical information regarding PSA and subsequent prostate biopsies is lacking, and so we were unable to explore whether differential biopsy rates may affect outcomes in obese men. These limitations notwithstanding, we believe that the findings are robust. They are buttressed by several design strengths. BRFSS is designed to ascertain disease risk behaviors in a large, nationally representative sample. The prevalence of obesity in our sample is similar to estimates from other sources.3 The association between BMI and prostate cancer screening persisted regardless of whether the classification of BMI was categorical or linear. Finally, the model controlled for important sociodemographic characteristics, such as race/ethnicity, income, education level and access to care. CONCLUSIONS To our knowledge our study is the first investigation of the association between obesity and prostate cancer screening behavior in men 40 years or older. Higher screening rates in obese men suggest that disparities in prostate cancer outcomes in obese men do not arise from differences in screening behavior but they may be related to differences in tumor biology, performance characteristics of the PSA test in obese men or differential prostate biopsy rates after an elevated PSA test. Further investigation is required to confirm our findings and explain differences in prostate cancer outcomes in obese men. ACKNOWLEDGMENTS Damon Seils assisted with the manuscript.
497
APPENDIX Survey Questions From the Prostate Cancer Screening Module of the 2002 Behavioral Risk Factor Surveillance System 1. A Prostate Specific Antigen Test, also called a PSA test, is a blood test used to check men for prostate cancer. Have you ever had a PSA test? Yes No Do not know/not sure Refused 2. How long has it been since you had your last PSA test? Within the past year (less than 12 months ago) Within the past 2 years (greater than 1 year but less than 2 years) Within the past 3 years (greater than 2 years but less than 3 years) Within the past 5 years (greater than 3 years but less than 5 years) 5 or More years ago Do not know/not sure Refused 3. A digital rectal examination is an examination in which a doctor, nurse, or other health professional places a gloved finger into the rectum to feel the size, shape, and hardness of the prostate gland. Have you ever had a digital rectal examination? Yes No Do not know/not sure Refused 4. How long has it been since your last digital rectal examination? Within the past year (less than 12 months ago) Within the past 2 years (greater than 1 year but less than 2 years) Within the past 3 years (greater than 2 years but less than 3 years) Within the past 5 years (greater than 3 years but less than 5 years) 5 or More years ago Do not know/not sure Refused 5. Have you ever been told by a doctor, nurse, or other health professional that you had prostate cancer? Yes No Do not know/not sure Refused
Abbreviations and Acronyms BMI ⫽ body mass index BRFSS ⫽ Behavioral Risk Factor Surveillance System DRE ⫽ digital rectal examination PSA ⫽ prostate specific antigen
REFERENCES 1.
Allison DB, Fontaine KR, Manson JE, Stevens J and VanItallie TB: Annual deaths attributable to obesity in the United States. JAMA 1999; 282: 1530. 2. Calle EE, Rodriguez C, Walker-Thurmond K and Thun MJ: Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003; 348: 1625. 3. Amling CL, Kane CJ, Riffenburgh RH, Ward JF, Roberts JL, Lance RS et al: Relationship between obesity and race in predicting adverse pathologic variables in patients undergoing radical prostatectomy. Urology 2001; 58: 723. 4. Freedland SJ, Aronson WJ, Kane CJ, Presti JC Jr, Amling CL, Elashoff D et al: Impact of obesity on biochemical control after radical prostatectomy for clinically localized prostate cancer: a report by the Shared Equal Access Regional Cancer Hospital database study group. J Clin Oncol 2004; 22: 446. 5. Rohrmann S, Roberts WW, Walsh PC and Platz EA: Family history of prostate cancer and obesity in relation to highgrade disease and extraprostatic extension in young men with prostate cancer. Prostate 2003; 55: 140.
498 6.
7.
8.
9.
10.
11. 12.
BODY MASS INDEX AND PROSTATE CANCER SCREENING Whittemore AS, Kolonel LN, Wu AH, John EM, Gallagher RP, Howe GR et al: Prostate cancer in relation to diet, physical activity, and body size in blacks, whites, and Asians in the United States and Canada. J Natl Cancer Inst 1995; 87: 652. Amling CL, Riffenburgh RH, Sun L, Moul JW, Lance RS, Kusuda L et al: Pathologic variables and recurrence rates as related to obesity and race in men with prostate cancer undergoing radical prostatectomy. J Clin Oncol 2004; 22: 439. Kane CJ, Bassett WW, Sadetsky N, Silva S, Wallace K, Pasta DJ et al: Obesity and prostate cancer clinical risk factors at presentation: data from CaPSURE. J Urol 2005; 173: 732. Wolk A, Mantzoros CS, Andersson SO, Bergstrom R, Signorello LB, Lagiou P et al: Insulin-like growth factor 1 and prostate cancer risk: a population-based, case-control study. J Natl Cancer Inst 1998; 90: 911. Moyad MA: Is obesity a risk factor for prostate cancer, and does it even matter? A hypothesis and different perspective. Urology, suppl., 2002; 59: 41. Yip I, Heber D and Aronson W: Nutrition and prostate cancer. Urol Clin North Am 1999; 26: 403. Baillargeon J, Pollock BH, Kristal AR, Bradshaw P, Hernandez J, Basler J et al: The association of body mass index and prostate-specific antigen in a population-based study. Cancer 2005; 103: 1092.
13.
14.
15. 16.
17.
18.
19.
Wee CC, McCarthy EP, Davis RB and Phillips RS: Screening for cervical and breast cancer: is obesity an unrecognized barrier to preventive care? Ann Intern Med 2000; 132: 697. Rosen AB and Schneider EC: Colorectal cancer screening disparities related to obesity and gender. J Gen Intern Med 2004; 19: 332. Prostate-specific antigen (PSA) best practice policy. American Urological Association. Oncology (Huntingt), 2000; 14: 267. Behavioral Risk Factor Surveillance System: Codebook 2002. Atlanta: United States Department of Health and Human Services, Centers for Disease Control and Prevention. Available at http://www.cdc.gov/brfss/technical_infodata/ surveydata/codebook_02.rtf. Accessed May 3, 2005. Rao JNK and Scott AJ: The analysis of categorical data from complex sample surveys: chi-squared tests for goodness of fit and independence in two-way tables. J Am Stat Assoc 1981; 76: 221. Sirovich BE, Schwartz LM and Woloshin S: Screening men for prostate and colorectal cancer in the United States: does practice reflect the evidence? JAMA 2003; 289: 1414. Breen N, Wagener DK, Brown ML, Davis WW and BallardBarbash R: Progress in cancer screening over a decade: results of cancer screening from the 1987, 1992, and 1998 National Health Interview Surveys. J Natl Cancer Inst 2001; 93: 1704.
Purpose: Obesity is associated with more advanced disease and worse outcomes in men with prostate cancer. To our knowledge the relationship between obesity and prostate cancer screening behavior in men 40 or older is unknown. Thus, we examined associations between body mass index and prostate cancer screening behavior. Materials and Methods: We used the 2002 Behavioral Risk Factor Surveillance System to study prostate cancer screening in a representative sample of 57,827 men 40 years or older. Primary outcomes were the proportion of men ever screened and the proportion screened in the last year for prostate cancer. Results: Obese men were more likely than normal weight men to have had a prostate specific antigen test (62.1% vs 56.1%, p ⬍0.001) and to have had a prostate specific antigen test in the last year (44.2% vs 38.2%, p ⬍0.001). After controlling for sociodemographic characteristics obese men remained more likely than normal weight men to have had a prostate specific antigen test (OR 1.46, 95% CI 1.33–1.61) and to have had a prostate specific antigen test in the last year (OR 1.42, 95% CI 1.30 –1.55). Respondents reporting an ongoing relationship with a physician (OR 2.88, 95% CI 2.57–3.22) and black nonHispanic men vs white men (OR 1.58, 95% CI 1.38 –1.81) were also more likely to have had a prostate specific antigen test in the last year. Conclusions: Obese men are more likely than normal weight men to be screened for prostate cancer. Associations between advanced stage, worse outcomes and obesity may not be explained by disparities in the screening of obese men for prostate cancer. Key Words: prostate, mass screening; obesity; prostate-specific antigen; prostatic neoplasms
recommend an individualized decision to screen, acknowledging the potential risks and benefits of PSA screening, and individual patient preferences.15 To our knowledge the relationship between obesity and prostate cancer screening behavior has not been explored in men 40 years or older. Thus, we used data from the 2002 BRFSS to examine the relationship between obesity and prostate cancer screening.
besity is a growing health concern in the United States and it influences many other health conditions, including cancer.1,2 Although there is conflicting evidence for an association between obesity and prostate cancer risk, a stronger link was established between obesity and biochemical recurrence, advanced stage and higher Gleason score.3– 8 Obesity may influence prostate cancer biology through increased insulin-like growth factor 1 and estrogenic compounds, and decreased sex hormone-binding globulin.9 –11 Obesity may also affect disease ascertainment in men with prostate cancer. For example, recent evidence suggests that lower PSA in obese and overweight men may mask prostate cancer.12 Alternatively obesity may influence cancer screening behavior, as shown for cervical, breast and colorectal cancers.13,14 Professional societies generally support prostate cancer screening in men 50 years or older who have a life expectancy of at least 10 years or in younger men who may be at increased risk for prostate cancer. They
O
MATERIALS AND METHODS Data were obtained from the 2002 BRFSS, an annual, population based telephone survey done by the Centers for Disease Control and Prevention. BRFSS queries a nationally representative sample of civilian noninstitutionalized adults residing in households in the United States regarding health care use and risk behaviors relating to 1 or more of the 10 leading causes of mortality. Using random digit dialing BRFSS identifies a probability sample of all households with telephones in each state, randomly selects 1 adult per household and interviews respondents about health behaviors.16 A total of 99,262 men completed the BRFSS survey in 2002. We excluded 34,743 men (35.0%) younger than 40 years from analysis because they were not asked questions about prostate cancer screening.16 We also excluded 2,524
Submitted for publication April 14, 2006. Study exempt from approval requirement according to the Duke University Health System institutional review board. * Correspondence: Division of Urology, Department of Surgery, Box 3707, Duke University Medical Center, Durham, North Carolina 27710 (telephone: 919-684-5057; FAX: 919-684-4611; e-mail: [email protected]).
0022-5347/07/1772-0493/0 THE JOURNAL OF UROLOGY® Copyright © 2007 by AMERICAN UROLOGICAL ASSOCIATION
493
Vol. 177, 493-498, February 2007 Printed in U.S.A. DOI:10.1016/j.juro.2006.09.059
494
BODY MASS INDEX AND PROSTATE CANCER SCREENING
participants who reported a history of prostate cancer, and 3,484 and 684 who did not respond to questions about prostate cancer screening, and to questions about height and weight used to determine BMI, respectively. Outcome Measures Prostate cancer screening. Men 40 years or older were asked a series of questions regarding prostate cancer screening (see Appendix). PSA testing was the primary outcome measure for analysis. Although the combination of serum PSA testing and DRE is the preferred screening method, the question structure did not address whether the PSA test and DRE occurred together (see Appendix). Because PSA is the more sensitive test for prostate cancer, we selected it as the primary outcome. Obesity. Respondents were classified by National Institutes of Health standards as normal weight (BMI less than 25.0 kg/m2), overweight (BMI 25.0 to 29.9 kg/m2) and obese (BMI 30 kg/m2 or greater). Obese patients were further divided into subgroups according to National Institutes of Health guidelines as class I—30.0 to 34.9, class II—35.0 to 39.9 and class III— 40.0 kg/m2 or greater. Other covariates. Participants were asked several questions related to sociodemographic characteristics. Participants were also asked about their general health and access to health care, including insurance status and whether they had a personal physician or health care provider. To control for potential frequent interactions with health care providers we included responses to the question “Have you ever been told by a doctor that you have diabetes?” States were grouped into census region according to United States Census Bureau classification. For purposes of analysis we combined some response categories. For education level we grouped responses of never attended, grades 1 to 8 and grades 9 to 11 into less than high school graduate. For marital status we classified men who were married or part of an unmarried couple as married and we classified men who were divorced, separated, never married or widowed as unmarried. We dichotomized the annual household income variable as less than $35,000 and $35,000 or greater. For employment status we classified students, homemakers, retirees and persons unable to work as not in the labor force. Statistical Analysis We used BRFSS sampling weights in all analyses. These weights account for demographic differences and the complex survey structure. The sampling weights also account for nonresponse and lack of coverage in segments of the population.16 Statistical analysis was performed using SAS® 9.1. The institutional review board of the Duke University Health System determined that the study was exempt from the requirement for approval. We compared weighted prostate cancer screening frequencies using the Rao-Scott chi-square test.17 We created a logistic regression model to explore prostate cancer screening behaviors based on categorical BMI (normal weight, overweight and obese) controlling for age, race/ethnicity, education level, marital status, annual household income, employment status, general health, insurance status, whether the respondent had a personal physician and con-
sidering the complex survey design. The incidence of missing responses (do not know/not sure or refused) for each variable was 1% to 7%. We retained respondents with missing data regarding race/ethnicity, income, education level, employment status, marital status, general health, insurance status, personal physician and diabetes mellitus, and we included dummy variables in the model to denote missing data. To examine the regression model sensitivity to the specification of the BMI variable we fit separate models with BMI as a linear variable and as a 4-level categorical variable (normal, overweight, obese [classes I and II] and extremely obese [class III]). Given the possibility of a nonlinear relationship between age and PSA testing, we treated age as a categorical variable (5-year increments) in the final model. We examined the sensitivity of findings to this specification by fitting a separate model with age as a linear variable. Results were considered statistically significant at 2-sided ␣ ⬍0.05.
RESULTS Table 1 shows the characteristics of the 57,827 respondents. Mean age was 55.4 years (95% CI 55.3–55.6) and 61% of respondents were 50 years or older. Of the respondents 48.0% were overweight and 24.9% were obese. Of the respondents 88.6% reported having health insurance and 82.3% reported having at least 1 personal physician. Obese respondents were younger than normal weight respondents (54.2 vs 56.7 years) and there was a sharp decrease in the proportion of obese men beyond age 69 years. Obese respondents were also much more likely than normal weight respondents to report a diagnosis of diabetes mellitus (43.8% vs 17.6%). Overall 59.9% of respondents (95% CI 59.2– 60.7) reported ever having a PSA test and 42.0% (95% CI 41.2– 42.7) reported having a PSA test in the last year. Of the men 74.6% reported ever having DRE (95% CI 74.0 –75.3). The proportion of respondents reporting DRE in the last year (42.7%, 95% CI 42.0 – 43.4) was similar to that for PSA testing in the last year. Table 2 shows screening rates by BMI category. Obese men were more likely than normal weight men to have ever had a PSA test (62.1% vs 56.1%, p ⬍0.001) and to have had a PSA test in the last year (44.2% vs 38.2%, p ⬍0.001). Similarly obese men were more likely than normal weight men to report having DRE. Overweight men were also more likely than normal weight men to have had PSA or DRE. The increased screening propensity in obese men persisted after controlling for sociodemographic characteristics and access to care. The odds for obese men ever having a PSA test were 1.37 times higher than the odds for normal weight men (95% CI 1.24 –1.51, table 3). Obese men were also more likely to have had a PSA test in the last year (OR 1.32, 95% CI 1.21–1.45) and more likely to report having DRE. Black nonHispanic men were more likely to undergo PSA testing, as were men from higher income households. Access to care had a strong association with the likelihood of prostate cancer screening. The odds of men having a PSA test in the last year were almost
BODY MASS INDEX AND PROSTATE CANCER SCREENING
495
TABLE 1. Respondent characteristics* Wtd % (95% CI)* Characteristic
No. Pts
Normal Wt
Mean ⫾ SE age: 40–49 50–59 60–69 70–79 80 or Older BMI Race: White Black Other Multiracial Hispanic No response Annual household income ($): Less than 35,000 35,000 or Greater Refused Do not know Education level: Less than high school graduate High school graduate 1–3 yrs college 4 or Greater yrs college No response Employment status: Employed Self-employed Unemployed Nonlabor force No response Marital status: Married Not married No response General health: Excellent Very good Good Fair Poor No response Health insurance coverage: Yes No No response Personal physician: Yes (at least 1) No No response Diabetes mellitus: No Yes No response Census region: Northeast Midwest South West
57,827 20,123 16,864 11,130 7,159 2,551 57,827
56.7 37.2 25.8 17.4 13.1 6.5 27.1
(0.17) (35.8–38.5) (24.5–27.0) (16.4–18.5) (12.3–14.0) (5.9–7.1) (26.4–27.7)
55.3 37.7 29.3 18.4 11.5 3.2 48.0
(0.11) (36.7–38.7) (28.3–30.2) (17.6–19.2) (10.8–12.1) (2.9–3.5) (47.3–48.7)
54.2 39.9 31.3 18.6 8.9 1.3 24.9
(0.15) (38.5–41.4) (29.9–32.6) (17.6–19.7) (8.1–9.6) (1.0–1.5) (24.3–25.5)
47,037 3,420 2,557 976 3,194 643
75.1 7.3 7.0 1.3 8.1 1.2
(73.6–76.6) (6.5–8.2) (6.0–8.0) (1.0–1.6) (7.0–9.1) (0.9–1.5)
76.7 7.3 4.2 1.3 9.6 0.9
(75.6–77.7) (6.7–7.9) (3.7–4.8) (1.1–1.6) (8.8–10.5) (0.7–1.0)
74.7 9.7 2.9 1.6 10.3 0.8
(73.2–76.2) (8.8–10.6) (2.4–3.4) (1.2–1.9) (9.0–11.5) (0.6–1.0)
19,890 31,751 1,823 4,363
36.3 52.9 3.3 7.5
(34.9–37.6) (51.6–54.4) (2.7–3.9) (6.8–8.1)
29.3 60.7 2.6 7.3
(28.4–30.3) (59.7–61.7) (2.3–2.9) (6.8–7.9)
34.9 56.2 3.0 5.9
(33.5–36.3) (54.8–57.7) (2.1–3.9) (5.2–6.5)
6,581 17,477 13,682 19,994 93
12.3 (11.4–13.3) 27.5 (26.3–28.8) 22.8 (21.6–24.0) 37.2 (35.9–38.5) 0.16 (0.07–0.25)
11.0 (10.3–11.8) 28.4 (27.5–29.2) 23.9 (23.0–24.8) 36.6 (35.6–37.6) 0.14 (0.06–0.21)
13.8 (12.5–15.0) 32.3 (31.0–33.7) 25.1 (23.9–26.4) 28.6 (27.3–29.9) 0.09 (0.03–0.15)
27,534 8,547 2,184 19,495 67
46.4 (45.0–47.8) 14.2 (13.2–15.2) 4.6 (3.9–5.2) 34.7 (33.4–36.0) 0.13 (0.07–0.20)
51.9 (50.9–53.0) 14.2 (13.5–14.9) 4.1 (3.6–4.6) 29.6 (28.7–30.5) 0.11 (0.05–0.17)
51.9 (50.5–53.4) 11.7 (10.8–12.6) 4.8 (4.1–5.4) 31.5 (30.0–32.9) 0.11 (0.05–0.16)
38,758 18,975 94
70.5 (69.3–71.7) 29.3 (28.1–30.5) 0.17 (0.02–0.31)
77.1 (76.3–77.9) 22.7 (21.9–23.4) 0.21 (0.06–0.36)
77.0 (75.8–78.1) 22.9 (21.7–24.1) 0.10 (0.04–0.15)
11,283 17,766 17,738 7,408 3,448 184
23.9 30.4 27.1 11.7 6.6 0.3
21.3 32.4 30.1 11.4 4.4 0.4
11.2 28.3 35.6 17.2 7.4 0.3
51,577 6,182 68
87.2 (86.2–88.2) 12.7 (11.7–13.7) 0.1 (0.03–0.15)
89.3 (88.6–90.0) 10.5 (9.8–11.2) 0.1 (0.07–0.20)
88.8 (87.8–89.8) 11.1 (10.1–12.1) 0.1 (0.03–0.12)
48,211 9,502 114
79.6 (78.4–80.9) 20.1 (18.9–21.4) 0.2 (0.1–0.4)
82.5 (81.6–83.4) 17.3 (16.4–18.2) 0.2 (0.1–0.3)
84.9 (83.7–86.2) 14.8 (13.6–16.0) 0.2 (0.1–0.4)
51,516 6,200 111
92.7 (91.9–93.5) 7.1 (6.3–7.9) 0.2 (0.1–0.3)
91.0 (90.4–91.6) 8.8 (8.2–9.4) 0.1 (0.1–0.2)
80.6 (79.4–81.8) 19.2 (18.1–20.4) 0.1 (0.1–0.2)
12,273 12,960 12,964 13,197
19.6 21.1 24.7 23.4
19.1 23.0 24.1 22.2
17.6 24.2 25.9 19.3
(22.7–25.0) (29.1–21.7) (25.8–28.3) (10.8–12.6) (5.9–7.3) (0.2–0.4)
(18.5–20.8) (20.2–22.1) (23.6–25.8) (22.0–24.9)
Overwt
(20.5–22.2) (31.5–33.4) (29.1–31.0) (10.7–12.1) (3.9–4.8) (0.2–0.6)
(18.3–20.0) (22.2–23.7) (23.2–24.9) (21.1–23.3)
Obese
(10.3–12.1) (26.9–29.7) (34.2–37.0) (16.1–18.4) (6.6–8.1) (0.2–0.3)
(16.5–18.7) (23.1–25.3) (24.7–27.1) (17.9–20.7)
* Male population in the United States (percents may not sum to 100 due to rounding).
3 times higher if the respondent reported having at least 1 personal physician (OR 2.95, 95% CI 2.63–3.32). There was no association between diabetes mellitus and PSA testing. Findings were not sensitive to the manner in which BMI was classified. In the categorical model obese men (OR 1.38, 95% CI 1.25–1.53) and extremely obese men (OR 1.22, 95% CI 0.94 –1.59) were more likely than normal weight men to report ever having a PSA test. Obese men were also more likely to have had a PSA test in the last year (OR 1.34, 95% CI 1.22–1.48). In the linear model each point increase in BMI increased the odds of ever having a PSA test (OR 1.02,
95% CI 1.01–1.03) and having a PSA test in the last year (OR 1.02, 95% CI 1.01–1.02). Findings were similar for DRE. As in the primary regression, influential covariates in these analyses included race/ethnicity, income, education level and access to care. Findings also were not sensitive to the manner in which age was classified. In the linear model obese men (OR 1.46, 95% CI 1.33–1.61) and overweight men (OR 1.28, 95% CI 1.18 –1.38) were more likely to undergo PSA testing than normal weight men after adjusting for all other covariates. Findings were similar in the 2 models when only men 50 years or older were included.
496
BODY MASS INDEX AND PROSTATE CANCER SCREENING TABLE 2. Screening rates by BMI
Screening Test PSA: Last yr Ever DRE: Last yr Ever
Wtd % (95% CI)* Normal Wt
Overwt
Obese
38.2 (36.8–39.5) 56.1 (54.7–57.5)
42.9 (41.9–44.0) 60.9 (59.9–61.9)
44.2 (42.8–45.7) 62.1 (60.7–63.6)
40.2 (38.9–41.5) 72.1 (70.3–73.0)
43.9 (42.9–45.0) 75.9 (75.0–76.9)
43.2 (41.7–44.6) 75.5 (74.1–76.9)
p ⬍0.001. * Male population in the United States.
these gaps have been closing recently and higher screening rates for cervical and prostate cancer have been documented in black patients.19 The higher rates may reflect patient and practitioner knowledge about the risk of prostate cancer in black men. Health insurance coverage and an ongoing rela-
TABLE 3. Odds of undergoing PSA test by respondent characteristic OR PSA Test (95% CI) Characteristic
DISCUSSION Our findings suggest that obese men are more likely than normal weight men to be screened for prostate cancer. Obese and overweight men are more likely than normal weight men to have ever had a PSA test and to have had a PSA test in the last year. These differences were not explained by respondent sociodemographic characteristics or access to care. Findings also were not sensitive to the manner in which BMI was classified since linear and categorical BMI demonstrated the association between higher BMI and prostate cancer screening. Thus, differences in prostate cancer screening behavior may not explain prostate cancer outcomes in obese men. This study extends a previous analysis of older BRFSS data, which showed a positive association between obesity and PSA testing in men older than 50 years.18 Our analysis included men 40 to 49 years old, which includes men at higher risk of prostate cancer who may elect earlier screening, such as black American men, and it provides a real world view of prostate cancer screening. In addition, the associations were not sensitive to different classifications of age or BMI. The model includes important covariates, such as a personal relationship with a physician, marital status and geographic location. Finally, the inclusion of dummy variables for missing responses (do not know or refused) accounts for potential associations between item nonresponse and screening behavior. We found a strong positive association between obesity and prostate cancer screening, although other studies documented negative associations between obesity and cancer screening.13,14 For example, Rosen and Schneider found that BRFSS respondents with BMI more than 35 kg/m2 were less likely than normal weight respondents to undergo screening for colorectal cancer.14 After controlling for sociodemographic variables this difference was entirely attributable to differences in BMI in women. Using data from the National Health Interview Survey Wee et al found that overweight and obese women were less likely than normal weight women to be screened for cervical and breast cancer.13 These differences may be attributable to the simplicity of serum PSA testing and DRE or to public perceptions of the importance of prostate cancer screening. Several other results of our analysis are noteworthy. Certain sociodemographic and access variables were associated with increased odds of prostate cancer screening. For example, higher income and education levels were associated with higher screening rates, as noted for other cancers.19 Black nonHispanic men were more likely to undergo PSA testing. Although racial/ethnic disparities in cancer screening exist,
Age: 40–45 46–50 51–55 56–60 61–65 66–70 71–75 76–80 81–85 Older than 85 BMI: Normal (less than 25.0 kg/m2) Overwt (25.0–29.9 kg/m2) Obese (30.0 kg/m2 or greater) Race: White nonHispanic Black nonHispanic Other nonHispanic Multiracial nonHispanic Hispanic No response Annual household income ($): Less than 35,000 35,000 or Greater Refused Do not know Education level: Less than high school graduate High school graduate Some college College graduate No response Employment status: Employed Self-employed Not employed Not in labor force No response Marital status: Not married Married No response General health: Excellent Very good Good Fair Poor No response Health insurance coverage: No Yes No response Personal physician: No Yes (at least 1) No response Diabetes mellitus: No Yes No response Census region: Northeast Midwest South West United States territory
Ever
Last Yr
1.00 2.20 (1.99–2.43) 4.13 (3.70–4.60) 6.82 (6.02–7.73) 7.63 (6.53–8.91) 9.92 (8.38–11.8) 11.9 (9.79–14.6) 10.6 (8.46–13.3) 7.21 (5.61–9.29) 5.95 (4.29–8.25)
1.00 1.97 (1.76–2.20) 3.53 (3.16–3.95) 4.87 (4.32–5.50) 5.57 (4.84–6.41) 6.87 (5.92–7.98) 8.39 (7.11–9.88) 6.46 (5.40–7.73) 4.45 (3.58–5.52) 3.87 (2.87–5.20)
1.00 1.23 (1.13–1.33) 1.37 (1.24–1.51)
1.00 1.20 (1.12–1.30) 1.32 (1.21–1.45)
1.00 1.77 (1.51–2.07) 0.62 (0.50–0.76) 1.00 (0.75–1.28) 1.19 (1.01–1.41) 0.96 (0.73–1.28)
1.00 1.58 (1.37–1.83) 0.75 (0.61–0.93) 1.02 (0.75–1.40) 1.30 (1.10–1.53) 1.29 (0.98–1.70)
1.00 1.57 (1.43–1.73) 1.50 (1.31–1.72) 0.94 (0.75–1.17)
1.00 1.47 (1.35–1.60) 1.49 (1.31–1.70) 0.92 (0.76–1.11)
1.00 1.51 (1.32–1.72) 1.87 (1.61–2.16) 2.12 (1.84–2.45) 1.32 (0.64–2.74)
1.00 1.38 (1.21–1.57) 1.55 (1.35–1.77) 1.73 (1.52–1.98) 1.05 (0.50–2.20)
1.00 1.07 (0.96–1.18) 1.02 (0.86–1.22) 1.38 (1.22–1.55) 1.40 (0.67–2.95)
1.00 0.98 (0.88–1.08) 0.94 (0.79–1.13) 1.24 (1.12–1.38) 1.18 (0.60–2.30)
1.00 1.18 (1.09–1.28) 1.44 (0.60–3.45)
1.00 1.10 (1.02–1.18) 2.27 (0.90–5.71)
1.00 0.94 (0.86–1.04) 0.98 (0.89–1.09) 1.02 (0.89–1.17) 0.90 (0.75–1.07) 0.65 (0.37–1.15)
1.00 0.93 (0.85–1.02) 1.06 (0.96–1.16) 0.97 (0.86–1.10) 0.91 (0.77–1.07) 0.90 (0.54–1.51)
1.00 1.52 (1.33–1.72) 0.83 (0.36–1.96)
1.00 1.55 (1.35–1.78) 0.77 (0.29–2.06)
1.00 2.60 (2.35–2.89) 0.83 (0.36–1.96)
1.00 2.95 (2.63–3.31) 1.59 (0.63–3.98)
1.00 1.12 (0.98–1.28) 1.20 (0.70–2.05)
1.00 1.07 (0.95–1.20) 0.94 (0.53–1.54)
1.00 0.93 (0.85–1.03) 1.20 (1.09–1.33) 1.00 (0.89–1.13) 1.25 (1.10–1.42)
1.00 0.95 (0.87–1.04) 1.20 (1.10–1.32) 0.95 (0.85–1.07) 1.14 (1.01–1.28)
BODY MASS INDEX AND PROSTATE CANCER SCREENING tionship with a physician were also strongly associated with prostate cancer screening. The reason for an association between obesity and prostate cancer screening is unclear. The analysis controlled for well-known sociodemographic confounders and access to care. Inclusion of a variable for diabetes mellitus was an attempt to control for more frequent interactions with physicians due to chronic health conditions. It may be that overweight and obese men have more interactions with physicians due to conditions other than diabetes and, therefore, they have more opportunities for screening. Alternatively providers may believe that obese men are a higher risk group. However, it is noteworthy that associations between worse outcomes, advanced stage, higher grade and obesity were published after the 2002 BRFSS survey was completed, and so they are unlikely to have influenced our findings. Our study has some limitations. The analysis is based on patient self-report, which may not correspond with actual screening behavior. Given the high prevalence of PSA testing in this survey, ORs may magnify small absolute differences in screening rates. However, even a small absolute increase in the screening rate in obese men supports the conclusion that differences in prostate cancer outcomes may not be explained by differential screening. The PSA test may not be exclusively used for prostate cancer screening, so that PSA testing rates may overestimate actual cancer screening rates. It is unclear how this source of error would affect our findings. However, respondents were asked about PSA testing in the context of prostate cancer screening, so that we anticipate small effects, if any, from this limitation. Finally, clinical information regarding PSA and subsequent prostate biopsies is lacking, and so we were unable to explore whether differential biopsy rates may affect outcomes in obese men. These limitations notwithstanding, we believe that the findings are robust. They are buttressed by several design strengths. BRFSS is designed to ascertain disease risk behaviors in a large, nationally representative sample. The prevalence of obesity in our sample is similar to estimates from other sources.3 The association between BMI and prostate cancer screening persisted regardless of whether the classification of BMI was categorical or linear. Finally, the model controlled for important sociodemographic characteristics, such as race/ethnicity, income, education level and access to care. CONCLUSIONS To our knowledge our study is the first investigation of the association between obesity and prostate cancer screening behavior in men 40 years or older. Higher screening rates in obese men suggest that disparities in prostate cancer outcomes in obese men do not arise from differences in screening behavior but they may be related to differences in tumor biology, performance characteristics of the PSA test in obese men or differential prostate biopsy rates after an elevated PSA test. Further investigation is required to confirm our findings and explain differences in prostate cancer outcomes in obese men. ACKNOWLEDGMENTS Damon Seils assisted with the manuscript.
497
APPENDIX Survey Questions From the Prostate Cancer Screening Module of the 2002 Behavioral Risk Factor Surveillance System 1. A Prostate Specific Antigen Test, also called a PSA test, is a blood test used to check men for prostate cancer. Have you ever had a PSA test? Yes No Do not know/not sure Refused 2. How long has it been since you had your last PSA test? Within the past year (less than 12 months ago) Within the past 2 years (greater than 1 year but less than 2 years) Within the past 3 years (greater than 2 years but less than 3 years) Within the past 5 years (greater than 3 years but less than 5 years) 5 or More years ago Do not know/not sure Refused 3. A digital rectal examination is an examination in which a doctor, nurse, or other health professional places a gloved finger into the rectum to feel the size, shape, and hardness of the prostate gland. Have you ever had a digital rectal examination? Yes No Do not know/not sure Refused 4. How long has it been since your last digital rectal examination? Within the past year (less than 12 months ago) Within the past 2 years (greater than 1 year but less than 2 years) Within the past 3 years (greater than 2 years but less than 3 years) Within the past 5 years (greater than 3 years but less than 5 years) 5 or More years ago Do not know/not sure Refused 5. Have you ever been told by a doctor, nurse, or other health professional that you had prostate cancer? Yes No Do not know/not sure Refused
Abbreviations and Acronyms BMI ⫽ body mass index BRFSS ⫽ Behavioral Risk Factor Surveillance System DRE ⫽ digital rectal examination PSA ⫽ prostate specific antigen
REFERENCES 1.
Allison DB, Fontaine KR, Manson JE, Stevens J and VanItallie TB: Annual deaths attributable to obesity in the United States. JAMA 1999; 282: 1530. 2. Calle EE, Rodriguez C, Walker-Thurmond K and Thun MJ: Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003; 348: 1625. 3. Amling CL, Kane CJ, Riffenburgh RH, Ward JF, Roberts JL, Lance RS et al: Relationship between obesity and race in predicting adverse pathologic variables in patients undergoing radical prostatectomy. Urology 2001; 58: 723. 4. Freedland SJ, Aronson WJ, Kane CJ, Presti JC Jr, Amling CL, Elashoff D et al: Impact of obesity on biochemical control after radical prostatectomy for clinically localized prostate cancer: a report by the Shared Equal Access Regional Cancer Hospital database study group. J Clin Oncol 2004; 22: 446. 5. Rohrmann S, Roberts WW, Walsh PC and Platz EA: Family history of prostate cancer and obesity in relation to highgrade disease and extraprostatic extension in young men with prostate cancer. Prostate 2003; 55: 140.
498 6.
7.
8.
9.
10.
11. 12.
BODY MASS INDEX AND PROSTATE CANCER SCREENING Whittemore AS, Kolonel LN, Wu AH, John EM, Gallagher RP, Howe GR et al: Prostate cancer in relation to diet, physical activity, and body size in blacks, whites, and Asians in the United States and Canada. J Natl Cancer Inst 1995; 87: 652. Amling CL, Riffenburgh RH, Sun L, Moul JW, Lance RS, Kusuda L et al: Pathologic variables and recurrence rates as related to obesity and race in men with prostate cancer undergoing radical prostatectomy. J Clin Oncol 2004; 22: 439. Kane CJ, Bassett WW, Sadetsky N, Silva S, Wallace K, Pasta DJ et al: Obesity and prostate cancer clinical risk factors at presentation: data from CaPSURE. J Urol 2005; 173: 732. Wolk A, Mantzoros CS, Andersson SO, Bergstrom R, Signorello LB, Lagiou P et al: Insulin-like growth factor 1 and prostate cancer risk: a population-based, case-control study. J Natl Cancer Inst 1998; 90: 911. Moyad MA: Is obesity a risk factor for prostate cancer, and does it even matter? A hypothesis and different perspective. Urology, suppl., 2002; 59: 41. Yip I, Heber D and Aronson W: Nutrition and prostate cancer. Urol Clin North Am 1999; 26: 403. Baillargeon J, Pollock BH, Kristal AR, Bradshaw P, Hernandez J, Basler J et al: The association of body mass index and prostate-specific antigen in a population-based study. Cancer 2005; 103: 1092.
13.
14.
15. 16.
17.
18.
19.
Wee CC, McCarthy EP, Davis RB and Phillips RS: Screening for cervical and breast cancer: is obesity an unrecognized barrier to preventive care? Ann Intern Med 2000; 132: 697. Rosen AB and Schneider EC: Colorectal cancer screening disparities related to obesity and gender. J Gen Intern Med 2004; 19: 332. Prostate-specific antigen (PSA) best practice policy. American Urological Association. Oncology (Huntingt), 2000; 14: 267. Behavioral Risk Factor Surveillance System: Codebook 2002. Atlanta: United States Department of Health and Human Services, Centers for Disease Control and Prevention. Available at http://www.cdc.gov/brfss/technical_infodata/ surveydata/codebook_02.rtf. Accessed May 3, 2005. Rao JNK and Scott AJ: The analysis of categorical data from complex sample surveys: chi-squared tests for goodness of fit and independence in two-way tables. J Am Stat Assoc 1981; 76: 221. Sirovich BE, Schwartz LM and Woloshin S: Screening men for prostate and colorectal cancer in the United States: does practice reflect the evidence? JAMA 2003; 289: 1414. Breen N, Wagener DK, Brown ML, Davis WW and BallardBarbash R: Progress in cancer screening over a decade: results of cancer screening from the 1987, 1992, and 1998 National Health Interview Surveys. J Natl Cancer Inst 2001; 93: 1704.