- Email: [email protected]
Cancer screening estimates for U.S. metropolitan areas
Cancer Screening Estimates for U.S. Metropolitan Areas David E. Nelson, MD, MPH, Julie Bolen, PhD, Stephen Marcus, PhD, Henry E. Wells, MS, Helen Meissner, PhD
Objectives:
To provide estimates of breast, cervical, and colorectal cancer screening for metropolitan areas in the United States.
Methods:
Behavioral Risk Factor Surveillance System (BRFSS) data from 1997 to 1999 were reweighted and analyzed for 69 U.S. metropolitan areas for the receipt of a Papanicolaou (Pap) test (ages ⱖ18 years); mammography (ages ⱖ40 years); fecal occult blood testing and sigmoidoscopy (ages ⱖ50 years). Stratified analyses by demographics were performed for 25 metropolitan areas with populations of ⱖ1.5 million.
Results:
Metropolitan estimates ranged from 64.6% to 82.0% for mammography and from 77.2% to 91.7% for Pap tests. There was much greater variability in estimates for colorectal cancer screening, with a 3.6-fold difference in the range of estimates for fecal occult blood testing (9.9% to 35.2%) and a 2.5-fold difference for sigmoidoscopy (17.3% to 43.3%). In the 25 largest areas, prevalence of cancer screening was generally lower for persons with a high school education or less and for those without health insurance. Compared with women aged 50 to 64 years, mammography estimates were lower for women aged 40 to 49 years in 13 of the 25 metropolitan areas. Pap testing was less common among women aged ⱖ65 years, and colorectal cancer screening was less common for persons aged 50 to 64 years.
Conclusions: Estimates of cancer screening varied substantially across metropolitan areas. Increased efforts to improve cancer screening are needed in many urban areas, especially for colorectal cancer screening. The BRFSS is a useful, inexpensive, and timely resource for providing metropolitan-area cancer screening estimates and may be used in the future to guide local or county-level screening efforts. (Am J Prev Med 2003;24(4):301–309)
Introduction
S
creening has been found to reduce morbidity and premature mortality from cervical, breast, and colorectal cancer.1–11 The National Health Interview Survey12–14 and the Behavioral Risk Factor Surveillance System (BRFSS)15,16 provide national and state estimates for receipt of cancer screening among adults. Although there have been efforts in Los Angeles County and elsewhere,17–26 metropolitan-level estimates for cancer screening are not widely available. Conducting surveys is expensive, and many health departments lack resources or infrastructure to collect such data. The lack of urban data on cancer screening is unfortunate. Sociodemographics of urban areas differ From the Division of Cancer Control and Population Sciences, National Cancer Institute (Nelson, Marcus, Meissner), Bethesda, Maryland; Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention (Bolen), Atlanta, Georgia; and Research Triangle Institute (Wells), Atlanta, Georgia Address correspondence to: David E. Nelson, MD, MPH, Centers for Disease Control and Prevention, 4770 Buford Hwy, NE, Mailstop K-50, Atlanta GA 30341-3717. E-mail: [email protected].
Am J Prev Med 2003;24(4) Published by Elsevier Inc.
from those of states; most urban areas have a higher concentration of racial/ethnic minority populations, who also are over-represented among the poor.27 Participation in cancer screening programs is low among many inner-city populations,28 –32 and national and state estimates may mask such differences.33,34 Relevant data can be used to help empower communities to address their own health issues,33–36 especially as many efforts to improve cancer screening occur locally,37– 41 and these data can help track progress toward Healthy People 2010 goals.42 One source for obtaining metropolitan area data is the BRFSS,43,44 as the Centers for Disease Control and Prevention (CDC) recently reweighted BRFSS data for metropolitan areas based on county of residence.45 For this study, the overall prevalence estimates for the receipt of a Papanicolaou (Pap) test, mammography, and colorectal cancer screening (fecal occult blood test [FOBT] and sigmoidoscopy) were determined for metropolitan areas with sufficient sample sizes. Screening estimates were then compared by age, gender, race/ethnicity, education level, and health insurance status for the 25 largest metropolitan areas.
0749-3797/03/$–see front matter doi:10.1016/S0749-3797(03)00024-2
301
Methods Data used in this study came from the BRFSS. Briefly, the BRFSS is a state-based system of health surveys of adults that is coordinated by CDC.46 – 48 Health-related data are obtained each year through telephone surveys of randomly selected persons aged ⱖ18 years. We analyzed data from 1997 to 1999, the most recent years with metropolitan-level data on cancer screening. Data on mammography and cervical cancer screening were obtained in each of these years, but colorectal cancer screening data was only available for 1997 and 1999. The total sample size ranged from 133,048 in 1997 to 159,921 in 1999. Median state-level annual response rates, based on persons actually reached by telephone, were 76.8% in 1997, 73.4% in 1998, and 68.4% in 1999. Because these were state-based surveys, response rates by metropolitan areas were unavailable.
Cancer Screening Definitions For the purposes of this study, mammography screening was the percentage of women aged ⱖ40 years who (1) answered “yes” to “A mammogram is an x-ray of each breast to look for breast cancer. Have you ever had a mammogram?” and (2) who reported receiving a mammogram sometime within the past 2 years when asked “How long has it been since you had your last mammogram?” Cervical cancer screening was the percentage of women aged ⱖ18 years (1) who answered “yes” to “A Pap smear is a test for cancer of the cervix. Have you ever had a Pap smear?”; (2) who reported receiving the test sometime within the past 3 years when asked “How long has it been since had your last Pap smear?”; and (3) who answered “no” to the question “Have you ever had a hysterectomy?” We assessed two measures for colorectal cancer screening among persons aged ⱖ50 years: FOBT and sigmoidoscopy. Receipt of FOBT was based on the percentage of respondents (1) who answered “yes” to “A blood stool test is a test that may use a special kit at home to determine whether the stool contains blood. Have you ever had this test using a home kit?”; and (2) who had reported receiving it within the past year or more recently when asked “When did you have your last blood stool test using a home kit?” The question on receipt of sigmoidoscopy was slightly different in 1997 than in 1999. In 1997, respondents were asked, “A sigmoidoscopy or proctoscopy is when a tube is inserted in the rectum to view the bowel for signs of cancer and other health problems. Have you ever had this exam?” In 1999, the question was changed to receiving sigmoidoscopy or colonoscopy. Persons who responded “yes” to this question, and who reported having received the test within the past ⱕ5 years when asked “When did you have your last sigmoidoscopy or proctoscopy [colonoscopy]?” were considered to have received colorectal cancer screening. We use the word “sigmoidoscopy” to describe this type of colorectal cancer screening across both survey years.
Metropolitan Areas and Statistical Analyses Self-reported county of residence was used to classify respondents as residents of metropolitan statistical areas, primary metropolitan statistical areas, or New England county metropolitan areas using census definitions.49 Data were then reweighted, based on the age, gender, and race/ethnicity
302
distribution, for each metropolitan area using intercensal estimates. All counties within the metropolitan areas were included regardless of state boundaries. To increase precision of the estimates, data were pooled across years (3 years for mammography and cervical cancer screening, 2 years for colorectal cancer screening). Analyses were limited to metropolitan areas with ⱖ300 respondents for all measures or that had a population of ⱖ1.5 million residents. Sixty-nine areas met our inclusion criteria. For the 25 metropolitan areas with intercensal population estimates of ⱖ1.5 million, additional analyses were conducted by age (18 to 34, 35 to 64, and ⱖ65 years for Pap test; 40 to 49, 50 to 64, and ⱖ65 years for mammography); gender (colorectal measures); race/ethnicity (non-Hispanic white, nonHispanic black, Hispanic); education (high school or less, more than high school), and health insurance status. Because of small numbers, analyses were restricted to groups with ⱖ50 persons. Due to overall small numbers, subgroup analyses for colorectal cancer screening were not performed for Oakland, Orange County, Riverside–San Bernardino, and San Diego in California. All statistical analyses were conducted in SAS (SAS Institute, Cary NC, 2001) and SUDAAN (Research Triangle Institute, Research Triangle Park NC, 2000). To provide additional context, pooled state estimates were analyzed for each measure. Metropolitan areas were grouped by census region (Northeast, Midwest, South, and West) and calculated for regional median and range values. For the 25 largest metropolitan areas, differences among subgroups groups within each area were assessed using two-sample t -tests for differences between estimates. Because of the potential problem of multiple comparisons, differences were considered significant only when p ⬍0.01.
Results Overall Estimates for 69 Metropolitan Areas Metropolitan estimates for mammography, cervical cancer screening, and colorectal cancer screening for the 69 metropolitan areas are listed in Table 1. Regional medians for both mammography (range for medians, 73.5% to 76.4%) and cervical cancer screening (range for medians, 85.1% to 86.9%) were similar across regions. Especially low estimates for both mammography and cervical cancer screening were noted for Charleston WV and Huntington–Ashland WV–KY–OH. There was much greater variation for estimates of colorectal cancer screening (Table 1). Regional median estimates for FOBT ranged from 18.0% to 24.3%, with lower estimates found in the South and West. However, there was more than a three-fold difference across individual metropolitan areas (range, 9.9% to 35.2%), with 11 areas having FOBT estimates of ⱕ15%. Regional estimates for sigmoidoscopy were similar, ranging from 32.7% in the Midwest to 33.9% in the Northeast. Estimates for individual metropolitan areas varied from 17.3% to 43.3%, with the lowest estimates in Huntington–Ashland WV–KY–OH, Oklahoma City OK, and Omaha NE–IA.
American Journal of Preventive Medicine, Volume 24, Number 4
Table 1. Overall estimates for cancer screening by metropolitan area,a 1997–1999 Colorectal cancer screening Region and metropolitan area Northeast Bergen–Passaic NJ Boston MA–NH Burlington VT Hartford CT Monmouth–Ocean NJ Nassau–Suffolk NY New Haven–Bridgeport– Stamford–Waterbury– Danbury CT Newark NJ New York NY Philadelphia PA–NJ Pittsburgh PA Providence–Warwick–Pawtucket RI Northeast median Range Midwest Chicago IL Cincinnati OH–KY–IN Cleveland–Lorain–Elyria OH Columbus OH Des Moines IA Detroit MI Indianapolis IN Kansas City MO–KS Milwaukee–Waukesha WI Minneapolis–St. Paul MN–WI Omaha NE–IA St. Louis MO–IL Sioux Falls SD Wichita KS Midwest median Range South Atlanta GA Baltimore MD Birmingham AL Charleston WV Charlotte–Gastonia–Rock Hill NC–SC Dallas TX Dover DE Ft. Lauderdale FL Greensboro–Winston–Salem– High Point NC Greenville–Spartanburg– Anderson SC Houston TX Huntington–Ashland WV– KY–OH Jacksonville FL Little Rock–North Little Rock AR Louisville KY–IN Memphis TN–AR–MS Miami FL Nashville TN New Orleans LA Norfolk–Virginia Beach– Newport News VA–NC
Mammography % ⴞ 95% CI
Pap test % ⴞ 95% CI
FOBT % ⴞ 95% CI
Sigmoidoscopy % ⴞ 95% CI
73.8 ⫾ 4.8 82.0 ⫾ 1.6 75.6 ⫾ 3.0 80.1 ⫾ 2.9 70.7 ⫾ 4.7 76.3 ⫾ 4.8 78.8 ⫾ 2.7
83.5 ⫾ 4.0 87.6 ⫾ 1.3 88.0 ⫾ 2.5 87.3 ⫾ 2.3 83.1 ⫾ 3.4 86.6 ⫾ 3.3 85.4 ⫾ 2.1
24.5 ⫾ 5.9 28.0 ⫾ 2.5 28.7 ⫾ 3.8 26.9 ⫾ 4.2 23.2 ⫾ 4.8 23.1 ⫾ 5.1 25.7 ⫾ 3.3
37.8 ⫾ 6.3 33.4 ⫾ 2.6 33.3 ⫾ 4.0 36.9 ⫾ 4.5 31.7 ⫾ 5.5 36.4 ⫾ 5.8 38.3 ⫾ 3.8
76.4 ⫾ 4.0 76.4 ⫾ 2.9 78.9 ⫾ 2.3 73.6 ⫾ 3.3 81.3 ⫾ 1.7 76.4 70.7–82.0
89.1 ⫾ 2.8 83.1 ⫾ 2.1 84.8 ⫾ 2.1 80.3 ⫾ 2.7 86.1 ⫾ 1.4 85.1 80.3–89.1
21.5 ⫾ 4.4 24.2 ⫾ 3.3 24.7 ⫾ 3.0 21.8 ⫾ 3.0 23.8 ⫾ 2.1 24.3 21.5–28.7
32.8 ⫾ 5.0 34.0 ⫾ 3.7 30.9 ⫾ 3.3 33.9 ⫾ 4.4 36.1 ⫾ 2.5 33.9 30.9–38.3
72.1 ⫾ 2.5 76.3 ⫾ 3.9 76.8 ⫾ 4.3 73.4 ⫾ 6.2 76.7 ⫾ 3.6 76.5 ⫾ 76.5 70.0 ⫾ 5.2 75.7 ⫾ 2.6 73.2 ⫾ 4.2 68.2 ⫾ 2.0 72.9 ⫾ 2.9 74.8 ⫾ 3.2 70.2 ⫾ 4.3 73.7 ⫾ 4.2 73.5 68.2–76.8
85.8 ⫾ 1.6 83.9 ⫾ 3.6 89.2 ⫾ 2.9 89.3 ⫾ 3.3 86.2 ⫾ 2.9 86.6 ⫾ 2.0 87.1 ⫾ 3.4 89.1 ⫾ 1.7 84.8 ⫾ 3.0 88.0 ⫾ 1.1 86.0 ⫾ 2.2 86.3 ⫾ 2.3 85.3 ⫾ 3.0 87.7 ⫾ 3.0 86.5 83.9–89.3
17.5 ⫾ 3.5 22.2 ⫾ 4.3 23.4 ⫾ 5.1 22.6 ⫾ 6.2 28.4 ⫾ 5.0 20.0 ⫾ 3.2 13.3 ⫾ 3.9 23.4 ⫾ 3.2 12.3 ⫾ 4.1 22.7 ⫾ 2.1 23.8 ⫾ 3.3 18.0 ⫾ 3.7 15.2 ⫾ 3.7 24.3 ⫾ 5.1 22.4 12.3–28.4
31.7 ⫾ 4.4 26.4 ⫾ 4.7 43.3 ⫾ 5.7 28.0 ⫾ 6.2 33.8 ⫾ 5.3 38.4 ⫾ 4.0 26.5 ⫾ 5.2 35.5 ⫾ 3.7 34.8 ⫾ 5.6 41.5 ⫾ 2.5 24.0 ⫾ 3.5 25.9 ⫾ 4.0 29.0 ⫾ 5.1 33.6 ⫾ 5.6 32.7 24.0–43.3
74.0 ⫾ 3.1 81.3 ⫾ 2.3 80.5 ⫾ 4.0 67.7 ⫾ 4.8 74.5 ⫾ 4.3
90.6 ⫾ 1.7 89.8 ⫾ 1.8 89.1 ⫾ 2.9 78.5 ⫾ 4.1 87.1 ⫾ 3.3
17.5 ⫾ 3.2 28.2 ⫾ 3.0 13.5 ⫾ 4.1 11.0 ⫾ 3.3 17.5 ⫾ 3.5
42.5 ⫾ 4.5 33.6 ⫾ 3.2 30.5 ⫾ 5.0 27.3 ⫾ 5.2 36.5 ⫾ 6.3
71.8 ⫾ 4.3 77.0 ⫾ 3.3 78.7 ⫾ 4.2 72.3 ⫾ 5.2
85.0 ⫾ 3.2 87.2 ⫾ 2.3 86.8 ⫾ 3.3 91.0 ⫾ 3.3
16.8 ⫾ 4.4 17.1 ⫾ 3.7 23.9 ⫾ 4.9 35.2 ⫾ 5.2
25.8 ⫾ 5.4 31.4 ⫾ 4.4 36.0 ⫾ 6.2 33.3 ⫾ 5.1
76.1 ⫾ 3.6
86.7 ⫾ 3.0
18.5 ⫾ 3.9
29.5 ⫾ 4.6
72.2 ⫾ 4.2 64.6 ⫾ 6.2
82.7 ⫾ 3.2 77.2 ⫾ 5.8
20.7 ⫾ 3.1 15.5 ⫾ 4.5
35.1 ⫾ 5.9 23.9 ⫾ 5.4
81.7 ⫾ 4.5 70.4 ⫾ 4.5
89.6 ⫾ 3.6 82.2 ⫾ 4.1
19.3 ⫾ 5.3 15.0 ⫾ 3.8
29.4 ⫾ 5.9 35.1 ⫾ 5.1
74.2 ⫾ 3.9 73.4 ⫾ 3.7 74.6 ⫾ 4.5 77.7 ⫾ 3.3 70.4 ⫾ 4.4 76.0 ⫾ 4.2
87.0 ⫾ 3.1 89.6 ⫾ 2.2 83.4 ⫾ 3.1 89.9 ⫾ 2.3 84.8 ⫾ 3.2 89.1 ⫾ 2.8
21.2 ⫾ 5.7 12.9 ⫾ 2.6 13.9 ⫾ 3.7 14.5 ⫾ 3.4 18.0 ⫾ 4.2 17.2 ⫾ 4.7
26.0 ⫾ 4.7 29.2 ⫾ 4.8 29.9 ⫾ 5.8 27.6 ⫾ 4.9 33.9 ⫾ 5.5 34.4 ⫾ 5.9 (continued on next page)
Am J Prev Med 2003;24(4)
303
Table 1. (continued) Colorectal cancer screening Region and metropolitan area Oklahoma City OK Richmond–Petersburg, VA Tampa–St. Petersburg–Clearwater FL Tulsa OK Washington DC–MD–VA–WV West Palm Beach–Boca Raton FL Wilmington–Newark DE–MD South median Range West Albuquerque NM Boise City ID Denver CO Honolulu HI Las Vegas NV–AZ Los Angeles–Long Beach CA Oakland CA Orange County CA Phoenix–Mesa AZ Portland–Vancouver OR–WA Reno NV Riverside–San Bernardino CA Salt Lake City–Ogden UT San Diego CA Seattle–Bellevue–Everett WA Tucson AZ West median Range Nationwide median Range
Mammography % ⴞ 95% CI
Pap test % ⴞ 95% CI
FOBT % ⴞ 95% CI
Sigmoidoscopy % ⴞ 95% CI
68.0 ⫾ 3.5 76.1 ⫾ 4.8 78.9 ⫾ 3.3 69.7 ⫾ 4.1 76.7 ⫾ 2.6 80.5 ⫾ 4.2 79.7 ⫾ 2.8 74.6 64.6–81.7
86.4 ⫾ 2.6 91.7 ⫾ 2.8 84.9 ⫾ 3.1 86.3 ⫾ 3.0 86.9 ⫾ 2.2 82.7 ⫾ 3.9 87.8 ⫾ 1.8 86.9 77.2–91.7
9.9 ⫾ 2.5 18.4 ⫾ 5.3 28.6 ⫾ 4.4 21.2 ⫾ 4.0 25.8 ⫾ 3.0 30.0 ⫾ 5.3 20.8 ⫾ 3.0 18.0 9.9–35.2
17.3 ⫾ 3.1 37.8 ⫾ 6.3 33.4 ⫾ 4.6 28.8 ⫾ 4.6 39.4 ⫾ 3.5 35.1 ⫾ 5.6 41.8 ⫾ 3.6 33.3 17.3–42.5
75.0 ⫾ 2.8 66.9 ⫾ 3.3 74.7 ⫾ 3.3 76.5 ⫾ 3.0 67.4 ⫾ 4.2 73.1 ⫾ 3.4 79.8 ⫾ 5.3 78.6 ⫾ 5.3 79.1 ⫾ 5.0 76.6 ⫾ 2.7 70.4 ⫾ 4.4 66.1 ⫾ 5.5 68.0 ⫾ 3.2 70.8 ⫾ 5.4 72.4 ⫾ 2.7 78.1 ⫾ 5.0 73.9 66.1–79.8 74.8 64.6–82.0
86.0 ⫾ 2.1 85.2 ⫾ 2.2 87.8 ⫾ 2.2 86.8 ⫾ 2.0 81.3 ⫾ 3.2 82.8 ⫾ 2.4 90.8 ⫾ 3.3 86.9 ⫾ 3.7 83.4 ⫾ 3.7 89.3 ⫾ 1.7 87.0 ⫾ 2.7 82.6 ⫾ 3.9 80.9 ⫾ 2.4 84.1 ⫾ 3.9 88.0 ⫾ 1.7 82.6 ⫾ 3.7 85.6 80.9–90.8 86.4 77.2–91.7
22.3 ⫾ 3.6 17.8 ⫾ 3.2 26.8 ⫾ 4.2 20.7 ⫾ 3.1 10.8 ⫾ 3.3 14.2 ⫾ 3.0 NA NA 19.1 ⫾ 4.1 25.6 ⫾ 3.1 16.6 ⫾ 4.1 NA 17.2 ⫾ 3.0 NA 27.2 ⫾ 3.2 23.4 ⫾ 5.5 19.9 10.8–27.2 21.2 9.9–35.2
35.6 ⫾ 3.9 31.1 ⫾ 4.1 30.6 ⫾ 4.2 40.1 ⫾ 4.0 29.2 ⫾ 4.7 35.1 ⫾ 4.4 NA NA 32.7 ⫾ 4.8 37.3 ⫾ 3.6 25.4 ⫾ 4.9 NA 34.9 ⫾ 3.8 NA 35.6 ⫾ 3.6 30.7 ⫾ 5.6 33.8 25.4–40.1 33.4 17.3–43.3
a
Because of insufficient sample size, no estimates are available for metropolitan areas in Maine, Mississippi, Montana, New Hampshire, North Dakota, or Wyoming. Alaska has no defined metropolitan statistical areas. CI, confidence interval; FOBT, fecal occult blood test; NA, not available, as sample size ⬍300; Pap test, Papanicolaou test.
Selected Demographic Estimates for the 25 Largest Metropolitan Areas Further analyses of mammography and cervical cancer screening by selected demographic measures for the 25 largest metropolitan areas are presented in Table 2. Mammography varied by age; with ages 50 to 64 years as the baseline, estimates were significantly lower in 13 areas among women aged 40 to 49 and for three areas among women aged ⱖ65 years. We found no significant differences for mammography estimates between blacks and whites, but estimates were significantly lower for Hispanics compared to whites in Riverside–San Bernardino CA (data not shown). In all areas, persons with more than a high school education were more likely than those with less than a high school education to report mammography screening (data not shown). Health insurance was associated with receipt of mammography, as estimates were significantly lower for uninsured women in 13 of 17 areas for which these comparisons could be made. Cervical cancer screening was significantly less common among women aged ⱖ65 years than among 304
women aged 18 to 34 years in 13 of 25 areas (Table 3). Estimates for cervical cancer screening were significantly higher for blacks than whites in five areas (Atlanta GA, Baltimore MD, Chicago IL, Houston TX, and Washington DC–MD–VA–WV), but we found no differences between Hispanics and whites in metropolitan areas with adequate sample sizes for such comparisons (data not shown). Cervical cancer screening estimates were lower for persons with less than a high school education in all but one of the largest metropolitan areas, with the differences significant in 13 areas. As with mammography, health insurance status was associated with cervical cancer screening, with estimates significantly lower in 11 of 23 metropolitan areas for the uninsured. Colorectal cancer screening estimates by demographic groups were available for 21 of the 25 largest metropolitan areas (Table 4). FOBT was less commonly reported for people aged 50 to 64 years compared with the ⱖ65 group, with significant differences found in 9 of 21 areas. In contrast, there were few differences in FOBT estimates by gender, race/ethnicity, or educa-
American Journal of Preventive Medicine, Volume 24, Number 4
Table 2. Mammography estimates for the largest U.S. metropolitan areas by age and health insurance status Age (years) % ⴞ 95% CI
Insurance status % ⴞ 95% CI
Metropolitan area
40–49
50–64
>65
Insured
Uninsured
Los Angeles–Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA Washington DC–MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Riverside–San Bernardino CA Minneapolis–St. Paul MN–WI San Diego CA Nassau–Suffolk NY Orange County CA St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg– Clearwater FL Seattle–Bellevue–Everett WA Oakland CA Cleveland–Lorain–Elyria OH Miami FL Newark NJ Median Range
63.8 ⫾ 5.7* 73.3 ⫾ 5.1* 65.5 ⫾ 4.1* 77.8 ⫾ 2.3* 73.6 ⫾ 4.3* 70.3 ⫾ 4.5* 69.2 ⫾ 5.5* 62.1 ⫾ 7.0* 69.6 ⫾ 5.0* 61.7 ⫾ 7.2* 72.2 ⫾ 11.5 54.2 ⫾ 9.8 60.0 ⫾ 3.5* 55.7 ⫾ 9.8* 71.1 ⫾ 8.8 71.0 ⫾ 9.4 70.8 ⫾ 5.9 79.6 ⫾ 4.1 71.2 ⫾ 6.7 70.4 ⫾ 6.7
79.0 ⫾ 5.6 83.8 ⫾ 4.5 76.1 ⫾ 4.4 87.6 ⫾ 3.2 83.3 ⫾ 4.1 83.8 ⫾ 4.0 80.3 ⫾ 4.7 81.4 ⫾ 5.2 82.0 ⫾ 4.6 77.0 ⫾ 6.7 81.0 ⫾ 6.8 68.2 ⫾ 9.9 74.9 ⫾ 3.3 85.0 ⫾ 7.5 81.7 ⫾ 7.4 78.8 ⫾ 9.3 79.9 ⫾ 5.3 82.3 ⫾ 4.1 77.5 ⫾ 6.3 77.5 ⫾ 6.3
77.9 ⫾ 5.8 72.2 ⫾ 5.1 74.8 ⫾ 4.0 80.7 ⫾ 2.8 79.6 ⫾ 4.0 76.5 ⫾ 4.5 79.8 ⫾ 4.8 75.9 ⫾ 8.9 69.6 ⫾ 6.3 78.7 ⫾ 7.8 83.9 ⫾ 5.3 76.0 ⫾ 8.7* 70.8 ⫾ 3.5 72.3 ⫾ 9.3 75.1 ⫾ 8.0 85.7 ⫾ 8.3 73.5 ⫾ 5.5 82.1 ⫾ 3.9 71.1 ⫾ 5.2 84.6 ⫾ 4.5*
77.8 ⫾ 3.3 78.1 ⫾ 2.9 73.7 ⫾ 2.5 83.0 ⫾ 1.6 79.9 ⫾ 2.4 77.8 ⫾ 2.7 79.0 ⫾ 2.9 75.9 ⫾ 4.4 75.2 ⫾ 3.1 74.8 ⫾ 4.3 NA NA 69.5 ⫾ 2.1 NA NA NA 76.4 ⫾ 3.2 83.1 ⫾ 2.3 74.8 ⫾ 3.4 83.3 ⫾ 3.2
49.5 ⫾ 10.5** 64.5 ⫾ 11.1 52.3 ⫾ 9.7** 63.9 ⫾ 9.0** 62.5 ⫾ 11.5** 61.8 ⫾ 10.2** 41.6 ⫾ 13.8** 52.2 ⫾ 11.3** 59.1 ⫾ 12.9 44.8 ⫾ 14.5** NA NA 37.8 ⫾ 11.0** NA NA NA 52.7 ⫾ 14.2** 56.0 ⫾ 11.6** 55.5 ⫾ 14.8 37.2 ⫾ 12.5**
65.2 ⫾ 4.6* 76.8 ⫾ 8.7 68.7 ⫾ 8.9 70.1 ⫾ 8.7 74.9 ⫾ 6.6 70.3 54.2–79.6
78.7 ⫾ 4.4 78.7 ⫾ 10.0 83.3 ⫾ 7.3 74.0 ⫾ 7.2 81.0 ⫾ 6.6 80.3 68.2–87.6
74.2 ⫾ 4.6 85.6 ⫾ 7.7 77.6 ⫾ 6.0 78.7 ⫾ 6.8 73.0 ⫾ 7.3 76.5 69.6–85.7
NA NA NA 80.4 ⫾ 4.4 77.8 ⫾ 4.0 77.8 69.5–83.3
NA NA NA 50.2 ⫾ 12.4** 62.8 ⫾ 16.0 52.7 37.2–64.5
* Difference statistically significant compared to persons aged 50 – 64 years; **difference statistically significant compared to persons with health insurance. CI, confidence interval; NA, not available because of insufficient sample size.
tion (data not shown). Sigmoidoscopy estimates were higher among persons aged ⱖ65 years and among men, with significant differences found in 7 of 21 areas by both age and gender. Unfortunately, small sample sizes preclude drawing any firm conclusions about race/ethnic differences in colorectal cancer screening in most areas for this measure (data not shown). Persons with lower levels of education were less likely to receive sigmoidoscopy, with significant differences in 6 of 21 areas. Because of small sample sizes, we were able only to compare estimates for FOBT and sigmoidoscopy by insurance status for eight metropolitan areas, but estimates were significantly lower for the uninsured in six of these areas for both screening tests (data not shown).
Discussion This study had several important findings. We demonstrated the ability of the BRFSS to provide cancer screening estimates in 69 metropolitan areas. Although regional median metropolitan estimates for screening were generally similar, they masked important intraand inter-regional differences. Differences were most pronounced for colorectal cancer screening; subopti-
mal estimates found in all areas indicate increased efforts are warranted to increase such screening nationwide. Although differences were smaller, substantial variation by metropolitan area for both breast and cervical cancer screening remained. We also found important differences among subpopulations within the 25 largest metropolitan areas and confirmed data from national and state studies that cervical and colorectal cancer screening estimates vary by age. Lower estimates for mammography among women aged 40 to 49 years, and lower estimates for cervical cancer screening among women aged ⱖ65 years were not surprising, given the ongoing debate about screening intervals, the recommended age when women should begin mammography screening, and whether regular cervical cancer screening can be discontinued after age 65. Furthermore, the U.S. Preventive Services Task Force, American Cancer Society, American College of Obstetricians and Gynecologists, and American Medical Association have differing recommendations for these tests.15,50 As for differences by education level, except for FOBT, people with lower levels of education were generally less likely to receive cancer screening in the 25 largest metropolitan areas.14,15 Am J Prev Med 2003;24(4)
305
Table 3. Cervical cancer screening estimates for the largest U.S. metropolitan areas, by selected demographics Age (years) % ⴞ 95% CI
Educational level % ⴞ 95% CI
Insurance status % ⴞ 95% CI
Metropolitan area
18–34
35–64
>65
>HS
Insured
Uninsured
Los Angeles–Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA–NJ Washington DC–MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Riverside–San Bernardino CA Minneapolis–St. Paul MN– WI San Diego CA Nassau–Suffolk NY Orange County CA St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg–Clearwater FL Seattle–Bellevue–Everett WA Oakland CA Cleveland–Lorain–Elyria OH Miami FL Newark NJ Median Range
81.8 ⫾ 3.9 82.4 ⫾ 3.5 87.5 ⫾ 2.5 88.0 ⫾ 2.3 85.6 ⫾ 4.5 85.7 ⫾ 3.8 86.8 ⫾ 3.3 86.7 ⫾ 4.1 94.1 ⫾ 2.2 87.9 ⫾ 4.6 81.9 ⫾ 7.0 89.9 ⫾ 4.8
85.7 ⫾ 3.1 88.5 ⫾ 2.4* 89.4 ⫾ 2.0 91.7 ⫾ 1.4* 89.0 ⫾ 2.2 89.5 ⫾ 3.1 88.9 ⫾ 2.5 81.2 ⫾ 4.8 91.0 ⫾ 2.4 84.0 ⫾ 4.7 84.5 ⫾ 4.5 79.7 ⫾ 6.1
73.4 ⫾ 8.9 66.9 ⫾ 6.8* 69.0 ⫾ 5.5* 72.5 ⫾ 4.1* 69.5 ⫾ 5.9* 78.3 ⫾ 5.1 77.7 ⫾ 6.5 71.3 ⫾ 13.4 75.9 ⫾ 7.4* 76.2 ⫾ 11.1 83.6 ⫾ 7.3 70.3 ⫾ 12.1*
80.5 ⫾ 3.8 80.1 ⫾ 3.5 79.7 ⫾ 3.1 81.8 ⫾ 2.4 78.9 ⫾ 3.1 79.2 ⫾ 4.9 82.7 ⫾ 3.3 81.3 ⫾ 4.8 84.5 ⫾ 3.7 79.6 ⫾ 5.8 86.8 ⫾ 4.1 79.4 ⫾ 5.6
84.7 ⫾ 3.0 85.6 ⫾ 2.5 89.7 ⫾ 1.7** 90.9 ⫾ 1.4** 90.1 ⫾ 3.0** 90.6 ⫾ 2.1** 89.5 ⫾ 2.4** 83.8 ⫾ 4.3 93.6 ⫾ 1.7** 88.9 ⫾ 3.4** 79.2 ⫾ 6.2 85.9 ⫾ 5.3
86.9 ⫾ 2.3 85.2 ⫾ 2.1 87.3 ⫾ 1.6 88.6 ⫾ 1.2 85.6 ⫾ 2.2 88.6 ⫾ 1.9 88.3 ⫾ 1.9 83.7 ⫾ 3.7 92.2 ⫾ 1.6 87.6 ⫾ 3.1 85.2 ⫾ 3.9 86.1 ⫾ 3.9
69.7 ⫾ 6.3*** 72.2 ⫾ 6.2 74.7 ⫾ 6.3*** 75.8 ⫾ 6.3*** 76.8 ⫾ 7.6 76.9 ⫾ 8.2*** 71.1 ⫾ 8.8*** 79.9 ⫾ 6.3 79.4 ⫾ 7.3*** 76.0 ⫾ 9.0 74.9 ⫾ 10.5 66.7 ⫾ 11.3***
89.4 ⫾ 2.0 90.0 ⫾ 1.4
75.0 ⫾ 4.3*
80.6 ⫾ 2.5
91.9 ⫾ 1.2** 89.1 ⫾ 1.2
75.3 ⫾ 6.1***
82.9 ⫾ 6.5 86.9 ⫾ 6.7 83.3 ⫾ 6.9 90.1 ⫾ 3.9 90.8 ⫾ 3.1 90.2 ⫾ 4.2 89.4 ⫾ 4.6
89.2 ⫾ 4.5 90.7 ⫾ 3.6 88.7 ⫾ 4.3 88.9 ⫾ 2.9 92.5 ⫾ 2.0 82.5 ⫾ 3.7* 85.5 ⫾ 4.4
68.0 ⫾ 14.2 72.7 ⫾ 10.6 93.3 ⫾ 6.5 69.8 ⫾ 7.3* 77.2 ⫾ 5.8* 55.7 ⫾ 7.9 76.5 ⫾ 8.1*
79.6 ⫾ 7.0 84.0 ⫾ 6.0 82.3 ⫾ 6.7 82.8 ⫾ 3.9 86.9 ⫾ 3.0 72.5 ⫾ 4.6 80.4 ⫾ 4.9
87.2 ⫾ 4.5 88.2 ⫾ 3.2 89.7 ⫾ 4.3** 88.9 ⫾ 2.8 91.8 ⫾ 2.1** 87.5 ⫾ 3.1** 88.5 ⫾ 3.8
87.1 ⫾ 3.9 88.3 ⫾ 3.2 91.2 ⫾ 3.4 87.6 ⫾ 2.3 90.6 ⫾ 1.8 82.1 ⫾ 2.8 85.2 ⫾ 3.3
72.0 ⫾ 11.2 76.0 ⫾ 14.4 67.1 ⫾ 12.4*** 74.3 ⫾ 10.6 82.5 ⫾ 6.6 62.4 ⫾ 11.2 83.5 ⫾ 7.0
88.6 ⫾ 2.7 89.0 ⫾ 6.8 93.5 ⫾ 4.3 82.4 ⫾ 5.5 85.1 ⫾ 5.1 87.5 81.8–94.1
90.8 ⫾ 2.1 93.8 ⫾ 3.0 89.7 ⫾ 4.3 86.9 ⫾ 3.8 85.3 ⫾ 4.1 88.9 79.7–93.8
72.5 ⫾ 6.9* 77.6 ⫾ 14.2 78.9 ⫾ 7.6* 73.9 ⫾ 9.9 67.4 ⫾ 9.7* 73.4 55.7–93.3
81.5 ⫾ 3.9 89.9 ⫾ 6.1 85.6 ⫾ 4.8 80.1 ⫾ 5.2 74.0 ⫾ 5.6 80.6 72.5–89.9
90.3 ⫾ 1.8** 91.1 ⫾ 3.8 92.6 ⫾ 3.4 86.0 ⫾ 3.7 88.2 ⫾ 3.6** 88.9 79.2–93.6
89.0 ⫾ 1.7 NA NA 86.0 ⫾ 3.4 84.6 ⫾ 3.2 87.3 82.1–92.2
78.5 ⫾ 6.7*** NA NA 74.9 ⫾ 7.3*** 67.9 ⫾ 11.5 74.9 62.4–83.5
Difference statistically significant compared to persons aged 18 –34 years; **difference statistically significant compared to persons with ⱕhigh school education; ***difference statistically significant compared to persons with health insurance. CI, confidence interval; HS, high school; NA, not available because of insufficient sample size.
*
Consistent with national and state level studies,14,15 few differences were found in the largest metropolitan areas among whites, blacks, and Hispanics for receipt of any of the four cancer screening tests. The small number of areas with sufficient data on colorectal cancer screening among Hispanics prevents drawing meaningful conclusions about this population. Comparisons between blacks and whites in this study should be interpreted with caution. Understanding the relationship between race/ethnicity and screening independent of socioeconomic status is difficult. Minorities often are over-represented in lower socioeconomic groups and have differential access to services,51 and healthcare delivery and access (e.g., having a usual source of care and health insurance) greatly increase the probability of receiving cancer screening tests.1,14,15 We confirmed findings from previous research on the association between lack of health insurance and lower prevalence of cancer screening,14,15 despite small sample sizes in many areas. Such findings suggest that efforts to increase cancer screening in underserved populations will benefit from broader 306
systematic changes that extend health insurance coverage. It should also be acknowledged that cultural and logistical factors are important determinants for receipt of cancer screening tests.52 Comparing these findings with other local surveys is difficult because of differing age groups and frequency for receiving screening, but our estimates were generally similar to local estimates from Los Angeles,18,19 Salt Lake City,24 Boston,22 and Boise23 during this same time period. This study had several limitations. Validity of selfreported county of residence was unknown, possibly resulting in misclassification of respondents. People without telephones were excluded from the surveys. Although 95% of U.S. households have telephones, people living in households without telephones are more likely to be younger, black, or Hispanic and to have lower income levels.53 Response rates by urban area could not be determined. Typical of other surveys,54 response rates declined from 1997 to 1999, and the effect on our estimates is not known. We were also
American Journal of Preventive Medicine, Volume 24, Number 4
Table 4. Demographic estimates of fecal occult blood testing and sigmoidoscopy for the largest U.S. metropolitan areas
Metropolitan area Los Angeles– Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA–NJ Washington DC– MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Minneapolis– St. Paul MN–WI Nassau–Suffolk NY St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg– Clearwater FL Seattle–Bellevue– Everett WA Cleveland–Lorain– Elyria OH Miami FL Newark NJ Median Range
FOBT
Sigmoidoscopy
Age (years) % ⴞ 95% CI
Age (years) % ⴞ 95% CI
Gender % ⴞ 95% CI
Education % ⴞ 95% CI
>65
50–64
>65
Male
Female
8.7 ⫾ 2.9
21.3 ⫾ 5.6*
29.9 ⫾ 5.7
41.7 ⫾ 6.8*
42.6 ⫾ 6.9
28.9 ⫾ 5.4** 26.4 ⫾ 6.3
42.6 ⫾ 5.9***
22.4 ⫾ 4.4 11.8 ⫾ 4.0 25.6 ⫾ 3.5 21.6 ⫾ 4.2 22.3 ⫾ 3.7
26.2 ⫾ 4.9 24.2 ⫾ 5.9* 30.7 ⫾ 3.6 27.9 ⫾ 4.4 31.1 ⫾ 4.9*
28.7 ⫾ 4.8 27.7 ⫾ 5.7 30.5 ⫾ 3.7 26.8 ⫾ 4.6 35.9 ⫾ 4.5
40.2 ⫾ 5.7* 36.3 ⫾ 6.6 36.7 ⫾ 3.8 35.1 ⫾ 4.8 44.7 ⫾ 5.2
39.7 ⫾ 6.3 35.0 ⫾ 7.5 39.6 ⫾ 4.4 36.8 ⫾ 5.4 43.8 ⫾ 5.5
29.8 ⫾ 4.4 29.3 ⫾ 5.2 28.4 ⫾ 3.1** 26.2 ⫾ 4.0** 35.7 ⫾ 4.4
31.4 ⫾ 5.1 29.7 ⫾ 6.3 28.2 ⫾ 3.8 26.6 ⫾ 4.2 34.6 ⫾ 4.4
37.5 ⫾ 5.5 33.7 ⫾ 6.2 37.6 ⫾ 3.7*** 36.8 ⫾ 5.2*** 42.1 ⫾ 4.8
18.0 ⫾ 4.2 11.1 ⫾ 4.4 16.6 ⫾ 4.0 14.5 ⫾ 5.2 17.3 ⫾ 5.6 18.4 ⫾ 2.5
22.5 ⫾ 5.1 26.9 ⫾ 10.4* 18.9 ⫾ 5.4 20.4 ⫾ 7.7 20.9 ⫾ 5.5 28.4 ⫾ 3.4*
31.6 ⫾ 5.1 34.2 ⫾ 7.5 35.6 ⫾ 5.5 20.2 ⫾ 6.5 24.8 ⫾ 6.5 37.0 ⫾ 3.2
46.8 ⫾ 6.2* 36.5 ⫾ 9.6 53.2 ⫾ 7.4* 35.0 ⫾ 9.4 41.1 ⫾ 6.7* 47.5 ⫾ 3.9*
45.8 ⫾ 6.5 39.5 ⫾ 9.2 48.8 ⫾ 7.3 31.4 ⫾ 9.3 37.5 ⫾ 7.7 43.7 ⫾ 3.7
32.6 ⫾ 4.9** 31.0 ⫾ 7.4 37.1 ⫾ 5.4 21.2 ⫾ 6.2 29.1 ⫾ 5.9 39.7 ⫾ 3.3
32.8 ⫾ 5.4 31.3 ⫾ 9.0 41.8 ⫾ 6.4 21.9 ⫾ 7.7 28.0 ⫾ 6.6 39.5 ⫾ 3.5
44.6 ⫾ 5.9*** 37.6 ⫾ 7.8 43.4 ⫾ 6.4 28.8 ⫾ 7.5 36.3 ⫾ 6.6 43.4 ⫾ 3.4
19.6 ⫾ 6.5 17.2 ⫾ 5.2 21.2 ⫾ 3.7 15.8 ⫾ 4.9 18.5 ⫾ 5.7
27.8 ⫾ 8.2 19.0 ⫾ 5.3 36.3 ⫾ 4.7* 27.2 ⫾ 6.1* 36.3 ⫾ 6.3*
36.4 ⫾ 8.1 23.0 ⫾ 5.4 27.3 ⫾ 4.2 29.3 ⫾ 6.1 28.1 ⫾ 6.8
36.5 ⫾ 8.6 29.3 ⫾ 6.1 40.7 ⫾ 4.8* 37.9 ⫾ 6.3 37.5 ⫾ 6.2
42.4 ⫾ 9.3 27.6 ⫾ 6.5 42.1 ⫾ 5.2 40.9 ⫾ 7.4 36.0 ⫾ 7.4
31.3 ⫾ 7.3 24.7 ⫾ 5.3 26.3 ⫾ 3.7** 28.5 ⫾ 5.4** 31.4 ⫾ 5.6
31.4 ⫾ 8.7 22.8 ⫾ 5.4 29.0 ⫾ 4.2 31.8 ⫾ 5.4 35.5 ⫾ 6.6
40.7 ⫾ 8.1 30.3 ⫾ 6.5 39.5 ⫾ 4.9*** 37.2 ⫾ 7.4 31.5 ⫾ 6.3
22.6 ⫾ 4.1
33.2 ⫾ 5.2*
32.9 ⫾ 4.6
39.1 ⫾ 5.6
38.8 ⫾ 6.0
32.8 ⫾ 4.3
31.5 ⫾ 6.3
37.7 ⫾ 4.4
21.1 ⫾ 7.7
25.8 ⫾ 6.8
38.8 ⫾ 8.3
48.1 ⫾ 7.7
50.2 ⫾ 9.3
37.9 ⫾ 7.1
39.4 ⫾ 7.6
49.0 ⫾ 8.9
12.0 ⫾ 4.9 18.8 ⫾ 5.9 18.4 8.7–25.6
16.0 ⫾ 5.4 24.7 ⫾ 6.7 26.2 16.0–36.3
23.2 ⫾ 7.6 30.0 ⫾ 6.7 29.9 20.2–38.8
37.3 ⫾ 8.1 36.1 ⫾ 7.5 37.9 29.3–53.2
34.9 ⫾ 9.8 42.4 ⫾ 8.6 39.7 27.6–50.2
26.0 ⫾ 6.7 25.8 ⫾ 5.9* 29.3 21.2–39.7
27.3 ⫾ 7.7 23.9 ⫾ 6.8 31.3 21.9–41.8
33.0 ⫾ 8.5 40.4 ⫾ 7.1 37.6 28.8–49.0
50–64
>HS
*
Difference is statistically significant compared to persons aged 50 – 64 years; **difference is statistically significant compared to males; difference is statistically significant compared to persons with ⱕhigh school education. CI, confidence interval; FOBT, fecal occult blood testing; HS, high school. ***
unable to examine trends between 1997 and 1999 because of data pooling. Self-reports overestimate receipt of breast, cervical, and colon cancer screening.55–57 Although we used the term screening, undoubtedly some respondents reported procedures or tests that were done for diagnostic purposes, which would contribute to overestimating the true extent of screening. The effect of pooling the slightly different questions on sigmoidoscopy in 1997 and 1999 is unclear, although any effect is likely to be consistent across areas. The role that local variation in uninsured rates had on our estimates is not known, but is likely to affect differences across areas because of the strong association between health insurance status and cancer screening.14,15,58 Differences by metropolitan areas may reflect variations by age, race/ethnicity, and education level.59 – 62 Furthermore, our estimates were for entire metropolitan areas, but screening estimates are likely to vary by geographic districts within each areas (e.g., central cities versus suburbs). Despite pooling data across
years, the numbers of respondents was small for certain subpopulations in many areas, reducing the precision of our estimates. State sample sizes vary widely because of differences in financial support.48 As a result, small metropolitan areas in some states had large sample sizes, while other states had larger metropolitan areas with smaller samples, preventing us from including areas such as San Antonio TX and San Francisco CA because of insufficient sample size. Despite these limitations, the BRFSS offers important benefits for making metropolitan area estimates. The system uses standardized protocols and methodology for data collection, and no new data collection is required. An additional advantage is timeliness, as BRFSS final annual data are available within 4 to 6 months of collection. It is a flexible system that allows for the incorporation of state- or local-area–relevant questions on cancer screening. Perhaps most important, use of existing BRFSS data would result in substantial cost savings to metropolitan area health departments around the United States. The annualized cost Am J Prev Med 2003;24(4)
307
for the Los Angeles County CA survey was estimated at $900,00019; in contrast, reweighting the entire BRFSS to provide metropolitan estimates for the entire United States cost about $100,000. In summary, BRFSS data could be used to help improve cancer prevention and to control efforts at the metropolitan level. The system provides an inexpensive, timely, and readily available source of data that could be used to examine current estimates of cancer screening, evaluate the impact of local efforts, and provide a baseline against which data from more localized surveys can be compared. Local health departments should consider routinely using these reweighted data as part of their ongoing surveillance activities for cancer screening and other measures (e.g., risk factors such as smoking and physical activity) that are already being routinely collected as part of the BRFSS. We are grateful to the state BRFSS coordinators, Donna Brogan, Jill Dever, Glen Laird, and Bill Scott, for their assistance.
References 1. Cramer DW. The role of cervical cytology in the declining morbidity and mortality of cervical cancer. Cancer 1974;34:2018 –27. 2. Benedet JL, Anderson GH, Matistic JP. A comprehensive program for cervical cancer detection and management. Am J Obstet Gynecol 1992;166: 1254 –9. 3. Gotzsche PC, Olsen O. Is screening for breast cancer with mammography justifiable? Lancet 2000;355:129 –34. 4. Kerlikowske K, Grady D, Rubim SM, et al. Efficacy of screening mammography: a meta-analysis. JAMA 1995;273:149 –54. 5. U.S. Preventive Services Task Force. Guide to clinical preventive services. 2nd ed. Baltimore, MD: Williams & Wilkins, 1996. 6. Kronborg O, Fenger C, Olsen J, et al. Randomised study of screening for colorectal cancer with faecal-occult– blood screening. Lancet 1996;348: 1467–71. 7. Hardcastle JD, Chamberlain JO, Robinson MHE, et al. Randomised controlled trial of faecal-occult– blood screening for colorectal cancer. Lancet 348;1472–7. 8. Mandel JS, Church TR, Ederer F, Bond JH. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood. J Natl Cancer Inst 1999;91:434 –7. 9. Selby JV, Freidman GD, Quesenberry CP, et al. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992;356:653–7. 10. Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy an colorectal cancer mortality. J Natl Cancer Inst. 1992;84:1572–5. 11. Slattery ML, Edwards SL, Ma KN, et al. Colon cancer screening, lifestyle, and risk of colon cancer. Cancer Causes Control 2000;11:555–63. 12. Rakowski W, Rimer BK, Bryant SA. Integrating behavior and intention regarding mammography by respondents in the 1990 National Health Interview Survey of health promotion and disease prevention. Public Health Rep 1993;108:605–24. 13. Pearlman DN, Rakowski W, Ehrich B, et al. Breast cancer screening practices among Black, Hispanic, and White women: reassessing differences. Am J Prev Med 1996;12:327–37. 14. Breen N, Wagener DK, Brown ML, et al. 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 –13. 15. Blackman DK, Bennett EM, Miller DS. Trends in self-reported use of mammograms (1989 –1997) and Papanicolaou tests (1991–1997)—Behavioral Risk Factor Surveillance System. MMWR Surveill Summ 1999;48:1–22.
308
16. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System. Available at www.cdc.gov/brfss. Accessed on March 31, 2002. 17. Health Resources and Services Administration. Community Health Status Indicators Project. Rockville, MD: Health Resources and Services Administration. Available at: www.communityhealth.hrsa.gov. Accessed on March 31, 2002. 18. Los Angeles County Department of Health Services. The health of Angelenos: a comprehensive report of the health of the residents of Los Angeles County. Los Angeles, CA: Los Angeles County Department of Health Services, 2000. Available at: http://lapublichealth.org/ha/reports/ angelenos/hofa.pdf. Accessed on April 12, 2002. 19. Simon PA, Wold CM, Cousineau MR, et al. Meeting the data needs of a local heatlh department: the Los Angeles County health survey. Am J Public Health 2001;91:1950 –2. 20. Northridge ME, Morabia A, Ganz ML, et al. Contribution of smoking to excess mortality in Harlem. Am J Epidemiol 1998;147:250 –8. 21. Parchman M, Byrd T. Access to and use of ambulatory health care by a vulnerable Mexican American population on the U.S.–Mexico border. J Health Care Poor Underserved 2001;12:404 –14. 22. Massachusetts Department of Public Health. Behavioral Risk Factor Survey 1994 —1996 city report: Boston. Boston: Massachusetts Department of Public Health, 1998. 23. Idaho Department of Health and Welfare. Idaho Behavioral Risk Factor Surveillance System, 1999 survey data. Boise: Idaho Department of Health and Welfare, 2000. 24. Utah Department of Health. Utah Behavioral Risk Factor Surveillance System local health district report. Salt Lake City: Utah Department of Health, 2000. 25. Virginia Department of Health. Healthy Virginia communities: a report on year 2000 health status and risk reduction indicators for the Commonwealth of Virginia and health districts. Richmond: Virginia Department of Health, 1997. 26. Wisconsin Department of Health and Family Services. County and regional estimates of behavioral risk factors and health screening practices, 1989 – 1994. Madison: Wisconsin Department of Health and Family Services, 1997. 27. U.S. Census Bureau. Census 2000 summary file 1 (SF 1) 100-percent data. Washington, DC: U.S. Census Bureau. Available at: http://factfinder. census.gov/home/en/sf1.html. Accessed on April 12, 2002. 28. Burack RC, Gimotty PA. Promoting screening mammography in inner-city settings. The sustained effectiveness of computerized reminders in a randomized controlled trial. Med Care 1997;35:921–31. 29. Doyle JP, Parker RM, Jacobson TA, McNagny SE. Breast and cervical cancer screening in an inner-city medical walk-in clinic: taking advantage of an often missed opportunity. Am J Prev Med 1996;12:345–50. 30. Taylor VM, Thompson B, Montano DE, Mahloch J, Johnson K, Li S. Mammography use among women attending an inner-city clinic. J Cancer Educ 1998;13:96 –101. 31. Weber BE, Reilly BM. Enhancing mammography use in the inner city. A randomized trial of intensive case management. Arch Intern Med 1997; 157:2345–9. 32. Whitman S, Ansell D, Lacey L, et al. Patterns of breast and cervical cancer screening and three public health centers in an inner-city urban area. Am J Public Health 1991;81:1651–3. 33. Leviton LC, Snell E, McGinnis M. Urban issues in health promotion strategies. Am J Public Health 2000;90:863–6. 34. Freudenberg N. Time for a national agenda to improve the health of urban populations. Am J Public Health 2000;90:837–40. 35. Fielding JE, Sutherland CE, Halfon N. Community health report cards: results of a national survey. Am J Prev Med 1999;17:79 –86. 36. Institute of Medicine. The future of public health. Washington DC: National Academy Press, 1988. 37. Marcus AC, Crane LA. A review of cervical cancer intervention research: implications for public health programs and future research. Prev Med 1998;27:13–31. 38. Rimer BK. Interventions to enhance cancer screening—a brief review of what works and what is on the horizon. Cancer 1998;83(suppl S):1770 –4. 39. Champion VL, Ray DW, Heilman DK, Springston JK. A tailored intervention for mammography among low-income African-American women. J Psychosocial Oncol 2000;18:1–13. 40. Shenson D, Cassarino L, DiMartino D. Improving access to mammograms through community-based influenza clinics—a quasi-experimental study. Am J Prev Med 2001;20:97–102.
American Journal of Preventive Medicine, Volume 24, Number 4
41. Hiatt RA, Pasick RJ, Stewart S, et al. Community-based cancer screening for underserved women: design and baseline findings from the breast and cervical cancer intervention study. Prev Med 2001;33:190 –203. 42. U.S. Department of Health and Human Services. Healthy People 2010, 2nd ed. Washington DC: U.S. Department of Health and Human Services, 2000. 43. Centers for Disease Control and Prevention. 1991 Behavioral Risk Factor Surveillance System quality control report. Atlanta GA: Centers for Disease Control and Prevention, 1992. 44. Centers for Disease Control and Prevention. 1999 Behavioral Risk Factor Surveillance System quality control report. Atlanta GA: Centers for Disease Control and Prevention, 2000. 45. Centers for Disease Control and Prevention. Cigarette smoking in 99 metropolitan areas—United States, 2000. MMWR Morb Mortal Wkly Rep 2001;50:1107–13. 46. Marks JS, Hogelin GC, Gentry EM, et al. The behavioral risk factor surveys: I. state-specific prevalence estimates of behavioral risk factors. Am J Prev Med 1985;1:1–8. 47. Powell-Griner E, Anderson JE, Murphy W. State- and sex-specific prevalence of selected characteristics—Behavioral Risk Factor Surveillance System, 1994 and 1995. MMWR Morb Mortal Wkly Rep 1997;46:1–31. 48. Nelson DE, Holtzman D, Waller M, et al. Objectives and design of the Behavioral Risk Factor Surveillance System. In: Proceedings of the 1998 American Statistical Association Section on Survey Research Methods, Dallas, TX, August 9 –13, 1998. Alexandria VA: American Statistical Association, 1998:214 – 8. 49. U.S. Office of Management and Budget. Metropolitan areas defined. Washington, DC: Office of Management and Budget, June 30, 1993. (1990 CPH-S-11). 50. Rathore S, McGreevey JD, Schulman KA, Atkins D. Mandated coverage for cancer-screening services: whose guidelines do states follow? Am J Prev Med 2000;19:71–8. 51. Mills RJ. Health insurance coverage: 2000. Current Population Reports P60-215. Washington, DC: U.S. Census Bureau, 2001.
52. Kelaher M., Stellman JM. The impact of Medicare funding on the use of mammography among older women: implications for improving access to screening. Prev Med 2000;31:658 –64. 53. U.S. Bureau of the Census. Statistical brief: phoneless in America. Washington, DC: U.S. Department of Commerce, Economics and Statistics Administration, 1994. 54. Massey JT, O’Connor D, Krotki K. Response rates in random digit dialing (RDD) telephone surveys. In: Proceedings of the 1997 American Statistical Association Section on Survey Research Methods, Anaheim, CA, August 10 –14, 1997. Alexandria, VA: American Statistical Association, 1997:707–12. 55. Gordon NP, Hiatt RA, Lampert DI. Concordance of self-reported data and medical record audit for six cancer screening procedures. J Natl Cancer Inst 1993;85:556 –70. 56. Hiatt RA, Perez-Stable EJ, Quesenberry C, et al. Agreement between self-reported early cancer detection practices and medical audits among Hispanic and non-Hispanic white health plan members in Northern California. Prev Med 1995;24:278 –85. 57. Warnecke RB, Sudman S, Johnson TP, et al. Cognitive aspects of recalling and reporting health-related events: Papanicolaou smears, clinical breast exam, and mammogram. Am J Epidemiol 1997;146:982–92. 58. Frazier E, Jiles R, Mayberry R. Use of screening mammography and clinical breast examinations among black, Hispanic, and white women. Prev Med 1996;25:118 –25. 59. U.S. Census Bureau. Metropolitan area rankings for persons 65 years of age and over, 1996. Press release, July 21, 1998. Available at: www.census.gov/ Press-Release/metro09.prn. Accessed on April 16, 2002. 60. U.S. Census Bureau. Metropolitan area rankings for Hispanic population, 1996. Press Release, July 21, 1998. Available at: www.census.gov/PressRelease/metro13.prn. Accessed on April 16, 2002. 61. U.S. Census Bureau. Metropolitan area rankings for Black population, 1996. Press Release, July 21, 1998. Available at: www.census.gov/PressRelease/metro11.prn. Accessed on April 16, 2002. 62. Gaquin DA, DeBrandt KA, eds. Education statistics of the United States. 3rd ed. Lanham, MD: Bernan, 2001.
Am J Prev Med 2003;24(4)
309
Objectives:
To provide estimates of breast, cervical, and colorectal cancer screening for metropolitan areas in the United States.
Methods:
Behavioral Risk Factor Surveillance System (BRFSS) data from 1997 to 1999 were reweighted and analyzed for 69 U.S. metropolitan areas for the receipt of a Papanicolaou (Pap) test (ages ⱖ18 years); mammography (ages ⱖ40 years); fecal occult blood testing and sigmoidoscopy (ages ⱖ50 years). Stratified analyses by demographics were performed for 25 metropolitan areas with populations of ⱖ1.5 million.
Results:
Metropolitan estimates ranged from 64.6% to 82.0% for mammography and from 77.2% to 91.7% for Pap tests. There was much greater variability in estimates for colorectal cancer screening, with a 3.6-fold difference in the range of estimates for fecal occult blood testing (9.9% to 35.2%) and a 2.5-fold difference for sigmoidoscopy (17.3% to 43.3%). In the 25 largest areas, prevalence of cancer screening was generally lower for persons with a high school education or less and for those without health insurance. Compared with women aged 50 to 64 years, mammography estimates were lower for women aged 40 to 49 years in 13 of the 25 metropolitan areas. Pap testing was less common among women aged ⱖ65 years, and colorectal cancer screening was less common for persons aged 50 to 64 years.
Conclusions: Estimates of cancer screening varied substantially across metropolitan areas. Increased efforts to improve cancer screening are needed in many urban areas, especially for colorectal cancer screening. The BRFSS is a useful, inexpensive, and timely resource for providing metropolitan-area cancer screening estimates and may be used in the future to guide local or county-level screening efforts. (Am J Prev Med 2003;24(4):301–309)
Introduction
S
creening has been found to reduce morbidity and premature mortality from cervical, breast, and colorectal cancer.1–11 The National Health Interview Survey12–14 and the Behavioral Risk Factor Surveillance System (BRFSS)15,16 provide national and state estimates for receipt of cancer screening among adults. Although there have been efforts in Los Angeles County and elsewhere,17–26 metropolitan-level estimates for cancer screening are not widely available. Conducting surveys is expensive, and many health departments lack resources or infrastructure to collect such data. The lack of urban data on cancer screening is unfortunate. Sociodemographics of urban areas differ From the Division of Cancer Control and Population Sciences, National Cancer Institute (Nelson, Marcus, Meissner), Bethesda, Maryland; Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention (Bolen), Atlanta, Georgia; and Research Triangle Institute (Wells), Atlanta, Georgia Address correspondence to: David E. Nelson, MD, MPH, Centers for Disease Control and Prevention, 4770 Buford Hwy, NE, Mailstop K-50, Atlanta GA 30341-3717. E-mail: [email protected].
Am J Prev Med 2003;24(4) Published by Elsevier Inc.
from those of states; most urban areas have a higher concentration of racial/ethnic minority populations, who also are over-represented among the poor.27 Participation in cancer screening programs is low among many inner-city populations,28 –32 and national and state estimates may mask such differences.33,34 Relevant data can be used to help empower communities to address their own health issues,33–36 especially as many efforts to improve cancer screening occur locally,37– 41 and these data can help track progress toward Healthy People 2010 goals.42 One source for obtaining metropolitan area data is the BRFSS,43,44 as the Centers for Disease Control and Prevention (CDC) recently reweighted BRFSS data for metropolitan areas based on county of residence.45 For this study, the overall prevalence estimates for the receipt of a Papanicolaou (Pap) test, mammography, and colorectal cancer screening (fecal occult blood test [FOBT] and sigmoidoscopy) were determined for metropolitan areas with sufficient sample sizes. Screening estimates were then compared by age, gender, race/ethnicity, education level, and health insurance status for the 25 largest metropolitan areas.
0749-3797/03/$–see front matter doi:10.1016/S0749-3797(03)00024-2
301
Methods Data used in this study came from the BRFSS. Briefly, the BRFSS is a state-based system of health surveys of adults that is coordinated by CDC.46 – 48 Health-related data are obtained each year through telephone surveys of randomly selected persons aged ⱖ18 years. We analyzed data from 1997 to 1999, the most recent years with metropolitan-level data on cancer screening. Data on mammography and cervical cancer screening were obtained in each of these years, but colorectal cancer screening data was only available for 1997 and 1999. The total sample size ranged from 133,048 in 1997 to 159,921 in 1999. Median state-level annual response rates, based on persons actually reached by telephone, were 76.8% in 1997, 73.4% in 1998, and 68.4% in 1999. Because these were state-based surveys, response rates by metropolitan areas were unavailable.
Cancer Screening Definitions For the purposes of this study, mammography screening was the percentage of women aged ⱖ40 years who (1) answered “yes” to “A mammogram is an x-ray of each breast to look for breast cancer. Have you ever had a mammogram?” and (2) who reported receiving a mammogram sometime within the past 2 years when asked “How long has it been since you had your last mammogram?” Cervical cancer screening was the percentage of women aged ⱖ18 years (1) who answered “yes” to “A Pap smear is a test for cancer of the cervix. Have you ever had a Pap smear?”; (2) who reported receiving the test sometime within the past 3 years when asked “How long has it been since had your last Pap smear?”; and (3) who answered “no” to the question “Have you ever had a hysterectomy?” We assessed two measures for colorectal cancer screening among persons aged ⱖ50 years: FOBT and sigmoidoscopy. Receipt of FOBT was based on the percentage of respondents (1) who answered “yes” to “A blood stool test is a test that may use a special kit at home to determine whether the stool contains blood. Have you ever had this test using a home kit?”; and (2) who had reported receiving it within the past year or more recently when asked “When did you have your last blood stool test using a home kit?” The question on receipt of sigmoidoscopy was slightly different in 1997 than in 1999. In 1997, respondents were asked, “A sigmoidoscopy or proctoscopy is when a tube is inserted in the rectum to view the bowel for signs of cancer and other health problems. Have you ever had this exam?” In 1999, the question was changed to receiving sigmoidoscopy or colonoscopy. Persons who responded “yes” to this question, and who reported having received the test within the past ⱕ5 years when asked “When did you have your last sigmoidoscopy or proctoscopy [colonoscopy]?” were considered to have received colorectal cancer screening. We use the word “sigmoidoscopy” to describe this type of colorectal cancer screening across both survey years.
Metropolitan Areas and Statistical Analyses Self-reported county of residence was used to classify respondents as residents of metropolitan statistical areas, primary metropolitan statistical areas, or New England county metropolitan areas using census definitions.49 Data were then reweighted, based on the age, gender, and race/ethnicity
302
distribution, for each metropolitan area using intercensal estimates. All counties within the metropolitan areas were included regardless of state boundaries. To increase precision of the estimates, data were pooled across years (3 years for mammography and cervical cancer screening, 2 years for colorectal cancer screening). Analyses were limited to metropolitan areas with ⱖ300 respondents for all measures or that had a population of ⱖ1.5 million residents. Sixty-nine areas met our inclusion criteria. For the 25 metropolitan areas with intercensal population estimates of ⱖ1.5 million, additional analyses were conducted by age (18 to 34, 35 to 64, and ⱖ65 years for Pap test; 40 to 49, 50 to 64, and ⱖ65 years for mammography); gender (colorectal measures); race/ethnicity (non-Hispanic white, nonHispanic black, Hispanic); education (high school or less, more than high school), and health insurance status. Because of small numbers, analyses were restricted to groups with ⱖ50 persons. Due to overall small numbers, subgroup analyses for colorectal cancer screening were not performed for Oakland, Orange County, Riverside–San Bernardino, and San Diego in California. All statistical analyses were conducted in SAS (SAS Institute, Cary NC, 2001) and SUDAAN (Research Triangle Institute, Research Triangle Park NC, 2000). To provide additional context, pooled state estimates were analyzed for each measure. Metropolitan areas were grouped by census region (Northeast, Midwest, South, and West) and calculated for regional median and range values. For the 25 largest metropolitan areas, differences among subgroups groups within each area were assessed using two-sample t -tests for differences between estimates. Because of the potential problem of multiple comparisons, differences were considered significant only when p ⬍0.01.
Results Overall Estimates for 69 Metropolitan Areas Metropolitan estimates for mammography, cervical cancer screening, and colorectal cancer screening for the 69 metropolitan areas are listed in Table 1. Regional medians for both mammography (range for medians, 73.5% to 76.4%) and cervical cancer screening (range for medians, 85.1% to 86.9%) were similar across regions. Especially low estimates for both mammography and cervical cancer screening were noted for Charleston WV and Huntington–Ashland WV–KY–OH. There was much greater variation for estimates of colorectal cancer screening (Table 1). Regional median estimates for FOBT ranged from 18.0% to 24.3%, with lower estimates found in the South and West. However, there was more than a three-fold difference across individual metropolitan areas (range, 9.9% to 35.2%), with 11 areas having FOBT estimates of ⱕ15%. Regional estimates for sigmoidoscopy were similar, ranging from 32.7% in the Midwest to 33.9% in the Northeast. Estimates for individual metropolitan areas varied from 17.3% to 43.3%, with the lowest estimates in Huntington–Ashland WV–KY–OH, Oklahoma City OK, and Omaha NE–IA.
American Journal of Preventive Medicine, Volume 24, Number 4
Table 1. Overall estimates for cancer screening by metropolitan area,a 1997–1999 Colorectal cancer screening Region and metropolitan area Northeast Bergen–Passaic NJ Boston MA–NH Burlington VT Hartford CT Monmouth–Ocean NJ Nassau–Suffolk NY New Haven–Bridgeport– Stamford–Waterbury– Danbury CT Newark NJ New York NY Philadelphia PA–NJ Pittsburgh PA Providence–Warwick–Pawtucket RI Northeast median Range Midwest Chicago IL Cincinnati OH–KY–IN Cleveland–Lorain–Elyria OH Columbus OH Des Moines IA Detroit MI Indianapolis IN Kansas City MO–KS Milwaukee–Waukesha WI Minneapolis–St. Paul MN–WI Omaha NE–IA St. Louis MO–IL Sioux Falls SD Wichita KS Midwest median Range South Atlanta GA Baltimore MD Birmingham AL Charleston WV Charlotte–Gastonia–Rock Hill NC–SC Dallas TX Dover DE Ft. Lauderdale FL Greensboro–Winston–Salem– High Point NC Greenville–Spartanburg– Anderson SC Houston TX Huntington–Ashland WV– KY–OH Jacksonville FL Little Rock–North Little Rock AR Louisville KY–IN Memphis TN–AR–MS Miami FL Nashville TN New Orleans LA Norfolk–Virginia Beach– Newport News VA–NC
Mammography % ⴞ 95% CI
Pap test % ⴞ 95% CI
FOBT % ⴞ 95% CI
Sigmoidoscopy % ⴞ 95% CI
73.8 ⫾ 4.8 82.0 ⫾ 1.6 75.6 ⫾ 3.0 80.1 ⫾ 2.9 70.7 ⫾ 4.7 76.3 ⫾ 4.8 78.8 ⫾ 2.7
83.5 ⫾ 4.0 87.6 ⫾ 1.3 88.0 ⫾ 2.5 87.3 ⫾ 2.3 83.1 ⫾ 3.4 86.6 ⫾ 3.3 85.4 ⫾ 2.1
24.5 ⫾ 5.9 28.0 ⫾ 2.5 28.7 ⫾ 3.8 26.9 ⫾ 4.2 23.2 ⫾ 4.8 23.1 ⫾ 5.1 25.7 ⫾ 3.3
37.8 ⫾ 6.3 33.4 ⫾ 2.6 33.3 ⫾ 4.0 36.9 ⫾ 4.5 31.7 ⫾ 5.5 36.4 ⫾ 5.8 38.3 ⫾ 3.8
76.4 ⫾ 4.0 76.4 ⫾ 2.9 78.9 ⫾ 2.3 73.6 ⫾ 3.3 81.3 ⫾ 1.7 76.4 70.7–82.0
89.1 ⫾ 2.8 83.1 ⫾ 2.1 84.8 ⫾ 2.1 80.3 ⫾ 2.7 86.1 ⫾ 1.4 85.1 80.3–89.1
21.5 ⫾ 4.4 24.2 ⫾ 3.3 24.7 ⫾ 3.0 21.8 ⫾ 3.0 23.8 ⫾ 2.1 24.3 21.5–28.7
32.8 ⫾ 5.0 34.0 ⫾ 3.7 30.9 ⫾ 3.3 33.9 ⫾ 4.4 36.1 ⫾ 2.5 33.9 30.9–38.3
72.1 ⫾ 2.5 76.3 ⫾ 3.9 76.8 ⫾ 4.3 73.4 ⫾ 6.2 76.7 ⫾ 3.6 76.5 ⫾ 76.5 70.0 ⫾ 5.2 75.7 ⫾ 2.6 73.2 ⫾ 4.2 68.2 ⫾ 2.0 72.9 ⫾ 2.9 74.8 ⫾ 3.2 70.2 ⫾ 4.3 73.7 ⫾ 4.2 73.5 68.2–76.8
85.8 ⫾ 1.6 83.9 ⫾ 3.6 89.2 ⫾ 2.9 89.3 ⫾ 3.3 86.2 ⫾ 2.9 86.6 ⫾ 2.0 87.1 ⫾ 3.4 89.1 ⫾ 1.7 84.8 ⫾ 3.0 88.0 ⫾ 1.1 86.0 ⫾ 2.2 86.3 ⫾ 2.3 85.3 ⫾ 3.0 87.7 ⫾ 3.0 86.5 83.9–89.3
17.5 ⫾ 3.5 22.2 ⫾ 4.3 23.4 ⫾ 5.1 22.6 ⫾ 6.2 28.4 ⫾ 5.0 20.0 ⫾ 3.2 13.3 ⫾ 3.9 23.4 ⫾ 3.2 12.3 ⫾ 4.1 22.7 ⫾ 2.1 23.8 ⫾ 3.3 18.0 ⫾ 3.7 15.2 ⫾ 3.7 24.3 ⫾ 5.1 22.4 12.3–28.4
31.7 ⫾ 4.4 26.4 ⫾ 4.7 43.3 ⫾ 5.7 28.0 ⫾ 6.2 33.8 ⫾ 5.3 38.4 ⫾ 4.0 26.5 ⫾ 5.2 35.5 ⫾ 3.7 34.8 ⫾ 5.6 41.5 ⫾ 2.5 24.0 ⫾ 3.5 25.9 ⫾ 4.0 29.0 ⫾ 5.1 33.6 ⫾ 5.6 32.7 24.0–43.3
74.0 ⫾ 3.1 81.3 ⫾ 2.3 80.5 ⫾ 4.0 67.7 ⫾ 4.8 74.5 ⫾ 4.3
90.6 ⫾ 1.7 89.8 ⫾ 1.8 89.1 ⫾ 2.9 78.5 ⫾ 4.1 87.1 ⫾ 3.3
17.5 ⫾ 3.2 28.2 ⫾ 3.0 13.5 ⫾ 4.1 11.0 ⫾ 3.3 17.5 ⫾ 3.5
42.5 ⫾ 4.5 33.6 ⫾ 3.2 30.5 ⫾ 5.0 27.3 ⫾ 5.2 36.5 ⫾ 6.3
71.8 ⫾ 4.3 77.0 ⫾ 3.3 78.7 ⫾ 4.2 72.3 ⫾ 5.2
85.0 ⫾ 3.2 87.2 ⫾ 2.3 86.8 ⫾ 3.3 91.0 ⫾ 3.3
16.8 ⫾ 4.4 17.1 ⫾ 3.7 23.9 ⫾ 4.9 35.2 ⫾ 5.2
25.8 ⫾ 5.4 31.4 ⫾ 4.4 36.0 ⫾ 6.2 33.3 ⫾ 5.1
76.1 ⫾ 3.6
86.7 ⫾ 3.0
18.5 ⫾ 3.9
29.5 ⫾ 4.6
72.2 ⫾ 4.2 64.6 ⫾ 6.2
82.7 ⫾ 3.2 77.2 ⫾ 5.8
20.7 ⫾ 3.1 15.5 ⫾ 4.5
35.1 ⫾ 5.9 23.9 ⫾ 5.4
81.7 ⫾ 4.5 70.4 ⫾ 4.5
89.6 ⫾ 3.6 82.2 ⫾ 4.1
19.3 ⫾ 5.3 15.0 ⫾ 3.8
29.4 ⫾ 5.9 35.1 ⫾ 5.1
74.2 ⫾ 3.9 73.4 ⫾ 3.7 74.6 ⫾ 4.5 77.7 ⫾ 3.3 70.4 ⫾ 4.4 76.0 ⫾ 4.2
87.0 ⫾ 3.1 89.6 ⫾ 2.2 83.4 ⫾ 3.1 89.9 ⫾ 2.3 84.8 ⫾ 3.2 89.1 ⫾ 2.8
21.2 ⫾ 5.7 12.9 ⫾ 2.6 13.9 ⫾ 3.7 14.5 ⫾ 3.4 18.0 ⫾ 4.2 17.2 ⫾ 4.7
26.0 ⫾ 4.7 29.2 ⫾ 4.8 29.9 ⫾ 5.8 27.6 ⫾ 4.9 33.9 ⫾ 5.5 34.4 ⫾ 5.9 (continued on next page)
Am J Prev Med 2003;24(4)
303
Table 1. (continued) Colorectal cancer screening Region and metropolitan area Oklahoma City OK Richmond–Petersburg, VA Tampa–St. Petersburg–Clearwater FL Tulsa OK Washington DC–MD–VA–WV West Palm Beach–Boca Raton FL Wilmington–Newark DE–MD South median Range West Albuquerque NM Boise City ID Denver CO Honolulu HI Las Vegas NV–AZ Los Angeles–Long Beach CA Oakland CA Orange County CA Phoenix–Mesa AZ Portland–Vancouver OR–WA Reno NV Riverside–San Bernardino CA Salt Lake City–Ogden UT San Diego CA Seattle–Bellevue–Everett WA Tucson AZ West median Range Nationwide median Range
Mammography % ⴞ 95% CI
Pap test % ⴞ 95% CI
FOBT % ⴞ 95% CI
Sigmoidoscopy % ⴞ 95% CI
68.0 ⫾ 3.5 76.1 ⫾ 4.8 78.9 ⫾ 3.3 69.7 ⫾ 4.1 76.7 ⫾ 2.6 80.5 ⫾ 4.2 79.7 ⫾ 2.8 74.6 64.6–81.7
86.4 ⫾ 2.6 91.7 ⫾ 2.8 84.9 ⫾ 3.1 86.3 ⫾ 3.0 86.9 ⫾ 2.2 82.7 ⫾ 3.9 87.8 ⫾ 1.8 86.9 77.2–91.7
9.9 ⫾ 2.5 18.4 ⫾ 5.3 28.6 ⫾ 4.4 21.2 ⫾ 4.0 25.8 ⫾ 3.0 30.0 ⫾ 5.3 20.8 ⫾ 3.0 18.0 9.9–35.2
17.3 ⫾ 3.1 37.8 ⫾ 6.3 33.4 ⫾ 4.6 28.8 ⫾ 4.6 39.4 ⫾ 3.5 35.1 ⫾ 5.6 41.8 ⫾ 3.6 33.3 17.3–42.5
75.0 ⫾ 2.8 66.9 ⫾ 3.3 74.7 ⫾ 3.3 76.5 ⫾ 3.0 67.4 ⫾ 4.2 73.1 ⫾ 3.4 79.8 ⫾ 5.3 78.6 ⫾ 5.3 79.1 ⫾ 5.0 76.6 ⫾ 2.7 70.4 ⫾ 4.4 66.1 ⫾ 5.5 68.0 ⫾ 3.2 70.8 ⫾ 5.4 72.4 ⫾ 2.7 78.1 ⫾ 5.0 73.9 66.1–79.8 74.8 64.6–82.0
86.0 ⫾ 2.1 85.2 ⫾ 2.2 87.8 ⫾ 2.2 86.8 ⫾ 2.0 81.3 ⫾ 3.2 82.8 ⫾ 2.4 90.8 ⫾ 3.3 86.9 ⫾ 3.7 83.4 ⫾ 3.7 89.3 ⫾ 1.7 87.0 ⫾ 2.7 82.6 ⫾ 3.9 80.9 ⫾ 2.4 84.1 ⫾ 3.9 88.0 ⫾ 1.7 82.6 ⫾ 3.7 85.6 80.9–90.8 86.4 77.2–91.7
22.3 ⫾ 3.6 17.8 ⫾ 3.2 26.8 ⫾ 4.2 20.7 ⫾ 3.1 10.8 ⫾ 3.3 14.2 ⫾ 3.0 NA NA 19.1 ⫾ 4.1 25.6 ⫾ 3.1 16.6 ⫾ 4.1 NA 17.2 ⫾ 3.0 NA 27.2 ⫾ 3.2 23.4 ⫾ 5.5 19.9 10.8–27.2 21.2 9.9–35.2
35.6 ⫾ 3.9 31.1 ⫾ 4.1 30.6 ⫾ 4.2 40.1 ⫾ 4.0 29.2 ⫾ 4.7 35.1 ⫾ 4.4 NA NA 32.7 ⫾ 4.8 37.3 ⫾ 3.6 25.4 ⫾ 4.9 NA 34.9 ⫾ 3.8 NA 35.6 ⫾ 3.6 30.7 ⫾ 5.6 33.8 25.4–40.1 33.4 17.3–43.3
a
Because of insufficient sample size, no estimates are available for metropolitan areas in Maine, Mississippi, Montana, New Hampshire, North Dakota, or Wyoming. Alaska has no defined metropolitan statistical areas. CI, confidence interval; FOBT, fecal occult blood test; NA, not available, as sample size ⬍300; Pap test, Papanicolaou test.
Selected Demographic Estimates for the 25 Largest Metropolitan Areas Further analyses of mammography and cervical cancer screening by selected demographic measures for the 25 largest metropolitan areas are presented in Table 2. Mammography varied by age; with ages 50 to 64 years as the baseline, estimates were significantly lower in 13 areas among women aged 40 to 49 and for three areas among women aged ⱖ65 years. We found no significant differences for mammography estimates between blacks and whites, but estimates were significantly lower for Hispanics compared to whites in Riverside–San Bernardino CA (data not shown). In all areas, persons with more than a high school education were more likely than those with less than a high school education to report mammography screening (data not shown). Health insurance was associated with receipt of mammography, as estimates were significantly lower for uninsured women in 13 of 17 areas for which these comparisons could be made. Cervical cancer screening was significantly less common among women aged ⱖ65 years than among 304
women aged 18 to 34 years in 13 of 25 areas (Table 3). Estimates for cervical cancer screening were significantly higher for blacks than whites in five areas (Atlanta GA, Baltimore MD, Chicago IL, Houston TX, and Washington DC–MD–VA–WV), but we found no differences between Hispanics and whites in metropolitan areas with adequate sample sizes for such comparisons (data not shown). Cervical cancer screening estimates were lower for persons with less than a high school education in all but one of the largest metropolitan areas, with the differences significant in 13 areas. As with mammography, health insurance status was associated with cervical cancer screening, with estimates significantly lower in 11 of 23 metropolitan areas for the uninsured. Colorectal cancer screening estimates by demographic groups were available for 21 of the 25 largest metropolitan areas (Table 4). FOBT was less commonly reported for people aged 50 to 64 years compared with the ⱖ65 group, with significant differences found in 9 of 21 areas. In contrast, there were few differences in FOBT estimates by gender, race/ethnicity, or educa-
American Journal of Preventive Medicine, Volume 24, Number 4
Table 2. Mammography estimates for the largest U.S. metropolitan areas by age and health insurance status Age (years) % ⴞ 95% CI
Insurance status % ⴞ 95% CI
Metropolitan area
40–49
50–64
>65
Insured
Uninsured
Los Angeles–Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA Washington DC–MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Riverside–San Bernardino CA Minneapolis–St. Paul MN–WI San Diego CA Nassau–Suffolk NY Orange County CA St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg– Clearwater FL Seattle–Bellevue–Everett WA Oakland CA Cleveland–Lorain–Elyria OH Miami FL Newark NJ Median Range
63.8 ⫾ 5.7* 73.3 ⫾ 5.1* 65.5 ⫾ 4.1* 77.8 ⫾ 2.3* 73.6 ⫾ 4.3* 70.3 ⫾ 4.5* 69.2 ⫾ 5.5* 62.1 ⫾ 7.0* 69.6 ⫾ 5.0* 61.7 ⫾ 7.2* 72.2 ⫾ 11.5 54.2 ⫾ 9.8 60.0 ⫾ 3.5* 55.7 ⫾ 9.8* 71.1 ⫾ 8.8 71.0 ⫾ 9.4 70.8 ⫾ 5.9 79.6 ⫾ 4.1 71.2 ⫾ 6.7 70.4 ⫾ 6.7
79.0 ⫾ 5.6 83.8 ⫾ 4.5 76.1 ⫾ 4.4 87.6 ⫾ 3.2 83.3 ⫾ 4.1 83.8 ⫾ 4.0 80.3 ⫾ 4.7 81.4 ⫾ 5.2 82.0 ⫾ 4.6 77.0 ⫾ 6.7 81.0 ⫾ 6.8 68.2 ⫾ 9.9 74.9 ⫾ 3.3 85.0 ⫾ 7.5 81.7 ⫾ 7.4 78.8 ⫾ 9.3 79.9 ⫾ 5.3 82.3 ⫾ 4.1 77.5 ⫾ 6.3 77.5 ⫾ 6.3
77.9 ⫾ 5.8 72.2 ⫾ 5.1 74.8 ⫾ 4.0 80.7 ⫾ 2.8 79.6 ⫾ 4.0 76.5 ⫾ 4.5 79.8 ⫾ 4.8 75.9 ⫾ 8.9 69.6 ⫾ 6.3 78.7 ⫾ 7.8 83.9 ⫾ 5.3 76.0 ⫾ 8.7* 70.8 ⫾ 3.5 72.3 ⫾ 9.3 75.1 ⫾ 8.0 85.7 ⫾ 8.3 73.5 ⫾ 5.5 82.1 ⫾ 3.9 71.1 ⫾ 5.2 84.6 ⫾ 4.5*
77.8 ⫾ 3.3 78.1 ⫾ 2.9 73.7 ⫾ 2.5 83.0 ⫾ 1.6 79.9 ⫾ 2.4 77.8 ⫾ 2.7 79.0 ⫾ 2.9 75.9 ⫾ 4.4 75.2 ⫾ 3.1 74.8 ⫾ 4.3 NA NA 69.5 ⫾ 2.1 NA NA NA 76.4 ⫾ 3.2 83.1 ⫾ 2.3 74.8 ⫾ 3.4 83.3 ⫾ 3.2
49.5 ⫾ 10.5** 64.5 ⫾ 11.1 52.3 ⫾ 9.7** 63.9 ⫾ 9.0** 62.5 ⫾ 11.5** 61.8 ⫾ 10.2** 41.6 ⫾ 13.8** 52.2 ⫾ 11.3** 59.1 ⫾ 12.9 44.8 ⫾ 14.5** NA NA 37.8 ⫾ 11.0** NA NA NA 52.7 ⫾ 14.2** 56.0 ⫾ 11.6** 55.5 ⫾ 14.8 37.2 ⫾ 12.5**
65.2 ⫾ 4.6* 76.8 ⫾ 8.7 68.7 ⫾ 8.9 70.1 ⫾ 8.7 74.9 ⫾ 6.6 70.3 54.2–79.6
78.7 ⫾ 4.4 78.7 ⫾ 10.0 83.3 ⫾ 7.3 74.0 ⫾ 7.2 81.0 ⫾ 6.6 80.3 68.2–87.6
74.2 ⫾ 4.6 85.6 ⫾ 7.7 77.6 ⫾ 6.0 78.7 ⫾ 6.8 73.0 ⫾ 7.3 76.5 69.6–85.7
NA NA NA 80.4 ⫾ 4.4 77.8 ⫾ 4.0 77.8 69.5–83.3
NA NA NA 50.2 ⫾ 12.4** 62.8 ⫾ 16.0 52.7 37.2–64.5
* Difference statistically significant compared to persons aged 50 – 64 years; **difference statistically significant compared to persons with health insurance. CI, confidence interval; NA, not available because of insufficient sample size.
tion (data not shown). Sigmoidoscopy estimates were higher among persons aged ⱖ65 years and among men, with significant differences found in 7 of 21 areas by both age and gender. Unfortunately, small sample sizes preclude drawing any firm conclusions about race/ethnic differences in colorectal cancer screening in most areas for this measure (data not shown). Persons with lower levels of education were less likely to receive sigmoidoscopy, with significant differences in 6 of 21 areas. Because of small sample sizes, we were able only to compare estimates for FOBT and sigmoidoscopy by insurance status for eight metropolitan areas, but estimates were significantly lower for the uninsured in six of these areas for both screening tests (data not shown).
Discussion This study had several important findings. We demonstrated the ability of the BRFSS to provide cancer screening estimates in 69 metropolitan areas. Although regional median metropolitan estimates for screening were generally similar, they masked important intraand inter-regional differences. Differences were most pronounced for colorectal cancer screening; subopti-
mal estimates found in all areas indicate increased efforts are warranted to increase such screening nationwide. Although differences were smaller, substantial variation by metropolitan area for both breast and cervical cancer screening remained. We also found important differences among subpopulations within the 25 largest metropolitan areas and confirmed data from national and state studies that cervical and colorectal cancer screening estimates vary by age. Lower estimates for mammography among women aged 40 to 49 years, and lower estimates for cervical cancer screening among women aged ⱖ65 years were not surprising, given the ongoing debate about screening intervals, the recommended age when women should begin mammography screening, and whether regular cervical cancer screening can be discontinued after age 65. Furthermore, the U.S. Preventive Services Task Force, American Cancer Society, American College of Obstetricians and Gynecologists, and American Medical Association have differing recommendations for these tests.15,50 As for differences by education level, except for FOBT, people with lower levels of education were generally less likely to receive cancer screening in the 25 largest metropolitan areas.14,15 Am J Prev Med 2003;24(4)
305
Table 3. Cervical cancer screening estimates for the largest U.S. metropolitan areas, by selected demographics Age (years) % ⴞ 95% CI
Educational level % ⴞ 95% CI
Insurance status % ⴞ 95% CI
Metropolitan area
18–34
35–64
>65
>HS
Insured
Uninsured
Los Angeles–Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA–NJ Washington DC–MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Riverside–San Bernardino CA Minneapolis–St. Paul MN– WI San Diego CA Nassau–Suffolk NY Orange County CA St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg–Clearwater FL Seattle–Bellevue–Everett WA Oakland CA Cleveland–Lorain–Elyria OH Miami FL Newark NJ Median Range
81.8 ⫾ 3.9 82.4 ⫾ 3.5 87.5 ⫾ 2.5 88.0 ⫾ 2.3 85.6 ⫾ 4.5 85.7 ⫾ 3.8 86.8 ⫾ 3.3 86.7 ⫾ 4.1 94.1 ⫾ 2.2 87.9 ⫾ 4.6 81.9 ⫾ 7.0 89.9 ⫾ 4.8
85.7 ⫾ 3.1 88.5 ⫾ 2.4* 89.4 ⫾ 2.0 91.7 ⫾ 1.4* 89.0 ⫾ 2.2 89.5 ⫾ 3.1 88.9 ⫾ 2.5 81.2 ⫾ 4.8 91.0 ⫾ 2.4 84.0 ⫾ 4.7 84.5 ⫾ 4.5 79.7 ⫾ 6.1
73.4 ⫾ 8.9 66.9 ⫾ 6.8* 69.0 ⫾ 5.5* 72.5 ⫾ 4.1* 69.5 ⫾ 5.9* 78.3 ⫾ 5.1 77.7 ⫾ 6.5 71.3 ⫾ 13.4 75.9 ⫾ 7.4* 76.2 ⫾ 11.1 83.6 ⫾ 7.3 70.3 ⫾ 12.1*
80.5 ⫾ 3.8 80.1 ⫾ 3.5 79.7 ⫾ 3.1 81.8 ⫾ 2.4 78.9 ⫾ 3.1 79.2 ⫾ 4.9 82.7 ⫾ 3.3 81.3 ⫾ 4.8 84.5 ⫾ 3.7 79.6 ⫾ 5.8 86.8 ⫾ 4.1 79.4 ⫾ 5.6
84.7 ⫾ 3.0 85.6 ⫾ 2.5 89.7 ⫾ 1.7** 90.9 ⫾ 1.4** 90.1 ⫾ 3.0** 90.6 ⫾ 2.1** 89.5 ⫾ 2.4** 83.8 ⫾ 4.3 93.6 ⫾ 1.7** 88.9 ⫾ 3.4** 79.2 ⫾ 6.2 85.9 ⫾ 5.3
86.9 ⫾ 2.3 85.2 ⫾ 2.1 87.3 ⫾ 1.6 88.6 ⫾ 1.2 85.6 ⫾ 2.2 88.6 ⫾ 1.9 88.3 ⫾ 1.9 83.7 ⫾ 3.7 92.2 ⫾ 1.6 87.6 ⫾ 3.1 85.2 ⫾ 3.9 86.1 ⫾ 3.9
69.7 ⫾ 6.3*** 72.2 ⫾ 6.2 74.7 ⫾ 6.3*** 75.8 ⫾ 6.3*** 76.8 ⫾ 7.6 76.9 ⫾ 8.2*** 71.1 ⫾ 8.8*** 79.9 ⫾ 6.3 79.4 ⫾ 7.3*** 76.0 ⫾ 9.0 74.9 ⫾ 10.5 66.7 ⫾ 11.3***
89.4 ⫾ 2.0 90.0 ⫾ 1.4
75.0 ⫾ 4.3*
80.6 ⫾ 2.5
91.9 ⫾ 1.2** 89.1 ⫾ 1.2
75.3 ⫾ 6.1***
82.9 ⫾ 6.5 86.9 ⫾ 6.7 83.3 ⫾ 6.9 90.1 ⫾ 3.9 90.8 ⫾ 3.1 90.2 ⫾ 4.2 89.4 ⫾ 4.6
89.2 ⫾ 4.5 90.7 ⫾ 3.6 88.7 ⫾ 4.3 88.9 ⫾ 2.9 92.5 ⫾ 2.0 82.5 ⫾ 3.7* 85.5 ⫾ 4.4
68.0 ⫾ 14.2 72.7 ⫾ 10.6 93.3 ⫾ 6.5 69.8 ⫾ 7.3* 77.2 ⫾ 5.8* 55.7 ⫾ 7.9 76.5 ⫾ 8.1*
79.6 ⫾ 7.0 84.0 ⫾ 6.0 82.3 ⫾ 6.7 82.8 ⫾ 3.9 86.9 ⫾ 3.0 72.5 ⫾ 4.6 80.4 ⫾ 4.9
87.2 ⫾ 4.5 88.2 ⫾ 3.2 89.7 ⫾ 4.3** 88.9 ⫾ 2.8 91.8 ⫾ 2.1** 87.5 ⫾ 3.1** 88.5 ⫾ 3.8
87.1 ⫾ 3.9 88.3 ⫾ 3.2 91.2 ⫾ 3.4 87.6 ⫾ 2.3 90.6 ⫾ 1.8 82.1 ⫾ 2.8 85.2 ⫾ 3.3
72.0 ⫾ 11.2 76.0 ⫾ 14.4 67.1 ⫾ 12.4*** 74.3 ⫾ 10.6 82.5 ⫾ 6.6 62.4 ⫾ 11.2 83.5 ⫾ 7.0
88.6 ⫾ 2.7 89.0 ⫾ 6.8 93.5 ⫾ 4.3 82.4 ⫾ 5.5 85.1 ⫾ 5.1 87.5 81.8–94.1
90.8 ⫾ 2.1 93.8 ⫾ 3.0 89.7 ⫾ 4.3 86.9 ⫾ 3.8 85.3 ⫾ 4.1 88.9 79.7–93.8
72.5 ⫾ 6.9* 77.6 ⫾ 14.2 78.9 ⫾ 7.6* 73.9 ⫾ 9.9 67.4 ⫾ 9.7* 73.4 55.7–93.3
81.5 ⫾ 3.9 89.9 ⫾ 6.1 85.6 ⫾ 4.8 80.1 ⫾ 5.2 74.0 ⫾ 5.6 80.6 72.5–89.9
90.3 ⫾ 1.8** 91.1 ⫾ 3.8 92.6 ⫾ 3.4 86.0 ⫾ 3.7 88.2 ⫾ 3.6** 88.9 79.2–93.6
89.0 ⫾ 1.7 NA NA 86.0 ⫾ 3.4 84.6 ⫾ 3.2 87.3 82.1–92.2
78.5 ⫾ 6.7*** NA NA 74.9 ⫾ 7.3*** 67.9 ⫾ 11.5 74.9 62.4–83.5
Difference statistically significant compared to persons aged 18 –34 years; **difference statistically significant compared to persons with ⱕhigh school education; ***difference statistically significant compared to persons with health insurance. CI, confidence interval; HS, high school; NA, not available because of insufficient sample size.
*
Consistent with national and state level studies,14,15 few differences were found in the largest metropolitan areas among whites, blacks, and Hispanics for receipt of any of the four cancer screening tests. The small number of areas with sufficient data on colorectal cancer screening among Hispanics prevents drawing meaningful conclusions about this population. Comparisons between blacks and whites in this study should be interpreted with caution. Understanding the relationship between race/ethnicity and screening independent of socioeconomic status is difficult. Minorities often are over-represented in lower socioeconomic groups and have differential access to services,51 and healthcare delivery and access (e.g., having a usual source of care and health insurance) greatly increase the probability of receiving cancer screening tests.1,14,15 We confirmed findings from previous research on the association between lack of health insurance and lower prevalence of cancer screening,14,15 despite small sample sizes in many areas. Such findings suggest that efforts to increase cancer screening in underserved populations will benefit from broader 306
systematic changes that extend health insurance coverage. It should also be acknowledged that cultural and logistical factors are important determinants for receipt of cancer screening tests.52 Comparing these findings with other local surveys is difficult because of differing age groups and frequency for receiving screening, but our estimates were generally similar to local estimates from Los Angeles,18,19 Salt Lake City,24 Boston,22 and Boise23 during this same time period. This study had several limitations. Validity of selfreported county of residence was unknown, possibly resulting in misclassification of respondents. People without telephones were excluded from the surveys. Although 95% of U.S. households have telephones, people living in households without telephones are more likely to be younger, black, or Hispanic and to have lower income levels.53 Response rates by urban area could not be determined. Typical of other surveys,54 response rates declined from 1997 to 1999, and the effect on our estimates is not known. We were also
American Journal of Preventive Medicine, Volume 24, Number 4
Table 4. Demographic estimates of fecal occult blood testing and sigmoidoscopy for the largest U.S. metropolitan areas
Metropolitan area Los Angeles– Long Beach CA New York NY Chicago IL Boston MA–NH Philadelphia PA–NJ Washington DC– MD–VA–WV Detroit MI Houston TX Atlanta GA Dallas TX Phoenix–Mesa AZ Minneapolis– St. Paul MN–WI Nassau–Suffolk NY St. Louis MO–IL Baltimore MD Pittsburgh PA Tampa–St. Petersburg– Clearwater FL Seattle–Bellevue– Everett WA Cleveland–Lorain– Elyria OH Miami FL Newark NJ Median Range
FOBT
Sigmoidoscopy
Age (years) % ⴞ 95% CI
Age (years) % ⴞ 95% CI
Gender % ⴞ 95% CI
Education % ⴞ 95% CI
>65
50–64
>65
Male
Female
8.7 ⫾ 2.9
21.3 ⫾ 5.6*
29.9 ⫾ 5.7
41.7 ⫾ 6.8*
42.6 ⫾ 6.9
28.9 ⫾ 5.4** 26.4 ⫾ 6.3
42.6 ⫾ 5.9***
22.4 ⫾ 4.4 11.8 ⫾ 4.0 25.6 ⫾ 3.5 21.6 ⫾ 4.2 22.3 ⫾ 3.7
26.2 ⫾ 4.9 24.2 ⫾ 5.9* 30.7 ⫾ 3.6 27.9 ⫾ 4.4 31.1 ⫾ 4.9*
28.7 ⫾ 4.8 27.7 ⫾ 5.7 30.5 ⫾ 3.7 26.8 ⫾ 4.6 35.9 ⫾ 4.5
40.2 ⫾ 5.7* 36.3 ⫾ 6.6 36.7 ⫾ 3.8 35.1 ⫾ 4.8 44.7 ⫾ 5.2
39.7 ⫾ 6.3 35.0 ⫾ 7.5 39.6 ⫾ 4.4 36.8 ⫾ 5.4 43.8 ⫾ 5.5
29.8 ⫾ 4.4 29.3 ⫾ 5.2 28.4 ⫾ 3.1** 26.2 ⫾ 4.0** 35.7 ⫾ 4.4
31.4 ⫾ 5.1 29.7 ⫾ 6.3 28.2 ⫾ 3.8 26.6 ⫾ 4.2 34.6 ⫾ 4.4
37.5 ⫾ 5.5 33.7 ⫾ 6.2 37.6 ⫾ 3.7*** 36.8 ⫾ 5.2*** 42.1 ⫾ 4.8
18.0 ⫾ 4.2 11.1 ⫾ 4.4 16.6 ⫾ 4.0 14.5 ⫾ 5.2 17.3 ⫾ 5.6 18.4 ⫾ 2.5
22.5 ⫾ 5.1 26.9 ⫾ 10.4* 18.9 ⫾ 5.4 20.4 ⫾ 7.7 20.9 ⫾ 5.5 28.4 ⫾ 3.4*
31.6 ⫾ 5.1 34.2 ⫾ 7.5 35.6 ⫾ 5.5 20.2 ⫾ 6.5 24.8 ⫾ 6.5 37.0 ⫾ 3.2
46.8 ⫾ 6.2* 36.5 ⫾ 9.6 53.2 ⫾ 7.4* 35.0 ⫾ 9.4 41.1 ⫾ 6.7* 47.5 ⫾ 3.9*
45.8 ⫾ 6.5 39.5 ⫾ 9.2 48.8 ⫾ 7.3 31.4 ⫾ 9.3 37.5 ⫾ 7.7 43.7 ⫾ 3.7
32.6 ⫾ 4.9** 31.0 ⫾ 7.4 37.1 ⫾ 5.4 21.2 ⫾ 6.2 29.1 ⫾ 5.9 39.7 ⫾ 3.3
32.8 ⫾ 5.4 31.3 ⫾ 9.0 41.8 ⫾ 6.4 21.9 ⫾ 7.7 28.0 ⫾ 6.6 39.5 ⫾ 3.5
44.6 ⫾ 5.9*** 37.6 ⫾ 7.8 43.4 ⫾ 6.4 28.8 ⫾ 7.5 36.3 ⫾ 6.6 43.4 ⫾ 3.4
19.6 ⫾ 6.5 17.2 ⫾ 5.2 21.2 ⫾ 3.7 15.8 ⫾ 4.9 18.5 ⫾ 5.7
27.8 ⫾ 8.2 19.0 ⫾ 5.3 36.3 ⫾ 4.7* 27.2 ⫾ 6.1* 36.3 ⫾ 6.3*
36.4 ⫾ 8.1 23.0 ⫾ 5.4 27.3 ⫾ 4.2 29.3 ⫾ 6.1 28.1 ⫾ 6.8
36.5 ⫾ 8.6 29.3 ⫾ 6.1 40.7 ⫾ 4.8* 37.9 ⫾ 6.3 37.5 ⫾ 6.2
42.4 ⫾ 9.3 27.6 ⫾ 6.5 42.1 ⫾ 5.2 40.9 ⫾ 7.4 36.0 ⫾ 7.4
31.3 ⫾ 7.3 24.7 ⫾ 5.3 26.3 ⫾ 3.7** 28.5 ⫾ 5.4** 31.4 ⫾ 5.6
31.4 ⫾ 8.7 22.8 ⫾ 5.4 29.0 ⫾ 4.2 31.8 ⫾ 5.4 35.5 ⫾ 6.6
40.7 ⫾ 8.1 30.3 ⫾ 6.5 39.5 ⫾ 4.9*** 37.2 ⫾ 7.4 31.5 ⫾ 6.3
22.6 ⫾ 4.1
33.2 ⫾ 5.2*
32.9 ⫾ 4.6
39.1 ⫾ 5.6
38.8 ⫾ 6.0
32.8 ⫾ 4.3
31.5 ⫾ 6.3
37.7 ⫾ 4.4
21.1 ⫾ 7.7
25.8 ⫾ 6.8
38.8 ⫾ 8.3
48.1 ⫾ 7.7
50.2 ⫾ 9.3
37.9 ⫾ 7.1
39.4 ⫾ 7.6
49.0 ⫾ 8.9
12.0 ⫾ 4.9 18.8 ⫾ 5.9 18.4 8.7–25.6
16.0 ⫾ 5.4 24.7 ⫾ 6.7 26.2 16.0–36.3
23.2 ⫾ 7.6 30.0 ⫾ 6.7 29.9 20.2–38.8
37.3 ⫾ 8.1 36.1 ⫾ 7.5 37.9 29.3–53.2
34.9 ⫾ 9.8 42.4 ⫾ 8.6 39.7 27.6–50.2
26.0 ⫾ 6.7 25.8 ⫾ 5.9* 29.3 21.2–39.7
27.3 ⫾ 7.7 23.9 ⫾ 6.8 31.3 21.9–41.8
33.0 ⫾ 8.5 40.4 ⫾ 7.1 37.6 28.8–49.0
50–64
>HS
*
Difference is statistically significant compared to persons aged 50 – 64 years; **difference is statistically significant compared to males; difference is statistically significant compared to persons with ⱕhigh school education. CI, confidence interval; FOBT, fecal occult blood testing; HS, high school. ***
unable to examine trends between 1997 and 1999 because of data pooling. Self-reports overestimate receipt of breast, cervical, and colon cancer screening.55–57 Although we used the term screening, undoubtedly some respondents reported procedures or tests that were done for diagnostic purposes, which would contribute to overestimating the true extent of screening. The effect of pooling the slightly different questions on sigmoidoscopy in 1997 and 1999 is unclear, although any effect is likely to be consistent across areas. The role that local variation in uninsured rates had on our estimates is not known, but is likely to affect differences across areas because of the strong association between health insurance status and cancer screening.14,15,58 Differences by metropolitan areas may reflect variations by age, race/ethnicity, and education level.59 – 62 Furthermore, our estimates were for entire metropolitan areas, but screening estimates are likely to vary by geographic districts within each areas (e.g., central cities versus suburbs). Despite pooling data across
years, the numbers of respondents was small for certain subpopulations in many areas, reducing the precision of our estimates. State sample sizes vary widely because of differences in financial support.48 As a result, small metropolitan areas in some states had large sample sizes, while other states had larger metropolitan areas with smaller samples, preventing us from including areas such as San Antonio TX and San Francisco CA because of insufficient sample size. Despite these limitations, the BRFSS offers important benefits for making metropolitan area estimates. The system uses standardized protocols and methodology for data collection, and no new data collection is required. An additional advantage is timeliness, as BRFSS final annual data are available within 4 to 6 months of collection. It is a flexible system that allows for the incorporation of state- or local-area–relevant questions on cancer screening. Perhaps most important, use of existing BRFSS data would result in substantial cost savings to metropolitan area health departments around the United States. The annualized cost Am J Prev Med 2003;24(4)
307
for the Los Angeles County CA survey was estimated at $900,00019; in contrast, reweighting the entire BRFSS to provide metropolitan estimates for the entire United States cost about $100,000. In summary, BRFSS data could be used to help improve cancer prevention and to control efforts at the metropolitan level. The system provides an inexpensive, timely, and readily available source of data that could be used to examine current estimates of cancer screening, evaluate the impact of local efforts, and provide a baseline against which data from more localized surveys can be compared. Local health departments should consider routinely using these reweighted data as part of their ongoing surveillance activities for cancer screening and other measures (e.g., risk factors such as smoking and physical activity) that are already being routinely collected as part of the BRFSS. We are grateful to the state BRFSS coordinators, Donna Brogan, Jill Dever, Glen Laird, and Bill Scott, for their assistance.
References 1. Cramer DW. The role of cervical cytology in the declining morbidity and mortality of cervical cancer. Cancer 1974;34:2018 –27. 2. Benedet JL, Anderson GH, Matistic JP. A comprehensive program for cervical cancer detection and management. Am J Obstet Gynecol 1992;166: 1254 –9. 3. Gotzsche PC, Olsen O. Is screening for breast cancer with mammography justifiable? Lancet 2000;355:129 –34. 4. Kerlikowske K, Grady D, Rubim SM, et al. Efficacy of screening mammography: a meta-analysis. JAMA 1995;273:149 –54. 5. U.S. Preventive Services Task Force. Guide to clinical preventive services. 2nd ed. Baltimore, MD: Williams & Wilkins, 1996. 6. Kronborg O, Fenger C, Olsen J, et al. Randomised study of screening for colorectal cancer with faecal-occult– blood screening. Lancet 1996;348: 1467–71. 7. Hardcastle JD, Chamberlain JO, Robinson MHE, et al. Randomised controlled trial of faecal-occult– blood screening for colorectal cancer. Lancet 348;1472–7. 8. Mandel JS, Church TR, Ederer F, Bond JH. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood. J Natl Cancer Inst 1999;91:434 –7. 9. Selby JV, Freidman GD, Quesenberry CP, et al. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992;356:653–7. 10. Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy an colorectal cancer mortality. J Natl Cancer Inst. 1992;84:1572–5. 11. Slattery ML, Edwards SL, Ma KN, et al. Colon cancer screening, lifestyle, and risk of colon cancer. Cancer Causes Control 2000;11:555–63. 12. Rakowski W, Rimer BK, Bryant SA. Integrating behavior and intention regarding mammography by respondents in the 1990 National Health Interview Survey of health promotion and disease prevention. Public Health Rep 1993;108:605–24. 13. Pearlman DN, Rakowski W, Ehrich B, et al. Breast cancer screening practices among Black, Hispanic, and White women: reassessing differences. Am J Prev Med 1996;12:327–37. 14. Breen N, Wagener DK, Brown ML, et al. 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 –13. 15. Blackman DK, Bennett EM, Miller DS. Trends in self-reported use of mammograms (1989 –1997) and Papanicolaou tests (1991–1997)—Behavioral Risk Factor Surveillance System. MMWR Surveill Summ 1999;48:1–22.
308
16. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System. Available at www.cdc.gov/brfss. Accessed on March 31, 2002. 17. Health Resources and Services Administration. Community Health Status Indicators Project. Rockville, MD: Health Resources and Services Administration. Available at: www.communityhealth.hrsa.gov. Accessed on March 31, 2002. 18. Los Angeles County Department of Health Services. The health of Angelenos: a comprehensive report of the health of the residents of Los Angeles County. Los Angeles, CA: Los Angeles County Department of Health Services, 2000. Available at: http://lapublichealth.org/ha/reports/ angelenos/hofa.pdf. Accessed on April 12, 2002. 19. Simon PA, Wold CM, Cousineau MR, et al. Meeting the data needs of a local heatlh department: the Los Angeles County health survey. Am J Public Health 2001;91:1950 –2. 20. Northridge ME, Morabia A, Ganz ML, et al. Contribution of smoking to excess mortality in Harlem. Am J Epidemiol 1998;147:250 –8. 21. Parchman M, Byrd T. Access to and use of ambulatory health care by a vulnerable Mexican American population on the U.S.–Mexico border. J Health Care Poor Underserved 2001;12:404 –14. 22. Massachusetts Department of Public Health. Behavioral Risk Factor Survey 1994 —1996 city report: Boston. Boston: Massachusetts Department of Public Health, 1998. 23. Idaho Department of Health and Welfare. Idaho Behavioral Risk Factor Surveillance System, 1999 survey data. Boise: Idaho Department of Health and Welfare, 2000. 24. Utah Department of Health. Utah Behavioral Risk Factor Surveillance System local health district report. Salt Lake City: Utah Department of Health, 2000. 25. Virginia Department of Health. Healthy Virginia communities: a report on year 2000 health status and risk reduction indicators for the Commonwealth of Virginia and health districts. Richmond: Virginia Department of Health, 1997. 26. Wisconsin Department of Health and Family Services. County and regional estimates of behavioral risk factors and health screening practices, 1989 – 1994. Madison: Wisconsin Department of Health and Family Services, 1997. 27. U.S. Census Bureau. Census 2000 summary file 1 (SF 1) 100-percent data. Washington, DC: U.S. Census Bureau. Available at: http://factfinder. census.gov/home/en/sf1.html. Accessed on April 12, 2002. 28. Burack RC, Gimotty PA. Promoting screening mammography in inner-city settings. The sustained effectiveness of computerized reminders in a randomized controlled trial. Med Care 1997;35:921–31. 29. Doyle JP, Parker RM, Jacobson TA, McNagny SE. Breast and cervical cancer screening in an inner-city medical walk-in clinic: taking advantage of an often missed opportunity. Am J Prev Med 1996;12:345–50. 30. Taylor VM, Thompson B, Montano DE, Mahloch J, Johnson K, Li S. Mammography use among women attending an inner-city clinic. J Cancer Educ 1998;13:96 –101. 31. Weber BE, Reilly BM. Enhancing mammography use in the inner city. A randomized trial of intensive case management. Arch Intern Med 1997; 157:2345–9. 32. Whitman S, Ansell D, Lacey L, et al. Patterns of breast and cervical cancer screening and three public health centers in an inner-city urban area. Am J Public Health 1991;81:1651–3. 33. Leviton LC, Snell E, McGinnis M. Urban issues in health promotion strategies. Am J Public Health 2000;90:863–6. 34. Freudenberg N. Time for a national agenda to improve the health of urban populations. Am J Public Health 2000;90:837–40. 35. Fielding JE, Sutherland CE, Halfon N. Community health report cards: results of a national survey. Am J Prev Med 1999;17:79 –86. 36. Institute of Medicine. The future of public health. Washington DC: National Academy Press, 1988. 37. Marcus AC, Crane LA. A review of cervical cancer intervention research: implications for public health programs and future research. Prev Med 1998;27:13–31. 38. Rimer BK. Interventions to enhance cancer screening—a brief review of what works and what is on the horizon. Cancer 1998;83(suppl S):1770 –4. 39. Champion VL, Ray DW, Heilman DK, Springston JK. A tailored intervention for mammography among low-income African-American women. J Psychosocial Oncol 2000;18:1–13. 40. Shenson D, Cassarino L, DiMartino D. Improving access to mammograms through community-based influenza clinics—a quasi-experimental study. Am J Prev Med 2001;20:97–102.
American Journal of Preventive Medicine, Volume 24, Number 4
41. Hiatt RA, Pasick RJ, Stewart S, et al. Community-based cancer screening for underserved women: design and baseline findings from the breast and cervical cancer intervention study. Prev Med 2001;33:190 –203. 42. U.S. Department of Health and Human Services. Healthy People 2010, 2nd ed. Washington DC: U.S. Department of Health and Human Services, 2000. 43. Centers for Disease Control and Prevention. 1991 Behavioral Risk Factor Surveillance System quality control report. Atlanta GA: Centers for Disease Control and Prevention, 1992. 44. Centers for Disease Control and Prevention. 1999 Behavioral Risk Factor Surveillance System quality control report. Atlanta GA: Centers for Disease Control and Prevention, 2000. 45. Centers for Disease Control and Prevention. Cigarette smoking in 99 metropolitan areas—United States, 2000. MMWR Morb Mortal Wkly Rep 2001;50:1107–13. 46. Marks JS, Hogelin GC, Gentry EM, et al. The behavioral risk factor surveys: I. state-specific prevalence estimates of behavioral risk factors. Am J Prev Med 1985;1:1–8. 47. Powell-Griner E, Anderson JE, Murphy W. State- and sex-specific prevalence of selected characteristics—Behavioral Risk Factor Surveillance System, 1994 and 1995. MMWR Morb Mortal Wkly Rep 1997;46:1–31. 48. Nelson DE, Holtzman D, Waller M, et al. Objectives and design of the Behavioral Risk Factor Surveillance System. In: Proceedings of the 1998 American Statistical Association Section on Survey Research Methods, Dallas, TX, August 9 –13, 1998. Alexandria VA: American Statistical Association, 1998:214 – 8. 49. U.S. Office of Management and Budget. Metropolitan areas defined. Washington, DC: Office of Management and Budget, June 30, 1993. (1990 CPH-S-11). 50. Rathore S, McGreevey JD, Schulman KA, Atkins D. Mandated coverage for cancer-screening services: whose guidelines do states follow? Am J Prev Med 2000;19:71–8. 51. Mills RJ. Health insurance coverage: 2000. Current Population Reports P60-215. Washington, DC: U.S. Census Bureau, 2001.
52. Kelaher M., Stellman JM. The impact of Medicare funding on the use of mammography among older women: implications for improving access to screening. Prev Med 2000;31:658 –64. 53. U.S. Bureau of the Census. Statistical brief: phoneless in America. Washington, DC: U.S. Department of Commerce, Economics and Statistics Administration, 1994. 54. Massey JT, O’Connor D, Krotki K. Response rates in random digit dialing (RDD) telephone surveys. In: Proceedings of the 1997 American Statistical Association Section on Survey Research Methods, Anaheim, CA, August 10 –14, 1997. Alexandria, VA: American Statistical Association, 1997:707–12. 55. Gordon NP, Hiatt RA, Lampert DI. Concordance of self-reported data and medical record audit for six cancer screening procedures. J Natl Cancer Inst 1993;85:556 –70. 56. Hiatt RA, Perez-Stable EJ, Quesenberry C, et al. Agreement between self-reported early cancer detection practices and medical audits among Hispanic and non-Hispanic white health plan members in Northern California. Prev Med 1995;24:278 –85. 57. Warnecke RB, Sudman S, Johnson TP, et al. Cognitive aspects of recalling and reporting health-related events: Papanicolaou smears, clinical breast exam, and mammogram. Am J Epidemiol 1997;146:982–92. 58. Frazier E, Jiles R, Mayberry R. Use of screening mammography and clinical breast examinations among black, Hispanic, and white women. Prev Med 1996;25:118 –25. 59. U.S. Census Bureau. Metropolitan area rankings for persons 65 years of age and over, 1996. Press release, July 21, 1998. Available at: www.census.gov/ Press-Release/metro09.prn. Accessed on April 16, 2002. 60. U.S. Census Bureau. Metropolitan area rankings for Hispanic population, 1996. Press Release, July 21, 1998. Available at: www.census.gov/PressRelease/metro13.prn. Accessed on April 16, 2002. 61. U.S. Census Bureau. Metropolitan area rankings for Black population, 1996. Press Release, July 21, 1998. Available at: www.census.gov/PressRelease/metro11.prn. Accessed on April 16, 2002. 62. Gaquin DA, DeBrandt KA, eds. Education statistics of the United States. 3rd ed. Lanham, MD: Bernan, 2001.
Am J Prev Med 2003;24(4)
309