Journal of Community Health

, Volume 33, Issue 1, pp 10–21

Previous Cancer Screening Behavior as Predictor of Endoscopic Colon Cancer Screening Among Women Aged 50 and Over, in NYC 2002

Authors

    • Department of EpidemiologyJoseph A. Mailman School of Public Health at Columbia University
    • Department of Environmental HealthJoseph A. Mailman School of Public Health at Columbia University
    • Center for Urban Epidemiologic StudiesThe New York Academy of Medicine
  • Christina Chan
    • Center for Urban Epidemiologic StudiesThe New York Academy of Medicine
  • David Vlahov
    • Center for Urban Epidemiologic StudiesThe New York Academy of Medicine
  • Maria K. Mitchell
    • AMDeC Foundation
  • Stephen B. Johnson
    • Department of Biomedical InformaticsColumbia University
  • Harold Freeman
    • Ralph Lauren Center for Cancer Care and Prevention
Original paper

DOI: 10.1007/s10900-007-9067-3

Cite this article as:
Guerrero-Preston, R., Chan, C., Vlahov, D. et al. J Community Health (2008) 33: 10. doi:10.1007/s10900-007-9067-3

Abstract

Colon cancer screening rates in women are low. Whether screening for breast and cervical cancer is associated with colon cancer screening behavior is unknown but could provide linkage opportunities. To identify the extent to which both breast and cervical cancer screening increases uptake of colon cancer screening among women in New York City. Women at least 50 years old completed questionnaires for the New York Cancer Project. Analyses compared rates of endoscopic colon cancer screening with adherence to screening recommendations for breast and cervical cancer. Of the 3,386 women, 87.8% adhered to breast and cervical cancer screening guidelines, yet only 42.1% had received endoscopic colon cancer screening. Most women with colon cancer screening (95%) also reported past mammogram and Pap-smear. In multivariable analysis, women who adhered to the other two procedures were more likely to have had colon cancer screening than women with no prior history (OR = 4.4; CI = 2.36, 8.20), after accounting for age, race/ethnicity, insurance status, family history of cancer and income. Significant predictors of endoscopic colon cancer screening included: age over 65 years (OR = 1.63; CI = 1.23, 2.15) with 50–65 years old as the reference, any health insurance (OR = 2.18; CI = 1.52, 3.13) and a family history of cancer (OR = 1.38; CI = 1.17, 1.61). Colorectal cancer screening remains low, even among women who undergo other cancer screening tests. Opportunities to link cancer screening tests to encourage colon cancer screening merit closer attention.

Keywords

Cancer screening behaviorPredictors of colon cancer screening behaviorSocial and health care determinants

Introduction

Cancer is the second leading cause of death in the United States; despite advances in detection and treatment, cancer remains a major public health problem [1]. Colorectal cancer is the second leading cause of cancer deaths (after lung) in New York City and the leading cause of cancer death among non-smokers [2, 3]. The US Preventive Services Task Force and the American Cancer Society recommend fecal occult blood testing, flexible sigmoidoscopy, colonoscopy, and even perhaps double-contrast barium enema as acceptable colorectal cancer screening tests. The New York City Department of Health and Mental Hygiene recommends that everyone age 50 or older should see a doctor and get a colonoscopy, because this is the only screening technique that also includes treatment, if polyp removal is indicated. This paper examines if there is an association between compliance with cervical and breast cancer screening recommendations and endoscopic colon cancer screening recommendations among women in New York City age 50 or older, after accounting for race, ethnicity and income.

Reported rates for colon cancer screening in population based studies remain low, regardless of ethnicity, while there has been a steady increase in the use of other cancer screening tests [4, 5]. Analyses of the National Health Interview Surveys have shown that rates of screening for colon, breast and cervical cancers improved between 1987 and 2003, yet rates for colorectal screening in 1998 were lower than levels observed for breast cancer screening in 1987 [6]. Despite a small improvement, colorectal screening rates in 2001 were still lagging substantially behind breast and cervical cancer screening rates and reported rates of colon cancer screening use in 2003 were still low for both, men (32%) and women (30%) [7, 8].

Endoscopic colon cancer screening use has been significantly associated with higher knowledge, lower barriers, higher benefits, higher self-efficacy, and provider recommendation [9]. Awareness campaigns and patient educational aids, including decision tools implemented in multiple sites, such as worksites, community centers, health care systems, and physician offices, increase the percent of eligible Americans who understand their personal risk, the need for screening, and the options available to them [10]. Physician’s support of endoscopic colon cancer screening has also been positively associated with the receipt of endoscopic colon cancer screening [11]. Population based studies have also identified ethnicity, educational attainment, former smoker status, health insurance, having a usual source of care, or if they talked to a general doctor, as factors associated with the use of colon cancer screening tests [8].

Colorectal cancer racial disparities have emerged and widened for three decades. These temporal trends probably reflect complicated racial differences between screening practice patterns and etiologic factors [12]. Some of the etiologic factors that may be related to these disparities are political, socioeconomic and discriminatory forces, which have segregated people of color into communities with some of the highest indices of urban poverty and material deprivation in the US [13]. For example, screening guidelines recommend periodic testing for adults aged 50 and older, yet a number of studies report that these tests are especially underused among minority, elderly, less educated, and poor women [14].

Adjusting for age, gender, access to care (as income and insurance), and risk profile (as cancer in family, smoking, and obesity), Blacks and Hispanics are less likely to have been screened than Whites [15]. Hispanics have lower rates of screening for cervical, breast, and colon cancer than Non-Hispanics [16]. In addition, the percentage of State populations below poverty level is inversely correlated with reported colorectal cancer screening received by people over 50 years old [17].

A possible target population for focused, community based, participatory interventions aimed at increasing endoscopic colon cancer screening rates in New York City, are those women who regularly get a mammography and Pap-smear done as part of their cancer prevention strategies. Correlates of colorectal cancer screening test use are similar to those observed for mammography and Pap-smear tests and include race; ethnicity; age; education; income; health insurance coverage; having a usual source of health care, a recent physician visit, use of other cancer screening tests, and a recommendation from a physician for screening [8]. The purpose here is to determine if the rates of endoscopic colon cancer screening across racial and ethnic groups are higher among women in NYC who have already undergone breast and cervical cancer screening.

Methods

A baseline cross-section from a prospective cohort was utilized for this study. Using venue-based sampling with quotas, 18,187 adults aged 30 years or older were recruited in 14 enrollment sites set up across the five boroughs of New York City over a 2-year period. Sites included six medical centers, two community hospitals, and six community based health centers. In addition New York Blood Center enrolled individuals into the project as part of its routine donor blood drives. Enrollment sites were selected according to their location to ensure every borough in New York City was covered. Each enrollment site had a team of interviewers, phlebotomists, a project coordinator/outreach worker, and (for medical centers only) a faculty member who served as the site co-primary investigator for the project.

To be eligible, enrollees had to be 30 years of age or older, reside in the New York Tri-State area, and have a literacy level sufficient to complete a simple mail-out follow-up questionnaire. Enrollees completed a screening questionnaire, a baseline questionnaire, underwent venipuncture, and provided contact information as well as identifiers to facilitate registry linkages. The questionnaire included the following: demographic information, personal medical history, substance use, reproductive history and body measurements. Registry linkages include the New York State Department of Vital records; the National Death Index; the Center for tumor registries for New York, Connecticut, and New Jersey and signed medical release forms to facilitate access to the medical records.

The sample was 39% male, 37% older than 50 years, 53% white, 20% African American, 18% Hispanic, and 9% Asian. In terms of family history of cancer, 21% reported mother, 21% reported father, and 5.9% reported both parents with cancer; 8.5% reported any sibling with cancer. At baseline, 1,231 participants reported prior cancer, including breast (33%), skin (23%), colorectal (8%), cervix (8%), lymph (7%), prostate (7%), uterus (5%). For the cancer-free group, 60.1% were female, 33.9% were at least 50 years old, 58.9% were white, 20.6% were African American, 9.6% were Asian, 7.2% were other, 3.7% refused to answer and for ethnicity 18.5% reported Hispanic. The distribution of racial/ethnic groups for those older than 50 years was statistically similar to the New York City 2000 Census. In terms of family history of cancer among the cancer-free participants 20.2% reported cancer in the mother, 20.9% reported cancer in the father, 5.6% in both parents, and 7.8% in siblings [15, 18].

This study was limited to 3,386 women ages 50+ who reported no prior history of hysterectomy, breast cancer, cervical cancer, and/or colon cancer. The percentages of women over fifty that showed adherence to current screening guidelines for a mammography, Pap-smear and endoscopic colon cancer screening (combined variable that includes both, colonoscopy or sigmoidoscopy) were compared among different ethnic groups and other relevant demographic and behavioral data reported on questionnaire.

To assess the role of socio-demographic characteristics (age, education, place of residence, country of origin and race/ethnicity), access to care (as income and insurance), and risk profile (as cancer in immediate family and history of smoking) we examined the relation between each of these factors with colon cancer screening by recent cancer screening behavior, using chi-square and Fischer exact tests, as measures of significance. The role of cancer screening behaviors as predictors of colon cancer screening were assessed with a multivariable logistic regression model, adjusted for potential confounders.

Results

The race/ethnicity characteristics of the women in the study were: 56% Non-Hispanic White, 10% Non-Hispanic Black, 14% Hispanic, 11% Non-Hispanic Asian, and 8% other. The distribution of demographics in this sample was found to be similar to the 2000 Census, utilizing chi-square tests (data not shown). Nearly 45% had household incomes of $50,000 or less, 13.6% had not completed high school, and more than a third of the women (35%) were born outside of the U.S. Almost 70.0% had private health insurance, less than 10% had no insurance and 56.9% had never smoked.

Table 1 shows associations of colon cancer screening with demographic, racial/ethnic, access to care and personal risk characteristics by rates of recent screening behaviors. Most women, 87.8%, had both a mammography and a Pap-smear within the recommended time interval of the American Cancer Society guidelines, 4.9% had mammography but did not have a Pap-smear, 3.6% had a Pap-smear but did not have a mammogram, and 3.6% had neither screening procedure [19]. Rates for endoscopic colon cancer screening were 42.1% for women who had both cancer screening tests done, 17.4% for women who had only mammography, 21.3% for women who had only a Pap-smear and 9.8% for women who had neither screening procedure (P = .0001), with 39% overall reporting colon cancer screening.
Table 1

Characteristics of 2,059 women aged 50 and over who received both mammography and Pap-smear, mammography alone, Pap-smear alone, or neither type of screening, NYC 2002

  

Recent screning statusa

All women

Received both mammography and Pap-smear

Needs Pap-smear only

Needs mammography only

Needs both types of screening

Received screening colonoscopy

Total

3,386 (100%)

2,974 (87.8%)

n = 167 (4.9%)

n = 122 (3.6%)

n = 123 (3.6%)

n = 1,320 (38.9%)

Age

50–54

1,345 (39.7)

1,195 (40.2)

54 (32.3)

54 (44.3)

42 (34.1)

422 (32)

55–59

1,031 (30.4)

914 (30.7)

41 (24.6)

41 (33.6)

35 (28.5)

431 (32.7)

60–64

595 (17.6)

515 (17.3)

39 (23.4)

14 (11.5)

27 (22)

261 (19.8)

65–69

292 (8.6)

253 (8.5)

19 (11.4)

8 (6.6)

12 (9.8)

150 (11.4)

70–74

71 (2.1)

63 (2.1)

3 (1.8)

3 (2.5)

2 (1.6)

35 (2.7)

75+

52 (1.5)

34 (1.1)

11 (6.6)

2 (1.6)

5 (4.1)

21 (1.6)

Race/ethnicity

NH-White

1,912 (56.5)

1,715 (57.7)

80 (47.9)

55 (45.1)

62 (50.4)

865 (65.5)

NH-Black

331 (9.8)

298 (10)

9 (5.4)

13 (10.7)

11 (8.9)

114 (8.6)

NH-Asian

489 (14.4)

433 (14.6)

20 (12)

20 (16.4)

16 (13)

139 (10.5)

Hispanic

388 (11.5)

302 (10.2)

40 (24)

19 (15.6)

27 (22)

116 (8.8)

NH-Other

266 (7.9)

226 (7.6)

18 (10.8)

15 (12.3)

7 (5.7)

86 (6.5)

Income

Missing

37 (1.1)

24 (0.8)

4 (2.4)

3 (2.5)

6 (4.9)

7 (0.5)

Less than $15,000

435 (12.8)

352 (11.8)

36 (21.6)

22 (18)

25 (20.3)

136 (10.3)

$15,000–30,000

428 (12.6)

365 (12.3)

30 (18)

21 (17.2)

12 (9.8)

148 (11.2)

$30,001–50,000

586 (17.3)

521 (17.5)

25 (15)

18 (14.8)

22 (17.9)

223 (16.9)

$50,001–100,000

908 (26.8)

831 (27.9)

29 (17.4)

22 (18)

26 (21.1)

392 (29.7)

Greater than $100,000

466 (13.8)

437 (14.7)

12 (7.2)

13 (10.7)

4 (3.3)

228 (17.3)

Don’t know

212 (6.3)

174 (5.9)

11 (6.6)

12 (9.8)

15 (12.2)

58 (4.4)

Education

Missing

21 (0.6)

14 (0.5)

2 (1.2)

3 (2.5)

2 (1.6)

4 (0.3)

Less than HS

462 (13.6)

374 (12.6)

40 (24)

19 (15.6)

29 (23.6)

131 (9.9)

HS or equiv

821 (24.2)

721 (24.2)

39 (23.4)

35 (28.7)

26 (21.1)

277 (21)

Some college

657 (19.4)

578 (19.4)

28 (16.8)

25 (20.5)

26 (21.1)

260 (19.7)

College grad or more

1,410 (41.6)

1,277 (42.9)

56 (33.5)

37 (30.3)

40 (32.5)

642 (48.6)

Smoking history

Current

380 (11.2)

327 (11)

20 (12)

17 (13.9)

16 (13)

108 (8.2)

Former

1,078 (31.8)

987 (33.2)

35 (21)

27 (22.1)

29 (23.6)

521 (39.5)

Never

1,928 (56.9)

1,660 (55.8)

112 (67.1)

78 (63.9)

78 (63.4)

691 (52.3)

Foreign vs. US born

Unknown

31 (0.9)

26 (0.9)

2 (1.2)

2 (1.6)

1 (0.8)

14 (1.1)

Born in USA

2,171 (64.1)

1,962 (66)

75 (44.9)

64 (52.5)

70 (56.9)

928 (70.3)

Foreign born

1,184 (35)

986 (33.2)

90 (53.9)

56 (45.9)

52 (42.3)

378 (28.6)

Borough

Bronx

294 (8.7)

265 (8.9)

11 (6.6)

14 (11.5)

4 (3.3)

98 (7.4)

Brooklyn

777 (22.9)

663 (22.3)

50 (29.9)

29 (23.8)

35 (28.5)

298 (22.6)

Manhattan

705 (20.8)

637 (21.4)

24 (14.4)

19 (15.6)

25 (20.3)

342 (25.9)

Queens

736 (21.7)

618 (20.8)

47 (28.1)

32 (26.2)

39 (31.7)

267 (20.2)

Staten Island

481 (14.2)

438 (14.7)

18 (10.8)

16 (13.1)

9 (7.3)

152 (11.5)

Health insurance

Private

2,340 (69.1)

2,114 (71.1)

93 (55.7)

72 (59)

61 (49.6)

967 (73.3)

Medicare

317 (9.4)

272 (9.1)

21 (12.6)

10 (8.2)

14 (11.4)

164 (12.4)

Medicaid

313 (9.2)

260 (8.7)

25 (15)

16 (13.1)

12 (9.8)

105 (8)

Other

152 (4.5)

135 (4.5)

7 (4.2)

5 (4.1)

5 (4.1)

69 (5.2)

Self

98 (2.9)

86 (2.9)

7 (4.2)

2 (1.6)

3 (2.4)

45 (3.4)

No insurance

311 (9.2)

229 (7.7)

28 (16.8)

21 (17.2)

33 (26.8)

46 (3.5)

aDuring last 2 years for mammography and last 3 years for Pap-smear

Among women who had endoscopic colon cancer screening, 95% had also reported past mammogram and Pap-smear. However, for women who only had one type of screening (mammography or pap-smear, but not both), only 19% also had received endoscopic colon cancer screening.

Bivariable differences in colon cancer screening behavior by prior history of cancer screening were seen among the four groups of screening behaviors. Women with a prior history of undergoing mammography and Pap-smear screening tests were more likely to be NH-White with higher income and education, had insurance coverage, no smoking history, and were 50–54 years old, born in the USA.

The race/ethnicity breakdown of recent cancer screening behavior, for both cervical and breast cancer, was: 90% NH-white, 90% NH-black, 88.5% Hispanic, 78% NH-Asian, and 85% other. The race/ethnicity breakdown of having had endoscopic colon cancer screening within the last 5 years was: 45% NH-white, 34% NH-black, 28% Hispanic, 30% NH-Asian, and 32% for others (P < .001) (Fig. 1). Among peoples of Caribbean descent the rates of endoscopic colon cancer screening did not vary by race or heritage [Black from West Indies (31%, P < 0.06), Puerto Rican (29%, P < 0.005), Dominican (26%, P < 0.05), data not shown].
https://static-content.springer.com/image/art%3A10.1007%2Fs10900-007-9067-3/MediaObjects/10900_2007_9067_Fig1_HTML.gif
Fig. 1

Race/ethnicity breakdown of recent cancer screening behavior, contrasting endoscopic colon cancer screening (dark bars) among women who have been screened for both cervical and breast cancer (white bars), within recommended time frames

Table 2 shows crude and adjusted odds ratios of the association between predictive factors and endoscopic colon cancer screening behavior. These unadjusted odds ratio show that prior screening behavior and cancer risk profile characteristics, as well as demographic, behavioral, and access to care factors, all increase the likelihood of having had a colon cancer screening test during the past 5 years.
Table 2

Predictors of colorectal cancer screening among women 50 years+ in New York City (2002)

 

Total n

% with colonoscopy

Crude odds ratio

Adjusted odds ratio

OR

Lower CI

Upper CI

OR

Lower CI

Upper CI

All

3,386

39

      

Screening behavior (mampap)

Need Both

123

0.9

1.0

1.0

Need Mam

122

2

2.51

1.20

5.23*

1.60

0.72

3.57

Need Pap

167

2.2

1.94

0.95

3.98*

1.27

0.59

2.73

Need Neither

2,974

94.9

6.74

3.70

12.27*

4.40

2.36

8.20

Race/ Ethnicity (racenew)

White

1,912

65.5

1.0

1.0

Black

331

8.6

0.64

0.50

0.81*

0.71

0.54

0.93

Hispanic

489

10.5

0.48

0.39

0.60*

0.53

0.41

0.68

Other

654

15.3

0.54

0.45

0.65*

0.86

0.68

1.09

Age—older than 65 years (agegt65)

50–65 years

3,060

88

1.0

1.0

65+ years

326

12

1.56

1.24

1.96*

1.63

1.23

2.15

Any health insurance (anyins)

No

364

5.2

1.0

1.0

Yes

3,022

94.8

3.08

2.34

4.05*

2.18

1.52

3.13

Family history of cancer (famhx)

No

1,619

40.8

1.0

 

1.0

Yes

1,767

59.2

1.59

1.38

1.83*

1.38

1.17

1.61

Income (incnew)

<$15,000

435

10.3

1.0

1.0

$15,000–29,999

428

11.2

1.16

0.88

1.54

0.96

0.71

1.30

$30,000–49,999

586

16.9

1.35

1.04

1.76*

0.95

0.71

1.27

$50,000–99,999

908

29.7

1.67

1.31

2.13*

1.09

0.83

1.44

$100,000 or more

466

17.3

2.11

1.60

2.77*

1.27

0.93

1.73

OR = odds ratio; CI = confidence interval

P < .001

The multivariable logistic regression model fitted to predict endoscopic colon cancer screening in the past 5 years shows that women with other cancer screening behavior were more likely than those without other site screening to have had endoscopic colon cancer screening (OR = 4.4; CI = 2.36, 8.20), even after accounting for age, race/ethnicity, insurance status, family history of cancer and income. Predictors in the model, which significantly increased the likelihood of having endoscopic colon cancer screening history, after simultaneously adjusting for the other co-variates were: age over 65 years (OR = 1.63; CI = 1.23, 2.15) with age between 50 and 65 years old as the reference; having any health insurance (OR = 2.18; CI = 1.52, 3.13); and a family history of cancer (OR = 1.38; CI = 1.17, 1.61).

Predictors in the model that significantly reduced the likelihood of having endoscopic colon cancer screening history, after adjusting for all other co-variates were NH-Black (OR = 0.71; CI = 0.54, 0.93); and Hispanic (OR = 0.53; CI = 0.41, 0.68) when compared to NH-Whites. Overall, the profile for women in this sample to have had endoscopic colon cancer screening included white, >65 years old, having insurance, higher income and a family history of cancer.

Discussion

The major findings of this paper are first that rates of endoscopic colon cancer screening in women over 50 years old in New York City are low, when compared to screening behavior for both breast and cervical cancer; and second, that rates were low for endoscopic colon cancer screening even for those women that adhered to breast and cervical cancer screening guidelines. However, having received both, a Pap-smear and a mammography, increases four-fold the likelihood of having received endoscopic colon cancer screening, when compared to women without any recent history of breast and cervical cancer screening behavior. Combined, these data signal a missed opportunity by primary care providers who routinely prescribe or conduct mammography and Pap-smear tests.

Lack of patient awareness and lack of physician recommendation for screening are key barriers to obtaining colorectal screening [20]. Information is limited about patients’ preferences for discussing cancer risks and risk management with primary care physicians [21, 22]. Embarrassment around the symptoms, ignorance, and fear, are barriers that need to be overcome in order to reduce the number of people affected by the disease [23]. Factors that may also differentiate likelihood of screening include insurance coverage, source of care, lower income, and age after accounting for sex, racial/ethnic group and educational level. Other factors that could influence personal decision or provider referral for screening include personal risk factors for colon cancer, such as family history, obesity, physical exercise, and smoking [24].

There might be cognitive, psychosocial, and health services factors that account for the missed opportunity identified in this paper and merit further research. Cognitive theory suggests that perception of cancer risk and perceived benefits of screening might vary by socioeconomic group, influencing health care accessibility and use of cancer screening services [25, 26]. It has been reported that income, insurance and family history of cancer are factors associated with colon cancer screening for NH-Whites, NH-Blacks, NH-Asians and Hispanics adults over 50 years of age [16].

Psychosocial factors, such as higher stress and lower social support, may explain in part why people from lower socio-economic (SES) environments are less likely to participate in endoscopic colon cancer screening. A recent comparison between psychosocial and cognitive models of socioeconomic variation in participation in colon cancer screening provided better support for cognitive that psychosocial factors [25]. Younger age and less education have also been shown to be significant barriers to colon cancer screening in low-income, multi-ethnic populations, a finding that is consistent with our results [27].

Frequent use of health services and having private insurance are the strongest predictors of cancer screening behavior in community-based studies [28]. The use of preventive and cancer screening services is a health behavior associated with better health outcomes for the elderly diagnosed with cancer. Black and white patients (of higher and lower SES) who used more of the preventive and cancer screening services were at a lower risk of having late stage colon cancer than their counterparts who used fewer of these services [29]. The organizational and structural components of health services organizations may also influence treatments and outcomes in cancer, spanning the continuum of cancer care from screening to diagnosis, recovery or death [30]. Differences exist in cancer patient outcomes depending on the health care delivery system in which they are enrolled. Patients enrolled in managed care services had 9% greater survival in hazards ratio if they had breast cancer, and 6% if they had colorectal cancer, when compared to patients enrolled in fee for service systems [31]. Structural factors such as inadequate endoscopy suite time and space, and inadequate staffing have been identified as barriers to performing more screening colonoscopies in New York City hospitals, negatively impacting earlier detection and prevention rates [32].

Another one of the strongest predictors of cancer screening behavior in community-based studies is primary care physician recommendations for screening. Studies have shown that one of the barriers that limit the number of people who receive screening colonoscopies is the lack of primary care physician referrals [33]. Physician’s support of colon cancer screening has been positively associated with the receipt of colonoscopy [11]. Conversely, lack of physician recommendation for screening are key barriers to obtaining colorectal screening [20]. There have been several different attempts to increase colorectal cancer screening. This lack of referrals might be due to several factors, including beliefs about cancer risk and lack of updated cancer prevention information available for both, providers and consumers [34].

Awareness campaigns and patient educational aids, including decision tools, implemented in multiple sites, such as worksites, community centers, health care systems, and physician offices, increase the percent of eligible Americans who understand their personal risk, the need for screening, and the options available to them [10].

Ongoing research, undertaken to understand the experiences, treatments, and outcomes of patients with colorectal cancer in the United States, may impact both, primary care providers and consumers’ knowledge and behaviors related to colon cancer screening [35]. Further research on cognitive, psychosocial, and organizational factors should be conducted to understand and impact the missed opportunity that has been identified in this paper, and thus, increase the rates of colon cancer screening among women over 50 years old.

Colonoscopy has been suggested as the preferred method of screening for colorectal cancer in women [36]. Colonoscopy is the gold standard for colon cancer screening, against which all other diagnostic tests are measured, yet it is not without pitfalls and should be used in conjunction with other screening methods [37]. Utilizing colonoscopy as the primary colon cancer screening test will escalate screening costs, thus representing an effective screening barrier to uninsured and sub-insured poor people of different ethnic backgrounds. Health services utilization, socioeconomic characteristics and insurance status play a fundamental role in colon cancer prevention. Wide spread colonoscopy use may ultimately increase the existing disparities in incidence, prognosis and date of presentation of colon cancer, if compensatory measures are not taken or an affordable alternative is shown to be as effective.

This study has several limitations. First, the sample represents volunteers for a study and therefore may not be a representative sample of New Yorkers, however, comparing our demographics to that of the US Census in New York City showed similar proportions (data not shown), and our overall rates for endoscopic colon cancer screening were similar to the random digit dial study of the New York City Department of Health around the same time, suggesting that our results are not entirely out of line [38]. Second, the study did not distinguish between the use of sigmoidoscopy and colonoscopy for diagnosis and their use for screening, which makes it impossible to estimate the use of screening colonoscopy. In addition it underestimates the overall rates of colorectal cancer screening since it does not take into consideration the use of Fecal Occult Blood Test or double-contrast barium enema, both recommended colorectal cancer screening tests. Third, despite the number of published studies that underline the importance of physician recommendation to patients to seek colorectal screening examination, there were no items in our questionnaire pertaining to the strength of such recommendation.

The temporal associations between the three different types of cancer screening tests examined in this analysis, breast, cervical and colon could not be determined because the data are cross-sectional. In addition, the data was limited to self-report of previously selected individual predictors of colon cancer screening rates. The inaccuracy of self reports introduces bias into the analysis and leads to overestimation of the results.

With limitations acknowledged, these data indicate that an opportunity to increase colorectal cancer screening exists by encouraging those providers and patients already utilizing other cancer screening tests. Promoting such linkages merits closer attention.

Acknowledgments

This research was supported in part by funds from the federal government NCI grant number 5T32CA009529-20, NIA grant number 2P30AG-15294; and NCMHD grant number 5S21MD008130-02.

Copyright information

© Springer Science+Business Media, LLC 2007