Colon cancer screening practices in New York City, 2003

Results of a large random-digit dialed telephone survey

Authors


Abstract

BACKGROUND

New York City (NYC) has one of the highest concentrations of gastroenterologists in the country, yet only 33% of colorectal cancers in NYC are diagnosed early, and approximately 1500 New Yorkers die from colorectal cancer each year.

METHODS

Using data from a large, local, random-digit dialed telephone survey (n = 9802), the authors of the current study described types of colorectal cancer screening modalities and characteristics of adults undergoing screening within a recommended timeframe. Multivariate analyses were used to examine demographic, behavioral, socioeconomic, and neighborhood-level predictors of screening participation, with particular attention to factors associated with colonoscopy, the recommended screening modality in NYC.

RESULTS

Fifty-five percent of NYC adults aged ≥ 50 years reported a recent colorectal cancer screening test, and 42% reported a colonoscopy within the past 10 years. After multiple statistical adjustments, groups with the lowest likelihood of screening were the poor (odds ratio [OR], 0.66; 95% confidence interval [CI], 0.53–0.83) and uninsured (OR, 0.31; 95% CI, 0.20–0.48), as well as Asians (OR, 0.46; 95% CI, 0.29–0. 72), and current smokers (OR, 0.62; 95% CI, 0.50–0.78). Colonoscopy was less frequently reported by non-Hispanic Black New Yorkers and by women; both groups reported higher use of fecal occult blood tests. Less than 10% of adult New Yorkers reported a sigmoidoscopy in the past 5 years.

CONCLUSIONS

Low screening uptake in NYC leaves nearly 1 million New Yorkers, particularly poor and uninsured adults, at risk for undetected colorectal cancer. Colonoscopy screening programs in NYC should address health care and socioeconomic barriers and target racial and ethnic minorities and women. Cancer 2005. © 2005 American Cancer Society.

Each year, approximately 140,000 new cases of colorectal cancer are diagnosed in the United States (U.S.) and approximately 50,000 persons die from this disease.1 Screening uptake has been low even though screening tests for colorectal cancer can reduce cancer incidence by discovering precancerous polyps and detect cancer early, which results in earlier stage diagnosis and reduced mortality.2 In 2001, only half (53%) of U.S. adults aged ≥ 50 were screened within the recommended time interval.3 Screening rates for colorectal cancer are significantly lower than those for cervical cancer (83%) and breast cancer (70%), both of which have improved substantially over the past few decades.4 An increase in colorectal cancer screening rates to these levels would result in significant reductions in colorectal cancer incidence and mortality.

New York City (NYC) has a large number of acute-care hospitals and outpatient endoscopy clinics, as well as one of the highest concentrations of gastroenterologists in the country.5–7 Nonetheless, only 33% of colorectal cancer diagnoses in NYC are made in the early stage of disease,8 and more than 1500 New Yorkers die from colorectal cancer each year.9 In 2003, the New York City Department of Health and Mental Hygiene (NYC DOHMH), together with a coalition of academic medical centers and community based organizations, issued specific guidelines for colorectal cancer screening in NYC.10 The NYC coalition departed from the option-based approach recommended by most leading national organizations, which suggests a variety of screening tests for colorectal cancer for asymptomatic persons ≥ 50 years of age. The standard options include a choice of either yearly fecal occult blood tests (FOBT), flexible sigmoidoscopy every 5 years, or colonoscopy every 10 years.11–13 Instead, the NYC coalition recommended colonoscopy every 10 years as the preferred colorectal cancer screening test, with yearly FOBT as an acceptable, although not optimal, alternative for those unwilling or unable to undergo colonoscopy.

The rationale for the preferred recommendation was based on reported higher sensitivity of colonoscopy compared with other screening modalities14–18 and on the local capacity in NYC to conduct colonoscopies.6

The current study describes colorectal cancer screening practices among adult New Yorkers aged ≥ 50 years in 2003, before widespread dissemination of the new guidelines. We used data from a large, population-based, telephone survey of nearly 10,000 adults. The survey's large sample size and study design permitted a detailed assessment of cancer screening practices. This detailed analysis of individual- and neighborhood-level factors associated with colon cancer screening practices will guide the NYC campaign to increase colonoscopy screening and will provide baseline measures with which the campaign can be evaluated.

MATERIAL AND METHODS

Study Population

The New York City Community Health Survey (CHS) is an annual cross-sectional telephone survey conducted by the NYC DOHMH. The CHS uses a stratified two-stage cluster design based on random-digit dialing to select a representative sample of noninstitutionalized adults, aged ≥ 18 years. Survey enrollment is stratified across 34 neighborhoods in NYC, based on adjacent zip code aggregations (median census population = 240,000), using participant quotas to ensure robust neighborhood-level estimates. Data from each neighborhood are pooled to produce representative estimates for the adult population of NYC.

The sampling frame for 2003 was constructed through a list of random digit telephone numbers provided by a commercial vendor; 10 attempts were made to reach each selected household. Upon agreement to participate in the survey, one adult was randomly selected from each household. Interviewers were capable of conducting interviews in 26 languages, including the most commonly spoken languages in NYC: English, Spanish, Chinese, Russian, Korean, Greek, Yiddish, Polish and Haitian Creole. Surveys were defined as incomplete if the respondent began the interview but terminated it without answering key demographic questions (i.e., age, gender, race) or providing an enumeration of telephone lines in the household (required for sample selection weighting). A detailed description of the survey is available elsewhere.19 Between May and July 2003, a total of 9802 adults completed an interview from a total of 16,633 households contacted (cooperation rate = 59%, response rate = 26%).

Survey Questionnaire

The CHS questionnaire consists primarily of standardized questions adapted from the Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance Survey (BRFSS).20 Topic areas include health status, personal health behaviors, health care access, and preventive health service practices, including cancer screening.

Among all respondents aged ≥ 50 years, self-reported information on colorectal cancer screening practices determined screening prevalence estimates (see Table 1). The series of modified BRFSS questions were used to assess the proportion of adults who 1) had ever had any recommended colorectal cancer screening test, and 2) had a test with the recommended frequency, according to national guidelines. Screening tests were described to respondents to improve their recall.

Table 1. Colon Cancer Screening Questions in the 2003 Community Health Survey
Endoscopy procedures
Participants were asked if they ever had a sigmoidoscopy or colonoscopy:
“Sigmoidoscopy and colonoscopy are exams in which a tube is inserted in the rectum to view the bowel for signs of cancer or other health problems. Have you ever had either of these exams?”
Those who responded yes were also asked:
“Do you know which exam you had performed? A sigmoidoscopy involves examination of only the lower colon, while a colonoscopy involves examination of the entire colon. Colonoscopy usually involves taking medicine to make you have many watery stools the night before the exam and getting medicine through a needle in the arm to make you sleepy during the procedure.”
Fecal occult blood test screening
All participants were asked if they ever had a fecal occult blood test:
“A test for blood in your stool is where you have a bowel movement and use a stick to smear a small sample of it on a special card. Have you ever had this test?”
Screening test within recommended time period
For each type of screening test, respondents were asked when they last had their test, and responses were categorized as:
“within the past year”
“within the past 1 year (FOBT only)”
“within the past 5 years”
“within the past 10 years”
“more than 10 years ago”
“don't know or not sure”

Additional covariates of interest were defined and grouped into four domains. Domains included: demographic covariates, such as age (50–64 yrs vs. ≥ 60 yrs), race and ethnicity (non-Hispanic Black, non-Hispanic White, Hispanic, and Asian), and gender; personal risk behaviors, such as current smoking and physical inactivity; socioeconomic and access to care covariates, such as personal household income, health insurance status, and whether the participant had a personal doctor; and neighborhood/contextual variables, such as neighborhood income level. Neighborhood income was grouped according to the percentage of families ≤ 200% federal poverty level; neighborhoods with 45% or more of families meeting this definition were identified as low income. Medium income was defined as 30–45% of families in a neighborhoos, and high-income neighborhoods had ≤ 30%.

Statistical Analyses

Descriptive statistics were used to determine prevalence of screening patterns, and 95% confidence intervals for both prevalence and affected population estimates were calculated with the exact binomial distribution. Prevalence estimates were age-adjusted to the 2000 U.S. population age structure.

Differences in screening practices were examined across covariates. Univariate associations were tested at the 0.05 significance level with the chi-square statistic, and stratified analyses were used to examine nonhomogeneity of associations across other covariates. Multiple logistic regression was used to identify independent predictors of colorectal cancer screening and to adjust for confounding. Two models were constructed with different dependent outcomes: having undergone any colon cancer test in a recommended time frame (every 10 years for colonoscopy, every year for FOBT, and every 5 years for flexible sigmoidoscopy) and having undergone a colonoscopy within the past 10 years.

All analyses applied sampling weights to account for the survey's complex sampling design. Poststratification weights extrapolated survey findings to neighborhood populations, based on the respondent's age, gender, and race. Data were analyzed with SAS (Version 8.02; SAS Institute Inc., Cary, NC). Standard errors for all point estimates were estimated with SUDAAN (version 8.0; Research Triangle Institute, Research Triangle Park, NC).

RESULTS

Of the 9802 adults surveyed, 3606 were aged ≥ 50 years. Among these respondents, 55% reported having a colorectal cancer screening test (Table 2) within a nationally recommended timeframe. Screening levels were nearly 60% among those aged ≥ 65 years. The proportion screened was higher among non-Hispanic Whites (60%) than among non-Hispanic Black (52%), Hispanic (52%), and Asian (36%) adults. Adults born outside of the U.S. reported significantly lower levels of any colorectal cancer screening than adults born in the U.S.

Table 2. Age-Adjusted Percentage of Adult New Yorkers Aged ≥ 50 Years Who Reported Colon Cancer Screening Tests, by Cancer Screening Type, NYC Community Health Survey 2003.
CharacteristicFOBT, past yrSigmoidoscopy, past 5 yrsColonoscopy, past 10 yrsAny timely screening
%95% CI range%95% CI range%95% CI range%95% CI range
  • FOBT: fecal occult blood test; CI: confidence interval; NH: non-Hispanic.

  • a

    Age adjusted to the 2000 U.S. population, except age stratifications.

Total31.930.0–33.89.58.3–10.841.739.7–43.755.153.2–57.1
Agea        
 50–6430.027.5–32.58.67.1–10.437.735.1–40.351.949.2–54.5
 ≥ 6534.331.4–37.210.58.7–12.646.443.4–49.559.056.1–61.8
Race or ethnicity        
 NH White31.529.0–34.19.98.4–11.748.145.4–50.960.357.8–62.8
 NH Black37.933.9–42.17.04.9–10.035.031.1–39.052.348.2–56.4
 Hispanic29.124.5–34.213.09.4–17.638.433.3–43.751.746.6–56.7
 Asian24.216.6–33.95.01.9–12.524.317.0–33.336.728.1–46.2
 Other36.925.3–50.29.74.3–20.334.722.8–48.951.838.4–64.8
Birthplace        
 United States33.831.6–36.110.48.9–12.144.942.5–47.359.256.9–61.5
 Foreign born28.325.1–31.87.75.9–9.935.432.0–39.047.844.3–51.3
Gender        
 Male31.428.5–34.411.89.7–14.345.242.1–48.457.054.0–59.9
 Female32.430.0–34.97.96.6–9.439.236.6–41.854.051.4–56.6

Important distinctions emerged when type of screening modality was examined separately (Table 2). Colonoscopy during the past 10 years was the most commonly reported screening test (42%), followed by FOBT during the past year (32%). Less than 10% of adult New Yorkers reported having a sigmoidoscopy in the past 5 years. Colonoscopy was the most frequently reported screening test in all race and ethnic groups, although non-Hispanic White adults had a significantly higher rate of colonoscopy (and overall colorectal screening) than other groups. Non-Hispanic Black adults reported higher levels of FOBT screening than other racial and ethnic groups. Reports of FOBT screening during the preceding year were particularly high among non-Hispanic Black women (40%, 95% confidence interval [CI], 35%–45%).

Men reported endoscopic screening modalities (colonoscopy and sigmoidoscopy) more frequently than women, whereas reports of FOBT were similar for both genders. The gender difference in overall colorectal cancer screening was greater for older adults compared with younger adults. Colorectal screening was significantly higher among men aged ≥ 65 years compared with men aged 50–64 years (63% vs. 52%, odds ratio [OR], 1.3; 95%CI, 1.0–1.6), whereas differences in overall colorectal cancer screening frequency among older women compared with younger women were more modest (57% vs. 52%; OR, 1.6; 95%CI 1.2–2.0). Gender differences in colorectal screening were most evident among Asians, with Asian men being twice as likely as Asian women to have received any timely colorectal cancer screening test (50% vs. 23%; OR, 2.9; 95%CI, 1.4–6.3).

Use of other clinical preventive services and healthy behaviors were both positively associated with having received a recent colon cancer screening test. Current smokers were less likely to have received a timely screening than nonsmokers (44% vs. 57%; OR, 0.62; 95% CI, 0.50–0.78), and physically inactive adults were less likely to be screened than those reporting some physical activity (50% vs. 59%; OR, 0.75; 95% CI, 0.63–0.89). Similarly, adults who did not receive an influenza vaccination in the past 12 months were less likely to have been screened than those who did obtain the vaccination (46% vs. 66%, OR, 0.44; 95% CI, 0.37–0.53).

The likelihood of having a recent colon cancer screening test (any modality) was negatively associated with socioeconomic status and access to care (Table 3). Higher screening prevalence was reported among adults with higher levels of both education and household income, relative to those with lower levels. Screening was also more common among those with private insurance and with a personal doctor. Among persons with no insurance, only 24% reported having had a screening test with the recommended frequency.

Table 3. Age-Adjusted Percentage of Adult New Yorkers Aged ≥ 50 Years Who Reported Any Timely Colon Cancer Screening Tests, by Selected Socioeconomic Characteristics, NYC Community Health Survey 2003
CharacteristicAny timely screening
%95% CI range
Total55.153.2–57.1
Education  
 Less than high school degree49.244.7–53.7
 High school degree51.647.9–55.2
 Some college5752.5–61.4
 College graduate61.157.8–64.3
Insurance status  
 Private58.452.9–63.7
 Medicaid or Medicare53.850.7–56.9
 No insurance24.414.8–37.5
Household income in U.S. dollars  
 < 25,00048.044.9–51.2
 25,000–49,99954.350.2–58.4
 50,000–74,99963.256.5–69.3
 ≥ 75,00072.967.5–77.7
 Unknown53.646.2–60.9
Neighborhood income level  
 Low income54.550.9–58.0
 Medium income50.747.3–54.0
 High income60.357.0–63.4

Levels of colorectal cancer screening at the recommended frequency were also examined across strata of neighborhood income levels. Respondents from high-income neighborhoods reported higher levels of cancer screening than those from lower income neighborhoods, although a dose-response association was less apparent than for personal household income. More importantly, perhaps, rates of colon cancer screening among those lacking health insurance were low in each neighborhood median income group except in the poorest neighborhoods (Fig. 1).

Figure 1.

Age-adjusted prevalence of timely colon cancer screening, by neighborhood income measures and type of insurance, New York City, 2003.

Multivariate Analyses

Simultaneous adjustment for multiple covariates tended to reduce the magnitude of observed differences in recent colorectal cancer screening by any modality across racial and ethnic groups and among foreign-born persons (Table 4). In contrast, the adjusted odds ratio of having a timely screening test among women compared with men was lower after adjustment (adjusted odds ratio [AOR], 0.85, 95% CI, 0.71–1.0) than in univariate analyses (OR, 0.91; 95%CI, 0.78–1.07). Significant independent predictors of not having received any colorectal cancer screening test with the recommended frequency included not having health insurance, being Asian, being < 65 years old, being female, having a household income of < $25,000, living in a medium income neighborhood, being physically inactive, and currently smoking. Lack of health insurance was the strongest independent predictor of not having received a colorectal cancer screening test within a recommended timeframe.

Table 4. Univariate and Multivariate Associations Between Timely Colon Cancer Screening Test and Covariates, and Colonoscopy in the Past 10 Years and Covariates, NYC 2003
CharacteristicAny timely colorectal cancer screening testColonoscopy, past 10 years
Adjusted OR95% CI rangeAdjusted OR95% CI range
  1. OR: odds ratio; CI: confidence interval; NH: non-Hispanic.

Age    
 50–640.800.64–0.990.680.54–0.86
 ≥ 651.0 1.0 
Race or ethnicity    
 NH White1.0 1.0 
 NH Black0.920.74–1.130.720.58–0.91
 Hispanic1.000.75–1.340.890.66–1.20
 Asian0.410.26–0.640.360.22–0.58
Birthplace    
 United States1.0 1.0 
 Foreign born0.860.70–1.050.900.73–1.11
Gender    
 Male1.0 1.0 
 Female0.860.70–1.050.740.62–0.89
Education    
 Some college education1.0 1.0 
 High school degree or less0.860.72–1.040.870.72–1.05
Insurance type    
 Private1.0 1.0 
 Medicaid or Medicare1.020.81–1.280.890.71–1.13
 Uninsured0.310.20–0.480.390.23–0.65
Household income in U.S. dollars    
 < 25,0000.680.54–0.850.700.55–0.89
 25,000–74,9990.860.69–1.070.900.72–1.12
 ≥ 75,0001.0 1.0 
Neighborhood income level    
 Low income0.970.77–1.220.900.72–1.14
 Medium income0.760.61–0.930.730.59–0.90
 High income1.0 1.0 
Physical activity    
 Some activity1.0 1.0 
 No activity0.740.63–0.880.870.73–1.04
Smoking    
 Nonsmoker1.0 1.0 
 Current smoker0.620.49–0.780.580.45–0.75

With the same covariates, a separate model was constructed examining predictors of having a screening colonoscopy in the past 10 years (Table 4). In this colonoscopy-specific model, gender and race and ethnicity remained significant predictors after adjustment for other covariates. Factors significantly associated with not having a colonoscopy within the past 10 years included age < 65 years (AOR, 0.68; 95%CI, 0.54–0.86), Black or Asian ethnicity (Black AOR 0.72, 95%CI, 0.58–0.91; Asian AOR, 0.36, 95%CI, 0.22–0.58), and female gender (AOR, 0.74, 95% CI, 0.62–0.89). Socioeconomic predictors remained similar in magnitude and significance to those associated with any timely cancer screening.

DISCUSSION

The current study presents comprehensive findings on colorectal cancer screening practices from one of the largest local, random-digit dialed telephone health surveys of adults in the U.S. We found low uptake of colon cancer screening in all groups we examined (overall 55%), leaving nearly 1 million New Yorkers aged ≥ 50 years at risk for having undetected colon cancer. Particularly low levels of screening were identified among the poor and uninsured, as well as among Asian adults, women, and current smokers.

Because a major initiative has been initiated in NYC to promote colonoscopy as the preferred screening modality for all men and women aged ≥ 50 years, we paid particular attention to colonoscopy screening practices. Colonoscopy was less frequently reported by women and by non-Hispanic Black New Yorkers. The lower uptake, or prevalence, of colonoscopy among non-Hispanic Blacks is of particular concern, given the high rates of colon cancer mortality in this population. In 2002, there were 31 colorectal cancer deaths for every 100,000 non-Hispanic Black men living in NYC—30% higher than for non-Hispanic White men, 41% higher than for Hispanic men, and nearly 50% higher than for Asian men.21 Although colorectal cancer mortality rates are lower in women than men, mortality is also highest among non-Hispanic Black women in NYC compared with women in other race and ethnic groups. Our assessment of type of screening modality shows that, despite the lower prevalence of colonoscopy, non-Hispanic Black adults in NYC are being screened for colon cancer by any modality as frequently as other race and ethnic groups. Similar national studies did not find higher uptake of FOBT among Black adults compared with other race andethnic groups.22, 23 Exactly why this may be the case in NYC, and whether it reflects patient or provider preference, is unclear. FOBT has been shown to be more common in the Northeast compared with other regions.22 Gender-specific attitudes have been reported elsewhere; factors such as embarrassment and fear about undergoing an endoscopic procedure and preference regarding the endoscopist's gender are commonly reported barriers to screening among women.23–27 Previous studies have also shown that women who preferred a female endoscopist were significantly less likely to have been screened.27

The low rate of reported sigmoidoscopy examinations in the past 5 years—less than 10%—suggests that sigmoidoscopy is a less preferred method of screening in NYC. It is unclear whether low uptake of sigmoidoscopy reflects patient or provider preference. Unfortunately, national surveys on screening practices do not distinguish between colonoscopic versus sigmoidoscopic screening; thus, we do not know whether this low uptake of sigmoidoscopy is unique to NYC.22 However, findings from a national random sample of Medicare beneficiaries suggest that the number of annual sigmoidoscopies has been declining since 1992, while colonoscopy procedures have gradually increased.28 It is also possible that the large number of gastroenterologists in the NYC area reduces the widespread use of sigmoidoscopy, which may be more commonly performed in settings that rely more heavily on general practitioners for colon cancer screening.29

We observed a strong, graded association between colon cancer screening and socioeconomic status. The strongest predictor of not being screened for colon cancer was lack of health insurance. Lack of insurance is a well recognized barrier to all types of cancer screening.27, 30 Surprisingly, the only group of uninsured adults receiving somewhat comparable levels of colorectal cancer screening as the insured were those living in the poorest neighborhoods, suggesting that screening programs have some impact in reaching the uninsured in very poor neighborhoods, most likely a result of the concentration of community health centers and public hospitals in poorer neighborhoods. Programs are perhaps less effective at doing so in the rest of the city. We know of no published findings that suggest either under- or overreporting bias tendencies for self-reported cancer screening associated with income or socioeconomic status.

Although colorectal cancer mortality rates in NYC are lowest among Asian adults, screening rates in this population are troublingly low, particularly among Asian women. Less than 1 in 4 (22%) Asian women reported having had a colorectal cancer screening test within a recommended timeframe, and only 13% of Asian women report undergoing a colonoscopy in the past 10 years. The challenges of breast and cervical cancer screening programs in successfully covering Asian communities have been recognized for some time,31–32 but the disparity in uptake of colon cancer screening has not been reported previously.

We also found that personal health risk behaviors, such as physical inactivity and smoking, were associated with lower levels of colon cancer screening. The negative association between cigarette smoking and screening for colorectal cancer is consistent with other reports that have identified lower rates of clinical preventative service use among smokers. As with many other cancers, smoking has been shown to increase the lifetime risk for colorectal cancer.33 Targeted efforts on the part of healthcare providers may help to improve screening uptake in these at-risk populations.

Despite the large and random nature of the survey design from which these findings were generated, this study has several limitations. First, the CHS is a telephone survey of the noninstitutionalized adult population and does not include persons without home telephones or institutionalized populations, such as nursing home residents and prisoners, thus limiting ability to generalize these findings to those populations. Second, not all of the contacted households agreed to take part in the survey; screening practices of nonresponders may be different from those who took part in the survey, although we have no evidence to suggest that this is the case.

Third, these data are based on self-reported behavior. Prior assessments of self-reported colorectal cancer screening behavior against medical records have suggested that the validity is moderate, with a tendency to overestimate the frequency with which one has been screened if asked over the telephone.34, 35 In addition, no attempt was made to distinguish between diagnostic and screening tests in the questions asked. The result is a likely overestimation of screening rates. This only serves to reinforce our finding that current screening levels are low for colorectal cancer. We have no reason to suspect that overreporting is more prominent in certain demographic groups.

Finally, we were unable to investigate other important potential predictors of cancer screening behavior, such as family history of colorectal cancer, referral patterns, and perceptions of the importance of screening. Each of these is a known factor influencing the likelihood of a timely colorectal cancer screening among persons aged > 50 years. The roles of provider referral patterns and other neighborhood-level factors in determining screening outcomes may also be important, particularly in the urban setting.

Screening for colorectal cancer (including strategies that solely promote colonoscopy every 10 years) is more cost-effective than either breast or cervical cancer screening.36 Nearly half of New Yorkers aged ≥ 50 years are not undergoing screening within the recommended schedule. The provider community can prevent many of the deaths due to colorectal cancer in NYC, as clinician recommendation remains one of the most powerful determinants of whether a patient undergoes colorectal cancer screening.37 Resources and interventions designed to increase screening practices will need to target poor and uninsured communities. Colonoscopy is increasingly being promoted as the preferred procedure for colon cancer screening in NYC. Ongoing collection of detailed information on the cancer screening practices throughout the city will serve as a guide to direct and evaluate these targeted efforts.

Acknowledgements

The authors thank all Citywide Colon Cancer Control Coalition (C5) members, particularly those on the Surveillance and Research committee, for their helpful comments on early drafts on this manuscript. These include Vivek Gumaste, Lynn Couey, Anna Bennet, Birgit Bogler, and Gabe Feldman. Additional comments from Mary T. Bassett and Kelly J. Henning were also greatly appreciated.

Ancillary