- Email: [email protected]
Colorectal cancer screening: Physician recommendation is influential advice to Marylanders
Preventive Medicine 41 (2005) 367 – 379 www.elsevier.com/locate/ypmed
Colorectal cancer screening: Physician recommendation is inf luential advice to Marylanders Amy Gilbert, M.P.H.a, Norma Kanarek, Ph.D.b,T a
The Johns Hopkins Bloomberg School of Public Health, Department of Epidemiology, 615 North Wolfe Street, Baltimore, MD 21205, USA b The Johns Hopkins Bloomberg School of Public Health, Department of Environmental Health Sciences, Room E7038, 615 North Wolfe Street, Baltimore, MD 21205, USA Available online 5 April 2005
Abstract Background. In comparison to the United States, Maryland is facing a significantly higher burden of colorectal cancer incidence and mortality. The primary objective of this study was to determine the predictors of colorectal cancer screening use in Maryland. Methods. We performed secondary analyses on Maryland Cancer Survey 2002 data from 2994 respondents to investigate important predictors for individual colorectal cancer screening tests. CRC screening outcomes were defined as (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, or (3) colonoscopy within the past 10 years. Results. We found that clinician recommendation for a screening test is the best predictor in both age categories (50–64 years and 65+ years); it is a very strong indicator and consistently improves the odds of use by a factor of at least 8 for any screening test. Conclusions. There remains a great need for improved colorectal cancer screening in Maryland. According to our results, it is clear that the most influential way to improve overall colorectal cancer screening for each test and both age groups is to increase clinician recommendation for these tests. D 2005 Elsevier Inc. All rights reserved. Keywords: FOBT; Sigmoidoscopy; Colonoscopy; Colorectal cancer screening; Colon cancer screening
Introduction Colorectal cancer (CRC) is the second most common fatal malignancy in Maryland [1]. According to the Maryland Cigarette Restitution Fund’s 2003 Annual Cancer Report, 2778 new CRC cases were diagnosed and 1158 deaths due to CRC were recorded in Maryland during 2000. In comparison to the United States, Maryland is facing a significantly higher burden of CRC incidence and mortality. Effective CRC screening procedures allow for the detection and removal of precursor lesions and facilitate earlier identification of malignancies at stages that are more
T Corresponding author. Fax: +1 410 955 0617. E-mail addresses: [email protected] (A. Gilbert)8 [email protected] (N. Kanarek). 0091-7435/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2005.01.008
amenable to treatment [2,3]. The burden of CRC can be reduced by increased colorectal cancer screening and adherence to guidelines such as those issued by the American Cancer Society (ACS) [4]. For average risk individuals ages 50 years and older, the ACS lists the following as acceptable options for CRC screening: (1) annual fecal occult blood test, (2) flexible sigmoidoscopy every 5 years, or (3) colonoscopy every 10 years with the caveat that completion of timely fecal occult blood test and sigmoidoscopy is preferred over the individual tests alone. In practice, however, the tests are conducted in different settings and at different times. According to 2001 BRFSS results for Maryland, 44.4% of residents 50 years and older reported FOBT use within the past 2 years and 52.2% reported ever having had a sigmoidoscopy or colonoscopy [1]. Fifty-eight percent of Maryland Cancer Survey 2002 (MCS) responders 50 years
368
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
of age or older reported ever having had a sigmoidoscopy or colonoscopy [5,6]. This indicates that at least 42% of residents age 50 or older have not been screened by a method that visualizes the colon. The primary objective of this study was to determine the predictors of CRC screening test use among Marylanders 50 years or older. Additionally, we sought to assess and quantify the outstanding need for CRC screening by individual CRC test and age.
Background Studies have previously shown an association between many factors and CRC screening. The literature suggests that age and sex influence screening behavior. Thomas et al. found peak compliance for annual FOBT to be around age 70 with lower screening seen among the youngest (55 years or younger) and the oldest (80 years or older) [7]. Lemon et al. found that men aged 65–74 were more likely to be currently CRC screened than men aged 50–64 [8]. Women perceive themselves to be at lower risk for CRC than men, as do their healthcare providers [9]. This may lead to differences in screening adherence between males and females. Race may also play a role in CRC screening. Escarce et al. demonstrated that after adjusting for age and sex, White elderly persons are significantly more likely than Black elderly persons to receive sigmoidoscopy or colonoscopy, a difference not entirely accounted for by access to health care [10]. Studies also propose that there may be differences in CRC screening practices in urban and rural settings [11]. Rural residents may have less access to the facilities, instruments, and trained physicians needed for CRC screening. Health status may also influence the utilization of preventive health practices. Data indicates that breast and cervical cancer screening rates decrease as comorbidity increases [12]. This same effect may exist for CRC screening. Health insurance coverage and access to health care have been shown to be associated with CRC screening participation [8,11]. However, a study in Washington State concluded that use of CRC screening tests did not substantially change once Medicare coverage was available [13]. This and other literature reinforce that medical coverage alone is not enough to effectively increase CRC screening utilization. In studying factors associated with screening sigmoidoscopy among general medicine patients, Lewis and Jensen found that patients were five times more likely (OR = 5.02) to get a sigmoidoscopy if a clinician had advised it [14]. Holt concluded that recommendation and demonstration of concern by a physician may be the primary motivating factors in screening compliance [15]. Similarly, McCarthy and Moskowitz noted evidence that 50–75% of patients offered screening sigmoidoscopy will accept [16]. Unfortunately, primary care physician recommendations for CRC screening have been shown to be inconsistent and inappropriate [17]. Knowledge of CRC and the availability of screening can also be implicated in compliance [9]. Furthermore, family history of
CRC, non-smoking status, and higher education have all been found to be associated with higher levels of screening [8].
Methods Data source The Maryland Cigarette Restitution Fund, created in 2000 with tobacco settlement money by the Maryland State Legislature, established the Cancer Prevention, Education, Screening, and Treatment (CPEST) Program under the Center for Cancer Surveillance and Control at the Department of Health and Mental Hygiene (DHMH). The Surveillance and Evaluation Unit of the CPEST Program commissioned the MCS, a population-based, random digit dial, computer-assisted land line telephone interview utilizing list-assisted disproportionate stratified (urban/rural) sampling. The MCS was conducted by a research team in the Department of Epidemiology and Preventive Medicine at the University of Maryland School of Medicine in Baltimore. The survey asked questions about cancer screening practices, knowledge of cancer and cancer screening, and cancer risk behaviors for selected cancers among English-speaking Marylanders age 40 and older residing in private residences. A total of 5071 interviews were completed from 84,172 phone numbers called. The Council of American Survey Research Organizations response rate was 38.4% (completed interviews/[known eligible + presumed eligible]) [5]. The completion rate (completed interviews/known eligible) was 65.4%.
Design We obtained the MCS data set [6] and performed secondary analyses to investigate CRC screening usage in accordance with ACS screening guidelines for fecal occult blood test (FOBT), sigmoidoscopy, and colonoscopy. The three screening tests were evaluated independently and respondents reporting use of more than one test were counted more than once. Survey questions did not allow a dual response to sigmoidoscopy and colonoscopy use, however. Several distinct factor groups were considered and evaluated as contributors to CRC screening: (1) demographics [age, sex, race, Hispanic ethnicity, marital status], (2) socioeconomic status [urban/rural residence, education, income], (3) health status [self-reported health status], (4) health behaviors [current smoking status, alcohol consumption, body mass index (BMI)], (5) personal cancer predictors [level of concern about cancer, family history of CRC, knowledge of CRC screening], and (6) health system predictors [clinician recommendation for screening test, usual source of care, health insurance, inability to receive needed medical care]. Institutional Review Boards of the
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Maryland DHMH and the Johns Hopkins Bloomberg School of Public Health approved this study.
Data analysis CRC screening test outcomes were defined as (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, or (3) colonoscopy within the past 10 years. CRC screening test outcomes were determined using survey variables that indicated whether FOBT, sigmoidoscopy, and/or colonoscopy had been performed, and if so, the time since each exam was done. Respondent observations where race was missing were not included in the initial data set since this variable was required for assigning a weight to the respondent. Data from 418 participants were dropped from analysis due to ddon’t knowT or drefusedT responses for most covariates. Missing values for income (n = 512) and BMI (n = 113) were retained and analyzed as separate categories. In initial multivariate analyses, there was a significant interaction for age and health insurance. Thus, age groups were considered in separate models (50–64 years and 65+ years). Age was tested as an additional continuous variable, found not to add to either age group model, and was not retained. Due to DHMH data reporting guidelines for small numbers, 24 observations reporting no health insurance in the 65+ age category were dropped, leaving this group with 100% health insurance coverage. As a result, health insurance was not evaluated in the 65+ age group. Survey responses analyzed totaled 1730 in the 50–64 age group and 1264 in the 65+ age group. All CRC screening outcomes, Hispanic ethnicity, current smoking status, current alcohol consumption, knowledge of screening tests for CRC, ever had clinician recommendation for FOBT, ever had clinician recommendation for sigmoidoscopy or colonoscopy, usual source of care, health insurance, and inability to receive needed medical care were dichotomous variables coded byesQ (1) or bnoQ (0). Health insurance was assigned bnoQ (0) if no current health insurance or no health insurance in the past 12 months was reported. Family history of CRC byesQ (1) reflects reported CRC diagnosis of any first-degree relative. Respondents were considered to have a usual source of care if at least one primary health care provider was reported (1). Race was categorized as White, Black, or Other. Indicator variables were used to identify marital status as (a) married or partner of an unmarried couple; (b) divorced, widowed, or separated; and (c) never married. Education categories were defined as (a) less than high school; (b) high school graduate; (c) 1–3 years of college; (d) college graduate; (e) master’s degree, advanced professional or doctoral degree. Categories for annual household income were represented in thousands of dollars by (a) less than 25; (b) 25–49.999; (c) 50–74.999; and (d) 75+. Self-reported health status was grouped as (a) excellent, very good, or
369
good; (b) fair; and (c) poor. The indicator variable for BMI was defined as (a) not overweight or obese; and (b) overweight or obese. Respondents were asked about their concern about getting cancer in the future and the categories were (a) very concerned; (b) somewhat concerned; and (c) not at all concerned. Unweighted multiple logistic regression analysis was performed by age category (50–64 years and 65+ years) for (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, (3) colonoscopy within the past 10 years, and (4) screening colonoscopy within the past 10 years. A timely screening colonoscopy included only those that were reported as routine/screening exams or prompted by family history. Since our primary objective was etiological in nature, the most conservative approach was to perform an unweighted regression analysis [18]. Likelihood ratio tests were done to evaluate the statistical significance ( P b 0.05) of the following four variable groups for each outcome: health status, health behaviors, personal cancer predictors, and health system predictors. In accordance with known associations, the demographics and socioeconomic status variable groups (except for marital status) were retained in all models as covariates. Completion of timely FOBT was included as a covariate in the colonoscopy model. Models were then run with only significant variable groups retained. Following that, individual non-significant variables (likelihood ratio tests P N 0.05) within the variable groups were removed to assemble the most parsimonious model for each outcome and age category. Final models (Table 3) estimated levels of CRC screening in each age group, controlling for other important factors. Models were assessed for goodness of fit using the Hosmer–Lemeshow test. Stata version 8.0 was used to analyze these data.
Results Respondent characteristics Weighted respondent characteristics are provided in Table 1 by age. In the 50–64 year group, 85 respondents (5.30%) reported having had a timely FOBT and timely sigmoidoscopy; 240 respondents (13.80%) reported having had a timely FOBT and timely colonoscopy. In the 65+ year group, 75 respondents (6.08%) reported having had a timely FOBT and timely sigmoidoscopy; 275 respondents (23.38%) reporting having had a timely FOBT and timely colonoscopy. Of the timely colonoscopies, 23.29% and 33.03% were reported as screening in the 50–64 and 65+ age groups, respectively. Tables 2a and 2b summarize CRC screening in Maryland by age group and screening test. Usage in Maryland is higher for older adults on every screening test. Colonoscopy use is most prevalent (ages 50– 64, 36.44%; ages 65 and above, 51.44%), FOBT is next (30.83% and 39.23%), and sigmoidoscopy is least (11.09% and 12.29%). Use was consistently and statistically signifi-
370
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 1 Respondent characteristics by age category, Maryland Cancer Survey (2002)
Table 1 (continued) Weighted percent (SE)
Weighted percent (SE) 50–64 yearsa (n = 1730) Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No
65+ yearsb (n = 1264)
49.17 [1.41] 50.83 [1.41]
41.09 [1.61] 58.91 [1.61]
71.93 [1.37] 22.32 [1.24] 5.75 [0.87]
80.15 [1.42] 16.56 [1.33] 3.29 [0.68]
2.50 [0.45] 97.50 [0.45]
1.10 [0.29] 98.90 [0.29]
72.16 [1.18]
59.58 [1.51]
23.06 [1.11]
37.81 [1.48]
4.78 [0.52]
2.61 [0.49]
78.05 [0.72] 21.95 [0.72]
76.73 [0.94] 23.27 [0.94]
9.59 27.83 20.54 21.71 20.33
15.45 31.70 20.44 18.26 14.15
13.26 23.16 16.26 34.59 12.73
[0.86] [1.24] [1.12] [1.19] [1.13]
[1.20] [1.45] [1.26] [1.28] [1.12]
[0.95] [1.15] [1.04] [1.36] [0.93]
26.13 [1.36] 28.35 [1.43] 11.83 [1.11] 11.66 [1.07] 22.03 [1.32]
82.65 [1.07] 13.03 [0.94] 4.32 [0.60]
78.11 [1.32] 16.84 [1.19] 5.05 [0.70]
20.55 [1.17] 79.45 [1.17]
7.40 [0.90] 92.60 [0.90]
46.09 [1.40] 53.91 [1.40]
53.50 [1.60] 46.50 [1.60]
33.86 [1.33] 62.27 [1.36] 3.87 [0.50]
42.76 [1.57] 54.37 [1.59] 2.87 [0.50]
36.50 [1.38] 50.22 [1.41] 13.28 [0.94]
29.60 [1.47] 44.24 [1.58] 26.16 [1.43]
11.79 [0.87] 88.21 [0.87]
14.12 [1.12] 85.88 [1.12]
Knowledge of screening tests for CRC Yes No Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No CRC test use No timely test Timely FOBT onlyd Timely sigmoidoscopy onlye Timely colonoscopy onlyf Timely FOBT and timely sigmoidoscopyg Timely FOBT and timely colonoscopyh Timely sigmoidoscopy and timely colonoscopyi
50–64 yearsa (n = 1730)
65+ yearsb (n = 1264)
95.14 [0.72] 4.86 [0.72]
93.57 [0.76] 6.43 [0.76]
38.86 [1.37] 61.14 [1.37]
44.64 [1.59] 55.36 [1.59]
60.00 [1.39] 40.00 [1.39]
70.55 [1.46] 29.55 [1.46]
93.37 [0.68] 6.63 [0.68]
97.03 [0.53] 2.97 [0.53]
90.89 [0.77] 9.11 [0.77]
100.00c 0c
5.54 [0.64] 94.46 [0.64]
1.79 [0.40] 98.21 [0.40]
40.76 11.72 5.78 22.63 5.30
26.50 9.77 6.21 28.06 6.08
[0.01] [0.01] [0.01] [0.01] [0.01]
13.81 [0.01] 0
[0.01] [0.01] [0.01] [0.01] [0.01]
23.38 [0.01] 0
a
Total Maryland population for 50–64 years of age = 739,400. Total Maryland population for 65+ years of age = 504,000. c Twenty-four observations reporting no health insurance in the 65+ year age group were dropped. d FOBT within 1 year. e Sigmoidoscopy within 5 years. f Colonoscopy within 10 years. g FOBT within 1 year and sigmoidoscopy within 5 years. h FOBT within 1 year and colonoscopy within 10 years. i Sigmoidoscopy within 5 years and colonoscopy within 10 years. b
cantly higher among those who reported receiving a clinician recommendation for screening, having a usual source of care, or having health insurance (ages 50–64 only). These are differences most relevant to our study; as illustrated in Tables 2a and 2b, other statistically significant differences were also seen in these data. Multiple logistic regression Table 3 depicts the final multiple logistic regression models for CRC screening in Maryland. Though we modeled compliance to timely FOBT and timely sigmoidoscopy, which was basically no different from timely FOBT
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
alone, and compliance to any CRC screening test, which was very similar to the model for colonoscopy, we present the models for each test separately. Predictors of use, as assessed by logistic regression analysis, vary by age and screening test. Model results for persons aged 50–64 FOBT. The only significant predictor of FOBT use was clinician recommendation for the test (OR = 36.89; 95% CI: 27.06–50.29; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 4.91, P = 0.55). Sigmoidoscopy. Sigmoidoscopy use was significantly lower among Blacks when compared to Whites (OR = 0.57; 95% CI: 0.35–0.93; P = 0.026). Rural residents are less likely have had a timely sigmoidoscopy than urban residents (OR = 0.61; 95% CI: 0.42–0.88; P = 0.009). Persons in fair health are less likely to have had a timely sigmoidoscopy than those in excellent, very good, or good health (OR = 0.49; 95% CI: 0.27–0.89; P = 0.019). Current smokers are less likely to have had a timely sigmoidoscopy than non-smokers (OR = 0.39; 95% CI: 0.22–0.69; P = 0.001). Family history of CRC was borderline significant and indicated that persons with family history were less likely to have had a timely sigmoidoscopy (OR = 0.60; 95% CI: 0.36–0.99; P = 0.050). Clinician recommendation for sigmoidoscopy increases the odds of use (OR = 11.09; 95% CI: 6.07– 20.25; P b 0.001). The final model fits the data (Hosmer– Lemeshow test statistic = 9.92, P = 0.13). Colonoscopy. Reporting a family history of CRC significantly increases the odds of colonoscopy use (OR = 2.56; 95% CI: 1.78–3.68; P b 0.001). Independently, reporting having a clinician recommendation for the test raised the odds of colonoscopy by more than 30:1 (OR = 31.76; 95% CI: 21.14–47.73; P b 0.001). Usual source of care was also important and positively associated with colonoscopy use (OR = 2.83; 95% CI: 1.52–5.27; P = 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 1.64, P = 0.94). Screening colonoscopy. Women have decreased odds of screening colonoscopy use (OR = 0.66; 95% CI: 0.51–0.86; P = 0.002). Individuals of Black race or Hispanic ethnicity have increased odds of screening colonoscopy use (OR = 1.87; 95% CI: 1.34–2.62; P b 0.001 and OR = 2.26; 95% CI: 1.05–4.88; P = 0.038, respectively). Income also appears to play a role in screening colonoscopy use for 50- to 64-year-olds. Those in annual income brackets of $25,000–49,999 and $75,000+ (compared to b$25,000) have increased odds of screening colonoscopy use (OR = 1.66; 95% CI: 1.01–2.73; P = 0.05 and OR = 1.81; 95% CI: 1.09–3.01; P = 0.02, respectively). Family history of CRC improves the odds of screening colonoscopy (OR = 2.71;
371
95% CI: 1.93–3.81; P b 0.001). Clinician recommendation significantly raised the odds of screening colonoscopy (OR = 18.26; 95% CI: 11.27–29.57; P b 0.001). Usual source of care was also positively associated with screening colonoscopy use (OR = 3.10; 95% CI: 1.41–6.83; P = 0.005). The final model fits the data (Hosmer–Lemeshow test statistic = 1.57, P = 0.95). Model results for persons aged 65+ FOBT. Blacks were more likely to have had a timely FOBT than Whites (OR = 2.20; 95% CI: 1.38–3.51; P = 0.001). Those with poor health status had almost 4 times the odds of FOBT use (OR = 3.75; 95% CI: 1.87–7.52; P b 0.001). Those who reported a clinician recommendation for FOBT had over 26 times the odds of use (OR = 26.73; 95% CI: 19.25–37.11; P b 0.001). Having a usual source of care improves the odds of having a timely FOBT (OR = 3.28; 95% CI: 1.06–10.17; P = 0.039). The model fits the data (Hosmer–Lemeshow test statistic = 6.67, P = 0.35). Sigmoidoscopy. An annual income of $50,000–$74,999 compared to b$25,000 significantly improves the odds of sigmoidoscopy use (OR = 2.64; 95% CI: 1.30–5.37; P = 0.008). Clinician recommendation for the test improves sigmoidoscopy use by more than 11:1 (OR = 11.62; 95% CI: 5.04–26.76; P b 0.001). The model fits the data (Hosmer–Lemeshow test statistic = 3.45, P = 0.75). Colonoscopy. Current smokers have a borderline statistically significant decreased odds of colonoscopy use (OR = 0.59; 95% CI: 0.35–1.01; P = 0.055). Current drinkers also have a decreased odds of colonoscopy use (OR = 0.74; 95% CI: 0.55–0.99; P = 0.045). Reporting having a family history of CRC improves the odds of colonoscopy (OR = 2.23; 95% CI: 1.48–3.38; P b 0.001). Reporting having a clinician recommendation for colonoscopy significantly increases the odds of use by more than 20:1 (OR = 21.71, 95% CI: 14.87–31.72; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 4.38, P = 0.63). Screening colonoscopy. Family history of CRC increases the odds of screening colonoscopy use (OR = 2.60; 95% CI: 1.83–3.72; P b 0.001). Clinician recommendation for colonoscopy significantly improves the odds of screening colonoscopy (OR = 8.70; 95% CI: 5.82–12.99; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 6.28, P = 0.39).
Discussion The main objective of this analysis was to determine what factors were the primary predictors of CRC screening test use in Maryland. This study was able to take a comprehensive approach to understanding the factors that
372
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 2a Colorectal cancer screening as reported by Maryland Cancer Survey 2002 respondents
Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No
50–64 years FOBTa
50–64 years Sigmoidoscopyb
50–64 years Colonoscopyc
50–64 years Screening Colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
30.69 [26.87–34.80] 30.96 [27.82–34.24]
12.04 [9.57–15.03] 10.17 [8.23–12.49]
38.21 [34.16–42.43] 34.72 [31.51–38.08]
26.09 [22.55–29.97] 20.58 [17.93–23.52]
30.55 [27.80–33.44] 35.67 [29.79–42.01] 15.55 [8.07–27.84]
11.98 [10.10–14.14] 7.73 [4.96–11.86] 12.98 [6.22–25.14]
36.93 [34.05–39.91] 36.45 [30.55–42.79] 30.14 [18.47–45.11]
22.62 [20.18–25.26] 27.28 [21.90–33.41] 16.18 [8.42–28.85]
30.02 [15.40–50.26] 30.85 [28.35–33.47]
9.44 [2.76–27.65] 11.13 [9.51–12.99]
48.74 [31.00–66.80] 36.12 [33.51–38.82]
45.87 [28.14–64.72] 22.71 [20.49–25.10]
30.80 [27.84–33.94]
11.98 [9.98–14.31]
37.25 [34.08–40.52]
24.37 [21.62–27.35]
30.98 [26.16–36.26]
8.87 [6.39–12.20]
34.69 [29.89–39.82]
20.75 [16.91–25.20]
30.42 [21.23–41.50]
8.26 [4.04–16.14]
32.64 [23.30–43.61]
19.23 [11.82–29.72]
30.85 [27.91–33.95] 30.74 [26.67–35.14]
12.02 [10.08–14.28] 7.75 [5.71–10.45]
36.51 [33.43–39.71] 36.17 [31.91–40.67]
23.61 [20.96–26.49] 22.15 [18.56–26.21]
19.15 29.36 31.72 32.97 35.16
[12.81–27.62] [24.91–34.23] [26.34–37.64] [27.62–38.80] [29.59–41.18]
5.86 [2.89–11.53] 11.12 [8.11–15.05] 10.34 [7.19–14.64] 12.21 [8.95–16.44] 13.07 [9.58–17.59]
29.36 31.09 36.95 40.76 41.96
[21.57–38.58] [26.67–35.88] [31.37–42.91] [34.99–46.80] [36.02–48.13]
19.29 18.43 23.75 26.19 28.27
[12.83–27.95] [14.79–22.72] [19.10–29.13] [21.37–31.65] [22.99–34.22]
29.89 24.67 31.06 32.85 37.22
[23.07–37.73] [20.30–37.73] [25.12–37.69] [28.55–37.47] [30.22–44.80]
10.12 7.40 10.22 13.13 14.34
28.56 33.98 37.92 40.55 36.04
[21.84–36.40] [29.06–39.26] [31.51–44.78] [35.94–45.33] [29.23–43.46]
14.37 22.04 26.06 26.56 22.42
[9.74–20.17] [17.80–26.97] [20.35–32.71] [22.61–30.92] [16.80–29.27]
[6.34–15.79] [5.01–10.82] [6.75–15.17] [10.28–16.63] [9.72–20.66]
30.91 [28.21–33.76]
11.74 [9.95–13.81]
36.29 [33.45–39.22]
23.86 [21.41–26.50]
30.78 [24.08–38.39] 29.33 [18.50–43.15]
7.41 [4.21–12.71] 9.64 [3.96–21.63]
38.28 [31.20–45.90] 33.76 [22.12–47.77]
23.56 [17.39–31.10] 11.57 [6.04–21.03]
26.64 [21.26–32.81] 31.91 [29.15–34.80]
5.23 [2.88–9.30] 12.60 [10.73–14.74]
28.59 [23.14–34.74] 38.47 [35.56–41.46]
18.54 [14.24–23.78] 24.52 [21.98–27.25]
33.86 [30.12–37.82] 28.23 [25.02–31.68]
11.61 [9.27–14.44] 10.64 [8.55–13.17]
36.42 [32.63–40.40] 36.45 [32.94–40.10]
22.68 [19.46–26.26] 23.81 [20.79–27.12]
32.03 [27.80–36.57] 30.31 [27.18–33.63] 28.72 [18.69–41.38]
14.12 [11.11–17.78] 9.72 [7.86–11.95] 6.55 [2.52–16.00]
33.55 [29.29–38.10] 37.68 [34.36–41.13] 41.63 [29.57–54.78]
20.93 [17.40–24.97] 24.40 [21.51–27.53] 26.08 [16.34–38.93]
30.70 [26.52–35.22] 29.81 [26.52–33.33] 35.02 [28.07–42.68]
9.77 [7.35–12.86] 12.51 [10.22–15.22] 9.34 [5.62–15.14]
38.34 [33.85–43.04] 35.04 [31.58–38.66] 36.49 [29.45–44.16]
23.59 [19.89–27.74] 22.38 [19.43–25.63] 25.91 [19.49–33.57]
32.78 [26.21–40.09] 30.57 [27.92–33.35]
9.37 [5.80–14.79] 11.32 [9.59–13.31]
57.14 [49.32–64.61] 33.67 [30.95–36.50]
43.65 [36.20–51.39] 20.57 [18.30–23.05]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
373
Table 2a (continued)
Knowledge of screening tests for CRC Yes No Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No Total use, weighted % (95% CI) a b c d e f
50–64 years FOBTa
50–64 years Sigmoidoscopyb
50–64 years Colonoscopyc
50–64 years Screening Colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
31.74 [29.19–34.40] 13.01 [6.41–24.61]
11.59 [9.92–13.51] 1.18 [0.16–8.11]
37.11 [34.46–39.84] 23.28 [13.44–37.22]
23.99 [21.68–26.46] 9.67 [4.49–19.58]
70.72 [66.56–74.55] 5.48 [4.12–7.25]
N/Ae N/Ae
N/Ae N/Ae
N/Ae N/Ae
N/Af N/Af
17.41 [14.9–20.25] 1.60 [0.85–2.99]
57.32 [53.82–60.75] 5.10 [3.34–7.71]
36.52 [33.22–39.96] 3.44 [2.08–5.63]
32.26 [29.65–34.98] 10.69 [6.23–17.75]
11.36 [9.69–13.28] 7.18 [3.02–16.11]
38.26 [35.54–41.06] 10.70 [6.21–17.83]
24.50 [22.14–27.02] 6.26 [2.81–13.36]
32.09 [29.46–34.85] 18.2 [12.43–25.86]
11.65 [9.92–13.62] 5.51 [2.64–11.15]
37.82 [35.07–40.66] 22.59 [16.30–30.42]
24.51 [22.12–27.08] 11.10 [6.75–17.70]
29.80 [20.26–41.50] 30.89 [28.34–33.56] 30.83 [28.30–33.36]
5.44 [2.11–13.31] 11.42 [9.75–13.33] 11.09 [9.38–12.80]
36.25 [25.93–48.02] 36.45 [33.78–39.20] 36.44 [33.81–39.07]
17.04 [10.40–26.65] 23.66 [21.34–26.14] 23.29 [20.96–25.63]
FOBT within 1 year. Sigmoidoscopy within the past 5 years. Colonoscopy within the past 10 years. Screening colonoscopy within the past 10 years. Recommendation for FOBT was assessed only for FOBT compliance. Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy.
influence CRC screening test use. We found that clinician recommendation for a screening test is the best predictor for all forms of screening in both age categories; it is a very strong indicator and consistently improves the odds of use by a factor of at least 8 in all instances. During the uptake phase of mammography, clinician recommendation has been cited for significantly improving screening use as well [19–22]. In this study, the predictors of use differ by age group (50–64 years and 65+ years) and screening test (FOBT, sigmoidoscopy, colonoscopy, and screening colonoscopy). Following clinician recommendation, the strongest predictor of less sigmoidoscopy use in the 50–64 group is current smoking status (decreased use in current smokers, compared to never or former smoker), then fair health status (compared to excellent, very good, or good health), family history of CRC (compared to no first-degree relatives affected), Black race (compared to Whites), and place of residence (compared to urban). For colonoscopy and screening colonoscopy in the 50–64 group, family history improves the odds of use. There were additional predictors for screening colonoscopy in this younger group. Males were more likely than females, Blacks more likely than Whites, and Hispanics more likely than non-Hispanics to have a screening colonoscopy. Higher
income ($25,000–49,999 and N$75,000) and usual source of care also improved the odds of screening colonoscopy among the 50- to 64-year-olds. In the 65+ group, again after physician recommendation, FOBT use is less among those with poor health status (compared to excellent, very good, or good health) and more for those with a usual source of care (compared to no usual source of care), and Black race (compared to Whites). Sigmoidoscopy use in the 65+ group is significantly improved among those with an annual household income of $50,000–$74,999 as compared to those earning less than $25,000. In addition to physician recommendation, colonoscopy and screening colonoscopy use in the 65+ group is improved for those with a family history of CRC. The odds of colonoscopy are decreased for current drinkers aged 65+. Why these variables play a role in CRC screening use can be hypothesized. Less use among current smokers and current drinkers may be explained because this group, in general, may not be as health conscious as non-smokers or non-drinkers. Fair health status may prevent a person from taking part in preventive health practices like CRC screening. However, increased use was noted among those with poor health. We found that persons aged 50–64 in fair health
374
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 2b Colorectal cancer screening as reported by Maryland Cancer Survey 2002 respondents
Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No Knowledge of screening tests for CRC Yes No
65+ years FOBTa
65+ years Sigmoidoscopyb
65+ years Colonoscopyc
65+ years Screening colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
38.93 [33.85–44.27] 39.44 [35.75–43.25]
14.18 [10.91–18.24] 10.97 [8.84–13.54]
52.83 [47.49–58.10] 50.47 [46.66–54.27]
35.97 [31.06–41.18] 30.99 [27.60–34.60]
37.27 [34.07–40.58] 48.87 [40.15–57.66] 38.55 [20.59–60.28]
12.81 [10.72–15.23] 9.13 [5.17–15.62] 15.67 [6.21–34.29]
51.67 [48.32–55.01] 53.11 [44.26–61.77] 37.31 [20.6–57.73]
33.70 [30.58–36.96] 32.45 [24.99–40.93] 19.83 [8.07–41.06]
29.23 [10.90–58.24] 39.34 [36.28–42.49]
10.8 [2.81–33.61] 12.31 [10.39–14.52]
53.97 [28.43–77.59] 51.41 [48.25–54.55]
30.85 [12.59–58.00] 33.06 [30.17–36.07]
41.8 [37.56–46.16]
14.68 [11.90–17.99]
52.02 [47.65–56.35]
34.97 [30.97–39.20]
35.12 [30.93–39.56] 40.04 [23.47–59.25]
8.52 [6.45–11.16] 12.43 [4.99–27.72]
50.86 [46.42–55.29] 46.54 [28.80–65.20]
30.10 [26.18–34.34] 31.19 [16.19–51.55]
40.17 [36.52–43.93] 36.13 [31.23–41.34]
12.41 [10.15–15.09] 11.89 [8.95–15.62]
52.46 [48.68–56.21] 48.06 [42.89–53.28]
33.53 [30.11–37.14] 31.39 [26.70–36.49]
40.03 34.93 44.55 40.84 38.24
[32.24–48.36] [29.94–40.26] [37.93–51.37] [33.42–48.70] [30.45–46.68]
6.68 9.25 13.15 15.93 19.3
[3.77–11.54] [6.66–12.71] [9.22–18.43] [11.00–22.51] [13.49–26.84]
46.84 51.51 55.21 54.02 47.49
[38.63–55.23] [46.16–56.83] [48.44–61.80] [46.29–61.56] [39.26–55.86]
29.99 34.74 38.53 29.66 28.94
38.43 38.69 38.66 41.31 40.08
[32.73–44.46] [33.15–44.55] [29.60–48.57] [32.20–51.06] [33.81–46.69]
6.06 12.99 22.03 18.75 10.15
[3.95–9.18] [9.57–17.38] [14.96–31.20] [12.29–27.53] [6.94–14.59]
46.77 55.09 47.04 54.41 53.06
[41.00–52.62] [49.25–60.79] [37.37–56.94] [44.74–63.76] [46.38–59.63]
28.13 [23.20–33.65] 37.11 [31.64–42.93] 30.53 [22.07–40.54] 33.29 [24.93–42.85] 34.82 [28.87–41.28]
[22.89–38.22] [29.81–40.02] [32.15–45.34] [23.17–37.09] [22.10–36.89]
38.66 [35.24–42.19] 38.89 [31.72–46.57] 49.25 [35.43–63.19]
13.4 [11.17–15.99] 8.26 [4.98–13.38] 8.6 [3.55–19.39]
50.18 [46.64–53.72] 54.51 [46.89–61.93] 60.59 [46.55–73.07]
33.36 [30.12–36.77] 32.73 [25.99–40.28] 28.94 [17.72–43.51]
26.57 [17.69–37.86] 40.24 [37.08–43.49]
6.71 [3.00–14.30] 12.74 [10.73–15.06]
32.29 [22.40–44.07] 52.97 [49.73–56.18]
20.10 [12.40–30.90] 34.07 [31.07–37.19]
40.25 [36.16–44.47] 38.06 [33.60–42.73]
9.66 [7.44–12.45] 15.32 [12.30–18.94]
51.74 [47.56–55.89] 51.09 [46.36–55.79]
32.26 [28.51–36.26] 33.92 [29.61–38.51]
37.53 [33.11–42.16] 39.85 [35.64–44.21] 52.91 [35.77–69.40]
11.33 [8.79–14.48] 12.59 [9.96–15.81] 20.93 [9.84–39.1]
51.7 [47.00–56.38] 51.79 [47.46–56.09] 40.75 [25.23–58.37]
33.95 [29.57–38.62] 32.96 [29.12–37.05] 20.71 [10.31–37.24]
36.69 [31.37–42.36] 42.28 [37.79–46.90] 36.94 [30.87–43.46]
11.18 [8.03–15.35] 15.48 [12.46–19.06] 8.16 [5.24–12.51]
53.22 [47.36–58.99] 54.23 [49.65–58.75] 44.69 [38.50–51.04]
34.83 [29.62–40.43] 34.25 [30.01–38.77] 28.94 [23.51–35.04]
45.15 [36.96–53.61] 38.26 [35.01–41.62]
10.02 [5.84–16.66] 12.66 [10.61–15.05]
68.49 [59.95–75.94] 48.63 [45.27–52.01]
49.78 [41.49–58.09] 30.28 [27.26–33.48]
40.11 [36.95–43.36] 26.39 [17.34–37.99]
12.96 [10.95–15.28] 2.53 [0.62–9.77]
52.75 [49.50–55.98] 32.27 [22.03–44.56]
33.76 [30.78–36.87] 22.53 [13.59–34.98]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
375
Table 2b (continued)
Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No Total use, weighted % (95% CI) a b c d e f g
65+ years FOBTa
65+ years Sigmoidoscopyb
65+ years Colonoscopyc
65+ years Screening colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
73.34 [68.98–77.29] 11.73 [9.25–14.76]
N/Ae N/Ae
N/Ae N/Ae
N/Ae N/Ae
N/Af N/Af
16.76 [14.18–19.71] 1.59 [0.63–3.93]
68.74 [65.20–72.08] 9.97 [7.15–13.75]
43.74 [40.09–47.46] 7.39 [5.07–10.65]
39.95 [36.86–43.13] 15.67 [6.50–33.17]
12.51 [10.57–14.74] 5.20 [0.72–29.16]
51.88 [48.69–55.05] 37.02 [21.90–55.20]
33.31 [30.40–36.36] 23.90 [11.67–42.75]
N/Ag N/Ag
N/Ag N/Ag
N/Ag N/Ag
N/Ag N/Ag
31.66 [15.18–54.52] 39.37 [36.30–42.52] 39.23 [36.15–42.31]
7.04 [1.63–25.69] 12.39 [10.46–14.61] 12.29 [10.25–14.33]
52.41 [31.15–72.84] 51.42 [48.25–54.57] 51.44 [48.30–54.58]
13.63 [3.89–38.07] 33.39 [30.49–36.42] 33.03 [30.07–35.99]
FOBT within 1 year. Sigmoidoscopy within the past 5 years. Colonoscopy within the past 10 years. Screening colonoscopy within the past 10 years. Recommendation for FOBT was assessed only for FOBT compliance. Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy. Health insurance coverage was 100% in the 65+ year age group after dropping 24 observations reporting no insurance.
are less likely to have had a timely sigmoidoscopy than those in excellent, very good, or good health. In contrast, for ages 65+, poor health status was associated with increased odds of FOBT use. These findings suggest that self-reported health status is an important indicator of CRC screening use and may influence use of other preventive health services. Another explanation is that those with usual source of care, higher incomes, or urban residences are likely to have improved access to health care resources and thus are more likely to participate in CRC screening. After controlling for these and other factors, health insurance in the 50–64 age group, however, did not affect CRC screening use. A family history of CRC would indicate high-risk status for which colonoscopy would be the most appropriate form of screening. This explains why family history of CRC is associated with increased odds of colonoscopy and screening colonoscopy, with a decreased odds of sigmoidoscopy, irrespective of age. It would stand to reason that family history of CRC would additionally lead to a decrease in FOBT usage, although family history does not appear to influence FOBT compliance one way or the other in these data. The difference in sigmoidoscopy use and screening colonoscopy use between Whites and Blacks may be explained in part by the increased risk of CRC in Black individuals. Black Marylanders aged 50–64 are significantly less likely than White residents to receive timely sigmoido-
scopy screening. This may be because they are receiving FOBT and/or colonoscopy instead (Tables 2a, 2b). The odds ratios for Black race do not reach statistical significance (FOBT: OR = 1.40, P = 0.079; colonoscopy: OR = 1.29, P = 0.128), but are suggestive that FOBT and colonoscopy use may be higher in Blacks as compared to Whites for this younger age group. In fact, Blacks have increased odds of screening colonoscopy use over Whites for those 50–64 years of age (OR = 1.87, P b 0.001). In this age group, other generally underscreened groups, males and Hispanics, also have significantly increased odds of screening colonoscopy. At ages 65+, Black residents are significantly more likely to have had a timely FOBT than White residents. Still in the older age group, the odds of sigmoidoscopy use remain lower for Blacks, while the odds of colonoscopy use remain higher (though neither are statistically significant). These findings suggest that FOBT, as opposed to no screening for this highrisk group, may be preferentially used in this older age group with more pressing health concerns or comorbidities. Overall compliance was not presented here because of lack of agreement on what bcompleteQ for colorectal cancer means. However, here we found that screened persons whose CRC screening may be bcompleteQ according to ACS guidelines are the minority. For those 50- to 64-year-olds, 5.78% have received timely sigmoidoscopy alone and 11.72% have received timely FOBT alone while 41.73%
376
Table 3 Final multiple logistic regression models for colorectal cancer screening in Maryland, Maryland Cancer Survey (2002) 50–64 years FOBT odds ratio (95% CI)
Sigmoidoscopy odds Colonoscopy odds Screening colonoscopy ratio (95% CI) ratio (95% CI) odds ratio (95% CI)
0.93 [0.72–1.20]
1.14 [0.86–1.50]
Referent Referent 1.01 [0.76–1.35] 0.88 [0.63–1.22]
Referent 0.82 [0.64–1.05]
Referent 0.66 [0.51–0.86]T
Referent 1.01 [0.72–1.40]
Referent 1.06 [0.72–1.55]
Referent 0.86 [0.64–1.16]
Referent 0.76 [0.57–1.02]
Referent 1.40 [0.96–2.04] 0.53 [0.23–1.20] 0.78 [0.30–2.00]
Referent 1.29 [0.93–1.80] 0.84 [0.44–1.61] 1.23 [0.57–2.67]
Referent 1.87 [1.34–2.62]T 1.04 [0.52–2.07] 2.26 [1.05–4.88]T
Referent 2.20 [1.38–3.51]T 1.35 [0.54–3.36] 0.47 [0.13–1.67]
Referent 0.85 [0.45–1.61] 2.19 [0.85–5.67] 1.17 [0.30–4.53]
Referent 1.40 [0.89–2.20] 0.60 [0.26–1.40] 1.35 [0.41–4.40]
Referent 1.45 [0.96–2.19] 0.50 [0.20–1.26] 1.13 [0.38–3.42]
Referent
Referent
Referent
0.72 [0.48–1.08]
1.23 [0.91–1.67]
0.95 [0.71–1.28]
1.03 [0.36–3.00]
1.00 [0.43–2.36]
1.25 [0.55–2.86]
Referent 0.57 [0.35–0.93]T 1.43 [0.65–3.13] 0.48 [0.14–1.66]
Referent Referent 1.02 [0.75–1.40] 0.61 [0.42–0.88]T
Referent 1.05 [0.80–1.37]
Referent 1.09 [0.82–1.45]
Referent 1.01 [0.73–1.41]
Referent 1.18 [0.80–1.76]
Referent 0.85 [0.63–1.15]
Referent 0.89 [0.67–1.20]
Referent
Referent
Referent
Referent
Referent
Referent
Referent
1.06 [0.57–1.98] 1.11 [0.50–2.46]
0.81 [0.48–1.37]
0.65 [0.37–1.14]
1.34 [0.82–2.19]
1.03 [0.52–2.05]
1.20 [0.77–1.87]
1.15 [0.75–1.77]
1.28 [0.66–2.48] 0.88 [0.38–2.03] 1.27 [0.64–2.50] 0.92 [0.40–2.12] 1.42 [0.71–2.84] 0.85 [0.36–2.00]
0.92 [0.53–1.60] 0.86 [0.49–1.51] 0.86 [0.48–1.52]
0.81 [0.45–1.45] 0.82 [0.46–1.47] 0.82 [0.45–1.49]
1.67 [0.97–2.85] 1.31 [0.74–2.34] 0.95 [0.51–1.77]
1.22 [0.59–2.53] 1.44 [0.68–3.04] 1.38 [0.63–3.03]
1.31 [0.80–2.16] 1.22 [0.71–2.08] 1.15 [0.65–2.04]
1.17 [0.73–1.88] 0.73 [0.44–1.23] 0.87 [0.50–1.51]
Referent 0.70 [0.42–1.15] 1.35 [0.77–2.37] 0.91 [0.54–1.53] 1.23 [0.70–2.19]
Referent 1.22 [0.78–1.89] 1.07 [0.67–1.73] 1.18 [0.75–1.85] 1.18 [0.72–1.94]
Referent 1.66 [1.01–2.73]TT 1.68 [0.98–2.87] 1.81 [1.09–3.01]T 1.51 [0.87–2.66]
Referent 0.89 [0.58–1.36] 1.08 [0.59–1.98] 1.64 [0.88–3.05] 1.43 [0.92–2.23]
Referent 1.67 [0.95–2.92] 2.64 [1.30–5.37]T 1.79 [0.85–3.76] 1.16 [0.63–2.16]
Referent 0.91 [0.61–1.34] 0.65 [0.37–1.13] 0.82 [0.46–1.45] 1.01 [0.67–1.53]
Referent 1.17 [0.80–1.70] 0.81 [0.47–1.42] 1.12 [0.64–1.95] 1.18 [0.80–1.74]
Referent
Referent 0.73 [0.38–1.38] 0.77 [0.39–1.53] 0.86 [0.46–1.61] 1.14 [0.58–2.23]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
FOBT within 1 yeara Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing
65+ years Sigmoidoscopy odds Colonoscopy odds Screening colonoscopy FOBT odds ratio ratio (95% CI) ratio (95% CI) odds ratio (95% CI) (95% CI)
Health insurance Unable to receive needed medical care Constant coefficientd Goodness of fit test: Hosmer–Lemeshow v 2, P value
Referent
Referent
0.49 [0.27–0.89]T 0.69 [0.28–1.72]
1.13 [0.74–1.71] 3.75 [1.87–7.52]T
0.39 [0.22–0.69]T
36.89 [27.06–50.29]T N/Ac
0.049 [0.019–0.124] 4.91, P = 0.55
0.59 [0.35–1.01] 0.62 [0.36–1.09] 0.74 [0.55–0.99]T 0.97 [0.73–1.30]
0.60 [0.36–0.99]TT
2.55 [1.78–3.68]T 2.71 [1.93–3.81]T
N/Ab
N/Ab
N/Ab
11.09 [6.07–20.25]T
31.76 [21.14–47.73]T
18.25 [11.27–29.57]T
2.83 [1.52–5.27]T
3.10 [1.41–6.83]T
3.28 [1.06–10.17]T
0.014 [0.004–0.043] 1.64, P = 0.95
0.009 [0.002–0.035] 1.57, P = 0.95
0.019 [0.005–0.078] 6.67, P = 0.35
0.074 [0.021–0.254] 9.92, P = 0.13
26.73 [19.25–37.11]T N/Ac
0.60 [0.34–1.04] 3.60 [0.83–15.57]
2.23 [1.48–3.38]T 2.60 [1.83–3.72]T
N/Ab
N/Ab
N/Ab
11.62 [5.04–26.76]T
21.71 [14.87–31.72]T
8.70 [5.82–13.00]T
0.002 [0.0004–0.017] 3.45, P = 0.75
0.133 [0.054–0.328] 4.38, P = 0.63
0.132 [0.055–0.320] 6.28, P = 0.39
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Health status Excellent, very good, or good Fair Poor Health behavior Current smokers Current alcohol drinkers Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Knowledge of screening tests for CRC Health system predictors Ever had recommendation for FOBT Ever had recommendation for sigmoidoscopy or colonoscopy Usual source of care
a
Timely FOBT was only evaluated in the Colonoscopy Model. Recommendation for FOBT was assessed only for FOBT. c Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy. d Note that this is the exponentiated coefficient. T Statistically significant (P b 0.05). TT P = 0.05. b
377
378
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
have received a timely colonoscopy, or a timely sigmoidoscopy with timely FOBT. Comparable percentages for the 65 and older group are 6.21%, 9.77%, 57.52%. The outreach population with no screening tests is larger—40.76% for the younger and 26.50% for the older screening age group. Since colonoscopy is a diagnostic tool as well as a fairly widely implemented screening test in Maryland, we modeled all colonoscopy and screening colonoscopy alone for both age groups. The demographic factors of gender, race, and income were influential among the 50- to 64-yearolds for screening colonoscopy but not for any colonoscopy. In the older adults group, screening colonoscopy use was less among those who report alcohol use. Increasing the use of screening colonoscopy may be amenable to programmatic targeting of specific subpopulations.
exploration of these nuances is warranted. Even without the particulars, knowledge of the differences discovered in this study can instruct program planning to increase CRC screening test use in Maryland, and perhaps in other states as well. Additionally, the literature indicates that age-related vulnerability factors may necessitate special interventions targeted for older persons [27]. This study suggests that health care providers (i.e., general practice, family medicine, internal medicine) should be the target for future interventions to increase clinician recommendation for CRC screening tests and to address the lower use of sigmoidoscopy, even in the 50- to 64-year-olds. There remains a great need for improved CRC screening in Maryland. According to our results, it is clear that the most influential way to improve overall CRC screening for each test and both age groups is to increase clinician recommendation for these tests.
Limitations These data and analyses have some limitations that are common in telephone surveys. First, the MCS only included responses from residents with a land line phone, who are English speakers, and have a residence (the institutionalized population was excluded). In addition, these data are based solely on self-report and therefore may be hindered by recall or information bias. As a result, the estimates provided here may be high since lower-income groups may not have been represented and also because self-report may lead to an overestimation of screening prevalence and an underestimation of screening timeliness [23,24]. One study did demonstrate, however, that the sensitivity and specificity for self-report of CRC screening tests are nearly 90% or above when carefully worded questions are used (as was the case in this survey) [25]. To learn the relationship of a particular variable and outcome while controlling for all other variables, multivariate analysis is an extremely useful tool. Yet in performing an unweighted logistic regression analysis, we assumed that all observations were independent. If this is not true for these survey data, the measures of association will remain the same but the standard errors may be underestimations that could lead to inaccurate conclusions regarding statistical significance.
Conclusions In summary, the most compelling conclusion is that clinician recommendation is the best predictor of CRC screening use. There is a unique profile for each type of screening test (FOBT, sigmoidoscopy, or colonoscopy), and these profiles differ by age group (50–64 years and 65+ years). Research suggests that the perceived benefits and barriers of screening differ by type of screening test, thereby influencing use [26]. Nevertheless, clinician advice is a strong factor in these data and may be driving the results seen here. It is unclear why a particular factor, such as race, predicts use for one test or age group and not another; future
Acknowledgments We would like to thank Ebenezer Israel1 , Eileen Steinberger1, Annette Hopkins1, Min Zhan1, and Carmela Groves2 for designing, collecting, and weighting the MCS data. We would also like to acknowledge Helio Lopez2, Scott Zeger3, and Michael Griswold3 for their statistical assistance.
References [1] Maryland Department of Health and Mental Hygiene. Annual cancer report. Available at: http://www.fha.state.md.us/cancer/html/crf_ann_ can_rpt.html. Accessed November 6, 2003 and December 16, 2004. [2] Vijan S, Hwang EW, Hofe TP, Hayward RA. Which colon cancer screening test: a comparison of costs, effectiveness, and compliance. Am J Med 2001;111:593 – 601. [3] Frazier AL, Colditz GA, Fuchs CS, Kuntz KM. Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 2001;284:1954 – 61. [4] Smith RA, Cokkinides V, Eyre HJ. American cancer society guidelines for the early detection of cancer, 2003. CA Cancer J Clin 2003;53:27 – 43. [5] Steinberger1 E, Israel1 E, Hopkins1 A, Zhan1 M, Uman1 J, Glover1 M, Groves2 C, Bienia2 M, Dwyer2 D. Maryland Cancer Survey Report 2002. 1Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD and 2Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD, 2002 (www.fha.state.md.us/cancer/pdf/ MCS_Report_2002-V3.pdf). [6] Steinberger1 E, Israel1 E, Hopkins1 A, Zhan1 M, Groves2 C. Maryland Cancer Survey 2002. 1Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD and 2Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD, 2002.
1 Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD. 2 Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD. 3 Johns Hopkins Bloomberg School of Public Health.
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379 [7] Thomas W, White CM, Mah J, et al. Longitudinal compliance with annual screening for fecal occult blood. Minnesota colon cancer control study. Am J Epidemiol 1995;142:176 – 82. [8] Lemon S, Zapka J, Puleo E, et al. Colorectal cancer screening participation: comparisons with mammography and prostate-specific antigen screening. Am J Public Health 2001;91:1264 – 72. [9] Weitzman ER, Zapka J, Estabrook B, et al. Risk and reluctance: understanding impediments to colorectal cancer screening. Prev Med 2001;32:502 – 13. [10] Escarce JJ, Epstein KR, Colby DC, Schwartz JS. Racial differences in the elderly’s use of medical procedures and diagnostic tests. Am J Public Health 1993;83:948 – 54. [11] Walsh JME, Terdiman JP. Colorectal cancer screening. JAMA 2003;289:1297 – 302. [12] Kiefe CI, Funkhouser E, Fouad MN, May DS. Chronic disease as a barrier to breast and cervical cancer screening. J Gen Int Med 1998;13:357 – 65. [13] Ko CW, Kreuter W, Baldwin L. Effect of medicare coverage on use of invasive colorectal cancer screening tests. Arch Int Med 2002;162: 2581 – 6. [14] Lewis SF, Jensen NM. Screening sigmoidoscopy. J Gen Int Med 1996;11:542 – 4. [15] Holt Jr WS. Factors affecting compliance with screening sigmoidoscopy. J Fam Pract 1991;32:585 – 9. [16] McCarthy BD, Moskowitz MA. Screening flexible sigmoidoscopy: patient attitudes and compliance. J Gen Int Med 1993;8: 120 – 5. [17] Cooper GS, Fortinsky RH, Hapke R, Landefeld S. Primary care physician recommendations for colorectal cancer screening. Arch Int Med 1997;157:1946 – 50. [18] Korn EL, Graubard BI. Epidemiologic studies utilizing surveys:
[19]
[20]
[21] [22]
[23]
[24]
[25]
[26]
[27]
379
accounting for the sampling design. Am J Public Health 1991; 81:1166 – 73. Taplin SH, Urban N, Taylor VM, Savarino J. Conflicting national recommendations and the use of screening mammography: does the physician’s recommendation matter? J Am Board Fam Pract 1997;10: 88 – 95. Coll PP, O’Connor PJ, Crabtree BF, Besdine RW. Effects of age, education, and physician advice on utilization of screening mammography. J Am Geriatr Soc 1989;37:957 – 62. Zapka JG, Costanza ME, Harris DR, et al. Impact of a breast cancer screening community intervention. Prev Med 1993;22:34 – 53. Bobo JK, Shapiro JA, Schulman J, Wolters CL. On-schedule mammography rescreening in the national breast and cervical cancer early detection program. Cancer Epidemiol Biomarkers Prev 2004;13: 620 – 30. 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;85:566 – 70. Lipkus IM, Rimer BK, Lyna PR, Pradhan AA, Conaway M, Woods-Powell CT. Colorectal screening patterns and perceptions of risk among African-American users of a community health center. J Commun Health 1996;21:409 – 27. Baier M, Calonge N, Cutter G, et al. Validity of self-reported colorectal cancer screening behavior. Cancer Epidemiol Biomarkers Prev 2000;9:229 – 32. James AS, Campbell MK, Hudson MA. Perceived barriers and benefits to colon cancer screening among African Americans in North Carolina: how does perception relate to screening behavior? Cancer Epidemiol Biomarkers Prev 2002;11:529 – 34. Clipp EC, Carver EH, Pollak KI, et al. Age-related vulnerabilities of older adults with colon adenomas. Cancer 2004;100:1085 – 94.
Colorectal cancer screening: Physician recommendation is inf luential advice to Marylanders Amy Gilbert, M.P.H.a, Norma Kanarek, Ph.D.b,T a
The Johns Hopkins Bloomberg School of Public Health, Department of Epidemiology, 615 North Wolfe Street, Baltimore, MD 21205, USA b The Johns Hopkins Bloomberg School of Public Health, Department of Environmental Health Sciences, Room E7038, 615 North Wolfe Street, Baltimore, MD 21205, USA Available online 5 April 2005
Abstract Background. In comparison to the United States, Maryland is facing a significantly higher burden of colorectal cancer incidence and mortality. The primary objective of this study was to determine the predictors of colorectal cancer screening use in Maryland. Methods. We performed secondary analyses on Maryland Cancer Survey 2002 data from 2994 respondents to investigate important predictors for individual colorectal cancer screening tests. CRC screening outcomes were defined as (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, or (3) colonoscopy within the past 10 years. Results. We found that clinician recommendation for a screening test is the best predictor in both age categories (50–64 years and 65+ years); it is a very strong indicator and consistently improves the odds of use by a factor of at least 8 for any screening test. Conclusions. There remains a great need for improved colorectal cancer screening in Maryland. According to our results, it is clear that the most influential way to improve overall colorectal cancer screening for each test and both age groups is to increase clinician recommendation for these tests. D 2005 Elsevier Inc. All rights reserved. Keywords: FOBT; Sigmoidoscopy; Colonoscopy; Colorectal cancer screening; Colon cancer screening
Introduction Colorectal cancer (CRC) is the second most common fatal malignancy in Maryland [1]. According to the Maryland Cigarette Restitution Fund’s 2003 Annual Cancer Report, 2778 new CRC cases were diagnosed and 1158 deaths due to CRC were recorded in Maryland during 2000. In comparison to the United States, Maryland is facing a significantly higher burden of CRC incidence and mortality. Effective CRC screening procedures allow for the detection and removal of precursor lesions and facilitate earlier identification of malignancies at stages that are more
T Corresponding author. Fax: +1 410 955 0617. E-mail addresses: [email protected] (A. Gilbert)8 [email protected] (N. Kanarek). 0091-7435/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2005.01.008
amenable to treatment [2,3]. The burden of CRC can be reduced by increased colorectal cancer screening and adherence to guidelines such as those issued by the American Cancer Society (ACS) [4]. For average risk individuals ages 50 years and older, the ACS lists the following as acceptable options for CRC screening: (1) annual fecal occult blood test, (2) flexible sigmoidoscopy every 5 years, or (3) colonoscopy every 10 years with the caveat that completion of timely fecal occult blood test and sigmoidoscopy is preferred over the individual tests alone. In practice, however, the tests are conducted in different settings and at different times. According to 2001 BRFSS results for Maryland, 44.4% of residents 50 years and older reported FOBT use within the past 2 years and 52.2% reported ever having had a sigmoidoscopy or colonoscopy [1]. Fifty-eight percent of Maryland Cancer Survey 2002 (MCS) responders 50 years
368
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
of age or older reported ever having had a sigmoidoscopy or colonoscopy [5,6]. This indicates that at least 42% of residents age 50 or older have not been screened by a method that visualizes the colon. The primary objective of this study was to determine the predictors of CRC screening test use among Marylanders 50 years or older. Additionally, we sought to assess and quantify the outstanding need for CRC screening by individual CRC test and age.
Background Studies have previously shown an association between many factors and CRC screening. The literature suggests that age and sex influence screening behavior. Thomas et al. found peak compliance for annual FOBT to be around age 70 with lower screening seen among the youngest (55 years or younger) and the oldest (80 years or older) [7]. Lemon et al. found that men aged 65–74 were more likely to be currently CRC screened than men aged 50–64 [8]. Women perceive themselves to be at lower risk for CRC than men, as do their healthcare providers [9]. This may lead to differences in screening adherence between males and females. Race may also play a role in CRC screening. Escarce et al. demonstrated that after adjusting for age and sex, White elderly persons are significantly more likely than Black elderly persons to receive sigmoidoscopy or colonoscopy, a difference not entirely accounted for by access to health care [10]. Studies also propose that there may be differences in CRC screening practices in urban and rural settings [11]. Rural residents may have less access to the facilities, instruments, and trained physicians needed for CRC screening. Health status may also influence the utilization of preventive health practices. Data indicates that breast and cervical cancer screening rates decrease as comorbidity increases [12]. This same effect may exist for CRC screening. Health insurance coverage and access to health care have been shown to be associated with CRC screening participation [8,11]. However, a study in Washington State concluded that use of CRC screening tests did not substantially change once Medicare coverage was available [13]. This and other literature reinforce that medical coverage alone is not enough to effectively increase CRC screening utilization. In studying factors associated with screening sigmoidoscopy among general medicine patients, Lewis and Jensen found that patients were five times more likely (OR = 5.02) to get a sigmoidoscopy if a clinician had advised it [14]. Holt concluded that recommendation and demonstration of concern by a physician may be the primary motivating factors in screening compliance [15]. Similarly, McCarthy and Moskowitz noted evidence that 50–75% of patients offered screening sigmoidoscopy will accept [16]. Unfortunately, primary care physician recommendations for CRC screening have been shown to be inconsistent and inappropriate [17]. Knowledge of CRC and the availability of screening can also be implicated in compliance [9]. Furthermore, family history of
CRC, non-smoking status, and higher education have all been found to be associated with higher levels of screening [8].
Methods Data source The Maryland Cigarette Restitution Fund, created in 2000 with tobacco settlement money by the Maryland State Legislature, established the Cancer Prevention, Education, Screening, and Treatment (CPEST) Program under the Center for Cancer Surveillance and Control at the Department of Health and Mental Hygiene (DHMH). The Surveillance and Evaluation Unit of the CPEST Program commissioned the MCS, a population-based, random digit dial, computer-assisted land line telephone interview utilizing list-assisted disproportionate stratified (urban/rural) sampling. The MCS was conducted by a research team in the Department of Epidemiology and Preventive Medicine at the University of Maryland School of Medicine in Baltimore. The survey asked questions about cancer screening practices, knowledge of cancer and cancer screening, and cancer risk behaviors for selected cancers among English-speaking Marylanders age 40 and older residing in private residences. A total of 5071 interviews were completed from 84,172 phone numbers called. The Council of American Survey Research Organizations response rate was 38.4% (completed interviews/[known eligible + presumed eligible]) [5]. The completion rate (completed interviews/known eligible) was 65.4%.
Design We obtained the MCS data set [6] and performed secondary analyses to investigate CRC screening usage in accordance with ACS screening guidelines for fecal occult blood test (FOBT), sigmoidoscopy, and colonoscopy. The three screening tests were evaluated independently and respondents reporting use of more than one test were counted more than once. Survey questions did not allow a dual response to sigmoidoscopy and colonoscopy use, however. Several distinct factor groups were considered and evaluated as contributors to CRC screening: (1) demographics [age, sex, race, Hispanic ethnicity, marital status], (2) socioeconomic status [urban/rural residence, education, income], (3) health status [self-reported health status], (4) health behaviors [current smoking status, alcohol consumption, body mass index (BMI)], (5) personal cancer predictors [level of concern about cancer, family history of CRC, knowledge of CRC screening], and (6) health system predictors [clinician recommendation for screening test, usual source of care, health insurance, inability to receive needed medical care]. Institutional Review Boards of the
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Maryland DHMH and the Johns Hopkins Bloomberg School of Public Health approved this study.
Data analysis CRC screening test outcomes were defined as (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, or (3) colonoscopy within the past 10 years. CRC screening test outcomes were determined using survey variables that indicated whether FOBT, sigmoidoscopy, and/or colonoscopy had been performed, and if so, the time since each exam was done. Respondent observations where race was missing were not included in the initial data set since this variable was required for assigning a weight to the respondent. Data from 418 participants were dropped from analysis due to ddon’t knowT or drefusedT responses for most covariates. Missing values for income (n = 512) and BMI (n = 113) were retained and analyzed as separate categories. In initial multivariate analyses, there was a significant interaction for age and health insurance. Thus, age groups were considered in separate models (50–64 years and 65+ years). Age was tested as an additional continuous variable, found not to add to either age group model, and was not retained. Due to DHMH data reporting guidelines for small numbers, 24 observations reporting no health insurance in the 65+ age category were dropped, leaving this group with 100% health insurance coverage. As a result, health insurance was not evaluated in the 65+ age group. Survey responses analyzed totaled 1730 in the 50–64 age group and 1264 in the 65+ age group. All CRC screening outcomes, Hispanic ethnicity, current smoking status, current alcohol consumption, knowledge of screening tests for CRC, ever had clinician recommendation for FOBT, ever had clinician recommendation for sigmoidoscopy or colonoscopy, usual source of care, health insurance, and inability to receive needed medical care were dichotomous variables coded byesQ (1) or bnoQ (0). Health insurance was assigned bnoQ (0) if no current health insurance or no health insurance in the past 12 months was reported. Family history of CRC byesQ (1) reflects reported CRC diagnosis of any first-degree relative. Respondents were considered to have a usual source of care if at least one primary health care provider was reported (1). Race was categorized as White, Black, or Other. Indicator variables were used to identify marital status as (a) married or partner of an unmarried couple; (b) divorced, widowed, or separated; and (c) never married. Education categories were defined as (a) less than high school; (b) high school graduate; (c) 1–3 years of college; (d) college graduate; (e) master’s degree, advanced professional or doctoral degree. Categories for annual household income were represented in thousands of dollars by (a) less than 25; (b) 25–49.999; (c) 50–74.999; and (d) 75+. Self-reported health status was grouped as (a) excellent, very good, or
369
good; (b) fair; and (c) poor. The indicator variable for BMI was defined as (a) not overweight or obese; and (b) overweight or obese. Respondents were asked about their concern about getting cancer in the future and the categories were (a) very concerned; (b) somewhat concerned; and (c) not at all concerned. Unweighted multiple logistic regression analysis was performed by age category (50–64 years and 65+ years) for (1) FOBT within the past year, (2) sigmoidoscopy within the past 5 years, (3) colonoscopy within the past 10 years, and (4) screening colonoscopy within the past 10 years. A timely screening colonoscopy included only those that were reported as routine/screening exams or prompted by family history. Since our primary objective was etiological in nature, the most conservative approach was to perform an unweighted regression analysis [18]. Likelihood ratio tests were done to evaluate the statistical significance ( P b 0.05) of the following four variable groups for each outcome: health status, health behaviors, personal cancer predictors, and health system predictors. In accordance with known associations, the demographics and socioeconomic status variable groups (except for marital status) were retained in all models as covariates. Completion of timely FOBT was included as a covariate in the colonoscopy model. Models were then run with only significant variable groups retained. Following that, individual non-significant variables (likelihood ratio tests P N 0.05) within the variable groups were removed to assemble the most parsimonious model for each outcome and age category. Final models (Table 3) estimated levels of CRC screening in each age group, controlling for other important factors. Models were assessed for goodness of fit using the Hosmer–Lemeshow test. Stata version 8.0 was used to analyze these data.
Results Respondent characteristics Weighted respondent characteristics are provided in Table 1 by age. In the 50–64 year group, 85 respondents (5.30%) reported having had a timely FOBT and timely sigmoidoscopy; 240 respondents (13.80%) reported having had a timely FOBT and timely colonoscopy. In the 65+ year group, 75 respondents (6.08%) reported having had a timely FOBT and timely sigmoidoscopy; 275 respondents (23.38%) reporting having had a timely FOBT and timely colonoscopy. Of the timely colonoscopies, 23.29% and 33.03% were reported as screening in the 50–64 and 65+ age groups, respectively. Tables 2a and 2b summarize CRC screening in Maryland by age group and screening test. Usage in Maryland is higher for older adults on every screening test. Colonoscopy use is most prevalent (ages 50– 64, 36.44%; ages 65 and above, 51.44%), FOBT is next (30.83% and 39.23%), and sigmoidoscopy is least (11.09% and 12.29%). Use was consistently and statistically signifi-
370
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 1 Respondent characteristics by age category, Maryland Cancer Survey (2002)
Table 1 (continued) Weighted percent (SE)
Weighted percent (SE) 50–64 yearsa (n = 1730) Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No
65+ yearsb (n = 1264)
49.17 [1.41] 50.83 [1.41]
41.09 [1.61] 58.91 [1.61]
71.93 [1.37] 22.32 [1.24] 5.75 [0.87]
80.15 [1.42] 16.56 [1.33] 3.29 [0.68]
2.50 [0.45] 97.50 [0.45]
1.10 [0.29] 98.90 [0.29]
72.16 [1.18]
59.58 [1.51]
23.06 [1.11]
37.81 [1.48]
4.78 [0.52]
2.61 [0.49]
78.05 [0.72] 21.95 [0.72]
76.73 [0.94] 23.27 [0.94]
9.59 27.83 20.54 21.71 20.33
15.45 31.70 20.44 18.26 14.15
13.26 23.16 16.26 34.59 12.73
[0.86] [1.24] [1.12] [1.19] [1.13]
[1.20] [1.45] [1.26] [1.28] [1.12]
[0.95] [1.15] [1.04] [1.36] [0.93]
26.13 [1.36] 28.35 [1.43] 11.83 [1.11] 11.66 [1.07] 22.03 [1.32]
82.65 [1.07] 13.03 [0.94] 4.32 [0.60]
78.11 [1.32] 16.84 [1.19] 5.05 [0.70]
20.55 [1.17] 79.45 [1.17]
7.40 [0.90] 92.60 [0.90]
46.09 [1.40] 53.91 [1.40]
53.50 [1.60] 46.50 [1.60]
33.86 [1.33] 62.27 [1.36] 3.87 [0.50]
42.76 [1.57] 54.37 [1.59] 2.87 [0.50]
36.50 [1.38] 50.22 [1.41] 13.28 [0.94]
29.60 [1.47] 44.24 [1.58] 26.16 [1.43]
11.79 [0.87] 88.21 [0.87]
14.12 [1.12] 85.88 [1.12]
Knowledge of screening tests for CRC Yes No Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No CRC test use No timely test Timely FOBT onlyd Timely sigmoidoscopy onlye Timely colonoscopy onlyf Timely FOBT and timely sigmoidoscopyg Timely FOBT and timely colonoscopyh Timely sigmoidoscopy and timely colonoscopyi
50–64 yearsa (n = 1730)
65+ yearsb (n = 1264)
95.14 [0.72] 4.86 [0.72]
93.57 [0.76] 6.43 [0.76]
38.86 [1.37] 61.14 [1.37]
44.64 [1.59] 55.36 [1.59]
60.00 [1.39] 40.00 [1.39]
70.55 [1.46] 29.55 [1.46]
93.37 [0.68] 6.63 [0.68]
97.03 [0.53] 2.97 [0.53]
90.89 [0.77] 9.11 [0.77]
100.00c 0c
5.54 [0.64] 94.46 [0.64]
1.79 [0.40] 98.21 [0.40]
40.76 11.72 5.78 22.63 5.30
26.50 9.77 6.21 28.06 6.08
[0.01] [0.01] [0.01] [0.01] [0.01]
13.81 [0.01] 0
[0.01] [0.01] [0.01] [0.01] [0.01]
23.38 [0.01] 0
a
Total Maryland population for 50–64 years of age = 739,400. Total Maryland population for 65+ years of age = 504,000. c Twenty-four observations reporting no health insurance in the 65+ year age group were dropped. d FOBT within 1 year. e Sigmoidoscopy within 5 years. f Colonoscopy within 10 years. g FOBT within 1 year and sigmoidoscopy within 5 years. h FOBT within 1 year and colonoscopy within 10 years. i Sigmoidoscopy within 5 years and colonoscopy within 10 years. b
cantly higher among those who reported receiving a clinician recommendation for screening, having a usual source of care, or having health insurance (ages 50–64 only). These are differences most relevant to our study; as illustrated in Tables 2a and 2b, other statistically significant differences were also seen in these data. Multiple logistic regression Table 3 depicts the final multiple logistic regression models for CRC screening in Maryland. Though we modeled compliance to timely FOBT and timely sigmoidoscopy, which was basically no different from timely FOBT
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
alone, and compliance to any CRC screening test, which was very similar to the model for colonoscopy, we present the models for each test separately. Predictors of use, as assessed by logistic regression analysis, vary by age and screening test. Model results for persons aged 50–64 FOBT. The only significant predictor of FOBT use was clinician recommendation for the test (OR = 36.89; 95% CI: 27.06–50.29; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 4.91, P = 0.55). Sigmoidoscopy. Sigmoidoscopy use was significantly lower among Blacks when compared to Whites (OR = 0.57; 95% CI: 0.35–0.93; P = 0.026). Rural residents are less likely have had a timely sigmoidoscopy than urban residents (OR = 0.61; 95% CI: 0.42–0.88; P = 0.009). Persons in fair health are less likely to have had a timely sigmoidoscopy than those in excellent, very good, or good health (OR = 0.49; 95% CI: 0.27–0.89; P = 0.019). Current smokers are less likely to have had a timely sigmoidoscopy than non-smokers (OR = 0.39; 95% CI: 0.22–0.69; P = 0.001). Family history of CRC was borderline significant and indicated that persons with family history were less likely to have had a timely sigmoidoscopy (OR = 0.60; 95% CI: 0.36–0.99; P = 0.050). Clinician recommendation for sigmoidoscopy increases the odds of use (OR = 11.09; 95% CI: 6.07– 20.25; P b 0.001). The final model fits the data (Hosmer– Lemeshow test statistic = 9.92, P = 0.13). Colonoscopy. Reporting a family history of CRC significantly increases the odds of colonoscopy use (OR = 2.56; 95% CI: 1.78–3.68; P b 0.001). Independently, reporting having a clinician recommendation for the test raised the odds of colonoscopy by more than 30:1 (OR = 31.76; 95% CI: 21.14–47.73; P b 0.001). Usual source of care was also important and positively associated with colonoscopy use (OR = 2.83; 95% CI: 1.52–5.27; P = 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 1.64, P = 0.94). Screening colonoscopy. Women have decreased odds of screening colonoscopy use (OR = 0.66; 95% CI: 0.51–0.86; P = 0.002). Individuals of Black race or Hispanic ethnicity have increased odds of screening colonoscopy use (OR = 1.87; 95% CI: 1.34–2.62; P b 0.001 and OR = 2.26; 95% CI: 1.05–4.88; P = 0.038, respectively). Income also appears to play a role in screening colonoscopy use for 50- to 64-year-olds. Those in annual income brackets of $25,000–49,999 and $75,000+ (compared to b$25,000) have increased odds of screening colonoscopy use (OR = 1.66; 95% CI: 1.01–2.73; P = 0.05 and OR = 1.81; 95% CI: 1.09–3.01; P = 0.02, respectively). Family history of CRC improves the odds of screening colonoscopy (OR = 2.71;
371
95% CI: 1.93–3.81; P b 0.001). Clinician recommendation significantly raised the odds of screening colonoscopy (OR = 18.26; 95% CI: 11.27–29.57; P b 0.001). Usual source of care was also positively associated with screening colonoscopy use (OR = 3.10; 95% CI: 1.41–6.83; P = 0.005). The final model fits the data (Hosmer–Lemeshow test statistic = 1.57, P = 0.95). Model results for persons aged 65+ FOBT. Blacks were more likely to have had a timely FOBT than Whites (OR = 2.20; 95% CI: 1.38–3.51; P = 0.001). Those with poor health status had almost 4 times the odds of FOBT use (OR = 3.75; 95% CI: 1.87–7.52; P b 0.001). Those who reported a clinician recommendation for FOBT had over 26 times the odds of use (OR = 26.73; 95% CI: 19.25–37.11; P b 0.001). Having a usual source of care improves the odds of having a timely FOBT (OR = 3.28; 95% CI: 1.06–10.17; P = 0.039). The model fits the data (Hosmer–Lemeshow test statistic = 6.67, P = 0.35). Sigmoidoscopy. An annual income of $50,000–$74,999 compared to b$25,000 significantly improves the odds of sigmoidoscopy use (OR = 2.64; 95% CI: 1.30–5.37; P = 0.008). Clinician recommendation for the test improves sigmoidoscopy use by more than 11:1 (OR = 11.62; 95% CI: 5.04–26.76; P b 0.001). The model fits the data (Hosmer–Lemeshow test statistic = 3.45, P = 0.75). Colonoscopy. Current smokers have a borderline statistically significant decreased odds of colonoscopy use (OR = 0.59; 95% CI: 0.35–1.01; P = 0.055). Current drinkers also have a decreased odds of colonoscopy use (OR = 0.74; 95% CI: 0.55–0.99; P = 0.045). Reporting having a family history of CRC improves the odds of colonoscopy (OR = 2.23; 95% CI: 1.48–3.38; P b 0.001). Reporting having a clinician recommendation for colonoscopy significantly increases the odds of use by more than 20:1 (OR = 21.71, 95% CI: 14.87–31.72; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 4.38, P = 0.63). Screening colonoscopy. Family history of CRC increases the odds of screening colonoscopy use (OR = 2.60; 95% CI: 1.83–3.72; P b 0.001). Clinician recommendation for colonoscopy significantly improves the odds of screening colonoscopy (OR = 8.70; 95% CI: 5.82–12.99; P b 0.001). The final model fits the data (Hosmer–Lemeshow test statistic = 6.28, P = 0.39).
Discussion The main objective of this analysis was to determine what factors were the primary predictors of CRC screening test use in Maryland. This study was able to take a comprehensive approach to understanding the factors that
372
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 2a Colorectal cancer screening as reported by Maryland Cancer Survey 2002 respondents
Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No
50–64 years FOBTa
50–64 years Sigmoidoscopyb
50–64 years Colonoscopyc
50–64 years Screening Colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
30.69 [26.87–34.80] 30.96 [27.82–34.24]
12.04 [9.57–15.03] 10.17 [8.23–12.49]
38.21 [34.16–42.43] 34.72 [31.51–38.08]
26.09 [22.55–29.97] 20.58 [17.93–23.52]
30.55 [27.80–33.44] 35.67 [29.79–42.01] 15.55 [8.07–27.84]
11.98 [10.10–14.14] 7.73 [4.96–11.86] 12.98 [6.22–25.14]
36.93 [34.05–39.91] 36.45 [30.55–42.79] 30.14 [18.47–45.11]
22.62 [20.18–25.26] 27.28 [21.90–33.41] 16.18 [8.42–28.85]
30.02 [15.40–50.26] 30.85 [28.35–33.47]
9.44 [2.76–27.65] 11.13 [9.51–12.99]
48.74 [31.00–66.80] 36.12 [33.51–38.82]
45.87 [28.14–64.72] 22.71 [20.49–25.10]
30.80 [27.84–33.94]
11.98 [9.98–14.31]
37.25 [34.08–40.52]
24.37 [21.62–27.35]
30.98 [26.16–36.26]
8.87 [6.39–12.20]
34.69 [29.89–39.82]
20.75 [16.91–25.20]
30.42 [21.23–41.50]
8.26 [4.04–16.14]
32.64 [23.30–43.61]
19.23 [11.82–29.72]
30.85 [27.91–33.95] 30.74 [26.67–35.14]
12.02 [10.08–14.28] 7.75 [5.71–10.45]
36.51 [33.43–39.71] 36.17 [31.91–40.67]
23.61 [20.96–26.49] 22.15 [18.56–26.21]
19.15 29.36 31.72 32.97 35.16
[12.81–27.62] [24.91–34.23] [26.34–37.64] [27.62–38.80] [29.59–41.18]
5.86 [2.89–11.53] 11.12 [8.11–15.05] 10.34 [7.19–14.64] 12.21 [8.95–16.44] 13.07 [9.58–17.59]
29.36 31.09 36.95 40.76 41.96
[21.57–38.58] [26.67–35.88] [31.37–42.91] [34.99–46.80] [36.02–48.13]
19.29 18.43 23.75 26.19 28.27
[12.83–27.95] [14.79–22.72] [19.10–29.13] [21.37–31.65] [22.99–34.22]
29.89 24.67 31.06 32.85 37.22
[23.07–37.73] [20.30–37.73] [25.12–37.69] [28.55–37.47] [30.22–44.80]
10.12 7.40 10.22 13.13 14.34
28.56 33.98 37.92 40.55 36.04
[21.84–36.40] [29.06–39.26] [31.51–44.78] [35.94–45.33] [29.23–43.46]
14.37 22.04 26.06 26.56 22.42
[9.74–20.17] [17.80–26.97] [20.35–32.71] [22.61–30.92] [16.80–29.27]
[6.34–15.79] [5.01–10.82] [6.75–15.17] [10.28–16.63] [9.72–20.66]
30.91 [28.21–33.76]
11.74 [9.95–13.81]
36.29 [33.45–39.22]
23.86 [21.41–26.50]
30.78 [24.08–38.39] 29.33 [18.50–43.15]
7.41 [4.21–12.71] 9.64 [3.96–21.63]
38.28 [31.20–45.90] 33.76 [22.12–47.77]
23.56 [17.39–31.10] 11.57 [6.04–21.03]
26.64 [21.26–32.81] 31.91 [29.15–34.80]
5.23 [2.88–9.30] 12.60 [10.73–14.74]
28.59 [23.14–34.74] 38.47 [35.56–41.46]
18.54 [14.24–23.78] 24.52 [21.98–27.25]
33.86 [30.12–37.82] 28.23 [25.02–31.68]
11.61 [9.27–14.44] 10.64 [8.55–13.17]
36.42 [32.63–40.40] 36.45 [32.94–40.10]
22.68 [19.46–26.26] 23.81 [20.79–27.12]
32.03 [27.80–36.57] 30.31 [27.18–33.63] 28.72 [18.69–41.38]
14.12 [11.11–17.78] 9.72 [7.86–11.95] 6.55 [2.52–16.00]
33.55 [29.29–38.10] 37.68 [34.36–41.13] 41.63 [29.57–54.78]
20.93 [17.40–24.97] 24.40 [21.51–27.53] 26.08 [16.34–38.93]
30.70 [26.52–35.22] 29.81 [26.52–33.33] 35.02 [28.07–42.68]
9.77 [7.35–12.86] 12.51 [10.22–15.22] 9.34 [5.62–15.14]
38.34 [33.85–43.04] 35.04 [31.58–38.66] 36.49 [29.45–44.16]
23.59 [19.89–27.74] 22.38 [19.43–25.63] 25.91 [19.49–33.57]
32.78 [26.21–40.09] 30.57 [27.92–33.35]
9.37 [5.80–14.79] 11.32 [9.59–13.31]
57.14 [49.32–64.61] 33.67 [30.95–36.50]
43.65 [36.20–51.39] 20.57 [18.30–23.05]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
373
Table 2a (continued)
Knowledge of screening tests for CRC Yes No Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No Total use, weighted % (95% CI) a b c d e f
50–64 years FOBTa
50–64 years Sigmoidoscopyb
50–64 years Colonoscopyc
50–64 years Screening Colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
31.74 [29.19–34.40] 13.01 [6.41–24.61]
11.59 [9.92–13.51] 1.18 [0.16–8.11]
37.11 [34.46–39.84] 23.28 [13.44–37.22]
23.99 [21.68–26.46] 9.67 [4.49–19.58]
70.72 [66.56–74.55] 5.48 [4.12–7.25]
N/Ae N/Ae
N/Ae N/Ae
N/Ae N/Ae
N/Af N/Af
17.41 [14.9–20.25] 1.60 [0.85–2.99]
57.32 [53.82–60.75] 5.10 [3.34–7.71]
36.52 [33.22–39.96] 3.44 [2.08–5.63]
32.26 [29.65–34.98] 10.69 [6.23–17.75]
11.36 [9.69–13.28] 7.18 [3.02–16.11]
38.26 [35.54–41.06] 10.70 [6.21–17.83]
24.50 [22.14–27.02] 6.26 [2.81–13.36]
32.09 [29.46–34.85] 18.2 [12.43–25.86]
11.65 [9.92–13.62] 5.51 [2.64–11.15]
37.82 [35.07–40.66] 22.59 [16.30–30.42]
24.51 [22.12–27.08] 11.10 [6.75–17.70]
29.80 [20.26–41.50] 30.89 [28.34–33.56] 30.83 [28.30–33.36]
5.44 [2.11–13.31] 11.42 [9.75–13.33] 11.09 [9.38–12.80]
36.25 [25.93–48.02] 36.45 [33.78–39.20] 36.44 [33.81–39.07]
17.04 [10.40–26.65] 23.66 [21.34–26.14] 23.29 [20.96–25.63]
FOBT within 1 year. Sigmoidoscopy within the past 5 years. Colonoscopy within the past 10 years. Screening colonoscopy within the past 10 years. Recommendation for FOBT was assessed only for FOBT compliance. Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy.
influence CRC screening test use. We found that clinician recommendation for a screening test is the best predictor for all forms of screening in both age categories; it is a very strong indicator and consistently improves the odds of use by a factor of at least 8 in all instances. During the uptake phase of mammography, clinician recommendation has been cited for significantly improving screening use as well [19–22]. In this study, the predictors of use differ by age group (50–64 years and 65+ years) and screening test (FOBT, sigmoidoscopy, colonoscopy, and screening colonoscopy). Following clinician recommendation, the strongest predictor of less sigmoidoscopy use in the 50–64 group is current smoking status (decreased use in current smokers, compared to never or former smoker), then fair health status (compared to excellent, very good, or good health), family history of CRC (compared to no first-degree relatives affected), Black race (compared to Whites), and place of residence (compared to urban). For colonoscopy and screening colonoscopy in the 50–64 group, family history improves the odds of use. There were additional predictors for screening colonoscopy in this younger group. Males were more likely than females, Blacks more likely than Whites, and Hispanics more likely than non-Hispanics to have a screening colonoscopy. Higher
income ($25,000–49,999 and N$75,000) and usual source of care also improved the odds of screening colonoscopy among the 50- to 64-year-olds. In the 65+ group, again after physician recommendation, FOBT use is less among those with poor health status (compared to excellent, very good, or good health) and more for those with a usual source of care (compared to no usual source of care), and Black race (compared to Whites). Sigmoidoscopy use in the 65+ group is significantly improved among those with an annual household income of $50,000–$74,999 as compared to those earning less than $25,000. In addition to physician recommendation, colonoscopy and screening colonoscopy use in the 65+ group is improved for those with a family history of CRC. The odds of colonoscopy are decreased for current drinkers aged 65+. Why these variables play a role in CRC screening use can be hypothesized. Less use among current smokers and current drinkers may be explained because this group, in general, may not be as health conscious as non-smokers or non-drinkers. Fair health status may prevent a person from taking part in preventive health practices like CRC screening. However, increased use was noted among those with poor health. We found that persons aged 50–64 in fair health
374
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Table 2b Colorectal cancer screening as reported by Maryland Cancer Survey 2002 respondents
Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Hispanic Non-Hispanic Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing Health status Excellent, very good, or good Fair Poor Health behaviors Current smokers Yes No Current alcohol drinkers Yes No Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Yes No Knowledge of screening tests for CRC Yes No
65+ years FOBTa
65+ years Sigmoidoscopyb
65+ years Colonoscopyc
65+ years Screening colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
38.93 [33.85–44.27] 39.44 [35.75–43.25]
14.18 [10.91–18.24] 10.97 [8.84–13.54]
52.83 [47.49–58.10] 50.47 [46.66–54.27]
35.97 [31.06–41.18] 30.99 [27.60–34.60]
37.27 [34.07–40.58] 48.87 [40.15–57.66] 38.55 [20.59–60.28]
12.81 [10.72–15.23] 9.13 [5.17–15.62] 15.67 [6.21–34.29]
51.67 [48.32–55.01] 53.11 [44.26–61.77] 37.31 [20.6–57.73]
33.70 [30.58–36.96] 32.45 [24.99–40.93] 19.83 [8.07–41.06]
29.23 [10.90–58.24] 39.34 [36.28–42.49]
10.8 [2.81–33.61] 12.31 [10.39–14.52]
53.97 [28.43–77.59] 51.41 [48.25–54.55]
30.85 [12.59–58.00] 33.06 [30.17–36.07]
41.8 [37.56–46.16]
14.68 [11.90–17.99]
52.02 [47.65–56.35]
34.97 [30.97–39.20]
35.12 [30.93–39.56] 40.04 [23.47–59.25]
8.52 [6.45–11.16] 12.43 [4.99–27.72]
50.86 [46.42–55.29] 46.54 [28.80–65.20]
30.10 [26.18–34.34] 31.19 [16.19–51.55]
40.17 [36.52–43.93] 36.13 [31.23–41.34]
12.41 [10.15–15.09] 11.89 [8.95–15.62]
52.46 [48.68–56.21] 48.06 [42.89–53.28]
33.53 [30.11–37.14] 31.39 [26.70–36.49]
40.03 34.93 44.55 40.84 38.24
[32.24–48.36] [29.94–40.26] [37.93–51.37] [33.42–48.70] [30.45–46.68]
6.68 9.25 13.15 15.93 19.3
[3.77–11.54] [6.66–12.71] [9.22–18.43] [11.00–22.51] [13.49–26.84]
46.84 51.51 55.21 54.02 47.49
[38.63–55.23] [46.16–56.83] [48.44–61.80] [46.29–61.56] [39.26–55.86]
29.99 34.74 38.53 29.66 28.94
38.43 38.69 38.66 41.31 40.08
[32.73–44.46] [33.15–44.55] [29.60–48.57] [32.20–51.06] [33.81–46.69]
6.06 12.99 22.03 18.75 10.15
[3.95–9.18] [9.57–17.38] [14.96–31.20] [12.29–27.53] [6.94–14.59]
46.77 55.09 47.04 54.41 53.06
[41.00–52.62] [49.25–60.79] [37.37–56.94] [44.74–63.76] [46.38–59.63]
28.13 [23.20–33.65] 37.11 [31.64–42.93] 30.53 [22.07–40.54] 33.29 [24.93–42.85] 34.82 [28.87–41.28]
[22.89–38.22] [29.81–40.02] [32.15–45.34] [23.17–37.09] [22.10–36.89]
38.66 [35.24–42.19] 38.89 [31.72–46.57] 49.25 [35.43–63.19]
13.4 [11.17–15.99] 8.26 [4.98–13.38] 8.6 [3.55–19.39]
50.18 [46.64–53.72] 54.51 [46.89–61.93] 60.59 [46.55–73.07]
33.36 [30.12–36.77] 32.73 [25.99–40.28] 28.94 [17.72–43.51]
26.57 [17.69–37.86] 40.24 [37.08–43.49]
6.71 [3.00–14.30] 12.74 [10.73–15.06]
32.29 [22.40–44.07] 52.97 [49.73–56.18]
20.10 [12.40–30.90] 34.07 [31.07–37.19]
40.25 [36.16–44.47] 38.06 [33.60–42.73]
9.66 [7.44–12.45] 15.32 [12.30–18.94]
51.74 [47.56–55.89] 51.09 [46.36–55.79]
32.26 [28.51–36.26] 33.92 [29.61–38.51]
37.53 [33.11–42.16] 39.85 [35.64–44.21] 52.91 [35.77–69.40]
11.33 [8.79–14.48] 12.59 [9.96–15.81] 20.93 [9.84–39.1]
51.7 [47.00–56.38] 51.79 [47.46–56.09] 40.75 [25.23–58.37]
33.95 [29.57–38.62] 32.96 [29.12–37.05] 20.71 [10.31–37.24]
36.69 [31.37–42.36] 42.28 [37.79–46.90] 36.94 [30.87–43.46]
11.18 [8.03–15.35] 15.48 [12.46–19.06] 8.16 [5.24–12.51]
53.22 [47.36–58.99] 54.23 [49.65–58.75] 44.69 [38.50–51.04]
34.83 [29.62–40.43] 34.25 [30.01–38.77] 28.94 [23.51–35.04]
45.15 [36.96–53.61] 38.26 [35.01–41.62]
10.02 [5.84–16.66] 12.66 [10.61–15.05]
68.49 [59.95–75.94] 48.63 [45.27–52.01]
49.78 [41.49–58.09] 30.28 [27.26–33.48]
40.11 [36.95–43.36] 26.39 [17.34–37.99]
12.96 [10.95–15.28] 2.53 [0.62–9.77]
52.75 [49.50–55.98] 32.27 [22.03–44.56]
33.76 [30.78–36.87] 22.53 [13.59–34.98]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
375
Table 2b (continued)
Health system predictors Ever had recommendation for FOBT Yes No Ever had recommendation for sigmoidoscopy or colonoscopy Yes No Usual source of care Yes No Health insurance Yes No Unable to receive needed medical care Yes No Total use, weighted % (95% CI) a b c d e f g
65+ years FOBTa
65+ years Sigmoidoscopyb
65+ years Colonoscopyc
65+ years Screening colonoscopyd
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
Weighted % (95% CI)
73.34 [68.98–77.29] 11.73 [9.25–14.76]
N/Ae N/Ae
N/Ae N/Ae
N/Ae N/Ae
N/Af N/Af
16.76 [14.18–19.71] 1.59 [0.63–3.93]
68.74 [65.20–72.08] 9.97 [7.15–13.75]
43.74 [40.09–47.46] 7.39 [5.07–10.65]
39.95 [36.86–43.13] 15.67 [6.50–33.17]
12.51 [10.57–14.74] 5.20 [0.72–29.16]
51.88 [48.69–55.05] 37.02 [21.90–55.20]
33.31 [30.40–36.36] 23.90 [11.67–42.75]
N/Ag N/Ag
N/Ag N/Ag
N/Ag N/Ag
N/Ag N/Ag
31.66 [15.18–54.52] 39.37 [36.30–42.52] 39.23 [36.15–42.31]
7.04 [1.63–25.69] 12.39 [10.46–14.61] 12.29 [10.25–14.33]
52.41 [31.15–72.84] 51.42 [48.25–54.57] 51.44 [48.30–54.58]
13.63 [3.89–38.07] 33.39 [30.49–36.42] 33.03 [30.07–35.99]
FOBT within 1 year. Sigmoidoscopy within the past 5 years. Colonoscopy within the past 10 years. Screening colonoscopy within the past 10 years. Recommendation for FOBT was assessed only for FOBT compliance. Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy. Health insurance coverage was 100% in the 65+ year age group after dropping 24 observations reporting no insurance.
are less likely to have had a timely sigmoidoscopy than those in excellent, very good, or good health. In contrast, for ages 65+, poor health status was associated with increased odds of FOBT use. These findings suggest that self-reported health status is an important indicator of CRC screening use and may influence use of other preventive health services. Another explanation is that those with usual source of care, higher incomes, or urban residences are likely to have improved access to health care resources and thus are more likely to participate in CRC screening. After controlling for these and other factors, health insurance in the 50–64 age group, however, did not affect CRC screening use. A family history of CRC would indicate high-risk status for which colonoscopy would be the most appropriate form of screening. This explains why family history of CRC is associated with increased odds of colonoscopy and screening colonoscopy, with a decreased odds of sigmoidoscopy, irrespective of age. It would stand to reason that family history of CRC would additionally lead to a decrease in FOBT usage, although family history does not appear to influence FOBT compliance one way or the other in these data. The difference in sigmoidoscopy use and screening colonoscopy use between Whites and Blacks may be explained in part by the increased risk of CRC in Black individuals. Black Marylanders aged 50–64 are significantly less likely than White residents to receive timely sigmoido-
scopy screening. This may be because they are receiving FOBT and/or colonoscopy instead (Tables 2a, 2b). The odds ratios for Black race do not reach statistical significance (FOBT: OR = 1.40, P = 0.079; colonoscopy: OR = 1.29, P = 0.128), but are suggestive that FOBT and colonoscopy use may be higher in Blacks as compared to Whites for this younger age group. In fact, Blacks have increased odds of screening colonoscopy use over Whites for those 50–64 years of age (OR = 1.87, P b 0.001). In this age group, other generally underscreened groups, males and Hispanics, also have significantly increased odds of screening colonoscopy. At ages 65+, Black residents are significantly more likely to have had a timely FOBT than White residents. Still in the older age group, the odds of sigmoidoscopy use remain lower for Blacks, while the odds of colonoscopy use remain higher (though neither are statistically significant). These findings suggest that FOBT, as opposed to no screening for this highrisk group, may be preferentially used in this older age group with more pressing health concerns or comorbidities. Overall compliance was not presented here because of lack of agreement on what bcompleteQ for colorectal cancer means. However, here we found that screened persons whose CRC screening may be bcompleteQ according to ACS guidelines are the minority. For those 50- to 64-year-olds, 5.78% have received timely sigmoidoscopy alone and 11.72% have received timely FOBT alone while 41.73%
376
Table 3 Final multiple logistic regression models for colorectal cancer screening in Maryland, Maryland Cancer Survey (2002) 50–64 years FOBT odds ratio (95% CI)
Sigmoidoscopy odds Colonoscopy odds Screening colonoscopy ratio (95% CI) ratio (95% CI) odds ratio (95% CI)
0.93 [0.72–1.20]
1.14 [0.86–1.50]
Referent Referent 1.01 [0.76–1.35] 0.88 [0.63–1.22]
Referent 0.82 [0.64–1.05]
Referent 0.66 [0.51–0.86]T
Referent 1.01 [0.72–1.40]
Referent 1.06 [0.72–1.55]
Referent 0.86 [0.64–1.16]
Referent 0.76 [0.57–1.02]
Referent 1.40 [0.96–2.04] 0.53 [0.23–1.20] 0.78 [0.30–2.00]
Referent 1.29 [0.93–1.80] 0.84 [0.44–1.61] 1.23 [0.57–2.67]
Referent 1.87 [1.34–2.62]T 1.04 [0.52–2.07] 2.26 [1.05–4.88]T
Referent 2.20 [1.38–3.51]T 1.35 [0.54–3.36] 0.47 [0.13–1.67]
Referent 0.85 [0.45–1.61] 2.19 [0.85–5.67] 1.17 [0.30–4.53]
Referent 1.40 [0.89–2.20] 0.60 [0.26–1.40] 1.35 [0.41–4.40]
Referent 1.45 [0.96–2.19] 0.50 [0.20–1.26] 1.13 [0.38–3.42]
Referent
Referent
Referent
0.72 [0.48–1.08]
1.23 [0.91–1.67]
0.95 [0.71–1.28]
1.03 [0.36–3.00]
1.00 [0.43–2.36]
1.25 [0.55–2.86]
Referent 0.57 [0.35–0.93]T 1.43 [0.65–3.13] 0.48 [0.14–1.66]
Referent Referent 1.02 [0.75–1.40] 0.61 [0.42–0.88]T
Referent 1.05 [0.80–1.37]
Referent 1.09 [0.82–1.45]
Referent 1.01 [0.73–1.41]
Referent 1.18 [0.80–1.76]
Referent 0.85 [0.63–1.15]
Referent 0.89 [0.67–1.20]
Referent
Referent
Referent
Referent
Referent
Referent
Referent
1.06 [0.57–1.98] 1.11 [0.50–2.46]
0.81 [0.48–1.37]
0.65 [0.37–1.14]
1.34 [0.82–2.19]
1.03 [0.52–2.05]
1.20 [0.77–1.87]
1.15 [0.75–1.77]
1.28 [0.66–2.48] 0.88 [0.38–2.03] 1.27 [0.64–2.50] 0.92 [0.40–2.12] 1.42 [0.71–2.84] 0.85 [0.36–2.00]
0.92 [0.53–1.60] 0.86 [0.49–1.51] 0.86 [0.48–1.52]
0.81 [0.45–1.45] 0.82 [0.46–1.47] 0.82 [0.45–1.49]
1.67 [0.97–2.85] 1.31 [0.74–2.34] 0.95 [0.51–1.77]
1.22 [0.59–2.53] 1.44 [0.68–3.04] 1.38 [0.63–3.03]
1.31 [0.80–2.16] 1.22 [0.71–2.08] 1.15 [0.65–2.04]
1.17 [0.73–1.88] 0.73 [0.44–1.23] 0.87 [0.50–1.51]
Referent 0.70 [0.42–1.15] 1.35 [0.77–2.37] 0.91 [0.54–1.53] 1.23 [0.70–2.19]
Referent 1.22 [0.78–1.89] 1.07 [0.67–1.73] 1.18 [0.75–1.85] 1.18 [0.72–1.94]
Referent 1.66 [1.01–2.73]TT 1.68 [0.98–2.87] 1.81 [1.09–3.01]T 1.51 [0.87–2.66]
Referent 0.89 [0.58–1.36] 1.08 [0.59–1.98] 1.64 [0.88–3.05] 1.43 [0.92–2.23]
Referent 1.67 [0.95–2.92] 2.64 [1.30–5.37]T 1.79 [0.85–3.76] 1.16 [0.63–2.16]
Referent 0.91 [0.61–1.34] 0.65 [0.37–1.13] 0.82 [0.46–1.45] 1.01 [0.67–1.53]
Referent 1.17 [0.80–1.70] 0.81 [0.47–1.42] 1.12 [0.64–1.95] 1.18 [0.80–1.74]
Referent
Referent 0.73 [0.38–1.38] 0.77 [0.39–1.53] 0.86 [0.46–1.61] 1.14 [0.58–2.23]
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
FOBT within 1 yeara Demographic characteristics Sex Male Female Race White Black Other Hispanic ethnicity Marital status Married or partner of an unmarried couple Divorced, widowed, or separated Never married Socioeconomic status Place of residence Urban Rural Education Less than high school High school graduate/GED 1–3 years of college College graduate Master’s, advanced professional, or doctoral degree Annual income, in thousands of dollars b25 25–49.999 50–74.999 75+ Missing
65+ years Sigmoidoscopy odds Colonoscopy odds Screening colonoscopy FOBT odds ratio ratio (95% CI) ratio (95% CI) odds ratio (95% CI) (95% CI)
Health insurance Unable to receive needed medical care Constant coefficientd Goodness of fit test: Hosmer–Lemeshow v 2, P value
Referent
Referent
0.49 [0.27–0.89]T 0.69 [0.28–1.72]
1.13 [0.74–1.71] 3.75 [1.87–7.52]T
0.39 [0.22–0.69]T
36.89 [27.06–50.29]T N/Ac
0.049 [0.019–0.124] 4.91, P = 0.55
0.59 [0.35–1.01] 0.62 [0.36–1.09] 0.74 [0.55–0.99]T 0.97 [0.73–1.30]
0.60 [0.36–0.99]TT
2.55 [1.78–3.68]T 2.71 [1.93–3.81]T
N/Ab
N/Ab
N/Ab
11.09 [6.07–20.25]T
31.76 [21.14–47.73]T
18.25 [11.27–29.57]T
2.83 [1.52–5.27]T
3.10 [1.41–6.83]T
3.28 [1.06–10.17]T
0.014 [0.004–0.043] 1.64, P = 0.95
0.009 [0.002–0.035] 1.57, P = 0.95
0.019 [0.005–0.078] 6.67, P = 0.35
0.074 [0.021–0.254] 9.92, P = 0.13
26.73 [19.25–37.11]T N/Ac
0.60 [0.34–1.04] 3.60 [0.83–15.57]
2.23 [1.48–3.38]T 2.60 [1.83–3.72]T
N/Ab
N/Ab
N/Ab
11.62 [5.04–26.76]T
21.71 [14.87–31.72]T
8.70 [5.82–13.00]T
0.002 [0.0004–0.017] 3.45, P = 0.75
0.133 [0.054–0.328] 4.38, P = 0.63
0.132 [0.055–0.320] 6.28, P = 0.39
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
Health status Excellent, very good, or good Fair Poor Health behavior Current smokers Current alcohol drinkers Body mass index Not overweight or obese Overweight or obese Missing Personal cancer predictors Level of concern about cancer Very concerned Somewhat concerned Not at all concerned Family history of CRC Knowledge of screening tests for CRC Health system predictors Ever had recommendation for FOBT Ever had recommendation for sigmoidoscopy or colonoscopy Usual source of care
a
Timely FOBT was only evaluated in the Colonoscopy Model. Recommendation for FOBT was assessed only for FOBT. c Recommendation for sigmoidoscopy or colonoscopy was assessed only for sigmoidoscopy and colonoscopy. d Note that this is the exponentiated coefficient. T Statistically significant (P b 0.05). TT P = 0.05. b
377
378
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379
have received a timely colonoscopy, or a timely sigmoidoscopy with timely FOBT. Comparable percentages for the 65 and older group are 6.21%, 9.77%, 57.52%. The outreach population with no screening tests is larger—40.76% for the younger and 26.50% for the older screening age group. Since colonoscopy is a diagnostic tool as well as a fairly widely implemented screening test in Maryland, we modeled all colonoscopy and screening colonoscopy alone for both age groups. The demographic factors of gender, race, and income were influential among the 50- to 64-yearolds for screening colonoscopy but not for any colonoscopy. In the older adults group, screening colonoscopy use was less among those who report alcohol use. Increasing the use of screening colonoscopy may be amenable to programmatic targeting of specific subpopulations.
exploration of these nuances is warranted. Even without the particulars, knowledge of the differences discovered in this study can instruct program planning to increase CRC screening test use in Maryland, and perhaps in other states as well. Additionally, the literature indicates that age-related vulnerability factors may necessitate special interventions targeted for older persons [27]. This study suggests that health care providers (i.e., general practice, family medicine, internal medicine) should be the target for future interventions to increase clinician recommendation for CRC screening tests and to address the lower use of sigmoidoscopy, even in the 50- to 64-year-olds. There remains a great need for improved CRC screening in Maryland. According to our results, it is clear that the most influential way to improve overall CRC screening for each test and both age groups is to increase clinician recommendation for these tests.
Limitations These data and analyses have some limitations that are common in telephone surveys. First, the MCS only included responses from residents with a land line phone, who are English speakers, and have a residence (the institutionalized population was excluded). In addition, these data are based solely on self-report and therefore may be hindered by recall or information bias. As a result, the estimates provided here may be high since lower-income groups may not have been represented and also because self-report may lead to an overestimation of screening prevalence and an underestimation of screening timeliness [23,24]. One study did demonstrate, however, that the sensitivity and specificity for self-report of CRC screening tests are nearly 90% or above when carefully worded questions are used (as was the case in this survey) [25]. To learn the relationship of a particular variable and outcome while controlling for all other variables, multivariate analysis is an extremely useful tool. Yet in performing an unweighted logistic regression analysis, we assumed that all observations were independent. If this is not true for these survey data, the measures of association will remain the same but the standard errors may be underestimations that could lead to inaccurate conclusions regarding statistical significance.
Conclusions In summary, the most compelling conclusion is that clinician recommendation is the best predictor of CRC screening use. There is a unique profile for each type of screening test (FOBT, sigmoidoscopy, or colonoscopy), and these profiles differ by age group (50–64 years and 65+ years). Research suggests that the perceived benefits and barriers of screening differ by type of screening test, thereby influencing use [26]. Nevertheless, clinician advice is a strong factor in these data and may be driving the results seen here. It is unclear why a particular factor, such as race, predicts use for one test or age group and not another; future
Acknowledgments We would like to thank Ebenezer Israel1 , Eileen Steinberger1, Annette Hopkins1, Min Zhan1, and Carmela Groves2 for designing, collecting, and weighting the MCS data. We would also like to acknowledge Helio Lopez2, Scott Zeger3, and Michael Griswold3 for their statistical assistance.
References [1] Maryland Department of Health and Mental Hygiene. Annual cancer report. Available at: http://www.fha.state.md.us/cancer/html/crf_ann_ can_rpt.html. Accessed November 6, 2003 and December 16, 2004. [2] Vijan S, Hwang EW, Hofe TP, Hayward RA. Which colon cancer screening test: a comparison of costs, effectiveness, and compliance. Am J Med 2001;111:593 – 601. [3] Frazier AL, Colditz GA, Fuchs CS, Kuntz KM. Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 2001;284:1954 – 61. [4] Smith RA, Cokkinides V, Eyre HJ. American cancer society guidelines for the early detection of cancer, 2003. CA Cancer J Clin 2003;53:27 – 43. [5] Steinberger1 E, Israel1 E, Hopkins1 A, Zhan1 M, Uman1 J, Glover1 M, Groves2 C, Bienia2 M, Dwyer2 D. Maryland Cancer Survey Report 2002. 1Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD and 2Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD, 2002 (www.fha.state.md.us/cancer/pdf/ MCS_Report_2002-V3.pdf). [6] Steinberger1 E, Israel1 E, Hopkins1 A, Zhan1 M, Groves2 C. Maryland Cancer Survey 2002. 1Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD and 2Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD, 2002.
1 Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, MD. 2 Center for Cancer Surveillance and Control, Maryland Department of Health and Mental Hygiene, Baltimore, MD. 3 Johns Hopkins Bloomberg School of Public Health.
A. Gilbert, N. Kanarek / Preventive Medicine 41 (2005) 367–379 [7] Thomas W, White CM, Mah J, et al. Longitudinal compliance with annual screening for fecal occult blood. Minnesota colon cancer control study. Am J Epidemiol 1995;142:176 – 82. [8] Lemon S, Zapka J, Puleo E, et al. Colorectal cancer screening participation: comparisons with mammography and prostate-specific antigen screening. Am J Public Health 2001;91:1264 – 72. [9] Weitzman ER, Zapka J, Estabrook B, et al. Risk and reluctance: understanding impediments to colorectal cancer screening. Prev Med 2001;32:502 – 13. [10] Escarce JJ, Epstein KR, Colby DC, Schwartz JS. Racial differences in the elderly’s use of medical procedures and diagnostic tests. Am J Public Health 1993;83:948 – 54. [11] Walsh JME, Terdiman JP. Colorectal cancer screening. JAMA 2003;289:1297 – 302. [12] Kiefe CI, Funkhouser E, Fouad MN, May DS. Chronic disease as a barrier to breast and cervical cancer screening. J Gen Int Med 1998;13:357 – 65. [13] Ko CW, Kreuter W, Baldwin L. Effect of medicare coverage on use of invasive colorectal cancer screening tests. Arch Int Med 2002;162: 2581 – 6. [14] Lewis SF, Jensen NM. Screening sigmoidoscopy. J Gen Int Med 1996;11:542 – 4. [15] Holt Jr WS. Factors affecting compliance with screening sigmoidoscopy. J Fam Pract 1991;32:585 – 9. [16] McCarthy BD, Moskowitz MA. Screening flexible sigmoidoscopy: patient attitudes and compliance. J Gen Int Med 1993;8: 120 – 5. [17] Cooper GS, Fortinsky RH, Hapke R, Landefeld S. Primary care physician recommendations for colorectal cancer screening. Arch Int Med 1997;157:1946 – 50. [18] Korn EL, Graubard BI. Epidemiologic studies utilizing surveys:
[19]
[20]
[21] [22]
[23]
[24]
[25]
[26]
[27]
379
accounting for the sampling design. Am J Public Health 1991; 81:1166 – 73. Taplin SH, Urban N, Taylor VM, Savarino J. Conflicting national recommendations and the use of screening mammography: does the physician’s recommendation matter? J Am Board Fam Pract 1997;10: 88 – 95. Coll PP, O’Connor PJ, Crabtree BF, Besdine RW. Effects of age, education, and physician advice on utilization of screening mammography. J Am Geriatr Soc 1989;37:957 – 62. Zapka JG, Costanza ME, Harris DR, et al. Impact of a breast cancer screening community intervention. Prev Med 1993;22:34 – 53. Bobo JK, Shapiro JA, Schulman J, Wolters CL. On-schedule mammography rescreening in the national breast and cervical cancer early detection program. Cancer Epidemiol Biomarkers Prev 2004;13: 620 – 30. 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;85:566 – 70. Lipkus IM, Rimer BK, Lyna PR, Pradhan AA, Conaway M, Woods-Powell CT. Colorectal screening patterns and perceptions of risk among African-American users of a community health center. J Commun Health 1996;21:409 – 27. Baier M, Calonge N, Cutter G, et al. Validity of self-reported colorectal cancer screening behavior. Cancer Epidemiol Biomarkers Prev 2000;9:229 – 32. James AS, Campbell MK, Hudson MA. Perceived barriers and benefits to colon cancer screening among African Americans in North Carolina: how does perception relate to screening behavior? Cancer Epidemiol Biomarkers Prev 2002;11:529 – 34. Clipp EC, Carver EH, Pollak KI, et al. Age-related vulnerabilities of older adults with colon adenomas. Cancer 2004;100:1085 – 94.