Managed Care Patients' Preferences, Physician Recommendations, and Colon Cancer Screening

July 21, 2014
Sarah Hawley, PhD, MPH

,
Sarah Lillie, PhD

,
Greg Cooper, MD

,
Jennifer Elston Lafata, PhD

Volume 20, Issue 7

CRC screening rates in a large managed care organization were low. Among those screened, use was associated more with physicians' recommendations than with patient preferences.

Objective

To evaluate associations between patients’ preferences for attributes of different colorectal (CRC) screening modalities, physician CRC screening recommendations during periodic health exams, and subsequent utilization of screening 12 months later in a large health maintenance organization (HMO).

Study Design

Multi-method study including baseline surveys from average-risk HMO members joined with audio recordings of 415 periodic health exams (PHEs) and electronic medical record (EMR) data.

Methods

Patient ratings of test attributes were used to create an algorithm reflecting type and strength of CRC screening modality preference at baseline. Physician recommendations were obtained from audio recordings. Attribute-based test preferences and physician recommendations were compared with CRC test use using chisquare tests. Associations between attribute-based preferences and physician recommendations were assessed using logistic regression.

Results

Based on attribute rankings, most participants had a weak preference for colonoscopy (COL) (41%), an unclear preference (22.4%), or a weak preference for fecal occult blood testing (FOBT) (18.6%). About half (56%) of patients were screened at 12 months and there was no statistical association between attribute preferences and type of test received. Patients were significantly more likely to receive a recommendation including a test other than COL when they had an attribute-based test preference for FOBT (odds ratio [OR]: 2.17; 95% CI, 1.26-3.71; P <.01).

Conclusions

CRC screening test use in this large HMO was generally low. It was not associated with patients’ preferences for different attributes of CRC screening tests but was associated with physician recommendations. Physicians may have better success in getting patients to screen if they consider preferences for test attributes.

Am J Manag Care. 2014;20(7):555-561

  • Colorectal cancer (CRC) screening test use in a large health maintenance organization (HMO) was relatively low (56%).

  • Patients had CRC screening preferences that could be linked to attributes of different tests.

  • HMO members more often received the screening test that was recommended by their physician, usually colonoscopy, than the test that was consistent with their preferences.

  • Physicians appeared willing to consider preferences for tests other than colonoscopy, so adherence with CRC may be increased if patients' preferences can be incorporated into recommendations.

Studies have found variation in the proportion of adults who are adherent with colorectal cancer (CRC) screening, with reports ranging from 45% to 70%,1-3 yet there remains consensus that there is room for improvement, particularly in managed care settings where there is generally coverage for all types of CRC screening. There are several options for CRC screening,4-6 but increasingly, only colonoscopy (COL) and fecal occult blood testing (FOBT) are used in clinical practice.7,8 Low rates of CRC screening uptake, combined with the existence of more than 1 appropriate test, have led some to suggest that offering patients the test they prefer may be an effective method for increasing CRC screening adherence.9,10

This argument is supported by several studies that have documented variation in preferences for CRC screening tests across populations.9,11-19 Overall, these studies have shown that patient preferences can be linked to specific attributes of the screening tests, such as the preparation involved.13-18 Yet only a few studies have evaluated associations among patient preferences for key attributes of CRC screening tests, test ordering by physician, and test utilization.11,14,19

Missing from these studies is an examination of whether attribute-based test preferences are incorporated into physician CRC screening recommendations. The importance of having a physician recommendation in screening uptake has been well documented.18-20 Yet increasingly, physicians are recommending COL and often do not offer patients a choice of screening options.21 The recent National Institutes of Health State-of-the-Science Conference Statement on CRC screening has called for a continued need to understand the role of patient preferences in CRC screening adherence.22

We used data from an observational study conducted in a managed care setting to better understand associations among the preferences of average-risk adults for key attributes of CRC screening tests, physician test recommendations, and subsequent use of CRC screening. We had 3 objectives: 1) to describe the distribution of preferences for key attributes of CRC screening tests and to link these attributes to existing CRC screening modalities; 2) to evaluate the association between physician recommendation for CRC screening modalities and attribute-based CRC screening test preferences; and 3) to compare CRC screening utilization 12 months post visit with attribute-based CRC screening test preferences and physician recommendation.

METHODS

The data used in this analysis came from a study of patient-provider discussions about CRC screening conducted in a large health maintenance organization (HMO) located in southeast Michigan (R01CA112379). Additional details about the study are provided elsewhere.1,13,23,24

Participant Eligibility Criteria and Recruitment

Participating physicians (N = 64) were salaried family and general internal medicine physicians affiliated with a managed care group in southeast Michigan. Participating physicians agreed to allow scheduled periodic health exam (PHE) visits of their eligible patients to be audiorecorded, with patient consent.

Patients (N = 500) were insured, aged 50 to 80 years, and due for CRC screening (eg, never screened with any of the recommended tests or overdue for screening based on US Preventive Services Task Force [USPSTF] guidelines) at the time of a PHE with a study-participating physician from February 2007 to June 2009. All patient participants in the study were able to receive either FOBT or COL for screening purposes without a co-payment as part of their managed care coverage. Patients identi- fied via electronic medical record (EMR) and appointment scheduling data were sent an introduction letter and called approximately 2 weeks prior to their PHE to confirm eligibility, and to complete a pre-visit telephone survey. Participants were asked to arrive at their scheduled PHE early to enable completion of informed consent prior to audio-recording of the PHE. Participants received a $25 gift card. All aspects of the study were approved by the Institutional Review Boards of the Henry Ford Medical Group, Case Western Reserve University, and Virginia Commonwealth University.

Data Sources & Measures

There were 3 primary outcomes for this analysis: 1) attribute-based CRC screening test preferences prior to the PHE, generated from patient survey responses to a list of key CRC screening attributes; 2) primary care physician recommendation for CRC screening test(s) that occurred during the PHE obtained via office visit audio recording; and 3) CRC test utilization 12 months following the PHE obtained via EMR data. Each is described below.

Table 1

Attribute-based CRC screening test preferences: In the pre-visit telephone survey, patients ranked 7 key attributes of screening tests. These attributes were obtained from prior literature that had assessed CRC screening preferences via a list of attributes.9,13-18 Based on this literature, the attributes selected for the survey were test accuracy, preparation required, complications/side effects, need for sedation, frequency of the test, degree of pain/discomfort, and whether a stool sample was necessary. Participants ranked each attribute first, second, and third most important when deciding which screening test to use. We developed an algorithm () from patient rankings to categorize patients into attribute-based CRC test preference groups. Our algorithm was based on other attribute-based preference elicitation methods, including conjoint analysis (CA) and the analytic hierarchy process (AHP), which require respondents to engage with complex scenarios or systematic deconstruction of decisions into subsets of attributes. Both CA and AHP have been used to assess preferences for CRC screening, and research has found that certain key attributes can be linked to preferences for specific CRC tests.9,25,26 Because these methods are conceptually difficult and typically cannot be done via a telephone survey, we developed a telephone-based approach to assessing them in our study and used our algorithm to link them to preferences for different CRC tests.

Using participant rankings, the following CRC screening test preference structures were generated: 1) strong COL preference (accuracy first, preparation and complications not ranked); 2) weak COL preference (accuracy first,preparation or complications second or third); 3) strong FOBT preference (preparation or complications first, accuracy not ranked); and 4) weak FOBT preference (preparation or complications second, accuracy second or third). Those whose attribute rankings did not fall into one of these 4 groups were categorized into a fifth group labeled “unclear preference.” An unclear preference included individuals who ranked any of the other attributes in their top 3, but where it was not possible to link these preferences to a test. Because of the small number of patients with strong preferences for either FOBT or COL, for analyses we consolidated the 2 FOBT preference groups and the 2 COL preference groups to create a final preference measure with 3 groups: FOBT, COL, or unclear.

Physician CRC screening recommendations: Physician CRC modality recommendation during the PHE was obtained from audio recordings. Details regarding the coding of the content of the PHE related to CRC screening discussions are provided elsewhere.23,24 Because of the high rate of recommendation for COL, we categorized recommendations into 2 groups: COL-only versus COL-plus another test, which was almost universally FOBT (eg, “COL-plus”). All visits included in this analysis were PHEs of patients who were eligible or overdue for CRC screening. Visits by patients who had received a prior screening test in the recommended time frame were not eligible for this study, and visits in which COL was recommended as an appropriate follow-up to a positive FOBT were not included in this study.

CRC Test Utilization

CRC test use in the 12 months post PHE was collected from the EMR. Because of the small number of patients receiving barium enema or sigmoidoscopy (N = 5), we excluded those patients and created a variable measuring CRC screening receipt at 12 months with the following categorization: 1) COL only; 2) FOBT only; 3) COL and FOBT; 4) no screening.

Independent Variables

The primary covariates used in this analysis were patient demographics that were assessed via patient self- report from the pre-visit survey. These characteristics included: age, race (white vs African American/other), gender, education (high school or less vs some college or more) and income (<$40,000 vs $40,000 to <$80,000 vs >$80,000 or more).

ANALYSIS

The analyses followed the research objectives and were conducted accordingly in 3 phases.

Attribute-Based CRC Screening Test Preferences

First, we generated a description of the sample according to the patients’ demographic factors. To assess the importance of each attribute, we calculated the proportion of patients who ranked each attribute 1st, 2nd, and 3rd. We then applied our algorithm to categorize baseline CRC test preferences into the 5 categories as described above. Next, we evaluated associations between baseline preferences and patient demographic factors using X2 tests. We then conducted multinomial logistic regression (MNL) to further evaluate associations between baseline preferences and participants’ demographic characteristics using the 3-level preferences variable (COL-strong/weak, FOBT−strong/weak, unclear), with COL−strong/weak as the referent category.

Physician CRC Screening Recommendations

We conducted a logistic regression of our 2-level recommendation variable (COL-only vs COL-plus) using the latter as the referent category. We conducted this regression in a stepwise manner, with Model 1 including only patient demographics as independent variables, and Model 2 including both demographics and baseline test preferences. In these analyses we adjusted for clustering by the physician’s identification number.

Associations With CRC Test Utilization

We compared both baseline test preferences and physician recommendations with CRC screening test utilization 12 months following the PHE using X2 tests.

RESULTS

Participant Characteristics

Table 2

The characteristics of the participants and nonparticipants are described in detail elsewhere.23,24 Briefly, 47% of physicians and 50% of patients agreed to participate. Physician participants did not differ from nonparticipants in age or gender, but were significantly more likely to be African American or a family medicine physician. Patient participants did not differ from non participants in race or marital status, but were significantly younger and more likely to be female.23,24 Among the 500 consenting patient participants, there were 485 audible office visit recordings. Excluded from consideration in the current analyses are visits for which the audio recording indicated the patient was not due for CRC screening (n = 12), the patient had screening scheduled at the time of the visit (n = 25), or was in the midst of a related diagnostic workup (n = 1). Also excluded are patients for whom the baseline telephone survey was not available (n = 3). The resultant sample consists of 444 unique patient visits. Of these, 93.5% had a discussion of CRC screening in the PHE as documented in the audio recording (N = 415) which is the denominator used in this analysis. Patients were on average 59 years old (SD = 7.9), and typically white (65.5%) and female (64.1%). Most patients were highly educated (72.5% had attended at least some college) and just over one-third had a household income of over $80,000 (35.9%) ().

Attribute-Based CRC ScreeningTest Preferences

Using our 5-level categorization, we found few patients strongly preferred COL (7.0%) or strongly preferred FOBT (11.1%). Most had a weak preference for COL (41.0%), followed by an unclear preference (22.4%) and a weak preference for FOBT (18.6%). We did not find statistically significant differences in attribute-based CRC screening test preferences by patient race, gender, education or income. The results of the MNL model with our 3-level categorization of baseline preferences were consistent with those of the bivariate analysis (data not shown).

Physician CRC Screening Recommendations

Among PHEs with a physician recommendation for CRC screening, COL was mentioned and recommended for CRC screening in almost all PHEs (N = 412; 99.0%), alone or in conjunction with mentioning other tests. In 29.1% of all visits (N = 120), the physician recommended both COL and FOBT. Other CRC screening tests (barium enema and sigmoidoscopy) were recommended in a minority of PHEs, and only as another option to doing COL (2 visits each, 0.98% in total). Although not consistent with guidelines for CRC screening, these visits where COL plus another test were recommended did occur and were considered “COL-plus” recommendations as described above.

In Model 1, men were significantly less likely than women to receive a COL-plus recommendation than a COL-only (OR: 0.52; 95% confidence interval [CI], 0.27- 0.97, P <.05) (Table 3). In Model 2, when baseline CRC test preferences were included, patients with a baseline FOBT preference were more likely than those with a baseline COL preference to receive a COL-plus than a COL-only recommendation (OR: 2.17; 95% CI, 1.26-3.71; P <.01). Similarly, patients with an unclear baseline testpreference were more likely to receive a COL and/or FOBT recommendation than a COL- only recommendation (OR: 2.13; 95% CI, 1.17-3.88; P <.05).

Associations With CRC Test Utilization

Overall, just over half of the patients (55.7%) had received CRC screening 12 months post PHE. Of these, 67.2% had COL only, 19.5% FOBT only, and 13.3% had FOBT, followed by COL (eg, 32.8% had FOBT alone or followed by COL due to a positive result). Although the majority of patients in each attribute-based preference group receiving any CRC test received COL only, 40.6% of those with a baseline preference for COL received COL, while fewer patients with a baseline FOBT or unclear preference received COL (33.3% and 34.4%, respectively). Those with a pre-visit attribute-based preference for FOBT had a higher rate of receipt of FOBT initially (21.1%) than those with an attribute-based preference for COL (16.2%) or unclear preference (17.8%) (Figure 1).

Figure 2

Physician modality recommendation was also associated with CRC test utilization. Patients who were recommended COL-only received COL more often than those recommended COL-plus (44.8% vs 18.3%, P <.001). Compared with patients with a COL-only recommendation, patients who received the COL-plus recommendation more often received FOBT initially (32.5% vs 12.4%, P <.001) ().

DISCUSSION

Our results from a large managed care population eligible and due for CRC screening confirm the findings of others that average-risk patients have distinct attitudes towards the attributes of CRC screening tests.9,15-18,25,26 Our work differed from prior research in that we were able to use attribute rankings to distinguish between stronger and weaker preferences for the 2 most commonly used tests in clinical practice, COL and FOBT. We found that, based on our attribute-based ranking algorithm, few patients had a strong preference for either of these tests (7.0% and 11.1%, respectively). Rather, more patients had a weak preference for the 2 leading test options (41.0% for COL and 18.6% for FOBT), and a sizable group had an unclear test preference (22.4%), which could include preferences for COL or FOBT that may have been associated with unmeasured attributes or factors. These results suggest that while most patients do have distinct attitudes toward key attributes of different CRC screening tests, they may need assistance in clarifying which attributes matter most strongly to them in making a screening decision.

Many patients in our study with an identifiable preference for a CRC screening test based on these key attributes had not been screened 12 months later, and among those who did get screened, we found no association between attribute-based preferences and type of test received. The lack of concordance between test preference and test utilization is consistent with 1 other study.14 Our results also confirm the importance of physician recommendations in screening uptake. Among patients who received a COL-only recommendation, most who got screened by 12 months did so with COL (78.3% of those screened) com- pared with FOBT (21.7%), while the reverse was true for those who received a COL-plus recommendation (64% of those screened received FOBT vs 36% who received COL).

Our results also suggest that physicians may be open to considering the patients’ preferences. Our finding that COL was the recommended test in nearly every visit in which screening was discussed is in line with the growing trend toward physician recommendation of COL in primary care.4,5 When another CRC screening test was recommended, it was always in conjunction with COL, a result consistent with a recent study which found that physicians do not offer a menu of modality choices, but instead often recommend COL alone or in addition to other options.21 This is also consistent with prior work indicating that patients who preferred COL were more likely to have a CRC screening test ordered than those who preferred another test.11 In our study, when patients presented with an attribute-based preference for FOBT or had an unclear preference, they were significantly more likely to receive the COL-plus recommendation versus COL-only. While we cannot be sure from our data, this result suggests that physicians may have been willing to offer a second test option in addition to COL when faced with patients who questioned the initial COL recommendation or clearly indicated a preference for attributes of another test.

Some key limitations should be acknowledged. We used a limited number of attributes in our algorithm, and there may have been unmeasured attributes or factors that could have contributed to preferences for specific tests. Although our algorithm generated CRC test preference patterns similar to those found in other studies that used more established preference elicitation methods,9,25,26 we were not able to directly validate our method against existing preference elicitation methods. However, our preference results were also similar to studies that simply asked patients to indicate their preference for different CRC tests,15,16 and to one study that asked patients to assess the importance of test attributes.19 Further work to evaluate the validity of a simplified algorithm like the one developed in this study is needed before such approaches can be widely applied to preference assessment.

The study was not designed to determine the motivation for recommending more than 1 test in a visit. For instance, we did not code for what patients said during the PHE (eg, questioning the test recommended) that might have prompted an FOBT recommendation in addition to COL. While offering test options is consistent with CRC screening guidelines,4 the practice of recommending 2 tests concurrently is not. Screening was assessed 12 months following the PHE, and may have been influenced by factors not measured in our study. While we did not address important structural factors influencing patient preferences for CRC screening such as logistical barriers or cost of screening, all participants in our study were insured with CRC screening coverage, via a managed care organization in southeast Michigan. However, as a result, we cannot generalize our results to other primary care patients, especially those who are uninsured.

Overall, our results suggest that physicians do not appear to be incorporating preferences for CRC screening test attributes into preventive health visits. Although previous research has shown that physicians have misconceptions about patient preferences for CRC screening tests and test attributes,8,15,27-29 we found that physicians appeared to be more open to considering other options when patients preferred attributes consistent with a test other than COL. A recent study found higher rates of CRC screening ad- herence when patients were offered a choice than when recommended COL only,27 suggesting that being open to other options could be an effective method for increasing screening rates. These results suggest that primary care clinicians may have better success in getting patients to screen if patients are queried specifically about which attributes of screening they prioritize. Interventions designed to prompt physicians to assess preferences for key CRC screening attributes may be one mechanism for increasing adherence and improving shared decision making in clinical practice.Author Affiliations: From Division of General Medicine, University of Michigan Health System, Ann Arbor, MI, and Ann Arbor VA Medical Center, Ann Arbor, MI (SH); Department of Health Behavior and Health Education, University of Michigan School of Public Health, Ann Arbor, MI (SL); Case Western Reserve University Comprehensive Cancer Center, Cleveland, OH (GC); Department of Social and Behavioral Health, Virginia Commonwealth University School of Medicine, Ricmond, VA (JEL).

Funding Source: Support provided by Grant R01 CA112379.

Author Disclosures: The authors have no conflicts of Interest to disclose. All authors contributed to the conception, design and/or analyses of the data presented in the manuscript. All authors have reviewed and approved the final version.

Authorship Information: Concept and design (SH, JEL); acquisition of data (JEL, GC); analysis and interpretation of data (SH, SL); drafting of the manuscript (SH, JEL, GC, SL); critical revision of the manuscript for important intellectual content (SH, JEL, GC, SL); statistical analysis (SH, JEL, SL); provision of study materials or patients (JEL, GC); obtaining funding (JEL); administrative, technical, or logistic support (SH, JEL); supervision (JEL, GC).

Address correspondence to: Sarah T. Hawley, PhD, MPH, Associate Professor, Division of General Medicine, University of Michigan, North Campus Research Complex, 2800 Plymouth Rd, Ann Arbor, MI 48109. E-mail: sarahawl@umich.edu.1. Shires DA, Divine G, Schum M, et al. Colorectal cancer screening use among insured primary care patients. Am J Manag Care. 2011;17: 480-488.

2. Klabunde CN, Cronin KA, Breen N, et al.Trends in colorectal cancer test use among vulnerable populations in the United States. Cancer Epi Biomarker Prev. 2011;20:1611-1621.

3. Schroy PC III, Geller AC, Crosier Wood M, et al. Utilization of colorectal cancer screening tests. Prev Med. 2001;33:381-391.

4. US Preventive ServicesTask Force. Screening for colorectal cancer: US Preventive ServicesTask Force recommendation statement. Ann Intern Med. 2008;149:627-637.

5. Levin B, Lieberman DA, McFarland B, et al; American Cancer Society Colorectal Cancer Advisory Group; US Multi-SocietyTask Force; American College of Radiology Colon Cancer Committee. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps, 2008: a joint guideline from the American Cancer Society,the US Multi-Society Task Force on colorectal cancer, and the American College of Radiology. CA Cancer J Clin. 2008;58:130-160.

6. Whitlock E, Lin J, Liles E, Beil T, Fu R. Screening for colorectal cancer: a targeted, updated systematic review for the U.S. Preventive Services Task Force. Ann Intern Med. 2008;149:638-658.

7. Klabunde CN, Lanier D, Nadel MR, et al. Colorectal cancer screening by primary care physicians: recommendations and practices, 2006— 2007. Am J Prev Med. 2009;37:8-16.

8. McQueen A, Bartholomew LK, Greisinger AJ, et al. Behind closed doors: physician-patient discussions about colorectal cancer screening. J Gen Intern Med. 2009;24(11):1228-1235.

9. Hawley ST, Volk RJ, Krishnamurthy P, et al. Preferences for colorectal cancer screening among racially/ethnically diverse primary care patients. Med Care. 2008;46:S10-S16.

10. Woolf S. Overcoming the barriers to change: screening for colorectal cancer. Am Fam Physician. 2000;61:1621-1628.

11. Schroy PC III, Emmons K, Peters E, et al.The impact of a novel computer-based decision aid on shared decision making for colorectal cancer screening: a randomized trial. Med Decis Making. 2011;31:93-107.

12. DeBourcy AC, Lichtenberger S, Felton S, et al. Community-based preferences for stool cards versus colonoscopy in colorectal cancer screening. J Gen Intern Med. 2008;23:169-174.

13. Lafata JE, Divine G, Moon C, Williams LK. Patient-physician colorectal cancer screening discussions and screening use. Am J Prev Med. 2006;31:202—209.

14. Hawley ST, McQueen A, Bartholomew LK, et al. Preferences for colorectal cancer screening tests and screening test use in a large multispecialty primary care practice. Cancer. 2012;118:2726-2734.

15. Ling BS, Moskowitz MA, Wachs D, Pearson B, Schroy PC. Attitudes to- ward colorectal cancer screening tests. J Gen Int Med. 2001;16:822-830.

16. Marshall DA, Johnson FR, Kulin NA, et al. How do physician as- sessments of patient preferences for colorectal cancer screening tests differ from actual preferences? a comparison in Canada and the United States using a stated-choice survey. Health Econ. 2009;18:1420-1439.

17. Pignone M, Bucholtz D, Harris R. Patient preferences for colon cancer screening. J Gen Intern Med. 1999;14:432-437.

18. Shokar NK, Carlson CA, Weller SC. Informed decision making changes test preferences for colorectal cancer screening in a diverse population. Ann Fam Med. 2010;8:141-150.

19. Wolf RL, Basch CE, Brouse CH, Shmukler C, Shea S. Patient preferences and adherence to colorectal cancer screening in an urban population. Am J Public Health. 2006;96:809-811.

20. Feldstein AC, Perrin N, Liles EG, et al. Primary care colorectal cancer screening recommendation patterns: associated factors and screening outcomes. Med Decis Making. 2012;32:198-208.

21. Zapka JM, Klabunde CN, Arora NK, et al. Physicians’ colorectal cancer screening discussion and recommendation patterns. Cancer Epidemiol Biomarkers Prev. 2011;20:509-521.

22. Steinwachs D, Allen JD, Barlow WE, et al. National Institutes of Health State-of-the-Science Conference statement: enhancing use and quality of colorectal cancer screening. Ann Intern Med. 2010;152:663-667.

23. Wunderlich T, Cooper G, Divine G, et al. Inconsistencies in patient perceptions and observer ratings of shared decision making: the case of colorectal cancer screening. Patient Educ Couns. 2010;80:358-363.

24. Lafata JE, Cooper GS, Divine G, et al. Patient-physician colorectal cancer screening discussions: delivery of the 5A’s in practice. Am J Prev Med. 2011;41(5):480-486.

25. Marshall DA, Johnson FR, Phillips KA, et al., Measuring patient preferences for colorectal cancer screening using a choice-format survey. Value Health. 2007;10(5):415-30.

26. Dolan JG, Boohaker E, Allison J, et al. Patients’ preferences and priorities regarding colorectal cancer screening. Med Decis Making. 2013;33(1):59-70.

27. Inadomi JM, Vijan S, Janz NK, et al. Adherence to colorectal cancer screening: a randomized clinical trial of competing strategies. Arch Intern Med. 2012;172:575-582.

28. Cooper G, Fortinsky R, Hapke R, Landefeld C. Factors associated with the use of flexible sigmoidoscopy as a screening test for the detection of colorectal carcinoma by primary care physicians. Cancer. 1998;82:1476-1481.

29. Klabunde CN, Vernon SW, Nadel MR, et al. Barriers to colorectal cancer screening: a comparison of reports from primary care physicians and average-risk adults. Med Care. 2005;43:939-944.