A team of content and methodology experts at the University of New Mexico Cancer Center developed the primary care physician survey in 2005. Some items were adapted from a National Cancer Institute survey on CRC screening barriers (15). We revised the survey after pilot testing it with a sample of 5 primary care physicians who assessed content and face validity. The final 34-item survey was divided into 6 sections that addressed the practice strategies and system support for performing screening (patient education, reminders, tracking), the rationale for recommending screening (influential factors), cancer screening beliefs and practices (perceived effectiveness, testing recommendations, patient adherence), screening barriers (patient, provider, system), practice characteristics (type, location, size), and physician characteristics (demographics, specialty). (Readers can obtain a copy of the survey from the corresponding author.) Screening practice items were based on USPSTF recommendations (16) for average-risk patients aged 50 years or older and included the options of colonoscopy, flexible sigmoidoscopy, barium enema, or fecal occult blood testing (FOBT). We used 6-point rating scales ranging from "never"/"not" = 1 to "most"/"very" = 6 to measure survey items assessing screening beliefs, practices, and barriers. The human research review committee of the University of New Mexico approved the physician survey.

We identified potential respondents from the 2006 New Mexico Board of Medical Examiners provider list, which contains the self-identified specialties and addresses for licensed physicians in New Mexico. The eligibility criterion was a primary specialty listing of internal medicine, family medicine, general practice, or geriatrics. We excluded physicians with secondary specialties that indicated that they were not primary care providers, including medical subspecialties, urgent care, or administration.

We mailed 964 surveys in June 2006 accompanied by a cover letter signed by leaders of primary care professional organizations and the New Mexico Department of Health. We subsequently mailed surveys approximately 3 and 5 months later to nonrespondents. We did not offer compensation for participating in the survey. On the basis of returned mailings, updated Medical Examiners lists, and survey responses, we excluded 123 physicians from our denominator because we could not locate them and another 127 physicians who either were no longer licensed in New Mexico (retired or moved) or who were not primary care providers. We were left with a denominator of 714 respondents and 216 completed surveys, for a response rate of 30%.

The study participants were a population-based sample of respondents to the 2004 New Mexico BRFSS, a federally funded annual survey implemented in collaboration with state health departments. Random-digit dialing methods were used to derive a probability sample of households with telephones to collect data on health-related behaviors and risk factors for respondents aged 18 years or older. In 2004, the New Mexico Department of Health designed a module that was administered to 6,390 adults aged 50 years or older that asked about CRC screening with a home FOBT and lower endoscopy. Overall, 3,355 participants who reported being not current or never tested with either home FOBT or lower endoscopy were asked to provide the most important reason for the lack of testing. These participants included 1,695 who had never performed a home FOBT and 798 who had not been tested in the past year, as well as 1,504 who never had undergone a sigmoidoscopy or colonoscopy and 297 who had not been tested in the past 5 years.

For the physician survey, we used descriptive statistics to characterize providers, practices, screening practices, and ratings of barriers to screening. We conducted a series of exploratory factor analyses on the patient, provider, and system barriers assessed by the survey to confirm that survey items clustered within these 3 distinct groups of barriers. We used factorial within-subjects multivariate analyses of variance to directly compare physician rating levels within each set of barriers to CRC screening (20).

To account for the complex survey design of the BRFSS, we used Stata 9.0 (StataCorp LP, College Station, Texas) to estimate percentages and their 95% confidence intervals (CIs) by demographic characteristics. All estimates represent weighted population-based estimates for residents aged 50 years or older who responded to the 2004 New Mexico BRFSS.

Among the physician respondents who completed these survey items, most were middle-aged white men who practiced in urban areas (Table 1). Almost half practiced in physician-owned, single-specialty practices, and most worked in small groups.

Few practices used an electronic medical record to identify patients due for CRC screening and most did not have a mechanism to ensure that patients completed FOBT tests (Table 2). However, few respondents found it difficult to arrange CRC screening.

The most influential factors for screening, on a 6-point scale from 1 (not at all influential) to 6 (very influential), were evidence from the medical literature and professional guidelines. Respondents considered screening colonoscopy to be the most effective strategy for reducing CRC mortality. They were less enthusiastic about FOBT, flexible sigmoidoscopy, and barium enema.

Colonoscopy was recommended at least some of the time for screening average-risk patients by 94% of respondents compared with 86% for FOBT. Overall, 74% of respondents never recommended flexible sigmoidoscopy, and 74% never recommended barium enema. Nearly all respondents reported that they began screening average-risk patients at age 50 for each of the screening procedures except FOBT, for which 23% reported that they began screening patients in their 40s. The mean (standard deviation) screening interval in years was 1.4 (1.6) for FOBT, 8.2 (2.6) for colonoscopy, 6.3 (2.4) for barium enema, and 4.5 (1.3) for flexible sigmoidoscopy. Nearly all respondents indicated that they had an upper age limit for screening with colonoscopy, barium enema, and flexible sigmoidoscopy; about half indicated that they would not screen beyond age 80. However, 63% of respondents reported not having an upper age limit for FOBT screening.

In the past year, respondents estimated offering CRC screening to a median (interquartile range) 80% (70%-95%) of their patients aged 50 years or older. Overall, respondents estimated that 50% (40%-70%) of their eligible patients were current with CRC screening recommendations.

Only 3 patient barriers were rated 4 or higher on the 6-point rating scale: embarrassment/anxiety, fear of pain, and inadequate insurance (Table 3). Physician barriers were all rated below the midpoint of 3.5 on the 6-point rating scale, indicating that physicians did not see themselves as often creating barriers to screening. Similarly, physicians generally did not perceive system factors, which were categorized into implementing screening and obtaining lower endoscopy, as creating barriers to screening.

Among the 3,355 respondents to the BRFSS module questions about lack of screening, 58% were aged 50 to 64 years, 47% were men, 67% were non-Hispanic whites, and 26% were Hispanic. By far, the most frequently cited response for never having undergone CRC screening or not being current with either FOBT or lower endoscopy was that a doctor never suggested testing (Table 4). A substantial proportion of participants indicated that lack of symptoms was the reason for not being screened. Few reported testing being distasteful or embarrassing, having fear of cancer, or having fear of pain with testing as barriers. Furthermore, access and cost issues were not frequently listed as barriers for FOBT, though they were slightly more common for endoscopic procedures.

Our study had some potential limitations. The overall response rate to the physician survey was low, creating a potential selection bias if respondents were not representative of the population of New Mexico primary care physicians. However, a recent New Mexico Health Policy Commission (NMHPC) report suggests that the demographics of our sample were consistent with statewide data on primary care physicians (41). In 2008, the NMHPC reported that 43% of primary care physicians were aged 55 or older (vs 39% in our sample), 56.9% were men (vs 68%), and 48% were in the county containing Albuquerque (vs 47%). We also know that our denominator of potentially eligible physicians was not accurate and that we likely underestimated our response rate. We relied on the Board of Medical Examiners physician listings, which do not consistently characterize specialty or training status, so we may have misclassified specialists and trainees as being primary care physicians. Contact information is updated only every 3 years, so we could not be certain that we had correct addresses. However, our results in terms of practice patterns, system support, and barriers are consistent with those of other surveys reporting higher response rates (15,33). We were also unable to verify provider responses regarding screening practices and adherence.

The BRFSS data are subject to selection and recall bias, although reports of physicians failing to recommend screening are supported by national surveys (15) as well as directly observed patient encounters (31). Social desirability bias may have caused respondents to minimize fear and embarrassment as screening barriers. Finally, our data are ecologic; physician and BRFSS respondents are not directly linked, which could result in differing perceptions of barriers, particularly related to access.

Our results suggest that CRC screening in New Mexico could be facilitated by information systems that readily identify patients who are due for screening and track test results. Physicians may also increase screening by educating patients about cancer and the rationale and options for screening. Physicians in a state with limited resources for cancer screening should also avoid potentially inefficient (not having an upper age limit for FOBT screening) and impractical (emphasizing screening colonoscopy) screening strategies.