Costs associated with colorectal cancer (CRC) are projected to be between $11.4 and $14 billion in 2020, creating a substantial burden for Medicare enrollees and the Medicare program (1). Illness and death associated with CRC can be reduced through early detection and treatment. Four recommended tests for CRC screening of persons of average risk (2) are covered under Medicare: 1) annual fecal occult blood testing (FOBT); 2) sigmoidoscopy or 3) barium enema every 4 years; or 4) colonoscopy every 10 years. Despite having Medicare coverage for CRC tests, less than half of Medicare enrollees have received CRC tests at the recommended intervals (3)

Much debate about health care reform centered on providing health insurance coverage to uninsured Americans. Although lack of insurance coverage is a leading reason for lack of access to health care, insurance coverage does not ensure access to health care. Aday et al (4) have used the term "realized access" to highlight the importance of factors in addition to insurance coverage and availability of providers that influence consumer access to health care, including the "transactions between patients and providers during the process of care delivery."

Physician influence on patients' CRC screening status is well documented. Physician recommendation has been associated with increased likelihood of screening (5,6), and the lack of recommendation has been identified as a barrier (7,8). Having a recent well visit (6), being seen in group practices rather than solo practices (9), and visiting internal medicine physicians rather than other primary care providers (10) have all been linked with increased CRC testing. Understanding physicians' influence on the CRC screening status of Medicare enrollees is, however, complicated by the fragmented care received by most Medicare enrollees. One study showed that most Medicare enrollees see 7 physicians in 4 practices in a given year (9).

The objective of this study was to examine the role of physician visits among Medicare enrollees with and without CRC testing — enrollees who access this preventive service and enrollees with "unrealized access." Our aim was to identify potential opportunities to increase CRC testing in the untested population.

The study included 1,108,424 North Carolina and South Carolina Medicare enrollees aged 65 to 85 (Table 1). Of these, 46.5% were not compliant and not tested in 2005; 21.9% had a CRC test in 2005; and 31.6% were compliant with CRC tests and were excluded from further analyses. Of those who needed a CRC test in 2005 (n = 757,594), 68% did not receive one. Compared with enrollees who had any CRC test in 2005, those who did not receive a test were more likely to be of minority race or Hispanic ethnicity (20.4% vs 14.4%); to have originally entered Medicare because of a disabling condition (11.1% vs 8.8%); and to be eligible for a state buy-in program (18.3% vs 10.3%). Those with no test in 2005 were more likely to be classified as unknown in terms of comorbid conditions (15.6% vs 4.4%), likely because of a limited number of health claims in 2004.

Lack of a physician visit during the previous year was higher among enrollees with no CRC tests compared with enrollees who had a CRC test in 2005 (19.8% vs 5.4%) (Table 2). Those not tested in 2005 had, on average, fewer physician visits than those who were tested (4.7 visits vs 7 visits). Among enrollees who were not tested in 2005, 45% received care exclusively from primary care physicians (Table 3). Enrollees without a test in 2005 were less likely than enrollees with a test to have received care from both primary care and nonprimary care physicians (17% vs 24%) and from physicians with primary care, nonprimary care, and mixed specialties (10% vs 18%).

Of all Medicare patients who received care from at least 3 different types of physicians (primary care, nonprimary care, and mixed-specialty physicians), only half had a CRC test in 2005 (Figure 1). Less than one-third of patients who had only primary care, nonprimary care, or mixed-specialty visits had a CRC test.

Medicare patients with obstetrician/gynecologist visits were most likely to be tested in 2005 (47%), followed by patients who had seen physicians with multiple specialties (Figure 2). Patients who had visits with family or general practice physicians were least likely to be tested (29%). No primary care specialty was without missed opportunities: less than half of the patients with primary care visits in 2005 had a CRC test, regardless of the primary care specialty of the physicians.

Model 1 (Table 4) revealed a strong, positive association between the number of visits and having a test, with a monotonic increase in the likelihood of having a CRC test as the number of visits increased. Enrollees with 1 to 5 visits were more than twice as likely to have had a test as those with no visits (OR, 2.6; 95% CI, 2.6-2.7). Enrollees with 21 or more visits were 8 times as likely to have had a test as those with no visits (OR, 8.3; 95% CI, 7.9-8.6).

Model 2 (Table 4) showed that enrollees whose visits included contact with a primary care physician were significantly more likely to be tested in 2005 than enrollees with nonprimary care visits. Odds ratios for all categories of physician visits that included primary care visits were significantly greater than 1. Although the number of visits remained an important predictor, the effect was attenuated by type of physician included in the model.

In Model 3 (Table 4), enrollees who saw family or general medicine physicians were less likely to have had a CRC test (OR, 0.61; 95% CI, 0.60-0.62) than those who visited physicians with internal medicine, preventive medicine, or geriatric specialties. Enrollees who saw obstetrician/gynecologists were more likely to have had a CRC test (OR, 1.23; 95% CI, 1.17-1.29).

In all 3 models, demographic characteristics were significantly associated with receipt of CRC test. Medicare enrollees who were older, minority race or Hispanic ethnicity, or male were less likely to be tested. Enrollees originally eligible because of disability and those entitled to state buy-in (a marker for low income) were also less likely to have had a CRC test. Enrollees with evidence of any comorbid conditions were less likely to have had a CRC test. State variation was also observed: enrollees from South Carolina were more likely to have had CRC tests than those from North Carolina.

Almost half (47%) of the Medicare population in 2 Southern states was in need of, and did not receive, a CRC test in 2005, a finding consistent with earlier studies showing low CRC test use rates despite Medicare coverage (14-16). We found that having any physician office visit, increased number of office visits, and visits with primary care physicians were associated with increased CRC test use. We also found that a number of enrollees did not regularly see physicians, and many of those who did have physician contact remained untested.

We identified missed opportunities by all provider types to promote CRC testing. For most specialties, most of their Medicare patients were in need of a CRC test at the time of the visit and did not receive one. The characteristics of the untested group reveal a potentially vulnerable population — those with Medicaid eligibility, who may have disabling conditions and multiple chronic conditions. Most are accessing medical care, but their access to preventive services remains "unrealized." Our findings highlight the gap between coverage and receipt of CRC test and demonstrate the influence of physician visits. We identify 3 categories of Medicare enrollees who are untested: those with no visits, those with primary care visits, and those with visits to multiple types of physicians. This categorization suggests that different strategies may be needed to improve CRC testing in each group.

Previous research has shown preventive or periodic health exams are associated with CRC test use (17). Yet, we found almost 20% of enrollees who needed a CRC test had no physician contact during the year of observation. In 2011, Medicare will allow 1 annual wellness visit during which counseling on CRC screening could occur. However, prior experience indicates that coverage of a wellness visit by itself will not assure Medicare enrollees use it — in 2006 and 2007, only 3% of enrollees used the "Welcome to Medicare" visit (18). Outreach to new Medicare enrollees, and "in-reach" to recall patients who have not recently been seen, may improve CRC testing rates (19).

As a stand-alone strategy, encouraging enrollees to visit primary care physicians may not result in substantial improvements in CRC testing at the population level (20). In our cohort, many enrollees remained untested regardless of the frequency of visits. Our findings are consistent with earlier studies reporting that contact with physicians is not sufficient to ensure that CRC testing will be completed (5,15,21,22). Greater understanding of tools and systems is needed to support physicians in the promotion of CRC testing. Factors identified as barriers to physicians offering CRC testing include the limited amount of time for a patient visit, competing priorities, and lack of office systems to facilitate screening (23-25). Increased CRC screening referrals have been documented through the use of simple tools such as chart stickers and reminders (26). A toolbox containing sample policies, reminder systems, and communication approaches is available from the National Colorectal Cancer Roundtable to assist physicians in increasing CRC testing (www.nccrt.org).

Physician attitudes toward cancer screening may vary by specialty. The higher rate of CRC testing we found among enrollees with an obstetrician/gynecologist visit is consistent with findings from an earlier study showing significantly higher rates of mammography referral for patients treated by obstetrician/gynecologists compared with other primary care physicians (27).

Although strategies to improve CRC testing typically target primary care physicians (28,29), we found that a combination of primary care and specialist visits was associated with increased CRC testing. A recent study reported that specialists spend a substantial portion of their time providing routine care and preventive services to their patients (30). Electronic health records (EHRs) offer potential for improving CRC testing among all physicians, including specialists, by alerting physicians to patients' test status. EHRs could identify and recall patients for screening, remind office staff to counsel patients about screening, and monitor screening compliance. An EHR system with the ability to exchange information across settings may be the only feasible way to manage the multiplicity of communication that needs to occur between the many physicians providing care to Medicare enrollees.