|Home | About | Journals | Submit | Contact Us | Français|
Rising colorectal cancer (CRC) screening rates in the last decade are attributable almost entirely to increased colonoscopy use. Little is known about factors driving the increase, but primary care physicians (PCPs) play a central role in CRC screening delivery.
Explore PCP attitudes toward screening colonoscopy and their associations with CRC screening practice patterns.
Cross-sectional analysis of data from a nationally representative survey conducted in 2006–2007.
1,266 family physicians, general practitioners, general internists, and obstetrician-gynecologists.
Physician-reported changes in the volume of screening tests ordered, performed or supervised in the past 3 years, attitudes toward colonoscopy, the influence of evidence and perceived norms on their recommendations, challenges to screening, and practice characteristics.
The cooperation rate (excludes physicians without valid contact information) was 75%; 28% reported their volume of FOBT ordering had increased substantially or somewhat, and the majority (53%) reported their sigmoidoscopy volume decreased either substantially or somewhat. A majority (73%) reported that colonoscopy volume increased somewhat or substantially. The majority (86%) strongly agreed that colonoscopy was the best of the available CRC screening tests; 69% thought it was readily available for their patients; 59% strongly or somewhat agreed that they might be sued if they did not offer colonoscopy to their patients. All three attitudes were significantly related to substantial increases in colonoscopy ordering.
PCPs report greatly increased colonoscopy recommendation relative to other screening tests, and highly favorable attitudes about colonoscopy. Greater emphasis is needed on informed decision-making with patients about preferences for test options.
Over the past decade, the US Preventive Services Task Force, American Cancer Society, American College of Gastroenterology, and other organizations have published guidelines recommending a number of testing options for colorectal cancer (CRC) screening.1,2 CRC screening rates continue to be suboptimal,3,4 and guidelines continue to encourage screening by one of several effective tests. The prevalence of screening by colonoscopy specifically, however, has greatly increased compared with other testing options.5–7 National trend data show a rise in CRC screening rates since 2000 that is attributable almost entirely to increased use of colonoscopy.
Even though the increase in colonoscopy use has been dramatic, little is known about what factors influenced this increase. Numerous studies have illustrated that patient, provider, practice, and health system factors affect primary care physician (PCP) behavior related to cancer screening, 8–10 and the pivotal role of PCPs in recommending and increasing CRC screening is well established.11 However, PCPs’ opinions about colonoscopy have not previously been assessed.
In this study, using data from a nationally representative survey of practicing PCPs, we (1) investigate physicians’ reports of changes in the past 3 years in the volume of CRC screening tests that they order, perform or supervise, with a particular focus on colonoscopy; (2) explore physicians’ attitudes related to colonoscopy and potential associations of their attitudes with screening colonoscopy practice patterns; and (3) investigate the relationships of selected known or postulated factors with the reported changes in volume of colonoscopy ordering. These data may be helpful for understanding practice variation and to guide thinking about future research priorities and potential interventions.
A nationally representative sample of PCPs participated in a survey between September 2006 and May 2007, sponsored by the National Cancer Institute (NCI) in collaboration with the Agency for Healthcare Research and Quality (AHRQ) and the Centers for Disease Control and Prevention (CDC). Eligible physicians included office-based family physicians, general internists, general practitioners, and obstetrician/gynecologists (OB/GYNs) aged 75 or younger. A sample of PCPs was selected from the American Medical Association’s Physician Masterfile using the four specialties as sampling strata. Prior to selection, the sampling frame database was sorted by age, gender, US Census region, and urban-rural practice location within each stratum. Detailed survey content and sampling strategy are reported elsewhere.12,13 The study was reviewed by the institutional review boards of the NCI and CDC and determined to be exempt.
In September 2006, 1,975 PCPs were sent a questionnaire on colorectal and lung cancer screening via express mail, and several methods were used to encourage participation.14 These methods included letters of endorsement from national physician organizations, postage-paid return envelopes, a $50 honorarium, two additional mailings, up to three follow-up telephone reminders to non-respondents, and the option to complete the survey by telephone. Additional details have been published elsewhere.12
The dependent measure of interest was physicians’ reports of changes in the volume of screening tests they ordered, performed, or supervised in the past 3 years for FOBT, sigmoidoscopy, and colonoscopy. Response categories were on a 5-point scale (increased substantially, increased somewhat, stayed the same, decreased somewhat, or decreased substantially).
Guided by the social ecological perspective—which recognizes the influence of multi-level factors on behavior—and the Theory of Planned Behavior,15,16 we assessed the relationship of selected factors with test ordering reports. The Theory of Planned Behavior highlights the potential influence on behavior of a physician’s attitudes, perceived norms, and perceived challenges.
Physicians’ attitudes toward colonoscopy screening included agreement with three statements: “it is the best of the available CRC screening tests,” “it is readily available for my patients,” and “I worry that I might be sued if I do not offer screening colonoscopy to my patients.” Responses to these statements were measured on a 4-point scale, from “strongly agree” to” strongly disagree.” Additionally, we created a summary measure of the strength of agreement across the three items: strongly agree with none of the three items, strongly agree with one of the three items, strongly agree with two, or strongly agree with all three items.
Three items related to reported influence of perceived norms. Respondents rated local collegial norms, patient preferences, and national recommendations (USPSTF, ACS, and published clinical evidence) on a scale of “very influential,” “somewhat influential,” and “not influential” to their practice of CRC screening. Respondents were also given a “not applicable/not familiar” choice.
With respect to perceived challenges, they were asked how influential two items were in their recommendations for CRC screening: the cost of screening tests for patients with no third party coverage and availability of reimbursement by third party payers, including Medicare and Medicaid.
Additional measures included individual physician characteristics, that is specialty, age, sex, race, and medical school affiliation. Practice context measures included size (number of physicians), geographic location (urban, large rural city/town, small/isolated rural town), and the practice’s panel of patients, including percent of uninsured patients. Practice systems support measures included whether CRC screening guidelines had been implemented, whether CRC screening reminders were provided to the physician and/or patients, and type of medical records system used (paper, in transition, or electronic).
We used descriptive statistics to examine reported changes in ordering by test type, reports of agreement with attitudes concerning colonoscopy, and the distribution of factors potentially associated with physicians’ colonoscopy test ordering. Further, we used chi-square statistics to assess the bivariate associations of these measures with the reported change in colonoscopy ordering. Variables with an association at p<0.10 were retained for multivariate models. Polytomous logistic regression models assessed factors associated with changes in colonoscopy ordering. We estimated two models: one with the three attitude items included separately as covariates and one with a summary attitude measure as a single covariate. We used the statistical program SUDAAN version 9.1 to account for the complex survey design and to incorporate survey weights to obtain national estimates. Analyses were conducted in 2010–2011.
As previously reported, 1,266 PCPs responded to the survey on colorectal and lung cancer screening.12 The survey’s absolute response rate was 69.3% and cooperation rate (excludes physicians lacking valid contact information) was 75.0%. Physicians’ personal and practice characteristics are reported in Table 1. These characteristics produced by the weighted analyses are nationally representative of practicing PCPs in the US. The majority of respondents were male, non-Hispanic white, not affiliated with a medical school, in solo or small group practices, and in urban environments. The majority reported that a low proportion of their patients were uninsured, CRC screening guidelines were implemented in the practice, and they used paper rather than electronic medical records. A minority reported systems in place for CRC screening reminders for the physician or the patient.
As reported in Figure 1, 28% of respondents reported their volume of FOBT ordering had increased substantially or somewhat. In contrast, the majority (53%) reported their sigmoidoscopy volume had decreased either substantially or somewhat. The vast majority (73%) reported that colonoscopy volume had increased somewhat or substantially.
Figure 2 and Table 2 (see “total” column) show PCPs’ attitudes about colonoscopy. The majority (86%) strongly agreed that colonoscopy was the best of the available CRC screening tests. A smaller proportion (69%) thought it was readily available for their patients. Asked whether they might be sued if they did not offer colonoscopy to their patients, 59% either strongly or somewhat agreed and 42% somewhat or strongly disagreed. The majority of respondents (53%) strongly agreed with two of the three items, and 14% agreed with all three items.
As shown in Table 2, 56% of PCPs indicated that local collegial norms were very or somewhat influential in their screening practices and 80% that patient preferences were very or somewhat influential. A majority reported ACS (69%) and USPSTF guidelines (68%) and published clinical evidence (72%) as very influential. In response to two measures related to financial issues, 81% reported that the cost of CRC screening tests for patients with no third party coverage was very or somewhat influential, while 65% reported that reimbursement availability by third party payers was very or somewhat influential.
Table 2 also shows the bivariate relationships of attitudes, perceived norms, and challenges with reported changes in colonoscopy ordering. The three individual colonoscopy attitude measures were each significantly related to reporting substantial increases in colonoscopy ordering as was the summary measure (Fig. 3). Table 3 shows the findings of the regression analyses. Agreement with the three individual attitude measures was significantly related to reporting substantial increases in colonoscopy ordering compared to other reports. Those who opined it was the best available test, strongly or somewhat agreed it was readily available, or strongly or somewhat agreed or somewhat disagreed they worried they might be sued were more likely to report a substantial increase in colonoscopy ordering.
With respect to other physician perception measures (influence of norms and patient preference and challenges), only one was independently related to increased colonoscopy ordering. PCPs who indicated that availability of third party reimbursement for CRC screening tests was very influential in their recommendations were significantly more likely to report substantially increased colonoscopy ordering than were PCPs who said that third party reimbursement was somewhat or not influential.
Several individual-level demographic measures were independently associated, notably specialty and age. Two practice-level measures, having CRC screening guidelines implemented in the practice and having EMR partially or fully in place, were related to reporting a substantial increase in colonoscopy ordering.
Table 4 shows the findings of the polytomous logistic regression model that included the attitude summary measure (other item findings not reported in table). Those who strongly agreed with one or more of the three colonoscopy attitude items were significantly more likely to have substantially or somewhat increased their ordering of colonoscopy. The magnitude of the odds ratios increased incrementally with the number of items with which the physician agreed. Other significant items in this model remained essentially the same as the first model.
This national survey confirms other studies documenting greatly increased recommendation and use of colonoscopy for CRC screening.5–7,17 Only a decade ago, a national survey conducted by NCI showed that very small proportions of PCPs endorsed colonoscopy as the CRC screening test they most often recommended to their patients.18 Following that survey, in 2001, the Centers for Medicare and Medicaid Services added coverage of screening colonoscopy for average-risk Medicare beneficiaries.19 In 2002, the US Preventive Services Task Force (USPSTF) added colonoscopy to the menu of recommended options for CRC screening,20 but did not single out one test as more effective than other recommended options, which included FOBT and sigmoidoscopy. Thus, when our data were collected, all major organizations were unanimous in saying that any of several CRC screening strategies were appropriate. Colonoscopy also was not considered the best test in quantitative analyses performed by the USPSTF and the Institute of Medicine.21 Further, guidelines explicitly recommended that physicians discuss available test options with their patients and that choices among options be made based on cost, availability, convenience, and personal preference. However, as we and others have documented, discussing these options with patients does not appear to be routine practice for most PCPs,22 and PCPs’ discussions with patients about CRC screening are cursory and omit explanation of procedure risks.23,24 Instead, colonoscopy—the most expensive option that also carries significant risk 25—has become the screening test choice in the US.
Why has colonoscopy recommendation and ordering overwhelmed other colorectal cancer screening tests? Our data provide insights into influences that may be driving this change. A large proportion of PCPs (70%) viewed colonoscopy as the best of the available colorectal cancer screening tests. The majority (64%) strongly agreed that colonoscopy testing was readily available for their patients. More than half (58%) strongly or somewhat agreed that they worried they might be sued if they did not offer screening colonoscopy to their patients. Strong agreement with each of these three attitude items was independently and significantly related to PCPs’ reports of substantially increased colonoscopy ordering. To our knowledge, these attitude questions about being the best available test and fear of lawsuits have not been asked before in a national survey. They provide a credible explanation for increased referral to colonoscopy but do not elucidate how these opinions were shaped.
The proportion of PCPs feeling strongly about being sued is curious.26 Interestingly, gastroenterologists may fear being sued because of injury during colonoscopy, but PCPs may be concerned about not recommending the perceived “gold standard.”27 Gastroenterologists who perform the procedure may be particularly important in influencing its use.28 For example, one newspaper advertisement was titled: "Your golden years deserve the gold standard of colon cancer screening."29 The number of websites and blogs advocating colonoscopy30 as the gold standard may reflect marketing to PCPs as well as to the public who increasingly use the web for health information.31
From the community perspective, it is important to note that 31% of respondents reported that colonoscopy was not readily available for their patients. If accurate, this perception may reflect lower specialist capacity in certain geographical areas or regional variation in practice patterns.32 Medicare coverage of screening colonoscopy may have lessened the concern about patients not having insurance and increased the ordering of colonoscopy.33 These effects may increase following implementation of the Affordable Health Care Act, which provides coverage without co-payments.34
However, access to CRC screening remains an issue given persistent evidence of disparities for certain racial/ethnic groups, individuals with low socioeconomic status, and the uninsured.17 Subramanian and colleagues35 argue that, at the population level and in an era of tight budgets, screening patients with FOBT as opposed to colonoscopy would result in more people being screened and a greater gain in life-years. As discussed in the 2010 NIH State of the Science Conference on Enhancing Use and Quality of Colorectal Cancer Screening, continued heavy reliance on screening colonoscopy is unlikely to facilitate attainment of high rates of CRC screening because current capacity to provide this procedure to the millions of adults aged 50–75 who have not been screened may be insufficient.4
Practice level measures were also significantly related to reports of increased colonoscopy ordering. Numerous studies have highlighted the major role of reminder systems in increasing screening prevalence.9 In our analyses, while reminders were not independently related to reports of increased colonoscopy recommendation, physicians in practices with partial or full electronic medical record systems were more likely to report substantial increases in colonoscopy ordering. Perhaps EMRs cue more efficiently and therefore result in more ordering.36,37
We found that OB/GYNs were less likely than other PCPs to report substantial increases in colonoscopy ordering. This difference may be a reflection of the younger patient population that is typically seen by OB/GYNs. It also could be related to their propensity to use in-office FOBT.38 Another differentiating physician characteristic was age. Older PCPs were more likely to report substantially increased use of screening colonoscopy. The explanation for this is speculative, but perhaps younger physicians have been ordering more colonoscopy right along while older physicians report an increase given changes in coverage over the years.
We acknowledge several limitations of this study. While we investigated three potential attitudes’ relationship to increases in screening colonoscopy, there may indeed be other factors worthy of investigation. Additionally, our study does not include the patient perspective. It is based on a physician survey, and direct patient reports therefore were beyond its scope. We attempted to assess the extent to which physicians were influenced by patients’ preferences, but our measure may have been too indirect and subject to socially desirable responses. A more direct assessment, such as “How often do your patients ask specifically for screening colonoscopy,” could be informative in future research. Our assessment of what physicians thought was the best test was also limited, and we did not ask how often PCPs use personalized discussion to determine the best test for each patient. This issue should be addressed in future study. For example, how would recommendations be modified in the presence or absence of a family history of colorectal cancer or a family or personal history of adenomatous polyps?1,21 The survey items were not explicit about the categorization of FOBT and FIT tests. Finally, we analyzed physicians’ reports of their practice patterns without corroborating evidence from chart reviews or insurance claims. Some studies have reported that physicians overestimate their screening recommendations,39 while a more recent report documented the reliability of physician self-report of preventive care activities.40 Other validation studies support use of survey data for profiling CRC screening trends and patterns.7
This study strongly confirms the significant increases in colonoscopy ordering for CRC screening in other national reports, and the findings provide insight into factors related to PCPs’ recommendations and ordering practices. It is important to note, particularly in an era of increasing social media, that additional normative factors may influence attitudes, as well as other factors that we were unable to assess in this survey. Clearly, influences in addition to practice guidelines are shaping PCPs’ recommendations. Perhaps engaging clinicians in discussion about screening test efficacy, availability, costs, and harms, as well as the legitimacy of litigation fears, could promote more acceptance of recommendations that call for patient-centered decision making. While CRC screening use has been increasing in the US, a significant proportion of the adult population remains unscreened, and encouraging patient and physician discussions about options for screening may help to increase appropriate use and address disparities. Interventions might include those to promote increased CRC screening and shared decision making since both are consistent with health care reform and other efforts that emphasize evidence-based practice, the patient-centered medical home, and enhanced access and cost-effectiveness.41
Funding support for this study was provided by the National Cancer Institute (contract no. N02-PC-51308); the Agency for Healthcare Research and Quality (inter-agency agreement nos. Y3-PC-5019-01 and Y3-PC-5019-02); and the Centers for Disease Control and Prevention (inter-agency agreement no. Y3-PC-6017-01). The findings and conclusions in this report are those of the authors and do not necessarily represent the views or official position of the National Cancer Institute, the Agency for Healthcare Research and Quality, or the Centers for Disease Control and Prevention.
The authors declare they have no conflicts of interest.