The RTI International-University of North Carolina Evidence-based Practice Center (RTI‐UNC EPC) prepared this report for the National Institutes of Health (NIH) State-of-the-Science Conference on Enhancing Use and Quality of Colorectal Cancer Screening, which is scheduled for February 2010. This chapter summarizes and discusses the findings of our review of peer-reviewed literature concerning improving the appropriate use and quality of colorectal cancer (CRC) screening.

We adopted three outcomes on which to focus: use of CRC screening, patient-physician discussions about CRC screening, and quality of CRC screening. The screening tests included in our review are the at-home fecal occult blood test (FOBT), flexible sigmoidoscopy (FS), colonoscopy, and double contrast barium enema. We attempted to find studies on the uses of tests recently introduced to clinical practice for CRC screening, including the fecal immunochemical test, fecal DNA testing, and computed tomographic colonoscopy, but found no studies concerning the trends in use and quality of these tests. We further examined “appropriate” use in terms of three constructs: underuse, overuse, and misuse. This report presents findings from a systematic review of literature from January 1998 to September 2009 of four key questions (KQs):

  • KQ 2: What factors influence the use of CRC screening?
  • KQ 3: What strategies are effective in increasing the appropriate use of CRC screening and followup?
  • KQ 4: What are the current and projected capacities to deliver CRC screening and surveillance at the population level?
  • KQ 5: What are the effective approaches for monitoring the use and quality of CRC screening?

We also present background information on trends in the use and quality of CRC screening (KQ 1), relying on national studies and relevant articles from our extensive search for KQs 2 through 5. Finally, we comment on research needs (KQ 6).

Results for KQ 2 are largely descriptive. KQ 3, KQ 4, and KQ 5 are more analytic; each asks for information about the effectiveness of different approaches and an interpretation of comparisons presented in study analyses. For this reason, we provide strength of evidence evaluations for KQ 3, KQ 4, and KQ 5 but not for KQ 2; the strength of evidence tables and overall grades can be found in Chapter 4 in the relevant sections. We refer readers to Chapter 2 for methods for rating the quality (internal validity, or risk of bias) of individual studies and for grading the overall strength of evidence for specific groups of studies.

In the remainder of this chapter, we first give an overall summary of our findings, for all KQs. We then consider some implications of our findings and discuss the limitations of the review. Finally, we present suggestions for future research (KQ 6).

Summary of Findings

As summarized in Table 54, our extensive literature review for KQ 1 found many problems of underuse, overuse, and misuse of CRC screening. To guide our systematic reviews for KQs 2 to 5, we developed an analytic framework (Figure 1, Chapter 2). Our review helped us to specify better the important factors in the analytic framework that may be helpful in considering ways to improve the appropriate use and quality of CRC screening.

Table 54. Summary of the evidence and strength of evidence grades by key question.

Table 54

Summary of the evidence and strength of evidence grades by key question.

From the patient’s point of view in Figure 1, we found that access to health care in general (including having health insurance and a regular source of primary care) is a necessary predisposing factor to CRC screening. Our KQ 2 review found specifically that having health insurance and a regular physician are strongly associated with higher levels of CRC screening. People without health insurance and a regular source of primary care have very low screening rates. This is not surprising; the nature of CRC screening, and the absence of a national program outside of primary care to deliver screening to the uninsured, is such that having a regular source of primary care is essential to improving CRC screening rates.

As shown in the analytic framework (Figure 1), however, access to care alone is insufficient to guarantee high levels of screening. After patients have access to care, they still need a simple and reliable mechanism by which to engage with physicians and/or others in the health care system to understand the idea of screening and the pros and cons of screening strategies. Few health care systems build such discussions into routine care, as shown by our KQ 1 finding of suboptimal numbers and quality of discussions about CRC screening. With such a varied range of screening strategies for CRC screening, this lack of a mechanism to promote and assist patient understanding and choice is a major barrier to appropriate use. An important finding in our KQ 2 review is that unscreened patients did not know about the need for CRC screening, and “just didn’t think about it.” The great majority of patients with physicians who recommend screening actually complete screening. Ideally, this recommendation would be accompanied by a reasonable discussion of screening options.

Helping patients understand CRC screening entails more than giving information in a one-sided, noninteractive manner. Our KQ 3 review found that small media messages with such materials as brochures alone were ineffective in increasing appropriate screening. Certain decision aid designs may be a useful approach in assisting patients to understand the pros and cons of screening and to make informed decisions about which screening strategy is right for them. The evidence to date on the effectiveness of decision aids is insufficient to determine the best design and delivery models; more research is needed.

In developing systems that can reliably help patients understand CRC screening and choose an appropriate screening strategy, several groups of disadvantaged patients need special attention. Patients who are not fluent in the English language, patients whose culture differs from that of the prevailing US medical culture, and (probably) patients with low levels of health literacy need specially designed approaches to help them understand CRC screening. Our KQ 2 review showed the screening rates of Hispanic people and of people not acculturated into the United States to be significantly lower than those of non-Hispanic whites and/or those born or living in the United States for longer periods of time.

Because of the few studies of the association between physician characteristics or health care systems and CRC discussions/screening, we cannot say whether certain types of physicians, or certain types of systems within which physicians work, are or are not more associated with appropriate screening. Some evidence indicates (KQ 2) that system factors such as involving nonphysician staff and having electronic medical records may be associated with appropriate screening.

Although access to care (e.g., a regular source of primary care that one visits at least annually), together with health insurance coverage for screening, combined with patient understanding and physician recommendation of screening, increases appropriate screening for many people, though some groups still need further assistance with completing screening. This step, between patient decision and appropriate test use, appears straightforward in the analytic framework (Figure 1); we found it is often more complex than depicted in this framework.

Because CRC screening strategies require people to carry out unusual procedures of preparation and testing, and then to navigate the medical system to complete screening, assisting people in the details of completing screening is sometimes necessary to reach high rates of appropriate screening. Our KQ 3 review found that, in some populations, employing more intensive one-on-one approaches, eliminating or reducing structural barriers for patients, and making overall system changes successfully increase appropriate screening. These sometimes intensive approaches are likely not necessary for all populations, although reducing barriers and streamlining and simplifying the screening process are likely to be helpful for all.

We developed our analytic framework and conducted our review with the understanding that, from the major guideline groups, a range of appropriate options for CRC screening exists. In contrast to this view, however, the United States might decide to favor a strategy of preferring an initial colonoscopy over other strategies. Our review of KQ 4 indicates a considerable degree of uncertainty about whether the nation has existing—or even latent—capacity to meet the need in this latter assumption. That is, we cannot conclude that the country can either conduct “catch-up” screening of people who have not been screened or to continue steady-state screening and the resulting surveillance for the longer term. Thus, if the United States were to embark on an initial colonoscopy-preferred strategy, and if the above approaches to increasing screening use were effective, then inadequate capacity to screen the eligible population is a real possibility.

Almost all the literature we found and reviewed for KQ 2 and KQ 3 focused on the problem of underuse of CRC screening. Despite our finding in KQ 1 that CRC screening discussions are also underused, we uncovered little evidence concerning factors associated with, or interventions to improve, underuse of screening discussions.

KQ 1 also showed considerable problems with overuse and misuse of CRC screening. No studies reviewed in KQ 2 examined factors associated with overuse of screening; no studies reviewed in KQ 3 examined interventions to reduce overuse of CRC screening. Similarly, little of the literature for KQ 3 examined interventions to reduce misuse in screening.

As shown in our analytic framework (Figure 1), an important (and probably necessary) factor between decisions about screening and appropriate use (minimizing overuse and misuse as well as underuse) is monitoring. Our review in KQ 1 found several monitoring systems for self-report of CRC screening use; these include the National Health Interview Survey (NHIX) and the Behavioral Risk Factor Surveillance System (BRFSS). However, few systems monitor overuse and misuse. We found no systems in the United States for reducing overuse and no corrective steps to minimize misuse. We found no direct evidence about monitoring to review in KQ 5. Table 54 highlights our primary findings and conclusions from KQs 1 to 5.

We have adapted general recommendations for monitoring systems (Table 55) to show what types of data systems might be considered. Some initial systems are being started and could be encouraged and expanded. A national program of breast and cervical cancer screening and a mammography consortium both provide important information to monitor screening for these cancers. A complementary approach might be to expand data collection in the Surveillance, Epidemiology and End Results (SEER) program areas to include screening rates and even misuse data; such information might then be correlated with incidence and pathology data from SEER.

Table 55. Features of an ideal monitoring system for CRC screening use and quality.

Table 55

Features of an ideal monitoring system for CRC screening use and quality.

Implications of This Review

Although we found a gradual increase in CRC screening over the past 10 to 12 years, this increase still leaves the nation at a lower screening rate for CRC than for breast (or even prostate) cancer.194 Finding interventions to increase appropriate CRC screening has clearly been challenging, perhaps more so than for other cancers. Perhaps because of the complexity (and even invasiveness) of the CRC tests, or because of the problem of having to choose among screening strategies, many people have not understood the need for CRC screening, and others have not been able to complete screening. Medical practice systems have often been inadequate in informing patients, discussing their questions, and assisting them in the complexities of completing CRC screening.

Our summary of our findings highlights certain aspects of our analytic framework (Figure 1) and points to a logical series of steps to improve appropriate CRC screening. The first step concerns access to health care, including having health insurance and a regular source of primary care, as a necessary predisposing factor. The nature of CRC screening is that a physician (or nonphysician medical staff) must be involved in the decision to screen and in the completion of testing.

After access to care, the second step is to find ways for all patients to engage in a discussion at some level. The design and intensity of the discussion will depend on the patient’s prior understanding of CRC screening and the health care system. Such discussions take place with a trained health educator (e.g., physician or nonphysician staff), perhaps with an effective and tested decision aid, and focus on the pros and cons of CRC screening and the various screening options open to the patient. This interaction would need to be different for people with special circumstances, such as lack of fluency in English or lack of acculturation to the United States, than for most patients.

The third step in this progression is to simplify procedures for completing CRC screening for everyone. This includes providing proactive assistance to people from disadvantaged groups to complete screening after the screening decision has been made.

The further implications of this review are related to the implications for interventions to increase appropriate CRC screening use and quality (including reducing underuse, overuse, and misuse) and to three cross-cutting themes that underlie our findings: access to CRC screening; communication about CRC screening; and the organization of CRC screening. These three issues are among the strongest, potentially modifiable barriers to improving the use and quality of CRC screening.

Interventions to improve screening. Although we found high strength of evidence and positive effects for patient reminders, one-on-one interactions, eliminating structural barriers, and overall system changes as interventions to improve screening, still not clear is whether any set of interventions would effectively increase appropriate screening rates to high levels across the country. The health system may or may not have the ability to implement these interventions on a broad scale within medical practices and for the general population. To implement and maintain these interventions properly, an effective monitoring and feedback system (KQ 5) is needed. These systems are not in place in most primary care practices or health care systems.

How to overcome the focus in US medical care on nonpreventive care, and how to overcome the time and cost barriers to implementing and maintaining the systems within busy primary care practices, is also unclear. For example, incentives to primary care practices for improving CRC screening rates may or may not work. Partly because of the lack of positive incentives and the required time and effort from primary care practices, the durability of interventions that initially seem successful may be questionable. Another important issue is how interventions to improve CRC screening integrate with other medical practice systems.

Finally, the cost-effectiveness of the sometimes intensive interventions to gain sometimes small increases in screening is also unknown. Until these more fundamental issues are resolved, the question of whether widespread implementation of any interventions will have a large, sustained effect at reasonable cost (including time and effort of the patient, the physician, and the medical practice) remains unanswered.

Access to CRC screening. A critical underlying issue in this literature is access to care, a necessary precursor to access to CRC screening. Among the more striking findings from our review of factors associated with lower rates of CRC screening (KQ 2) is that people without health insurance, people with no source of usual care, people with no recent physician visits, and people with lower income status have quite low CRC screening rates. Improved communication and medical care organization can be effective only for people who are connected to a primary care provider.

Communication about CRC screening. One positive finding of this report is the overall importance of communication specific to CRC screening between medical staff and patients in improving appropriate CRC screening (i.e., reducing underuse, overuse, and misuse). CRC screening requires a great deal of patient understanding and effort (e.g., knowing which tests to take and when, and how to get them done). Communicating such information to patients and guiding them in making decisions specific to their medical and family history all take time. To make appropriate decisions about individually optimal screening, to carry out the preparation and follow-through correctly, and to obtain screening at recommended intervals all require patient knowledge, motivation, and assistance from medical personnel. When few CRC discussions take place (KQ 1), when many eligible patients do not know that they should be screened (KQ 2), when medical personnel make few recommendations for screening (KQ 2), when many people do not receive periodic health examinations (at which some time might be devoted to discussions of CRC screening [KQ 2]), and when few intensive one-on-one interventions exist to assist patients to decide, prepare, and follow-through (KQ 3), suboptimal screening rates should not be surprising.

An instructive case study for the importance of communication is the situation of Hispanic and Asian populations in the United States, especially because these groups have low rates of CRC screening (KQ 2). Although access to care certainly accounts for some of the disparity in screening rates for Hispanics, even when studies adjust for access, multiple good-quality studies using national population-based data show that screening rates for Hispanics or Asians continue to remain below those of non-Hispanic whites.1,46,111,120,141,147,151 This finding suggests that other factors, such as language and cultural differences, are also likely to be important determinants of screening.

Determining whether differences in CRC screening test use are mediated primarily through differences in language or differences in cultural beliefs about health and prevention is challenging, since language use is often a central part of the definition of acculturation.121,149 Determining whether lower screening rates in Hispanics or Asians is driven mainly by cultural beliefs, by possible distrust of the medical health system, or by language is important. If the differences in screening test use reflect true differences in informed choices not to have screening based on culturally mediated values and preferences, then some of the difference in screening test use may be appropriate. However, accumulating evidence suggests that language and possibly literacy barriers contribute to lack of knowledge about the risk that CRC poses and about the potential benefits of screening. Poor communication, at the level of the health care system as a whole, at the community level, and at the level of the patient-physician interaction, clearly contributes to low CRC screening rates in racial and ethnic groups. Language and literacy barriers likely lead to even poorer communication among subpopulations that prefer to obtain health information in a language other than English.

Organization of CRC screening and monitoring. CRC screening in the United States requires the involvement of primary care physicians. Most receive no regular feedback on their CRC screening rates, as might occur in the Veterans Health Administration (VA) or other integrated health care system. Few medical practices involve nonphysician office staff in discussing CRC screening with patients; few reach out to patients who have not been screened or who miss screening appointments. As suggested by the VA’s success with CRC screening (KQ 1), by the association of use of nonphysician staff with higher CRC screening rates (KQ 2), and by randomized controlled trials (RCTs) of organizational change (KQ 3) to improve screening, organizational change supported by monitoring and feedback systems (KQ 5) could have a positive effect on screening. Nonetheless, drawing conclusions on how to reduce overuse and misuse will always be difficult without adequate monitoring of these outcomes.

A second important aspect of organization is external to the primary care practice. It involves coordination of various parts of the health care system involved in CRC screening. Because these parts of the health care system are often fragmented, barriers arise that patients must navigate to complete screening. These same barriers work against monitoring the progress of patients as they move through the system or even providing assistance to those who cannot surmount the obstacles. Finally, these barriers create problems for providing consistent and timely information to patients and for establishing systems to reduce overuse and misuse.

Limitations of this Review

Limitations of the Evidence Base

Reporting. Our ability to draw conclusions on the effectiveness of CRC screening interventions is limited by the relative paucity of detail on specific elements of the interventions. Studies inconsistently adhered to reporting standards such as STROBE218 and CONSORT,219 making critical appraisal of internal validity and assessment of applicability challenging. In particular, many studies did not report on the intensity of the intervention (e.g., the number and length of sessions and the time period of interaction with clients), the existence of protocols governing the intensity of intervention, or fidelity to such protocols. In addition, a number of studies used multiple components of interventions (e.g., reminders paired with one-on-one interactions) to increase CRC screening but only provided overall findings. Reporting findings in this way made discerning the incremental impact of each component difficult if not impossible to assess. CRC screening interventions represent an opportunity to translate effective interventions into a variety of clinical settings; the absence of information on fidelity limits their translation.

Heterogeneity of the interventions and the intervention sites. Categorizing the interventions was complicated by the heterogeneity of approaches, even for interventions that we eventually placed in the same category. The problem of classification was also complicated by the diversity of the sites in which the interventions occurred. In a sense, for example, an intervention that would be considered a “reminder” in one location might be considered a small media intervention in a different location.

Choice of appropriate comparators. The evidence base for interventions is marked by heterogeneity in comparators in addition to appreciable diversity in the CRC screening measure itself. Although appropriate comparators can and should differ by the specific outcomes being addressed, studies often did not justify the choice of comparator(s), either on its own merits or in relation to usual care. In most cases for studies included in KQ 3, investigators did not define “usual care”; this ambiguity hampers accurate interpretation of comparisons. For that reason, our conclusions regarding the effectiveness of CRC screening interventions are necessarily limited.

We also note that a potential Hawthorne effect may exist for studies comparing CRC screening interventions with usual care as opposed to a “sham” control. In cases involving comparisons of different types of CRC screening interventions, all interventions may receive a Hawthorne boost. This possibility makes distinguishing the different effects of the various interventions difficult.

Choice of appropriate outcomes. As with the comparators, we encountered problems assessing the studies for each key question because of the way researchers defined and operationalized CRC screening. For studies that examined factors related to screening (KQ 2), many investigators used different definitions for CRC screening, partly because of a national trend toward colonoscopy and away from FOBT and sigmoidoscopy during the period of this review. Some teams considered subjects screened if they had ever received one type of test; others were more precise in including both the test and the recommended timeframe in their calculation of up-to-date screenings; while others included any CRC test code in their analysis, regardless of whether the test was provided as a screening or diagnostic procedure, thereby increasing the challenge in determining which factors were truly related to screening.

Even with these variations in the guidelines, how researchers operationalized the outcome of being up to date was inconsistent in this body of literature. Thus, assessing both the effectiveness of interventions (KQ 3) and the factors associated with screening (KQ 2) was overly challenging, particularly with regard to assessing the appropriateness of screening.

KQ 4 also presented challenges in assessing capacity outcomes and systematically applying these to a wide range of modeling assumptions; these problems in turn made synthesizing the findings difficult. Studies pertaining to this topic were also inconsistent in how they considered the difference between screening and diagnostic colonoscopy. Some modeling studies were unclear about whether they included surveillance colonoscopy in their calculation that would result from increased number of screening tests.

Across KQs 2, 3, and 4, we observed a heavy reliance on self-reported data that are not verified through other sources; this problem, too, complicates drawing reliable conclusions. Questions to assess self-report were not standardized, despite an NCI-led effort to develop standardized survey questions that have been subsequently evaluated in validation studies.211–214

Study design and sample size. Most KQ 2 studies were cross-sectional rather than cohort designs; thus, we could not easily examine time relationships. For this reason, there were no studies that examined factors associated with appropriate annual or serial use of FOBT, for example, and only focused on one-time or current use. Many studies did not report a priori hypotheses about their primary outcomes. Limited sample sizes resulted in studies that were not powered to find differences between experimental and control or comparison groups when such differences might in fact have existed.

In addition, the time periods in which investigators followed patients during intervention implementation for KQ 3 studies or measured capacity for KQ 4 studies varied considerably. Sometimes time frames were not specified at all. Again, these deficiencies hampered our ability to draw any conclusions across the studies. For KQ 3 studies in particular, time periods for following patients ranged from 3 to 24 months. Establishing a more common time for followup would improve the overall strength of evidence for these studies.

Appropriate adjustment for confounding. The evidence base is also limited by variations in the specific confounders and effect modifiers that investigators included or controlled for in their analyses. This issue arose particularly for examining factors influencing screening (KQ 2) and for quantifying capacity and projected demand (KQ 4). Omitting important confounders and effect modifiers (e.g., patient factors known to impact test use, temporal factors such as large macro-media campaigns such as when Katie Couric had a colonoscopy on national television), especially cointerventions in comparison arms, limits the interpretability and utility of the evidence from such investigations. Furthermore, using the studies that did account for confounders and effect modifiers is hampered by the lack of consistent definition and inclusion of key variables.

These deficiencies together appreciably limit the consistency and validity of the evidence. As a result, we found several bodies of evidence for important outcomes that we could grade only as low strength of evidence.

Limitations of the Review

We limited our search to articles published in English, primarily because the focus of this review was the United States. Issues of the use and quality of CRC screening likely vary by country. Our review does not address the nature, outcomes, or interventions developed in other countries. We excluded RCTs with samples sizes less than 30 and observational studies with samples sizes less than 100. We also limited the studies to those reporting on data collected from January 1998.

For time and resource reasons, we did not conduct dual independent, blinded review of articles for abstraction of information into evidence tables. Instead, one reviewer performed the initial review, and a second reviewer examined that input and recommended changes or corrections when needed. These two reviewers reconciled any differences by consensus discussion. We did apply dual independent review for assessing the quality of individual articles and grading the strength of evidence, and often involved a third team member to resolve disagreements about these issues. These are, generally speaking, standard approaches for the RTI-UNC EPC.

The paucity of similar articles—taking populations, patient characteristics, settings, and the heterogeneity and complexity of the interventions and the outcomes measured—precluded any efforts to pool findings statistically.

Future Research Directions

The last key question (KQ 6) is to assess “What research is needed to make the most progress and have the greatest public health impact in promoting the appropriate use of colorectal cancer screening?” We found numerous gaps in the available research that could be addressed to help us better understand and influence CRC screening rates. We summarize our suggestions for future research in Table 56.

Table 56. Suggested research agenda to improve the appropriate use and quality of CRC screening (priority areas in bold font).

Table 56

Suggested research agenda to improve the appropriate use and quality of CRC screening (priority areas in bold font).

The priority for research should be RCTs of interventions to implement appropriate CRC screening (i.e., minimizing underuse, overuse, and/or misuse) and monitoring linked to improvement initiatives. In our review, we became aware of multiple studies of the operating characteristics of potential new CRC tests. Although improving screening tests is a reasonable research agenda (especially finding ways to reduce the need for the most invasive and expensive tests), screening could be balanced with research to find ways to implement screening programs that we already know are effective, working to minimize underuse, overuse, and misuse. To focus research primarily on developing newer screening tests without placing higher priority on implementation of the existing effective tests leaves the people of the United States with inadequate screening. At least as important as newer screening tests are improved access, improved communication, and improved organization. We present in this review results of the uses of CRC tests within the VA system, where access to health care and insurance coverage are addressed by being members of that system, demonstrating that in this system, the use of screening is greater than among the general public. We found that rates among respondents in a nationally representative sample of respondents in the 2005 National Health Interview Survey (NHIS) who reported being covered by military insurance were statistically significantly more likely to have been screened when compared to all other insured and uninsured respondents (31.6 percent of those without insurance versus as high as 43.0 percent among insured respondents, compared to 67.9 percent among those with military insurance [P < 0.0001]).46 These findings indicate that when an organization is designed to provide screening to a population with consistent access to care, CRC screening rates can increase to levels seen for breast cancer screening with mammography.

Not only must we understand the organizational and system features important to increasing screening, but research also needs to examine the effectiveness of strategies that target several of the screening steps discussed above. Only when all three steps are accomplished—access to primary care, discussion and recommendation, and providing assistance and reducing barriers to complete screening—would we expect screening rates to markedly improve. The interventions of client or patient reminders, one-on-one interactions, and interventions to eliminate structural barriers seem to hold promise in increasing screening. Their impact could be increased if combined with further interventions to assist patients in traversing the health care system to complete screening. Patient reminders were an effective intervention in increasing cancer screening rates (including CRC screening) in a 2002 meta-analysis.220 In that meta-analysis, organizational change (such as the use of separate prevention clinics, use of a planned prevention visit, designation of nonphysician staff to do specific preventive care activities) was the most potent intervention in increasing preventive care.220 This study suggests that a combination of interventions may have the greatest impact on screening rates.

Interventions should be tested that work to optimize CRC screening together with other appropriate screening programs. Some of these interventions could target clinicians. We included two studies that examined the impact of provider-level interventions (for screening186 or surveillance colonoscopy188). Considering the central place that clinicians and their staff have in the screening steps, this is a potentially promising target to improve screening rates, particularly if it increases discussions between patients and providers.

In addition, cost-effectiveness studies of successful interventions to improve screening and monitoring, and then pragmatic trials that are focused on implementation of successful strategies within actual primary care practice are urgently needed. Different intensities of interventions, and even wholly different interventions, will likely be needed for different populations. Interventions should be targeted at the specific steps that are problems for specific populations (e.g., those who speak other languages than English at home could likely benefit from more basic interventions to increase awareness and discussions, whereas those who are already obtaining screening on an irregular basis may benefit most from patient reminders).

Further, we also need continued research into measuring current volume and projected demand for screening strategies. Finally, we found little evidence that adequate monitoring systems that assess the full spectrum of appropriate CRC screening (including overuse, underuse, and misuse) are in widespread use, and are being used to improve screening. Such monitoring systems are critically important for continued improvement of CRC screening. There is clearly a large and important research agenda for the future.

This research should target more than overcoming the underuse of CRC screening, as important as that is. We found little research interest in reducing underuse of patient-physician discussions about CRC screening, or in reducing overuse and misuse of CRC screening. This research should be a priority in that the issues of high overuse and high misuse are prevalent in today’s US health care.

Conclusions

Our review suggests that the United States is yet some distance from fully realizing the promise of appropriate and high-quality CRC screening. Problems of underuse, overuse, and misuse are not being adequately addressed at present. By focusing our research effort on the issues that matter most—access to screening, communication between patient and medical staff, the organization of care—and by further researching how to implement effective and cost-effective strategies into actual primary care practice, we will have the greatest opportunity to reduce the burden of suffering of CRC for the people of the United States.