Systematic Review: Enhancing the Use and Quality of Colorectal Cancer Screening FREE

National guidelines from the U.S. Preventive Services Task Force (1) and the American Cancer Society, American College of Radiology, and U.S. Multi-Society Task Force on Colorectal Cancer (2) recommend screening persons at average risk for colorectal cancer (CRC). Although the guidelines have different approaches to screening, they both recommend discussing CRC screening with eligible persons and offering a range of screening options. However, for CRC screening to contribute to a reduction in CRC mortality without unreasonable harms and costs, it must be offered to persons who have a reasonable probability of net benefit and be conducted effectively and efficiently. We consider issues of use and quality under the terms underuse, overuse, and misuse(3). Underuse (not screening persons who have a reasonable probability of net benefit) of CRC screening and lack of discussions about CRC screening remain a clear problem (4 - 14). However, emerging evidence suggests that overuse (screening persons or using screening methods with little potential for net benefit) (15 - 23) and misuse (screening in ways that reduce net benefit) (24 - 46) are also important problems.

Self-reported screening rates from national surveys, which are probably overestimates of actual screening, have increased from less than 25% in the late 1980s to about 50% to 60% in 2005 to 2006 (4 - 9). The increased screening rate can be attributed entirely to increased use of screening colonoscopy after Medicare started reimbursing for colonoscopy in 2001; screening with fecal occult blood test (FOBT) and sigmoidoscopy decreased during this period (9,47 - 52). Although we found no national data on discussions between primary care physicians and patients about CRC screening, several studies agree that these discussions, when they occur at all, seldom include consideration of harms, patient preferences, and patient choice (10 - 14). Furthermore, when these discussions are held, physicians routinely recommend colonoscopy and rarely mention other tests (53 - 54).

In addition to underuse, screening can be overused when persons and populations who are unlikely to benefit (or may even be harmed) are screened or when persons are screened or have screening-related procedures too often. Overuse occurs in several ways. The first is screening persons older than 85 years (1) or screening persons with severe comorbid conditions. Several studies (15 - 17) show that persons with severe comorbid conditions are screened about as often as persons without comorbid conditions. A second type of overuse is postpolypectomy surveillance colonoscopy at intervals that are too frequent (18 - 23). A third type is polypectomy for small polyps that are 5 mm or less in size. Because removal of these polyps yields little benefit (55 - 57) and increases the risk for complications (24,37), it is unclear whether aggressive attempts to remove polyps (the current policy) are warranted or should be considered overuse.

Misuse is also well documented. This includes use of in-office rather than home FOBT (25 - 26); poor FOBT tracking; patients not being adequately encouraged to return FOBT cards (27); lack of adequate follow-up of positive FOBT results (28 - 34); and colonoscopy that does not reach the cecum (58), has too short a withdrawal time (43 - 46), misses important lesions, or has high rates of adverse events (22,38 - 42).

To address these issues, we conducted a systematic review to inform the 2010 National Institutes of Health (NIH) State-of-the-Science Conference on Enhancing the Use and Quality of Colorectal Cancer Screening. As part of our review, we aimed to summarize evidence on factors that influence CRC screening and on strategies that increase the appropriate use (that is, minimizing underuse, overuse, and misuse) of CRC screening and screening discussions.

We followed a standard protocol for the systematic review. A technical report that details methods and results for each research question that guided the review is available at www.ahrq.gov/downloads/pub/evidence/pdf/crcuse/crcuse.pdf(3).

We address 2 of 6 key questions developed by a panel from the National Cancer Institute and the Office of Medical Applications of Research of NIH.

1. What factors influence the use of CRC screening?

2. Which strategies increase the appropriate use of CRC screening and follow-up?

We searched MEDLINE, the Cochrane Library, and the Cochrane Central Register of Controlled Trials (supplemented by hand-searches) for English-language articles published from January 1998 to September 2009. We searched data sources by using Medical Subject Heading (MeSH) terms and keywords. The MeSH terms included colorectal neoplasms, colonoscopy, sigmoidoscopy (including flexible sigmoidoscopy), and major headings included mass screening. Keywords included stool test, FOBT, and DNA stool. Appendix Table 1 lists complete inclusion and exclusion criteria.

Two reviewers independently screened each title and abstract identified from our searches. If either judged the title and abstract as potentially eligible for inclusion, the full text was reviewed. Two reviewers independently evaluated each full-text article to determine whether it should be included. Disagreements were resolved by team consensus.

For included studies, 1 reviewer extracted data on design, participant characteristics, statistical analyses, and results by using a structured form. This reviewer also used standard, predefined criteria to assign an initial quality rating (that is, assessment of internal validity) (59). A second reviewer evaluated the data extraction for accuracy and independently assigned a quality rating. Differences in quality ratings were resolved through consensus involving a third reviewer. We excluded studies receiving poor-quality ratings from our analyses.

We used a descriptive approach to summarize study characteristics and outcomes for all research questions. For each intervention category, we synthesized the results qualitatively and graded the strength of evidence for each intervention category.

For key question 2 (strategies to increase the appropriate use of screening), 2 reviewers graded the overall strength of evidence for each intervention category on the basis of an approach devised for the Agency for Healthcare Research and Quality (AHRQ) Effective Health Care Program (59 - 60). Disagreements were discussed among study investigators to reach consensus. Appendix Table 2 defines the strength-of-evidence grades.

The AHRQ, in partnership with the NIH Office of Medical Applications of Research, suggested the initial questions. The AHRQ provided copyright release for this manuscript. Representatives of both the AHRQ and the Office of Medical Applications of Research were informed of key methodological decisions but did not participate in the literature search, determination of study eligibility criteria, data analysis or interpretation, preparation of the review, or interpretation of the results.

We screened 3029 titles and abstracts and evaluated 861 full-text articles. A total of 116 publications met inclusion criteria for the full report (3). Of these, 93 studies (96 articles) were included for the questions addressed in this article (Appendix Figure). We included 72 studies (74 articles) addressing the key question regarding factors associated with screening and 21 studies (22 articles) addressing the question regarding strategies to increase appropriate screening.

We categorized factors associated with the use of CRC screening tests into 5 domains: patient factors, physician factors (including physician characteristics, physician–patient connectedness, and physician recommendations about screening), patient–physician communication factors, the periodic health examination, and system factors. We further categorized the patient factors into 4 groups: demographic characteristics, access to care, personal health or risk factors, and psychosocial factors.

All but 2 studies focused on factors associated with underuse of CRC screening. Those 2 studies focused on misuse by examining factors associated with lack of follow-up of an abnormal FOBT result. None focused on factors associated with CRC screening discussions or overuse of CRC screening.

Factors associated with higher rates of CRC screening include having insurance (61 - 67), having regular access to care (61,63,65,68 - 70), having a physician recommendation for screening (61,63,67,71 - 78), having effective patient–provider communication (79 - 81), participating in regular screenings for other types of cancer (61,63,65,82 - 83), being non-Hispanic white (71,84 - 85), having a higher education level (61,63,65), having a higher household income, and having a family history of CRC or personal history of another type of cancer (61,63,68). Sociodemographic groups that were much less likely to be screened included U.S. Hispanic and Asian persons (65,71,84,86), persons less acculturated to the United States (86 - 93), and persons born outside the United States (13,16,18,20).

Although 1 included study examined the association between screening and periodic health examinations (94), several additional studies (68,76) provided adequate supporting evidence to conclude that persons who have periodic health examinations have increased CRC screening rates. System-level factors associated with CRC screening, such as having nonclinician support for screening, using reminder systems, and helping patients keep appointments, increase CRC screening rates within systems (95 - 100).

We classified strategies into those that targeted the patient, provider, or health system. We included 15 studies (101 - 115) that examined patient-level interventions, 2 studies (114,116) that targeted provider-level interventions, and 5 studies (6 articles) (117 - 122) that targeted elements of a particular health care system (some studies had more than 1 focus). A single study (105) examined a decision aid's effect on CRC screening discussions, and all studies focused on increasing screening. We found no controlled trials of either fair or good quality that tested interventions targeting a community or studies focusing on reducing overuse or misuse.

We divided the patient-level interventions into 6 categories: eliminating structural barriers, one-on-one interactions, patient reminders, small media (for example, print or video) without decision aids, small media with decision aids, and group education (Table 1). A total of 5 studies (101,103,107,111,113) reported on interventions that included components from more than 1 category. However, for most studies, findings were not presented in a way that allowed us to assess the independent effect of each component.

All studies of patient-level interventions focused on reducing underuse of CRC screening or follow-up after receiving a positive FOBT result (that is, misuse) (Table 1). All but 4 studies (106,111 - 113) enrolled patients of clinics or health plans with ready access to care. The baseline screening rates were usually not reported and were probably quite varied.

All studies of interventions to eliminate structural barriers (101,103,107,111,115) provided FOBT tests and instructions to use at home. One study (107) also addressed language and cultural barriers. These studies all showed increased screening rates (range of absolute increase, 15% to 42%). We graded the strength of evidence for this body of evidence as high.

Studies of one-on-one interactions included intensive in-person contact with patients by a nurse (101), by a health educator (107), or on the phone (108). These interventions all increased screening rates (range of absolute increase, 15% to 42%). We graded the strength of this body of evidence as high.

These studies (103,110 - 111,114) all used mailed reminders and targeted persons who were due for rescreening. They found small to moderate increases in CRC screening (range, 5% to 15%). We graded the strength of this body of evidence as high.

Studies of small-media interventions (for example, education print or video) either mailed educational materials to patients before an appointment or asked patients to review the materials while waiting for an appointment. These studies (102 - 104,113) showed no statistically significant increases in CRC screening. We graded the strength of this body of evidence as high.

Decision aids are materials that provide information on risks and benefits of screening and screening options through written materials, a video presentation, or an interactive Web site. Findings of studies examining the effect of interventions based on decision aids were mixed, with 2 of 3 studies (105 - 106) showing benefit (14% and 23%), and 1 study (109) showing no effect. We graded the strength of this body of evidence as low.

Group education interventions targeted members of churches (113) or civic clubs (112). Findings were mixed, with 1 study (112) showing a negative effect on screening (−7%), and another study (113) finding a small positive effect that was not statistically significant (12%). We graded the strength of this body of evidence as low.

One study (116) of a physician-reminder intervention provided letters to physicians of patients who were overdue for surveillance colonoscopy (Table 2). This study found a 5% increase in surveillance follow-up. Another study (114) provided electronic reminders to clinicians during a visit with a patient who was due for screening. This study found that the reminders had no effect. We graded the strength of this body of evidence as low.

Evidence from 5 studies (6 articles) (117 - 121) was available for evaluating various system-level interventions (for example, changes to improve referral of patients for screening or to provide a patient navigator or someone in a similar role) to help patients navigate the health care system (Table 2). These studies found consistently positive effects on screening (absolute increase in screening rates, 7% to 28%). We graded the strength of this body of evidence as high.

Our review confirmed that important problems exist in the underuse, overuse, and misuse of CRC screening. We also found underuse and low quality of discussions between patients and health professionals about CRC screening, even though national guideline groups (1 - 2) recommend such discussions. Most of the research we reviewed about CRC screening focused on the accuracy of various screening tests rather than improving the implementation of screening strategies that are known to be effective. Almost all of the research about implementation of CRC screening focused on increasing use, whereas minimal research has examined improving discussions about CRC screening or reducing overuse and misuse of CRC screening tests.

We found that the factors most strongly associated with CRC screening involved lack of general access to health care (such as having health insurance and a regular source of care). A special case of lack of access involves disadvantaged populations, including Hispanic and Asian persons, persons born outside the United States, and persons with limited English-language proficiency. These groups have additional vulnerabilities that contribute to screening disparities even after socioeconomic and access factors are accounted for.

On the basis of the review, we found that access to care alone is insufficient to guarantee appropriate levels of screening. Patients still need a simple, reliable mechanism by which to engage physicians and others in the health care system to understand the idea of screening as well as the pros and cons of different screening strategies and how to complete screening. Few health care systems build these discussions into routine care (36,123).

Even with access to care and some understanding of the importance of screening, the complexity of screening tests and the screening process also leads to suboptimum screening rates and quality. Few health care systems have organized programs to assist patients in completing appropriate screening, track the return of FOBT cards, complete endoscopy screening, and obtain appropriate follow-up of abnormal findings (36,123).

We found that unscreened patients often did not know about the need for CRC screening (63,65) and that knowledge of screening options other than colonoscopy was limited. Ideally, a recommendation for screening would be accompanied by a reasonable discussion of screening options, including both the expected benefits and the time, effort, costs, degree of discomfort, and risks associated with each recommended strategy. However, evidence (13 - 14,53 - 54) suggests that if provider–patient “discussions” happen at all, they often include a recommendation for colonoscopy. With several tests available for CRC screening, the lack of a mechanism to promote and assist patients in understanding screening options is a barrier to appropriate use.

Despite the challenges, we found good evidence that several interventions do improve appropriate CRC screening use for some populations. Eliminating structural barriers by such interventions as mailing FOBT cards to patients, providing one-on-one counseling by nonphysician staff to help persons understand CRC screening, and reminding patients when they need screening are effective approaches. Although simple “one-way” information brochures seem ineffective in improving CRC screening, limited evidence suggests that interactive decision aids may be useful in helping patients to understand the pros and cons of different screening strategies and to make informed decisions about which screening strategy is right for them.

Screening strategies for CRC require the ability to carry out specific instructions for preparation and testing, as well as the ability to navigate the medical system to complete screening. Because of this, one-on-one assistance with the logistics of screening is sometimes needed to reach high rates of appropriate screening. Such intensive approaches are probably not necessary for all populations, although reducing barriers and standardizing the screening process are likely to be helpful for all.

Although we found strong evidence of positive effects for several types of interventions, the capacity to actually deliver these interventions is hindered most fundamentally by policies that determine how health care is structured and paid for. It remains unclear whether any specific intervention or program of combined interventions would effectively increase screening rates across the country. Given the variation of practice models, practice sizes, degree of system integration, and availability of tests, our ability to implement organized screening programs and interventions on a broad scale within and across medical practices is uncertain at best. To implement and maintain such interventions properly, whether at a national level or even at a practice level, we need effective monitoring and feedback systems. Such systems seem to be available and linked to active CRC screening programs only in a few large, integrated health systems and a small fraction of individual practices.

Overcoming the time and cost barriers to implementing and maintaining screening systems in busy primary care practices will require a shift in focus away from short-term and episodic care. Such a shift will almost certainly require a change in provider-, practice-, and system-level incentives for implementing and integrating appropriate long-term and preventive care systems. Because of the misalignment in incentives and the required time and effort from primary care practices, the sustainability of interventions that may initially be successful is uncertain.

Our review has some limitations. First, selective reporting of positive results and publication bias may have influenced our findings and conclusions. Second, despite a vigorous attempt to identify relevant articles, it is possible that some were missed or inappropriately excluded. Third, we excluded randomized, controlled trials with samples sizes less than 30, observational studies with samples sizes less than 100, and studies conducted before January 1998. Inclusion of some or all of these studies could have influenced our findings.

Our review points to opportunities for future research. Perhaps most strikingly, we found a mismatch between the focus of most of the published literature (which was on the accuracy of various CRC screening tests) and the need to improve appropriate use and quality of CRC screening strategies, for which there is already good evidence of net benefit in a population. Although it is reasonable for research to examine the characteristics of newer screening tests (especially ones that lead to less invasive initial screening), pursuit of this goal may have distracted researchers from exploring how to implement screening strategies that we already know are effective. To focus research primarily on developing newer screening tests without placing higher priority on implementing the existing effective tests leaves patients with many newer tests but inadequate screening. Additional imbalances are the lack of research on the problems of overuse, including assessing the net benefit (benefits minus harms) of aggressive attempts to identify and remove all polyps (regardless of size), as well as the limited research on misuse of CRC screening.

Further research should examine the cost-effectiveness of strategies to implement and maintain durable CRC screening programs. However, our ability to develop these durable strategies is constrained by a lack of incentives for primary care practices to implement these programs. Programs must be cost-effective from a societal perspective but also provide a business case for individual practices to adopt and maintain the program.

Until these fundamental issues are addressed, widespread implementation of any interventions may not have a large, sustained effect at reasonable costs (including time and effort of the patient, physician, and medical practice). Therefore, another priority for future research should be pragmatic, generalizable trials of cost-effective interventions to implement appropriate CRC screening (that is, minimizing underuse, overuse, and misuse) and monitoring, which is linked to incentives to maintain effective interventions over time.

Focusing research efforts on the issues that matter most—access to screening, communication between patients and medical providers, and organization of care—and further researching how to implement effective and cost-effective strategies into primary care practice will afford us the greatest opportunity to reduce the burden of suffering from CRC.