3Overview of Trends in Use and Quality of CRC Screening

Trends in Incidence and Mortality from Colorectal Cancer

Colorectal cancer is the third most common nonskin cancer among men and among women; an estimated 146,970 people in the United States were newly diagnosed with this disease in 2009.³² The overall age-adjusted incidence rate for CRC has decreased in both men and women and in all ethnic groups since the mid-1980s, with an overall 3 percent annual decline between 1998 and 2005.³³ CRC incidence is higher among non-Hispanic blacks than among non-Hispanic whites; it is lower among Asian-Pacific Islanders and Hispanics than among non-Hispanic whites.

Colorectal cancer is the also the third-highest cause of cancer death among men and women; an estimated 49,920 deaths were attributed to this disease in 2009 in the United States.³² The overall age-adjusted mortality rate from CRC has decreased in both men and women since the mid-1980s; the annual percent decline between 2002 and 2005 was 4.3 percent.³³ CRC mortality rates declined for non-Hispanic black, non-Hispanic white, and Asian-Pacific Islander men and women. The rates dropped for Hispanic men but not for Hispanic women.³³ CRC mortality is higher in non-Hispanic blacks than non-Hispanic whites; it is lower in Asian-Pacific Islanders and Hispanics than in non-Hispanic whites. The gap in CRC mortality between non-Hispanic blacks and non-Hispanic whites did not change between 1997 and 2005.³⁴

Measures of CRC Screening

Several approaches have been used for measuring the percentage of a population that is up to date on CRC screening according to the national guidelines. Research studies of this question have most often used patient self-reports, but administrative databases, medical record reviews, and physician reports have also been used. A field study for the National Committee on Quality Assurance (NCQA) compared three different measurement approaches for assessing rates of CRC screening: patient surveys, administrative datasets, and a hybrid approach that performed medical record review for patients who did not have evidence of screening by administrative data.³⁵ Among the five health plans examined in the NCQA study, two did not show much difference between administrative and hybrid approaches, but the other three plans had 5 percent to 15 percent higher rates by the hybrid approach than by administrative data alone.³⁵

In all five plans, patient surveys (surveys patterned on the standard questions used by the Behavioral Risk Factor Surveillance System [BRFSS] from the Centers for Disease Control and Prevention [CDC]) gave screening rates higher than the hybrid and administrative approaches; the differences ranged from 2.4 percent to 23.3 percent for the hybrid approach and from 7.9 percent to 34.8 percent for the administrative approach. The differences between the survey and administrative approaches were lower for fecal occult blood test (FOBT) screening (difference ranged from 0.4 percent to 11.3 percent) than for flexible sigmoidoscopy (FS) and colonoscopy (difference for colonoscopy ranged from 7.1 percent to 26.9 percent).

One major reason for higher estimates from surveys is that nonrespondents are likely to have had less screening over time than respondents. Thus, one would expect that surveys would overestimate screening when response rates are low. When response rates are high, other studies have found a smaller degree of overestimation of screening rates, although some overestimation is still present.^36–38 Other studies have found that self-report overestimates screening rates more with FOBT than with colonoscopy.^39–40 Ultimately, because of changes in guidelines, as well as how questions are asked and current use is operationalized, measures of CRC screening have been challenging to standardize. For this reason, drawing valid conclusions on use is problematic.

Changes in Medicare Coverage of CRC Screening

In January 1998, the Centers for Medicare and Medicaid Services started covering CRC screening for Medicare beneficiaries; the tests included FOBT and FS as recommended by the American Cancer Society (ACS). On July 1, 2001, Medicare extended coverage for screening to colonoscopy every 10 years.

Changes Over Time in National Surveys of Screening

We found reports of screening rates from large, national surveys in two major sources: the National Health Interview Survey (NHIS), administered by the National Center for Health Statistics (NCHS), and BRFSS. NHIS is a personal household interview that contains a core set of questions plus additional supplements on specific topics. The CRC screening questions were revised in the late 1990s. ⁴¹ Before 2000, the NHIS did not distinguish between home and office FOBT and did not distinguish among endoscopic tests (e.g., proctoscopy, FS, colonoscopy). In addition, the 2000 NHIS asked about screening longer than 3 years before the survey. ⁴² Thus, screening rates before 2000 included some number of office FOBTs within the previous 1 or 2 years and proctoscopy (as well as FS and colonoscopy) within the previous 3 years. Starting in 2000, up-to-date screening is defined as home FOBT within the previous year, FS within the previous 5 years, or colonoscopy within the previous 10 years. The earlier rates from NHIS are thus likely an overestimate of the actual screening rates at the time (because of including in-office FOBT and proctoscopy, and how questions were asked) compared with rates starting in 2000. Also, since respondents had been asked about endoscopy use in the past 3 years only, this rate could be an underestimate of screening for these tests. NHIS interviewers read test descriptions to all eligible respondents for the first time in 2003. ⁴³

BRFSS is a national, random-digit-dial telephone survey administered in the United States to respondents 18 and older. BRFSS asked about FS and proctoscopy (not distinguishing between them) until 1999, when the question was changed to ask about FS or colonoscopy (again, not distinguishing between them). Before 2001, BRFSS did not allow for screening intervals longer than 5 years. Thus, BRFSS estimates before 2001 are for FOBT within the past year or lower endsocopy within the past 5 years. Starting in 2001, most estimates are for FOBT in the past year or lower endoscopy within the previous 10 years.

BRFSS response rates vary by state. For 1997, the overall median response rate by state was 62.1 percent, in 1999 it was 55.2 percent, and in 2001 it was 51.1 percent (range 33.3 percent to 81.5 percent). In 2002, the median response rate was 58.3 percent; in 2004 it was 52.7 percent, and in 2006 it was 51.4 percent. Thus, not all state estimates have the same validity.⁴⁴ About 3 percent of respondents were eliminated from the 2002 and 2004 analyses because they refused to answer or did not know the answer; in 2006, 4.5 percent were eliminated.⁴⁵

Estimates from NHIS and BRFSS

In 1987 by NHIS data (Table 3), 22 percent of men and 24.2 percent of women had had an FOBT within the previous 2 years or FS, proctoscopy, or colonoscopy within 3 years.⁴¹ For women, these screening rates increased to 28.2 percent in 1992 and 30.2 percent in 1998. For men, rates increased to 29.4 percent in 1992 and 37.1 percent in 1998.⁴¹ In 2000 (using the more restrictive definition of screening), 37.1 percent of both men and women had had at least one of these tests.⁴² In 2003, NHIS found that 46.5 percent of men and 43.1 percent of women had been screened⁴³ and in 2005, 50.0 percent of both men and women had been screened.⁴⁶

Table 3

Trends in screening according to the National Health Interview Survey.

The 1997 BRFSS (Table 4) found that 41 percent of respondents ages 50 years and older had had either an FOBT in the previous year or lower endoscopy (either FS or proctoscopy) in the previous 5 years. In 1999, this percentage had increased to 44 percent.⁴⁷ The 2001 BRFSS found that 53.1 percent of people in this age group reported having either an FOBT within the previous year or lower endoscopy (either FS or colonoscopy) within the previous 10 years.⁴⁴ In 2002, this percentage was 53.9 percent; in 2004, it was 56.8 percent and in 2006, it was 60.8 percent.⁴⁵

Table 4

Trends in CRC screening according to the Behavioral Risk Factor Surveillance Survey.

Population Subgroups

The changes in definitions of tests and testing intervals noted above cloud the data concerning CRC screening rates among population subgroups, including racial, ethnic, age, sex or gender, income, and educational groups. One BRFSS study used common coding and standard definitions over the years 2002 to 2006 for the data in Table 5.⁴⁵ Although the absolute percentages here are slightly higher than those from the NHIS (partly because of higher response rates in NHIS and the use of telephone rather than in-person interviews), the trends are the same in both surveys. Higher overall absolute screening rates are seen in older versus younger people, in white versus black populations, and in non-Hispanic versus Hispanic people. Higher education, higher income, and health insurance coverage are also associated with higher screening rates.

Table 5

Percentage of respondents 50 years of age or order who reported receiving a fecal occult blood test within 1 year and/or a lower endoscopy within 10 years, by selected characteristics—BRFSS, United States, 2002, 2004, and 2006.

Medical Practice Rates

Several studies provided information about CRC screening rates in medical practices, although we found no practice with uniform methods that could provide trend data over time. One chart review study of a sample of 12 diverse primary care practices in Michigan in 2003 found that CRC screening rates varied from 24 percent to 60 percent of eligible patients being up to date (FOBT in the past year, FS in the previous 5 years, or colonoscopy in the previous 10 years).⁴⁸ Another study examined CRC screening for 21,833 patients who were continuous members of an integrated health plan in the Midwest for the 5-year period ending December 31, 2003. Using automated records, the authors classified 54 percent of patients as being up to date for CRC screening (having received at least three FOBT kits, one FS, one colonoscopy, or one barium enema over that period).⁴⁹

Frequency of Discussions about CRC Screening

We found no trend data about this topic, but we did find several relevant articles. One study from 1998 to 2006 in southern California collected data from surveys with 191 physicians and 5,978 patients, asking about previous screening and discussions about several conditions, including FOBT and FS.⁵⁰ In this study, 37 percent of patients had discussed FOBT with their physician and 31 percent had discussed FS.

A second study audiotaped interactions between patients of the Veterans Health Administration (VA) eligible for CRC screening and their physicians.⁵¹ The study defined nine elements of informed decisionmaking and scored the occurrence of each element in 91 audiotapes of patients who had a CRC screening test ordered during that visit. Informed decisionmaking elements included such issues as discussion of the patient’s role in decisionmaking, discussion of alternatives, discussion of uncertainties, assessment of patient understanding, and asking for patient preferences. The median number of elements addressed was 1. No single element was addressed in more than 50 percent of interactions. Only 6 percent of interactions discussed uncertainties or patient understanding. A telephone and in-person survey asked 65 academic and community primary care physicians to present CRC screening to the investigator as if the investigator were a patient.⁵² Only 33.8 percent of respondents discussed the patient’s role in the decision, 16.9 percent discussed benefits and risks of screening strategies, and 10.8 percent provided alternative screening strategies.

A 2005 survey asked 270 primary care physicians connected with Northwestern University Feinberg School of Medicine to rate the importance of various general communication tasks relevant to CRC screening and to report how often they accomplish those tasks with screening-eligible patients.⁵³ Talking with patients was rated 9.5 out of 10 in importance; physicians reported that they accomplished this with 84.4 percent of patients. Discussing colonoscopy was rated 9.2; physicians reported accomplishing this with 84.8 percent of patients. Explaining test benefits was rated 9.0; physicians reported that they accomplished this for 79.3 percent of patients. Explaining test risks was rated 8.1; physicians reported this behavior for 63 percent of eligible patients. Eliciting patient views or preferences was rated 8.0; physicians reported accomplishing this for 65.7 percent of patients. Presenting more than one option was rated only 6.4 on the same scale and discussing FOBT was rated as 5.0; physicians reported accomplishing an FOBT discussion with 54 percent of eligible patients.

This same study also examined videotapes from an existing dataset of primary care encounters.⁵³ The authors found 18 videotaped encounters from a database of 271 interactions with patients’ ages 49 to 80 years in which the physician discussed CRC screening for the first time. Two authors viewed each videotape to determine to what extent physicians achieved the tasks they rated in the survey above. The benefits of the screening test were described in 28 percent of encounters; the risks were described in 0 percent of the encounters. In 28 percent of videotaped encounters in which CRC screening was discussed, physicians elicited patient views or preferences for CRC screening.

A survey of 2,501 patients of an integrated health care delivery system in southeastern Michigan who were continuously enrolled from 1999 to 2003 was able to link patients’ responses to an automated health record system to determine CRC screening over the 5-year study period.⁵⁴ Only 54 percent of this cohort was screened during the 5 years. About 80 percent of respondents (50.4 percent response rate) reported having a discussion with their physician about CRC screening. Of those having a discussion, 71 percent reported discussing colonoscopy and 41 percent FOBT. About 66 percent of patients reported that their physician discussed the pros and cons of different tests; 33 percent said that they had been asked about their preference for different types of tests and 39 percent were offered a choice among available tests. The association between those who had been offered a choice and receipt of a CRC screening test was negative; in this case, being offered a choice was associated with a lower screening rate. The usual length of these discussions and the relationship between patient report and actual discussion was not reported.

In this report, we distinguish between discussions of CRC screening between physicians and patients (covering such areas as pros and cons of screening options and eliciting patient preferences) as opposed to a simple physician recommendation of CRC screening (which is discussed in KQ 2). Although discussion and recommendation are not the same, recommendation would likely be a part of most discussions of CRC screening between physician and patient. Patient awareness of CRC screening is another likely result of CRC discussions. When there has been no physician recommendation and when patients are unaware of CRC screening, it is likely that there have been no discussions. Thus, lack of awareness and lack of a physician recommendation are two of the more frequent reasons that people who have not been screened give for not having obtained such tests.^21,55–57

Test-Specific Trends

Over time, the percentage of eligible people screened with FOBT and FS has declined while the percentage screened with colonoscopy has increased. For example, the proportion of BRFSS respondents who had had an FOBT within 1 year declined from 2002 to 2006: 21.6 percent in 2002, 18.5 percent in 2004, 16.2 percent in 2006. The percentage who had had a lower endoscopy (either FS or colonoscopy) in the previous 10 years increased over the same period: 44.8 percent in 2002, 50.1 percent in 2004, and 55.7 percent in 2006.⁴⁵

One national study examined the Medicare administrative database to determine trends in the use of various CRC screening tests between 1995 and 2003. Medicare started reimbursing for screening colonoscopy on July 1, 2001.⁵⁸ In 1995, 18.0 percent of Medicare beneficiaries received FOBT; in 2003, the figure was 14.3 percent. The percentage of people who received FS in 1995 was 3.9 percent, decreasing to 1.2 percent in 2003. The rate for colonoscopy, by contrast, rose: in 1995, 3.9 percent of Medicare beneficiaries received colonoscopy; in 2003, the figure was 9.4 percent. The relative decline in FS and the relative increase in colonoscopy was greater in white patients than in nonwhite patients. These changes were most pronounced after July 2001. These percentages are for screening received within a 1-year period, rather than the percentage of people who are up to date. A second analysis examined the test-specific trends within the Medicare population from 1998 to 2005, with similar findings.⁵⁹

Other studies using information from the administrative databases of health plans or large gastroenterology practices have also found increased use of screening colonoscopy after July 2001.^49,60–62

In an important study of trends in specific CRC screening test use between 1992 and 2002 in the Medicare population, use of FS increased from a mean rate per calendar-year quarter per 100,000 beneficiaries of 570.6 in 1996–1997 to 691.9 in 1999–2000 (after it was covered by Medicare in 1998) and then decreased to 267.5 in 2002–2003, after colonoscopy coverage started in 2001.⁶³ Colonoscopy use, by contrast, increased from a mean rate per quarter per 100,000 beneficiaries of 284.6 in 1996–1997 to 1,918.9 in 2002–2003. This study also found that the percentage of CRCs diagnosed at an early stage rose for proximal but not distal cancers after 2001, indicating the effect of colonoscopy in detecting proximal cancers. Even with this increase in screening associated with Medicare reimbursement, many Medicare beneficiaries remained unscreened.

A study of CRC screening test use from 1998 to 2003 in the VA system, in which physicians have no financial incentives to perform colonoscopy, found an increase in overall screening, driven primarily by an increased number of FOBTs.⁶⁴ FOBT as a proportion of all screening tests increased from 81.7 percent to 90.4 percent over the study period while screening colonoscopy declined from 5.7 percent to 4.7 percent and FS declined from 8.3 percent to 3.6 percent. A 2007 study of 17,252 patients in the Western Region Tricare Insurance system of the Department of Defense found that 71 percent of these beneficiaries were up to date with standard CRC screening guidelines, and 83 percent of those who were up to date had had a colonoscopy within the previous 10 years.⁶⁵

Trends toward screening colonoscopy may be less pronounced among disadvantaged groups than among the more advantaged. Although disadvantaged people (e.g., those without health insurance) are less up to date with screening, those who are screened may be more likely to be screened with FOBT than colonoscopy. One study conducted telephone interviews with 570 users of private physician offices (3 percent without insurance) and 500 registrants of county health centers (44 percent without insurance) in a single geographic area of New York State. Fifty-four percent of users of private physician offices and 28 percent of county health center registrants had had colonoscopy within the previous 10 years, while more county health center registrants had had an FOBT in the past year (31 percent private physician users versus 37 percent county health center registrants). Seventy percent of the private physician users and 55 percent of county health center registrants were up to date with national guidelines for CRC screening.⁶⁶

Beyond the United States, the International Colorectal Cancer Screening Network surveyed CRC screening programs that started before May 2004.⁶⁷ They found 10 organized CRC screening programs in seven countries. Of these, five used FOBT only, three used FS only, one used FOBT and FS, and one offered colonoscopy only. The program offering only colonoscopy was in Poland; the United States was not listed as having an organized program. The FOBT programs were split between gFOBT and iFOBT. A variety of pilot programs and research initiatives were also listed.

Patient Preferences for CRC Screening Tests

We found several studies that asked people about their preferences for CRC screening tests. In general, the studies found diversity of opinion, with some people preferring colonoscopy (often because of its accuracy) and others favoring FOBT (often to avoid the discomfort and inconvenience of colonoscopy).

One study recruited 323 colonoscopy-naïve supermarket shoppers from a low-to-middle‐class neighborhood in Denver, Colorado.⁶⁸ About half of respondents were non-Hispanic white with most of the rest evenly split between African-Americans and Latinos. After a description of the tests, 53 percent preferred FOBT and 47 percent preferred colonoscopy. Another study recruited 212 primary care patients from the waiting rooms of 3 community health centers and one academic medical center.⁶⁹ Patients were divided nearly equally among white, African‐American, and Hispanic people. Of the guideline-recommended tests, 37 percent preferred colonoscopy, 31 percent FOBT, 15 percent barium enema, and 9 percent sigmoidoscopy. One further study recruited 4,042 people who were participating in a multi-center study (84 sites) comparing fecal DNA testing with FOBT and colonoscopy.⁷⁰ Eighty-nine percent of participants were white. The participants were asked to complete a questionnaire after completing all three study tests. When asked which test they preferred for routine testing, 45 percent selected the fecal DNA test, 32 percent FOBT, and 15 percent colonoscopy.

Geographic Differences

We found no data on trends about differences in CRC screening rates by geographic factors; we did find several relevant reports. Using 2001 BRFSS estimates, states varied dramatically in the percentage of people having had an FOBT within the previous 2 years and in the percentage of people ever having had FS or colonoscopy.⁷¹ For FOBT for white men, the rates ranged from 14.3 percent in Alabama to 43.7 percent in Vermont. For FS/colonoscopy for white men, the rates ranged from 33.5 percent in Oklahoma to 63.5 percent in Delaware. For FOBT for white women, the rates ranged from 11.6 percent in Alabama to 46.7 percent in North Carolina. For FS/colonoscopy for white women, the rates ranged from 38.3 percent in Kentucky to 62.1 percent in North Dakota.

For FOBT for black men, the rates ranged from 4.7 percent in Alabama to 48.6 percent in North Carolina. For FS/colonoscopy for black men, the rates ranged from 13.7 percent in Tennessee to 56.4 percent in California. For FOBT in black women, the rates ranged from 10.5 percent in Alabama to 43.3 percent in Massachusetts. For FS/colonoscopy in black women, the rates ranged from 35.6 percent in New York to 59.2 percent in Virginia.

The 2004 BRFSS found variation among the states in the percentage of respondents ages 50 years and older reporting having had either an FOBT within the previous year or lower endoscopy within the previous 10 years.⁷² Rates ranged from 47.9 percent in Mississippi to 68.2 percent in Minnesota.

Health System Rates

The VA has a performance measure from medical record review for screening for people ages 50 to 80 years (FOBT within the past year, FS within the past 5 years, or colonoscopy within the past 10 years). With respect to being up to date on CRC screening, 78 percent of patients were up to date in 2007 and 79 percent in 2008.⁷³ The VA system has annual CRC screening rates from 1996 to the present. A few representative years are the following: 1996: 34 percent; 2000: 68 percent; 2004: 72 percent; and 2006: 76 percent.

NCQA, for its Healthcare Effectiveness Data and Information Set (HEDIS) commercial plans and using the same definition for being up to date as the VA, reported for 2007 that 55.6 percent of patients were up to date. The HEDIS measure is calculated from administrative data followed by a chart review for patients with no evidence of screening. No HEDIS trend data were available to us.

Overuse of CRC Screening

Although most of the previous discussion concerns underuse of CRC screening, overuse is also a concern. The two aspects of overuse for which we found evidence in the literature are overuse in people who, because of severe comorbidity or advanced age, have little potential to benefit and overuse of surveillance colonoscopy. By surveillance colonoscopy, we are referring to colonoscopy for patients who have had a previous colonic polyp (and, usually, polypectomy).

Overuse among persons unlikely to benefit. We found no data concerning trends for overuse but did find several relevant reports. Overuse of CRC screening has been documented in three studies in the VA system, questioning whether some patients are being screened inappropriately.^74–76 Some patients are less likely than others to survive for the 5 to 10 years necessary to have a chance of benefit from screening. In one study, 18 percent of patients given an FOBT kit at a single VA facility had severe comorbidities.⁷⁶ In the other two studies, of multiple VA system sites, people with severe comorbidities were screened as often as people with no co-existing illnesses.^74–75

Recently, the US Preventive Services Task Force (USPSTF) recommended that people over age 75 not be screened routinely and that people over age 85 not be screened at all.³ Thus, screening people over age 85 may also be considered overuse of screening.

Overuse of surveillance colonoscopy. Another potential for overuse is the frequency of surveillance colonoscopy after polypectomy. A 1999–2000 survey of a nationally representative sample of 317 gastroenterologists and 125 general surgeons active in colonoscopy surveillance (response rate 83 percent) asked for their suggestions for surveillance colonoscopy for four clinical scenarios.⁷⁷ One scenario, the finding of a hyperplastic polyp, confers no additional CRC risk and requires no surveillance over routine screening. Yet 24 percent of gastroenterologists and 54 percent of general surgeons recommended surveillance colonoscopy, most of them at a frequency of 5 years or less. A second scenario, finding a single small adenoma less than 1.0 cm in size, is generally classified as a “low risk” situation, and the MSTF guideline is surveillance colonoscopy at 5 to 10 years.⁷⁸ Yet 52 percent of gastroenterologists and 77 percent of general surgeons recommended surveillance colonoscopy every 3 years or more often. The authors concluded that “these findings suggest considerable over-performance of surveillance colonoscopy.”⁷⁷ A similar study of primary care physicians found even more frequent recommendations for surveillance of low-risk patients.⁷⁹ A study of endoscopists’ recommendations for repeat colonoscopy in 10 primary care practices in Virginia and Maryland found that endoscopists often recommend colonoscopy more frequently than guidelines recommend.⁸⁰ The mean number of years in which repeat colonoscopy was recommended by endoscopists was 7.8 years following normal colonoscopy, 5.8 years following the finding of a hyperplastic polyp, and 4.4 years following the finding of 1 or 2 small adenomas.

An innovative followup study of 1,297 participants in the Polyp Prevention Trial (an RCT of a dietary intervention to prevent colorectal adenomas) found evidence of both underuse and overuse of surveillance colonoscopy. Among patients with high-risk adenomas (who, according to national guidelines, should receive surveillance in 3 years⁸¹), only 36 percent had received surveillance within 3 years, and only 65.2 percent had had a surveillance examination within 5 years. Among patients with low-risk adenomas (who should receive surveillance only between 5 and 10 years of initial screening), however, 39.7 percent had had a surveillance examination within 4 years.⁸²

Misuse Rates

We define misuse as performance of screening tests in such a way that benefits are reduced or harms are increased compared with optimal performance. “Optimal” performance is sometimes difficult to define. Thus, we provide frequencies for clearly suboptimal performance and harms that could be potentially reduced by improved procedures.

We found literature on three types of misuse regarding FOBT: use of in-office FOBT when the literature is clear that home FOBT is preferable, nonreturn of FOBT cards given to patients, and nonfollowup of positive FOBT results with a full colon examination. We also found literature on two types of misuse of colonoscopy: high rates of adverse events such as colonic perforation and bleeding and nondetection of important colonic lesions. We found little data concerning trends for these problems and thus present the current situation as documented in the literature.

Reliance on in-office FOBT is clearly a problem of misuse, substituting a less effective test for a more effective one.⁸³ A 1999–2000 national survey of primary care physicians found that 32.5 percent of physicians used in-office FOBT exclusively; another 41.2 percent used a combination of in-office and home-based FOBT.⁸⁴ Nearly one-third of patients in the 2000 NHIS who reported having an FOBT said that the only test they had had was an in-office FOBT.⁸⁴ Whether these percentages have changed after this study was done remains unclear.

Another type of misuse of CRC screening tests is nonreturn of FOBT cards given to patients. We found only one study concerning this issue, an RCT of an intervention to improve return of FOBT cards in a VA primary care clinic.⁸⁵ In the control (usual care) arm of this study, 51.3 percent of patients returned the FOBT cards they were given (mean time to return cards in this group was 143 days).

Still another type of misuse is nonfollowup of positive FOBT screening results. We found one study in an integrated health care system that examined trends between 1993 and 2005 in the percentage of positive FOBTs that were followed by a complete diagnostic examination within 1 year.⁸⁶ This percentage increased from between 57 percent and 64 percent in 1993–1996 to between 82 percent and 86 percent in 2000–2005. The authors noted the introduction during those periods of tracking systems and screening guidelines.

Other studies provided information about follow-up rates for positive FOBTs but not trends over time. Two studies from the VA (data from 2000–2002) have documented lack of followup of positive FOBTs. One study of national VA data found that 41 percent of patients with a positive FOBT had not received or been referred for a follow-up test (either colonoscopy or barium enema) within 6 months.⁸⁷ A second study at a single VA center examined chart reviews on 538 men who had had a positive FOBT. About 77 percent were referred to gastroenterology; only 44 percent underwent full colon examination within 12 months.⁸⁸ In a study of positive FOBTs (76 percent from a screening FOBT) in a large integrated health care system (data from 2004–2006), fewer than 10 percent of patients had no action taken; colonoscopy was completed in 62 percent within a year.⁸⁹ Three older single-institution studies^90–92 (one using 1986 data, one using 1998 data, and one using 1993 data) and one study of community medical practices (using 1994–1996 data)⁹³ examining positive FOBTs from screening programs found from 23 percent to 46 percent of patients had no follow-up colon evaluation.

A 1999–2000 survey of 182 health plans (52 percent response rate) by the National Cancer Institute found that only 41 percent of plans had any system for delivering and/or monitoring CRC screening use; 25 percent had a mechanism for reminding patients when they are due for screening; 16 percent had a system for reminding physicians when a patient is due. Fewer than 15 percent of plans monitored receipt of follow-up care after a positive FOBT.⁹⁴

Another form of misuse is a high rate of adverse events during or after colonoscopy. We found no data on trends for this topic, but we did find two important reports to highlight. One study examined the Medicare database to count adverse events requiring an emergency department visit or hospitalization within 30 days of a colonoscopy.²⁰ The risk of colonic perforation was about 0.6 per 1,000 colonoscopies. The risk of gastrointestinal bleeding or transfusions was 2.1 per 1,000 in a group that was screened and did not have a polypectomy and 8.7 per 1,000 in a group that had a polypectomy. Some patients also suffered a cardiovascular event within 30 days: 9.9 per 1,000 procedures in the screening but no polypectomy group and 23.4 per 1,000 in the polypectomy group. Adverse events increased with age; people over age 85 suffered more than twice as many adverse events as people ages 66 to 69. A systematic review that pooled US studies before January 2008 found a combined rate of serious complications of screening colonoscopy of 2.5 per 1,000 procedures, with 85 percent of the complications occurring in patients who had had a polypectomy.⁹⁵

Misuse of colonoscopy also includes lack of detection of important lesions. Studies have found that from 2.1 percent to 5.9 percent of people diagnosed with CRC had had a colonoscopy within 3 years of the cancer diagnosis,^96–98 raising the issue of missed cancers. One study of back-to-back colonoscopies done on the same day found that 6 percent of adenomas at least 1 cm in size and 13 percent of adenomas 6 to 9 mm in size were missed on the first colonoscopy.⁹⁹ Other studies of CRC found by short-term follow-up colonoscopy after previous colonoscopy have raised the same question.^81,100

One variable that has been studied to provide insight into important missed lesions at colonoscopy is the adenoma detection rate. Several studies have shown variation among endoscopists in this rate. One factor associated with adenoma detection rates at colonoscopy is withdrawal time, which is the time required for the endoscopist to withdraw the colonoscope after full insertion.^101–103 Although longer withdrawal times are associated with increased detection of advanced adenomas (i.e., adenomas greater than 1 cm in size, or with dysplastic or villous components), longer times are also associated with increased detection and removal of small, low-risk polyps of uncertain clinical importance. A follow-up study found that instituting a practice-wide policy of at least 8 minutes for withdrawal reduced variation in adenoma detection rates among endoscopists; specifically the new policy increased detection of any neoplasia from 23.5 percent to 34.7 percent and increased detection of advanced adenomas from 5.5 percent to 6.3 percent of subjects.¹⁰⁴ Thus, most of the increase in adenoma detection was due to detection of nonadvanced adenomas.

Another factor associated with lower adenoma detection rates is depth of insertion, in particular the percentage of colonoscopies in which the cecum was reached. One study used an Ontario, Canada, database to explore the percentage of colonoscopies that were coded as incomplete (i.e., did not reach the cecum), finding variation in incomplete rates.¹⁰⁵ Colonoscopies performed in a clinician’s office were more likely to be incomplete than ones performed in an academic center (24.6 percent versus 12.6 percent). The percentage of incomplete colonoscopies declined over time (18.9 percent in 1999 to 10 percent in 2003). Similar data are not available from the United States.