|Home | About | Journals | Submit | Contact Us | Français|
Individuals with a negative screening colonoscopy are recommended to repeat colonoscopy in ten years.
To assess the effectiveness and costs of colonoscopy versus other rescreening strategies following a negative colonoscopy.
Literature and SEER.
50-year-olds who had no adenomas or cancer detected at screening colonoscopy.
No further screening and rescreening at age 60 with ten-yearly colonoscopy, yearly highly-sensitive guaiac fecal occult blood testing (HSFOBT), yearly fecal immunochemical testing (FIT), or five-yearly computed tomographic colonography (CTC).
Lifetime number of colorectal cancer cases, life expectancy, and lifetime costs per 1000 individuals assuming: 1) perfect adherence and 2) imperfect adherence.
Rescreening with any modality yielded sizable reductions in colorectal cancer risk compared to no further screening (range 7.7 to 12.6 lifetime cases per 1000 (perfect adherence) and 17.7 to 20.9 lifetime cases per 1000 (imperfect adherence), versus 31.3 lifetime cases per 1000 with no further screening). For both adherence scenarios, the differences in life-years across rescreening strategies were small (range 30,893 to 30,902 per 1000 (perfect adherence) and 30,865 to 30,869 per 1000 (imperfect adherence)). Compared to continuing colonoscopy, rescreening with HSFOBT, FIT, or CTC had fewer complications and was less costly.
Results were sensitive to test-specific adherence rates.
Limited data on adherence with rescreening.
Compared with the currently-recommended strategy of continuing ten-yearly colonoscopy after an initial negative exam, rescreening at age 60 with yearly HSFOBT, yearly FIT, or five-yearly CTC provides approximately the same benefit in life-years with fewer complications and at a lower cost. Therefore it is reasonable to rescreen individuals with a negative colonoscopy with other modalities.
Screening has been shown to reduce colorectal cancer incidence (1–3) and mortality (1–6). Screening rates have increased substantially over the past decade (7–8). Although alternative screening approaches are sanctioned by guidelines (9–10), much of the rise in screening was driven by increased use of colonoscopy (7).
Colonoscopy is a recommended method for routine colorectal cancer screening (9–10) and is used for follow-up of individuals with positive results on other screening tests (9) such as fecal occult blood tests (FOBTs), and for surveillance of individuals with a family or personal history of adenomas or colorectal cancer (11–13). Although safe, colonoscopy can cause complications (14–16) that in rare cases may be fatal (14–15, 17). Moreover, it requires considerable resources. Thus, strategic use of colonoscopy should be a priority for healthcare delivery.
Population-based registry and claims database studies have shown that the risk of developing colorectal cancer for individuals with a negative colonoscopy is substantially lower than the risk for those who are unscreened (18–19). This has prompted consideration of whether the guideline recommendation to repeat colonoscopy ten years following a negative exam (9–10) is indeed necessary. These guidelines (and mathematical models that have evaluated them) have assumed that people use only one screening test throughout their lifetimes. We assessed whether alternative rescreening strategies for individuals following a negative screening colonoscopy could maximize benefits and minimize costs and harms.
We used the Simulation Model of Colorectal Cancer (SimCRC) to evaluate management strategies for 50-year-olds with neither adenomas nor colorectal cancer detected at their first screening colonoscopy. SimCRC is one of the models in the National Cancer Institute-sponsored Cancer Intervention and Surveillance Modeling Network (CISNET) and has been used to inform the US Preventive Services Task Force's colorectal cancer screening guidelines (20) and the Centers for Medicare and Medicaid Services coverage decisions for stool deoxyribonucleic acid testing (21) and computed tomographic colonography (CTC) screening (22).
The SimCRC model is programmed in Microsoft Visual C++ 2010 Express (Seattle, Washington). The model specifications have been described (23–24). Briefly, the model's natural history component tracks the development of adenomas and their possible progression to invasive colorectal cancer in the absence of screening. An individual enters the model at birth and over time, may develop one or more adenomas (Figure 1). Adenomas may grow in size from small (1–5mm) to medium (6–9mm) to large (≥10mm); some may progress to preclinical colorectal cancer. Preclinical colorectal cancer may progress in stage (I–IV) and may be detected by symptoms. Relative survival following cancer diagnosis depends on age and tumor site and stage (25). Individuals may die from causes other than colorectal cancer at any age (26).
SimCRC's screening component allows adenoma(s) and/or preclinical colorectal cancer(s) to be detected based on the sensitivity of the screening test for lesion(s) of that type/size and, for endoscopic tests, the depth of endoscope insertion. Non-adenomatous polyps are not explicitly modeled but are reflected in test false-positive rates, allowing individuals to be referred for follow-up and undergo polypectomy for non-adenomatous polyps. We assume each detected adenoma is removed, thereby preventing its potential progression to colorectal cancer. Individuals with screen-detected colorectal cancer may face a lower risk of cancer death if it is detected at an earlier stage.
Because the natural history of colorectal cancer is largely unobserved, there are limited data to directly inform some model parameters. We inferred their values by calibrating the model to data on the prevalence, size, location and multiplicity of adenomas from autopsy studies (27–36) and the incidence of colorectal cancer from the Surveillance, Epidemiology, and End Results (SEER) Program (25). We used SEER data from 1975–1979 since colorectal cancer screening was rarely performed during this period. The calibration approach and fit of the model to these data are provided elsewhere (22–24).
We evaluated five rescreening strategies for individuals with a negative colonoscopy at age 50: no further screening; continuing ten-yearly colonoscopy; or rescreening with a yearly highly-sensitive guaiac FOBT (HSFOBT), yearly fecal immunochemical test (FIT), or five-yearly CTC. These strategies are guideline-sanctioned options for routine screening for 50-year-olds (9–10) and are therefore reasonable alternatives for rescreening individuals with a negative colonoscopy. Rescreening was assumed to begin at age 60 (that is, ten years after the negative colonoscopy) for all strategies.
We assumed individuals with a positive HSFOBT or FIT or a CTC indicating lesion(s) 6mm or larger were referred for follow-up colonoscopy. Due to the possibility of systematic positive HSFOBTs or FITs due, for example, to persistent gastrointestinal bleeding unrelated to adenomas or colorectal cancer, individuals with no adenomas or colorectal cancer detected at follow-up were assumed to discontinue HSFOBT or FIT and resume screening with ten-yearly colonoscopy; individuals with a positive CTC but no adenomas or colorectal cancer detected at follow-up were assumed to continue CTC screening. If an adenoma was detected and removed at colonoscopy, the individual began colonoscopy surveillance consistent with guidelines (13). We assumed screening of individuals with no history of adenomas or colorectal cancer ended after age 75 (10), but surveillance of individuals with a history of adenoma(s) continued for life.
Reliable estimates for adherence are limited, yet adherence rates may have a major impact on results. Therefore, we considered two adherence scenarios: perfect and imperfect. Perfect adherence meant 100% participation with each test. With imperfect adherence, adherence following the initial negative colonoscopy varied by test and incorporated within-subject correlation for adherence with rescreening (Appendix Table 1). Adherence with HSFOBT was based on Department of Veterans Affairs' data (37). Among men who exclusively received guaiac FOBT, 42% received one over a five-year period, 26% received two, 18% received three, and 14% received four or more. For FIT, we assumed per-test adherence was 24% higher than with HSFOBT, based on the relative increase in uptake with FIT vs. guaiac FOBT in a Dutch screening program (38). For the first rescreening colonoscopy we assumed 52% average adherence based on adherence with a five-year repeat colonoscopy among those with a negative initial exam (39). We further assumed that individuals had, on average, only one of the two recommended rescreening colonoscopies (at age 60 or 70). For each follow-up and surveillance colonoscopy we assumed 94% average adherence (2, 38). In the absence of data for CTC, we assumed the average chance of adhering with the first CTC was the same as for repeat colonoscopy (that is, 52%) and that individuals had, on average, two CTCs by age 75.
The sensitivities and specificity for each screening modality are shown in Table 1. We assumed that 5% of individuals would require two colonoscopies to achieve a complete exam and that the cecum was eventually examined in 95% of individuals. The model incorporated the risks of complications, including perforation, bleeding, and other gastrointestinal events (Table 2). We assumed there were 51.9 deaths per 1000 perforations (17).
Costs of screening tests (Table 1) and complications (Table 2) were based on 2007 national-average Medicare payments and beneficiary co-payments (assuming these payments were applicable to 50-64-year-olds) and patient time costs (see Appendix Table 2). Since Medicare does not currently reimburse for a screening CTC, we used the payment for a diagnostic study. The cost of bowel preparation was estimated at $23 (46). An hour of time was valued at the 2010 median hourly wage rate for civilians, $18 (45).
The stage and phase specific costs of colorectal cancer care (Table 2) were based on analyses of SEER-Medicare linked data. The analyses used the methodology reported by Yabroff (49), with stage reclassified using the American Joint Committee on Cancer staging algorithm and costs in the last year of life stratified by cause of death. The estimates incorporate patient time costs and copayments (50).
All costs were expressed in 2010 dollars and were inflation-adjusted as needed using the Consumer Price Index (51).
We used SimCRC to estimate the number of colorectal cancer cases and deaths, life-years, perforations and other complications, procedures requiring bowel preparation, and lifetime colorectal screening- and cancer-related costs for a hypothetical cohort of 50-year-olds with a negative screening colonoscopy under two adherence scenarios and five rescreening strategies. Outcomes were tallied from the time of the negative colonoscopy at age 50 until death. Costs were tallied from the societal perspective.
We performed sensitivity analyses on: colonoscopy test characteristics and cecal intubation rate; CTC test characteristics; colonoscopy complication rates; colonoscopy cost; costs of cancer care; and adherence rates (see Tables 1 and and22 and Appendix Table 1 for values used in the sensitivity analyses).
The National Cancer Institute funded this research. The funding source had no role in the design and conduct of the study; management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; or decision to submit the manuscript for publication. Drs. M. Brown and R. Yabroff of the National Cancer Institute provided the costs of colorectal cancer care used in the base-case analysis.
SimCRC predicts that 15% of those with a negative colonoscopy at age 50 would have adenomas or colorectal cancer detected by colonoscopy at age 60. The corresponding estimate for those with a positive colonoscopy at age 50 is 31%, assuming no surveillance colonoscopies are performed between ages 50 and 60.
With no further screening, 31.3 per 1000 50-year-olds with a negative screening colonoscopy would be diagnosed with colorectal cancer in their lifetimes and 11.9 per 1000 would die from the disease. Compared to no further screening, all rescreening strategies yielded sizable reductions in colorectal cancer risk. With perfect adherence, continuing ten-yearly colonoscopy screening yielded the fewest cancer cases (7.7 per 1000) and deaths (2.4 per 1000), but the largest number of perforations and other complications (1.1 and 20.9 per 1000, respectively) (Table 3). Rescreening with CTC yielded slightly more cases (9.3 per 1000) and deaths (2.7 per 1000) than continuing colonoscopy and nearly halved the rates of perforation and other complications (0.7 and 10.1 per 1000, respectively), but had the largest number of procedures requiring bowel preparation (3,982 per 1000 vs. 2,592 with colonoscopy). Rescreening with HSFOBT or FIT yielded 11.4 and 12.6 cases and 3.2 and 3.5 deaths per 1000 respectively, with complication rates similar to CTC. The number of procedures requiring bowel preparation was 1,557 per 1000 for HSFOBT and 1,282 per 1000 for FIT. All rescreening strategies yielded comparable life-years, ranging from 30,893 per 1000 for FIT to 30,902 per 1000 for colonoscopy (Table 3), a difference of 3 days per person.
With lifetime screening- and cancer-related costs of $3,840 per person, continuing colonoscopy screening was the most costly strategy (Table 3). Compared to continuing colonoscopy, cost-savings were $166 per person with CTC, $771 per person with HSFOBT, and $781 per person with FIT. Discounted life-years and costs are available in Appendix Table 3.
With imperfect adherence, continuing colonoscopy yielded the fewest colorectal cancer cases (17.7 per 1000), followed closely by switching to CTC (17.8 per 1000) (Table 3). Switching to CTC yielded the fewest colorectal cancer deaths (6.1), compared with 6.4 deaths with colonoscopy, 6.4 deaths with FIT, and 6.7 deaths with HSFOBT (all deaths are per 1000 with a negative colonoscopy at age 50). Continuing colonoscopy yielded the highest rate of perforation and other complications (0.6 and 11.0 per 1000 respectively). Rates for CTC were 0.4 and 5.6 per 1000 respectively, but CTC required more procedures with bowel preparation (2,135 per 1000 vs. 1,361 with colonoscopy). The FOBT strategies had perforation and complication risks similar to CTC (0.3 and 5.2–5.6 per 1000 respectively) and required fewer procedures with bowel preparation (626–672 per 1000). The differences in life-years across rescreening strategies were small, ranging from 30,865 per 1000 for HSFOBT to 30,869 per 1000 for CTC, a difference of 1 day per person.
All other strategies yielded lower screening- and cancer-related costs than continuing colonoscopy ($3,084 per person with a negative colonoscopy at age 50 (Table 3)), with cost-savings from switching from colonoscopy to CTC of $91 per person, to FIT $450 per person, and to HSFOBT $495 per person.
While the absolute number of life-years changed with assumptions about colonoscopy test characteristics and cecal intubation rate (Appendix Table 4), CTC test characteristics (Appendix Table 5), and colonoscopy risks (data not shown), the differences in life-years across rescreening strategies remained small, 4 days or fewer per person. The cost-savings from rescreening with a strategy other than colonoscopy fell.
Since colonoscopies are performed in all rescreening strategies, the lifetime costs of all strategies changed with colonoscopy cost. If colonoscopy cost was half the base-case estimate, all rescreening strategies, including continuing colonoscopy, yielded similar lifetime costs. As colonoscopy cost increased above the base-case estimate, the cost-savings from rescreening with methods other than colonoscopy increased (Appendix Figure 1). Due to the small number of cancer cases, the findings were relatively insensitive to higher costs of cancer care (Appendix Figure 2).
The results were sensitive to test-specific (imperfect) adherence rates. If adherence with each rescreening strategy was lower (higher) than the base-case, the life-years remained similar across tests (Figure 2, A). However, if adherence with some strategies was higher than the base-case values while adherence with others was lower, the differences in life-years across tests increased to a maximum of 40 years per 1000 (30,846 life-years per 1000 if continue colonoscopy with low adherence versus 30,886 if rescreen with CTC with high adherence), which is 15 days per person. In such a case, switching to CTC was no longer cost-saving (lifetime costs $2,737 per person if continue colonoscopy with low adherence versus $3,319 if rescreen with CTC with high adherence) (Figure 2, B).
Colonoscopy is a well-accepted strategy for prevention of colorectal cancer death (52) and efforts to promote its use have increased the proportion of Americans who report having had the procedure (7–8). However, the value of alternative rescreening strategies for those with a negative initial exam is uncertain. Ideally, a randomized trial would address this question, but such a study is unlikely to be performed. Results from a validated simulation model can therefore be informative.
Since it is debatable whether policy decisions and clinical recommendations should be informed by analyses that assume perfect adherence or those that incorporate more realistic, but poorly-described imperfect adherence rates, we evaluated both adherence scenarios. Notably, conclusions were similar across scenarios. Compared with the currently-recommended strategy of continuing ten-yearly colonoscopy after an initial negative exam, all of the other rescreening options we examined provide approximately the same benefit in life-years with fewer complications and at a lower cost. Therefore it is reasonable to rescreen individuals with a negative colonoscopy with other modalities.
Our findings have several implications. Colonoscopy has become the accepted standard for colorectal cancer screening in the US. However, there are not enough trained colonoscopists to perform all of the necessary screening procedures. Using modalities other than colonoscopy for rescreening may help to solve this shortage as it would free up scarce colonoscopy personnel to perform more primary screening exams.
From a policy perspective, the potential cost-savings (in 2010 dollars) from switching to FIT or HSFOBT following a negative screening colonoscopy rather than continuing colonoscopy are considerable. For every individual who switches, $450 to $495 is saved over their lifetime (assuming imperfect adherence). Data from the 2008 National Health Information Survey (53) indicate that approximately 40% of 50–54-year-olds had an endoscopy within the recommended intervals, and 92% reported their most recent endoscopy was a colonoscopy. On average, no adenomas or colorectal cancer are detected in 82% of initial screening colonoscopies (39). This suggests that if the estimated 6.6M 50–54-year-olds who had a negative screening colonoscopy in 2008 (that is, 40% × 92% × 82% × 21.5M 50–54-year-olds (54)) were rescreened with yearly FIT or yearly HSFOBT, $3 billion could be saved over the course of their lives. The cost-savings from switching to five-yearly CTC following a negative colonoscopy are lower yet still sizable ($0.6 billion), although these savings could be at least partially offset by the costs of working up extracolonic findings.
Our analysis has a number of limitations. We did not consider the risks and costs of radiation exposure from CTC because the radiation-related cancer risk was estimated to be very small in comparison to the reduction in colorectal cancer risk from CTC screening (55). We also did not include the risks, potential benefits, or costs associated with the detection of incidental findings by CTC. The prevalence of clinically-significant incidental findings in asymptomatic populations ranges from 7% to 11%, and the average cost of their work-up (in US settings) has been estimated at $28 to $99 per person screened (56). When these costs, as well as any potential cost-savings (and gains in life-expectancy) associated with earlier detection of clinically-significant disease are confirmed, they should be included in the assessment of a CTC strategy.
Data from several studies suggest that colonoscopy may not offer as much protection from right-sided compared with left-sided disease (19, 57–59). We did not incorporate this into our analysis because the reasons for the difference remain unclear, but likely involve a combination of technical and biological factors that may affect the location-specific effectiveness of colonoscopy as well as other screening modalities. When additional data become available that confirm the magnitude of the effect and elucidate the mechanism, they should be incorporated into an assessment of all modalities.
There are limited data on test-specific adherence, particularly among those who already had a colonoscopy and had no adenomas or colorectal cancers detected. Imperiale (39) reported adherence of 52% with a repeat colonoscopy five years after a negative exam. It is unclear if adherence ten years after a negative colonoscopy would differ. In the absence of data for CTC, we assumed adherence with the first CTC was equal to that with a repeat colonoscopy (that is, 52%) and that individuals on average have two CTCs by age 75. Many have suggested that adherence with CTC for initial screening may be higher than with colonoscopy (60–62), although such claims have been based on small single-institution studies. A Dutch population-based study found that screening uptake was higher for CTC vs. colonoscopy (63). However, CTC was performed without cathartic bowel preparation. It is unclear if uptake would be higher if individuals had cathartic bowel preparation (as modeled in our analysis). Our estimates of adherence with FOBT were based on data from a Veteran population over a five-year period; adherence among the general screening population (and over longer periods of time) may differ. Furthermore, adherence with FOBT may differ among those who already opted for colonoscopy.
In conclusion, compared with the currently-recommended strategy of continuing colorectal cancer screening with ten-yearly colonoscopy following an initial negative exam, rescreening at age 60 with yearly HSFOBT, yearly FIT, or five-yearly CTC yield comparable life-years with fewer complications and at a lower cost. Therefore it is reasonable to rescreen individuals with a negative colonoscopy with other modalities.
We acknowledge Martin Brown, PhD, and Robin Yabroff, PhD, of the NCI for their assistance with obtaining colorectal cancer treatment costs using SEER-Medicare linked data and Eric (Rocky) Feuer, PhD, of the NCI for continued support of the work and infrastructure of the CISNET consortium. We thank Carolyn M. Rutter, PhD, of the Group Health Research Institute and Ann G. Zauber, PhD, of Memorial Sloan-Kettering Cancer Center for helpful comments and review of earlier versions of this article. None of the individuals acknowledged above received compensation for their contributions.
Funding support: Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under award number RC1CA147256. The model used in this analysis was also supported by the National Cancer Institute (U01CA088204 and U01CA152959).The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Primary Funding Source: National Cancer Institute of the National Institutes of Health.
Publisher's Disclaimer: This is the prepublication, author-produced version of a manuscript accepted for publication in Annals of Internal Medicine. This version does not include post-acceptance editing and formatting. The American College of Physicians, the publisher of Annals of Internal Medicine, is not responsible for the content or presentation of the author-produced accepted version of the manuscript or any version that a third party derives from it. Readers who wish to access the definitive published version of this manuscript and any ancillary material related to this manuscript (e.g., correspondence, corrections, editorials, linked articles) should go to www.annals.org or to the print issue in which the article appears. Those who cite this manuscript should cite the published version, as it is the official version of record.
Disclosures: Dr. Gazelle is a consultant for GE Healthcare. The authors declare no conflicts of interest.
Reproducible Research Statement Protocol: Available to approved individuals with written agreement from Dr. Knudsen (gro.ati-hgm@nesdunka).
Statistical code and data set: Simulation model available to approved individuals with written agreement from Dr. Knudsen.