Colorectal Cancer Screening Evidence Summary - 11 21 18 - KK - Edits

OHSU Health System
Office of Clinical Integration and Evidence-Based Practice

Colorectal Cancer Screening
Evidence Summary
Date started: April 2017

Date completed: October 2017
Content Expert Team Members:

Tara Bergeron, MSN, Quality Improvement Coordinator, David Lieberman, MD, Gastroenterology, OHSU
Tuality Health Alliance Christine Mullowney, MD, Family Medicine, OHSU
Mike Doorly, MD, General and Colorectal Surgery, Tuality Lisa Newman, BSN, Nursing, OHSU
Healthcare Paola Paz, MSN, Quality Improvement Coordinator,
Alice Fung, MD, Radiolgy/Diagnostic Imaging, OHSU Tuality Health Alliance
Daniel Herzig, MD, General Surgery, OHSU Moira Ray, MD, MPH, Family Medicine, OHSU
Alan Hodgson, BS, Clinical Reporting, OHSU Bruin Rugge, MD, MPH, Family Medicine, OHSU
Vicki Jakovec, MSN, Nursing, OHSU Mick Scanlan, MD, Pathology
Steven Kassakian, MD, Internal Medicine/Informatics, Cheri Warren, MS, Clinical Informatics, OHSU
OHSU Daisuke Yamashita, MD, MPH, Family Medicine, OHSU
Jeremy Lake, MD, Gastroenterology Tuality Healthcare
Mark Lovgren, Patient Representative
Office of Clinical Integration and Evidence-Based Practice Team:

Elizabeth Crabtree, PhD, MPH, Director of Clinical
Integration and EBP
Marcy Hager, MA, EBP Program Manager
Andrew Hamilton, MS/MLS, Liaison Librarian
Stephanie Halvorson, MD, Clinical Integration Medical
Director/Hospital Medicine
Tovah Kohl, MA, EBP Program Manager
Objective for the Review: To critically review the evidence Definitions:

on screening for colorectal cancer in adult patients Average Risk: Non-African American patients aged 50
years or older with no personal history of CRC or
adenomas, no inflammatory bowel disease, and with a
negative first-degree family history for CRC.
Increased Risk: African American patients or patients with
a personal or first-degree family history of CRC or
advanced adenomas, or patients with inflammatory
bowel disease.
1
OHSU Health System
Evidence Summary
Review Preparation:
In asymptomatic populations at general risk of CRC
1. What is the effectiveness (or comparative effectiveness) of screening programs based on any of the following
screening tests (alone or in combination) in reducing (a) incidence of and (b) mortality from colorectal cancer:
colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool screening tests, guaiac fecal
occult blood, fecal immunochemical, multitarget stool DNA tests, blood screening test, methylated SEPT9
DNA? Does effectiveness (or comparative effectiveness) vary by important subpopulations?
2. What are the test performance characteristics (e.g., sensitivity and specificity) of the following screening tests
(alone or in combination) for detecting (a) colorectal cancer, (b) advanced adenomas, and (c) adenomatous
polyps based on size: colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool
screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA
tests, blood screening test, methylated SEPT9 DNA? Do test performance characteristics vary by important
subpopulations?
3. a) What are the adverse effects (i.e., serious harms) of the different screening tests (either as single application
or in a screening program)? b) Do adverse effects vary by important subpopulations (e.g., age)?
4. Does using shared decision-making when determining appropriate screening test increase the rate of
completed CRC screens compared to a LIP prescribed test?
5. What is the comparative patient acceptance of the following screening programs: colonoscopy, flexible
sigmoidoscopy, computed tomographic colonography, stool screening tests, high-sensitivity guaiac fecal occult
blood, fecal immunochemical, or multitarget stool DNA tests, blood-screening test, methylated SEPT9 DNA?
6. What is the comparative cost-effectiveness of the following screening programs: colonoscopy, flexible
sigmoidoscopy, computed tomographic colonography, stool screening tests, high-sensitivity guaiac fecal occult
blood, fecal immunochemical, or multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA?
7. In screening for CRC, what tools and/or strategies are most effective in identifying high-risk patients?
Quality Measures:
Outcome Process
- Patient satisfaction with - Percentage of patients aged 50-75 years who had appropriate screening for
CRC screening colorectal cancer
- Percentage of patients aged 50-75 years receiving a screening colonoscopy
without biopsy or polypectomy and with an adequate prep who had a
recommended follow-up -interval of 10 years for repeat colonoscopy
documented in their colonoscopy report
- Percentage follow-up for positive screenings for other modalities
- Percentage of patients ≥ 50 years old with ≥ 1 conventional adenoma or
sessile serrated polyp or colorectal cancer detected during screening
colonoscopy
- Percentage of patients ≥ 18 years old receiving a surveillance colonoscopy,
with a history of a prior adenomatous polyp(s) in previous colonoscopy
findings, who had an interval of ≥ 3 years since their last colonoscopy
- Utilization of shared decision making tool
- Utilization of clinical scoring system
- Positive test referrals received
2
OHSU Health System
Evidence Summary

Existing External Guidelines/Pathways/Order Sets
Existing External Guidelines
External Guideline Organization and Author Last Update
Colorectal Cancer Screening: Recommendations for Physicians and
U.S. Multi-Society Task Force 2017
Patients from the U.S. Multi-Society Task Force on Colorectal Cancer
National Comprehensive Cancer Network Colorectal Cancer
National Comprehensive Cancer Network 2017
Guidelines
Colorectal Cancer: Screening US Preventative Services Task Force 2016
Colorectal Cancer Canadian Task Force on Preventative Health Care 2016
Colorectal Cancer Screening American College of Radiology 2013
Colorectal Cancer Screening American College of Physicians 2012
American College of Gastroenterology Guidelines for Colorectal

American College of Gastroenterology 2008
Cancer Screening 2008
The seven published clinical guidelines were evaluated for this review using the University of Pennsylvania’s Center for Evidence-Based Practice Trustworthy Guideline rating
scale. The scale is based on the Institute of Medicine’s “Standards for Developing Trustworthy Clinical Practice Guidelines” (IOM), as well as a review of the AGREE Enterprise
and Guidelines International Network domains.
MSTF NCCN USPSTF CTF 2016 ACR ACP ACG

Guideline Issuer 2017 2017 2016 2013 2012 2008
1. Transparency A B A A C B B
2. Conflict of interest B NR A A NR A NR
3. Development group C C A NR B B C
4. Systematic Review A B A A B B B
5. Supporting evidence A C A A C B A
3
Colorectal Cancer Screening Evidence Summary
6. Recommendations A C A A C B A
7. External Review B NR A A NR B NR
8. Currency and updates B B A B B B C
See appendix B for full description of the Trustworthy Guideline grading system.
External Guideline Screening Recommendations

Multi-Society Task Force 2017 • Recommend that clinicians offer CRC screening beginning at age 50 (strong recommendation, high-quality evidence).
• Suggest that sequential offers of screening tests, offering multiple screening options, and risk-stratified screening are all reasonable approaches to offering screening (weak
recommendation, low-quality evidence).
1. Recommends colonoscopy every 10 years or annual FIT as first-tier options for screening the average-risk persons for colorectal neoplasia (strong recommendation; moderate
quality evidence).
4
2. Recommends that physicians performing screening colonoscopy measure quality, including the adenoma detection rate (strong recommendation, high-quality evidence).
3. Recommends that physicians performing FIT monitor quality (strong recommendation, low-quality evidence).
4. Recommends CT colonography every 5 years or FIT-fecal DNA every 3 years (strong recommendation, low-quality evidence) or flexible sigmoidoscopy every 5 to 10 years
(strong recommendation, high-quality evidence) in patients who refuse colonoscopy and FIT.
5. Suggests that capsule colonoscopy (if available) is an appropriate screening test when patients decline colonoscopy, FIT, FIT-fecal DNA, CT colonography, and fl exible
sigmoidoscopy (weak recommendation, low-quality evidence).
6. We suggest against Septin9 for CRC screening (weak recommendation, low-quality evidence).
National Comprehensive Cancer
Network 2017
- CRC screening is recommended in adults aged 50 – 75 years. Because the risk of colorectal screening increases with age, the decision to screen between ages 76-85 y should be
individualized and include a discussion of the risks and benefits based on comorbidity status and estimated life expectancy. Individuals who have not been previously screened
are most likely to benefit in this age group.
- Screening should be individualized and include a discussion of the risks and benefits of each modality.
- If colonoscopy is incomplete or preparation is suboptimal, consider other screening modality or repeat colonoscopy within 1 year.
5
US Preventative Services Task Force

2016
Canadian Task Force on Preventative • Recommend screening adults aged 60 to 74 for CRC with FOBT (either gFOBT or FIT) every two years OR flexible sigmoidoscopy every 10 years.
Health Care 2016 (Strong recommendation; moderate quality evidence)
• Recommend screening adults aged 50 to 59 for CRC with FOBT (either gFOBT or FIT) every two years OR flexible sigmoidoscopy every 10 years.
(Weak recommendation; moderate quality evidence)
• Recommend not screening adults aged 75 years and over for CRC.
(Weak recommendation; low quality evidence)
• Recommend not using colonoscopy as a screening test for CRC.
(Weak recommendation; low quality evidence)
American College of Radiology 2013 ACR’s guideline focuses on nonradiologic tests for colorectal cancer screening. Overall, the most appropriate imaging test for colorectal cancer screening is CTC. However, CTC expertise
may not be available in all geographic areas. Thus, a Double-Contrast Barium Enema (DCBE) may be the only imaging option in a particular area, despite its lower performance profile. The
choice between these 2 tests may ultimately depend on local imaging expertise and on physician and patient preference.
Summary
• CTC has emerged as the leading imaging technique for colorectal cancer screening.
• DCBE remains an imaging test that may be appropriate for colorectal cancer screening, particularly when CTC is not available.
• CTC is the preferred test following an incomplete optical colonoscopy.
• Imaging tests including CTC and barium enema are usually not appropriate for colorectal cancer screening in high-risk patients with hereditary nonpolyposis colorectal cancer
and inflammatory bowel disease.
American College of Physicians 2012 • Guidance Statement 1:
ACP recommends that clinicians perform individualized assessment of risk for colorectal cancer in all adults.
6
• Guidance Statement 2:
ACP recommends that clinicians screen for colorectal cancer in average-risk adults starting at the age of 50 years and in high-risk adults starting at the age of 40 years or 10
years younger than the age at which the youngest affected relative was diagnosed with colorectal cancer.
ACP recommends using a stool-based test, flexible sigmoidoscopy, or optical colonoscopy as a screening test in patients who are at average risk. ACP recommends using optical
colonoscopy as a screening test in patients who are at high risk. Clinicians should select the test based on the benefits and harms of the screening test, availability of the
screening test, and patient preferences.
ACP recommends that clinicians stop screening for colorectal cancer in adults over the age of 75 years or in adults with a life expectancy of less than 10 years.
American College of CRC Screening Recommendations
Gastroenterology 2008 Preferred CRC screening recommendations
• Cancer prevention tests should be offered first. The preferred CRC prevention test is colonoscopy every 10 years, beginning at age 50. (Grade 1 B) Screening should begin at age
45 years in African Americans (Grade 2 C)
• Cancer detection test. This test should be offered to patients who decline colonoscopy or another prevention test. The preferred cancer detection test is annual FIT for blood
(Grade 1 B)
Alternative CRC prevention tests
• Flexible sigmoidoscopy every 5-10 years (Grade 2 B)
• CT colonography every 5 years (Grade 1 C)
Alternative cancer detection tests
• Annual Hemoccult Sensa (Grade 1 B)
• Fecal DNA testing every 3 years (Grade 2 B)
Recommendations for screening when family history is positive but evaluation for HNPCC considered not indicated
• Single first-degree relative with CRC or advanced adenoma diagnosed at age ≥ 60 years
Recommended screening: same as average risk (Grade 2 B)
• Single first-degree with CRC or advanced adenoma diagnosed at age < 60 years or two first-degree relatives with CRC or advanced adenomas.
Recommended screening: colonoscopy every 5 years beginning at age 40 years or 10 years younger than age at diagnosis of the youngest affected relative (Grade 2 B)
FAP
• Patients with classic FAP (>100 adenomas) should be advised to pursue genetic counseling and genetic testing, if they have siblings or children who could potentially benefit
from this testing (Grade 2 B)
• Patients with known FAP or who are at risk of FAP based on family history (and genetic testing has not been performed) should undergo annual flexible sigmoidoscopy or
colonoscopy, as appropriate, until such time as colectomy is deemed by physician and patient as the best treatment (Grade 2 B)
• Patients with retained rectum after subtotal colectomy should undergo flexible sigmoidoscopy every 6 – 12 months (Grade 2 B)
• Patients with classic FAP, in whom genetic testing is negative, should undergo genetic testing for bi-allelic MYH mutations. Patients with 10 – 100 adenomas can be considered
for genetic testing for attenuated FAP and if negative, MYH associated polyposis (Grade 2 C)
HNPCC
• Patients who meet the Bethesda criteria should undergo microsatellite instability testing of their tumor or a family member’s tumor and/or tumor immunohistochemical
staining for mismatch repair proteins (Grade 2 B)
• Patients with positive tests can be offered genetic testing. Those with positive genetic testing, or those at risk when genetic testing is unsuccessful in an affected proband,
should undergo colonoscopy every 2 years beginning at age 20 – 25 years, until age 40 years, then annually thereafter (Grade 2 B)
7
Adverse Effects Guideline Recommendations:

The 2016 United States Preventive Service Task Force states the rate of serious adverse events from colorectal cancer screening increases with age. The harms of screening for colorectal cancer in adults aged
50 to 75 years is small. Thus, the harms of screening for colorectal cancer in adults 76 years and older are small to moderate.
Level of Evidence for recommendations not provided
The 2013 American College of Radiology stated that potential adverse health effects associated with radiation exposure are an important factor to consider when selecting the appropriate imaging procedure.
Because there is a wide range of radiation exposures associated with different diagnostic procedures, a relative radiation level (RRL) indication has been included for each imaging examination.
Shared Decision-Making Guideline Recommendations:

The 2017 Multi-Society Task Force suggests three approaches to offering screening: (1) Multiple Options; (2) Sequential Testing; and (3) Risk stratified approach. When using the multiple option approach
with patients, MSTF suggested limited to 2 to 3 preferred options. If patients decline all the offered options, 1 or more of the other options can be offered.
Weak Recommendation, low-quality evidence
8
The 2016 United States Preventive Services Task Force states clinicians should consider engaging patients in informed decision making about the screening strategy that would most likely result in completion,
with high adherence over time, taking into consideration both the patient’s preferences and local availability.
Cost-Effectiveness Guideline Recommendations:

The 2017 Multi-Society Task Force cited evidence that colorectal cancer screening by any available modality is cost-effective compared with no screening and in some models screening results in cost savings.
The 2016 United States Preventive Services Task Force does not consider the costs of providing a service in this assessment.
The 2013 American College of Radiology demonstrated in a cost-effective analysis that a double-contrast barium enema (DCBE) performed every 5 – 10 years costs less than $22,000 per life-year saved for a
possible range of natural history.
9
High-Risk Patients Guideline Recommendations:

The 2017 Multi-Society Task Force:
For patients with family history of colorectal cancer and polyps:
• Suggests that persons with 1 first-degree relative with CRC or a documented advanced adenoma diagnosed at age <60 years or with 2 first-degree relatives with CRC and/or documented advanced
adenomas undergo colonoscopy every 5 years beginning 10 years younger than the age at which the youngest first-degree relative was diagnosed or age 40, whichever is earlier (weak
recommendation, low-quality evidence).
• Suggests that persons with 1 first-degree relative diagnosed with CRC or a documented advanced adenoma at age ≥60 years begin screening at age 40. The options for screening and the
recommended intervals are the same as those for average-risk persons (weak recommendation, very-low quality evidence).
• Suggests that persons with 1 or more first-degree relatives with a documented advanced serrated lesion (SSP or traditional serrated adenoma ≥10 mm in size or an SSP with cytologic dysplasia) should
be screened according to above recommendations for persons with a family history of a documented advanced adenoma (weak recommendation, very-low-quality evidence).
• Recommends that persons with 1 or more first-degree relatives with CRC or documented advanced adenomas, for whom we recommend colonoscopy, should be offered annual FIT if they decline
colonoscopy (strong recommendation, moderate-quality evidence).
Patient considerations regarding age and colorectal cancer risk:
• Recommends that screening begin in non-African American average-risk persons at age 50 years (strong recommendation; moderate-quality evidence).
• Suggests that screening begin in African Americans at age 45 years (weak recommendation, very-low-quality evidence).
• Recommends that adults age <50 years with colorectal bleeding symptoms (hematochezia, unexplained iron deficiency anemia, melena with a negative upper endoscopy) undergo colonoscopy or an
evaluation sufficient to determine a bleeding cause, initiate treatment, and complete follow-up to determine resolution of bleeding (strong recommendation, moderate-quality evidence).
• Suggests that persons who are up to date with screening and have negative prior screening tests, particularly colonoscopy, consider stopping screening at age 75 years or when life expectancy is less
than 10 years (weak recommendation, low-quality evidence).
• Recommends considering stopping screening at age 75 years or when life expectancy is less than 10 years (weak recommendation, low-quality evidence).
• Suggests that persons without prior screening should be considered for screening up to age 85, depending on consideration of their age and comorbidities (weak recommendation, low-quality
evidence).
The 2016 United States Preventive Services Task Force recommendations applies to all racial/ethnic groups, with the clear acknowledgement that efforts are needed to ensure that at-risk populations receive
recommended screening, follow-up, and treatment.
The 2009 American College of Gastroenterology stated:
• Screening is recommended in African Americans beginning at age 45 years. (Grade 2 C)

• A family history of only small tubular adenomas in first degree relatives is not considered to increase the risk of CRC. (Level of Evidence for recommendation not provided)
• Individuals with a single first-degree relative with CRC or advanced adenomas diagnosed at age >/= 60 years can be screened like average-risk persons. (Level of Evidence for recommendation not
provided)
10
Review of Relevant Evidence: Search Strategies and Databases Reviewed

Search Strategies Document Strategies Used
Search Terms/Strategies Used: PICO Questions 1, 2, 3, 4, 5, 6, 7: See Appendix C
Years Searched - All Questions PICO Questions 1, 2, 3: January 2015 – May 2017; PICO Questions 4,5,6,7: 2002 – May 2017
Language English
Age of Subjects Adults, >/= 18 years
Evidence Found with Searches

Check type of evidence found Summary of Evidence – All Questions Number of articles obtained
Systematic reviews/Meta-analysis 16
Randomized controlled trials 3
Non-randomized studies 66
Government/State agency regulations 2
Professional organization guidelines/white papers, etc. 5
Evaluating the Quality of the Evidence

The GRADE criteria were used to evaluate the quality of evidence presented in research articles reviewed during the development of this guideline. The table below defines how the quality of the evidence is
rated and how a strong versus weak recommendation is established. For more detailed information, see Appendix A.
Recommendation
STRONG Desirable effects clearly outweigh undesirable effects or vice versa
CONDITIONAL Desirable effects closely balanced with undesirable effects
Quality Type of Evidence

High Further research is very unlikely to change our confidence in the estimate of effect.
Moderate Further research is likely to have an important impact on our confidence in the estimate of effect and may
change the estimate.
Low Further research is very likely to have an important impact on our confidence in the estimate of effect and is
likely to change the estimate.
Very Low Any estimate of effect is very uncertain.
11
Question #1: In asymptomatic populations at general risk of CRC, what is the effectiveness (or comparative effectiveness) of screening programs based on any of the following
screening tests (alone or in combination) in reducing (a) incidence of and (b) mortality from colorectal cancer: colonoscopy, flexible sigmoidoscopy, computed tomographic
colonography, stool screening tests, guaiac fecal occult blood, fecal immunochemical, multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA? Does effectiveness
(or comparative effectiveness) vary by important subpopulations?
Primary Literature:
USPSTF 2016 Systematic Review Findings:

Overall Summary of Impact of Screening on Colorectal Cancer Incidence and Mortality
12
Overall Summary of Effectiveness of Screening on CRC Mortality
13
PICO Question: In asymptomatic populations at general risk of CRC, what is the effectiveness (or comparative effectiveness) of screening programs Lower Quality Rating if:
based on any of the following screening tests (alone or in combination) in reducing (a) incidence of and (b) mortality from colorectal cancer: Studies inconsistent (wide
variation of treatment effect across
colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool screening tests, guaiac fecal occult blood, fecal immunochemical,
studies, populations, interventions,
multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA? Does effectiveness (or comparative effectiveness) vary by important or outcomes varied)
subpopulations?
Studies are indirect
Outcome: Incidence; Modality: Colonoscopy (PICO question is quite different
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations from the available evidence in
Study Methods regard to population, intervention,
Total # of Studies: 4 # of Systematic Reviews: 3 # of RCTs: 0 # of Non-Randomized Studies: 1 # of Diagnostic Studies: 0 comparison, or outcome)
Canadian Task To synthesize Systematic Review with meta- 0 RCTs No RCTs that met inclusion criteria for Study Limitations =
Force on evidence on the analysis the benefits of CRC screening using None Studies are imprecise (When
Preventive Health, benefits and harms of colonoscopy Systematic Review
studies include few patients and few
2016, CMAJ screening and the test Review did not address
Canadian Medical properties of effective focused clinical question events and thus have wide
Association screening methods for Search was not detailed or confidence intervals and the results
Journal asymptomatic adults exhaustive are uncertain)
who are not at high Quality of the studies was
risk for colorectal not appraised or studies were of Publication Bias
cancer. low quality
(e.g. pharmaceutical company
Methods and/or results were
inconsistent across studies sponsors study on effectiveness of
drug, only small, positive studies
found)
Pan, J., et al, 2016, To evaluate the Systematic Review with meta- 11 observational Pooled analysis showed that Study Limitations =
American Journal magnitude of analysis studies were included in colonoscopy was associated with a 61% None Increase Quality Rating if:
of protection against CRC analysis, 5 were cohort RR reduction in CRC incidence (RR: 0.39; Systematic Review Large Effect
Gastroenterology by colonoscopy, with studies, 3 prospective 95% CI: 0.26–0.60; I 2 =93.6%), in Review did not address Dose-response gradient
screening and and 2 retrospective and patients with non-malignant findings, focused clinical question
Plausible confounders or other
diagnostic indications, 6 were case–control although there was high heterogeneity Search was not detailed or
biases increase certainty of effect
in patients with non- studies. A total of for the outcome of CRC incidence. exhaustive
malignant findings and 1,499,521 individuals Quality of the studies was
further determine the included, 1,296,605 in not appraised or studies were of Quality (certainty) of evidence for
potentially more the morality meta- low quality studies as a whole:
marked effect of analysis Methods and/or results were High
screening over inconsistent across studies Moderate
diagnostic
Low
colonoscopy in the
magnitude or Very Low
14
reductions in CRC
incidence
Tinmouth, J., To conduct a meta- Systematic Review with meta- 2 Systematic Reviews, 2 Study Limitations =
2016, Canadian analysis quantifying analysis Prospective Studies, 3 None
Journal of the magnitude of Case-Controlled Studies Systematic Review
Gastroenterology protection by Review did not address
& Hepatology colonoscopy, with focused clinical question
screening and Search was not detailed or
diagnostic indications, exhaustive
against CRC in patients Quality of the studies was
with non-malignant not appraised or studies were of
findings and low quality
demonstrating the Methods and/or results were
potentially more inconsistent across studies
marked effect of
screening over
diagnostic
colonoscopy
Samadder, N.J., et To examine whether Case-control study; Utah In a database with more Exposure to colonoscopy reduced the Study Limitations =
al, 2016, Clinical exposure to residents, 54 to 90 years old, than 7.2 million unique odds for a diagnosis of CRC; the odds None
Gastroenterology colonoscopy who received a CRC diagnosis individuals, 5,128 cases ratios (ORs) were 0.41 for any CRC (95% Non-Randomized Studies
& Hepatology decreases the odds of from 2000 through 2010 and 20,512 controls confidence interval [CI], 0.38-0.44), 0.58 Failure to develop and apply
CRC incidence in Utah (cases) were included in were found and 741 for proximal colon cancer (95% CI, 0.51- appropriate eligibility criteria
analysis. Age- and sex-matched cases (14%) and 5715 0.65), and 0.29 for distal colon or rectal Flawed measurement of
controls with no history of CRC controls (28%) received cancer (95% CI, 0.25-0.33). This finding both exposure and outcome
(controls) were selected for a colonoscopy was consistent among sexes, age groups, Failure to adequately control
each case. We determined and cancer stages. confounding
receipt of colonoscopy 6 Incomplete or inadequately
months to 10 years before the short follow-up
reference date for each case Differences in important
and control through prognostic factors at baseline
administrative claims data.
Colonoscopy exposure was
compared by using conditional
logistic regression
References:
1. Canadian Task Force on Preventive Health, C., et al. (2016). "Recommendations on screening for colorectal cancer in primary care." CMAJ Canadian Medical Association Journal 188(5): 340-348.
2. Pan, J., et al. (2016). "Colonoscopy Reduces Colorectal Cancer Incidence and Mortality in Patients With Non-Malignant Findings: A Meta-Analysis." American Journal of Gastroenterology 111(3): 355-365.
3. Tinmouth, J., et al. (2016). "Colorectal Cancer Screening in Average Risk Populations: Evidence Summary." Canadian Journal of Gastroenterology & Hepatology 2016: 2878149.
4. Samadder, N. J., et al. (2016). "Risk of Incident Colorectal Cancer and Death After Colonoscopy: A Population-based Study in Utah." Clinical Gastroenterology & Hepatology 14(2): 279-286.e271-272.
15
colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool screening tests, guaiac fecal occult blood, fecal immunochemical, variation of treatment effect across
multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA? Does effectiveness (or comparative effectiveness) vary by important
or outcomes varied)
subpopulations?
Outcome: Incidence; Modality: Sigmoidoscopy Studies are indirect
Author/Date Purpose of Study Study Design & Sample Outcomes Design Limitations (PICO question is quite different
Methods from the available evidence in
Total # of Studies: 3 # of Systematic Reviews: 3 # of RCTs: 0 # of Non-Randomized Studies: 0 # of Diagnostic Studies: 0 regard to population, intervention,
Canadian Task To synthesize evidence Systematic Review with 3 RCTS; Combined The length of follow-up across four studies Study Limitations = comparison, or outcome)
Force on on the benefits and meta-analysis sample of 243,917 ranged from seven years to 11.9 years. None
Preventive Health, harms of screening and (108,234 [I]; 135,683 There was a reduction in late stage CRC Systematic Review
Studies are imprecise (When
2016, CMAJ the test properties of [C]). All studies included using screening with FS compared to no Review did not address
Canadian Medical effective screening a mixed gender screen RR 0.75 (95%CI, 0.66, 0.86, I2=23%); focused clinical question studies include few patients and few
Association methods for population. Two studies ARR 1,733/million (1,011-2,368 fewer, NNS Search was not detailed or events and thus have wide
Journal asymptomatic adults included participants = 577 (95%CI, 422-989)). exhaustive confidence intervals and the results
who are not at high risk with age ranging from Quality of the studies was are uncertain)
for colorectal cancer. 55 to 64 years and one not appraised or studies were of
study with age ranging low quality
Publication Bias
from 55 to 74 years. Methods and/or results were
inconsistent across studies
sponsors study on effectiveness of
found)
Holme, O., 2017, To compare the Systematic Review with 287,928 individuals Screening reduced the incidence of Study Limitations = Increase Quality Rating if:
BMJ effectiveness of flexible meta-analysis; Pooled were included in the colorectal cancer in men (relative risk 0.76; None Large Effect
sigmoidoscopy in analysis of randomised pooled analysis; 115,139 95% confidence interval 0.70 to 0.83) and Systematic Review Dose-response gradient
screening for colorectal trials (the US Prostate, randomised to women (0.83; 0.75 to 0.92). No difference Review did not address Plausible confounders or other
cancer by patient sex Lung, Colorectal and screening and 172,789 in the effect of screening was seen focused clinical question
and age Ovarian cancer screening to usual care. between men younger than 60 and those Search was not detailed or
trial older than 60. Screening reduced the exhaustive
(PLCO), the Italian incidence of colorectal cancer in women Quality of the studies was Quality (certainty) of evidence for
Screening for Colon and younger than 60 (relative risk 0.71; 95% not appraised or studies were of studies as a whole:
Rectum trial (SCORE), and confidence interval 0.59 to 0.84), but not low quality High
the Norwegian Colorectal significantly in those aged 60 or older Methods and/or results were Moderate
Cancer (0.90; 0.80 to 1.02). inconsistent across studies
Low
Prevention trial
Very Low
(NORCCAP)).
16
Tinmouth, J., To conduct a meta- Systematic Review with 1 Systematic Review Study Limitations =
2016, Canadian analysis quantifying the Meta-Analysis and 4 RCTs None
Journal of magnitude of Systematic Review
Gastroenterology protection by Review did not address
& Hepatology colonoscopy, with focused clinical question
diagnostic indications, exhaustive
against CRC in patients Quality of the studies was
with non-malignant not appraised or studies were of
findings and low quality
demonstrating the Methods and/or results were
potentially more inconsistent across studies
marked effect of
17
screening over
diagnostic colonoscopy
References:
2. Holme, O., et al. (2017). "Effectiveness of flexible sigmoidoscopy screening in men and women and different age groups: pooled analysis of randomised trials." BMJ 356: i6673.
or outcomes varied)
subpopulations?
Outcome: Incidence; Modality: Fecal Occult Blood Test (FOBT) Studies are indirect
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations (PICO question is quite different
Study from the available evidence in
Canadian Task To synthesize Systematic Review with meta- 2 RCTS; Combined The length of follow-up across both studies Study Limitations = comparison, or outcome)
Force on evidence on the analysis sample of 220,283 ranged from 9 years to 19.5 years. None
Preventive benefits and (110,200 [I]; Screening with gFOBT compared to no Systematic Review
Health, 2016, harms of 110,083[C]). These screen had a pooled effect on late stage Review did not address
CMAJ Canadian screening and mixed gender studies CRC of RR 0.92 (95%CI, 0.85, 0.99, I2=0%); focused clinical question studies include few patients and few
Medical the test included in one study the ARR was 1,141/million (198 fewer to Search was not detailed or events and thus have wide
Association properties of participants aged 45 to 2,017 fewer). The NNS is 876 (95%CI, 496- exhaustive confidence intervals and the results
Journal effective 74 years and the other 5051) for the outcome of incidence of late Quality of the studies was are uncertain)
screening study aged 60 to 64 stage CRC. not appraised or studies were of
methods for years. low quality
Publication Bias
asymptomatic Methods and/or results were
adults who are inconsistent across studies
not at high risk sponsors study on effectiveness of
for colorectal drug, only small, positive studies
cancer. found)
Increase Quality Rating if:

Large Effect
Dose-response gradient
18
Quality (certainty) of evidence for

studies as a whole:
High
Moderate
Low
Very Low
References:
Modeling Studies Summaries

Author/Date Purpose of Study Study Design & Methods Outcomes
Berger, B.M., et al., 2016, American To demonstrate that the US Preventive Services Task Descriptive analysis of USPSTF modeling of the clinical
Journal of Managed Care Force (USPSTF) modeling screening outcomes included impact of CRC screening strategies. These scenarios
CRC incidence reduction. involved different modalities (hsFOBT, FIT, and mt-sDNA),
intervals (2y, 3y, and every 5 years), and screening age
groups (50-75, 50-80, 50-85, 55-75, 55-80, and 55-85 years).
All scenarios assumed 100% adherence to the screening
modality and interval. Screening scenarios were plotted
against life years gained (LYG) and colonoscopies performed
to create an “efficiency frontier,” a line connecting the most
efficient screening scenarios based on the trade-off
between LYG and colonoscopies. By excluding annual stool
tests, we recreated CISNET’s analysis of the “efficiency
frontier” and compared these screening scenarios with each
other and with 10y colonoscopy screening. Per CISNET, tests
were recommended if they were within 98% of the LYG at
the equivalent point on the efficiency frontier and
generated at least 90% of the LYG generated by 10y
screening colonoscopy.
Brenner, 2015, Clinical To quantify screening colonoscopy effects on Analyzed data from more than 4.4 million screening Overall, approximately 180,000 CRCs (1/28 screening
Gastroenterology & Hepatology prevention, early detection, and over diagnosis of CRC colonoscopies (conducted on individuals 55-79 years old colonoscopies) were estimated to have been prevented, and
in the 10 years since it was introduced in Germany from 2003 through 2012) available through the national more than 40,000 CRCs (1/121 screening colonoscopies) were
screening colonoscopy registry. CRCs prevented, detected detected earlier than they would have been without screening,
earlier than they would have been without screening, and compared with approximately 4,500 overdiagnoses (1/1089
overdiagnosed (cancers detected at screening colonoscopy screening colonoscopies). Almost all CRCs prevented or
that would not have become clinically manifest during the detected earlier than they would have been without screening
patient's lifetime) were estimated by Markov models. resulted from screening colonoscopies performed on
Model parameters included sex-specific and age-specific individuals up to 75 years old (97% and 89%, respectively),
findings at screening colonoscopy; mortality; rates of
19
transition from nonadvanced to advanced adenoma, whereas 28% of overdiagnoses occurred from screening
advanced adenoma to preclinical cancer, or preclinical colonoscopies of individuals older than 75 years old.
cancer to clinically manifest cancer; and protection from
screening colonoscopy. 4.4 million screenings, records from
almost 2 million men and more than 2.4 million women
were included in the database. Slightly more than half of
them were between 55 and 64 years of age; only a minority
of approximately 2.5% were 80 years or older.
Geurts, S.M., et al., 2015, British To project the impact of phasing once-only flexible Simulated the life-course of English residents reaching age By mid-2030, 8.5 million individuals will have been invited for
Journal of Cancer sigmoidoscopy (FS) at age 55 into the England National 55 from 2013 onwards. Model inputs included population once-only FS screening. Adding FS to gFOBT screening is
Health Service Bowel Screening Programme (NHSBCSP), numbers, invitation rates and CRC incidence and mortality estimated to prevent an extra 9,627 (10%) cases by mid-2030.
augmenting biennial gFOBT at ages 60-74, on CRC cases rates. The impact of gFOBT and FS alone on CRC incidence If FS uptake is 38% or 71%, respectively, an extra 7,379 (8%) or
and deaths prevented in England by mid-2030. and mortality were derived from published trials, assuming 13,689 (15%) cases by mid-2030.
an uptake of 50% for FS and 57% for gFOBT. For FS plus
gFOBT, we assumed the gFOBT effect to be 75% of the
gFOBT alone impact.
Jeon, J., et al., 2015, Cancer Causes & To compute the incremental benefit of colonoscopy Calibrated models of CRC incidence within a multistage Model mirrored trial data with 10-year CRC risk reductions
Control over flexible sigmoidoscopy (FSG) using a validated clonal expansion framework to data from: (1) San Francisco- after FSG screening at age 50 years of approximately one-
mathematical model, which reflects colorectal Oakland SEER registry (reference population) and (2) FSG third; the optimal age for a ‘once-only’ FSG exam was between
neoplasia growth characteristics while allowing long-term follow-up data from 50,757 individuals after a ages 50 and 60 years; and the greater benefit was for men
uncertainty in endoscopic detection and removal of negative FSG in the Kaiser Permanente system. Then compared with women. There were considerable incremental
adenomas. compared the residual CRC risks after FSG with full-length gains in reduction in CRC risk by colonoscopy compared with
colonoscopy. FSG with the greatest benefit for screening colonoscopy at age
50 years.
Wang, Y.R., et al, 2016, Digestion To determine whether colonoscopy use is associated 5,701 patients were identified in the Medicare by pulling a Of 5,701 patients in the colonoscopy group, 37 (0.65%)
with a decreased risk of CRC in patients 76-85 years old 5% random sample of the Surveillance, Epidemiology and patients were diagnosed with CRC, compared to 379 (1.55%)
in the United States. End Results-Medicare linked database 76-85 years old at out of 24,437 patients in the control group (p < 0.001). The
outpatient colonoscopy between January 1, 1998 and cumulative incidences of CRC were lower in the colonoscopy
December 31, 2002. Using the Kaplan-Meier method, the group compared to those in the control group (5-year distal
cumulative incidence of CRC was estimated in the above- CRC: 0.26 vs. 0.77%; 5-year proximal CRC: 0.43 vs. 0.79%, both
mentioned colonoscopy group and compared with the p < 0.05). In multivariate Cox regression, colonoscopy was
control group of patients without colonoscopy. All patients associated with decreased risk of all CRC (hazard ratio ((HR)
were followed until diagnosis of CRC or carcinoma in situ, 0.42, 95% CI 0.28-0.65), distal CRC (HR 0.36, 95% CI 0.18-0.70),
death or December 31, 2005. The multivariate Cox and proximal CRC (HR 0.53, 95% CI 0.30-0.92)).
proportional hazards model was used in statistical analysis.
References:
1. Berger, B. M., et al. (2016). "USPSTF colorectal cancer screening guidelines: an extended look at multi-year interval testing." American Journal of Managed Care 22(2): e77-81.
2. Brenner, H., et al. (2015). "Prevention, early detection, and overdiagnosis of colorectal cancer within 10 years of screening colonoscopy in Germany." Clinical Gastroenterology & Hepatology 13(4): 717-723.
3. Geurts, S. M., et al. (2015). "Likely effect of adding flexible sigmoidoscopy to the English NHS Bowel Cancer Screening Programme: impact on colorectal cancer cases and deaths." British Journal of Cancer 113(1): 142-149.
4. Jeon, J., et al. (2015). "Incremental benefits of screening colonoscopy over sigmoidoscopy in average-risk populations: a model-driven analysis." Cancer Causes & Control 26(6): 859-870.
5. Wang, Y. R., et al. (2016). "Decreased Risk of Colorectal Cancer after Colonoscopy in Patients 76-85 Years Old in the United States." Digestion 93(2): 132-138.
20
or outcomes varied)
subpopulations?
Outcome: Mortality; Modality: Colonoscopy Studies are indirect
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations (PICO question is quite different
Study Methods from the available evidence in
Canadian Task To synthesize Systematic Review with meta- 0 RCTs No RCTs that met inclusion criteria for the Study Limitations = comparison, or outcome)
Force on evidence on the analysis benefits of CRC screening using None
Preventive Health, benefits and harms colonoscopy Systematic Review
2016, CMAJ of screening and the Review did not address
Canadian Medical test properties of focused clinical question studies include few patients and few
Association effective screening Search was not detailed or events and thus have wide
Journal methods for exhaustive confidence intervals and the results
asymptomatic adults Quality of the studies was are uncertain)
who are not at high not appraised or studies were
risk for colorectal of low quality
Publication Bias
cancer. Methods and/or results
were inconsistent across
studies sponsors study on effectiveness of
Pan, J., et al, 2016, To evaluate the Systematic Review with meta- 11 observational Pooled analysis showed that colonoscopy Study Limitations = drug, only small, positive studies
American Journal magnitude of analysis studies were included was associated with a 61% reduction in None found)
of protection against in analysis, 5 were CRC mortality (RR: 0.39; 95% CI: 0.35– Systematic Review
Gastroenterology CRC by colonoscopy, cohort studies, 3 0.43; I 2 =12.0%) in patients with non- Review did not address Increase Quality Rating if:
with screening and prospective and 2 malignant findings. focused clinical question
Large Effect
diagnostic retrospective and 6 Search was not detailed or
indications, in were case–control exhaustive
patients with non- studies. A total of Quality of the studies was Plausible confounders or other
malignant findings 1,499,521 individuals not appraised or studies were biases increase certainty of effect
and further were included, of low quality
determine the 1,305,419 in the Methods and/or results Quality (certainty) of evidence for
potentially more morality meta-analysis were inconsistent across studies as a whole:
marked effect of studies
High
screening over
diagnostic Moderate
colonoscopy in the Low
magnitude or Very Low
21
reductions in CRC
mortality
Tinmouth, J., To conduct a meta- Systematic Review with Meta- 2 Systematic Reviews, The overall certainty of direct evidence Study Limitations =
2016, Canadian analysis quantifying Analysis 2 Prospective Studies, supporting the use of colonoscopy to None
Journal of the magnitude of and 1 Retrospective screen people at average risk for CRC was Systematic Review
Gastroenterology protection by Study very low when compared with no Review did not address
& Hepatology colonoscopy, with screening with a RR 0.32 (0.23-0.43). focused clinical question
diagnostic exhaustive
indications, against Quality of the studies was
CRC in patients with not appraised or studies were
non-malignant of low quality
findings and Methods and/or results
demonstrating the were inconsistent across
potentially more studies
marked effect of
screening over
diagnostic
colonoscopy
Samadder, N.J., et To examine whether Case-control study; Utah In a database with 12.8% of CRC cases who died of any cause Study Limitations =
al, 2016, Clinical exposure to residents, 54 to 90 years old, more than 7.2 million and 11.5% of cases who died of None
Gastroenterology colonoscopy who received a CRC diagnosis unique individuals, CRC/malignancy underwent colonoscopy Non-Randomized Studies
& Hepatology decreases the odds from 2000 through 2010 (cases) 5,128 cases and versus 25.3% and 26.2% of matched Failure to develop and
of incidence CRC and were included in analysis. Age- 20,512 controls were controls for each group, respectively. CRC apply appropriate eligibility
death from CRC in and sex-matched controls with found and 741 cases cases who died (both all-cause and CRC criteria
Utah no history of CRC (controls) (14%) and 5715 related) were significantly less likely than Flawed measurement of
were selected for each case. The controls (28%) controls to have undergone colonoscopy both exposure and outcome
receipt of colonoscopy was received a (OR, 0.40; 95% CI, 0.35–0.46 for all-cause Failure to adequately
determined at 6 months to 10 colonoscopy mortality; OR, 0.33; 95% CI, 0.28–0.39 for control confounding
years before the reference date CRC-related mortality) Incomplete or
for each case and control inadequately short follow-up
through administrative claims Differences in important
data. Colonoscopy exposure was prognostic factors at baseline
compared by using conditional
logistic regression
References:
2. Pan, J., et al. (2016). "Colonoscopy Reduces Colorectal Cancer Incidence and Mortality in Patients With Non-Malignant Findings: A Meta-Analysis." American Journal of Gastroenterology 111(3): 355-365.
4. Samadder, N. J., et al. (2016). "Risk of Incident Colorectal Cancer and Death After Colonoscopy: A Population-based Study in Utah." Clinical Gastroenterology & Hepatology 14(2): 279-286.e271-272.
22
colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool screening tests, guaiac fecal occult blood, fecal immunochemical, variation of treatment effect
across studies, populations,
interventions, or outcomes varied)
subpopulations?
Outcome: Mortality; Modality: Fecal Immunochemical Test (FIT/iFOBT) Studies are indirect
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations (PICO question is quite different
Study from the available evidence in
Canadian Task To synthesize Systematic Review with meta- 1 RCT; 192,261 (94,423 The length of follow-up was eight years. Study Limitations = comparison, or outcome)
Force on evidence on the analysis [I]; 97,838 [C]) The effect of iFOBT on CRC mortality was None
Preventive Health, benefits and individuals and RR 0.88 (95%CI, 0.72, 1.07, I2=NA); ARR Systematic Review Studies are imprecise (When
2016, CMAJ harms of included a mixed 277/million (631 fewer to 151 more). Review did not address
studies include few patients and
Canadian Medical screening and the gender population focused clinical question
Association test properties of ages 30 years and Search was not detailed or few events and thus have wide
Journal effective older. The screening exhaustive confidence intervals and the
screening arm received RPHA- Quality of the studies was results are uncertain)
methods for FOBT (FIT) test using not appraised or studies were of
asymptomatic single screen method. low quality Publication Bias
adults who are not The control group was Methods and/or results were (e.g. pharmaceutical company
at high risk for no screening. inconsistent across studies
colorectal cancer.
found)

Chiu, H.M., et al., To assess the Prospective Cohort Study; Follow- 1,160,895 subjects The actual effectiveness in reducing CRC Study Limitations =
2015, Cancer effectiveness of up with Taiwanese patients from mortality attributed to the FIT screening None Large Effect
fecal 2004 to 2009 was conducted to was 62% (relative rate for the screened Non-Randomized Studies Dose-response gradient
immunochemical compare CRC morality for an group vs the unscreened group, 0.38; 95% Failure to develop and apply Plausible confounders or other
testing (FIT) in exposed (screened) group and an confidence interval, 0.35-0.42) with a appropriate eligibility criteria biases increase certainty of effect
reducing unexposed (unscreened) group in maximum follow-up of 6 years. The 21.4% Flawed measurement of
colorectal cancer a population-based CRC screening coverage of the population receiving FIT both exposure and outcome Quality (certainty) of evidence for
(CRC) mortality in service targeting community led to a significant 10% reduction in CRC Failure to adequately control studies as a whole:
a large, residents of Taiwan who were 50 mortality (relative rate, 0.90; 95% confounding
population-based to 69 years old. confidence interval, 0.84-0.95) after High
Incomplete or inadequately
service screening adjustments for a self-selection bias. short follow-up Moderate
program Differences in important Low
prognostic factors at baseline Very Low
23
Giorgi Rossi, P., et To evaluate the Observational Study; An A total of 171,785 The rate of colonoscopy participation was Study Limitations =
al, 2015, Journal impact of organized screening program was people have been about 90%, and 2,896 cancers were None
of screening with implemented in 2005 in the invited, and recorded (1237 in the screening period). Non-Randomized Studies
Gastroenterology immunochemical province of Reggio Emilia approximately 70% Incidence-based mortality decreased by Failure to develop and apply
FOBT (FIT) on CRC (Northern Italy). The program have undergone FIT at 27% (95% CI, 15-37%). appropriate eligibility criteria
mortality invites the resident population least once (272,197 Flawed measurement of
aged 50-69 for FIT every 2 years. tests). both exposure and outcome
Subjects who test positive are Failure to adequately control
referred for colonoscopy. People confounding
aged 50-74 from 1997 to 2012 Incomplete or inadequately
were classified for exposure to short follow-up
screening according to age and Differences in important
period. Furthermore, two open prognostic factors at baseline
cohorts-one never screened
(aged 50-69 in 1997) and one
invited for screening (aged 50-69
in 2005)-were followed up for 8
years to measure morality.
Zorzi, M., et al., To evaluate the Observational Study; In the The 50-74-year-old Compared with 1995–2000, 2006–2011 Study Limitations =
2015, Gut impact of FIT- Veneto Region (Italy), biennial resident population of mortality rates were 22% lower in the ESA None
based screening FIT-based screening programmes the early screening and than in the LSA (rate ratio (RR) = 0.78; 95% Non-Randomized Studies
programmes on that invited 50–69-year-old late screening areas CI 0.68 to 0.89). The reduction was larger Failure to develop and apply
CRC morality residents were introduced in were 274,266 and in women (RR=0.64; CI 0.51 to 0.80) than appropriate eligibility criteria
different areas between 2002 and 348,674 subjects in men (RR=0.87; CI 0.73 to 1.04). Flawed measurement of
2009. Study compared CRC both exposure and outcome
respectively.
mortality rates from Failure to adequately control
1995 to 2011 between the areas confounding
where screening started in 2002– Incomplete or inadequately
2004 (early screening areas (ESA)) short follow-up
and areas that introduced the Differences in important
screening in 2008–2009 (late prognostic factors at baseline
screening areas (LSA)) using
Poisson regression models.
Available data on CRC incidence
rates (1995–2007) and surgical
resection rates (2001–2012) was
also compared.
References:
2. Chiu, H. M., et al. (2015). "Effectiveness of fecal immunochemical testing in reducing colorectal cancer mortality from the One Million Taiwanese Screening Program." Cancer 121(18): 3221-3229.
3. Giorgi Rossi, P., et al. (2015). "Impact of Screening Program on Incidence of Colorectal Cancer: A Cohort Study in Italy." American Journal of Gastroenterology 110(9): 1359-1366.
4. Zorzi, M., et al. (2015). "Impact on colorectal cancer mortality of screening programmes based on the faecal immunochemical test." Gut 64(5): 784-790.
24
interventions, or outcomes
subpopulations?
varied)
Outcome: Mortality; Modality: Sigmoidoscopy
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations
(PICO question is quite different
Total # of Studies: 4 # of Systematic Reviews: 4 # of RCTs: 0 # of Non-Randomized Studies: 0 # of Diagnostic Studies: 0
Canadian Task To synthesize Systematic Review with meta- 4 RCTs; Combined The length of follow-up across four studies Study Limitations = regard to population,
Force on evidence on the analysis sample of 413,955 14 ranged from six years to 11.9 years. The None intervention, comparison, or
Preventive Health, benefits and harms (165,333 [I]; 248,622 meta-analysis of primary screening with Systematic Review outcome)
2016, CMAJ of screening and [C]). All studies included flexible sigmoidoscopy showed a relative Review did not address focused
Canadian Medical the test properties a mixed gender reduction of 28% in CRC mortality with a clinical question Studies are imprecise (When
Association of effective population. Three pooled RR of 0.72 (95%CI, 0.65, 0.81, Search was not detailed or studies include few patients and
Journal screening methods studies included I2=0%); the ARR was 1,176/million (830- exhaustive
few events and thus have wide
for asymptomatic participants with ages 1,486 fewer). The NNS for the outcome of Quality of the studies was not
adults who are not ranging from 55 to 64 CRC mortality was 850 (95%CI, 673-1205). appraised or studies were of low confidence intervals and the
at high risk for years and one study quality results are uncertain)
colorectal cancer. included participants Methods and/or results were
with ages 55 to 74 inconsistent across studies Publication Bias
years. (e.g. pharmaceutical company
sponsors study on effectiveness
Holme, O., 2017, To compare the Systematic Review; Pooled 287,928 individuals A total of 373 individuals in the screening Study Limitations =
of drug, only small, positive
BMJ effectiveness of analysis of randomised trials were included in the group and 740 in the usual care group died None
(the US Prostate, Lung, pooled analysis; 115,139 from colorectal cancer. Overall, colorectal Systematic Review studies found)
flexible
Colorectal and Ovarian cancer randomised to cancer mortality was reduced by 27% Review did not address focused
sigmoidoscopy in
screening trial screening and 172,789 (relative risk 0.73; 95% confidence interval clinical question Increase Quality Rating if:
screening for (PLCO), the Italian Screening to usual care. 0.64 to 0.83), 33% in men (0.67; 95% 0.57 Search was not detailed or Large Effect
colorectal cancer by for Colon and Rectum trial to 0.80), and 18% in women (0.82; 0.67 to exhaustive Dose-response gradient
patient sex and age (SCORE), and the Norwegian 1.00, P=0.048). When the analyses were Quality of the studies was not Plausible confounders or
Colorectal Cancer restricted to the 55-64 year age group, appraised or studies were of low
other biases increase certainty
Prevention trial (NORCCAP)). colorectal cancer mortality was statistically quality
significantly reduced in men (0.70; 0.57 to Methods and/or results were of effect
0.86) and in younger women (0.68; 0.47 to inconsistent across studies
25
0.98), but not in women aged 60 years and Quality (certainty) of evidence
older (1.07; 0.77 to 1.48). for studies as a whole:
High
Moderate
Low
Very Low
Tang, V., et al., To determine the Systematic Review with meta- Four RCTs, total = Studies were similar for patients’ age (50- Study Limitations =
2015, BMJ time to benefit of analysis 459,814 74 years), length of follow-up (11.2-11.9 Systematic Review
using flexible years), and relative risk for colorectal Review did not address focused
sigmoidoscopy for cancer related mortality (0.69-0.78 with clinical question
colorectal cancer flexible sigmoidoscopy screening). For Search was not detailed or
screening every 1000 people screened at five and 10 exhaustive
years, 0.3 and 1.2 colorectal cancer related Quality of the studies was not
deaths, respectively, were prevented. It appraised or studies were of low
took 4.3 years (95% confidence interval quality
2.8 to 5.8) to observe an absolute risk Methods and/or results were
reduction of 0.0002 (one colorectal cancer inconsistent across studies
related death prevented for every 5000
flexible sigmoidoscopy screenings).
It took 9.4 years (7.6 to 11.3) to observe an
absolute risk reduction of 0.001 (one
colorectal cancer related death prevented
for every 1000 flexible sigmoidoscopy
screenings).
26
Tinmouth, J., To conduct a meta- Systematic Review with Meta- 1 systematic review and Study Limitations =
2016, Canadian analysis quantifying Analysis 4 RCTs None
Journal of the magnitude of Systematic Review
Gastroenterology protection by Review did not address focused
& Hepatology colonoscopy, with clinical question
diagnostic exhaustive
indications, against The magnitude of the effect on CRC Quality of the studies was not
CRC in patients with mortality (RR, 0.72; 95%CI, 0.65 to 0.80) appraised or studies were of low
non-malignant quality
findings and Methods and/or results were
demonstrating the inconsistent across studies
potentially more
marked effect of
screening over
diagnostic
colonoscopy
References:
2. Holme, O., et al. (2017). "Effectiveness of flexible sigmoidoscopy screening in men and women and different age groups: pooled analysis of randomised trials." BMJ 356: i6673.
3. Tang, V., et al. (2015). "Time to benefit for colorectal cancer screening: survival meta-analysis of flexible sigmoidoscopy trials.[Erratum appears in BMJ. 2015;350:h2228; PMID: 25910493]." BMJ 350: h1662.
subpopulations?
Outcome: Mortality; Modality: Fecal Occult Blood Test (FOBT) Studies are indirect
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations (PICO question is quite different
Canadian Task To synthesize Systematic Review with meta- 4 RCTs; Combined For colorectal specific mortality, the meta- Study Limitations = comparison, or outcome)
Force on evidence on the analysis sample of 313,180 analysis for screening with gFOBT None
Preventive benefits and harms (156,737 [I - compared to no screening found RR 0.82 Systematic Review
27
Health, 2016, of screening and intervention]; 156,443[C (95%CI, 0.73, 0.92, I2=67%), with an ARR Review did not address Studies are imprecise (When
CMAJ Canadian the test properties - control]). Two studies 2,654/ million (1,128-4,010 fewer). The focused clinical question studies include few patients and
Medical of effective included participants number needed to screen (NNS) was 377 Search was not detailed or few events and thus have wide
Association screening methods with ages ranging from (95%CI, 249-887). exhaustive
confidence intervals and the
Journal for asymptomatic 45 to 75 years, one Quality of the studies was
adults who are not study included not appraised or studies were of results are uncertain)
at high risk for participants ages 50 to low quality
colorectal cancer. 80 years and one study Methods and/or results were Publication Bias
included participants inconsistent across studies (e.g. pharmaceutical company
ages 60 to 64 years. The sponsors study on effectiveness of
Tinmouth, J., To conduct a meta- Systematic Review with Meta- 2 RCTs The overall certainty of the evidence was Study Limitations =
found)
2016, Canadian analysis quantifying Analysis; PubMed, EMBASE, and high, suggesting a definite reduction in None
Journal of the magnitude of conference abstracts were CRC-related mortality. The magnitude of Systematic Review
Gastroenterolog protection by searched through 30 April 2015. the effect was small with a RR 0.87; 95% CI Review did not address Increase Quality Rating if:
y & Hepatology colonoscopy, with The primary outcomes were 0.82 to 0.92. focused clinical question Large Effect
screening and overall CRC incidence and Search was not detailed or Dose-response gradient
diagnostic mortality. Pooled relative risks exhaustive Plausible confounders or other
indications, against (RRs) and 95% confidence Quality of the studies was biases increase certainty of effect
CRC in patients with intervals (CIs) were calculated not appraised or studies were of
non-malignant using random-effect models. low quality
findings and Methods and/or results were Quality (certainty) of evidence for
demonstrating the inconsistent across studies studies as a whole:
potentially more High
marked effect of Moderate
screening over Low
diagnostic Very Low
colonoscopy
Ananda, S., et al., To analyze the Prospective Study; Data on 3,743 patients Of 1,930 patients aged between 50 and 70 Study Limitations =
2016, Internal impact of the consecutive patients enrolled years, 141 (7.3%) had a NBCSP detected None
Medicine Journal National Bowel into a prospective, cancer, 1441 (74.7%) presented with Non-Randomized Studies
Cancer Screening comprehensive, symptoms and 266 (13.8%) were Failure to develop and apply
Programme multidisciplinary database at six diagnosed through screening outside of appropriate eligibility criteria
(NBCSP) launched Victorian hospitals were the NBCSP. Based on the American Society Flawed measurement of
in Australia in 2006 examined. Clinicopathologic and of Anaesthesiology score, the NBCSP both exposure and outcome
that mailed outcome data were compared patients were fitter. They had an earlier Failure to adequately control
invitations to for NBCSP and symptomatic stage of diagnosis and were more likely to confounding
people turning 55 presentation patients. be female and less likely to have Incomplete or inadequately
or 65 years to lymphovascular invasion or to present as short follow-up
undertake an emergency. NBCSP detected patients Differences in important
immunochemical- had a lower rate of recurrence (HR 0.17, P prognostic factors at baseline
28
based FOBT = 0.0001) and fewer deaths (HR 0.19, P =

screening. 0.005).
References:
3. Ananda, S., et al. (2016). "Survival impact of the Australian National Bowel Cancer Screening Programme." Internal Medicine Journal 46(2): 166-171.
subpopulations?
Outcome: Mortality; Modality: Fecal Immunochemical Test (FIT) vs Fecal Occult Blood Test (gFOBT) (PICO question is quite different
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations from the available evidence in
regard to population, intervention,
Study Methods
Total # of Studies: 1 # of Systematic Reviews: 1 # of RCTs: 0 # of Non-Randomized Studies: 0 # of Diagnostic Studies: 0 comparison, or outcome)
Tinmouth, J., To conduct a meta- Systematic Review with 1 Systematic Review While there were well-designed Study Limitations =
2016, Canadian analysis quantifying Meta-Analysis and 10 RCTs randomized controlled trials (RCTs) None Studies are imprecise (When
Journal of the magnitude of comparing FIT with gFOBT, the outcomes Systematic Review studies include few patients and
Gastroenterology protection by of these trials (participation and detection Review did not address few events and thus have wide
& Hepatology colonoscopy, with rates) were considered to be less focused clinical question
screening and important than CRC related mortality. Search was not detailed or
results are uncertain)
diagnostic However, it was anticipated that the exhaustive
indications, against reduction in CRC-related mortality and the Quality of the studies was
CRC in patients with complications resulting from screening not appraised or studies were of Publication Bias
non-malignant with FIT would be at least equivalent to low quality (e.g. pharmaceutical company
findings and those observed from screening with Methods and/or results were sponsors study on effectiveness of
demonstrating the gFOBT. inconsistent across studies drug, only small, positive studies
potentially more
found)
marked effect of
screening over
diagnostic Increase Quality Rating if:
colonoscopy Large Effect
29

studies as a whole:
High
Moderate
Low
Very Low
References:

Berger, B.M., et al., 2016, American To demonstrate that the US Preventive Services Descriptive analysis of USPSTF modeling of the clinical impact of
Journal of Managed Care Task Force (USPSTF) modeling screening outcomes CRC screening strategies. These scenarios involved different
included CRC mortality reduction. modalities (hsFOBT, FIT, and mt-sDNA), intervals (2y, 3y, and
every 5 years), and screening age groups (50-75, 50-80, 50-85,
55-75, 55-80, and 55-85 years). All scenarios assumed 100%
adherence to the screening modality and interval. Screening
scenarios were plotted against life years gained (LYG) and
colonoscopies performed to create an “efficiency frontier,” a line
connecting the most efficient screening scenarios based on the
trade-off between LYG and colonoscopies. By excluding annual
stool tests, the CISNET’s analysis of the “efficiency frontier” was
recreated and compared these screening scenarios with each
other and with 10y colonoscopy screening. Per CISNET, tests
were recommended if they were within 98% of the LYG at the
equivalent point on the efficiency frontier and generated at least
90% of the LYG generated by 10y screening colonoscopy.
Brenner, H., et al, 2015, Clinical To quantify the effects of screening colonoscopy on Analyzed data from screening colonoscopies (conducted on Overall, approximately 180,000 CRCs (1/28 screening
Gastroenterology & Hepatology prevention, early detection, and overdiagnosis of individuals 55–79 years old from 2003 through 2012) available colonoscopies) were estimated to have been prevented,
colorectal cancer (CRC) in the 10 years since its through the national screening colonoscopy registry. CRCs and more than 40,000 CRCs (1/121 screening
introduction in Germany prevented, detected earlier than they would have been without colonoscopies) were detected earlier than they would
screening, and overdiagnosed (cancers detected at screening have been without screening, compared with
colonoscopy that would not have become clinically manifest approximately 4500 overdiagnoses (1/1089 screening
during the patient’s lifetime) were estimated by Markov models. colonoscopies). Almost all CRCs prevented or detected
Model parameters included sex-specific and age-specific findings earlier than they would have been without screening
at screening colonoscopy; mortality; rates of transition from resulted from screening colonoscopies performed on
nonadvanced to advanced adenoma, advanced adenoma to individuals up to 75 years old (97% and 89%,
preclinical cancer, or preclinical cancer to clinically manifest respectively), whereas 28% of overdiagnoses occurred
cancer; and protection from screening colonoscopy. For each from screening colonoscopies of individuals older than 75
iteration and each transition, mortality was accounted for and years old.
obtained from general population life tables for the year 2010.
30
Geurts, S.M., et al., 2015, British Journal To project the impact of phasing once-only flexible Simulated the life-course of English residents reaching age 55 By mid-2030, 8.5 million individuals will have been invited
of Cancer sigmoidoscopy (FS) at age 55 into the England from 2013 onwards. Model inputs included population numbers, for once-only FS screening. Adding FS to gFOBT screening
National Health Service Bowel Screening invitation rates and CRC incidence and mortality rates. The is estimated to 2,207 (12%) deaths by mid-2030. If FS
Programme (NHSBCSP), augmenting biennial gFOBT impact of gFOBT and FS alone on CRC incidence and mortality uptake is 38% or 71%, respectively, an extra 1,691 (9%) or
at ages 60-74, on CRC cases and deaths prevented were derived from published trials, assuming an uptake of 50% 3,154 (17%) deaths will be prevented by mid-2030.
in England by mid-2030. for FS and 57% for gFOBT. For FS plus gFOBT, the gFOBT effect
was assumed to be 75% of the gFOBT alone impact.
References:
1. Berger, B. M., et al. (2016). "USPSTF colorectal cancer screening guidelines: an extended look at multi-year interval testing." American Journal of Managed Care 22(2): e77-81.
2. Geurts, S. M., et al. (2015). "Likely effect of adding flexible sigmoidoscopy to the English NHS Bowel Cancer Screening Programme: impact on colorectal cancer cases and deaths." British Journal of Cancer 113(1): 142-149.
31
Question #2: What are the test performance characteristics (eg, sensitivity and specificity) of the following screening tests (alone or in combination) for detecting (a) colorectal cancer,
(b) advanced adenomas, and (c) adenomatous polyps based on size: colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool screening tests, high-sensitivity
guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA? Do test performance characteristics vary by important
subpopulations?
Primary Literature:

Overall Summary of Diagnostic Accuracy of Screening
32
33
PICO Question: What are the test performance characteristics (eg, sensitivity and specificity) of the following screening tests (alone or in Lower Quality Rating if:
combination) for detecting (a) colorectal cancer, (b) advanced adenomas, and (c) adenomatous polyps based on size: colonoscopy, flexible Studies inconsistent (wide
sigmoidoscopy, computed tomographic colonography, stool screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or variation of treatment effect across
multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA? Do test performance characteristics vary by important subpopulations?
or outcomes varied)
Outcome: FIT Characteristics
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations Studies are indirect
Study Methods (PICO question is quite different
Total # of Studies: 13 # of Systematic Reviews: 2 # of Non-Randomized Studies: 11 from the available evidence in
Canadian Task To synthesize Systematic Review 40 studies ( 38 cohorts Result are reported in median with range. Study Limitations =
Force on evidence on the and 2 case control) None comparison, or outcome)
The overall sensitivity for iFOBT was
Preventative benefits and harms 81.5% (53.3%-100%) and a specificity of Systematic Review
Healthcare (2016) of screening and the Review did not address Studies are imprecise (When
95.0% (87.2%-96.9%) with median PPV
CMAJ Canadian test properties of focused clinical question studies include few patients and few
Medical effective screening 7.35% (4.0%-10.8%), NPV 100% (99.7%- Search was not detailed or events and thus have wide
Association methods for 100%) and NNS 209 (41-430). exhaustive confidence intervals and the results
Journal asymptomatic adults Quality of the studies was
are uncertain)
who are not at high not appraised or studies were of
risk for colorectal low quality
cancer. Methods and/or results were Publication Bias
inconsistent across studies (e.g. pharmaceutical company
found)

Hirai et al. (2016) To assess in a meta- Systematic Review with meta- Thirteen studies, with Proximal CRC: Study Limitations =
Large Effect
Alimentary analysis, the analysis 11 FIT cohorts with 20, Pooled Sensitivity: 71.1% (60.9–79.6%) None
Pharmacology & diagnostic accuracy 148 patients Pooled Specificity: 95.2% (92.1–97.1%) Systematic Review Dose-response gradient
Therapeutics of FOBTs for relative Pooled Positive LR 14.7 (9.5–22.7) Review did not address Plausible confounders or other
detection of CRC Pooled Negative LR 0.3 (0.2–0.4) focused clinical question biases increase certainty of effect
according to AUC 91% Search was not detailed or
anatomical location exhaustive Quality (certainty) of evidence for
of CRC. Distal CRC: Quality of the studies was studies as a whole:
Pooled Sensitivity 79.0% (69.2–86.3%) not appraised or studies were of
High
Pooled Specificity 95.2% (92.1–97.1%) low quality
Pooled Positive LR 16.4 (10.5–25.5) Methods and/or results were Moderate
Pooled Negative LR 0.2 (0.1–0.3) inconsistent across studies Low
AUC 95% Very Low
34
Huang, Y., et al. To evaluate and Prospective cohort study 1020 participants Study Limitations =
(2016). European understand the where three fecal samples None
Journal of Cancer superiority of were collected from each Diagnostic Studies
Prevention quantitative FIT, a participant by one optimized Patients not enrolled in
representative and two common sampling consecutive or random manner
randomly selected devices, and then tested by Case-control study
population in China both quantitative and No independent, blind
was invited for CRC qualitative FITS. Colonoscopy comparison between index test
screening. The was provided independently to and reference test
performances of five all participants. Not all patients received
featured screening reference test
strategies was Reference test not likely to
compared. correctly classify target
condition
Failure to include all patients
in analysis
Shahidi, N., et al. To assess FIT Prospective cohort study 20,322 participants with At the BCCSP FIT cut-off of ≥10 𝜇𝜇g/g (≥50 Study Limitations =
(2016). Canadian performance among where a single quantitative FIT a positive FIT who then ng/mL) the PPV for CRC, HRAs, all None
Journal of average-risk with a cut-off of ≥10 𝜇𝜇g/g (≥50 underwent a adenomas, and all neoplasia were 2.3%, Diagnostic Studies
Gastroenterology participants ng/mL) was used to screen colonoscopy. 20.4%, 52.0%, and 54.2%, respectively. Patients not enrolled in
& Hepatology of the British participant. Participants with consecutive or random manner
Columbia Colon Case-control study
35
Screening Program positive FIT results were When comparing a cut-off of ≥10 𝜇𝜇g/g No independent, blind
(BCCSP) referred for colonoscopy. (≥50 ng/mL) to ≥20 𝜇𝜇g/g (≥100 ng/mL) the comparison between index test
PPV for CRC, HRAs, and all neoplasia and reference test
increased by absolute values of 1.5%, Not all patients received
6.5%, and reference test
5.7%, respectively. The frequency of Reference test not likely to
missed CRCs and HRAs was 61 (13.6%) correctly classify target
and 1300 (32.4%), respectively. condition
in analysis
Jensen, C. D., et al. To asses FIT Retrospective cohort study 323,349 participants Study Limitations =
(2016). Annals of performance who completed the first None
Internal Medicine characteristics over 4 round of FIT and were Diagnostic Studies
rounds of annual followed-up for up to 4 Patients not enrolled in
screening. screening rounds. consecutive or random manner
Case-control study
No independent, blind
comparison between index test
and reference test
Not all patients received
reference test
Reference test not likely to
correctly classify target
condition
in analysis
36
Kapidzic, A., et al. The study assessed A prospective cohort study 10,008 screenees (age, In the first round, differences in PPV for Study Limitations =
(2015). Clinical FIT performance in was performed, in which 50–74 y) advanced neoplasia between men and None
Gastroenterology men and women. subjects were invited for first- were approached for women were significant only at a cut-off Diagnostic Studies
& Hepatology round screening and first-round screening level of 15 mg Hb/g feces (men: 51%; Patients not enrolled in
for second-round screening and 8316 95% CI, 45–58; women: 40%; 95% CI, 32– consecutive or random manner
with a single FIT. Subjects with screenees (age, 51–74 48; P= .032) Case-control study
a hemoglobin (Hb) y) were approached for No independent, blind
level of 10 mg hemoglobin second round In the second round, no differences comparison between index test
(Hb)/g (or ‡50 ng/mL) feces or screening. in PPV between men and women were and reference test
higher were referred for observed. Not all patients received
colonoscopy. reference test
The test characteristics were The NNscope for advanced neoplasia and Reference test not likely to
assessed by sex for a range of CRC was similar in both sexes. correctly classify target
FIT cut-off values. condition
In both rounds, sex was not associated Failure to include all patients
significantly with the PPV for advanced in analysis
neoplasia after adjusting for age.
A lower NNscreen to detect 1 advanced

neoplasia was seen in men at cut-off level
of 10 mg Hb/g feces: men, 44; 95% CI, 34–
59; women, 66; 95% CI, 50–91; P =.046;
cut-off level of 20 mg Hb/g feces: men,
59;95%CI, 44–83; women, 95;95%CI, 67–
143; P=.045; cut-off level of 25 mg Hb/g
feces: men, 65; 95%CI, 48–91; women,
115; 95%CI, 77–167; P=.028, respectively.
A significantly higher FPR in men was

found in both rounds at a cut-off level of
10 mg Hb/g feces (FPR round I, 6.3% in
men vs 4.1% in women, P < .001; FPR
round II, 4.6% in men vs 3.3% in women,
P=.017).
No differences were seen when

comparing the fecal hemoglobin
concentrations between true-positive
men and women
37
Symonds, E. L., et Examine whether Retrospective cohort study. 13,433 subjects, 21,929 Temperature: Study Limitations =
al. (2015). Journal demographic, Participant demographics, correct screening packs with temperature expressed as a None
of Medical pathological, temperature on sample continuous variable, a 1degree C Diagnostic Studies
Screening behavioural, postage day, and previous increase in temperature decreased the Patients not enrolled in
and environmental screening were recorded. chance of a positive result by 1.8% (OR consecutive or random manner
factors affected Outcomes from 0.982, 95% CI 0.973–0.991) Case-control study
haemoglobin colonoscopy performed within No independent, blind
concentration and a year following FIT were Gender: comparison between index test
positive rates where collected. Multivariate logistic Positive rates were higher in males. OR and reference test
FIT samples are regression identified 1.47 ( 95% CI 1.30-1.65 p<.001) Not all patients received
mailed. significant predictors reference test
of test positivity. Colonoscopy Outcome: Reference test not likely to
Likelihood of significant neoplasia was correctly classify target
higher after a positive FIT compared with condition
a negative test result (OR 2.21, Failure to include all patients
95% CI 1.65–2.98) in analysis
Multivariate logistic regression

analysis indicated that the site of the
significant neoplasia was a predictor for
FIT positivity, with distal neoplasia
significantly
more likely to result in a positive FIT than
proximal neoplasia (OR 2.10, 95% CI 1.17–
3.78)
38
Wong, M. C., et al. To identify Retrospective database review 4482 participants who The sensitivity, specificity, positive Study Limitations =
(2015) demographic factors of prospectively collected data. underwent both FIT and predictive values, and negative None
Gastrointestinal associated with false- colonoscopy in the first predictive values for advanced neoplas ia Diagnostic Studies
Endoscopy positive and false- year and 857 were 33.1%, 91.9%, 19.0%, and 96.0%, Patients not enrolled in
negative FIT results underwent colonoscopy respectively. consecutive or random manner
during CRC screening after negative FIT Participants 66 to 70 years of age had Case-control study
results for 3 years. higher sensitivity, whereas older age, No independent, blind
smoking, and use of aspirin/ nonsteroidal comparison between index test
anti-inflammatory drugs were associated and reference test
with lower specificity. Not all patients received
reference test
The rates of false-positive and false-negat Reference test not likely to
ive results were 8.1% and 66.9%, correctly classify target
respecively. Older age (66-70 years; condition
adjusted odds ratio (AOR 1.95; 95% (CI], Failure to include all patients
1.35-2.81; P < .001), smoking (AOR in analysis
1.68; 95% CI, 1.08-2.61; P = .020) , and the
presence of polypoid adenoma (AOR 1.71;
95% CI, 1.14-2.57; P = .009) were
associated with false -positive results.
Younger participants (AOR for elderly

participants 0.31) and the use of aspirin/
nonsteroidal anti-inflammatory drugs
(AOR 4.44) in participants with 1 FIT
with negative results and the absence of
high-grade dysplasia (AOR for presence
0.41) were associated with false-
negative results.
39
Wong, M. C., et al. To compare the Prospective cohort study 5343 subjects (50-70 Sensitivity Study Limitations =
(2015). Clinical accuracy of a years old) who received FIT detected distal advanced adenoma None
Gastroenterology qualitative FIT in 2 FITs (Hemosure; with 39.7% sensitivity (95% confidence Diagnostic Studies
& Hepatology identifying patients cutoff interval [CI], 32.0%-48.0%) vs proximal Patients not enrolled in
with proximal vs value, 10 mg advanced adenoma with 25.0% sensitivity consecutive or random manner
distal advanced hemoglobin/g feces) (95% CI,17.3%-34.6%; P [ .014), distal Case-control study
neoplasia and before colonoscopy advanced neoplasia with 40.0% sensitivity No independent, blind
evaluate whether (95% CI, 32.5%L47.9%) vs proximal comparison between index test
analysis of 2 advanced neoplasia with 27.9% sensitivity and reference test
specimens (95% CI, 20.0%-37.4%; P [ .039), and any Not all patients received
performed better distal adenoma ‡10 mm, irrespective of reference test
than analysis of 1 other lesion characteristics, Reference test not likely to
specimen. with 39.5% sensitivity (95% CI,31.0%- correctly classify target
48.7%) vs proximal adenoma with 25.3% condition
sensitivity (95% CI, 16.5%-36.6%; P [ .038). Failure to include all patients
in analysis
Specificity
The specificity of FIT in detecting CRC was
similar between the proximal and distal
colon. FIT detected distal lesions with
higher PPV than proximal lesions. One FIT
detected advanced neoplasia with 31.8%
sensitivity (95% CI, 25.9%-38.4%) and
92.4% specificity (95% CI, 91.6%-93.2%),
whereas 2 FITs detected advanced
neoplasia with 34.1% sensitivity (95% CI,
28.0%-40.8%; P [ .617) and 91.9%
specificity (95% CI, 91.0%-92.7%; P [ .327).
FIT detected distal advanced neoplasia
with greater sensitivity and higher PPV
than proximal advanced neoplasia.
40
Bujanda, L., et al. To evaluate the Prospective cohort study with A total of 238,647 Multivariate analysis confirmed that, in Study Limitations =
(2015) European characteristics of CRC two rounds of screening. individuals participated the second round, CRC diagnosed was None
Journal of detected in a second in the first round of FIT more often proximal (hazard ratio vs. first Diagnostic Studies
Gastroenterology round of FIT screening and 69,193 round, 2.4; 95% confidence interval, 1.3– Patients not enrolled in
& Hepatology screening after individuals in the 4.4; P =0.003) and the concentration of consecutive or random manner
negative results in a second round after a Hb/g faeces was lower (hazard ratio vs. Case-control study
first round. first negative result. first round, 2.1; 95% confidence interval, No independent, blind
1.3–3.5; P =0.003). comparison between index test
and reference test
Not all patients received
reference test
condition
in analysis
41
Chausserie, S., et To compare the Prospective cohort study. Each 18,290 subjects The performance of tests for detection of Study Limitations =
al. (2015). seasonal variation patient sent in 3 consecutive advanced neoplasia was compared None
International performance of FIT stool samples. gFOBT was according to seasons using Receiver Diagnostic Studies
Journal of Cancer and gFOBT for CRC performed on all 3 samples Operating Characteristics (ROC) curves, at Patients not enrolled in
screening and FIT was performed on the various FIT cut-off values. The consecutive or random manner
last 2 samples. positivity rate of FIT was significantly Case-control study
lower in the summer compared with No independent, blind
other seasons (2.3% versus 3.0%, p = comparison between index test
0.03), whilst the positivity rate of gFOBT and reference test
increased in the autumn (1.8% versus Not all patients received
1.5%, p = 0.11). reference test
FIT was clinically more effective than correctly classify target
gFOBT over the four season-specific ROC condition
curves. At the cut-off concentration used Failure to include all patients
in the study, the season-specific FIT/ in analysis
gFOBT ratios for true positive rates were:
2.8 (Autumn), 2 . 5 (Winter), 3.0 (Spring),
3.7 (Summer), and for false positive rates:
1.2 (Autumn), 1.5 (Winter), 1.8 (Spring), 0
.9 (Summer).
The seasonal variations of performance

of FIT led to improved gain in specificity
in the summer, without a decrease in
gain in sensitivity compared with gFOBT.
42
Redwood, D. G., To assess the Prospective, cross-sectional 661 participants Sensitivity by MT-sDNA increased with Study Limitations =
et al. (2016). accuracy of a study of asymptomatic adults adenoma size (to 80% for lesions 2:3 cm; None
Mayo Clinic multitarget stool undergoing screening P=.01 for trend) and substantially Diagnostic Studies
Proceedings DNA test (MT-sDNA) colonoscopy. exceeded FIT sensitivity at all adenoma Patients not enrolled in
compared with FIT sizes. For sessile serrated polyps larger consecutive or random manner
for detection of than 1 cm (n= 9), detection was 67% by Case-control study
screening-relevant MT-sDNA vs 11 % by FIT (P= .07) For CRC No independent, blind
colorectal neoplasia (n= l O) , detection was 100% by MT-sDNA comparison between index test
(SRN) in Alaska vs 80% by FIT (P= .48). Specificities were and reference test
Native people 93% and 96%, respectively (P= .03). Not all patients received
reference test
condition
in analysis
43
Jin, P., et al. Evaluate the Prospective cohort study. 476 patients The sensitivity and specificity of SEPT9 Study Limitations =
(2015). Journal of performance of the Patients with CRC, for CRC were 74.8% (95% confidence None
Gastroenterology second-generation adenomatous polyps, interval [CI]: 67.0–81.6%) and 87.4% (vs Diagnostic Studies
& Hepatology SEPT9 assay for the hyperplastic polyps, and a non-CRC, 95% CI: 83.5–90.6%), Patients not enrolled in
detection of healthy control group were respectively. SEPT9 was positive in 66.7% consecutive or random manner
colorectal neoplasm, included. The clinical status of of stage I, 82.6% of stage II, 84.1% of Case-control study
and compared it with all patients was verified by stage III, and 100% of stage IV CRCs. No independent, blind
FIT. colonoscopy. In all patients, comparison between index test
peripheral blood samples were The sensitivity of SEPT9 for advanced and reference test
taken for SEPT9 testing. For adenomas was 27.4% (95% CI:18.7– Not all patients received
177 patients, both SEPT9 and 37.6%). reference test
FIT were performed. Reference test not likely to
SEPT9 showed better performance in correctly classify target
CRC detection than FIT, but similar condition
among advanced adenomas. Failure to include all patients
in analysis
References:
1. Canadian Task Force on Preventive Health, C., et al. (2016). "Recommendations on screening for colorectal cancer in primary care." CMAJ Canadian Medical Association Journal 188(5): 340-348. (Systematic Review)
2. Hirai, H. W., et al. (2016). "Systematic review with meta-analysis: faecal occult blood tests show lower colorectal cancer detection rates in the proximal colon in colonoscopy-verified diagnostic studies." Alimentary
Pharmacology & Therapeutics 43(7): 755-764.
3. Huang, Y., et al. (2016). "Optimizing sampling device for the fecal immunochemical test increases colonoscopy yields in colorectal cancer screening." European Journal of Cancer Prevention 25(2): 115-122.
4. Shahidi, N., et al. (2016). "Correlating Quantitative Fecal Immunochemical Test Results with Neoplastic Findings on Colonoscopy in a Population-Based Colorectal Cancer Screening Program: A Prospective Study." Canadian
Journal of Gastroenterology & Hepatology 2016: 4650471.
5. Jensen, C. D., et al. (2016). "Fecal Immunochemical Test Program Performance Over 4 Rounds of Annual Screening: A Retrospective Cohort Study." Annals of Internal Medicine 164(7): 456-463.
6. Kapidzic, A., et al. (2015). "Gender Differences in Fecal Immunochemical Test Performance for Early Detection of Colorectal Neoplasia." Clinical Gastroenterology & Hepatology 13(8): 1464-1471.e1464.
7. Symonds, E. L., et al. (2015). "Factors affecting faecal immunochemical test positive rates: demographic, pathological, behavioural and environmental variables." Journal of Medical Screening 22(4): 187-193.
8. Wong, M. C., et al. (2015). "Factors associated with false-positive and false-negative fecal immunochemical test results for colorectal cancer screening." Gastrointestinal Endoscopy 81(3): 596-607.
9. Wong, M. C., et al. (2015). "Diagnostic Accuracy of a Qualitative Fecal Immunochemical Test Varies With Location of Neoplasia But Not Number of Specimens." Clinical Gastroenterology & Hepatology 13(8): 1472-1479.
10. Bujanda, L., et al. (2015). "Colorectal cancer in a second round after a negative faecal immunochemical test." European Journal of Gastroenterology & Hepatology 27(7): 813-818.
11. Chausserie, S., et al. (2015). "Seasonal variations do not affect the superiority of fecal immunochemical tests over guaiac tests for colorectal cancer screening." International Journal of Cancer 136(8): 1827-1834.
12. Plumb, A. A., et al. (2015). "Effect of faecal occult blood positivity on detection rates and positive predictive value of CT colonography when screening for colorectal neoplasia." Clinical Radiology 70(10): 1104-1109.
13. Redwood, D. G., et al. (2016). "Stool DNA Testing for Screening Detection of Colorectal Neoplasia in Alaska Native People." Mayo Clinic Proceedings 91(1): 61-70.
14. Jin, P., et al. (2015). "Performance of a second-generation methylated SEPT9 test in detecting colorectal neoplasm." Journal of Gastroenterology & Hepatology 30(5): 830-833.
44
or outcomes varied)
Outcome: gFOBT Test Characteristics
Author/Date Purpose of Study Study Design & Sample Outcomes Design Limitations Studies are indirect
Methods (PICO question is quite different
Total # of Studies: 2 # of Systematic Reviews: 1 # of Non-Randomized Studies: 1 from the available evidence in
Canadian Task To synthesize evidence Systematic Review 40 studies ( 38 Result are reported in median with range. The Study Limitations =
Force on on the benefits and cohorts and 2 case None comparison, or outcome)
overall sensitivity of gFOBT is 47.1% (12.9%-
Preventative harms of screening and control) 75.0%) and median specificity of 96.1% (90.1%- Systematic Review
Healthcare the test properties of Review did not address Studies are imprecise (When
98.1%) with PPV of 7.5 (1.5%-15%), NPV of
(2016). CMAJ effective screening focused clinical question studies include few patients and few
Canadian methods for 99.55% (99.5%-99.6%) and NNS 597 (239-936). Search was not detailed or events and thus have wide
Medical asymptomatic adults exhaustive confidence intervals and the results
Association who are not at high risk Quality of the studies was
are uncertain)
Journal for colorectal cancer. not appraised or studies were of
low quality
Methods and/or results were Publication Bias
found)

Azimafousse Evaluate the impact of Analysis of a cross- 98,031 people The PPV for detection of advanced neoplasia Study Limitations =
Large Effect
Assogba, G. F., et repeated gFOBT sectional study focused on was 24.5%, substantially higher in men than None
al. (2015). screening on the PPV for people with a positive women (30.7% vs 17.7%), and it increased with Diagnostic Studies Dose-response gradient
Cancer advanced colorectal gFOBT who were followed age. Advancing age (RR1.28, 95% CI 1.19- 1.39 Patients not enrolled in Plausible confounders or other
Epidemiology neoplasia and their up with a colonoscopy. for every 10-year increase in age),female consecutive or random manner biases increase certainty of effect
distribution according to gender (RR 1.31,95 % Cl 1.19- 1.44),and Case-control study
atomic subsite. subsequent screening ( RR 1.15, 95 % Cl 1.04- No independent, blind Quality (certainty) of evidence for
1.27) were significantly and independently comparison between index test studies as a whole:
associated with detection of proximal and reference test High
adenocarcinoma. The latter was also detected Not all patients received
at an advanced stage more often (RR.1.24, 95 % Moderate
reference test
CI: 1.09- 1.42 ). Early stages of invasive Reference test not likely to Low
adenocarcinoma (stages I and II) was more correctly classify target Very Low
condition
45
likely to be detected in a subsequent than an Failure to include all patients

initial screening (RR 1.07, 95% CI 1.01- 1.13) in analysis
References:
2. Azimafousse Assogba, G. F., et al. (2015). "Impact of subsequent screening episodes on the positive predictive value for advanced neoplasia and on the distribution of anatomic subsites of colorectal cancer: A population-
based study on behalf of the French colorectal cancer screening program." Cancer Epidemiology 39(6): 964-971.
studies, populations, interventions, or
outcomes varied)
Outcome: MtDNA Test Characteristics
Study Methods (PICO question is quite different from
Total # of Studies: 1 # of Systematic Reviews: 1 the available evidence in regard to
population, intervention, comparison,
Zhai, R. L., et al. To evaluate the Systematic Review with meta- 53 studies with 7524 Single-gene test: sensitivity 48%, Study Limitations =
(2016). Medicine diagnostic analysis patients specificity 97%, pooled DOR 20.35 (95% None or outcome)
performance of CI: 17.63-23.49), ROC curve .908 (std Systematic Review
stool DNA for CRC. error .013), LR 9.17 Review did not address Studies are imprecise (When
focused clinical question studies include few patients and few
Multi-gene test: sensitivity 78%, Search was not detailed or events and thus have wide confidence
specificity 93%, pooled DOR 31.64 (95% exhaustive intervals and the results are uncertain)
CI: 25.13-39.84), ROC .934( std. error Quality of the studies was
.011), LR 7.94 not appraised or studies were of
low quality Publication Bias
Methods and/or results were (e.g. pharmaceutical company
inconsistent across studies sponsors study on effectiveness of
found)

Large Effect
46


studies as a whole:
High
Moderate
Low
Very Low
References:
1. Zhai, R. L., et al. (2016). "The Diagnostic Performance of Stool DNA Testing for Colorectal Cancer: A Systematic Review and Meta-Analysis." Medicine 95(5): e2129.
Question #3: a) What are the adverse effects (ie, serious harms) of the different screening tests (either as single application or in a screening program)? b) Do adverse effects vary by
important subpopulations (eg, age)?
Primary Literature:

Overall Summary of Serious Adverse Events of Screening
47
48
PICO Question: a)What are the adverse effects (ie, serious harms) of the different screening tests (either as single application or in a screening Lower Quality Rating if:
program)? b) Do adverse effects vary by important subpopulations (eg, age)? Studies inconsistent (wide
variation of treatment effect
Outcome: Major Bleeding across studies, populations,
Modality: Colonoscopy interventions, or outcomes
varied)
Author/Date Purpose of Study Study Design & Sample Outcomes Design Limitations
Methods Studies are indirect
Total # of Studies: 2 # of Systematic Reviews: 2
Canadian Task To synthesizes the A systematic review of 16 studies which Screening Colonoscopy One uncontrolled Study Limitations = from the available evidence in
Force on benefits and harms of literature with meta- included a total of study reported no cases of bleeding that None regard to population,
Preventative Health screening for colorectal analysis. resulted in hospitalization by number of Systematic Review intervention, comparison, or
et al. (2016) CMAJ colonoscopies 0/324 events; proportion of Review did not address focused outcome)
0.0/1,000 (95%CI, 0.0-11.72). Reported by clinical question
49
Canadian Medical cancer (CRC) in 254,530 patients and number of patients, bleeding that required Search was not detailed or
Association Journal asymptomatic adults. 14,703 colonoscopies hospitalization occurred in 94 of 79,486 exhaustive Studies are imprecise (When
patients 1.08/1,000 (0.85-1.32). Quality of the studies was not studies include few patients and
appraised or studies were of low
few events and thus have wide
Follow-up Colonoscopy Three quality
Methods and/or results were confidence intervals and the
uncontrolled observational studies
reported bleeding requiring hospitalization inconsistent across studies results are uncertain)
by number of colonoscopies. The total
events were 68/14,379, 4.73/1,000 (95%CI, Publication Bias
3.59-5.87). Seven papers reported this (e.g. pharmaceutical company
outcome by number of patients finding 28 sponsors study on effectiveness of
bleeds requiring hospitalization for 25,178,
1.11/1,000 (95%CI, 0.62-1.57).
found)

Large Effect
Plausible confounders or
other biases increase certainty of
effect

studies as a whole:
High
Vermeer, N.C., et To evaluate potential A systematic review of all 24 studies. 2,531,186 The pooled overall risk of major bleeding Study Limitations = Moderate
al. (2017) Cancer harm as a result of mass literature with a meta- total colonoscopies after colonoscopy was 0.8/1000 (95% CI None Low
Treatment Reviews colorectal cancer analysis to examine the .18-1.63) Systematic Review
Very Low
screening in terms of pooled incidence of Review did not address focused
complications after major complications of clinical question
colonoscopy, morbidity colonoscopy. Search was not detailed or
and mortality following exhaustive
surgery, psychological Quality of the studies was not
distress and appraised or studies were of low
inappropriate use of the quality
screening test. Methods and/or results were
inconsistent across studies
References:
2. Vermeer, N. C., et al. (2017). "Colorectal cancer screening: Systematic review of screen-related morbidity and mortality." Cancer Treatment Reviews 54: 87-98.
50
Outcome: Major Bleeding studies, populations, interventions, or
Modality: Flexible Sigmoidoscopy outcomes varied)

the available evidence in regard to
Canadian Task This systematic A systematic review of 2 studies with 149,866 Flexible Sigmoidoscopy (FS) Major Study Limitations = population, intervention, comparison,
Force on review synthesizes literature with meta- flexible sigmoidoscopies bleeding requiring hospitalization from None or outcome)
Preventative the benefits and analysis. primary screening with FS was reported Systematic Review
Health et al. harms of screening by number of patients in two papers. The Review did not address Studies are imprecise (When
(2016) CMAJ for colorectal cancer total events were 14/149,866, 0.00/1,000 focused clinical question studies include few patients and few
Canadian Medical (CRC) in (95%CI, 0.04-0.15). Search was not detailed or
events and thus have wide confidence
Association asymptomatic adults; exhaustive
Journal provides diagnostic Quality of the studies was intervals and the results are
properties for not appraised or studies were of uncertain)
screening tests that low quality
show a positive Methods and/or results were Publication Bias
impact on mortality inconsistent across studies (e.g. pharmaceutical company
or incidence of late
stage CRC; and
answers contextual drug, only small, positive studies
questions such as found)
patient preferences
and values Increase Quality Rating if:
Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
51
Outcome: Anxiety and Depression studies, populations, interventions, or
Modality: Colonoscopy outcomes varied)

Vermeer, N.C., et To evaluate A systematic review 11 studies, 15,447 Five out of seven prospective studies Study Limitations = population, intervention, comparison,
al. (2017) Cancer potential harm as a patients reported an adverse effect on None or outcome)
Treatment result of mass psychological well-being in participants Systematic Review
Reviews colorectal cancer who received a positive test result. Largest Review did not address Studies are imprecise (When
screening in terms effects were observed before the focused clinical question studies include few patients and few
of complications screenings test, in anticipation of and Search was not detailed or
after colonoscopy, shortly after being informed about a exhaustive
morbidity and positive test result. One study reported no Quality of the studies was intervals and the results are
mortality following clinically relevant psychological effect of not appraised or studies were of uncertain)
surgery, participation in the mass CRC screening, low quality
psychological even a decreased anxiety and Methods and/or results were Publication Bias
distress and improvement in some dimensions of health inconsistent across studies (e.g. pharmaceutical company
inappropriate use related quality of life because of receiving a
of the screening negative result. The largest prospective
test. study (n = 3828) reported that patients drug, only small, positive studies
experienced some psychological distress up found)
to six weeks after the colonoscopy. In two
studies the same pattern of declining Increase Quality Rating if:
scores during follow- up were observed Large Effect
when comparing participants tested Dose-response gradient
positive with positive findings at work-up
(true positives) and participants tested
positive with negative findings at work-up biases increase certainty of effect
(false positives).
studies as a whole:
High
Moderate
Low
Very Low
References:
52
Outcome: Anxiety and Depression studies, populations, interventions, or
Modality: Flexible Sigmoidoscopy and FIT outcomes varied)

Total # of Studies: 1 # of RCTs: 1
Kirkoen B, et al. To evaluate the RCT. Participants were invited 3,216 participants Anxiety :The interaction effect for anxiety RCTS population, intervention, comparison,
(2016) British psychological to either flexible sigmoidoscopy completed both of time (baseline and result), screening Lack of blinding or outcome)
Journal of Cancer effects of (FS) screening, (FIT), or no screening and group (FIT and FS), and screening result Lack of allocation
participation in screening (control). Participants questionnaire ( 1,839 in (positive and negative) was not statistically concealment Studies are imprecise (When
the Bowel Cancer were asked to complete a Flexible Sigmoidoscopy significant (P = 0.89). No significant Stopped early for benefit studies include few patients and few
Screening in Hospital Anxiety and group and 1,377 in FIT increase in anxiety with positive test Incorrect analysis of ITT
Norway pilot Depression Scale and a health group) and 2,618 results ( P= .29 for FS and .40 for FIT). Selective reporting of
program related quality of life scale participants in the Negative test results had a significant measures (e.g., no effect intervals and the results are
questionnaire when invited to control group decrease in anxiety from baseline (P<.01 outcome) uncertain)
screening and again when for FIT and FS) Large losses to F/U
receiving screening result. Difference in important Publication Bias
Control group was invited to Depression: Participants receiving positive prognostic factors at baseline (e.g. pharmaceutical company
complete questionnaire only. or negative screening test did not report a studies
statistically significant change in depression
from baseline. ( P=.14 for posFS and .07 for
posFIT and P= .09 for negFS and .63 for found)
negFIT)
No changes observed in the study reached Large Effect
the criteria of clinical relevance. Dose-response gradient

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Kirkoen, B., et al. (2016). "Do no harm: no psychological harm from colorectal cancer screening." British Journal of Cancer 114(5): 497-504.
53
Outcome: Perforation studies, populations, interventions,
Modality: Colonoscopy or outcomes varied)
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations
Study
Total # of Studies: 2 # of Systematic Reviews 2
from the available evidence in
Vermeer, N.C., et To evaluate A systematic review of all 22 studies. 2,456,349 The pooled risk of perforation after Study Limitations = regard to population, intervention,
al. (2017) Cancer potential harm as a literature with meta-analysis colonoscopies colonoscopy was .07/1000 ( 95% CI .0006- None comparison, or outcome)
Treatment result of mass .17) Systematic Review
Reviews colorectal cancer Review did not address Studies are imprecise (When
screening in terms focused clinical question
of complications Search was not detailed or
after colonoscopy, exhaustive events and thus have wide
morbidity and Quality of the studies was confidence intervals and the results
mortality following not appraised or studies were are uncertain)
surgery, of low quality
psychological Methods and/or results Publication Bias
distress and were inconsistent across
inappropriate use studies
of the screening Differences in important
test. prognostic factors at baseline drug, only small, positive studies
found)

Large Effect
54
Canadian Task This systematic Systematic review of the 16 studies: 39,235 Colonoscopy: Eight uncontrolled Study Limitations = Plausible confounders or other
Force on review synthesizes literature with meta-analysis colonoscopies; 116,680 observational studies reported perforation None biases increase certainty of effect
Preventative the benefits and sigmoidoscopies; data for screening colonoscopy; three of Systematic Review
Health et al. harms of screening 340,869 patients those by number of colonoscopies. There Review did not address
(2016) CMAJ for colorectal were total of 16 events for 39,235 focused clinical question
Canadian Medical cancer (CRC) in colonoscopies 0.41/1,000 (95%CI 0.19-0.62). Search was not detailed or studies as a whole:
Association asymptomatic For the five papers that reported by number exhaustive High
Journal adults; provides of patients, there were 45 events in 84,850 Quality of the studies was Moderate
diagnostic patients 0.53/1,000 (95%CI 0.37-0.69). not appraised or studies were Low
properties for of low quality Very Low
screening tests that Methods and/or results
show a positive were inconsistent across
impact on mortality studies
or incidence of late Differences in important
stage CRC; and prognostic factors at baseline
answers contextual
questions such as
patient preferences
and values.
References:
Outcome: Perforation across studies, populations,
Modality: Flexible Sigmoidoscopy interventions, or outcomes varied)
Study
55
Canadian Task This systematic Systematic review of the 16 studies: 39,235 Flexible Sigmoidoscopy: Seven Study Limitations = regard to population, intervention,
Force on review synthesizes literature with meta-analysis colonoscopies; 116,680 uncontrolled observational studies reported None comparison, or outcome)
Preventative the benefits and sigmoidoscopies; perforation for screening with FS. For the Systematic Review
Health et al. harms of screening 340,869 patients three papers that reported number of Review did not address
(2016) CMAJ for colorectal sigmoidoscopies, there were three focused clinical question
Canadian cancer (CRC) in perforations for 116,680 sigmoidoscopies, Search was not detailed or studies include few patients and
Medical asymptomatic with the proportion of perforations being exhaustive few events and thus have wide
Association adults; provides 0.03/1,000 (95%CI, 0.0-0.07). Four papers Quality of the studies was confidence intervals and the results
Journal diagnostic reported perforations by number of patient not appraised or studies were of are uncertain)
properties for this event rate was 4 for 277,421 patients low quality
screening tests that with the proportion 0.01/1,000 (95%CI, 0.0- Methods and/or results were Publication Bias
show a positive 0.03) inconsistent across studies
impact on mortality Differences in important
or incidence of late prognostic factors at baseline sponsors study on effectiveness of
stage CRC; and drug, only small, positive studies
answers contextual found)
questions such as
patient preferences Increase Quality Rating if:
and values. Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
Outcome: Perforation across studies, populations,
Modality: CT Colonography
56
Author/Date Purpose of Study Study Design & Sample Outcomes Design Limitations interventions, or outcomes
Methods varied)
Canadian Task This systematic Systematic review of the 1 study, 11,707 CT Screening CT Colonography: One paper Study Limitations = (PICO question is quite different
Force on review synthesizes literature with meta-analysis colonography reported no perforations for screening None from the available evidence in
Preventative the benefits and with CT colonography with 0 events for Systematic Review
regard to population,
Health et al. harms of screening 11,707 tests 0.0/1,000 (95%CI, 0.0- 0.33) Review did not address
(2016) CMAJ for colorectal cancer focused clinical question intervention, comparison, or
Canadian Medical (CRC) in Search was not detailed or outcome)
Association asymptomatic adults; exhaustive
Journal provides diagnostic Quality of the studies was Studies are imprecise (When
properties for not appraised or studies were of studies include few patients and
screening tests that low quality few events and thus have wide
show a positive Methods and/or results were
impact on mortality inconsistent across studies
or incidence of late Differences in important results are uncertain)
stage CRC; and prognostic factors at baseline
answers contextual Publication Bias
questions such as (e.g. pharmaceutical company
patient preferences sponsors study on effectiveness of
and values.
found)

Large Effect
Plausible confounders or
other biases increase certainty of
effect

studies as a whole:
High
Moderate
Low
Very Low
References:
57
Outcome: Mortality across studies, populations,
Modality: Colonoscopy interventions, or outcomes varied)
Study
Canadian Task This systematic Systematic review of the 4 studies. 111,160 Colonoscopy: Death as a result of Study Limitations = regard to population, intervention,
Force on review synthesizes literature with meta-analysis patients and 38,472 colonoscopy screening was reported in one None comparison, or outcome)
Preventative the benefits and colonoscopies study by number of colonoscopies; total Systematic Review
Health et al. harms of screening events were 12 deaths for 38,472 Review did not address Studies are imprecise (When
(2016) CMAJ for colorectal colonoscopies 0.31/1,000 (95%CI, 0.18- focused clinical question
studies include few patients and
Canadian cancer (CRC) in 0.55). For the two studies reporting this Search was not detailed or
Medical asymptomatic outcome by number of patients, total exhaustive few events and thus have wide
Association adults; provides events were 2/70,828, resulting in a Quality of the studies was confidence intervals and the results
Journal diagnostic proportion of 0.02/1,000 (95%CI, 0.0-0.06). not appraised or studies were of are uncertain)
properties for low quality
screening tests that Methods and/or results were Publication Bias
show a positive inconsistent across studies
impact on mortality Differences in important
or incidence of late prognostic factors at baseline
stage CRC; and drug, only small, positive studies
answers contextual found)
questions such as
patient preferences Increase Quality Rating if:
and values. Large Effect
58
Vermeer, N.C., et To evaluate A systematic review of all 8 studies. 204,640 Reported mortality rates ranged from 0- Study Limitations =
al. (2017) Cancer potential harm as a literature. participants. 3.3% with the largest study reporting no None Quality (certainty) of evidence for
Treatment result of mass mortality rates at 30 days. Systematic Review studies as a whole:
Reviews colorectal cancer
Review did not address High
screening in terms
of complications focused clinical question Moderate
after colonoscopy, Search was not detailed Low
morbidity and or exhaustive Very Low
mortality following Quality of the studies was
surgery, not appraised or studies
psychological were of low quality
distress and Methods and/or results
inappropriate use
were inconsistent across
of the screening
test. studies
Differences in important
prognostic factors at baseline
References:
Outcome: Mortality across studies, populations,
Modality: Flexible Sigmoidoscopy interventions, or outcomes varied)
Author/Date Purpose of Study Study Design & Sample Outcomes Design Limitations
Methods
59
Canadian Task This systematic review Systematic review of the 4 studies. 111,160 Flexible Sigmoidoscopy: Death resulting Study Limitations = regard to population, intervention,
Force on synthesizes the benefits literature with meta- patients and 38,472 from screening with FS was reported in one None comparison, or outcome)
Preventative Health and harms of screening analysis colonoscopies study by the number of patients. There Systematic Review
et al. (2016) CMAJ for colorectal cancer were 6 deaths in 40,332 patients 0.15/1,000 Review did not address
Canadian Medical (CRC) in asymptomatic (95%CI, 0.07-0.32). focused clinical question
Association Journal adults; provides Search was not detailed or studies include few patients and
diagnostic properties for exhaustive few events and thus have wide
screening tests that show Quality of the studies was confidence intervals and the results
a positive impact on not appraised or studies were of are uncertain)
mortality or incidence of low quality
late stage CRC; and Methods and/or results were Publication Bias
answers contextual inconsistent across studies
questions such as patient Differences in important
preferences and values. prognostic factors at baseline sponsors study on effectiveness of
found)

Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:

Knudsen et al., 2016, Clinical Review To inform the USPTF by modeling the benefits, burden, Three models were used (SimCRC, MISCAN, and CRC-SPIN) The lifetime number of harms from screening was low, with
and Education and harms of CRC screening strategies. Each model consisted of a natural history component and a at most 23 per 1000 40-year-olds with colonoscopy screening
screening component, which were used to simulate every 5 years from ages 45-85.
individual life histories from birth to death under alternative
CRC screening strategies.
60
Question #4: Does using shared decision-making when determining appropriate screening test increase the rate of completed CRC screens compared to a LIP prescribed test?
Primary Literature:
PICO Question: Does using shared decision-making when determining appropriate screening test increase the rate of completed CRC screens Lower Quality Rating if:
compared to a LIP prescribed test? Studies inconsistent (wide
Outcome: CRC Screening Completed
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations or outcomes varied)
Study
Total # of Studies: 7 # of RCTs: 2 # of Non-Randomized Studies: 5
Schroy, P.C., et To assess the RCT; Asymptomatic, average-risk 825 participants; 280 Patients in the decision aid–alone group Study Limitations =
al., 2014, Am J impact of decision patients aged 50-75 participants decision aid + YDR, 269 were more likely to have a test ordered None
Prev Med aid-assisted from an urban, academic safety- decision aid alone, and than the control group at the 1-month RCTS from the available evidence in
shared decision net hospital and community 276 control (69.1% vs 60.5%, p<0.035); 3-month Lack of blinding regard to population, intervention,
making (SDM) on health center between were (71.8% vs 62.3%, p=0.019); 6-month Lack of allocation comparison, or outcome)
CRC screening randomized to one of two (77.0% vs 65.2%, p=0.002); and 12-month concealment
uptake intervention arms (decision aid (80.7% vs 71.4%, p=0.011) time points. Stopped early for benefit Studies are imprecise (When
plus personalized risk assessment The decision aid–alone group also was Incorrect analysis of ITT
[YourDiseaseRisk YDR] or decision more likely to have a test ordered than Selective reporting of
events and thus have wide
aid alone) or control arm. The the decision aid plus YDR group at each of measures (e.g., no effect
interventions took place just prior these points, but here the differences outcome) confidence intervals and the results
to a routine office visit with their were only signifıcant at 1 month (69.1% vs Large losses to F/U are uncertain)
primary care providers. 60.4%, p<0.031); 6 months (77.0% vs Difference in important
67.1%, p<0.010); and 12 months (80.7% prognostic factors at baseline Publication Bias
vs 73.6%, p=0.048). The pattern of test (e.g. pharmaceutical company
ordering was similar for the three groups;
regardless of patient preferences,
colonoscopy was the most commonly drug, only small, positive studies
ordered test (range, 79%–81%) followed found)
by FOBT (13%–19%); flexible
sigmoidoscopy (<2%); and barium enema Increase Quality Rating if:
(<2%). Large Effect
Reuland, D.S., et To determine the RCT; Average risk patients 265 participants Intervention participants were more Study Limitations = Dose-response gradient
al., 2017, JAMA combined effect between 50 – 75 years in two likely to complete CRC screening within 6 None
Intern Med of a CRC screening community health center months (68% vs 27%); adjusted- RCTS
decision aid and practices, 1 in North Carolina and difference, 40 percentage points (95% CI, Lack of blinding biases increase certainty of effect
patient navigation 1 in New Mexico, were 29-51 percentage points). The Lack of allocation
compared with randomized 1:1 to intervention or intervention was more effective in concealment Quality (certainty) of evidence for
usual care on CRC control arms. Intervention women than in men (50 vs 21 percentage Stopped early for benefit studies as a whole:
participants viewed a CRC point increase, interaction P = .02).
61
screening screening decision aid Incorrect analysis of ITT High

completion immediately before their clinician Selective reporting of Moderate
encounter. The decision aid measures (e.g., no effect Low
promoted screening and outcome)
Very Low
presented colonoscopy and fecal Large losses to F/U
occult blood testing as screening Difference in important
options. After the clinician prognostic factors at baseline
encounter, intervention patients
received support for screening
completion from a bilingual
patient navigator. Control
participants viewed a food safety
video before the encounter and
otherwise received usual care.
Hoffman, R.M., To describe National Internet Survey Study; 1,134 CRC participants; For all cancer screening decisions, Study Limitations =
et al., 2014, Am J decision making Adults aged >/= 50 years who 477 men and 657 providers usually (63%-71%) explained None
Prev Med processes and made cancer screening decisions women. that testing was optional, but less often Non-Randomized Studies
outcomes for (breast, BrCa; colorectal, CRC; asked women (43%-57%) than men (70%- Failure to develop and apply
cancer screening prostate, PCa) within the 71%) whether they wanted testing. Only appropriate eligibility criteria
discussions previous 2 years were invited to 27%-38% of participants reported SDM. Flawed measurement of
complete internet survey. Perceived high/average cancer risk and both exposure and outcome
Participants were asked about feeling highly informed were associated Failure to adequately control
their perceived cancer risk; how with confidence in the screening confounding
informed they felt about cancer decision. Among participants discussing Incomplete or inadequately
tests; whether their healthcare CRC screening, men had slightly higher short follow-up
provider addressed pros/cons of decision process scores than women (2.3 Differences in important
testing, presented the option of vs 2.0, p=0.03) and were more likely to prognostic factors at baseline
no testing, and elicited their report being asked by the healthcare
input; whether they were tested; provider whether they wanted testing
and their confidence in the (71% vs 57%, p=0.02). Overall, men and
screening decision. women (71% vs 69%, p<0.01) were
equally likely to undergo CRC screening
and indicate that they would definitely
make the same decision again (men =
69.6% vs women = 63%).
Lafata, J.E., 2014, To evaluate the Observational Study; Audio- 443 patients 93% (OR 2.22; 1.48-3.32) of patients Study Limitations =
Patient association of the recording of health exams among received a recommendation for screening None
Education and 5As (Assess, insured patients aged 50-80 years (Advise) and 53% were screened in the Non-Randomized Studies
Counseling Advise, Agree, and due for CRC screening were following year. The likelihood of Failure to develop and apply
Assist and joined with pre-visit patient screening increased as the number of 5A appropriate eligibility criteria
Arrange) surveys and screening use data steps increased: compared to patients Flawed measurement of
discussion during from an electronic medical whose visit contained no 5A step, those both exposure and outcome
the primary care record. whose visit contained 1-2 steps (OR =
62
office visits with 2.96 [95% CI 1.16, 7.53]) and 3 or more Failure to adequately control
patients’ steps (4.98 [95% CI 1.84, 13.44]) were confounding
subsequent significantly more likely to use screening. Incomplete or inadequately
colorectal cancer short follow-up
(CRC) screening Differences in important
use prognostic factors at baseline
Laiyemo, A.O., et To evaluate Cohort Study; 2007 Health 4,283 respondents CRC screening discussions occurred with Study Limitations =
al., 2014, Prev provider-patient Information National Trends included in analysis 3,320 (76.2%) respondents. None
Med communication Survey (HINTS) was used to Approximately 95% of these discussions Non-Randomized Studies
about CRC identify respondents who were at were with physicians. Overall, 2,793 Failure to develop and apply
screening with least 50 years of age and (62.6%) respondents were current with appropriate eligibility criteria
and without answered questions about their CRC screening regardless of the screening Flawed measurement of
specific screening communication with their care modality. Discussion about screening both exposure and outcome
modality providers and CRC screening (OR = 8.83; 95% CI: 7.20-10.84) and Failure to adequately control
recommendation uptake. Respondents were providers making a specific confounding
on patient tracked to see if compliant with recommendation about screening Incomplete or inadequately
compliance with CRC screening as the use of FOBT modality rather than leaving it to the short follow-up
screening within 1 year, sigmoidoscopy patient to decide (OR = 2.04; 95% CI: Differences in important
guidelines within 5 years, or colonoscopy 1.54-2.68) were associated with patient prognostic factors at baseline
within 10 years. compliance with CRC screening
guidelines.
Mosen, D.M., To examine Observational Study; Conducted 883 participants Average scores for comprehensives of Study Limitations =
2013, Am J association of at Kaiser Permanente Northwest. CRC discussion and perceived benefits None
Manag Care comprehensivene Participants overdue for CRC were 0.4 (range 0-1) and 4.0 (range 1.5), Non-Randomized Studies
ss of CRC screening received an automated respectively. 28.2% (n = 249 (completed Failure to develop and apply
screening telephone call (ATC) encouraging screening, 84% of whom had surveyed appropriate eligibility criteria
discussion by PCPs CRC screening. Participants assessments after their screening date. Flawed measurement of
with completion completed a survey on PCPs’ Of screeners, 95.2% completed the FIT. both exposure and outcome
of CRC screening discussion of CRC screening and More comprehensive discussion of CRC Failure to adequately control
patient beliefs regarding screening was associated with increased confounding
screening. Primary outcome screening (OR = 1.51, 95% CI 1.03 – Incomplete or inadequately
measure: receipt of CRC 2.21). Higher perceived benefits (OR = short follow-up
screening (assessed by EMR, 9 1.46, 95% CI = 1.13 – 1.90) and one or Differences in important
months after ATC). Primary more PCP visits (OR = 5.82, 95% CI = 3.87 prognostic factors at baseline
independent variable: – 8.74) were also associated with
comprehensiveness of CRC increased screening.
63
screening discussion by PCPs (7-

item scale). Secondary
independent variables: perceived
benefits of screening (4-item
scale assessing respondents’
agreement with benefits of
timely screening) and primary
care utilization (EMR; 9 months
after ATC).
Underhill, M.L. To assess how Cross-Sectional Study; The 7,674 individuals Perceived provider communication and Study Limitations =
and M.T. provider-patient association of provider relationship quality were associated with None
Kiviniemi, 2012, communication communication quality, both adherence to colonoscopy and with Non-Randomized Studies
Health Education and characteristics relationship, and colorectal ever having been screened. Predictive Failure to develop and apply
& Behavior of the patient- cancer screening was examined margins analyses indicated that appropriate eligibility criteria
provider within data from the 2007 Health increasing perceptions from lowest to Flawed measurement of
relationship may Information National Trends highest levels of communication and both exposure and outcome
related to Survey. relationship quality would be associated Failure to adequately control
screening with increases in screening rates confounding
behavior approaching 16 percentage points. Incomplete or inadequately
short follow-up
References:
1. Hoffman, R. M., et al. (2014). "Lack of Shared Decision Making in Cancer Screening Discussions." American Journal of Preventive Medicine 47(3): 251-259.
2. Lafata, J. E., et al. (2014). "Patient-physician colorectal cancer screening discussion content and patients' use of colorectal cancer screening." Patient Education & Counseling 94(1): 76-82.
3. Laiyemo, A. O., et al. (2014). "Influence of provider discussion and specific recommendation on colorectal cancer screening uptake among U.S. adults." Preventive Medicine 67: 1-5.
4. Mosen, D. M., et al. (2013). "More comprehensive discussion of CRC screening associated with higher screening." Am J Manag Care 19(4): 265-271. l
5. Reuland, D. S., et al. (2017). "Effect of Combined Patient Decision Aid and Patient Navigation vs Usual Care for Colorectal Cancer Screening in a Vulnerable Patient Population: A Randomized Clinical Tria." JAMA Intern Med
177(7): 967-974.
6. Schroy, P. C., 3rd, et al. (2012). "Aid-assisted decision making and colorectal cancer screening: a randomized controlled trial." American Journal of Preventive Medicine 43(6): 573-583.
7. Underhill, M. L. and M. T. Kiviniemi (2012). "The association of perceived provider-patient communication and relationship quality with colorectal cancer screening." Health Educ Behav 39(5): 555-563.
64
Question #5: What is the comparative patient acceptance of the following screening programs: colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool
screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA tests, blood-screening test, methylated SEPT9 DNA?
Primary Literature:
PICO Question: What is the comparative patient acceptance of the following screening programs: colonoscopy, flexible sigmoidoscopy, computed Lower Quality Rating if:
tomographic colonography, stool screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA tests, Studies inconsistent (wide
blood-screening test, methylated SEPT9 DNA? variation of treatment effect across
Modality: SEPT9 Test vs. Stool Screening Test
or outcomes varied)
Study Studies are indirect
Total # of Studies: 2 # of Systematic Reviews: 0 # of RCTs: 0 # of Non-Randomized Studies: 2 (PICO question is quite different
Adler et al (2014) This study Prospective Cohort Study. 172 subjects. 63 of 172 The top three reasons for rejecting Study Limitations = from the available evidence in
BMC assessed Participants were recruited during subjects were compliant colonoscopy were being uncomfortable None regard to population, intervention,
Gastroenterology patient regular consultations. All subjects to screening with the bowel preparation for Non-Randomized Studies comparison, or outcome)
willingness to were advised to undergo screening colonoscopy (37%). 106 colonoscopy (54%), fear of colorectal Failure to develop and apply
use non- by colonoscopy. Subjects who of the 109 subjects who cancer itself (44%) and fear that appropriate eligibility criteria Studies are imprecise (When
invasive stool refused colonoscopy were offered a refused colonoscopy colonoscopy would be painful (32%). Flawed measurement of studies include few patients and few
or blood choice of non-invasive tests. accepted an alternative These results were corroborated in a both exposure and outcome
based Subjects who selected stool testing non-invasive method follow-up question asking what would Failure to adequately control
screening received a collection kit and (97%). 90 selected the convince subjects to be screened by confounding confidence intervals and the results
tests after instructions; subjects who selected Septin9 blood test colonoscopy where 38% indicated an Incomplete or inadequately are uncertain)
refusing plasma testing had a blood draw (83%), 16 selected a improved bowel preparation, 29% short follow-up
colonoscopy. during the office visit. Patients who stool test (15%) and 3 indicated cancer prevention by Differences in important Publication Bias
were positive for either were refused any test (3%). polypectomy and 24% indicated that prognostic factors at baseline (e.g. pharmaceutical company
advised to have a diagnostic overcoming fears would change their sponsors study on effectiveness of
colonoscopy choice. In addition, when asked why they
chose one of the screening tests, 78% and
81% of subjects who had a blood and found)
stool test respectively, indicated ease of
getting the test. For those choosing the Increase Quality Rating if:
blood test, primary reasons for not Large Effect
choosing the stool test related to being Dose-response gradient
uncomfortable with specimen handling.
For those choosing the stool test, the
primary reason related to having used a biases increase certainty of effect
stool test in the past. As only 3 subjects
rejected any form of testing, limited Quality (certainty) of evidence for
survey data is available. studies as a whole:
65
High
Moderate
Low
Very Low
Benning, T. M., et To elicit Survey with Non-randomized 815 individuals aged 55- The combi-test ( blood+stool) was Study Limitations =
al. (2014). Acta individuals’ discrete choice experiment ( DCE). 75 years generally preferred over the blood and None
Oncologica preferences stool test and all three were preferred Non-Randomized Studies
for different over the option not to participate in Failure to develop and apply
non-invasive screening. The alternative specific constant appropriate eligibility criteria
CRC screening for the combi-test had the highest value Flawed measurement of
tests in a followed by the constants of the blood and both exposure and outcome
Dutch stool tests, respectively. The differences Failure to adequately control
screening between the tests are all significant and confounding
campaign. indicated a clear order in screening test Incomplete or inadequately
preference. There was significant negative short follow-up
effects for the sensitivity and risk reduction Differences in important
dummies (baseline is highest attribute prognostic factors at baseline
level) and a positive effect of strong level
of evidence (baseline is limited scientific
evidence) for all screening tests. This
indicates that as expected, more sensitive
tests, tests that lead to a higher risk
reduction of CRC death, and tests that are
supported by strong scientific evidence
are more likely to be chosen.
Taber et al (2014) To examine Prospective Cohort Study. 100 participants aged The respondents assigned screening Study Limitations =
American Journal attitudes of a Participants eligible for CRC 50-74 years with modalities a specific rank as well as None
of Health Behavior diverse screening completed cross-sectional average CRC risk. average rankings. SEPT9 was ranked first Non-Randomized Studies
community- surveys of their screening or second by 91% of respondents overall, Failure to develop and apply
based sample preferences following group with 58% ranking it first, 10% tied for first, appropriate eligibility criteria
toward SEPT9 discussions of colonoscopy, and 23% second. Preferences differed Flawed measurement of
for CRC sigmoidoscopy, FOBT, and SEPT9. sharply by race/ethnicity and screening both exposure and outcome
screening. status, such that 92.9% of unscreened Failure to adequately control
Whites ranked SEPT9 as their first choice, confounding
in contrast to only 31.3% of unscreened Incomplete or inadequately
Blacks. Colonoscopy was ranked first by short follow-up
31% of participants, tied for first by 9%, Differences in important
and ranked second by 39%. FOBT was prognostic factors at baseline
ranked first by only 1%.
66
References:
1. Adler, A., et al. (2014). "Improving compliance to colorectal cancer screening using blood and stool based tests in patients refusing screening colonoscopy in Germany." BMC Gastroenterology 14: 183 .
2. Benning, T. M., et al. (2014). "Preferences for potential innovations in non-invasive colorectal cancer screening: A labeled discrete choice experiment for a Dutch screening campaign." Acta Oncologica 53(7): 898-908 .
3. Taber, J. M., et al. (2014). "Preferences for blood-based colon cancer screening differ by race/ethnicity." American Journal of Health Behavior 38(3): 351-361 .
Modality: FIT vs colonoscopy
or outcomes varied)
Study Methods Studies are indirect
Total # of Studies: 6 # of Systematic Reviews: 1 # of RCTs: 2# of Non-Randomized Studies: 3 (PICO question is quite different
Bonello et al. To assess public Survey. A postal survey was 491 participants A small proportion of participants had no Study Limitations = from the available evidence in
(2016) BMC preferences for carried out during a separate returned eligible preference between the options (n = 40; None regard to population, intervention,
Gastroenterology colorectal cancer study on preferences for questionnaires 8.1 %; 95 % CI [6.1 %, 10.9 %)). The Non-Randomized Studies comparison, or outcome)
(CRC) surveillance different first-line CRC screening majority stated a preference for Failure to develop and apply
tests for modalities (non- or full-laxative surveillance with the stool test (n = 298; appropriate eligibility criteria Studies are imprecise (When
intermediate-risk computed tomographic 60.8 %; 95 % CI [56.4 %, 65.0 %)) with Flawed measurement of studies include few patients and few
adenomas, using a colonography, flexible colonoscopy preferred less frequently (n = both exposure and outcome
hypothetical sigmoidoscopy, or colonoscopy). 152; 31.0 %; [27.1 %, 35.3 %]). There was Failure to adequately control
scenario. Individuals were allocated at evidence of gender differences in confounding confidence intervals and the results
random to receive a pack surveillance preferences with women more Incomplete or inadequately are uncertain)
containing information on one likely to prefer the stool test than men short follow-up
first-line test, and a paragraph (66.7 % vs. 53.6 %, x2 [2, N = 490] = 9.028, Differences in important Publication Bias
describing CRC surveillance p = .011). They did not find evidence to prognostic factors at baseline (e.g. pharmaceutical company
recommendations for people suggest differences in surveillance sponsors study on effectiveness of
who are diagnosed with preferences by ethnicity, employment
intermediate-risk adenomas status, level of education, or indirect
during screening. All experience of CRC (all p-values > .05) found)
participants received a
description of two surveillance The majority of the participants who stated Increase Quality Rating if:
options: annual single-sample, a preference for surveillance with the Large Effect
home-based stool testing stool test stated that they did so because Dose-response gradient
(consistent with Faecal they would be tested more frequently
lmmunochemical Tests; FIT) or (62.1 %) and the test was more convenient
triennial colonoscopy. Invitees biases increase certainty of effect
(51.7 %). A large proportion also said they
were asked to imagine they had believed that it was less likely to cause side
67
been diagnosed with effects (39.9 %) and less likely to cause Quality (certainty) of evidence for
intermediate-risk adenomas, harm (32.6 %). The majority of participants studies as a whole:
and then complete a who stated a preference for colonoscopy High
questionnaire on their said that they believed the test was better
Moderate
surveillance preferences. at finding polyps or cancer (77.6 %) and
also more thorough (53.9 %). Almost one Low
in three who preferred a colonoscopy gave Very Low
the reason that they would have already
had one and so would be more familiar
with the test (29.6 %).
Bowyer et al. To examine Qualitative Study. Five semi- 28 adults When evaluating FIT in the context of a Study Limitations =
(2013) Journal of attitudes towards structured discussion groups surveillance program, all respondents None
Medical Screening an annual faecal were conducted with 28 adults readily made comparisons with related Non-Randomized Studies
immunochemical with different levels of CRC risk tests that they had been exposed to Failure to develop and apply
test for and experience of colonoscopy previously. Those with no experience of appropriate eligibility criteria
haemoglobin (FIT) or colonoscopic surveillance. surveillance were enthusiastic about an Flawed measurement of
versus three- Information was presented annual FIT to replace three-yearly both exposure and outcome
yearly sequentially using a step-by-step colonoscopy, because they felt that the Failure to adequately control
colonoscopic discussion guide. Results were higher testing frequency could improve confounding
surveillance of analyzed using thematic analysis detection of advanced lesions. Those with Incomplete or inadequately
individuals at experience of colonoscopic surveillance did short follow-up
intermediate risk not perceive FIT to be as accurate as Differences in important
of colorectal colonoscopy, and therefore either prognostic factors at baseline
cancer (CRC). preferred colonoscopy on its own or
wanted an annual FIT in addition to three-
yearly colonoscopy.
Daskalakis, C., et To investigate Randomized Controlled Trial. 945 participants Preference was not associated with overall Study Limitations =
al. (2014). how CRC Participants ages 50 to 79 years screening, but individuals who preferred None
Cancer screening test who were not up to date with FIT were more likely to complete FIT RCTS
Epidemiology, preference colorectal cancer screening screening (P = 0.005), whereas those who Lack of blinding
Biomarkers & operates together were randomized into one of preferred colonoscopy were more likely Lack of allocation
Prevention with test access three groups: (i) a usual case to perform colonoscopy screening (P = concealment
and navigation to control group; (ii) a standard 0.032). Stopped early for benefit
influence intervention (SI) group that Incorrect analysis of ITT
screening received non-tailored mailed Selective reporting of
adherence in access to both stool blood test measures (e.g., no effect
primary care. and colonoscopy; and (iii) a outcome)
tailored navigation intervention Large losses to F/U
(TNI) group that was provided Difference in important
mailed access and navigation prognostic factors at baseline
68
based on self-reported
screening test preference.
69
Singal A. G., et al To compare the RCT. Participants were randomly 5999 participants aged Study Limitations =
(2017) JAMA effectiveness of assigned to a mailed FIT 50-64 None
FIT outreach and outreach, mailed colonoscopy RCTS
colonoscopy outreach, or usual care with Lack of blinding
outreach to clinic based screening. Lack of allocation
increase concealment
completion of CRC Stopped early for benefit
screening process Incorrect analysis of ITT
within 3 years. Selective reporting of
measures (e.g., no effect
outcome)
Large losses to F/U
Difference in important
Wortley, S., et al. A systematic Systematic review 3 studies, 4902 One included study reported that both Study Limitations =
(2014). The review of discrete participants screened and screening-naive subjects None
Patient: Patient- choice preferred flexible sigmoidoscopy and Systematic Review
Centered experiments colonoscopy to FOBT screening. Another Review did not address
Outcomes (DCEs) of CRC found that fecal DNA testing, colonoscopy, focused clinical question
Research screening. and virtual colonoscopy (CT colonography) Search was not detailed or
were the most preferred tests. exhaustive
The third study reported similar results, Quality of the studies was
with respondents preferring newer tests not appraised or studies were of
such as virtual colonoscopy. low quality
Methods and/or results were
In most studies, FOBT was reported as the inconsistent across studies
least preferred option despite being the
test most commonly adopted in national
screening programs.
Xu, Y., et al. To compare Prospective Cohort study. 954 patients In the AHP analysis, the test accuracy was Study Limitations =
(2015). BMC patient Patients scheduled for a given the highest priority (0.457), followed None
70
Health Services preferences for colonoscopy were asked to by complications (0.321), and test Non-Randomized Studies
Research FIT or colonoscopy complete a FIT before preparation (0.223). Patients preferred Failure to develop and apply
colonoscopy prep. Following colonoscopy (0.599) compared with FIT appropriate eligibility criteria
both tests, patients completed a (0.401) when considering accuracy; Flawed measurement of
questionnaire which was based preferred FIT (0.589) compared with both exposure and outcome
on analytic hierarchy process colonoscopy (0.411) when considering Failure to adequately control
decision-making model. avoiding complications; and preferred FIT confounding
(0.650) compared with colonoscopy Incomplete or inadequately
(0.350) when considering test short follow-up
preparation. The overall aggregated Differences in important
priorities were 0.517 for FIT, and 0.483 for prognostic factors at baseline
colonoscopy, indicating patients slightly
preferred FIT over colonoscopy. Patients'
preferences were significantly different
before and after provision of detailed
information on test features (p < 0.0001).
References:
1. Bonello, B., et al. (2016). "Using a hypothetical scenario to assess public preferences for colorectal surveillance following screening-detected, intermediate-risk adenomas: annual home-based stool test vs. triennial
colonoscopy." BMC Gastroenterology 16: 113.
2. Bowyer, H. L., et al. (2013). "Patient attitudes towards faecal immunochemical testing for haemoglobin as an alternative to colonoscopic surveillance of groups at increased risk of colorectal cancer." Journal of Medical
Screening 20(3): 149-156.
3. Daskalakis, C., et al. (2014). "The effects of test preference, test access, and navigation on colorectal cancer screening." Cancer Epidemiology, Biomarkers & Prevention 23(8): 1521-1528.
4. Singal, A. G. et. Al. (2017) “Effect of Colonoscopy Outreach vs Fecal Immunochemical Test Outreach on Colorectal Cancer Screening Completion.” JAMA 318(9); 806-815.
5. Wortley, S., et al. (2014). "Assessing stated preferences for colorectal cancer screening: a critical systematic review of discrete choice experiments." The Patient: Patient-Centered Outcomes Research 7(3): 271-282.
6. Xu, Y., et al. (2015). "Comparison of patient preferences for fecal immunochemical test or colonoscopy using the analytic hierarchy process." BMC Health Services Research 15: 175.
Modality: FIT vs gFOBT
or outcomes varied)
Total # of Studies: 2 # of Systematic Reviews: 1# of RCTs: # 0 of Non-Randomized Studies: 1
71
Deutekom, M., et The aim of the Prospective cohort study. All 20,623 participants Study Limitations = (PICO question is quite different
al. (2010). present study invited received a FOBT kit, invited, 10,972 invitees None from the available evidence in
Scandanavian was to study including an invitation letter, participated Non-Randomized Studies regard to population, intervention,
Journal of differences in information leaflet, and a FOBT Failure to develop and apply
comparison, or outcome)
Gastroenterology patient without costs for invitees. From appropriate eligibility criteria
perception the municipal databases, random Flawed measurement of
between i-FOBT samples were taken according to both exposure and outcome Studies are imprecise (When
and g-FOBT and postal address and randomized to Failure to adequately control studies include few patients and few
differences in receive a 3-day guaiac-based confounding events and thus have wide
perception and FOBT (g-FOBT) or an automated Incomplete or inadequately confidence intervals and the results
participation semi-quantitative FIT. A short follow-up are uncertain)
rates among questionnaire was sent by mail 2 Differences in important
relevant weeks after initial invitation to all prognostic factors at baseline
subgroups in a invitees, accompanied by a letter, Publication Bias
population based which explained the purpose of (e.g. pharmaceutical company
study the survey. sponsors study on effectiveness of
found)
Wortley, S., et al. A systematic Systematic Review 1 study, 1920 The study that only considered gFOBT vs Study Limitations =
(2014). The review of participants FIT, did not provided enough detail to None
Patient: Patient- discrete choice permit conclusions about the specific type Systematic Review Large Effect
Centered experiments of FOBT test respondents preferred. Review did not address Dose-response gradient
Outcomes (DCEs) of CRC focused clinical question Plausible confounders or other
Research screening. Search was not detailed or biases increase certainty of effect
exhaustive
Quality of the studies was Quality (certainty) of evidence for
not appraised or studies were of
studies as a whole:
low quality
Methods and/or results were High
inconsistent across studies Moderate
Low
Very Low
References:
1. Deutekom, M., et al. (2010). "Comparison of guaiac and immunological fecal occult blood tests in colorectal cancer screening: the patient perspective." Scandinavian Journal of Gastroenterology 45(11): 1345-1349 .
2. Wortley, S., et al. (2014). "Assessing stated preferences for colorectal cancer screening: a critical systematic review of discrete choice experiments." The Patient: Patient-Centered Outcomes Research 7(3): 271-282.
72
Modality: CT Colonography vs. Colonoscopy
or outcomes varied)
Study Methods Studies are indirect
Total # of Studies: 4 # of Systematic Reviews: 1# of RCTs: 0 # of Non-Randomized Studies: 3 (PICO question is quite different
Ghanouni, A., et To examine public Quantitative Study. Six 30 adults CTC was favored on the parameters of Study Limitations = from the available evidence in
al. (2012). perceptions of and discussion groups were carried invasiveness, extra-colonic evaluation and None regard to population, intervention,
Patient preferences for out with 30 adults aged 49–60 interference with daily life, whereas Non-Randomized Studies comparison, or outcome)
Education and colonoscopy vs. CT years (60% female). sensitivity, avoiding false-positives and the Failure to develop and apply
Counseling colonography (CTC) Information about different capacity to remove polyps immediately appropriate eligibility criteria Studies are imprecise (When
as technologies for aspects of the tests (e.g. were perceived to be important Flawed measurement of studies include few patients and few
colorectal cancer sensitivity, practical issues) advantages of colonoscopy. There was no both exposure and outcome
(CRC) screening.. was presented sequentially strong preference for either test: with Failure to adequately control
using a semi-structured, step- 46% preferring colonoscopy vs. 42% for confounding confidence intervals and the results
by-step topic guide. CTC. Incomplete or inadequately are uncertain)
Discussions were recorded and short follow-up
analyzed using framework Differences in important Publication Bias
analysis. prognostic factors at baseline (e.g. pharmaceutical company
found)
Imaeda, A., et al. To assess patient Survey Study. MDS survey to 92 patients enrolled in After completing the computer task, Study Limitations =
(2010). Journal experiences with a elicit patients’ values for primary care clinics at a patients were asked to choose their None Increase Quality Rating if:
of General Maximum characteristics related to fecal VA hospital and preferred screening test: 62% chose Non-Randomized Studies Large Effect
Internal Differences Scaling occult blood testing, associated university. colonoscopy, 23% chose colon capsule, Failure to develop and apply Dose-response gradient
Medicine (MDS) tool for sigmoidoscopy, colonoscopy, 10% CT colonography, 4% FOBT and 1% appropriate eligibility criteria Plausible confounders or other
eliciting values about CT colonography and colon sigmoidoscopy. Subjects preferring Flawed measurement of biases increase certainty of effect
CRC screening test capsule endoscopy. colonoscopy assigned greater importance both exposure and outcome
characteristics and to the sensitivity of the CRC screening test Failure to adequately control Quality (certainty) of evidence for
determine whether (median importance=21.5 versus 19.6, confounding
patients vary in how p=0.01) compared to those preferring less studies as a whole:
they prioritize test invasive procedures. There were no other short follow-up High
characteristics and statistically significant differences in Differences in important Moderate
whether this importance ratings between those prognostic factors at baseline Low
variation relates to preferring colonoscopy and subjects Very Low
test preferences preferring less invasive procedures.
73
Lin, O. S., et al. A systematic review Systematic Review 23 studies (comprising Amongst the included studies, 16 Study Limitations =
(2012). Journal and meta-analysis on 5616 subjects) (comprising 3573 subjects) showed a None
of General patient preference statistically significant preference for CTC, Systematic Review
Internal for CTC versus three (927 subjects) showed a preference Review did not address
Medicine colonoscopy. for colonoscopy, and four (1116 subjects) focused clinical question
showed no difference in preference. Search was not detailed or
exhaustive
Stratified analysis revealed that studies Quality of the studies was
published in radiology journals (preference not appraised or studies were of
difference 0.590 [95 % CI 0.485, 0.694]) low quality
seemed more likely than studies in Methods and/or results were
gastroenterology (0.218 [-0.015-0.451]) or inconsistent across studies
general medicine journals (-0.158 [-0.389-
0.072]) to report preference for CTC
(p<0.001). Studies by radiology authors
showed a trend towards stronger
preference for CTC compared with studies
by gastroenterology authors.
Moawad, F. J., et The aim of this Survey Study. Average-risk 250 patients The most common reasons for undergoing Study Limitations =
al. (2010). project was to assess patients CTC included convenience (33.6%), None
American Journal patient preferences undergoing CRC screening recommendation by referring provider Non-Randomized Studies
of between completed a survey that (13.2%), and perceived safety (10.8%). Had Failure to develop and apply
Roentgenology colonoscopy and CTC assessed reasons for choosing CTC not been an available option, 91 of the appropriate eligibility criteria
in CTC in 250 patients (36%) would have foregone Flawed measurement of
an open access lieu of colonoscopy, CRC screening. Among the 57 patients who both exposure and outcome
system. compliance with CRC screening had experienced both CTC and Failure to adequately control
if CTC was not offered, and colonoscopy 95% (n = 54) preferred CTC. confounding
which of Incomplete or inadequately
the two tests they preferred. short follow-up
References:
1. Ghanouni, A., et al. (2012). "Public perceptions and preferences for CT colonography or colonoscopy in colorectal cancer screening." Patient Education & Counseling 89(1): 116-121 .
2. Imaeda, A., et al. (2010). "What is most important to patients when deciding about colorectal screening?" Journal of General Internal Medicine 25(7): 688-693 .
3. Lin, O. S., et al. (2012). "Preference for colonoscopy versus computerized tomographic colonography: a systematic review and meta-analysis of observational studies." Journal of General Internal Medicine 27(10): 1349-1360 .
4. Moawad, F. J., et al. (2010). "CT colonography may improve colorectal cancer screening compliance." AJR. American Journal of Roentgenology 195(5): 1118-1123 .
74
Modality: FOBT vs colonoscopy
or outcomes varied)
Total # of Studies: 6 # of Systematic Reviews: 1 # of RCTs: 0# of Non-Randomized Studies: 5 (PICO question is quite different
Hawley, S., et al. To evaluate Multi-method study including 64 physicians and 500 Based on attribute rankings, most Study Limitations = from the available evidence in
(2014). American associations baseline surveys from average- patients participants had a weak preference for None regard to population, intervention,
Journal of between patients' risk HMO members joined with colonoscopy (COL) (41%), an unclear Non-Randomized Studies comparison, or outcome)
Managed Care preferences for audio recordings of 415 periodic preference (22.4%), or a weak preference Failure to develop and apply
attributes of health exams (PHEs) and for fecal occult blood testing (FOBT) appropriate eligibility criteria Studies are imprecise (When
different electronic medical record (EMR) (18.6%). About half (56%) of patients were Flawed measurement of studies include few patients and few
colorectal (CRC) data. Methods Patient ratings of screened at 12 months and there was no both exposure and outcome
screening test attributes were used to statistical association between attribute Failure to adequately control
modalities, create an algorithm reflecting preferences and type of test received. confounding confidence intervals and the results
physician CRC type and strength of CRC Patients were significantly more likely to Incomplete or inadequately are uncertain)
screening screening modality preference at receive a recommendation including a test short follow-up
recommendations baseline. Physician other than COL when they had an Differences in important Publication Bias
during periodic recommendations were obtained attribute-based test preference for FOBT prognostic factors at baseline (e.g. pharmaceutical company
health exams, and from audio recordings. Attribute- (odds ratio [OR]: 2.17; 95% CI, 1.26-3.71; P sponsors study on effectiveness of
subsequent based test preferences and < .01)
utilization of physician recommendations were
screening 12 compared with CRC test use using found)
months later in a chisquare tests. Associations
large health between attribute-based Increase Quality Rating if:
maintenance preferences and physician Large Effect
organization recommendations were assessed Dose-response gradient
(HMO). using logistic regression.
hypothetical
scenario. biases increase certainty of effect
Hawley, S. T., et To describe Survey Study. Patients completed 212 patients Of the guideline-recommended tests, 37% Study Limitations = Quality (certainty) of evidence for
al. (2008). variation in CRC a preference assessment preferred colonoscopy, 31 % FOBT, 15% None studies as a whole:
Medical Care screening instrument. Participants were barium enema, and 9% flexible Non-Randomized Studies High
preferences asked to rate 8 hypothetical CRC sigmoidoscopy. Ratings of new Failure to develop and apply Moderate
among racially/ screening test scenarios technology tests (virtual colonoscopy and appropriate eligibility criteria
Low
ethnically diverse comprised of different FIT) were significantly (P < 0.05) higher Flawed measurement of
Very Low
combinations of 5 attributes and than ratings of guideline-recommended both exposure and outcome
6 scenarios designed to depict tests. The order of the importance of
75
primary care guideline-recommended CRC attributes was: what the test involved Failure to adequately control
patients. screening tests (eg, fecal occult (37%), accuracy (19%), frequency (17%), confounding
blood test, flexible discomfort ( 15%), and preparation ( 13%). Incomplete or inadequately
sigmoidoscopy, colonoscopy, and Part-worth utilities for 1 attribute showed short follow-up
double contrast barium enema) that collecting a stool sample was most Differences in important
including new technology (eg, preferable and endoscopy without prognostic factors at baseline
virtual colonoscopy, fecal sedation least preferable. Multivariate
immunochemical test). Responses regression found that race/ethnicity and
were used to calculate the overall specific test attributes were
importance of test attributes, the independently associated (P < 0.05) with
relative importance of attribute test preferences.
levels, and to identify factors
associated with preferences.
Hawley, S. T., et The purpose of Survey Study. After 1224 patients Most patients stated a test preference: Study Limitations =
al. (2012). Cancer this study was to administration of a baseline 34.7% indicated a preference for FOBT, None
identify factors survey, participants were 41.1% for COL, 12.7% for SIG, 5.7% for BE, Non-Randomized Studies
associated with randomized to 1 of 3 study and 5.8% did not report a preference. Failure to develop and apply
colorectal cancer groups stratified by sex and past Factors statistically significantly associated appropriate eligibility criteria
(CRC) screening CRC screening status (ever at P<.10 with baseline test preference for Flawed measurement of
test preference screened vs overdue for COL, SIG, or FOBT in univariable analyses both exposure and outcome
and examine the screening). The tailored were family history, ever had CRC Failure to adequately control
association intervention group participated in screening with SIG, physician confounding
between test an interactive, tailored computer recommendation for any CRC test, FOBT, Incomplete or inadequately
preference and program; the web site group and COL, preference for decision-making short follow-up
test completed. viewed general information about stage of readiness for CRC screening, test- Differences in important
CRC screening from a publicly specific self-efficacy for FOBT, COL or SIG, prognostic factors at baseline
available web site, Screen for Life, comparative perceived risk, worry, and
which is the national CRC perceived pros of CRC screening
awareness campaign from the
Centers for Disease Control and
Prevention; and the survey-only
control group received no
additional information about CRC
screening. As part of the study, all
participants completed a wellness
visit and exam. At the 12-month
follow-up, medical records were
reviewed to collect CRC screening
utilization data.
76
Imaeda, A., et al. To assess patient Survey Study. MDS survey to elicit 92 patients enrolled in After completing the computer task, Study Limitations =
(2010). Journal experiences with a patients’ values for characteristics primary care clinics at a patients were asked to choose their None
of General Maximum related to fecal occult blood VA hospital and preferred screening test: 62% chose Non-Randomized Studies
Internal Differences testing, sigmoidoscopy, associated university. colonoscopy, 23% chose colon capsule, Failure to develop and apply
Medicine Scaling (MDS) tool colonoscopy, CT colonography 10% CT colonography, 4% FOBT and 1% appropriate eligibility criteria
for eliciting values and colon capsule endoscopy. sigmoidoscopy. Subjects preferring Flawed measurement of
about CRC colonoscopy assigned greater importance both exposure and outcome
screening test to the sensitivity of the CRC screening test Failure to adequately control
characteristics and (median importance=21.5 versus 19.6, confounding
determine p=0.01) compared to those preferring less Incomplete or inadequately
whether patients invasive procedures. There were no other short follow-up
vary in how they statistically significant differences in Differences in important
prioritize test importance ratings between those prognostic factors at baseline
characteristics and preferring colonoscopy and subjects
whether this preferring less invasive procedures.
variation relates
to test
preferences
77
Palmer, R. C., et To understand In-depth personal interviews 60 participants Study Limitations =

al. (2010). how were conducted with 60 None
Journal of Cancer characteristics of Americans to understand if CRC Non-Randomized Studies
Education CRC screening test preferences exist and to Failure to develop and apply
tests influence identify what attributes of appropriate eligibility criteria
preferences and screening tests influence test Flawed measurement of
CRC screening test preferences. both exposure and outcome
choice. Failure to adequately control
confounding
short follow-up
78
Wortley, S., et al. A systematic Systematic Review 4 studies, 4843 Four studies included alternatives that Study Limitations =
(2014). The review of discrete participants described a variety of CRC screening None
Patient: Patient- choice techniques One study reported that both Systematic Review
Centered experiments screened and screening- naive subjects Review did not address
Outcomes (DCEs) of CRC preferred flexible sigmoidoscopy and focused clinical question
Research screening. colonoscopy to FOBT screening. Another Search was not detailed or
found that fecal DNA testing, colonoscopy, exhaustive
and virtual colonoscopy (CT colonography) Quality of the studies was
were the most preferred tests. A third, not appraised or studies were of
reported similar results, with respondents low quality
preferring newer tests such as virtual Methods and/or results were
colonoscopy. In most studies, FOBT was inconsistent across studies
reported as the least preferred option
despite being the test most commonly
adopted in national screening programs.
References:
1. Hawley, S., et al. (2014). "Managed care patients' preferences, physician recommendations, and colon cancer screening." American Journal of Managed Care 20(7): 555-561 .
2. Hawley, S. T., et al. (2008). "Preferences for colorectal cancer screening among racially/ethnically diverse primary care patients." Medical Care 46(9 Suppl 1): S10-16
3. Hawley, S. T., et al. (2012). "Preferences for colorectal cancer screening tests and screening test use in a large multispecialty primary care practice." Cancer 118(10): 2726-2734 .
4. Imaeda, A., et al. (2010). "What is most important to patients when deciding about colorectal screening?" Journal of General Internal Medicine 25(7): 688-693 .
5. Palmer, R. C., et al. (2010). "Colorectal cancer screening preferences among African Americans: which screening test is preferred?" Journal of Cancer Education 25(4): 577-581 .
6. Wortley, S., et al. (2014). "Assessing stated preferences for colorectal cancer screening: a critical systematic review of discrete choice experiments." The Patient: Patient-Centered Outcomes Research 7(3): 271-282 .
Question #6: What is the comparative cost-effectiveness of the following screening programs: colonoscopy, flexible sigmoidoscopy, computed tomographic colonography, stool
screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA tests, blood screening test, methylated SEPT9 DNA?
Primary Literature:
PICO Question: What is the comparative cost-effectiveness of the following screening programs: colonoscopy, flexible sigmoidoscopy, computed Lower Quality Rating if:
blood screening test, methylated SEPT9 DNA? variation of treatment effect across
Modality: FIT vs. Colonoscopy
or outcomes varied)
Total # of Studies: 1 # of Systematic Reviews: 0 # of RCTs: 0 # of Non-Randomized Studies: 1
79
Wong et al., To evaluate the Economic Evaluation; the 5,863 patients received Study Limitations = (PICO question is quite different
2015, Scientific cost incremental cost-effectiveness ratio yearly FIT and 4,869 None from the available evidence in
Reports effectiveness of (ICER) was assessed based on the received colonoscopy in Economic Evaluation regard to population, intervention,
FIT and detection rates of neoplastic lesions, China The research question is not
colonscopy to and costs including screening clearly stated
detect compliance, polypectomy, The perspective of interest is
colorectal colonoscopy complications, and not clear (ie., societal, patient, Studies are imprecise (When
neoplastic staging of CRC detected health system, payer) studies include few patients and few
lesions based The source(s) of events and thus have wide
on data from a effectiveness estimates are not confidence intervals and the results
5-year clearly stated are uncertain)
community The primary outcome
screening measures are not clearly stated
service The methods for the Publication Bias
estimation of quantities and unit (e.g. pharmaceutical company
costs are not described sponsors study on effectiveness of
Using FIT scheme as a control, the drug, only small, positive studies
incremental Cost-Effectiveness Ratio of
found)
screening colonoscopy was US$3,489,
US$27,962, US$922,762 and US$23,981 to
detect one adenoma, advanced neoplasia,
Large Effect
CRC, and a composite endpoint of
advanced neoplasia or stage I CRC Dose-response gradient
respectively. The respective incremental Plausible confounders or other
cost effectiveness ratio (ICER) of screening biases increase certainty of effect
colonoscopy to detect these lesions in
lower-risk subjects was US$3,597, Quality (certainty) of evidence for
US$39,513, -US$2,765,876 (dominated) studies as a whole:
and US$32,297 for lower-risk subjects. The High
corresponding ICER was US$3,153, Moderate
US$14,852, US$184,162 and US$13,919 for Low
high-risk individuals Very Low
References:
1. Wong, M. C., Ching, J. Y., Chan, V. C., & Sung, J. J. (2015). The comparative cost-effectiveness of colorectal cancer screening using faecal immunochemical test vs. colonoscopy. Scientific Reports, 5, 13568.
doi:https://dx.doi.org/10.1038/srep13568
80
tomographic colonography, stool screening tests, high-sensitivity guaiac fecal occult blood, fecal immunochemical, or multitarget stool DNA tests, Studies inconsistent (wide variation
of treatment effect across studies,
blood screening test, methylated SEPT9 DNA?
populations, interventions, or outcomes
Modality: Hybrid Screening Strategy varied)
Study Methods (PICO question is quite different from the
available evidence in regard to
Dinh et al, 2013, To investigate the Economic Evaluation; cost- Population that A hybrid screening strategy led to Study Limitations =
population, intervention, comparison, or
Clinical cost-effectiveness effectiveness analysis using the represents members of substantial reductions in CRC incidence None
Gastroenterology of a hybrid Kaiser Permanente and mortality, gains in quality-adjusted life Economic Evaluation outcome)
Archimedes Model to evaluate
and Hepatology screening strategy the effects of different CRC Northern California years (QALYs), and reductions in costs, The research question is
that was based on screening strategies on health comparable with those of the best single- not clearly stated Studies are imprecise (When studies
a fecal outcomes and costs related to test strategies. Screening by annual FIT of The perspective of include few patients and few events and
immunological CRC patients 50–65 years old and then a single interest is not clear (ie., thus have wide confidence intervals and
test (FIT) and colonoscopy when they were 66 years old societal, patient, health
the results are uncertain)
colonoscopy In simulation model, patients (FIT/COLOx1) reduced CRC incidence by system, payer)
receive annual or biennial FIT, 72% and gained 110 QALYs for every 1000 The source(s) of
people during a period of 30 years, effectiveness estimates are Publication Bias
beginning when patients are 50
compared with no screening. not clearly stated (e.g. pharmaceutical company sponsors
years old, with a single
colonoscopy when they are 66 The primary outcome study on effectiveness of drug, only
years old Compared with annual FIT, FIT/COLOx1 measures are not clearly small, positive studies found)
gained 1400 QALYs/100,000 persons at an stated
incremental cost of $9700/QALY gained The methods for the
and required 55% fewer FITs. estimation of quantities and
unit costs are not described Large Effect
Compared with FIT/COLOx1, colonoscopy Dose-response gradient
at 10-year intervals gained 500 Plausible confounders or other
QALYs/100,000 at an incremental cost of biases increase certainty of effect
$35,100/QALY gained but required 37%
more colonoscopies. Quality (certainty) of evidence for
studies as a whole:
Uncertainties associated with estimates of
FIT performance within a program setting High
and sensitivities for flat and right-sided Moderate
lesions are expected to have significant Low
impacts on the cost-effectiveness results Very Low
References:
1. Dinh, T., Ladabaum, U., Alperin, P., Caldwell, C., Smith, R., & Levin, T. R. (2013). Health benefits and cost-effectiveness of a hybrid screening strategy for colorectal cancer. Clinical Gastroenterology & Hepatology,
11(9), 1158-1166. doi:https://dx.doi.org/10.1016/j.cgh.2013.03.013
81
Modality: Colonoscopy vs. Sigmoidoscopy varied)
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations Studies are indirect
Study (PICO question is quite different from the
Sharaf et al, To explore the Economic Analysis; performed a Persons at average CRC Compared with FIT, flexible sigmoidoscopy Study Limitations =
2013, Am J comparative contemporary cost-utility analysis risk within the general (FS) and colonscopy both cost < $ 50,000 / None
Gastroenterol effectiveness using a Markov model validated US population were QALY gained when FIT per-cycle adherence Economic Evaluation outcome)
and cost- against data from randomized modeled. Screening was < 50 % . The research question is
effectiveness of controlled trials of FOBT and strategies included not clearly stated Studies are imprecise (When studies
colonoscopy vs. sigmoidoscopy those recommended by Colonscopy cost $ 56,800 / QALY gained vs. The perspective of include few patients and few events and
sigmoidoscopy the US Preventive FS in the base case. Colonoscopy cost < $ interest is not clear (ie., thus have wide confidence intervals and
and alternative Services Task Force 100,000 / QALY gained vs. FS when societal, patient, health
CRC screening colonoscopy yielded a relative risk of system, payer)
strategies proximal CRC of < 0.5 vs. no screening. In The source(s) of
probabilistic analyses, colonoscopy was effectiveness estimates are Publication Bias
cost-effective vs. FS at a willingness-to-pay not clearly stated (e.g. pharmaceutical company sponsors
threshold of $ 100,000 / QALY gained in 84 The primary outcome study on effectiveness of drug, only
% of iterations measures are not clearly small, positive studies found)
stated
AUTHORS’ CONCLUSION: Screening The methods for the
colonoscopy may be cost-effective estimation of quantities and
compared with FIT and sigmoidoscopy, unit costs are not described Large Effect
depending on the relative rates of Dose-response gradient
screening uptake and adherence and the Plausible confounders or other
protective benefi t of colonoscopy in the biases increase certainty of effect
proximal colon. Colonoscopy ’ s cost-
effectiveness compared with Quality (certainty) of evidence for
sigmoidoscopy is contingent on the ability
studies as a whole:
to deliver ~ 50 % protection against CRC in
the proximal colon High
Moderate
Low
Very Low
References:
1. Sharaf, R. N., & Ladabaum, U. (2013). Comparative effectiveness and cost-effectiveness of screening colonoscopy vs. sigmoidoscopy and alternative strategies. American Journal of Gastroenterology, 108(1), 120-132.
doi:https://dx.doi.org/10.1038/ajg.2012.
82
studies, populations, interventions, or
Modality: Colonoscopy vs. Colonography outcomes varied)
Study (PICO question is quite different from
Total # of Studies: 3 # of Systematic Reviews: 1 of RCTs: 0 # of Non-Randomized Studies: 2
Kriza et al, 2013, To examine Systematic Review of cost 9 cost-effectiveness There was considerable heterogeneity in Study Limitations =
European Journal cost- effectiveness studies studies comparing CTC modelling complexity and methodology. None
of Radiology effectiveness of and COL as a screening Different model assumptions and inputs Systematic Review or outcome)
CTC versus tool and providing had large effects on resulting cost- Review did not address
optical outcomes in life-years effectiveness of CTC and COL. CTC was focused clinical question Studies are imprecise (When
colonoscopy saved, published found to be dominant or cost-effective in Search was not detailed or studies include few patients and few
(COL) for CRC between January 2006 three studies, assuming the most exhaustive events and thus have wide confidence
screening and November 2012 favourable scenario. COL was found to be Quality of the studies was
intervals and the results are
not cost effective in one study not appraised or studies were
of low quality uncertain)
In the remaining studies where CTC was Methods and/or results
found to be not cost-effective, it would be were inconsistent across Publication Bias
possible to transform CTC towards cost- studies (e.g. pharmaceutical company
effectiveness when altering assumptions sponsors study on effectiveness of
on costs, adherence and for some studies, drug, only small, positive studies
naturalhistory assumptions and CTC
found)
accuracy
Pyenson et al, To compare the Economic Evaluation; used Medicare Sampling of US CTC is 29% less expensive than OC for the Study Limitations = Increase Quality Rating if:
2015, Abdom Medicare claims data, fee schedules, Medicare popluation Medicare population in the base scenario. None Large Effect
Imaging population cost established protocols, and other Although the CTC cost advantage is Economic Evaluation Dose-response gradient
of CRC sources to estimate CTC and OC per- increased or reduced under alternative The research question is Plausible confounders or other
screening of screen costs, including the costs of scenarios, it is always positive not clearly stated biases increase certainty of effect
average risk OC referrals for a subset of CTC The perspective of interest
individuals by patients. They then modeled and is not clear (ie., societal,
CTC vs. optical compared the Medicare costs of patient, health system, payer) Quality (certainty) of evidence for
colonoscopy patients who complied with CTC and The source(s) of studies as a whole:
(OC) OC screening recommendations and effectiveness estimates are High
tested alternative scenarios not clearly stated Moderate
The primary outcome Low
measures are not clearly Very Low
stated
83
The methods for the

estimation of quantities and
unit costs are not described
Zafar et al, 2015, To compare Retrospective Cohort Study; initial Medicare outpatients Higher adjusted costs per patient were Study Limitations =
Acad Radiol differences in OC patients were matched on aged >/= 66 years who revealed in the year after initial CTC None
Medicare costs county of residence and year of received initial CTC (n = compared to initial OC for outpatient Non-Randomized Studies
1 year after screening 531) or OC (n = 17,593) testing related to potential colonic ($50; Failure to develop and
initial between January 2007 95% confidence interval [CI], $12–$88; P = apply appropriate eligibility
computed and December 2008 .010) and extracolonic findings ($64; 95% criteria
tomographic CI, $23–$106; P = .002). Flawed measurement of
colonography both exposure and outcome
(CTC) or initial However, there were no differences in Failure to adequately
optical adjusted total costs per patient in the control confounding
colonoscopy year after either modality ($2065; 95% CI, Incomplete or
(OC) $1672–$5803; P = .28). inadequately short follow-up
Similarly, adjusted costs did not differ prognostic factors at baseline
between cohorts for inpatient ($267; 95%
CI, $1017–$1550; P = .68) or outpatient
care ($2828; 95% CI, $311–$5966; P = .08).
References:
1. Kriza, C., Emmert, M., Wahlster, P., Niederlander, C., & Kolominsky-Rabas, P. (2013). An international review of the main cost-effectiveness drivers of virtual colonography versus conventional colonoscopy for colorectal
cancer screening: is the tide changing due to adherence? European Journal of Radiology, 82(11), e629-636. doi:https://dx.doi.org/10.1016/j.ejrad.2013.07.019
2. Pyenson, B., Pickhardt, P. J., Sawhney, T. G., & Berrios, M. (2015). Medicare cost of colorectal cancer screening: CT colonography vs. optical colonoscopy. Abdominal Imaging, 40(8), 2966-2976.
doi:https://dx.doi.org/10.1007/s00261-015-0538-1
3. Zafar, H. M., Yang, J., Armstrong, K., & Groeneveld, P. (2015). Cost Differences After Initial CT Colonography Versus Optical Colonoscopy in the Elderly. Academic Radiology, 22(7), 807-813.
doi:https://dx.doi.org/10.1016/j.acra.2015.03.002
Modality: Colonoscopy or outcomes varied)
Study
84
Fitch et al, 2015, To determine Economic Evaluation; Monte Carlo Model based on US Found that increasing screening adherence Study Limitations = (PICO question is quite different
Am J Manag the value of simulation using a large multi-state patients from 50% to 100% would cost about $3 per None from the available evidence in
Care life-years saved cancer registry, a large national member per month (2013 US$) and reduce Economic Evaluation regard to population, intervention,
due to administrative claims database, and CRC treatment costs by about $1 per The research question is not
colorectal a model of CRC development based member per month. The cost per life-year clearly stated
cancer (CRC) on published clinical literature, saved is approximately $12,000, an The perspective of interest is
screening with estimated the impact of screening amount that is much lower than for not clear (ie., societal, patient, Studies are imprecise (When
colonoscopy for with colonoscopy on incidence of cervical or breast cancer screening and health system, payer) studies include few patients and few
the population CRC, aggregate cost of comparable to lung cancer screening The source(s) of events and thus have wide
aged 50 to 64 colonoscopies and CRC, and life- effectiveness estimates are not confidence intervals and the results
years years saved clearly stated are uncertain)
The primary outcome
measures are not clearly stated
The methods for the Publication Bias
estimation of quantities and unit (e.g. pharmaceutical company
costs are not described sponsors study on effectiveness of
found)

Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Fitch, K., Pyenson, B., Blumen, H., Weisman, T., & Small, A. (2015). The value of colonoscopic colorectal cancer screening of adults aged 50 to 64. American Journal of Managed Care, 21(7), e430-438.
85
Modality: Colonography varied)
Hanley et al, To review Systematic Review of cost- 16 cost-effectiveness or CTC appeared cost-effective versus no Study Limitations =
2012, evidence on, effectiveness and cost-utility cost-utility analyses of screening and, in general, compared to None
International and identify key analyses CTC-based screening (11 flexible sigmoidoscopy and fecal occult Systematic Review or outcome)
Journal of factors from the US) blood testing. Results were mixed Review did not address
Technology influencing, comparing CTC to colonoscopy. focused clinical question Studies are imprecise (When
Assessment in cost- Parameters most influencing cost- Search was not detailed studies include few patients and few
Health Care effectiveness of effectiveness included: CTC costs, or exhaustive events and thus have wide confidence
CTC screening screening uptake, threshold for polyp Quality of the studies was
intervals and the results are uncertain)
referral, and extra-colonic findings not appraised or studies
were of low quality
Methods and/or results Publication Bias
were inconsistent across (e.g. pharmaceutical company
studies sponsors study on effectiveness of
drug, only small, positive studies found)

Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Hanly, P., Skally, M., Fenlon, H., & Sharp, L. (2012). Cost-effectiveness of computed tomography colonography in colorectal cancer screening: a systematic review. International Journal of Technology Assessment in Health
Care, 28(4), 415-423. doi:https://dx.doi.org/10.1017/S0266462312000542
86
Modality: FIT varied)
Guy et al, 2014, To estimate the Economic Evaluation; using The target population The initial additional investment required Study Limitations =
International initial investment estimates from the literature, included all adults aged was estimated at $277.9 to $318.2 million None
Journal of required and the cost 50 years to 75 years in annually, with an estimated 8.7 to 9.4 Economic Evaluation outcome)
examined the total initial (first
Technology per person screened 2 years) annual cost and the the United States million individuals screened at a cost of The research question is
Assessment in of a nationwide FIT- cost per person screened regardless of insurance $32 to $39 per person screened. The not clearly stated Studies are imprecise (When studies
Health Care based colorectal under each intervention design status and CRC program was estimated to prevent 2900 to The perspective of include few patients and few events and
cancer screening screening status 3100 deaths annually interest is not clear (ie., thus have wide confidence intervals and
program among societal, patient, health
adults aged 50 years system, payer)
to 75 years The source(s) of
effectiveness estimates are Publication Bias
not clearly stated (e.g. pharmaceutical company sponsors
The primary outcome study on effectiveness of drug, only
measures are not clearly small, positive studies found)
stated
The methods for the
estimation of quantities and
unit costs are not described Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Guy, G. P., Jr., Richardson, L. C., Pignone, M. P., & Plescia, M. (2014). Costs and benefits of an organized fecal immunochemical test-based colorectal cancer screening program in the United States. Cancer, 120(15), 2308-
2315. doi:https://dx.doi.org/10.1002/cncr.28724
87
Modality: SEPT9 vs other modalitites varied)
Ladabaum et al, To estimate the Economic Evaluation; cost- Modeled the SEPT9 decreased colorectal cancer Study Limitations =
2013, American comparative utility analysis using a validated contemporary incidence by 35% to 41% and colorectal None
Association for effectiveness and population in the United Economic Evaluation outcome)
decision analytic model cancer mortality by 53% to 61% at costs
Caner Research cost effectiveness comparing mSEPT9, fecal occult States at average risk The research question is
of $8,400 to $11,500/quality-adjusted
Journal of colorectal cancer blood testing (FOBT), fecal for colorectal cancer, not clearly stated Studies are imprecise (When studies
screening with with age specific all- life year gained versus no screening. The perspective of interest
immunochemical testing (FIT), include few patients and few events and
emerging sigmoidoscopy, and cause mortality based is not clear (ie., societal, thus have wide confidence intervals and
biomarkers, on U.S. Life Tables from All established screening strategies patient, health system, payer)
colonoscopy, projecting lifetime the results are uncertain)
illustrated by a 2003 were more effective than mSEPT9. FIT The source(s) of
benefits and costs
methylated Septin was cost saving, dominated mSEPT9, effectiveness estimates are not
9 DNA plasma assay clearly stated Publication Bias
and was preferred among all the
(mSEPT9), versus The primary outcome (e.g. pharmaceutical company sponsors
alternatives.
established measures are not clearly study on effectiveness of drug, only
strategies stated small, positive studies found)
Screening uptake and longitudinal
The methods for the
adherence rates over time strongly estimation of quantities and
influenced the comparisons between Increase Quality Rating if:
unit costs are not described
strategies. At the population level, Large Effect
mSEPT9 yielded incremental benefit at Dose-response gradient
acceptable costs when it increased the Plausible confounders or other
fraction of the population screened biases increase certainty of effect
more than it was substituted for other
strategies Quality (certainty) of evidence for
studies as a whole:
High
Moderate
Low
Very Low
References:
1. Ladabaum, U., Allen, J., Wandell, M., & Ramsey, S. (2013). Colorectal cancer screening with blood-based biomarkers: cost-effectiveness of methylated septin 9 DNA versus current strategies. Cancer Epidemiology, Biomarkers
& Prevention, 22(9), 1567-1576. doi:https://dx.doi.org/10.1158/1055-9965.EPI-13-0204
88
Modality: Fecal DNA varied)
Skally et al, 2013, To identiy key Systematic Review of cost- 7 studies (4 from US) fDNA was cost-effective when Study Limitations =
Appl Health Econ variables that effectiveness studies that undertook an compared with no screening in six None
Health Policy impinge cost economic evaluation of studies. Compared with other Systematic Review or outcome)
effectiveness of fDNA, using either a screening modalities, fDNA was not Review did not address
fDNA as a CRC cost-effectiveness or considered cost-effective in any of focused clinical question Studies are imprecise (When
screening tool cost-utility analysis, the base-case analyses: in five Search was not detailed or studies include few patients and few
compared with other studies it was dominated by all exhaustive events and thus have wide confidence
relevant screening alternatives considered. Sensitivity Quality of the studies was not
intervals and the results are uncertain)
modalities and/or no analyses identified cost, compliance, appraised or studies were of low
screening and test parameters as key influential quality
parameters Methods and/or results were Publication Bias
drug, only small, positive studies found)

Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Skally, M., Hanly, P., & Sharp, L. (2013). Cost effectiveness of fecal DNA screening for colorectal cancer: a systematic review and quality appraisal of the literature. Applied Health Economics & Health Policy, 11(3), 181-192.
doi:https://dx.doi.org/10.1007/s40258-013-0010-8
89
Modality: Various or outcomes varied)
Study (PICO question is quite different
Barzi et al, 2017, To compare the Economic Evaluation; Markov A simulated sample of Colonoscopy emerged as the most effective Study Limitations =
Cancer outcomes of model representing the natural the US general strategy under the base and alternate cases. None
various history of CRC was built and population ages 50 to CT colonography and flexible sigmoidoscopy Economic Evaluation comparison, or outcome)
screening validated against empiric data from 75 years with an were the next 2 most effective strategies, The research question is not
strategies on screening trials as well as the average risk of CRC respectively. DNA testing was more effective clearly stated Studies are imprecise (When
CRC outcomes Microstimulation Screening Analysis was followed for up to than FOBT and FIT by a small margin. The perspective of interest is studies include few patients and few
(MISCAN) model. Thirteen screening 35 years or until death not clear (ie., societal, patient, events and thus have wide
strategies based on colonoscopy, Screening was associated with a 5% to health system, payer)
confidence intervals and the results
sigmoidoscopy, computed 23%relative-risk reduction and a 12% to 34% The source(s) of
tomographic colonography, as well cancer-specific mortality risk reduction effectiveness estimates are not are uncertain)
as fecal immunochemical, occult compared with no screening. The highest clearly stated
blood, and stool DNA testing were risk-reduction levels were associated with The primary outcome Publication Bias
compared with no screening – the colonoscopy strategy. measures are not clearly stated (e.g. pharmaceutical company
societal perspective The methods for the sponsors study on effectiveness of
Colonoscopy emerged as the most effective estimation of quantities and unit drug, only small, positive studies
screening strategy with the highest life years costs are not described
found)
gained (0.022 life years) and CRCs prevented
(n = 1068) and the lowest total costs
($2861). These values were 0.012 life years Increase Quality Rating if:
gained, 574 CRCs prevented, and a total cost Large Effect
of $3164, respectively, for FOBT; and 0.011 Dose-response gradient
life years gained, 647 CRCs prevented, and a Plausible confounders or other
total cost of $4296, respectively, for DNA biases increase certainty of effect
testing.
Authors noted that improved sensitivity or Quality (certainty) of evidence for

specificity of a screening test for CRC studies as a whole:
detection was not sufficient to close the High
outcomes gap compared with colonoscopy
90
Subramanian et To develop an Economic Evaluation; updated a US population On average, the incremental cost of risk Study Limitations = Moderate
al, 2017, Cancer innovative previously validated stratified screening for colorectal cancer None Low
Causes and model to assess microsimulation model consisting of compared to the current approach at 60% Economic Evaluation Very Low
Control the three interlinked components: risk and 80% compliance rates is $18,342 and The research question is not
effectiveness, assessment, natural history, and $23,961 per life year gained. The harms in clearly stated
cost, and harms screening/treatment modules. Used terms of false positives and perforations are The perspective of interest is
of risk stratified data from representative national consistently lower for personalized scenarios not clear (ie., societal, patient,
colorectal surveys and the literature to create across all compliance rates. False positives health system, payer)
cancer a synthetic population that mimics are reduced by more than 47.0% and The source(s) of
screening the family history and genetic perforations by at least 9.9%. There is effectiveness estimates are not
profile of the US population. Applied considerable uncertainty in the life years clearly stated
risk stratification based on gained, but the reduction in harms remains The primary outcome
published risk assessment tools to stable under all scenarios measures are not clearly stated
triage individuals into five risk The methods for the
categories: high, increased, estimation of quantities and unit
medium, decreased, and low costs are not described
See table 1 below
References:
1. Barzi, A., Lenz, H. J., Quinn, D. I., & Sadeghi, S. (2017). Comparative effectiveness of screening strategies for colorectal cancer. Cancer, 123(9), 1516-1527. doi:10.1002/cncr.30518
2. Subramanian, S., Bobashev, G., Morris, R. J., & Hoover, S. (2017). Personalized medicine for prevention: can risk stratified screening decrease colorectal cancer mortality at an acceptable cost? Cancer Causes & Control, 28(4),
299-308. doi:10.1007/s10552-017-0864-4
91
Table 1. Framework for Risk Stratification, Screening Schedules and Scenarios --- from Subramanian study
92
Question #7: In screening for CRC, what tools and/or strategies are most effective in identifying high-risk patients?
Primary Literature:
PICO Question: In screening for CRC, what tools and/or strategies are most effective in identifying high-risk patients? Lower Quality Rating if:
Modality: Clinical Scoring System; Outcome: CRC Studies inconsistent (wide
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations studies, populations, interventions,
Study Methods or outcomes varied)
Total # of Studies: 3 # of Systematic Reviews: 1 # of Non-Randomized Studies: 2
Usher-Smith, J.A., To systematically Systematic Review 18 papers describing Nine models were developed in primary care Study Limitations =
et al., 2016, review and 15 risk models were populations and six in secondary care. Four None Studies are indirect
Cancer Prevention compare the included had good discrimination (AUROC > 0.8) in Systematic Review (PICO question is quite different
Research performance of external validation studies, and sensitivity Review did not address from the available evidence in
models that predict and specificity ranged from 0.25 and 0.99 to focused clinical question regard to population, intervention,
the risk of 0.99 and 0.46 depending on the cut-off Search was not detailed or comparison, or outcome)
undiagnosed chosen. Seventeen variables exhaustive
prevalent primary were included in three or more models: four Quality of the studies was
CRC for demographic variables (age, sex, smoking, not appraised or studies were of
symptomatic alcohol); family history of low quality studies include few patients and few
individuals CRC; eight symptoms (rectal bleeding, Methods and/or results were events and thus have wide
change in bowel habit, diarrhoea, inconsistent across studies confidence intervals and the results
constipation, abdominal pain, weight loss, are uncertain)
loss of appetite, mucous in the stool);
abnormal rectal examination; and three
Publication Bias
investigations (haemoglobin, mean cell
volume, faecal occult blood testing). All
models included symptoms, four included sponsors study on effectiveness of
only symptoms, and most also included age drug, only small, positive studies
(n = 11) and sex (n = 9). found)

Large Effect
93

studies as a whole:
High
Moderate
Low
Very Low
Hsu, L., et al., To refine models Retrospective Study; 120,000 participants A CRC genetic risk score that independently Study Limitations =
2015, designed to Developed risk determination predicted which patients in the training set None
Gastroenterology determine risk of models based on sex, age, would develop CRC was identified. Non-Randomized Studies
CRC by family history, genetic risk Compared with determination of risk based Failure to develop and apply
incorporating score (number of risk alleles only on family history, adding the genetic appropriate eligibility criteria
information from carried at 27 validated risk score increased discriminatory accuracy Flawed measurement of
common genetic common CRC susceptibility from 0.51 to 0.59 (P=.0028) for men and both exposure and outcome
susceptibility loci loci), and history of endoscopic from 0.52 to 0.56 (P=.14) for women. The Failure to adequately control
examinations from data age- and sex-specific 10 y CRC absolute risk confounding
collected in six studies was calculated and estimates were based on Incomplete or inadequately
performed from 1990 through the number of risk alleles, family history, and short follow-up
2011 in the United States and history of endoscopic examinations. A model Differences in important
Germany. The model was that included a genetic risk score better prognostic factors at baseline
validated using data collected determined the recommended starting age
from approximately 1800 for screening in subjects with and without
participants in the Prostate, family histories of CRC. The starting age for
Lung, Colorectal, and Ovarian high-risk men (family history of CRC and
Cancer Screening Trial, genetic risk score=90%) was 42 y, and for
conducted from 1993 through low-risk men (no family history of CRC and
2001 in the United States. genetic risk score=10%) was 52 years. For
men with no family history and a high
genetic risk score (90%), the starting age
would be 47 years; this is an intermediate
value that is 5 years earlier than it would be
for men with a genetic risk score of 10%.
Similar trends were observed in women.
References:
1. Hsu, L., et al. (2015). "A model to determine colorectal cancer risk using common genetic susceptibility loci." Gastroenterology 148(7): 1330-1339.e1314.
2. Usher-Smith, J. A., et al. (2016). "Risk Prediction Models for Colorectal Cancer: A Systematic Review." Cancer Prevention Research 9(1): 13-26.
94
Modality: Clinical Scoring System; Outcome: Adenoma Risk Studies inconsistent (wide
Author/Date Purpose of Study Design & Methods Sample Outcomes Design Limitations studies, populations, interventions,
Study or outcomes varied)
Total # of Studies: 2 # of Non-Randomized Studies: 2
Shaukat, A., et To develop and Cohort Study; Baseline data from 451 subjects were Variables in the prediction of adenoma Study Limitations =
al., 2015, validate a men and women who underwent included in the risk for colonoscopy screening were age None Studies are indirect
Biomarkers & clinical score screening colonoscopy from the RCT development model and (likelihood ratio test for overall Non-Randomized Studies (PICO question is quite different
Prevention for predicting National Colonoscopy Study (NCS) 1,334 subjects were contribution to model, P < 0.001), male Failure to develop and apply from the available evidence in
risk of was used to develop and validate an included in the sex (P < 0.001), body mass index (P < appropriate eligibility criteria regard to population, intervention,
adenoma from adenoma risk model. The study validation of the risk 0.001), family history of at least one first- Flawed measurement of comparison, or outcome)
screening asked all participants to complete score degree relative with colorectal cancer (P = both exposure and outcome
colonoscopy baseline questionnaires on clinical 0.036), and smoking history (P < 0.001). Failure to adequately control Studies are imprecise (When
risk factors and family history. The The adjusted AUROCC of 0.67 [95% confounding
risk score developed on phase I confidence interval (CI), 0.61-0.74] for studies include few patients and few
participants was applied to phase II the derivation cohort was not statistically short follow-up events and thus have wide
participants. Risk score was significantly different from that in the Differences in important confidence intervals and the results
estimated from logistic regression validation cohort. The adjusted AUROCC prognostic factors at baseline are uncertain)
yielded an area under the receiver for the entire cohort was 0.64 (95% CI,
operating characteristics curve 0.60-0.67).
Publication Bias
(AUROCC).
Cao, Y., et al., To develop a Prospective cohort study; Utilized 4,881 asymptomatic A total of 330 (6.7%) men and 678 (3.8%) Study Limitations = sponsors study on effectiveness of
2015, risk assessment the Health Professionals Follow-up white men and 17,970 women were diagnosed with high-risk None drug, only small, positive studies
International tool for high- Study (HPFS) and Nurses’ Health women who underwent adenoma at first-time screening Non-Randomized Studies found)
Journal of Cancer risk colorectal Study (NHS) participant data. colonoscopy as first colonoscopy. The model for men included Failure to develop and apply
adenoma Participants were asked in time screening for CRC age, family history of colorectal cancer, appropriate eligibility criteria Increase Quality Rating if:
(advanced questionnaire whether polyps had BMI, smoking, sitting watching TV/VCR, Flawed measurement of Large Effect
adenoma or been diagnosed in the past two regular aspirin/NSAID use, physical activity, both exposure and outcome Dose-response gradient
>/= 3 years. When a diagnosis was and a joint term of multivitamin and Failure to adequately control
adenomas) that reported, investigators acquired alcohol. For women, the model included Plausible confounders or other
confounding
can be medical records and pathology age, family history of colorectal cancer, Incomplete or inadequately biases increase certainty of effect
implemented in reports. A risk assessment was BMI, smoking, alcohol, beef/pork/lamb as short follow-up
clinical/general conducting including age, personal main dish, regular aspirin/NSAID, calcium, Differences in important Quality (certainty) of evidence for
settings history of diabetes, family history of and oral contraceptive use. The C-statistic prognostic factors at baseline studies as a whole:
through colorectal cancer in first-degree of the model for men was 0.67 and 0.60
High
evaluating a relative, regular use of aspirin, for women (0.64 and 0.57 in cross-
comprehensive multivitamin use, body mass index, validation). Both models calibrated well. Moderate
list of risk height, calcium intake from The predicted risk of high-risk adenoma Low
factors supplement and food, supplemental for men in the top decile was 15.4% vs Very Low
vitamin D, red meat intake, smoking 1.8% for men in the bottom decile (Odds
95
status, alcohol intake, and sedentary Ratio [OR]=9.41), and 6.6% vs 2.1% for
behavior. Additional, oral women (OR=3.48).
contraceptive use, menopausal
status, and postmenopausal
hormone use was evaluated in
women.
References:
1. Cao, Y., et al. (2015). "Assessing individual risk for high-risk colorectal adenoma at first-time screening colonoscopy." International Journal of Cancer 137(7): 1719-1728.
2. Shaukat, A., et al. (2015). "Development and validation of a clinical score for predicting risk of adenoma at screening colonoscopy." Cancer Epidemiology, Biomarkers & Prevention 24(6): 913-920.
Modality: Clinical Scoring System; Outcome: Colorectal Neoplasia Studies inconsistent (wide
Author/Date Purpose of Study Design & Sample Outcomes Design Limitations studies, populations, interventions,
Wong, M.C., et al., To develop and Prospective Cohort Study; 5,220 asymptomatic The prevalence of colorectal neoplasia in Study Limitations =
2014, Gut validate a Random sampling of Chinese screening participants; the derivation and validation cohorts was None Studies are indirect
clinical scoring participants undergoing 2,000 participants 31.4% and 30.8%, respectively. Using the Non-Randomized Studies (PICO question is quite different
system to colonoscopy in Hong Kong. evaluated for independent scoring system developed, 78.9% and Failure to develop and apply from the available evidence in
predict the risks Independent risk factors were risk factors 21.1% in the validation cohort were appropriate eligibility criteria regard to population, intervention,
of colorectal evaluated from first cohort for classified as AR and HR, respectively. The Flawed measurement of comparison, or outcome)
neoplasia to colorectal neoplasia, defined as prevalence of colorectal neoplasia in the both exposure and outcome
better inform adenoma, advanced neoplasia, AR and HR groups was 27.1% and 44.6%, Failure to adequately control Studies are imprecise (When
screening colorectal cancer or any respectively. The subjects in the HR group confounding
participants combination therefore using had 1.65-fold (95% CI 1.49 to 1.83) studies include few patients and few
and facilitate binary regression analysis. The increased prevalence of colorectal short follow-up events and thus have wide
their screening ORs from cohort for significant neoplasia than the AR group. Differences in important confidence intervals and the results
test choice risk factors were used to prognostic factors at baseline are uncertain)
develop a scoring system
ranging from 0 to 6: 0 – 2
Publication Bias
“average risk” (AR) and 3 – 6
“high risk” (HR). The risk factors
examined included age, gender, sponsors study on effectiveness of
family history of CRC, smoking, drug, only small, positive studies
drinking (current drinkers of found)
alcohol for more than two times
per week vs those drinking less Increase Quality Rating if:
or non-drinkers), Body Mass
Large Effect
index (BMI), self-reported
96
medical conditions, use of non- Dose-response gradient

steroidal anti-inflammatory Plausible confounders or other
agents (NSAIDS) and aspirin. biases increase certainty of effect

studies as a whole:
High
Moderate
Low
Very Low
Ruco, A., et al., To conduct an Cohort Study; One cohort 5,137 asymptomatic The prevalence of AN in the study cohort Study Limitations =
2015, BMC external consisted of patients completing participants was 6.8 %. The likelihood of detecting AN None
Gastroenterology evaluation of a baseline questionnaire that increased from 3.6 to 13.1 % for those with Non-Randomized Studies
previously covered demographic a risk score of 1 to 6 respectively. The c- Failure to develop and apply
published risk information, history of prior statistic for the multivariable logistic appropriate eligibility criteria
index for colon examinations, medical model in our cohort was 0.64 (95 % CI = Flawed measurement of
advanced history, prior surgeries, smoking 0.61–067) indicating modest overlap both exposure and outcome
neoplasia (AN) history, alcohol consumption, between risk scores. Failure to adequately control
physical activity, non-steroidal confounding
anti-inflammatory drug (NSAID) Incomplete or inadequately
use, and family history of short follow-up
cancer. The external validation Differences in important
cohort assessed the strength of prognostic factors at baseline
the association between the
predictors of AN. The analysis
used the risk score developed
by Kaminski et al. with updated
definitions to include
97
participants as old as 74 years of

age.
Kim, D.H., et al., To develop and Prospective Cohort Study; 3,561 subjects in derivation 4.7% in the derivation cohort and 4.3% in Study Limitations =
2015, Journal of validate a risk Developed Korean Colorectal cohort and 1,316 subjects the validation cohort had a prevalence of None
Clinical stratification- Screening (KCS) score by in the validation cohort advanced neoplasia. After a multivariate Non-Randomized Studies
Gastroenterology based optimizing and adjusting Asia- analysis, KCS was developed as 0 to 8 Failure to develop and apply
screening Pacific Colorectal Screening points comprising of age, sex, body mass appropriate eligibility criteria
model for (APCS) score to predict index, smoking, and family history of CRC. Flawed measurement of
predicting advanced neoplasia in Using KCS scores to stratify the validation both exposure and outcome
colorectal asymptotic Korean population cohort, the prevalence of advanced Failure to adequately control
advance who received screening neoplasia in the 3 risk tiers (average, confounding
neoplasia in colonoscopy. The moderate, and high) were 2.0%, 3.7%, and Incomplete or inadequately
Korea KCS score was divided into 3 risk 10.9%, respectively. Moderate-risk and short follow-up
tiers: score 0 to 1, “average risk high-risk tiers showed 2.1- and 6.5-fold Differences in important
(AR)”; 2 to 3, “moderate risk increased prevalence, respectively, of prognostic factors at baseline
(MR)”; and 4 to 8, “high risk advanced neoplasia compared with
(HR).” Score validities were average risk tier. In addition, KCS score
assessed by receiver operating showed relatively good discriminative
characteristics (ROC) analysis, power (ROC=0.681) and higher sensitivity
and the predicted risks of compared with APCS score for the high-
advanced neoplasia were risk tier.
assessed by comparing risk tier
groups. Each subject in the
validation group was assigned a
personal risk score, as
98
calculated by the KCS scoring

method. The performance of
the KCS score to predict the risk
of advanced neoplasia was
evaluated by determining the
ORs for each risk tier, and the
validity of the KCS score was
assessed by ROC analysis.
Ladabaum, U., et To explore Prospective Study; Individuals 509 screenees; 256 women 58 had AN. The prevalence of AN Study Limitations =
al., 2016, Cancer whether the who underwent colonoscopy and 253 men increased progressively from 6% in the None
National Cancer were invited to complete a risk lowest risk-score quintile to 17% in the Non-Randomized Studies
Institute (NCI) factor questionnaire that highest risk-score quintile (P = .002). Risk- Failure to develop and apply
CRC risk- included the NCI tool questions score distributions in individuals with appropriate eligibility criteria
assessment before their colonoscopy. The versus without AN differed significantly Flawed measurement of
tool, which was 10-year NCI risk score was (median, 1.38 [0.90–1.87] vs 1.02 [0.62– both exposure and outcome
developed to calculated for each participant 1.57], respectively; P = .003), with Failure to adequately control
predict future using the appropriate algorithm substantial overlap. The discriminatory confounding
CRC risk, could for women and men, and the accuracy of the tool was modest, with Incomplete or inadequately
predict current association between AN at areas under the curve of 0.61 (95% short follow-up
AN prevalence colonoscopy and predicted CRC confidence interval [CI], 0.54–0.69) Differences in important
in a diverse risk was analyzed. overall, 0.59 (95% CI, 0.49–0.70) for prognostic factors at baseline
population, women, and 0.63 (95% CI, 0.53–0.73) for
thereby The predictors included in the men. The results did not change
allowing its use NCI CRC risk-assessment tool for substantively when the analysis was
in risk women are an age indicator, restricted to adenomatous lesions or to
stratification body mass index, estrogen screening procedures without any
for screening. status within the last 2 years, additional incidental indication.
servings of vegetables per day,
aspirin and nonsteroidal anti-
inflammatory drug use, prior
negative sigmoidoscopy and/or
colonoscopy, polyp history,
number of relatives with CRC,
and current vigorous leisure
time activity (with different
individual factors used to
99
predict proximal CRC, distal CRC,

or rectal cancer).
The predictors included in the

NCI CRC risk-assessment tool for
men are body mass index,
servings of vegetables per day,
aspirin and nonsteroidal anti-
inflammatory drug use, usual
number of cigarettes smoked
per day and years of smoking in
current and former smokers,
prior negative sigmoidoscopy
and/or colonoscopy, polyp
history, number of relatives
with CRC, and current vigorous
leisure time activity (with
different individual factors used
to predict proximal CRC, distal
CRC, or rectal cancer).
Park, Y.M., et al., To identify risk Cross-sectional Study; Clinical 2,781 subjects The prevalence of overall and advanced Study Limitations =
2017, BMC factors and data were collected on colorectal neoplasm was 20.2% in None
Gastroenterology develop a asymptomatic subjects aged 40– development cohort and 2.5% in validation Non-Randomized Studies
simple 49 years who underwent cohort. Older age (45–49 years), male sex, Failure to develop and apply
prediction colonoscopy for routine health positive serology of Helicobacter pylori, appropriate eligibility criteria
model for examination. Subjects were and high triglyceride and low high-density Flawed measurement of
advanced randomly allocated to a lipoprotein (HDL) levels were both exposure and outcome
colorectal development or validation set. independently associated with an Failure to adequately control
neoplasm in Logistic regression analysis was increased risk of advanced colorectal confounding
asymptomatic used to determine predictors of neoplasm. BMI (body mass index) was not Incomplete or inadequately
individuals advanced colorectal neoplasm. significant in multivariable analysis. We short follow-up
aged 40-49 A simple scoring model for developed a simple scoring model for Differences in important
advanced colorectal neoplasm = advanced colorectal neoplasm (range 0–9). prognostic factors at baseline
Age [0: 40–44, 1: 45–49 years] × A cutoff of ≥4 defined 43% of subjects as
1 + Sex [0: female, 1: male] × 2 + high risk for advanced colorectal
Serology of H. pylori [0: neoplasm (sensitivity, 79%; specificity,
negative, 1: positive] × 2 + High 58%; area under the receiver operating
triglyceride level [0: normal curve = 0.72) in the validation datasets.
range, 1: high] × 2 + Low HDL
level [0: normal range, 1: low] ×
2
The range of the total score for
this risk model was 0–9. This
100
model yielded an AUROC of 0.74

for predicting advanced
neoplasm in the development
set.
Schroy, P.C., et al., To develop and Cross-sectional Study; 3,543 asymptomatic, Smoking was the strongest predictor (net Study Limitations =
2015, American validate a Conducted analysis of patients mostly average-risk reclassification improvement [NRI], 8.4% None
Journal of clinical index undergoing screening patients 50- 79 years of and height the weakest (NRI, 1.5%). Using Non-Randomized Studies
Gastroenterology for estimating colonoscopy at two urban safety age undergoing screening a simplified weighted scored system Failure to develop and apply
the probability net hospitals. Final index colonoscopy based on 0.5 increments of the adjusted appropriate eligibility criteria
of ACN at consisted of 5 independent odds ratio, the risk of ACN ranged from Flawed measurement of
screening predictors of risk (age, smoking, 3.2% (95% CI, 2.6 to 3.9) for the low-risk both exposure and outcome
colonoscopy alcohol, intake, height and a group (score </=2) to 8.6% (95% CI, 7.4- Failure to adequately control
combined sex/race/ethnicity 9.7) for the intermediate/high-risk group confounding
variable). (score 3-11). The model had moderate to Incomplete or inadequately
good over discrimination (C-statistic, 0.69; short follow-up
95% CI, 0.66-0.72) and good calibration (P Differences in important
= 0.73 to 0.93). prognostic factors at baseline
101
References:
1. Kim, D. H., et al. (2015). "Development and validation of a risk stratification-based screening model for predicting colorectal advanced neoplasia in Korea." Journal of Clinical Gastroenterology 49(1): 41-49.
2. Ladabaum, U., et al. (2016). "Predicting advanced neoplasia at colonoscopy in a diverse population with the National Cancer Institute colorectal cancer risk-assessment tool." Cancer 122(17): 2663-2670.
3. Park, Y. M., et al. (2017). "A simple scoring model for advanced colorectal neoplasm in asymptomatic subjects aged 40-49 years." BMC Gastroenterology 17(1): 7.
4. Ruco, A., et al. (2015). "Evaluation of a clinical risk index for advanced colorectal neoplasia among a North American population of screening age." BMC Gastroenterology 15: 162.
5. Schroy, P. C., 3rd, et al. (2015). "A Risk Prediction Index for Advanced Colorectal Neoplasia at Screening Colonoscopy." American Journal of Gastroenterology 110(7): 1062-1071.
6. Wong, M. C., et al. (2014). "A validated tool to predict colorectal neoplasia and inform screening choice for asymptomatic subjects." Gut 63(7): 1130-1136.
Studies inconsistent (wide
Modality: Combining Clinical Scoring System and FIT; Outcome: Identification of High-Risk Patients variation of treatment effect across
or outcomes varied)
Study
Chiu, H.M., et al., To test an Prospective Study; Asymptomatic 5,657 subjects 646 subjects (11.4%) were considered LR, Study Limitations = Studies are indirect
2016, algorithm that individuals visiting bowel cancer 3,243 subjects (57.3%) were considered None (PICO question is quite different
Gastroenterology combined Asia- screening centers or general medical MR, and 1,768 subjects (31.3%) were Non-Randomized Studies from the available evidence in
Pacific outpatient clinics older than 40 considered HR for AN. The proportions of Failure to develop and apply regard to population, intervention,
Colorectal years old in 12 Asia-Pacific regions individuals with an AN in these groups appropriate eligibility criteria comparison, or outcome)
Screening were included. APCS scores were were 1.5%, 5.1%, and 10.9% respectively. Flawed measurement of
(APCS) scores calculated for each individual (0-1 = Compared with LR group, MR and HR both exposure and outcome
with fecal low risk [LR], 2 – 3 = medium risk subjects had a 3.4-fold increase and a 7.8- Studies are imprecise (When
Failure to adequately control
immunochemic [MR], and 4 – 7 = high risk [HR] for fold increase in risk for AN, respectively. A confounding studies include few patients and few
al test (FIT) in advanced neoplasm [AN]. LR and total of 70.6% of subjects with AN (95% Incomplete or inadequately events and thus have wide
colorectal MR subjects were offered FIT and CI: 65.6% - 75.1%) and 95.1% subjects short follow-up confidence intervals and the results
cancer referred for early colonoscopies. with invasive cancers (95% CI: 82.2% - Differences in important are uncertain)
screening 99.2%) were correctly instructed to prognostic factors at baseline
undergo early colonoscopy examination.
Publication Bias
found)

Large Effect
102


studies as a whole:
High
Moderate
Low
Very Low
Cubiella, J., et al., To develop and Retrospective Cohort Study; Using 1,572 patients in For CRC, the odds ratio (OR) of the Study Limitations =
2017, asses the data collected from 5 studies, the derivation cohort and variables included in the Score were: age None
International diagnostic FAST Score (Validation Cohort) was 3,976 patients in (years): 1.03 (95% confidence intervals Non-Randomized Studies
Journal of Cancer accuracy of the compared with the COLONPREDICT validation cohort (CI): 1.02–1.05), male sex: 1.6 (95% CI: Failure to develop and apply
FAST (Fecal Score (Derivation) 1.1–2.3) and f-Hb (0–<20 mg Hb/g feces): appropriate eligibility criteria
hemoglobin 2.0 (95% CI: 0.7–5.5), (20-<200 mg Hb/g): Flawed measurement of
concentration, Derivation cohort – Included 16.8 (95% CI: 6.6–42.0), _200 mg Hb/g: both exposure and outcome
Age, and Sex consecutive symptomatic patients 65.7 (95% CI: 26.3–164.1). The AUC for Failure to adequately control
Test) Score, an referred to colonoscopy from the CRC detection was 0.88 (95% CI: 0.85– confounding
easy to COLONPREDICT study. Patients 0.90) in the derivation and 0.91 (95% CI: Incomplete or inadequately
calculate collected one fecal sample from a 0.90–093; p50.005) in the validation short follow-up
prediction tool single bowel movement during the cohort. At the two Score thresholds with Differences in important
based not only week before the colonoscopy. 90% (4.50) and 99% (2.12) sensitivity for prognostic factors at baseline
on f-Hb, but CRC, the Score had equivalent sensitivity,
also on age and Validation cohort – FAST Score although the specificity was higher in the
sex studies included symptomatic validation cohort (p < 0.001). Accordingly,
patients recruited in five studies the validation cohort was divided into
evaluating the diagnostic accuracy three groups: high (21.4% of the cohort,
of different FIT analytical systems positive predictive value—PPV: 21.7%),
for CRC, advanced neoplasia (AN), intermediate (59.8%, PPV: 0.9%) and low
and significant colonic lesion (SCL) (18.8%, PPV: 0.0%) risk for CRC.
detection or exclusion in
symptomatic patients. Patients were
rapidly allocated to one of three risk
groups: high risk, medium risk and
low risk.
103
References:
1. Chiu, H. M., et al. (2016). "A Risk-Scoring System Combined With a Fecal Immunochemical Test Is Effective in Screening High-Risk Subjects for Early Colonoscopy to Detect Advanced Colorectal Neoplasms." Gastroenterology
150(3): 617-625.e613.
2. Cubiella, J., et al. (2017). "The fecal hemoglobin concentration, age and sex test score: Development and external validation of a simple prediction tool for colorectal cancer detection in symptomatic patients." International
Journal of Cancer 140(10): 2201-2211.
Modality: Patient Characteristics; Outcome: CRC Studies inconsistent (wide
Carroll, J.C., et To assess the Data from the BETTER 775 participants The mean age of participants was 52.5 Study Limitations = Studies are indirect
al., 2017, proportion of (Building on Existing Tools to years and 72% were female. A minimum None (PICO question is quite different
Canadian Family primary care patients Improve Chronic Disease of 12% of patients (range 12% to 36%) had Non-Randomized Studies from the available evidence in
Physician who report a family Prevention and Screening in a reported FH of 1 of 4 chronic diseases. Failure to develop and apply regard to population, intervention,
history (FH) of type 2 Primary Care) trial were used. Among patients with positive FH, the appropriate eligibility criteria comparison, or outcome)
diabetes, coronary Patients were mailed following proportions of patients had Flawed measurement of
artery disease, questionnaires. Baseline FH that FH recorded in the EMR compared both exposure and outcome
breast cancer, or and screening data were with the questionnaire: CRC, 12% in the Studies are imprecise (When
Failure to adequately control
colorectal cancer, obtained for enrolled patients EMR versus 14% on the questionnaire, confounding studies include few patients and few
assess concordance from the EMR and health kappa = 0.510. There was moderate Incomplete or inadequately events and thus have wide
of FH information questionnaires. agreement for CRC. The presence of FH short follow-up confidence intervals and the results
derived from the was a significant predictor of CRC Differences in important are uncertain)
electronic medical screening (odds ratio 1.9, 95% CI 1.1 to prognostic factors at baseline
record (EMR) 3.1).
compared with Publication Bias
patient-completed (e.g. pharmaceutical company
health sponsors study on effectiveness of
questionnaires; and drug, only small, positive studies
assess whether found)
appropriate
screening was
informed by risk
based solely on FH. Large Effect
104

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Carroll, J. C., et al. (2017). "Assessing family history of chronic disease in primary care: Prevalence, documentation, and appropriate screening." Canadian Family Physician 63(1): e58-e67.
Modality: Patient Characteristics; Outcome: Colorectal Adenoma Studies inconsistent (wide
Lee, S.E., et al., To investigate Cross-Sectional Study; Compared 2,819 patients Prevalence of colorectal adenoma and Study Limitations =
2016, World prevalence and prevalence and characteristics of advanced adenoma were 19.7% and 1.5%, None Studies are indirect
Journal of risk factors for colorectal and advanced respectively. As patient age increased, so Non-Randomized Studies (PICO question is quite different
Gastroenterology colorectal adenomas in patients aged < 50 did the prevalence of colorectal neoplasm. Failure to develop and apply from the available evidence in
neoplasms in years who underwent However, prevalence of advanced appropriate eligibility criteria regard to population, intervention,
adults aged < 50 colonoscopy screening with adenoma did not differ between age- Flawed measurement of comparison, or outcome)
years, for whom subjects aged >= 50 years. To groups 45-49 years and >= 50 years (OR = both exposure and outcome
screening is not evaluate risk factors for colorectal 0.43, 95%CI: 0.17-1.07, P = 0.070). In Failure to adequately control Studies are imprecise (When
recommended. and advanced adenoma in young younger age-group (< 50 years), colorectal confounding
adults, we used multivariable adenoma was significantly associated studies include few patients and few
logistic regression models. with older age, waist circumference (OR = short follow-up events and thus have wide
Colorectal neoplasm 1.72, 95%CI: 1.15-2.55, P = 0.008), and Differences in important confidence intervals and the results
characteristics were evaluated current smoking (OR = 1.60, 95%CI: 1.07- prognostic factors at baseline are uncertain)
and compared with those in older 2.41, P = 0.023). Alcohol consumption was
patients. an independent risk factor for colorectal
Publication Bias
advanced adenoma (OR = 3.69, 95%CI:
1.08-12.54, P = 0.037). Multiple neoplasms
and large neoplasms (>= 1 cm) were more sponsors study on effectiveness of
prevalent in subjects >= 50 years. drug, only small, positive studies
found)
Lieberman, D.A., To measure the Prospective study; Data were 327,785 average-risk Risk of large polyps progressively Study Limitations =
et al., 2014, prevalence of derived from the Clinical adults who underwent increases with advancing age beyond age None Increase Quality Rating if:
Gastroenterology significant Outcomes Research Initiative colorectal cancer 75 years in both men (z = 23.5, P < 0.001) Non-Randomized Studies
Large Effect
colorectal polyps (CORI), established in 1995 to and women (z = 17.8, P < 0.001). Women Failure to develop and apply
screening Dose-response gradient
in average-risk study endoscopy utilization and had lower risks than men in every age appropriate eligibility criteria
105
individuals and outcomes in diverse practice group, regardless of race. Blacks had Flawed measurement of Plausible confounders or other
to determine settings throughout the United higher risk than whites from ages 50 both exposure and outcome biases increase certainty of effect
differences States. The colonoscopy reports through 65 years (50-54 years (7.1% vs. Failure to adequately control
based on age, came from 84 practices, including 6.2%; OR: 1.17; 95% CI: 1.02-1.35), 55-59 confounding
sex, race, or community practice and years (8.5% vs. 7.4%; OR: 1.16; 95% CI: Incomplete or inadequately
ethnicity endoscopy centers (78.5%), 0.996-1.36), 60-64 years (11.5% vs. 8.6%; short follow-up studies as a whole:
academic centers (8.3%) and VA OR: 1.38; 95% CI:1.18-1.61), and Hispanics Differences in important High
medical centers (13.2%). had lower risk than whites from ages 50 prognostic factors at baseline Moderate
Colonoscopy indications and through 80 years (5.1% vs. 6.7%; Low
findings were analyzed for patient OR: 0.75; 95% CI 0.70 - 0.79). The Very Low
groups based on age, sex, and prevalence of large polyps was 6.2% in
race/ethnicity. white men 50-54 years old. The risk was
similar among the groups of white women
65-69 years old, black women 55-59 years
old, black men 50-54 years old, Hispanic
women 70-74 years old, and Hispanic men
55-59 years old. The risk of proximal large
polyps increased with age, female sex, and
black race. The proportion with one or
more large polyps steadily increased with
age.
References:
1. Lee, S. E., et al. (2016). "Characteristics of and risk factors for colorectal neoplasms in young adults in a screening population." World Journal of Gastroenterology 22(10): 2981-2992.
2. Lieberman, D. A., et al. (2014). "Race, ethnicity, and sex affect risk for polyps >9 mm in average-risk individuals." Gastroenterology 147(2): 351-358; quiz e314-355.
Modality: Patient Characteristics; Outcome: Colonic Neoplasia Studies inconsistent (wide
Zapatier, J., et al., To evaluate the Retrospective Study; Patients 4,443 patients; 1,197 Among men between 40 and 49 years, Study Limitations =
2015, European influence of BMI undergoing a first-time screening colonoscopies in increasing BMI [odds ratio (OR) = 1.05, None Studies are indirect
Journal of on colonic or average-risk colonoscopy were patients aged between 95% confidence interval (CI): 1.00-1.09] Non-Randomized Studies (PICO question is quite different
Gastroenterology neoplasia in included in the analysis. Data on 40 and 49 years and and BMI of at least 27 (OR=1.95, 95% CI: Failure to develop and apply from the available evidence in
& Hepatology average-risk demographics, smoking, and BMI 3,246 in those aged 1.15-3.29) were predictors of adenomas. appropriate eligibility criteria regard to population, intervention,
patients aged were collected and correlated to between 50 and 59 Younger men with a BMI of at least 27 Flawed measurement of comparison, or outcome)
between 40 and the presence of adenomas and years were more likely to have proximal both exposure and outcome
advanced adenomas. adenomas (OR=2.23, 95% CI: 1.14-4.37) Failure to adequately control
but not advanced adenomas. There was confounding
106
59 years, no relation between BMI and adenomas in Incomplete or inadequately Studies are imprecise (When
analyzed by sex younger women. Among women aged short follow-up studies include few patients and few
between 50 and 59 years, increasing BMI Differences in important events and thus have wide
(OR=1.03, 95% CI: 1.01-1.05) and a BMI of prognostic factors at baseline
confidence intervals and the results
at least 24 (OR=1.43, 95% CI: 1.06-2.94)
was found to be correlated with are uncertain)
adenomas, and increasing BMI was also
found to be associated with proximal Publication Bias
adenomas (OR=1.67, 95% CI: 1.13-2.45). (e.g. pharmaceutical company
Among men aged between 50 and 59 sponsors study on effectiveness of
years, there was no relation between drug, only small, positive studies
BMI and adenomas, but there was a
found)
positive correlation for advanced
adenomas (OR=1.05, 95% CI: 1.002-1.09).
Among women aged between 50 and 59 Increase Quality Rating if:
years, BMI was not predictive of Large Effect
advanced adenomas. Dose-response gradient

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Zapatier, J., et al. (2015). "Can adjusting BMI for age and sex provide for a better predictor of colonic neoplasia?" European Journal of Gastroenterology & Hepatology 27(8): 974-980.
Modality: Predictive Model; Outcome: Lynch Syndrome Identification Studies inconsistent (wide
Khan, O., et al., To evaluate the Retrospective Study; Collected 230 individuals 113 (49%) probands were MMR mutation Study Limitations =
2011, American test data on individuals undergoing carriers. Areas under the receiver None Studies are indirect
Journal of characteristics of genetic testing for MMR gene operator curves were 0.76, 0.78, and 0.82 Non-Randomized Studies (PICO question is quite different
Gastroenterology Lynch syndrome mutations. Each individual's risk for MMRPredict, PREMM(1,2,6), and Failure to develop and apply from the available evidence in
predictive of mutation was examined using MMRPro respectively. While similar in appropriate eligibility criteria
107
models in a MMRPredict, PREMM(1,2,6), and overall performance, study highlights Flawed measurement of regard to population, intervention,
tertiary referral MMRPro. Amsterdam and unique test characteristics of these three both exposure and outcome comparison, or outcome)
group at two US Bethesda criteria were also quantitative models including comparisons Failure to adequately control
academic determined. Testing of sensitivity and specificity. Moreover, confounding
centers characteristics were calculated characteristics were identified of mutation Incomplete or inadequately
for each of the models. Testing carriers who were missed by each model. short follow-up studies include few patients and few
characteristics was calculated for Differences in important events and thus have wide
each of the models. prognostic factors at baseline confidence intervals and the results
are uncertain)
Publication Bias
found)

Large Effect

studies as a whole:
High
Moderate
Low
Very Low
References:
1. Khan, O., et al. (2011). "Performance of Lynch syndrome predictive models in a multi-center US referral population." American Journal of Gastroenterology 106(10): 1822-1827; quiz 1828.
Modality: Serrated Polyps; Outcome: CRC Studies inconsistent (wide
108
Erichsen, R., et al., To study CRC Case-control study; Conducted a 272,342 colonoscopies; Seventy-nine cases and 142 controls had Study Limitations =
2016, risks associated nationwide population-based study 2,045 CRC cases and SSA/Ps (OR, 3.07; 95% confidence interval None Studies are indirect
Gastroenterology with serrated in Denmark with individuals who 8,105 CRC-free [CI], 2.30-4.10). SSA/Ps with cytology Non-Randomized Studies (PICO question is quite different
polyps had received a colonoscopy. For markers of dysplasia were associated Failure to develop and apply
individuals (control) from the available evidence in
each case and control, with a particularly high OR (4.76; 95% CI, appropriate eligibility criteria
investigators identified the first 2.59-8.73). Women with SSA/P had a regard to population, intervention,
Flawed measurement of
colorectal polyp(s) that underwent higher risk for CRC than men with SSA/P both exposure and outcome comparison, or outcome)
a biopsy or were excised during or (OR for women, 5.05; 95% CI, 3.05-8.37 vs Failure to adequately control
after the initial colonoscopy, and OR for men, 2.18; 95% CI, 1.24-3.82); confounding Studies are imprecise (When
obtained tissue blocks for patients with SSA/P proximal to the Incomplete or inadequately studies include few patients and few
hyperplastic lesions. Four expert splenic flexure had the highest risk for short follow-up events and thus have wide
pathologists reviewed these CRC (OR, 12.42; 95% CI, 4.88-31.58). The Differences in important confidence intervals and the results
lesions using current terminology OR for CRC was 4.84 in the 14 cases vs 17 prognostic factors at baseline
for serrated polyps. Logistic controls with TSAs (95% CI, 2.36-9.93), are uncertain)
regression was used to compute 2.51 in the 757 cases vs 1698 controls
odds ratios (ORs) to associate the with conventional adenomas (95% CI, Publication Bias
risk of CRC with polyp type and 2.25-2.80), and 1.30 in the 55 cases vs (e.g. pharmaceutical company
estimated the absolute risks by 235 controls with hyperplastic polyps sponsors study on effectiveness of
multiplying the risk in patients with (95% CI, 0.96-1.77). The 10-year risk for drug, only small, positive studies
no polyps by these ORs. CRC was 4.4% for patients with SSA/P with
found)
dysplasia, 4.5% for patients with TSAs, and
2.3% for patients with conventional
adenomas. Increase Quality Rating if:
Large Effect

studies as a whole:
High
Moderate
Low
Very Low
109
References:
1. Erichsen, R., et al. (2016). "Increased Risk of Colorectal Cancer Development Among Patients With Serrated Polyps." Gastroenterology 150(4): 895-902.e895.
110
Appendix A. GRADE criteria for rating a body of evidence on an intervention

Developed by the GRADE Working Group
Grades and interpretations:

High: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low: Any estimate of effect is very uncertain.
Type of evidence and starting level

Randomized trial–high
Observational study–low
Any other evidence–very low
Criteria for increasing or decreasing level

Reductions
Study quality has serious (–1) or very serious (–2) problems
Important inconsistency in evidence (–1)
Directness is somewhat (–1) or seriously (–2) uncertain
Sparse or imprecise data (–1)
Reporting bias highly probable (–1)
Increases
Evidence of association† strong (+1) or very strong (+2)
Dose-response gradient evident (+1)
All plausible confounders would reduce the effect (+1)
†Strong association defined as significant relative risk (factor of 2) based on consistent evidence from two or more studies with no plausible confounders
Very strong association defined as significant relative risk (factor of 5) based on direct evidence with no threats to validity.
111
Appendix B. Trustworthy Guideline rating scale

The University of Pennsylvania’s Center for Evidence-Based Practice Trustworthy Guideline rating scale is based on the Institute of Medicine’s
“Standards for Developing Trustworthy Clinical Practice Guidelines” (IOM), as well as a review of the AGREE Enterprise and Guidelines International
Network domains.
The purpose of this scale is to focus on the weaknesses of a guideline that may reduce the trust a clinical user can have in the guideline, and
distinguish weaknesses in documentation (e.g. guide-line does not have a documented updating process) from weaknesses in the guidance itself (e.g.
recommendations are outdated). Current quality scales like AGREE emphasize documentation. They are important checklists for developers of new
guidelines, but are less useful for grading existing guidelines. These scales also are harder for clinicians and other persons who are not methodology
experts to apply, and their length discourages their use outside formal technology assessment reports. This new scale is brief, balanced, and easy and
consistent to apply.
We do not attempt to convert the results of this assessment into a numeric score. Instead we present a table listing the guidelines and how they are
rated on each standard. This facilitates qualitative understanding by the reader, who can see for what areas the guideline base as a whole is weak or
strong as well as which guidelines are weaker or stronger.
1. Transparency
A Guideline development methods are fully disclosed.
B Guideline development methods are partially disclosed.
C Guideline development methods are not disclosed.
The grader must refer to any cited methods supplements or other supporting material when evaluating the guideline. Methods should include:
Who wrote the initial draft
How the committee voted on or otherwise approved recommendations
Evidence review, external review and methods used for updating are not addressed in this standard.
2. Conflict of interest
A Funding of the guideline project is disclosed, disclosures are made for each individual panelist, and financial or
other conflicts do not apply to key authors of the guideline or to more than 1 in 10 panel members).
B Guideline states that there were no conflicts (or fewer than 1 in 10 panel members), but does not disclose funding
source.
112
C Lead author, senior author, or guideline panel members (at least 1 in 10) have conflict of interest, or guideline
project was funded by industry sponsor with no assurance of independence.
NR Guideline does not report on potential conflict of interests.
For purposes of this checklist, conflicts of interest include employment by, consulting for, or holding stock in companies doing business in fields
affected by the guideline, as well as related financial conflicts. This definition should not be considered exclusive. As much as anything, this is a
surrogate marker for thorough reporting, since it may be assumed that guideline projects are funded by the sponsoring organization and many authors
think it unnecessary to report a non-conflict.
3. Guideline development group

A Guideline development group includes 1) methodological experts and clinicians and 2) representatives of multiple
specialties.
B Guideline development group includes one of the above, but not both.
C Guideline developers all from one specialty or organization, and no methodologists.
NR Affiliations of guideline developers not reported
The purpose of this standard is to ensure that supporters of competing procedures, or clinicians with no vested interest in utilization of one procedure
or another, are involved in development of the guideline. Both AGREE II and IOM call for patient or public involvement: very few guideline panels have
done so to date, so this is not necessary for guidelines to be rated A. Involvement of methodologists or HTA specialists in the systematic review is
sufficient involvement in the guideline development group for our purposes. In the absence of any description of the guideline group, assume the
named authors are the guideline group.
4. Systematic review
A Guideline includes a systematic review of the evidence or links to a current review.
B Guideline is based on a review which may or may not meet systematic review criteria.
C Guideline is not based on a review of the evidence.
In order to qualify as a systematic review, the review must do all of the following:
Describe itself as systematic or report search strategies using multiple databases
Define the scope of the review (including key questions and the applicable population)
Either include quantitative or qualitative synthesis of the data or explain why it is not indicated
Note: this element does not address the quality of the systematic review: simply whether or not it exists. Concerns about quality or bias of the review
will be discussed in text, where the analyst will explain whether the weaknesses of the review weaken the validity or reliability of the guideline.
113
Note: a guideline may be rated B on this domain even if the review on which it is based is not available to us. This potential weakness of the guideline
should be discussed in text of the report.
5. Grading the supporting evidence

A Specific supporting evidence (or lack thereof) for each recommendation is cited and
graded
B Specific supporting evidence (or lack thereof) for each recommendation is cited but
the recommendation is not graded.
C Recommendations are not supported by specific evidence.
To score a B on this domain there should be specific citations to evidence tables or individual references for each relevant recommendation in the
guideline, or an indication that no evidence was available. Any standardized grading system is acceptable for purposes of this rating. If a guideline
reports that there is no evidence available despite a thorough literature search, it may be scored B on this domain, or even A if evidence for other
recommendations is cited and graded.
6. Recommendations
A Considerations for each recommendation are documented (i.e. benefits and harms of a particular action, and/or strength
of the evidence); and recommendations are presented in an actionable form.
B Either one or the other of the above criteria is met.
C Neither of the above criteria are met
In order to be actionable, the guideline should specify the specific population to which the guideline applies, the specific intervention in question, and
the circumstances under which it should be carried out (or not carried out). The language used in the recommendations should also be consistent with
the strength of the recommendation (e.g. directive and active language like “should” or “should not” for strong recommendations, and passive language
like “consider” for weak recommendations). A figure or algorithm is considered actionable as long as it is complete enough to incorporate all the
applicable patients and interventions. Please see the forthcoming NICE manual (24) for a good discussion of actionability in guidelines.
7. External review
A Guideline was made available to external groups for review.
B Guideline was reviewed by members of the sponsoring body only.
C Guideline was not externally reviewed.
NR No external review process is described.
8. Updating and currency of guideline
114
A Guideline is current and an expiration date or update process is

specified.
B Guideline is current but no expiration date or update process is
specified.
C Guideline is outdated.
A guideline is considered current if it is within the developers’ stated validity period, or if no period or expiration data is stated, the guideline was
published in the past three years (NOTE: the specific period may be changed at the analyst’s discretion, based on whether the technology is mature
and whether there is a significant amount of recent evidence). A guideline must address new evidence when it is updated. A guideline which is simply
re-endorsed by the panel without searching for new evidence must be considered outdated.
115
Appendix C. Search Strategies

PICO 1-3:
Ovid MEDLINE Search Strategy
1 exp Colorectal Neoplasms/ (120015)
2 ((colorect$ or colon or colonic$ or bowel$ or rectum or rectal) adj3 (cancer$ or neoplas$ or tumor$
or tumour$ or adenocarcin$ or carcin$ or malig$)).mp. [mp=title, abstract, original title, name of
substance word, subject heading word, keyword heading word, protocol supplementary concept word,
rare disease supplementary concept word, unique identifier, synonyms] (144610)
3 1 or 2 (149646)
4 exp feces/ or ((feces or faeces or faecal or fecal or stool*) adj3 sampl*).mp. [mp=title, abstract,
original title, name of substance word, subject heading word, keyword heading word, protocol
supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms]
(48539)
5 exp Mass Screening/ (80519)
6 exp "Early Detection of Cancer"/ (14867)
7 5 or 6 (92250)
8 3 and 4 and 7 (503)
9 limit 8 to yr="2015 -Current" (97)
Search Terms/Strategies Used:
10 exp Colorectal Neoplasms/di, dg (21577)
11 ((colorect$ or colon or colonic$ or bowel$ or rectum or rectal) adj3 (cancer$ or neoplas$ or
tumor$ or tumour$ or adenocarcin$ or carcin$ or malig$) adj7 (screen* or ((earl* or routin*) adj5
(diagnos* or detect* or discover*)))).mp. [mp=title, abstract, original title, name of substance word,
subject heading word, keyword heading word, protocol supplementary concept word, rare disease
supplementary concept word, unique identifier, synonyms] (9618)
12 10 or 11 (25572)
13 exp feces/ or ((feces or faeces or faecal or fecal or stool*) adj3 sampl*).mp. (48539)
14 (screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))).mp. [mp=title, abstract,
(584716)
15 12 and 13 and 14 (720)
17 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 ((feces or faeces or
faecal or fecal or stool*) adj3 sampl*)).mp. (602)
116
18 3 and 17 (77)
20 9 or 16 or 19 (127)
21 limit 20 to (comparative study or controlled clinical trial or guideline or meta analysis or
randomized controlled trial or systematic reviews) (18)
22 exp Epidemiologic Studies/ (1633944)
23 20 and 22 (39)
24 21 or 23 (53)
25 limit 24 to english language (53)
26 limit 24 to abstracts (53)
27 25 or 26 (53)
28 20 not 27 (74)
AND

3 1 or 2 (149646)
4 exp sigmoidoscopy/ (2180)
7 5 or 6 (92250)
8 3 and 4 and 7 (895)
10 exp Colorectal Neoplasms/di (17939)
12 10 or 11 (22212)
13 (sigmoidoscop* or (sigmoid* adj7 (endoscop* or scope or scoping or scoped))).mp. [mp=title,
abstract, original title, name of substance word, subject heading word, keyword heading word, protocol
(3597)
117

(584716)
15 12 and 13 and 14 (1540)
17 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 (sigmoidoscop* or
(sigmoid* adj7 (endoscop* or scope or scoping or scoped)))).mp. [mp=title, abstract, original title,
name of substance word, subject heading word, keyword heading word, protocol supplementary
concept word, rare disease supplementary concept word, unique identifier, synonyms] (826)
18 3 and 17 (787)
20 9 or 16 or 19 (109)
23 20 and 22 (26)
24 21 or 23 (51)
27 25 or 26 (51)
28 20 not 27 (58)
AND

3 1 or 2 (149646)
4 (blood adj5 (occult or feces or faeces or faecal or fecal or stool*)).mp. (7255)
7 5 or 6 (92250)
8 3 and 4 and 7 (2463)
118

12 10 or 11 (25572)
13 (blood adj5 (occult or feces or faeces or faecal or fecal or stool*)).mp. (7255)
(584716)
15 12 and 13 and 14 (2859)
17 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 (blood adj5 (occult
or feces or faeces or faecal or fecal or stool*))).mp. (1867)
18 3 and 17 (1752)
20 9 or 16 or 19 (306)
23 20 and 22 (88)
24 21 or 23 (147)
27 25 or 26 (146)
28 20 not 27 (160)
AND

3 1 or 2 (149646)
4 exp feces/ and exp immunochemistry/ (343)
5 ((feces or faeces or faecal or fecal or stool*) adj7 immunochem*).mp. (617)
119
6 4 or 5 (861)
9 7 or 8 (92250)
10 3 and 6 and 9 (507)
14 12 or 13 (25572)
(584716)
16 6 and 14 and 15 (584)
18 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 ((feces or faeces or
faecal or fecal or stool*) adj7 immunochem*)).mp. [mp=title, abstract, original title, name of substance
word, subject heading word, keyword heading word, protocol supplementary concept word, rare
disease supplementary concept word, unique identifier, synonyms] (324)
19 3 and 18 (320)
21 11 or 17 or 20 (152)
24 21 and 23 (46)
25 22 or 24 (67)
28 26 or 27 (67)
29 21 not 28 (85)
AND
120

3 1 or 2 (149646)
4 exp Tomography, X-Ray Computed/ (273244)
7 5 or 6 (92250)
8 3 and 4 and 7 (495)
12 10 or 11 (25572)
13 ((comput* adj3 tomogra*) or ct scan* or cat scan* or (virtual* adj2 (colonogra* or
colonoscop*))).mp. (370260)
(584716)
15 12 and 13 and 14 (1062)
17 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 ((comput* adj3
tomogra*) or ct scan* or cat scan* or (virtual* adj2 (colonogra* or colonoscop*)))).mp. (3674)
18 3 and 17 (331)
20 9 or 16 or 19 (110)
23 20 and 22 (29)
24 21 or 23 (45)
121

27 25 or 26 (45)
28 20 not 27 (65)
AND

3 1 or 2 (149646)
4 exp Colonoscopy/ (19740)
7 5 or 6 (92250)
8 3 and 4 and 7 (3392)
12 10 or 11 (25572)
13 colonoscop*.mp. (25076)
(584716)
15 12 and 13 and 14 (5203)
17 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 colonoscop*).mp.
[mp=title, abstract, original title, name of substance word, subject heading word, keyword heading
word, protocol supplementary concept word, rare disease supplementary concept word, unique
identifier, synonyms] (3397)
18 3 and 17 (2951)
122
20 9 or 16 or 19 (865)
23 20 and 22 (325)
24 21 or 23 (429)
27 25 or 26 (428)
28 20 not 27 (437)
AND

3 1 or 2 (149646)
4 ((hematolog* or blood or plasma or serum) adj3 (test or tests or testing or tested or assay* or
sampl* or draw or drawing or drawn or drew or draws)).mp. (200978)
5 exp Blood Specimen Collection/ or exp Hematologic Tests/ or bl.fs. (939407)
6 4 or 5 (1012330)
9 7 or 8 (92250)
10 3 and 4 and 9 (1816)
14 12 or 13 (25572)
123
(584716)
16 6 and 14 and 15 (2702)
18 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj10 ((hematolog* or
blood or plasma or serum) adj3 (test or tests or testing or tested or assay* or sampl* or draw or
drawing or drawn or drew or draws))).mp. (5841)
19 3 and 18 (1362)
21 11 or 17 or 20 (329)
24 21 and 23 (120)
25 22 or 24 (175)
28 26 or 27 (175)
Database Searched Ovid MEDLINE, Cochrane Database of Systematic Reviews, National Guideline Clearinghouse
Years Searched - All Questions January 2015 – May 2017
Language English
PICO 4-5:
3 1 or 2 (149981)
124

6 4 or 5 (92445)
7 3 and 6 (8061)
or tumour$ or adenocarcin$ or carcin$ or malig$) adj7 (screen* or ((earl* or routin*) adj5 (diagnos* or
detect* or discover*)))).mp. [mp=title, abstract, original title, name of substance word, subject heading
word, keyword heading word, protocol supplementary concept word, rare disease supplementary
concept word, unique identifier, synonyms] (9639)
(586400)
11 8 and 10 (8551)
12 9 or 11 (12557)
13 7 or 12 (13490)
14 exp attitude to health/ (287414)
15 exp health behavior/ (118920)
16 exp sociologic factors/ or exp socioeconomic factors/ (254082)
17 14 or 15 or 16 (548046)
18 exp decision making/ (125680)
19 13 and 17 and 18 (139)
20 ((share or shared or sharing or shares or joint* or collaborat* or confer* or consult* or participat*
or interact* or conversation* or discuss*) adj7 (patient* or decision* or decid* or choice* or
choos*)).mp. (148137)
21 13 and 20 (511)
22 (patient* adj7 (decision* or decid* or choice* or choos* or counsel* or ((inform* or educat*) adj5
(opt or opts or opting or opted or option*)))).mp. [mp=title, abstract, original title, name of substance
word, subject heading word, keyword heading word, protocol supplementary concept word, rare
disease supplementary concept word, unique identifier, synonyms] (65144)
23 13 and 22 (318)
24 (patient* adj7 (prefer* or input* or suggest* or thought* or think* or consider* or feedback) adj5
(decision* or decid* or choice* or choos* or opt or opts or opting or opted or option)).mp. [mp=title,
abstract, original title, name of substance word, subject heading word, keyword heading word, protocol
(5096)
25 13 and 24 (34)
26 exp Physician-Patient Relations/ (41349)
125
27 13 and 26 (163)
28 19 or 21 or 23 or 25 or 27 (890)
Years Searched - All Questions 2002 – May 2017
Language English
PICO 6:
3 1 or 2 (149981)
6 4 or 5 (92445)
7 3 and 6 (8061)
or tumour$ or adenocarcin$ or carcin$ or malig$) adj7 (screen* or ((earl* or routin*) adj5 (diagnos* or
detect* or discover*)))).mp. [mp=title, abstract, original title, name of substance word, subject heading
word, keyword heading word, protocol supplementary concept word, rare disease supplementary
concept word, unique identifier, synonyms] (9639)
(586400)
11 8 and 10 (8551)
12 9 or 11 (12557)
13 7 or 12 (13490)
14 exp cost benefit analysis/ (53627)
15 13 and 14 (586)
126
16 ((compar* or effectiv* or relativ* or total or overall* or benefi* or valu* or evaluat* or estimat*)

adj5 (cost or costs or financ* or expens* or econom* or reimburs* or insur* or dollar* or fiscal*) adj7
(screen* or test* or procedur*)).mp. (12902)
17 13 and 16 (573)
18 ((compar* or effectiv* or relativ* or total or overall* or benefi* or valu* or evaluat* or estimat*)
adj5 (cost or costs or financ* or expens* or econom* or reimburs* or insur* or dollar* or fiscal*) adj7
(((feces or faeces or faecal or fecal or stool*) adj3 sampl*) or (sigmoidoscop* or (sigmoid* adj7
(endoscop* or scope or scoping or scoped))) or (blood adj5 (occult or feces or faeces or faecal or fecal
or stool*)) or ((feces or faeces or faecal or fecal or stool*) adj7 immunochem*) or ((comput* adj3
tomogra*) or ct scan* or cat scan* or (virtual* adj2 (colonogra* or colonoscop*))) or (colonoscop* or
((colon* or bowel*) adj3 (endoscop* or scope* or scoping))))).mp. [mp=title, abstract, original title,
name of substance word, subject heading word, keyword heading word, protocol supplementary
concept word, rare disease supplementary concept word, unique identifier, synonyms] (643)
19 13 and 18 (214)
20 17 or 19 (628)
21 15 or 20 (875)
25 23 or 24 (656)
Language English
PICO 7:
Search Terms/Strategies Used: substance word, subject heading word, keyword heading word, protocol supplementary concept word,
3 1 or 2 (151492)
127
6 4 or 5 (93462)
7 exp Risk/ (845308)
8 exp mortality/ (257936)
9 exp life tables/ (12955)
10 exp epidemiologic studies/ (1662464)
11 7 or 8 or 9 or 10 (2288842)
12 exp "sensitivity and specificity"/ (457054)
13 exp reproducibility of results/ (312603)
14 12 or 13 (650587)
15 3 and 6 and 7 and 12 (342)
16 (screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))).mp. (594407)
(diagnos* or detect* or discover*)))).mp. (9740)
18 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj7 (accura* or
predict*)).mp. [mp=title, abstract, original title, name of substance word, subject heading word,
keyword heading word, protocol supplementary concept word, rare disease supplementary concept
word, unique identifier, synonyms] (16078)
19 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj7 (accura* or predict*
or (success* adj3 rate*))).mp. (16214)
20 19 not 18 (136)
21 ((screen* or ((earl* or routin*) adj5 (diagnos* or detect* or discover*))) adj7 (accura* or predict*
or (success* adj3 (rate* or locat* or identif*)))).mp. (16396)
22 21 not 19 (182)
23 3 and 21 (754)
24 ((accura* or predict* or (success* adj3 (rate* or locat* or identif*))) adj10 ((colorect$ or colon or
colonic$ or bowel$ or rectum or rectal) adj3 (cancer$ or neoplas$ or tumor$ or tumour$ or
adenocarcin$ or carcin$ or malig$) adj7 (screen* or ((earl* or routin*) adj5 (diagnos* or detect* or
discover*))))).mp. [mp=title, abstract, original title, name of substance word, subject heading word,
keyword heading word, protocol supplementary concept word, rare disease supplementary concept
word, unique identifier, synonyms] (310)
25 7 and 23 (190)
26 3 and 6 and 11 and 14 (670)
27 11 and 14 and 17 (682)
28 3 and 11 and 21 (349)
29 25 or 26 or 27 or 28 (1139)
128
32 30 or 31 (1126)
34 limit 33 to (comparative study or controlled clinical trial or evaluation studies or guideline or meta
analysis or randomized controlled trial or systematic reviews or validation studies) (382)
35 33 not 34 (619)
Language English
129

Colorectal Cancer Screening Evidence Summary - 11 21 18 - KK - Edits

Uploaded by

Copyright:

Available Formats

Colorectal Cancer Screening Evidence Summary - 11 21 18 - KK - Edits

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Colorectal Cancer Screening Evidence Summary - 11 21 18 - KK - Edits

Uploaded by

Copyright:

Available Formats

OHSU Health System

Office of Clinical Integration and Evidence-Based Practice

Date started: April 2017

Content Expert Team Members:

Office of Clinical Integration and Evidence-Based Practice Team:

Objective for the Review: To critically review the evidence Definitions:

Colorectal Cancer Screening

Colorectal Cancer: Screening US Preventative Services Task Force 2016

Colorectal Cancer Canadian Task Force on Preventative Health Care 2016

Colorectal Cancer Screening American College of Radiology 2013

Colorectal Cancer Screening American College of Physicians 2012

American College of Gastroenterology Guidelines for Colorectal

MSTF NCCN USPSTF CTF 2016 ACR ACP ACG

8. Currency and updates B B A B B B C

External Guideline Screening Recommendations

US Preventative Services Task Force

Adverse Effects Guideline Recommendations:

Shared Decision-Making Guideline Recommendations:

Cost-Effectiveness Guideline Recommendations:

High-Risk Patients Guideline Recommendations:

The 2009 American College of Gastroenterology stated:

• Screening is recommended in African Americans beginning at age 45 years. (Grade 2 C)

Review of Relevant Evidence: Search Strategies and Databases Reviewed

Evidence Found with Searches

Evaluating the Quality of the Evidence

CONDITIONAL Desirable effects closely balanced with undesirable effects

Quality Type of Evidence

USPSTF 2016 Systematic Review Findings:

Overall Summary of Effectiveness of Screening on CRC Mortality

Increase Quality Rating if:

Quality (certainty) of evidence for

Modeling Studies Summaries

Increase Quality Rating if:

based FOBT = 0.0001) and fewer deaths (HR 0.19, P =

Quality (certainty) of evidence for

Modeling Studies Summaries

USPSTF 2016 Systematic Review Findings:

Increase Quality Rating if:

A lower NNscreen to detect 1 advanced

A significantly higher FPR in men was

No differences were seen when

Multivariate logistic regression

Younger participants (AOR for elderly

The seasonal variations of performance

Increase Quality Rating if:

likely to be detected in a subsequent than an Failure to include all patients

Increase Quality Rating if:

Plausible confounders or other

Quality (certainty) of evidence for

USPSTF 2016 Systematic Review Findings:

Increase Quality Rating if:

Quality (certainty) of evidence for

Author/Date Purpose of Study Design & Sample Outcomes Design Limitations

Quality (certainty) of evidence for

Author/Date Purpose of Study Design & Sample Outcomes Design Limitations

Author/Date Purpose of Study Design & Sample Outcomes Design Limitations

Quality (certainty) of evidence for

Increase Quality Rating if:

Quality (certainty) of evidence for

14 (screen* or ((earl* or routin) adj5 (diagnos or detect* or discover*))).mp. [mp=title, abstract,