Black and White Differences in Colorectal Cancer Screening and Screening Outcomes: A Narrative Review

Abstract

Racial disparities in colorectal cancer (CRC) incidence are widely documented. There are two potential mechanisms for these disparities: differences in access to screening, including screening follow-up, and differences in underlying risk of CRC. We reviewed the literature for evidence of these two mechanisms. We show that higher CRC incidence in blacks relative to whites emerged only after the dissemination of screening and describe evidence of racial disparities in screening rates. In contrast to the strong evidence for differences in CRC screening utilization, there is limited evidence for racial differences in adenoma prevalence. In general, black and white patients who are screened have similar adenoma prevalence, though there is some evidence that advanced adenomas and adenomas in the proximal colon are somewhat more likely in black than white patients. We conclude that higher rates of CRC incidence among black patients are primarily driven by lower rates of CRC screening. Our findings highlight the need to increase black patients’ access to quality screening to reduce CRC incidence and mortality.

Introduction

Racial disparities in colorectal cancer (CRC) incidence and mortality are well known and widely documented. Relative to non-Hispanic whites (‘whites’), non-Hispanic blacks (‘blacks’) tend to have higher rates of CRC incidence¹, earlier age at diagnosis^2–4, later stage at diagnosis⁵, and worse stage-specific CRC mortality^6–9. These disparities have resulted in calls for earlier initiation of CRC screening for black patients^10,11, especially black men¹².

CRC screening first became available in 1967¹³. In the late 1980s and early 1990s, randomized trials demonstrated the effectiveness of screening to reduce mortality^14–16, leading to CRC screening guidelines in 1995^17,18. Currently available CRC screening tests include stool tests (guaiac-based fecal occult blood tests (FOBT), fecal immunochemical tests (FIT), and stool DNA tests), flexible sigmoidoscopy, colonoscopy, and computed tomographic colonography. Colonoscopy is both a screening test and a diagnostic test used to evaluate symptoms and follow-up abnormal results from other tests. Screening colonoscopy has been shown to reduce both incidence and mortality, with incidence reductions due to primary prevention via the removal of adenomas¹⁹. As recently as 2008, all average-risk individuals were recommended to begin screening at age 50^20,21. In 2009, the American College of Gastroenterology recommended that average-risk African Americans begin screening at age 45²², citing the “high incidence and early onset of colorectal cancer in African Americans”¹⁰. The Institute for Clinical Systems Improvement made a similar recommendation in 2010²³. In 2012, the American College of Physicians recommended that African Americans begin screening at age 40²³. In 2017, the U.S. Multi-Society Task Force on Colorectal Cancer suggested that African Americans begin screening at age 45, but noted that the evidence to support this recommendation was of low quality²⁴. In 2018, the American Cancer Society issued a qualified recommendation that all individuals begin screening at age 45²⁵.

Black-White Differences in CRC Incidence

Before 1985, blacks and whites had similar overall CRC incidence rates and similar age-specific CRC incidence. Figure 1 shows overall CRC incidence rates for blacks and whites based on Surveillance, Epidemiology, and End Results (SEER) Program Data (estimation details are provided in the Appendix)^26,27. In 1985, President Ronald Reagan’s diagnosis with CRC raised awareness about CRC²⁸. After 1985, CRC incidence began to decline, especially in whites. The reasons for this decline are uncertain, but may be attributed to both changes in modifiable risk factors and increased screening²⁹. Figure 2, which is similar to an analyses of SEER data by Murphy and colleagues³⁰, shows that from 1980 to 1984 blacks and whites had similar age-specific CRC incidence rates, but in 2013–2017 black/white age-specific CRC incidence begins to diverge around age 50, the recommended age to initiate CRC screening.

Figure 1.

Colorectal cancer incidence per 100,000 by race from 1975–2017. Datapoints are from the SEER 9 Registry and the trend line is from a Joinpoint regression analysis.

Figure 2.

Colorectal cancer incidence per 100,000 by race and time period, from the SEER 9 Registry.

Two mechanisms could explain racial disparities in CRC incidence. Disparities could be attributable to differences in CRC screening utilization^6,31,32, or differences in risk^33,34. Because most CRC arises through the adenoma-carcinoma pathway³⁵, we use adenoma prevalence as a proxy for CRC risk but do not examine mechanisms underlying risk differences. Differences in risk could be driven by racial differences in modifiable risk factors such as diet, smoking, exercise, and obesity³⁶, by genetic differences^10,33, or by the interplay between modifiable and genetic risk factors. In this paper we review the literature for evidence of these two mechanisms.

Materials and Methods

We searched the PubMed and Web of Science databases to identify articles focused on differential risk of CRC by race and ethnicity or utilization of CRC screening (details, including search terms, are provided in the Appendix). We also searched for citations to two articles focused on racial differences in risk^37,38. Our search identified 2,085 articles. This review includes information from 17 articles that reported screening outcomes (16 articles) or behaviors (14 articles) separately for both black and white US study participants and included at least 100 participants of either black or white race. Screening outcomes included adenoma prevalence, screen-detected CRC rates, interval CRC rates, and screening behaviors (percent up-to-date with guidelines, percent who received a test, and percent who return for repeat testing). We excluded studies reporting only polyp prevalence, rather than adenoma prevalence, as polyps include lesions that may not harbor malignant potential.

Results

Evidence for Racial Differences in CRC Screening Utilization

CRC screening includes multiple steps: initiation of screening, receipt of a follow-up colonoscopy after an abnormal referral test (e.g., after an abnormal stool-based test or flexible sigmoidoscopy), repeat screening after a normal test, and colonoscopy surveillance after adenoma detection.

Table 1 compares screening rates of black and white participants at different steps in the screening process. In addition to observed percentages, Table 1 includes unadjusted odds ratios (ORs) and relative risks (RRs), and adjusted ORs (aORs) and adjusted RRs (aRR). Each study adjusted for somewhat different factors. When multiple adjusted estimates were provided, we focused on estimates that adjusted for factors common across studies: age, sex, income and education.

Table 1.

Evidence for Racial Differences in CRC Screening Utilization

				Unadjusted Percentage With the Screening Behavior				Odds Ratios and Risk Ratios* Black Relative to White With 95% Confidence Intervals
				Black		White
Outcome/data source	Data Type	Data Collection Years	Age Range (Years)	%	N	%	N	Unadjusted	Adjusted**
				Up-to-date with CRC Screening Guidelines
				Based on receipt of FOBT
NHIS (Anderson, 1995)39†	Survey	1987	>=50	15.8	—	22.4	—	—	—
NHIS (Anderson, 1995)39†	Survey	1992	>=50	22.4	—	26.6	—	—	—
MCBS (Doubeni, 2009)40	Survey	2000	65–80	—	761	—	6,705	0.56 (0.42,0.74)	0.84 (0.62,1.13)
MCBS (Doubeni, 2009)40	Survey	2005	65–80	—	636	—	6,023	0.67 (0.45,1.01)	0.80 (0.54,1.20)
Based on receipt of endoscopy
MCBS (Doubeni, 2009)40	Survey	2000	65–80	—	761	—	6,705	0.70 (0.57,0.86)	1.06 (0.88,1.30)
MCBS (Doubeni, 2009)40	Survey	2005	65–80	—	636	—	6,023	0.74 (0.57,0.96)	1.11 (0.86,1.45)
Based on receipt of FOBT/FIT or endoscopy
MEPS (Jerant, 2008)41	Survey	2001–2005	>=50	48.2	2,809	57.2	14,823	—	0.86 (0.77,0.95)
VHA (Burgess, 2010)42	Survey, EHR	2005–2006	50–75	72	328	77	1,827	0.74 —	0.91 —
BRFSS (Liss, 2014)43	Survey	2010	50–75	59.0	—	62.0	—	0.96 (0.94,0.98)	1.02 (1.00,1.04)
NHIS (Sauer, 2018)44	Survey	2010–2013	50–75	56.8	3,567	59.8	14,527	—	—
BRFSS (Sauer, 2018)44	Survey	2012–2014	50–75	67.5	34,191	67.1	362,721	—	—
Health Center Patient Survey (Lee, 2020)45	Survey	2014	>=50	61.3	—	58.6	—	1.07 —	1.28 (1.02,1.60)
BRFSS (May, 2019)46	Survey	2008	50–75	60.6	13,830	63.9	157,150	—	—
BRFSS (May, 2019)	Survey	2010	50–75	64.4	16,871	66.2	180,961	—	—
BRFSS (May, 2019)	Survey	2012	50–75	65.8	17,978	67.4	180,654	—	—
BRFSS (May, 2019)	Survey	2014	50–75	67.8	15,943	68.4	180,150	—	—
BRFSS (May, 2019)	Survey	2016	50–75	66.4	17,518	70.4	186,331	—	—
Receipt of CRC Screening
Screening colonoscopy
National Colonoscopy Study (Mendelsohn, 2017)32†	RCT	2000–2002 2004–2007	40–69	77.9	245	82.9	240	0.94 (0.86,1.02)	1.03 (0.96,1.11)
FOBT or endoscopy
TRICARE (Changoor, 2018)48†	Claims	2010–2013	50–53	56.5	5,062	53.5	17,258	1.14 (1.06,1.23)	1.20 (1.11,1.29)
FIT
ADVANCE (Huguet, 2019)49	EHR	2012–2013	50–64	10.5	32,101	9.2	76,344	—	—
ADVANCE (Huguet, 2019)49	EHR	2014–2015	50–64	17.9	32,936	15.4	81,006	—	—
Complete Screening Episode
PROSPR (Burnett-Hartman, 2016)50	EHR	2010–2012	50–75	50.9	157,999	56.2	938,295	—	0.89 (0.88,0.91)
Follow-up Colonoscopy after Referral
PLCO, 1st screening (Laiyemo, 2010)31	RCT	1993–2001	55–74	62.6	767	72.4	13,743	—	0.88 (0.83, 0.93)
PLCO, 2nd screening (Laiyemo, 2015)51	RCT	2009–2011	58–79	76.6	304	83.1	4,183	—	0.90 (0.84, 0.96)
VHA (Partin, 2017)52	EHR	2009–2011	50–85	55	13,618	48	49,692	1.33 (1.28,1.39)	1.19 (1.14,1.25)
PROSPR (Burnett-Hartman, 2016)50	EHR	2010–2012	50–75	51.9	9,954	56.2	56,298	—	0.94 (0.88,1.00)
Repeat Screening
PLCO, 2nd screening (Laiyemo, 2015)51	RCT	2009–2011	58–79	60.6	1609	69.2	31,117	—	—
PROSPR, surveillance colonoscopy (Chubak, 2020)53	EHR	2010–2014	50–89	50.3	—	47.4	—	—	1.17 (0.99,1.39)

Abbreviations: ADVANCE: Accelerating Data Value Across a National Community health center network; BRFSS: Behavioral Risk Factors Surveillance System; EHR: Electronic Health Record; FOBT: fecal occult blood test; FIT: fecal immunochemical test; NHIS: National Health Interview Survey; MCBS: Medicare Current Beneficiary Survey; PLCO: Prostate, Lung, Colorectal and Ovarian; PROSPR: Population-based Research to Optimize the Screening Process; RCT: Randomized Controlled Trial; SEER: VHA: Veterans Health Affairs.

^*Risk ratios are provided in italics to distinguish them from odds ratios.

^**Characteristics adjusted for in aOR and aRR estimates: MCBS⁴⁰: age, sex, education, income, insurance type, usual place of health care, marital status, body mass index (BMI), census division, residence in a metropolitan service area, delayed care due to cost, language of the interview, self-reported general health status, and history of non–skin cancer; MEPS⁴¹: age, sex, education, and income; VHA (Burgess)⁴²: age, education, income, family history of CRC, overall health, comorbidities, substance or psychiatric diagnoses; BRFSS (Liss)⁴³: age, sex, education, income, insurance coverage, usual source of care, checkup in past year; National Colonoscopy Study³¹: age, sex, and education; TRICARE⁴⁸: sex, marital status, beneficiary category (retired, dependent, active duty, other), self or sponsor rank (enlisted, officer, other), type of care facility (military or civilian), self or sponsor service (Air Force, Army, Navy, Coast Guard, Marine, other), region (South, Midwest, Northeast, West); PROSPR (Burnette-Hartman)⁵⁰: age, sex, income, insurance type, length of prior enrollment, type of residence, comorbidity, and study site; PLCO, first screening³¹: age, sex, education, BMI, smoking status, family history of CRC, history of CRC within 3 years of enrollment, history of colon polyps, and screening center; PLCO, second screening⁵¹: age, sex, education, BMI, smoking status, family history of CRC, year of repeat flexible sigmoidoscopy and screening center; VA (Partin)⁵²: age, comorbidity, prior colonoscopy, personal history of polyps, ordering physician, ordering facility. PROSPR (Chubak)⁵³: age, sex, index colonoscopy findings, health system, insurance type, comorbidity index;

^†People of Hispanic ethnicity were included in both black and white groups.

Up-to-date with CRC Screening Guidelines

Most studies examined whether individuals were up-to-date with CRC screening guidelines in place for average-risk individuals at the time of the study, generally defined as having a FOBT or FIT in the last one to three years, receipt of flexible sigmoidoscopy in the last five years, or colonoscopy in the last ten years.

There is evidence that for many years, blacks were less likely than whites to be up-to-date with CRC screening guidelines, and that these differences may be narrowing. In 1987 when the National Health Interview Survey (NHIS) first asked about CRC screening, 15.8% of blacks and 22.4% of whites reported having completed a FOBT in the last 3 years³⁹. Analysis of the 2000, 2003, and 2005 Medicare Current Beneficiary Surveys (MCBS) found that black beneficiaries were less likely than whites to be up-to-date with either FOBT or endoscopic screening, and that these differences could be accounted for by differences in education, income, marital status, type of health insurance, usual place of health care, and self-reported general health status⁴⁰. Three studies of CRC screening in 2001 to 2010 also found that blacks were less likely to be up-to-date with screening than whites, and that these differences could largely be accounted for by differences in education, income, and health insurance^40–43. Analysis of 2010–2013 NHIS data found that the percentage of adults 50 to 75 who were up-to-date with CRC screening was lower for blacks (56.8%, 95% CI (54.4%,59.0%)) compared to whites (59.8%, 95% CI (58.8,60.8))⁴⁴. Analysis of 2012–2014 Behavioral Risk Factor Surveillance System (BRFSS) survey data found similar rates of being up-to-date with screening for blacks (67.5%, 95% CI (66.6%,68.5%)) and whites (67.1%, 95% CI (66.8%,67.4%) ) ⁴⁴. A 2014 patient survey found that blacks were more likely than whites to be up-to-date with screening⁴⁵. In this study, blacks were somewhat more likely to be insured than whites (77.8% versus 75.5%). Analysis of BRFSS data demonstrates steady increases in the percent of both blacks and whites who are up-to-date with screening, with black/white differences narrowing until 2014 when increases stalled among blacks but continued among whites⁴⁶.

Receipt of CRC Screening

Receipt of CRC screening is a simpler outcome than being up-to-date with screening, because it is based solely on test completion and not on past testing. Studies examining receipt of screening were based on randomized controlled trials (RCT), claims, or electronic health record (EHR) data.

Evidence about black versus white differences in receipt of CRC screening is mixed. The National Colonoscopy Study⁴⁷ found no significant difference in the likelihood of undergoing screening colonoscopy for black and white participants³². A study of patients insured by TRICARE in 2010–2013 found that blacks were more likely than whites to be screened⁴⁸. Analysis of data from the Accelerating Data Value Across a National Community (ADVANCE) network of community health centers found no black/white differences in receipt of FOBT from 2012–2015⁴⁹. However, during approximately the same time period the Population-based Research to Optimize the Screening Process (PROSPR) study, which focused on members of a managed care organization, found that black patients were significantly less likely than white patients to complete CRC screening, including follow-up colonoscopy when indicated⁵⁰.

Three studies estimated black/white differences in receipt of follow-up colonoscopy after an abnormal referral test and estimated differences after adjusting for age, sex, comorbidity, and personal history of colon polyps (among other factors). The Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial used flexible sigmoidoscopy as a referral test and found that black participants were significantly less likely than whites to return for follow-up colonoscopy within one year of referral at both the first³¹ and second⁵¹ screening rounds. In contrast, among Veterans Health Affairs (VHA) patients in 2009–2011, black patients were significantly more likely than white patients to have follow-up colonoscopy within 6 months of an abnormal stool test⁵². Finally, analysis of PROSPR data found that black patients were less likely than whites to have follow-up colonoscopy within 90 days of an abnormal stool test⁵⁰.

Repeat Screening

We found only two studies that focused on racial differences in repeat screening. Black PLCO study participants were significantly less likely than white participants to return for screening 3 to 5 years after a normal screening flexible sigmoidoscopy (60.6% vs. 69.3%)⁵¹. Another study of patients in a managed care setting found no evidence of black/white differences in return for surveillance colonoscopy within 3.5 years of an index colonoscopy with high-risk findings⁵³.

Summary

We found evidence that black people have had lower rates of CRC screening than whites. Survey data, which include repeated cross-sectional sampling of the US population across multiple years, indicate potential narrowing of black/white differences over time. Evidence of black/white differences in screening rates from studies of patient populations, primarily based on claims and EHR data, is mixed. The combined evidence across these studies indicates that at least until the recent past, blacks were less likely to be adequately screened for CRC than whites and that socioeconomic factors, including health insurance, factor into these disparities. For example, while the PLCO trial covered the cost of the initial screening flexible sigmoidoscopy, it did not cover the cost of follow-up colonoscopy after an abnormal screening test, and this may have contributed to lower rates of follow-up colonoscopy in black study participants³¹. Only studies that examined black and white patients who had similar rates of insurance found blacks more likely to be screened than whites^45,48,52.

Evidence for Racial Differences in Adenoma Prevalence

The existence of racial disparities in CRC screening utilization does not rule out the potential for racial differences in CRC risk. If there are racial differences in CRC risk, then there should be evidence of differences in observable markers of the underlying disease process. These markers include adenoma prevalence and differences in adenoma size or histology, which could indicate differences in the rate of progression from a precursor lesion to cancer.

Table 2 compares findings at screening for black and white participants, including the unadjusted percentages of participants with outcomes of interest and ORs or RRs that estimate the strength of the relationship between race and screening outcomes. We include aORs and aRRs, though each study adjusted for somewhat different factors. Studies generally reported more than one outcome of interest. Findings by outcomes are summarized below. Consistent with the literature, we define advanced adenomas as adenomas that are ≥10mm or have high-grade dysplasia or villous histology.

Table 2.

Evidence for Racial Differences in Screening Outcomes

				Unadjusted Percentag With the Screening Outcome				Odds Ratios and Risk Ratios*Black Relative to White With 95% Confidence Intervals
				Black		White
Outcome/data source	Data Type	Data Collection Years	Age Range (Years)	%	N	%	N	Unadjusted	Adjusted**
				Adenoma Prevalence
				At Screening Colonoscopy
National Colonoscopy Study (Mendelsohn, 2017)32¶	RCT	2000–2002 2004–2007	40–69	25.1	191	26.1	199	0.93 (0.67,1.28)	0.86 (0.60,1.25)
Stony Brook University (Stein, 2010)57	Survey, Chart Review	2006–2007	>=40	35.8	356	35.7	67
Kaiser (Corely, 2013)38	EHR	2006–2008	>=50	27	892	24	13,266	—	1.15 (0.98,1.35)
Columbia University (Lebwohl, 2012)54	EHR	2006–2010	>=50	25.9	591	19.1	3542		1.76 (1.52.2.04)
Columbia University (Lebwohl, 2012)54	EHR	2006–2010	50–59	19.4	288	16.7	1494
Columbia University (Lebwohl, 2012)54	EHR	2006–2010	60–69	28.0	186	17.2	1168
Columbia University (Lebwohl, 2012)54	EHR	2006–2010	>=70	38.5	117	24.9	880
Temple University (Friedenberg, 2012)55	Chart Review	2007–2010	50–59	42.9	669	38.5	258
Nassau University Medical Center (Zheng, 2014)58	Chart Review	2007–2011	—	21.8	635	24.0	283	—	0.86 (0.60,1.22)
Multitarget Stool DNA (Cooper, 2018)56	Accuracy Study	2012–2015	40–80	38.9	265	33.9	495
CCPN (Eberth, 2018)59	EHR	2014–2016	45–64	32.2	483	39.5	335	0.73 (0.54,0.99)	0.76 (0.54,1.06)
At Follow-up Colonoscopy
PLCO, 1st screening (Laiyemo, 2010)31†	RCT	1993–2001	55–74	50.8	480	54.2	9944	—	1.01 (0.92,1.11)
PLCO, 2nd screening (Laiyemo, 2015)51‡	RCT	2009–2011	58–79	37.3	233	41.9	3,477	—	1.06 (0.89,1.26)
At Surveillance Colonoscopy
Pooled Chemoprevention Study Results (Wallace, 2015)60	Meta-Analysis of RCTs	1984–1998	>50	47.7	172	47.4	2,022	—	1.08 (0.92,1.27)
Advanced Adenoma Prevalence
At Screening Colonoscopy
National Colonoscopy Study (Mendelsohn, 2017)32¶	RCT	2000–2002 2004–2007	40–69	7.3	191	5.5	199	1.19 (0.56,2.52)	1.04 (0.39,2.74)
Stony Brook University (Stein, 2010)57	Survey, Chart Review	2006–2007	>=40	6.0	356	7.3	67
Columbia University (Lebwohl, 2012)54	EHR	2006–2010	>=50	5.4	591	3.7	3542		1.91 (1.27,2.86)
Temple University (Friedenberg, 2012)55	Chart Review	2007–2010	50–59	6.4	669	7.0	257
Boston Medical Center (Schroy, 2013)61	Survey, Chart Review	2005–2012	50–79	5.0	1,681	6.8	1,172	Women:
								1.23 (0.70,2.18)	1.32 (0.73,2.40)
								Men:
								0.59 (0.40,0.88)	0.59 (0.39,0.89)
Multitarget Stool DNA (Cooper, 2018)56	Accuracy Study	2012–2015	40–80	6.8	265	6.7	495
CCPN (Eberth, 2018)59	EHR	2014–2016	45–64	10.9	483	15.5	335
At Follow-up Colonoscopy
PLCO, 1st screening (Laiyemo, 2010)31†	RCT	1993–2001	55–74	23.1	480	22.3	9,944	—	1.11 (0.94,1.30)
PLCO, 2nd screening (Laiyemo, 2015)51‡	RCT	1996–2006	58–79	11.6	233	13.7	3,477	—	1.27 (0.90,1.79)
At Surveillance Colonoscopy
Pooled Chemoprevention Study Results (Wallace, 2015)60	Meta-Analysis of RCTs	1984–1998	>50	13.4	172	14.2	2,022	—	1.05 (0.71,1.56)
Boston Medical Center (Kwah, 2014)62	EHR	2005–2012	>=50	41.9	203	47.4	246	1.18 (0.65,2.16)	1.30 (0.69,2.40)
Prevalence of Proximal Adenomas
At Screening Colonoscopy
Kaiser (Corley, 2013)38	EHR	2006–2008	>=50	17.0	892	14.1	13,266	—	1.26 (1.04,1.54)
Temple University (Friedenberg, 2012)55	Chart Review	2007–2010	50–59	24.2	669	23.7	258
Nassau University Medical Center (Zheng, 2014)58	Chart Review	2007–2011	—	16.8	635	20.5	283	—	—
At Follow-up Colonoscopy
PLCO, 1st screening (Laiyemo, 2010)31	RCT	1993–2001	55–74	21.0	480	19.0	9944	—	1.09 (0.91,1.29)
PLCO, 2nd screening (Laiyemo, 2015)51	RCT	1996–2006	58–79	18.8	233	18.9	3,477	—	1.11 (0.84,1.47)
Prevalence of Advanced Proximal Adenomas
At Screening Colonoscopy
Boston Medical Center (Schroy, 2013)61	Survey, Chart Review	2005–2012	50–79	2.6	1,681	2.6	1,172	—	—
At Follow-up Colonoscopy
PLCO, 1st screening (Laiyemo, 2010)31†	RCT	1993–2001	55–74	8.5	480	5.5	9,944	—	1.56 (1.13,2.14)
PLCO, 2nd screening (Laiyemo, 2015)51‡	RCT	1996–2006	58–79	4.4	233	6.4	3,477	—	1.44 (0.84,2.48)
Screen Detected CRC
At Screening Colonoscopy
Boston Medical Center (Schroy, 2013)61	Survey, Chart Review	2005–2012	50–79	0.4	1,681	0.4	1,172	—	—
At Follow-up Colonoscopy
PLCO, 1st screening (Laiyemo, 2010)31†	RCT	1993–2001	55–74	2.1	480	1.5	9,944	1.58 (0.80,3.12)	—
PLCO, 2nd screening (Laiyemo, 2015)51‡	RCT	1996–2006	58–79	1.3	233	0.5	3,477	—	—
Interval CRC
SEER-Medicare (Fedewa, 2017)67	Claims	2002–2013	66–75	—	4,196	—	51,313	—	1.31 (1.13,1.15)§

Abbreviations: BRFSS: Behavioral Risk Factors Surveillance System; CCPN: Colorectal Cancer Prevention Network; EHR: Electronic Health Record; NHIS: National Health Interview Survey; PLCO: Prostate, Lung, Colorectal and Ovarian; PROSPR: Population-based Research to Optimize the Screening Process; RCT: Randomized Controlled Trial; SEER: Surveillance, Epidemiology, and End Results; VA: Veterans Health Administration.

^*Risk ratios are provided in italics to distinguish them from odds ratios.

^**Characteristics adjusted for in aOR and aRR estimates: Columbia University⁵⁴: age, sex, family history of CRC, insurance status (Medicaid vs. other), participation of a trainee in exam; Nassau University Medical Center⁵⁸: age, sex, tobacco use, body mass index (BMI), indication for colonoscopy, alcohol use, dyslipidemia, hypertension; National Colonoscopy Study³²: age, sex, and education; Kaiser³⁸: age, family history of CRC; CCPN⁵⁹: age, sex, education, rural residence, physician, language spoken; PLCO, first screening³¹: age, sex, smoking status, family history of CRC, BMI, education, history of CRC within 3 years of enrollment, history of colon polyps, and screening center; PLCO, second screening⁵¹: age, sex, smoking status, family history of CRC, BMI, education, year of repeat flexible sigmoidoscopy and screening center; Boston Medical Center⁶¹: age, smoking status, BMI, alcohol use, education, insurance type, NSAID use, aspirin use, use of birth control pills, use of hormone replacement therapy, red meat intake, multivitamin use, calcium intake, physical activity, diabetes, previous colonoscopy. Pooled Chemoprevention Studies⁶⁰: age, sex, study treatment assignment, and follow-up time; SEER-Medicare⁶⁷: age, sex, state of residence, poverty level, urban–rural classification, Charlson comorbidity score, diverticulitis, polyp removal at index colonoscopy, physician specialty, and physician polyp detection rate quartile;

^¶People of Hispanic ethnicity were included in both black and white groups.

^†The PLCO study used flexible sigmoidoscopy as the reference screening test. During the first screening, the percentage of participants with a polyp detected at flexible sigmoidoscopy was 25.5% (N=3,011) and 23.9% (N=57,561) in black and white participants, respectively.

^‡The PLCO study used flexible sigmoidoscopy as the reference screening test. During the second screening, the percentage of participants with a polyp detected at flexible sigmoidoscopy was 31.2% (N=975) and 19.4% (N=21,550) in black and white participants, respectively.

^§Estimated hazard ratio for interval CRC up to 59 months after colonoscopy.

Adenoma Prevalence

We found mixed evidence that adenoma prevalence differs for black and white people. Of eight studies that described racial differences in adenoma prevalence at screening colonoscopy, four estimated higher adenoma prevalence in blacks than whites^38,54–56, two estimated similar adenoma prevalence in blacks and whites^32,57, and two estimated lower adenoma prevalence in blacks than whites^58,59. The study that found the strongest evidence of higher adenoma risk in blacks relative to whites (aRR=1.76), estimated that adenoma prevalence was 2.7 percentage points higher in blacks than whites among participants 50 to 59 years old, 10.3 points higher among participants 60 to 69 years old, and a 13.6 points higher among participants 70 and older⁵⁴. The PLCO study found no detectable black versus white differences in adenoma prevalence among study participants who underwent follow-up colonoscopy, at either the first³¹ or second screen⁵¹, though only patients with abnormal findings at flexible sigmoidoscopy were referred to colonoscopy. Pooled analysis of data from three CRC chemoprevention studies, each with null findings, found no black versus white differences in adenoma prevalence at surveillance colonoscopy⁶⁰.

Advanced Adenoma Prevalence

We found mixed evidence for racial differences in risk for advanced adenomas. Two of seven studies that described racial differences in advanced adenoma prevalence at screening colonoscopy observed a higher prevalence of advanced adenomas in black patients³², while five found that blacks had similar or lower risk of advanced adenomas than whites^{55–57,59,61}. Two of three studies that reported aRRs (or aOR) of advanced adenomas in black versus white study participants at screening colonoscopy estimated an elevated risk of advanced adenomas in black participants⁵⁴, including one study that found black women were at greater risk of advanced adenomas than white women, but black men were at lower risk of advanced adenomas than white men⁶¹. The PLCO study found no racial differences in the prevalence of advanced adenomas detected at the follow-up colonoscopy after the first flexible sigmoidoscopy screening³¹, but analysis of the second screening that controlled for screening center found evidence that blacks had a higher prevalence of advanced adenomas than whites⁵¹. Two studies of surveillance colonoscopy each found that blacks were more likely to have advanced adenomas than whites^60,62.

Proximal Adenoma Prevalence

The location of adenomas is important because adenomas in the proximal colon may be more difficult to detect⁶³ and proximal cancers may have a lower five-year survival rate than distal colon and rectal cancers due to differences in disease characteristics^64,65. Consistent with the literature, we use proximal location to indicate the transverse colon, hepatic flexure, ascending colon, and cecum.

We found mixed evidence for racial differences in the prevalence of proximal adenomas based on results from three observational studies^38,55,58. One study found similar proximal adenoma prevalence in black and white patients⁵⁵. Another study found that blacks were significantly more likely to have proximal adenomas compared to whites, adjusting for only age and family history of CRC³⁸. The third study found that proximal adenoma prevalence was lower in black patients than in white patients⁵⁸. Finally, the PLCO study found that black and white study participants had similar odds of having a proximal adenoma detected at follow-up colonoscopy at both the first and second screening (3 to 5 years later), though only participants with findings at flexible sigmoidoscopy were referred to colonoscopy^31,51.

Prevalence of Advanced Proximal Adenomas

Evidence about the relationship between race and prevalence of advanced proximal adenomas was limited and mixed. An observational study carried out in Boston Medical Center safety-net clinics found that black and white patients had the same prevalence of advanced proximal adenomas⁶¹. In contrast, the PLCO study found that black participants were more likely to have advanced proximal adenomas than white participants^31,51. Estimated racial differences were observed at the first and second round of screening, but only reached statistical significance at the first round.

Interval CRC

Interval CRC refers to CRC detected after colonoscopy and before the next recommended screening exam⁶⁶. Analysis of SEER-Medicare data, based on 66- to 75-year olds who underwent colonoscopy between 2002 and 2011, found that black patients were at significantly greater risk of interval cancer than white patients; the estimated probability of an interval cancer up to 54 months after colonoscopy was 0.071 for black patients and 0.051 for white patients⁶⁷. Interval cancers can be attributed to provider characteristics (adenoma detection rates, rates of complete adenoma/serrated polyp resection), patient characteristics (adequate bowel cleansing), and disease characteristics (progression rates). Thus, it is unclear what drives differences in interval cancer rates.

Summary

We found mixed evidence for racial differences in the prevalence of adenomas and limited evidence that black people may be at greater risk for advanced and proximal adenomas compared to whites. Our findings are similar to results from a recent systematic review and meta-analysis of nine studies that found no difference in the prevalence of advanced adenomas in black (6.6%) and white (6.2%) people but did find evidence that blacks had a higher prevalence of advanced proximal adenomas than whites (3.3% versus 2.4%)⁶⁸.

It is difficult to draw conclusions about racial differences in adenoma risk in the context of screening disparities. Colonoscopy with polypectomy has long-term effects on adenoma prevalence. If whites are more likely to be screened than blacks, this could result in underestimation of both adenoma and advanced adenoma prevalence in whites compared to blacks, even in studies that focused on screening outcomes. For example, in a study that described black/white differences in adenoma prevalence by age, differences were small before age 60, when people are less likely to be previously screened, and large after age 60 when the chance of previous screening increases, especially among whites⁵⁴. On the other hand, if black people receive lower-quality colonoscopy than whites, this could result in underestimation of adenomas and advanced adenomas in blacks compared to whites. Within the PLCO study, racial differences in the prevalence of abnormal findings at the second screening flexible sigmoidoscopy were partly explained by black patients being more likely to have their colonoscopies at centers with lower adenoma detection rates⁵¹.

Discussion

Understanding what drives racial disparities will better equip us to address them. Disparities that are driven by disparate access to quality care point to the importance of improving access to and uptake of high-quality screening. Disparities that arise from disparate risk point to the need for more aggressive screening in higher-risk groups, including earlier initiation and shorter screening intervals, and further exploration of factors that drive increased risk. Based on available data, we found strong evidence for racial differences in CRC screening access, but limited evidence for racial differences in adenoma prevalence. In general, black and white patients who are screened have similar adenoma prevalence. While there is some indication that risk of advanced and proximal adenomas is higher in black than white people, these differences could be driven by differences in screening. Based on this review, we conclude that higher rates of CRC incidence among black patients is largely due to lower rates of screening.

Screening can reduce CRC incidence and mortality, but black Americans face greater barriers to screening than whites⁶⁹. In 2017, black adults younger than 65 were less likely than whites to have health insurance (85.9% versus 91.5%)⁷⁰, and people without health insurance are less likely to participate in CRC screening⁷¹. Racial barriers persist among the insured, where healthcare systems and physicians within these systems play an important role in promoting CRC screening. In spite of widespread knowledge that blacks have higher CRC incidence than whites, blacks are less likely than whites to receive a recommendation for CRC screening^72,73. Physicians are less likely to recommend screening based on their perception of a patient’s ability to pay⁷⁴. Physicians are also less likely to recommend CRC screening when they identify a shortage of specialists in the region, and blacks may be more likely than whites to live in such regions⁷⁵. There is also evidence that black patients receive colonoscopies from physicians with lower adenoma detection rates than white patients ^51,67,76. From the patient perspective, trust in an individual’s primary care provider is the most important factor in the decision to undergo CRC screening, and the degree to which trust plays a role in the decision varies across different racial and ethnic groups⁷⁷.

Our report has several important limitations. This narrative review lacked the rigor of a systematic review and we do not provide pooled estimates of screening outcomes. We described differences in CRC screening and risk for black and white patients within the US and did not address differences across other racial or ethnic groups or in other countries. In addition, because of wide heterogeneity in genetics and ancestry for both “blacks” and “whites” in the US, comparisons between these groups could mask true risk differences in more homogeneous genetic subgroups. We relied primarily on adenoma prevalence as a proxy for underlying CRC risk. We did not address racial differences in the molecular characteristics of detected lesions which may be related to differences in disease progression and survival after diagnosis. Nor did we examine differences in the prevalence of sessile serrated polyps, which could be related to differences in the effectiveness of screening, and both CRC incidence and mortality⁷⁸. Finally, there is potential for bias in study estimates, especially from observational studies. In particular, black/white differences in screening uptake could result in differential underestimation of adenoma and advanced adenoma prevalence, and people at increased risk of CRC due to modifiable risk factors, such as smoking, may be underrepresented in studies of screened populations. In spite of these limitations, we feel that the available research, taken together, provides a compelling argument that the higher CRC incidence observed in blacks relative to whites is primarily due to differences in screening utilization.

It is imperative that we seek ways to reduce racial disparities in CRC screening, but how? Multi-level solutions are needed, including national and state policies to improve access to screening and follow-up of abnormal screening tests, health systems solutions to promote screening, and patient education efforts⁷⁹. Health care organization can ensure that testing is accessible by mailing FITs to all patients⁸⁰ and using patient navigators to assist patients^12,81. Overall, our review suggests that equal access to care can reduce screening disparities.