Modeling disparities in colorectal cancer outcomes: colonoscopy follow-up and quality are key
Year: 2024
Type: editorial
Abstract: Colorectal cancer (CRC) is the second-leading cause of cancer-related death in the United States, and racial and ethnic disparities in outcomes have been well described (1). Non-Hispanic Black Americans are 17% more likely to be diagnosed with CRC and 34% more likely to die from the disease than non-Hispanic White Americans (2). Multiple factors contribute to the disparities between Black and White individuals, and 1 explanation points to the lower utilization of screening among Black Americans (1). From 2001 to 2009, the Delaware Cancer Consortium increased screening colonoscopy uptake in Black and White residents to an equivalent 74%, which helped eliminate the preexisting disparity in CRC incidence in Delaware and nearly eliminated the disparity in CRC mortality (3). Data from the National Health Interview Survey show that CRC screening rates among Black and White Americans have been similar since 2013 (4), however, so the persistent disparities observed in the national incidence and mortality data require an explanation beyond screening uptake. In this issue of the Journal, Alagoz et al. (5) used the CRC-Adenoma Incidence and Mortality microsimulation model to estimate the effect of 2 quality metrics on disparities in CRC outcomes for Black and White individuals: 1) follow-up colonoscopy rate after an abnormal stool test and 2) endoscopist adenoma detection rate. These are arguably the 2 most important quality metrics in a screening setting that uses a combination of noninvasive testing and colonoscopy. Individuals with an abnormal fecal immunochemical test (FIT) have an approximately 10-fold higher risk of CRC than individuals undergoing a screening colonoscopy and a 25-fold higher risk than individuals with a normal FIT result (6,7). The US Multi-Society Task Force on Colorectal Cancer recommends that for a FIT-based screening program, the colonoscopy completion rate after an abnormal FIT should be at least 80% (6). The endoscopist adenoma detection rate is inversely associated with the risk of subsequent CRC and fatal CRC (8), and the recommended minimum adenoma detection rate is 25% for a mixed-sex patient population (9). Alagoz et al. (5) modeled the lifetime CRC outcomes per 1000 Black adults, assuming that the follow-up colonoscopy rate and endoscopist adenoma detection rate for Black individuals are 15% and 10% lower, respectively, than for White individuals. The results showed that eliminating the disparity in follow-up colonoscopy rate would reduce CRC incidence and mortality by 5.2% and 9.3%, respectively, in Black individuals, whereas life-years gained would increase by 3.4%. Erasing the disparity in endoscopist adenoma detection rate would reduce both CRC incidence and mortality by 9.4% and increase life-years gained by 3.7%. Finally, if Black and White Americans had equivalent rates for both metrics, CRC incidence and mortality would decrease by 14.6% and 18.7%, respectively, for Black Americans, and life-years gained would improve by 7.1%. These important findings help quantify the impact of 2 established CRC screening quality indicators on racial disparities. As with any model, details in the input parameters and assumptions are important to consider when interpreting the results. To estimate the difference in follow-up colonoscopy rates, the authors conducted a literature review and identified 2 studies that purportedly had a relative risk of 0.85 for Black vs White individuals (10,11). In 1 of the studies, the minimally adjusted odds ratio was 0.85, but the fully adjusted odds ratio that also accounted for insurance and comorbidities was 0.94, with a confidence interval that spanned 1, leading the study authors to conclude that Black and White individuals had similar odds of follow-up colonoscopy (10). Alagoz et al. (5) appear to address this point by having stated that using odds ratios that adjust for variables with race-specific disparities, such as insurance and comorbidities, is not generalizable. Although this is a sensible acknowledgment of the social determinants of health that explain these differences, the same reasoning was not consistently applied to the risk estimate of the other study (11), which was a hazard ratio of 0.85 that was adjusted for multiple variables, including insurance and comorbidities. If the fully adjusted risk estimates were used from both studies, then the estimated follow-up colonoscopy rate difference between Black and White individuals would be less than 15%, and the impact of this quality metric on CRC disparities would be attenuated. For the adenoma detection rate, the authors identified 5 studies on literature review in which endoscopist adenoma detection rate for Black adults was 3%-26% lower than for White adults. Although the 10% deficit that the authors chose for the base case appears conservative given the range of risk estimates, a weighted average or meta-analysis of the risk estimates would result in a smaller difference. The 2 biggest studies, which combined accounted for 97% of the patients in the 5 studies, showed that endoscopist adenoma detection rate was only 3%-5% lower for Black than for White individuals (12,13). In the study that reported the biggest (26%) difference in adenoma detection rate between Black and White individuals, the difference was not statistically significant (14). This finding was also supported by a large registry study that was not included in the literature review but found that the adjusted odds ratio for adenoma prevalence was 0.92 for Black vs White individuals among 2.8 million persons aged 50 years and older who received screening colonoscopies (15). The above points illustrate the challenges in selecting the appropriate inputs for modeling, especially when there is a paucity of data. Given the uncertainty in the base case scenario that the authors acknowledge, it is especially important to consider the range of possible estimates and corresponding outcomes in the sensitivity analysis they provide. In addition to uncertainties regarding the input values, it is also important to emphasize that the CRC-Adenoma Incidence and Mortality model framework does not incorporate the serrated pathway to CRC, which accounts for approximately 25% of cancers (16). This point is relevant because data suggest that the detection rate of serrated lesions is also lower in Black than in White individuals (15,17). Despite the inherent difficulties with parameter selection that accompany most modeling studies, Alagoz et al. (5) have successfully highlighted the importance of follow-up colonoscopy and adenoma detection rate not only to prevent CRC overall but also to reduce racial disparities. Addressing these disparities will not be easy, but fortunately there are already several evidence-based interventions that can improve these metrics. Follow-up colonoscopy rates can be improved with reminders from health-care professionals and patient navigation (18). The endoscopist adenoma detection rate can be improved by optimizing bowel preparation, applying several endoscopic techniques, and using a variety of adjunctive tools, including artificial intelligence (19). More targeted interventions to improve these CRC screening quality indicators in Black Americans—and perhaps in other racial and ethnic minority groups, as well—are necessary to improve CRC outcomes equitably. No new data were generated or analyzed for this editorial. Peter Liang, MD, MPH (Conceptualization; Writing—original draft; Writing—review & editing); Divya Bhatt, MD (Conceptualization; Writing—original draft; Writing – review & editing). No funding was used for this editorial. Peter Liang has served as a consultant for Guardant Health and Natera and received research support from Freenome. Divya Bhatt has no disclosures or conflicts of interest.
DOI: 10.1093/jnci/djae176
Authors: Peter S Liang, Divya Bhatt
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $5,708
OpenAlex ID: W4401447192