Society: AGA

Background: Health status and expected mortality are important considerations in assessing the benefits of colorectal cancer (CRC) screening, particularly for older adults given that CRC screening is recommended for adults with life expectancy ≥10 years. We analyzed past-year CRC screening according to a 10-year mortality index among community-dwelling adults aged 65-84 years.
Methods: We estimated 10-year mortality risk using a validated index comprising age, sex, body mass index, smoking history, comorbidities (diabetes, cancer, chronic obstructive pulmonary disease), number of overnight admissions, perceived health status, and functional limitations (difficulty walking several blocks, instrumental activities of daily life dependency) among 40,842 adults aged 65-84 years in the nationwide National Health Interview Survey in survey years 2000, 2003, 2005, 2008, 2010, 2013, 2015, and 2018. The 10-year predicted mortality risk from the lowest to highest quintiles of risk index were 12%, 24%, 39%, 58%, and 79%, respectively. We assessed the receipt of past-year CRC screening, including invasive (colonoscopy and sigmoidoscopy) and non-invasive (CT colonography and stool-based tests) tests, within 12 months according to quintiles of mortality risk.
Results: Among adults aged 65-84 years, prevalence of past-year CRC screening did not differ by quintile of mortality risk index (23.7%, 25.0%, 24.1%, 22.8%, and 22.5%, respectively, from the lowest to highest quintile, p >0.05, Figure 1). Compared to adults in the lowest quintile of mortality index, adults in the highest quintile had similar odds of past-year screening (OR 0.94, 95% CI: 0.85-1.03, p >0.05). About one-quarter (27.9%) of past-year CRC screening occurred in those with 10-year mortality risk >50%. There were some differences in past-year CRC screening by quintiles of mortality risk across strata of age. For example, adults aged 75-79 years in the lowest mortality quintile had lower past-year CRC screening compared to adults aged 70-74 years in the highest mortality quintiles (18.1% vs. 26.5%, p <0.001). This pattern was consistent for invasive and non-invasive screening tests.
Conclusion: Past-year CRC screening does not differ according to expected 10-year mortality risk. Current age-based approaches to CRC screening may result in under-screening for older, healthier adults but over-screening for younger adults with comorbidities. These data have implications for optimal use of limited endoscopic capacity, as well as possible overdiagnosis. Personalized screening with incorporation of individual factors such as life expectancy should be considered when recommending CRC screening.

<b>Figure 1</b>. Past-year colorectal cancer screening uptake (A), invasive screening (B), and non-invasive screening (C) according to quintiles of mortality index and age group. Mortality index derived from Schonberg <i>et al</i>, <i>J Am Geriatr Soc</i> 2017;<b>65</b>:1310-5.

Figure 1. Past-year colorectal cancer screening uptake (A), invasive screening (B), and non-invasive screening (C) according to quintiles of mortality index and age group. Mortality index derived from Schonberg et al, J Am Geriatr Soc 2017;65:1310-5.

Background: Colorectal cancer (CRC) incidence and mortality are higher in rural compared to urban areas, partly due to lower rural CRC screening rates. Racial and ethnic disparities also exist in CRC incidence and mortality. To reach the National Colorectal Cancer Roundtable goal of 80% screening in every community, we need to better understand the impact of both race/ethnicity and geographic location on CRC screening disparities.
Aim: To examine racial/ethnic and geographic variation in CRC screening across primary care clinics in Wisconsin (WI) utilizing a novel six-category rural-urban geodisparity model.
Methods: EHR data from 1/1/2008 to 12/31/2018 were reviewed from 14 WI health systems. Data included CRC screening test completion and multiple patient, provider, clinic, and health system characteristics. Adults aged 50-75 yrs who have not had a total colectomy and are currently served by one of the health systems were included. Completed CRC screening was defined as: (1) FOBT/FIT within 1 yr, (2) multi-target stool DNA within 3 yrs, (3) sigmoidoscopy or CT colonography within 5 yrs, or (4) colonoscopy within 10 yrs. Race/ethnicity was self-identified by the patient. Clinic location was defined by a novel geodisparity model that incorporates information on regional healthcare capacity and health needs in WI ZIP Code Tabulation Areas. Average predicted screening rates were calculated for the 12 groups of race/ethnicity (White, Non-white) combined with geolocation (urban advantaged, urban, urban underserved, rural advantaged, rural, rural underserved) from a hierarchical model accounting for patient-, provider-, and clinic-level characteristics, as well as interactions between race/ethnicity and geolocation.
Results: 462,742 patients eligible for CRC screening, served by 2,334 providers in 271 clinics across 14 health systems were included. Overall, Non-White patients had lower screening rates compared to Whites (70% vs 80%) and patients living in underserved areas had lower screening rates (rural, 70%; urban, 62%) than other geolocations [Table 1]. Non-White patients in all three categories of rural communities had predicted screening rates ≤62%. Both White and Non-White patients living in urban underserved communities had predicted screening rates ≤65% [Table 2].
Conclusions: When considering the impact of race/ethnicity alone or geolocation alone, Non-White patients and patients in underserved areas (both rural and urban) are less likely to complete CRC screening. The interaction of race/ethnicity and geolocation shows that Non-White patients living in a rural area are even less likely to complete CRC screening than their White counterparts, as well as compared to Non-White patients in urban advantaged and urban areas. Interventions to increase CRC screening in racial/ethnic minorities should also address barriers specific to their geolocations.

Robert Greenlee

Appearances