(267) Impact of using limited duration prevalence (LDP) vs. complete prevalence (CP) on racial/ethnic distributions derived from the Surveillance, Epidemiology, and End Results (SEER) database

Publication Number: 600

Background: The SEER database from US National Cancer Institute is frequently used to estimate cancer prevalence and assess the racial/ethnic distribution among cancer patients. The most recent SEER data includes 21 registries, which cover 48% of the US population and capture incident cancers in 2000-2019, and can be used to derive the 19-year population-based LDP. LDP, which consists of point prevalence on 1 Jan 2019, does not capture cancer cases diagnosed before 2000 who were alive on 1 Jan 2019. To correct for lack of complete capture, SEER provides a correction factor, the Complete Index (CI), calculated from cancer incidence and survival, to derive CP.

Objectives: To assess the impact of deriving racial/ethnic distribution from LDP vs. CP based on SEER data, using multiple myeloma (MM) and acute lymphocytic leukemia (ALL) as case examples.

Methods: We estimated the racial/ethnic distribution among patients with MM and ALL using both LDP and CP from the SEER 21 registry data. ALL was selected because it is the most common childhood cancer with high survival rate, while MM is a recurring and relapsing cancer and common in older adults. LDP estimates were derived for MM and ALL by age (5-year intervals from birth to 85 years), sex (man/woman), and race/ethnicity (non-Hispanic White [NHW], non-Hispanic Black [NHB], and Hispanic [H]) and standardized to the 2022 US Census population. CP was estimated based on age-sex-race-disease-specific CI provided by SEER.

Results: Compared to LDP, CP was 20% higher for MM (145,509 vs. 175,170), and the race-ethnicity distribution was similar between LDP and CP (NHW: 64% vs 65%, NHB: 22% vs 21%, H: 10% vs 10%). CP for ALL was 64% higher than LDP (57,114 vs. 93,645) and the racial/ethnic distribution was similar (NHW: 54% vs 55% NHB: 6% vs 6%, H: 31% vs 30%). For ALL, the CI differs substantially across age groups, ranging from 100% in children (i.e., LDP captured all cases) to 14-16% in females aged 40-44 years old (i.e., LDP captured a small proportion of ALL survivors), whereas the CI in general differs by 5-30 percentage points between race/ethnicity groups. Less variation of CI was observed for MM across age groups and between race/ethnicity groups

Conclusions: The availability of CI varies by race/ethnicity and cancer type. In the absence of CI for certain cancers or racial/ethnic groups, the use of LDP may still inform racial/ethnic distribution in prevalent cancer patients, while adjustment for completeness remains necessary for the purpose of estimating disease burden.