SSAT Home  |  Past Meetings
Society for Surgery of the Alimentary Tract

Back to 2021 Abstracts

Douglas S. Swords*, Timothy E. Newhook, Ching-Wei D. Tzeng, Yun Chun, Thomas Aloia, Ahmed Kaseb, Jean-Nicolas Vauthey, Hop Tran Cao
Surgical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX

Introduction: Low socioeconomic status (SES) is associated with lower rates of treatment and shorter overall survival (OS) in hepatocellular carcinoma (HCC), as in most cancers. We sought to estimate the extent to which SES-based disparities in overall survival (OS) are mediated by surgical treatment disparities in patients with stage I-II HCC.

Methods: We used the National Cancer Database to analyze patients ages 18-75 who were diagnosed with AJCC 8th edition stage I-II HCC in 2008-2016. Exclusions included missing SES, surgery contraindications/refusal, not receiving treatment at the reporting facility, and death within 1 month of diagnosis. We used inverse odds weighting mediation analysis to estimate the proportion of SES-based OS disparities that were mediated by disparities in use of treatment (i.e. the proportion mediated [PM]) after adjusting for baseline covariates (sex, age, race/ethnicity, stage, cancer history, Charlson-Deyo score, and year). SES was defined as low (quartile 1 zip code-level income and education levels), high (quartile 4 income and education) and middle (all other patients). Surgical treatments and facility factors were the mediators. Chemotherapy and radiation use were not analyzed as mediators because they were associated with worse OS, likely because they were given to higher risk patients. Confidence intervals were obtained using bootstrapping (500 iterations). In a secondary analysis we additionally adjusted for MELD score (45.3% missing) and AFP elevation (19.1% missing) using multiple imputation by chained equations (50 imputed data sets).

Results: Among 46,003 patients, SES was low in 21.3%, middle in 63.1%, and high in 15.6%. Use of each treatment modality increased in a graded fashion with increasing SES except ablation, which had similar rates for middle and high SES patients (Table Part A). Five-year OS by surgical modality is shown in Figure A. Five-year OS was 36.8% for low SES patients, 40.1% for middle SES, and 45.4% for high SES (Figure B). The most important mediator was use of surgical treatments, which mediated 51.5% of OS disparities for low SES patients and 42.1% for middle SES and was driven primarily by disparities in use of transplant (Table Part B). Facility-level mediators significantly mediated OS disparities, but the composite PMs were only slightly larger than the PMs for use of treatment. Results were similar in the secondary analyses in which missing values of MELD score and AFP were imputed (Table Part C).

Conclusions: These analyses suggest that approximately half of SES-based OS disparities are mediated by differential use of surgical treatments, and use of transplant is the single most important mediator. Eliminating treatment disparities would be expected to significantly narrow survival disparities.

Table. Use of Surgical Treatment by Socioeconomic Status and Results of Inverse Odds Weighted Causal Mediation Analyses

Figure. Overall survival by use of treatment (A) and socioeconomic status (B)

Back to 2021 Abstracts