Supplementary Figure S1 presents a series of plots assessing the proportionality assumption in a Cox proportional hazards model by examining the scaled Schoenfeld residuals plotted against time. Each plot corresponds to a different covariate included in the model. The residuals are displayed to investigate any systematic patterns that might suggest violations of the proportional hazards assumption. The lack of distinct trends or patterns across the series of plots supports the assumption's validity for the included covariates.
Funding
National Cancer Institute (NCI)
United States Department of Health and Human Services
Find out more...Fogarty International Center (FIC)
National Center for Advancing Translational Sciences (NCATS)
United States Department of Health and Human Services
Find out more...Vanderbilt University (VU)
ARTICLE ABSTRACT
Two-thirds of global cancers occur in low/middle income countries (LMIC). Northern Central America is the largest LMIC region in the Western Hemisphere and lack cancer registries to guide cancer control. We conducted a gastric cancer survival study in rural Western Honduras, characterized as having among the highest gastric cancer incidence rates in Latin America.
The cohort of incident gastric cancer diagnosed between 2002 and 2015 was studied with active follow-up with household visits. The regional gastric cancer registry was primary for case identification, with completeness examination with hospital data and national death certificates. Cox regression models were used for survival calculations.
Survival follow-up was achieved in 741/774 patients (95.7%). Household interviews were conducted in 74.1% (n = 549); 65.7% were male, median age at diagnosis was 64 years, 24.5% were <55 years; 43.9% of tumors had pyloric obstruction; 45.2%, 43.2%, and 7.3% of histology was intestinal, diffuse, and mixed, respectively. A total of 24.7% patients received treatment. The 5-year survival rates were 9.9% for both males and females, 7.7% for age <45, and 7.9% for diffuse gastric cancer. Median survival time was 4.8 months [95% confidence interval (CI), 4.2–5.6]. In the final Cox regression model including age, sex, Lauren subtype, and poverty index, only treatment was significantly associated with survival (hazard ratio = 2.43, 95% CI, 1.8–3.2).
Markedly low gastric cancer 5-year survival rates are observed in rural Central America. The majority of patients present with advanced disease and a minority have access to therapy.
The findings have implications for cancer control in the Central America LMICs and for US Latino populations.See related commentary by Riquelme and Abnet, p. 1550
