Using structural equation modeling to examine socioeconomic variations with clinical features, treatment received and mortality in patients with colorectal cancer
Introduction: Structural Equation Modeling (SEM) is a statistical methodology for representing, assessing, and testing theoretical network with relations between variables. Based on the fact that there has been a socioeconomic gradient in mortality among colorectal cancer patients, we examine pathways linking SES to treatment of choice, clinical features and mortality among cancer patients and explain the direct and indirect effects on each variable. Objective: The aim of this study is to develop and test a conceptual model of the mechanisms between SES, treatment received, clinical features and mortality among colorectal cancer patients in the Surveillance, Epidemiology and End Results (SEER)-Medicare database with the use of Structural Equation Modeling. Methods: Data was extracted from SEER-Medicare database. The conceptual model was evaluated by latent variables modeling approach and exploratory factor analysis was performed to identify latent variables. The hypothesized exogenous variables involved tumor stage, tumor size, comorbidity score, income, non-high school graduates, poverty, radiation, chemotherapy surgery, and mortality. The proposed models were tested for acceptable model fit by confirmatory factor analysis, and structural equation modeling analysis was performed to determine if all parameters are significantly different from zero. All of the statistical analyses were carried out by SAS 9.1 software. Results: The t-statistics of all latent variables are significant at p-value of 0.05. All of the three latent variables appeared to be significant predictors of the outcome measure (i.e., mortality). The results confirmed our research hypothesis that mortality is significantly correlated with treatment, clinial features and SES, where the lower the socioeconomic status, the higher the mortality of colorectal cancer patients. Every unit increase of SES accounts for 0.069% decrease in mortality; treatment received (either radiation, chemothrapy, or surgery ) accounts for 0.77% decrease in mortality; and advanced stage of cancer and higher comorbidity score accounts for 0.62% increase in the mortality of colorectal cancer patients. Conclusion: Our study demonstrated socioeconomic status, treatment received and clinical features are the key factors associated with mortality among cancer patients. The results from the analysis of structural equation modeling indicated that all three latent variables in this study are significant predictors of mortality among patients with colorectal cancer.
Ling, Kok Yan, "Using structural equation modeling to examine socioeconomic variations with clinical features, treatment received and mortality in patients with colorectal cancer" (2013). Texas Medical Center Dissertations (via ProQuest). AAI1552535.