Comorbidity measures to predict clinical and economic outcomes among elderly gynecologic cancer survivors
MetadataShow full item record
The incidence of gynecologic cancers increases with age, and elderly cancer survivors are more likely to have additional comorbid conditions. However, little is known about the relationships between comorbidity and health outcomes among elderly gynecologic cancer survivors. The primary purpose of this study is to examine the relationships between comorbidity and health outcomes, and the secondary purpose is to compare the performance of commonly used comorbidity indices to predict health outcomes among elderly gynecologic cancer survivors. This retrospective data analysis study used the 2007-2010 SEER-Medicare data. The study population was elderly gynecologic cancer survivors in the US. The primary independent variable was each comorbidity index: diagnosis-based indices (Charlson Comorbidity Index (CCI), Elixhauser Index (EI), National Cancer Institute comorbidity (NCI) index) and medication-based indices (the Chronic Disease Score (CDS) and RxRisk). The dependent variables were: survival (overall survival time and one-year mortality), the numbers of healthcare utilization events (emergency room (ER)/inpatient visits, outpatient visits, and office-based practitioner visits), and healthcare costs (ER/inpatient visit, outpatient visit, office-based practitioner visit, pharmacy, and total healthcare). Cox models with time-dependent covariates, Poisson regressions, negative binomial regressions, and gamma regressions with a log link were used. A total of 4,063 survivors were included. Among them, 27.59% died within one year after diagnosis, and the mean (SD) of total healthcare costs was $40,605 ($34,014). The diagnosis-based indices were associated with a shorter overall survival time and an increased mortality and outperformed the medication-based indices in predicting them. Regarding healthcare utilization and costs, the CCI and CDS-1 scores were better predictors for ER/inpatient visit-related outcomes and total healthcare cost, while the CDS-2 and RxRisk scores were better predictors for office-based practitioner visit-related outcomes. None of the comorbidity indices were significant predictors for outpatient visit-related outcomes and prescription costs. Since the ability of the comorbidity indices varied depending on the outcome of interest, the outcome along with the purpose of the study should be considered in selecting an appropriate comorbidity index. This study provides evidence that clinicians can use in developing better treatment plans for specific conditions, that researchers can use in choosing the best comorbidity index, and that payers can use in their budgeting by identifying comorbid conditions with higher costs.