Do factors affecting time from symptom onset to diagnosis or treatment of ovarian cancer also affect survival?

Abstract

<h3>Objectives:</h3> Little is known about whether factors affecting time from symptom onset to diagnosis or treatment of ovarian cancer also affect overall survival (OS) or cancer-specific survival (CSS). In this study, we aim to identify variables associated with improved time to diagnosis, treatment, and survival of ovarian cancer. <h3>Methods:</h3> A population-based study of the Surveillance, Epidemiology, and End Results (SEER)-Medicare database was conducted on patients ≥66 years old from 1992-2015 with a diagnosis of stage II-IV primary epithelial ovarian cancer. Regression models evaluated time from first coded symptom in a claim to diagnosis and treatment, OS, and CSS by age, race, comorbidity score, marital status, region, poverty level, year of diagnosis, urban/rural location, number of physician visit types in the 30 days prior to diagnosis, and emergency room presentation. <h3>Results:</h3> We identified 18693 women for analysis. Faster time to diagnosis was associated with BlackBlack race (p=.001) and both faster time to diagnosis and treatment were associated with unmarried status (diagnosis p<.001, treatment p<.001) and mostly rural (p=.011, p=.003) or all rural (p<.001, p=.003) location. Conversely, both slower time to diagnosis and treatment were associated with older age (p<.001, p<.001), higher comorbidity score (p<.001, p<.001), non-Northeast region (p<.001, p=.001), >4 physician visit types (p<.001, p<.001), emergency department presentation (p<.001, p<.001), and diagnosis after 1992-1999 (p<.001, p<.001). Five-year OS and CSS were 22.4% and 39.5%, respectively. Despite being associated with faster time to diagnosis and/or treatment, there were lower 5-year OS and CSS associated with unmarried status (OS p<.001, CSS p<.001), all rural location (p<.001, p<.001), and BlackBlack race (p<.001, p=.025). Lower 5-year OS and CSS were also associated with older age (p<.001, p<.001), higher comorbidity score (p<.001, p<.001), emergency presentation (p<.001, p<.001), and higher poverty level (p=.018, p=.048). Despite being associated with slower time to diagnosis and treatment, higher 5-year CSS was associated with >4 physician visit types (p=.002) and both higher 5-year OS and CSS were associated with diagnosis after 1992-1999 (p<.001, p<.001). Compared to 1992-1999, the hazard ratios of 5-year cancer-specific death after 5 years decreased for each subsequent time period (2000-2005 HR=.90; 2006-2010 HR=.82; 2011-2015 HR=.72). <h3>Conclusions:</h3> In our SEER-based database study, variables associated with faster time to diagnosis and treatment of ovarian cancer often correlated with worse OS and CSS. We hypothesize that the discordance between time from diagnosis to treatment and survival seen based on clinical and demographic variables may be due to unmeasured social determinants of health or deviations from guideline-concordant care. Future research to understand this discordance is warranted.