Socioeconomic disparities in cancer survival: Relation to stage at diagnosis, treatment, and centralization of patients to accredited hospitals, 2005–2014, Japan

Abstract

Background: Cancer survival varies by socioeconomic status in Japan. We examined the extent to which survival disparities are explained by factors relevant to cancer control measures (promoting early-stage detection, standardizing treatment, and centralizing patients to government-accredited cancer hospitals [ACHs]). Methods: From the Osaka Cancer Registry, patients diagnosed with solid malignant tumors during 2005–2014 and aged 15–84 years (N = 376,077) were classified into quartiles using the Area Deprivation Index (ADI). Trends in inequalities were assessed for potentially associated factors: early-stage detection, treatment modality, and utilization of ACH (for first contact/diagnosis/treatment). 3-year all-cause survival was computed by the ADI quartile. Multivariable Cox regression models were used to assess survival disparities and their trends through a series of adjustment for the potentially associated factors. Results: During 2005–2014, the most deprived ADI quartile had lower rates than the least deprived quartile for early-stage detection (42.6% vs. 48.7%); receipt of surgery (58.1% vs. 64.1%); and utilization of ACH (83.5% vs. 88.4%). While rate differences decreased for receipt of surgery and utilization of ACH (Annual Percent Change = −3.2 and − 11.9, respectively) over time, it remained unchanged for early-stage detection. During 2012–2014, the most deprived ADI quartile had lower 3-year survival than the least deprived (59.0% vs. 69.4%) and higher mortality (Hazard Ratio [HR] = 1.32, adjusted for case-mix): this attenuated with additional adjustment for stage at diagnosis (HR = 1.23); treatment modality (HR = 1.20); and utilization of ACH (HR = 1.19). Conclusions: Despite improvements in equalizing access to quality cancer care during 2005–2014, survival disparities remained. Interventions to reduce inequalities in early-stage detection could ameliorate such gaps.