Racial and socioeconomic disparities in survival improvement of eight cancers

Cancer is a leading cause of death globally, with survival varying substantially by cancer site (e.g., better in breast and prostate, worse in lung and pancreatic). Over the past two decades, treatments have advanced and improved survival, but improvements vary by site. Within a cancer type, survival also varies by race, age, sex, and socioeconomic status (SES), contributing to disparities. Prior studies show worse survival for black patients versus white in breast and prostate cancers, female advantages in lung cancer, and better outcomes for younger age at diagnosis and higher SES. While overall disparities have been well studied, disparities in survival improvements over time are less explored. A prior SEER-based study (1990–2009) found significant survival disparities by age and narrowed racial disparities (except ovarian cancer). Given substantial survival gains in the past decade due to new therapies (including immunotherapies) and prior work in prostate cancer, this study evaluates temporal improvements in cancer-specific survival (CSS) across race, age, SES, and sex in eight common cancers (breast, ovary, prostate, colon, rectum, liver, lung, pancreas) using SEER data from 2004–2018.

The paper situates its work within evidence that: (1) black patients have worse survival than white patients in breast and prostate cancers; (2) females often have better outcomes than males in lung cancer; (3) younger age at diagnosis and higher socioeconomic status are associated with better outcomes across many cancers. A prior SEER analysis (1990–2009) reported disparities in survival improvement by age and narrowed racial disparities except in ovarian cancer. Additional literature cited documents treatment differences by race (e.g., lower likelihood of recommended surgery among African Americans with early-stage NSCLC; more aggressive colorectal cancer treatment among white patients) and global trends in cancer incidence, mortality, and survival as well as the emergence of immunotherapy contributing to recent survival gains.

Data source: SEER 18 registries Incidence-Based Mortality for patients diagnosed 2004–2018 across eight common cancer sites (per Supplementary Table 1; breast, ovary, prostate, colon, rectum, liver, lung, pancreas referenced throughout the text). Total N=1,875,281.

Inclusion criteria: (1) Type of Reporting Source = Hospital inpatient/outpatient or clinic; (2) Only one primary cancer (Sequence number: One primary only); (3) Age at diagnosis 40–85 years.

Outcomes and variables: Overall and cancer-specific survival (CSS) from SEER variables Vital status recode, SEER cause-specific death classification, and Survival months. Race/ethnicity from Race and origin recode (NHW, NHB, NHAPI, Hispanic), categorized as White (Non-Hispanic White), Black (Non-Hispanic Black), Asian (Non-Hispanic Asian or Pacific Islander), and Hispanic (all races). Staging: For all except breast, stages from Derived AJCC Stage Group 6th ed (2004–2015), Derived SEER Combined Stage Group (2016–2017), and Derived EOD 2018 Stage Group (2018+). For breast: Breast - Adjusted AJCC 6th Stage (1988–2015), SEER Combined Stage Group (2016–2017), EOD 2018 Stage Group (2018+). Age from Age recode with single ages and 85+. Sex from Sex. Socioeconomic status proxied by area-level Median household income inflation adjusted to 2019.

Subgroup definitions: Age groups: 40–55 (Younger), 56–70 (Middle), 71–85 (Older). Income groups: < $60,000 (Low), $60,000–$74,999 (Intermediate), > $75,000 (High). Sex groups as recorded. White race, younger age, low income, and female sex used as reference categories when examining subgroup effects (except when stratifying by sex-specific cancers where applicable).

Temporal bins for improvement analysis: Year of diagnosis grouped into three 5-year periods: 2004–2008 (reference), 2009–2013, 2014–2018.

Statistical analysis: Cancer-specific survival analyzed using R package survival. Multivariable Cox proportional hazards regression used to estimate disparities and temporal improvements, adjusting for race, age group, income group, sex (where applicable), and stage. Hazard ratios (HRs) with 95% confidence intervals (CIs) reported. Significance threshold P < 0.05. Additional descriptive analysis compared shifts in stage distribution across time periods by race (Table 1).

• Sample and scope: Analysis of 1,875,281 patients with eight common cancers (breast, ovary, prostate, colon, rectum, liver, lung, pancreas) in SEER (2004–2018).

• Baseline CSS disparities (multivariable Cox):

  • Race: For all cancer sites except lung, black patients had significantly lower CSS than white patients; disparities were largest in breast cancer. Asian patients had significantly longer CSS across all cancer sites; notably in lung cancer adjusted HR=0.73 (95% CI 0.72–0.74). Hispanic patients had higher CSS in lung cancer (HR=0.92, 95% CI 0.90–0.93) but nonsignificant differences in other sites.
  • Age: Older age at diagnosis associated with worse CSS across all cancers. Example HRs (vs 40–55): breast 56–70 HR=1.19; 71–85 HR=1.86; prostate 56–70 HR=1.34; 71–85 HR=3.40.
  • Stage: Higher stage strongly associated with worse CSS in every cancer. Examples: breast stage IV HR≈73.81 vs stage I; colon stage IV HR≈37.82; lung stage IV HR≈8.05; pancreas stage IV HR≈4.31.
  • Income: Higher area-level income associated with better CSS across all cancers. Examples: breast >$75k HR=0.82 (0.80–0.84) vs <$60k; liver >$75k HR=0.81–0.84 range noted across sites (e.g., 0.81–0.91); pancreas >$75k HR=0.91 (0.89–0.92) vs <$60k (site-specific values provided in figures).
  • Sex: In cancers affecting both sexes, males had significantly higher hazard (worse CSS) than females.

• Temporal improvements in CSS (2009–2013 and 2014–2018 vs 2004–2008):

  • All races generally improved over time across most sites.
  • White patients: significant improvement in all cancers across both later periods.
  • Asian patients: significant improvements in most cancers; colon had less pronounced improvement.
  • Hispanic patients: significant improvements across cancers and periods except prostate.
  • Black patients: greatest improvement in CSS from 2014–2018 vs 2004–2008 for breast, ovary, colon, rectum, liver, lung, and pancreas. Nonetheless, persistent gaps remained: in breast, ovary, and prostate, black patients in 2014–2018 still had worse CSS than white patients in 2004–2008.
  • Illustrative HRs by race and time (vs white 2004–2008 referent):
    • Breast (black): 2004–2008 HR=1.55; 2009–2013 HR=1.35; 2014–2018 HR=1.26.
    • Prostate (black): 1.32; 1.28; 1.14 across the three periods.
    • Lung (black): 1.05; 0.89; 0.72 showing substantial improvement.
    • Colon (black): 1.24; 1.16; 1.01 approaching parity by 2014–2018.

• Age-related improvements: Significant CSS improvements for all age groups across all cancers. In colon, lung, and pancreas, the youngest (40–55) had the greatest improvement; in ovary and liver, older groups achieved the greatest improvement. For colon and rectum, the 56–70 group had the lowest improvement.

• Income-related improvements: In most sites (except breast and liver), high-income patients (> $75k) had the greatest survival improvement, especially 2014–2018. Example: prostate high-income HR=0.73 (0.68–0.78) vs 2004–2008, compared to low-income HR=0.83 (0.78–0.89) and intermediate-income HR=0.87 (0.82–0.93).

• Sex-related improvements: Females experienced greater improvements in colon, rectum, liver, and lung; males had greater improvement in pancreas.

• Stage migration (2009–2013 and 2014–2018 vs 2004–2008): Increased stage I and decreased stage IV diagnoses over time across many sites and races, particularly in liver and pancreas. Black patients showed increased stage I diagnoses across all eight cancers and decreased stage IV in breast, rectum, liver, and pancreas, potentially contributing to observed survival gains.

The study set out to quantify whether improvements in cancer-specific survival over time have been equitably shared across race, age, sex, and socioeconomic subgroups. Findings show broad CSS improvements from 2004 to 2018 across eight common cancers, but with persistent and, in some dimensions, widening disparities. Black patients—despite experiencing some of the largest gains in several cancers—still lag behind white patients, with black CSS in breast, ovary, and prostate (2014–2018) remaining worse than white CSS from a decade earlier (2004–2008). Asian patients consistently demonstrated survival advantages, and Hispanic patients showed advantages in lung cancer.

Age, stage, sex, and SES all contribute meaningfully to disparities. Older age and advanced stage strongly predict worse CSS, highlighting the central importance of early detection and optimized management for older adults. Higher income correlates with better CSS and greater improvement over time for most cancers, suggesting that access to screening, timely diagnosis, and advanced treatments may be unevenly distributed. Sex differences favor females in multiple cancers, consistent with prior literature.

Temporal analyses and stage distribution shifts suggest that earlier diagnosis—reflected by increases in stage I and decreases in stage IV—may partially explain improvements, particularly among black patients in several cancers. However, enduring gaps indicate that improvements in access, guideline-concordant care, and treatment equity remain necessary. The results underscore the need to identify and mitigate structural and clinical factors (e.g., screening access, treatment allocation, adherence to recommended therapies) that produce unequal outcomes across populations.

Using SEER data from 2004–2018, the study provides an atlas of temporal improvements in cancer-specific survival across eight major cancers and reveals persistent disparities by race, age, sex, and income. While survival improved for all groups, black patients continue to experience worse outcomes in several cancers despite notable gains, and high-income patients have benefited disproportionately from recent improvements. Future work should identify and address drivers of inequity—such as access to early detection, quality of care, and treatment uptake—and prioritize interventions for patient groups with persistently lower CSS. Additional research integrating molecular tumor characteristics, more granular individual-level socioeconomic metrics, and detailed treatment data could clarify mechanisms and guide targeted strategies to narrow disparities.

• Potential unmeasured confounders and unrecorded variables in SEER; variations in data coding and reporting; missing data.
• Possible patient migration between SEER registries and differential loss to follow-up; early censoring could bias survival estimates.
• SEER’s hospital-based data may not capture care in private clinics, outpatient radiation centers, nursing homes, or other outpatient settings, which are common for certain cancers (e.g., breast, prostate).
• Lack of molecular receptor and biomarker data across all cancers limits insights into biology-driven differences by race/ethnicity.
• Socioeconomic status measured via area-level income, which may misclassify individual SES, especially in large, heterogeneous counties.