The impact of socioeconomic status on cancer care and survival

Reporter: Gita Suneja, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: June 7, 2011

Presenter: Sandra L. Wong
Presenter's Affiliation: University of Michigan


  • Race, less education, low income, unemployment, and lack of health insurance have all been found to be predictors for receipt of suboptimal health care.
  • A growing body of literature has confirmed that disparities exist with regard to access to cancer screening, as well as treatment services for patients with cancer, based on these factors.
  • Low socioeconomic status (SES) is a risk factor that accounts for combined effects of education, income, and employment status variables.
  • Recent data has shown that low SES predicts for poor outcomes in cancer patients. Most SES research has focused on the effect of low SES on health behaviors, access to care, and mortality, but fewer studies have focused on the overall quality of care delivered.
  • The goal of this study was to determine the impact of SES on the quality of cancer care delivered, as well as survival, for patients with lung, esophagus, and pancreas primary cancers.
    • These cancers account for over 35% of all cancer deaths in the US annually and a large proportion of cancer costs.
    • Despite minimal improvement in survival rates over time, the aggressiveness with which patients are treated and followed has increased dramatically.
  • The hypothesis of the authors was that low SES predicts for lower aggressiveness of cancer care in poor prognosis cancers.


  • Data was abstracted from the SEER-Medicare database, a population-based cancer registry linked with national Medicare claims. The database is limited to fee-for-service claims in patients ³ 65 years of age.
  • All incident cases from 1992-2002 were included. Follow-up data was available through 2005.
  • The cohort included 68,167 lung cancer patients, 4,350 esophageal cancer patients, and 12,779 pancreatic cancer patients.
  • A summary measure of SES was constructed based on U.S. ZIP code data from the 2000 census. This included wealth/income level, education, and employment status (Diax-Roux, NEJM 2001).
    • Each beneficiary was assigned an SES summary score based on his/her zip code of residence.
    • Patients were grouped into 5 quintiles of SES with SES 1 being the lowest SES and SES 5 being the highest.
  • Patients were then assigned to a hospital based on where the majority of inpatient admissions or hospital-based services were performed. If this data was not available, patients were assigned to the hospital where their primary care physician was based.
  • The endpoints were treatment and 2-year survival rates.


  • Similar proportions of distant disease at presentation were seen across the SES quintiles, 35% for the lung cancer cohort, 24-25% for the esophageal cancer, cohort and 45-47% for the pancreatic cancer cohort. No disparities in cancer stage were identified.
  • Low SES patients were more likely to be black, require urgent/emergent hospital admission, and to be treated at a low-volume, non-teaching hospital. This trend was observed in lung, esophageal, and pancreatic cancer cohorts.
  • High proportions of patients in all SES quintiles received imaging for cancer staging. The majority of the patients had CT scans for staging, and they were received by >90% in all SES quintiles.
  • Specialized testing, such as PET scans, were more likely to be performed in higher SES patients in the lung cancer cohort (10% high SES vs. 5% low SES). This trend was not observed in the esophageal and pancreatic cancer patients.
  • For all three cancer types, higher utilization of surgery, chemotherapy, and radiation therapy was seen in high SES patients. In other words, more cancer-directed services were utilized in the higher SES groups.
  • Low SES patients were more likely to receive no cancer-directed treatment at all. No cancer-directed therapy was delivered to 45% of the low SES lung, 35% of low SES esophageal, and 60% of lowest SES pancreatic cancer patients.
    • High percentages of high SES patient were also found to receive no cancer-directed treatment -- 35% for lung, 25% for esophageal, and 50% for pancreatic cancer.
  • The crude hazard ratio for mortality demonstrated improved survival in higher SES patients. However, on adjusted analysis, taking into account other patient characteristics and treatment, a significant difference in survival by SES quintile was seen only in patients with lung cancer. This difference in survival in lung cancer was marginally significant (HR 1.11, 95% CI 1.08-1.15).
  • The authors also sorted hospitals into low and high SES groups based on the average SES of the patient population.
    • The highest SES hospitals treated a greater proportion of younger, healthier patients (assessed using the comorbidity index). They treated less black patients and had less urgent/emergent admissions. High SES hospitals tended to be teaching hospitals with high patient volume. These trends were observed in lung, esophageal, and pancreatic cohorts.
  • Interestingly, low SES patients treated in a high SES hospital had outcomes similar to high SES patients. Similarly, high SES patients treated in a low SES hospital had worse outcomes. In other words, the effect of the hospital attenuated the individual SES outcomes.

Authors’ Conclusions

  • Variation in cancer treatment occurs across SES groups, with the lowest SES patients more likely to receive no cancer treatment.
  • Despite this difference in treatment, higher SES patients do not have significantly improved survival rates for poor prognosis cancers.
  • Reducing variation in treatment delivery may improve healthcare efficiency without changing patient outcomes.


  • This study suggests that patients with low SES are at risk for receiving an inferior quality of cancer care. The implications are large, particularly given the current economic recession and the increasing number of patients who fall into a low SES category.
  • Interestingly, this study found that distribution of stage was equivalent across all SES groups; this contrasts with other studies suggesting that patients with lower SES are more likely to present at a later stage. This finding may be due to the lack of screening studies for lung, esophageal, and pancreatic cancer, leading to less disparity in staging at the time of presentation with clinical symptoms.
  • There are several notable limitations to this study:
    • SEER-Medicare represents a retrospective, population-based cohort and therefore direct correlation cannot be assessed.
    • The study cohort consists of Medicare beneficiaries over the age of 65 years, and the results may not be generalizable to younger patients. However, over 2/3 of patients with these three cancer diagnoses are Medicare beneficiaries.
    • SES definitions are based on area-level data instead of patient-level data, which may lead to misclassification of patient SES (i.e. a high SES patient living in a low SES zip code).
    • Patient and physician preferences are not captured in the SEER-Medicare database.
  • Additionally, the study captures treatments delivered in a hospital-based setting, whereas the majority of cancer treatments are now delivered in the outpatient setting. This may have greatly impacted the results and account for the surprisingly large number of low and high SES patients recorded as having received no treatment.
  • Nonetheless, these compelling results generate numerous hypotheses on how SES may affect cancer care and survival, and merit further investigation.
  • Finally, recognition that overall outcomes did not appear to differ based on the level of cancer-directed care delivered raises the question of whether some patients, particularly high SES patients, with aggressive cancer diagnoses, may be receiving over-treatment.