Associations between socioeconomic status and cancer survival: choice of SES indicator may affect results

Ann Epidemiol. 2006 Jul;16(7):521-8. doi: 10.1016/j.annepidem.2005.10.002. Epub 2006 Jan 4.


Purpose: Two previous studies, by Gorey et al. and Boyd et al., compared associations between socioeconomic status (SES) and cancer survival in Canada and the United States. Both studies used SES information from population censuses linked to cancer registries. This study investigates why two similar studies led to apparently conflicting results.

Methods: We conducted analyses following analytic details provided by the previously published studies to describe cancer survival in Toronto, Canada, and Detroit, MI. We examined the effects of choice of census indicators and census levels on the observed SES-related gradients in cancer survival.

Results: Significant associations between SES and cancer survival were observed in Toronto for several major disease sites when median household income was used as an SES indicator. Associations were weaker when a poverty indicator was used. In Detroit, similar SES gradients were observed by using both income and poverty as SES indicators. When SES quintiles were represented by income ranks, SES-associated survival gradients were much steeper in Detroit than Toronto. When SES was described by the median income in each quintile, gradients were similar in the two cities.

Conclusions: The apparent contradiction in results of two previous studies is related to the choice of SES indicators. Poverty may not be an indicator of choice for such an intercountry comparison.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Data Interpretation, Statistical
  • Humans
  • Income*
  • Neoplasms / economics
  • Neoplasms / epidemiology*
  • Neoplasms / mortality*
  • Ontario / epidemiology
  • Research Design*
  • Risk
  • Socioeconomic Factors
  • Survival Analysis
  • United States / epidemiology