Place of death of Houston area residents with cancer over a two-year period

J Pain Symptom Manage. 2003 Jul;26(1):637-43. doi: 10.1016/s0885-3924(03)00204-5.


The majority of cancer patients wish to die at home. Improved understanding of place of death and its relevant demographic predictors is important for the planning of palliative cancer care programs. The purpose of this study was to determine the place and predictors of site of death in cancer patients in a major U.S. metropolitan area. Death certificate data over two years were analyzed for Houston area residents with cancer who died in the Houston area. Information was obtained on factors that might be associated with the place where cancer patients die. For the purpose of this study, we looked at the following variables: primary site of cancer (hematological, breast, genitourinary, gastrointestinal, lung, and other); black, white, Hispanic, or Asian; age at death; marital status; sex; whether or not veteran of U.S. armed forces; levels of education; and area of residency within the Houston area. Univariate and multivariate analyses were performed. The majority of patients died in the hospital (51-52% both years), with the next most frequently occurring group dying at home (34-35% both years). Stepwise multivariate analysis resulted in a 6-variable logistic regression model. In this model, the odds of dying in hospital were increased by a factor of 2.7 if the patient had a hematological cancer (P<0.0001), a factor of 1.6 if the patient lived in Harris County (P<0.0001), and a factor of 1.5 if the patient was black (P<0.0001). Further characterization of factors associated with increased risk of hospital death rate is needed and systems should be developed to enable the majority of cancer patients to access palliative care services in the multiple settings in which they die.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Death Certificates
  • Female
  • Home Care Services / statistics & numerical data*
  • Hospitalization / statistics & numerical data*
  • Humans
  • Logistic Models
  • Male
  • Neoplasms / mortality*
  • Palliative Care / statistics & numerical data*
  • Texas / epidemiology
  • Urban Population / statistics & numerical data