A comparison between Cox proportional hazard models and logistic regression on prognostic factors in gastric cancer

East Afr J Public Health. 2009 Apr;6 Suppl(1):20-2. doi: 10.4314/eajph.v6i3.45766.

Abstract

Background: Despite its decreasing prevalence in industrialized nations, gastric cancer remains one of the most frequent cancers in the world. Since stomach cancer often was not detected until an advanced state, survival rate was rather low.

Aim: The aim of the present study is to compare the Cox proportional hazard model and Logistic regression model to estimate prognostic of patients with gastric cancer.

Material and methods: To determine the independent prognostic factors reducing survival time for gastric cancer, we compared the parametric methods (logistic regression) and non-parametric methods (Cox proportional hazard models) applied to patients who registered in one cancer registry center located in southern Iran.

Results: Of 442, 266 (60.2 %) died. In multivariate analyses using the Cox proportional hazard model, Age at diagnosis (P = 0.018, Hazard rate = 1.84), grade of tumor (P = 0.018, Hazard rate = 1.56), and metastasis (P = 0.004, Hazard rate = 1.53) were the most independent prognostic factors. As well as, using the stepwise logistic regression model, Age at diagnosis, (P = 0.005, Odds Ratio = 1.01), grade of tumor (P = 0.025, OR = 1.95), and metastasis (P < 0.001, OR = 2.81) were also the most independent factors who affected on survival.

Conclusion: Although regression coefficients are not all the same, these three factors are the most prognostic factors that affect on survival of gastric patients in both multivariate analyses.

Publication types

  • Comparative Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Age Factors
  • Female
  • Humans
  • Iran / epidemiology
  • Logistic Models*
  • Male
  • Middle Aged
  • Prognosis*
  • Proportional Hazards Models*
  • Registries
  • Risk Factors
  • Stomach Neoplasms / diagnosis
  • Stomach Neoplasms / mortality*
  • Survival Analysis
  • Time Factors