A Bayesian model for age, period, and cohort effects on mortality trends for lung cancer, in association with gender-specific incidence and case-fatality rates

J Thorac Oncol. 2009 Feb;4(2):167-71. doi: 10.1097/JTO.0b013e318194fabc.


Introduction: To study time trends in lung cancer mortality by separating the incidence and case-fatality rates, in association with age, period, and cohort effects.

Methods: Lung cancer cases (n = 44,139) diagnosed between 1996 and 2002 in Taiwan were analyzed by decomposing the time trend in mortality into incidence and case-fatality rates. Descriptive data, together with periodical treatment distribution (surgery, chemotherapy, and others) were analyzed using a Bayesian age, period, and cohort (BAPC) model.

Results: Midterm mortality (2-year age-adjusted standardized mortality rate) has been decreasing for male lung cancer patients since about 2000, mainly because of a decrease in incidence during this period. For women, 2-year age-adjusted standardized mortality rate has been slightly increasing, mainly as a result of increasing incidence. There were small improvements (3-6%) in the short-term (1-year) case-fatality rate, possibly owing to increased utilization (approximately 15-18%) of chemotherapy. The midterm (2-year) case-fatality rate remained roughly the same, especially for men.

Conclusions: Using a new BAPC model, we found that the trends in mortality for lung cancer paralleled the changes in incidence, with opposite effects in men and women. Increased utilization of chemotherapy might have partly accounted for the small improvement in the case-fatality rate. The contributions of other unmeasured factors such as staging and histologic distribution remain to be clarified in future studies.

Publication types

  • Comparative Study

MeSH terms

  • Age Factors
  • Aged
  • Bayes Theorem
  • Cohort Effect
  • Cohort Studies
  • Female
  • Humans
  • Incidence
  • Lung Neoplasms / mortality*
  • Lung Neoplasms / therapy
  • Male
  • Models, Statistical*
  • Mortality / trends*
  • Sex Factors
  • Taiwan
  • Time Factors