The scope of the proposed study is to develop and test new statistical procedures for age-period-cohort (APC) analysis of cancer incidence and mortality rates. Guided by an expanding literature on statistical methods and experience in statistical modeling and analyses for APC studies, this research plan addresses the development of models that aim to resolve the identifiability problem. A new statistical methodology is developed through eigen-analysis, principal component analysis, shrinkage models and reduce models with bias correction methods. These models address the problem of estimation and bias in log-linear regression models with exact collinearity between the effects of age, period and cohort in APC analysis. A unique estimator is identified, which generates accurate trend estimation with minimal bias or no bias. These new statistical models provide estimates of the age, period and cohort effects of cancer incidence and mortality, which exploit fully cancer summary rate data arising from the Surveillance, Epidemiology, and End Results (SEER) and other cancer registries. These APC models also permit estimation of future cancer rates. The methodology allows for trend estimation, construction of confidence intervals, bias correction and forecasting. The performance of the proposed methods will be evaluated with simulated and real cancer incidence and mortality rate data. The methodology will be applied to several studies on cancer incidence and mortality rates, including one study on mortality rate of cervical cancer in Ontario, six studies on incidence rates of breast cancer, prostate cancer, male and female colon cancer, male and female lung cancer in Connecticut. The methodology will also be applied to studies on national mortality rates of various types of cancer in different ethnic groups, such as White, Black, Asian/Pacific islanders and Hispanics, to investigate the differences between these groups as outlined in the Healthy People 2000. The ultimate goal of the investigation is to assess trend estimation and forecasting in APC analyses. The methods proposed in this investigation have immediate application to these studies, and offer an array of promising techniques for use in APC analysis of cancer incidence and mortality.
| Fu, Wenjiang J (2015) Racial-Sex Disparities--A Challenging Battle Against Cancer Mortality in the USA. J Racial Ethn Health Disparities 2:158-66 |
