Objective: To predict the trend of hepatocellular carcinoma (HCC) mortality and investigate the features of its mortality including age, period, and birth cohort in males living in Haimen city of Jiangsu province, China.
Methods: Grey model (GM) was modeled using standardized mortality rate (SMR) of HCC from 1993 to 2006, and was applied to predicting SMR until 2012. Based on the mortality density (MD) for a four-year period, the goodness-of-fit of models and comparisons between models were evaluated so as to obtain the best one among these models including the effects of intercept, age-period-cohort (APC), age-period (AP), age-cohort (AC), period-cohort(PC), and APC. Both APC full model and the best model were used to estimate effects of age, period, and cohort on HCC mortality. In addition, MD form 2005 to 2012 was predicted by the best model.
Results: Predictions based on GM (1,1) showed that SMR was 48.578 pre 100 000 population (relative error=-1.267%) in 2007 year, which declined between 2008 and 2012. The lowest value was 45.578 pre 100 000 people (in the 2012 year). The results of fitted models and comparisons between models showed that AP model was the best one (ΔG(2) = 9.065, AIC = 202.544). The curvatures of the effects of the three factors from APC model suggested that significances existed in changes of curvatures of 36.5 - 40.5 years old-(-0.368) and 64.5 - 68.5 years old-(-0.489) as well as in the change of 1956 - 1959 birth cohort (C(2)(1949.5, 1967.5) = -0.492). The estimation of relative risks for AP model showed that the age effects were upward to 64.5 - 68.5 years old-, then downward; and that the period effects were found to be declined between 1993 and 2004. Predictions based on AP model suggested the decrease of HCC mortality.
Conclusion: The slightly decreasing trend of HCC mortality for males might be explained by age, period and a minor birth cohort effects in Haimen of China.