Aims: The main purpose of this paper was to assess the effect of age, period, and cohort on stroke mortality rates among a Lithuanian urban population aged 25-64 years (1041 men and 724 women) between 1980 and 2004.
Methods: Routine stroke mortality data were obtained from official Kaunas region mortality register by codes 430-438 and I60-I69 in the 9th and in the 10th revisions of the International Classifications of Diseases (ICD), respectively. Mortality rates per 100,000 persons for men and women were age-adjusted using the age distribution of the European Standard Population. Age-specific mortality rates were analysed by sex, period, and birth cohort in eight 5-year age groups and five 5-year age groups. Goodness of fit of the Poisson regression models were evaluated using Pearson and Freeman-Tukey residuals. The age-period and age-period-cohort models provided a significantly better fit than a model with the factors "age'' and "cohort''.
Results: During the study period, mortality rates decreased from 46.8 to 33.0 per 100,000 for men, and from 20.2 to 18.1 per 100,000 for women (average annual decrease of -1.3%, p<0.1 for men, and -1.6%, p<0.03 for women). An age effect was present in both sexes. The definite upward period effect was observed from 1990 to 1994 both among men and women, and was followed by a sharp fall during 2000-4. Cohort and period effects have contained relevant information which partially explained trends in stroke mortality among a 25-64 year-old Lithuanian urban population.
Conclusions: During the period of 1980-2004, the mortality trend declined among women only. The period effect contains relevant information for the explanation of increasing mortality rates during 2000-4 among men and women. The Poisson regression models could be applied for the examination and explanation of the different causes of the population mortality.