Abstract
The study was conducted to estimate the effects of age, year of diagnosis (period) and year of birth (cohort) on the time trend of female lung cancer incidence between 1976 and 2000 in Hong Kong. Data were obtained from a population‐based cancer registry and the Census and Statistics Department. Using the Hong Kong general population of 1990 as the reference, age‐adjusted incidence rates were computed by using direct standardization. We also analyzed secular trends in 2 consecutive periods, 1976–1990 and 1991–2000, using a Poisson regression model to estimate the annual percentage change in incidence rate. Age, period and cohort effects were assessed employing the method proposed by Clayton and Schifflers with 3 Poisson regression models fitted separately for age alone, age and period, and age and cohort. The age‐adjusted incidence rate increased steadily up to 1990, when a peak of 40.1 per 100,000 was reached. Thereafter, a downward trend was observed. The overall annual increase was 4.2% during the period 1976–1990 with the fastest increase observed in women aged 80–84 years. An overall decrease of 2.3% was seen in the period 1991–2000, with the greatest reduction occurring in the 35–39 age group. After the adjustment for age, significant period and cohort effects on the time trend of incidence were observed. The age‐cohort model provided a better description of the data than the age‐period model. The risk of successive 5‐year birth cohorts increased from 1896 to 1925 and then decreased. The time trend of lung cancer incidence of Hong Kong female in 1976–2000 had a marked birth cohort effect. Our results imply that major sources of exposure, which carry an increased risk of lung cancer for Hong Kong women, have been becoming less common in recent decades. Domestic air pollution, poor nutrition and tobacco smoking might have played important roles in the change of risks in successive generations and provide guidance for future interventions. © 2004 Wiley‐Liss, Inc.