Changing rates of adenocarcinoma and adenosquamous carcinoma of the cervix in England

Background A recent analysis showed little or no effect of screening on the incidence of adenocarcinoma of the cervix between 1971 and 1992. We have used additional data on cancers diagnosed in 1993-94 in England and up to 1997 in five English cancer registries to investigate more recent trends. Methods After imputing the number of adenocarcinomas in women with unknown histology, we fitted an age-cohort model to 8062 adenocarcinomas of the cervix diagnosed in England between 1971 and 1987. Predictions from this model were applied to the more recent data on 5854 cases. Residual effects were plotted against year of diagnosis in each of four age-groups. Findings We estimated the underlying risk of cervical adenocarcinoma to be 14 times (95% CI 11-19) greater in women born in the early 1960s than in cohorts born before 1935. An age-cohort model fitted the data for England well up to 1987, but substantially overestimated the numbers of adenocarcinomas in young women from 1990 onwards. In 1996-97 the incidence rate in women aged 25-54 years was less than 40% of that predicted from the age-cohort model. Interpretation The substantial increase in cervical adenocarcinoma in recent years is largely a birth-cohort effect presumably associated with greater exposure to human papillomavirus after the sexual revolution in the 1960s. The relative decline in younger women observed in more recent years suggests an effect of cervical screening.