Population screening for cervical cancer in The Netherlands. A report by the Evaluation Committee.

A programme of population-based screening for cervical cancer in three adjacent regions in the Netherlands is described. The target female population aged 35-54 was 370,000. Screening was performed three times with intervals of three years during the period 1976-84. The total numbers of histologically proven abnormalities were 406, 1025 and 215 for severe dysplasia, carcinoma in situ and invasive carcinoma respectively. Detection rates were highest in the more urbanized regions and they fell from 3-5 per 1000 in the first to less than 2 per 1000 in the third screen. Specificity of screening was high enough to reach a predictive value of a positive test of 80% in the first screen and of 60-70% in the second and third screen. Population-based cervical cancer registries enabled the investigators to calculate the sensitivity of screening; if defined as the percentage of invasive cancers appearing within three years of cytological test result showing less than severe dysplasia, sensitivity was 60-70%. Further cytological follow-up of cytological smears with moderate dysplasia could be shown to increase sensitivity to about 80%. A case-control study showed that screening protects against a diagnosis of invasive cervical cancer (odds ratio 0.22 with 95% Cl 0.07-0.69) within the next five years. Prognosis of cervical cancer detected early is generally good. Survival statistics from the programme compare favourably with those from the 'normal care' system. Still, the stage distribution of the latter category has much improved over the last decades, probably due to introduction of the cervical spatula in general practice in 1975. Cost-effectiveness studies and future government health policy are discussed.