Response to the letter from Drs Oortmarssen and Habbema

Sir-Our primary purpose was to develop a consistent model based on Swedish statistics to describe the natural history of cervical neoplasia from the time when the patient develops cancer in situ (CIS) up to the time of death (Gustafsson & Adanmi, 1989). Conditions in Sweden and the need for a dynamic model favoured a mathematical approach rather than a statistical one. This meant that the concepts, assumptions , definitions, etc., had to be somewhat different from those usually used in a statistical context. For many years Sweden has had a national cancer register which includes all cases of cancer in situ. However, 75-80% of the smears are taken outside the organised screening programme , and these examinations are not registered but all findings of cancer in situ and invasive cancer must be reported. Thus, while this situation is ideal for relating the number of in situ cases found and eliminated to the decrease in invasive cancer and mortality, it is difficult to assess matters such as cost-effectiveness, test sensitivity, participation patterns, etc. Concerning the main issues raised in the letter from van Oortmarssen and Habbema: 1. These authors claim that our estimate of the proportion (P) of new in situ cases that would progress into invasive cancer without screening (12.2%) is too low. 2. They estimate the detection rates of cancer in situ from our results and calculate that these would correspond to a detection rate of between 25 and 30 per 1,000 women on a first screening of women aged 30 to 45 years. 3. They claim that our figures imply that 85% of the cases of cancer in situ detected by screening are treated unnecessarily. (In our paper (Gustafsson & Adami, 1989) we stated that 80% is 'typical' although we said the figure ranged between 77% and 85% since it varies with age.) In model terms these statements all question whether the rates of cancer in situ in Sweden are correct. The incidence rates of invasive cancer and the mortality rates in countries like Holland, the UK and Canada are rather similar to those in Sweden for the period studied (Canadian Task Force on Screening, 1976). But the in situ figures are several times higher in Sweden than estimates from corresponding countries. In Sweden about 4,000 cases of cancer in situ are reported annually; this figure should be compared to the reported number of cases of invasive cancer, …