Screening for cancer.

The screening programme should be planned as an experiment before large scale application. Otherwise, due to the different biases inherent in nonexperimental research in general, there is a high probability of arriving at inconclusive evidence. Too often a favourable natural history is presumed, including the assumption that treatment of a screen-detected case postpones death. After such an assumption is made it is very difficult to carry out research on whether such prerequisites are true independently of the correctness or falsity these assumptions. The screening programme requires not only a valid test but also a valid programme, including elements such as the quality of the test material (cervical smear), the quality of the laboratory, the attendance of the population and the availability of referral and treatment facilities. The only valid measure for the efficacy of a screening programme is mortality (and sometimes incidence) from the disease. There are many other indicators, called process measures, all of which may show a favourable effect even in cases where the programme is inefficient. As for the application of screening for cancer in the Nordic countries, the present state of the art is as follows: Screening for cervical cancer is a well established practice performed by the health services. Every year about 1000 cases of invasive disease are likely to be prevented in the five Nordic countries as a result of cytological screening. The effectiveness of screening for breast cancer is also established by modern scientific methods. The large-scale Swedish experiments clearly show the potential of screening based on mammography.(ABSTRACT TRUNCATED AT 250 WORDS)