CANCER MAPPING IN ALPINE REGIONS
Cancer mapping has a long tradition in cancer epidemiology. Maps have been produced for decades and some of them have been regarded as milestones, see for example the Cancer Mapping of Scotland . Both layout and methods have changed since then. Layout has mostly changed due to progress in IT-technology and methods ever since programmes for sophisticated smoothing methods have become widely available [2,3]. Our main goal is to show a stable estimate of the regional distribution in our study area.
The study area covers fourteen registries in the eastern part of Switzerland, the western part of Austria, Northern Italy and the Alpine parts of Slovenia. The study area covers about 6.2 million people, hence more than half of the Alpine population. The entire study area is located in Alpine regions, where we are confronted with sparse populations. Regional units are small with an average population size of 20,000 inhabitants, the smallest units not having less than 10,000. There are only few larger cities with more than 100,000 people. We present incidence and mortality data for the years 2001 to 2005, but shorter periods for some registries due to the availability of incidence data.
Incidence data are collected by cancer registries that are part of Cancer Incidence in Five Continents . Mortality data are official data in the whole area, with the exception of South Tyrol where a local procedure for collecting mortality data was implemented. We present information for main cancer sites and omit rarer cancer sites in order to provide more stable estimates.
Also, for main cancer sites we are confronted with small numbers per unit, and thus need to apply smoothing methods. We show only smoothed maps, which should give a more stable estimate of the underlying pattern. In addition to maps, we also present additional material for those readers interested in details. Cancer registries require funding, and some people ask whether it is still necessary to run cancer registries. Regional distribution of cancer incidence and mortality is one of the important public health questions, where cancer registries can play a major role. Maps are very easy to read, which means the importance of cancer registries can be demonstrated to a broad audience.
We hope that our material will be widely discussed and will lead to in-depth investigations of those sites where we see clear risk gradients in regional distribution. The material is presented in such a way that not only specialists should be able to read the maps. This of course entails some danger, because the interpretation of cancer maps requires great skill. Higher rates in some regions can be due to a) risk factors, b) screening activities (which can directly influence rates namely low incidence, for example for cervical cancer, and at least temporarily high incidence, for example for prostate cancer), c) differences in outcome if we see high risk in mortality and not in incidence, d) documentation or registration bias, and finally e) random variation.