The following countries are included (in parentheses abbreviations used in some tables etc.):

In the text and some graphs, the following combinations of countries have been used:

The average populations of the small areas varied from 10,400 in Finland and Norway to almost 200,000 in the former FRG (Table) 

In Russia the incidence and mortality data were not available by the small units (rayons) but only for counties (oblasts), which were divided into three categories: capital, other urban, and rural parts. Still, the average population of the units (excluding the huge St. Petersburg with a population of 5 million) was some 350,000 and the areas were very large in space. Because the coordinates and total populations of the towns were known, the urban oblast rates could be located to those map coordinates and weighted with population weights. Only the names - not locations or populations - of the rural rayons were known. Therefore a practical decision was made that each of the rayons within an oblast have equal population size and that the rayons are evenly located throughout the area of the oblast.

Cancer mortality data were available from all these areas for most of the cancer sites. Incidence data of satisfactory quality were lacking for part of Germany (former FRG except Saarland) and for Poland (except for cities of the Warsaw, Cracow and Wroclaw).

The number of cancer cases and cancer deaths for each cancer site together with the number of person-years were collected by small-area, sex and age (0-4, ... , 85+). The age-specific incidence and mortality rates were calculated by dividing the number of new cases or cancer deaths by the population in each stratum. The age-adjusted rates were calculated from the age-specific rates; the weights of the "world standard population"  were used. For restricted age ranges, truncated age-adjusted rates (world standard) were calculated.

The period of the incidence and mortality presentation is mainly between 1981 and 1990. However, there are some exceptions (although called "1981-90" in the headers):

The incidence and mortality rates were calculated for all sites combined and for 30 separate cancer sites. The list of sites, together with inclusion and exclusion criteria, is given as a table. It should be taken into account that the basic data for cancer incidence come from specialised cancer registries while the mortality rates mainly stem from the national statistical offices. These offices do not necessarily apply exactly the same coding rules or may have a different amount of diagnostic details available, which may cause incompability between mortality and incidence rates for certain areas.

There were insuperable problems in collecting incidence and mortality rates for skin melanoma and other skin cancers in a comparable way from different areas. Therefore, all skin cancers were excluded from the total cancer. [The rates for skin melanoma are yet shown as maps on the CD-ROM for those areas for which they were available.]

There were also problems in separating cancers of the colon and the rectum in a unique way, but the sum of these was comparable over all areas covered in this Atlas. The situation was similar with cancers of the oral cavity and pharynx: their sum was more comparable than the components separately. Therefore, also the sum categories for colon + rectum and oral cavity + pharynx are shown.

The individual rates for small-areas are not shown on the maps as such but as floating weighted averages of several neighbouring rates. By doing so the disturbing chance variation is reduced and the spatial trends become understandable. However, the rates for large cities (population > 350,000), in which the numbers of cases are satisfactory, are shown as such, i.e., non-smoothed. In some instances, neighbouring cities are combined if this corresponds to the actual situation in terms of behaviour of the inhabitants. E.g., the capital of Finland, Helsinki, together with the adjacent cities of Espoo, Vantaa and Kauniainen, are shown as one unit called "Greater Helsinki". The other combined cities are Greater Copenhagen, Odense, Århus and Aalborg in Denmark; Greater Oslo in Norway; Greater Warsaw, Cracow, Lodz and Poznan in Poland; and Greater Stockholm, Gothenburg and Malmö in Sweden.

The rate for each grid (size 2x2 km) on the map was defined as a weighted average of the age-adjusted incidence or mortality rates in the small-areas with population centres within 150 km from the middle of the grid. The weights were inversely associated with the distance; the weight was halved at the distance of 25 km (for the exact shape of the weighting function cf. Figure).

In addition, the weights were made directly proportional to the sizes of the populations within the 150 km circle. Cities with at least 350,000 inhabitants were excluded from the smoothing and their rates were shown as such as circles with a diameter relative to the population size and the colour indicating the incidence or mortality rate in the city. A white screen was used to dim the colour of areas with less than 1 inhabitant per km². For details on the map projection cf. Projection.

No smoothing was done over the borders between the nations or the three main areas within Germany (former GDR, Saarland, other parts of the former FRG).

The main type of scale used throughout this Atlas is a relative one with 19 colours varying from blue and green for low rates to red and violet for high rates [cf. graph]. The reference rate giving the mid-point of each scale is the combined age-adjusted incidence or mortality rate in Denmark, Finland, Iceland, Norway, Sweden, Estonia and Latvia for the cancer site and sex in question. These were the only countries with both mortality and incidence rates available for all individual sites in question. The total population of seven reference countries is 26 million

The reference rate falls in the middle of the scale (yellow), and each step between colours corresponds to a 1.15-fold relative increase in the incidence or mortality rate. The lower limit of the highest category is therefore always 3.28 (1.15^8.5) times the reference value, and the upper limit of the lowest category 30.5% (1.15^-8.5) of the reference. The relative difference between the lowest and highest category is thus > 10.8; if the geographical variation is smaller, the extreme colours are not used in the maps.

.

In order to compare the absolute incidence or mortality rates of various cancers, the same scale has to be used. To illustrate this, some maps were also produced with a fixed colour scale originally developed for the German mortality data (Eur J Cancer 1994; 30A:699-706). In that 19-colour scale the lightest colours indicate low rates and the darkest ones high rates. The lowest class limit always corresponds to an age-adjusted incidence or mortality rate of 5/10⁶ and the highest one a rate of 855/10⁶. The difference between steps increases according to a series 10, 15, 20, ..., 115, 120/10⁶. [cf. graph]

In addition to the maps, some basic descriptive data about cancer incidence and mortality are shown for different sites. The incidence and mortality rates for whole countries - summed up from the small-area data - are shown as bar diagrams for each cancer type, by sex. If there was marked variation in age pattern between the countries, the age-incidence or age-mortality curves are also shown. Because it was impossible to show all country-specific curves, countries with similar-shaped age curves were combined and their weighted averages shown. To avoid a dominance of large populations (with sometimes poorer data quality) each country was given a weight of 1.0 except for Iceland (population 0.2 million), which got a weight of 0.2.

Mortality to incidence rate ratios (small-area-based maps or bar diagrams on a country level) are sometimes shown as quality indicators of the rates or as an indirect illustration of survival or the level of diagnostic activity. In these maps a fixed light-to-dark colour scale with an absolute increase of 0.05 between the categories is used. [cf. graph] In the production of mortality/incidence ratio maps the smoothed value for both the incidence and mortality rate was first calculated for each grid, and the mortality/incidence ratio for that grid was based on these smoothed values. This approach makes it possible to calculate mortality/incidence rates even in a case when areal units for mortality and incidence are not exactly the same.

Short descriptions of cancer incidence and mortality registration in each country and of the country itself have been written by the co-authors from each of the participating countries.

The maps do not show geographical details such as rivers and mountains or names of main cities, etc. In the CD-ROM, however, one can easily check the names and populations of the main cities. If the general geography is not known to the reader, it is recommended to have a standard atlas at hand when studying the cancer maps.

There are no maps about the prevalence of aetiological factors of cancer, either. It would have been practically impossible to collect that kind of data in a comparable format for the different areas covered by the present Atlas. However, as far as there is firm knowledge about geographical variation of strong aetiological factors, e.g., within individual countries, these facts have been utilized when commenting the site-specific results.

In addition to this CD, there is a printed version of the Atlas. The book contains the full text of the Atlas but only a subset of the maps and other graphs to be found on the CD.