Findings

Social status

There are cancers known to be more common in lower social classes than in high ones (Pukkala 1995, Kogevinas et al. 1997). These cancer sites typically showed a geographical incidence trend increasing from west to east, most strikingly so for cancers of the oesophagus and stomach in both sexes, and larynx and lung in males. Because the survival rate is also lower in eastern parts, the geographical west-east gradient was even stronger in mortality maps of these cancer sites.

On the other hand, cancers which are commonest in higher social classes showed a decreasing west-east gradient. Strongest geographical trends of this type were seen for skin melanoma, and cancers of the colon, breast, prostate, and brain. In these sites the better survival in the west resulted in a smaller variation in the mortality maps than in the respective incidence maps .

Aetiology

The aim of this study was not to estimate formally associations between observed geographical variation in cancer frequency and aetiological factors. The first reason for not doing this was the lack of strictly comparable data on the prevalence of aetiological factors by small area. Even if we had had those data, we still would have met the general problem of ecological studies, namely that the area-specific averages rarely represent homogenously enough the behaviour of the individuals in the area, and the observed relative risks therefore tend to be strongly diluted. Still, there are some obvious associations between known strong factors and cancer, which can be seen from our cancer maps without any statistical analyses and where the knowledge about the existence of risk factors in various areas is good enough to allow conclusions.

From international statistics it can be seen that consumption of strong alcohol and tobacco in eastern areas of Northern Europe has been much higher than, say, in Scandinavia (Eurostat Year Book 2000). Consequently we saw dramatically higher rates of alcohol and tobacco related cancers in the east. Within Nordic countries, women in Iceland and Denmark have smoked more than in the remaining countries, and the map of lung cancer well reflects this fact. In all countries covered by our Atlas, the habit of smoking among women has been accepted earlier in cities than in rural areas. Consequently, in the maps of strongly smoking-related cancers larger cities tend to show systematically higher rates than the less urbanised areas around them .

Smoking and alcohol consumption are examples of a cancer-related life habits, the prevalence of which varies greatly between males and females. Therefore, incidence and mortality maps for males and females look different for cancer sites strongly associated with tobacco and alcohol. On the other hand, for some other cancers the similarity of geographical patterns in men and women suggests that the aetiological factors must be similar in men and women. Such factors include environment (drinking water, air pollution, trace elements in the ground, radiation) and to a large extent also diet. According to the present understanding, only some dietary components may have such a strong population attributable importance that they can be seen in the cancer incidence rates for entire municipalities. The similarity of the present sex-specific cancer maps suggests - not unexpectedly - that diet probably plays a central role in causation of cancers of at least stomach and colon.

Diagnostics

In cancers which are more common in the west than east, the level of diagnostic activity often affects the incidence and - to a considerably smaller extent - mortality. The commonest cancer sites in Nordic countries today, i.e., breast cancer in females and prostate cancer in males serve as important examples of diagnostic activity dependent cancers. Part of the variation in the incidence of thyroid cancer depends on how actively the relatives of thyroid cancer patients are screened. For example, the high-incidence area in Finland has changed location over decades according to how much resources has been put into the screening of thyroid cancer in each of the health care regions.

This Atlas includes some maps demonstrating age-specific incidence and mortality rates in contrast to most atlases published before. The age-specific maps give essential extra information. For example, there was relatively little difference in overall cancer mortality in males between areas among pensioners, but a huge variation in working ages. The age-curves of most cancer sites suggest that there have been differences by age between countries in the way in which the health care system aims at getting cancers diagnosed. The age-incidence curves have a similar shape in all countries up to the age of about 65 incidence, all sites, females, whereas in the oldest age-groups there are relatively much less cases diagnosed in the post-socialistic countries than in the west. It seems that the system has taken care of cancer diagnoses until pension age, but after that the people should have taken care for their health independently, which has not been very successful.

Evidently, in the older ages in post-socialistic countries mainly patients with serious symptoms went to a doctor, and therefore their cancers are, on average, at more advanced stages than those in western countries. The impression of clearly poorer survivalrates in east is greatly affected by this different stage distribution; if survival rates are adjusted for stage, there are usually only small differences in survival rates (cf. Estonia-Finland breast cancer survival: Karjalainen et al. 1989). There is no unambiguous answer on how aggressive the search for cancers should be: in Denmark the health care system is less active in tracing non-symptomatic latent cancers than in the neighbouring countries. Survival rates in Denmark are somewhat poorer than in other Nordic countries (Berrino et al. 1999) but people are more satisfied with their health care system than in any European country (Mossialos 1997).

Small-areas vs. larger areas

The geographical patterns seen in cancer maps are often merely variations between entire countries than between small-areas within a country. One example of an intra-country variation — in addition to the urban-rural variation demonstrated above — is the decreasing trend towards the north in the incidence of skin melanoma observed within all Nordic countries. If the association between sun radiation and skin melanoma would not have been known before, this geographical trend might have given a hint towards this aetiology. The maps on incidence of testicular cancer may be used, e.g., when considering the possible role of semen quality (sperm count) — the geographical variation of which is also rather well-known — as a risk factor of testicular cancer. The maps on thyroid cancer show high rates in mountain areas (Zatonski et al. 1996), among populations consuming much fish along the Atlantic coast, and also in areas where endemic goiter has been common. All this suggests that an abnormal intake of iodine may increase the risk of getting thyroid cancer. If it would have been possible to draw the maps by histology, we might have been able to demonstrate that the papillary type of thyroid cancer is common in iodine-rich areas while the follicular type is associated with iodine deficiency.

Certain cancers (e.g. Hodgkin´s disease, myeloma, leukaemia, and cancers of the pancreas and bone) showed no geographical variation whatsoever. This should not necessarily be interpreted as if these cancers would not have external risk factors at all, even though the aetiology of these cancers is largely unknown. The risk factors may be so evenly distributed among populations that variation between sub-populations does not show up as geographical differences. Occupation-specific risks serve as an example of this problem: even if the relative risk might be large, the proportion of exposed persons is so small that the few extra cases do not contribute much to the cancer rate of the whole municipality. An exception of this is lip cancer, which is linked to outdoor work (especially when combined with smoking; Lindqvist et al. 1979). Farmers and fishermen are at increased risk. Because the proportion of farmers is sufficiently large in rural municipalities, their high lip cancer incidence rate affects the rate of agricultural areas so strongly that it is clearly observable in cancer incidence maps. Lip cancer is the only site showing systematically lower rates in cities than in rural areas around them.

If a cancer rate in a small-area is very high and the population of the municipality sufficiently large in comparison to the surrounding areas, that point of observation may "jump out" of a smoothed map and thereby attract the interest of researchers. The high rate of bladder cancer in Greater Copenhagen is clearly visible in maps for both males and females in both incidence and mortality maps. If there would have been no smoothing, there would have been similar red spots allover the map due to random variation, and this single important observation would have gained less interest. Reasons behind the high rate in Copenhagen have been studied before (Jensen et al. 1986), but no good explanation has been found.

back