Mortality Atlas of Cancer and Other Causes.
Spain 1975-1986.

Guide for reading and interpreting maps and graphs

With the intention of aiding the reader to draw his/her own conclusions from the various maps, graphs and tables, a few general pointers for interpretation purposes have been set out below.

1. Dettection of the existence of geographical patterns

This amounts to attempting to answer the question: "Could chance alone explain the distribution of mortality due to any given cause?" The first step is to examine Rates Map . These maps are artificially colour-compensated. There will always be three provinces in those shades that denote scale extremes.

One has to try and perceive whether there is a discernable gradient in the shading of the scale colours, i.e., whether there is a rise in mortality from north to south, the interior to the coast or vice versa. Examples of such gradients are :

as well as many other (non-cancer) causes of death which exhibit a marked and defined north-south pattern (cardiovascular diseases in general and COPD).

All these causes quoted here by way of example present statistically significant D cluster-indices . In other words, chance alone cannot explain such geographical distribution or, put another way, there may be environmental causes at play which could possibly account for this distribution. Similar distribution patterns can in general be construed as diseases with similar causes. The clearest example of this is provided by tumours associated with the habit of smoking :

which display a pattern similar to and coinciding with that evinced by cardiovascular diseases.

2. Distinguishing causes of death having environmental components from those having genetic components

Diseases with a pronounced environmental component in their etiology will present rates having a relatively wider range (i.e., difference between highestand lowest rates). In order to check for this, observe the histogram featuring the rankings of the adjusted rates and the range. A useful exercise might be to calculate a ratio between the highest and lowest rates. For instance, cancer of the buccal cavity and pharynx in males yields a rate ratio of 4:1 (Cadiz 10:6, Avila 2:6). As it happens however, this ratio -regardless of which malignant tumour is being studied- always exceeds 2, and there is the possibility of outliers (e.g., Ceuta and Melilla). Of greater use then is to determine whether the set of provinces lying close to the Spanish average (provinces at the intermediate level) is very numerous or, what amounts to the same thing, whether the highest peak in the smoothed histogram coincides with or approaches the average.

3. Gender-related differences in mortality

Ranked histograms for males and females are drawn to the same scale, that is to say, the Y axes are the same length. This enables comparison of the "thickness" of the histograms, thereby highlighting the differences. All tumours associated with the habit of cigarette smoking (e.g., buccal cavity and pharynx , esophagus , larynx , and lung ) and consumption of alcoholic drinks (larynx and esophagus) exhibit considerable differences as between the sexes.

4. Time trends

Predominance of one of the two shades used in the third map (time trend) indicates -in addition to intensity- the precise direction of trend in Spain.

For example, stomach cancer shows a predominance of blue (decline) in all provinces, with the possibility of some provinces being pinpointed where the fall-off is slightly smaller. Comparisons should be run between maps plotting male and those plotting female trends, since there are divergent gender-related patterns for some important causes associated with cigarette smoking, such as lung cancer, where observation of females reveals which provinces are undergoing a shift in direction of trend.

5. Competitive causes

One of the reasons for including other causes of mortality in this study was to check their role as competitive causes for cancer insofar as distribution was concerned. An example of this is the marked north-south pattern for all cardiovascular diseases, which might conceivably be concealing the pattern for frequent malignant tumours, such as those found in cancer of the lung and esophagus. However, those provinces with the highest mortality rates for cardiovascular diseases also have the highest rates for cancer.