Dutch Cancer Atlas

What is IKNL’s purpose with the Dutch Cancer Atlas?

The Atlas’s purpose is described in more detail in Purpose of the Dutch Cancer Atlas. In short, the goals of the atlas are:

• To raise awareness: To let people know if and how cancer rates vary in different regions and where possible help them understand the meaning and causes of variation.
• To encourage research: To inspire more studies to investigate why cancer rates are different in various regions.
• To promote prevention: To motivate actions that can prevent cancer and reduce its impact.

Who is the Dutch Cancer Atlas intended for?

• For anyone seeking information about the incidence of cancer in specific regions.
• For researchers and healthcare professionals who require input for further research.
• For governments, policymakers, healthcare organizations, healthcare professionals and citizens looking for opportunities for preventive measures.

Why is there geographical variation for some types of cancer?

For some cancer types, such as skin cancer, cervical cancer and lung cancer, there is a lot of geographical variation in the atlas, while for others, like breast cancer, there is little to no variation. Differences between regions may be caused by several factors, such as differences in lifestyle choices (like past exposure to sunlight), environmental factors (such as air pollution from radon or wood burning), diagnostic practices, access to healthcare (which can be affected by socioeconomic disparities). Differences can also simply be caused by chance. In the atlas, the section 'Info on cancer types' explains what is shown on the map for each type of cancer. Where possible, this information includes interpretation of the variation and its potential causes.

It is important to note while someone may have one or more risk factors, this does not necessarily mean they will inevitably develop cancer. In fact, some individuals may develop cancer without having (known) risk factors. Also, it is generally not known to what extent a specific risk factor contributed to an individual's cancer development. The risk factors mentioned in the text are therefore meant as possible explanations for the variation seen on the maps. In many cases, the exact reasons for geographical differences in cancer incidence and the role of each individual risk factor remain unclear.

I live in a red area, does this mean that I have a higher risk of getting cancer?

The Dutch Cancer Atlas shows the average risk of cancer in an area by showing whether a particular cancer type occurs more or less often than expected based on the Dutch average. This does not necessarily reflect the risk of getting cancer for every individual in that particular area. A higher diagnosis rate in a particular region likely reflects the general health and lifestyle of people living in that area. It can also be the result of particular environmental factors, data registration, diagnostic practices and access to healthcare. In many cases, the cause of variation is unknown and variation may also simply be due to chance. The section "Info on cancer types" in the atlas explains what is shown on the map for each type of cancer. Where possible, this information includes interpretation of the variation and its potential causes.

What does the Dutch Cancer Atlas show about asbestos-related cancer in the Tata Steel area?

In the areas near Tata Steel, the incidence of mesothelioma (asbestos-related cancer) is noticeably higher. Between 2011 and 2022, this form of cancer was diagnosed more than twice as often north of Tata Steel compared to what would be expected based on the national average in the Netherlands. However, it is important to keep in mind that the absolute number of cases will still be low, since mesothelioma is a rare type of cancer; just over 3 out of every 100,000 people in Netherlands are diagnosed with mesothelioma.

Why is this happening and is some of the incidence caused by Tata Steel?

The primary cause of mesothelioma is asbestos exposure. This form of cancer can develop after exposure to asbestos fibres and typically occurs 20 to 30 years or even longer after exposure. So, the people diagnosed in the last ten years likely were exposed to asbestos at least 20 years ago. In general, mesothelioma is more prevalent among men than women, also in the area around Tata Steel. This is due to the fact that most asbestos exposure in the past took place in the work environment, particularly in shipbuilding and industry, where men make up the vast majority of the workforce.

However, it is unknown where people with mesothelioma living in the area around Tata Steel at the time of diagnosis were exposed to asbestos fibres. It is unclear to IKNL (Netherlands Comprehensive Cancer Organisation) to what extent asbestos was used in the vicinity of the blast furnaces at Tata Steel. It is known that in the past asbestos was used in shipbuilding. In the area around Tata Steel, there was a significant amount of shipbuilding in the past. Therefore, the higher number of mesothelioma diagnoses is likely not solely attributable to Tata Steel. Since the 1990s, preventive measures have been taken and working with asbestos is no longer allowed due to its significant health risks.

What about the air quality around Tata Steel? Doesn’t that pose an increased risk of cancer?

The atlas shows that in the area around Tata Steel the lung cancer rate is approximately 14% to 29% higher than the Dutch average. In general, about 80 percent of all lung cancer diagnoses is estimated to be caused by smoking, 10 percent by air pollution (including second-hand smoke, radon in homes, emissions from factories, wood burning, etc.) and the cause of the remaining 10 percent is unknown. However, based on the atlas alone it is not possible to determine how many additional cases of lung cancer are attributable to the air quality in the area. The RIVM (National Institute for Public Health and the Environment) monitors air quality around Tata Steel. More information can be found (in Dutch) at RIVM's website.

If I look at the numbers for 'men' and 'women,' in some cases they do not seem to add up when compared to the numbers for 'all persons’. Why is that?

This is for example the case for head and neck cancer in Lelystad. In that area, it is estimated that the diagnosis rate in men is 43% higher than the Dutch average and in women 17% higher. One would expect that the estimate for all persons would lie somewhere between 17 and 43%. However, the diagnosis rate for all persons is 50% higher than the Dutch average.

This is because the atlas uses smoothing, and head and neck cancer is relatively rare. In statistics, the more data is available for analysis, the more reliable the outcome will be. Therefore, the atlas uses data from a specific area as well as from its surrounding areas. In the example of Lelystad, the atlas can use more data for ‘all individuals’ compared to ‘men’ or ‘women’. So, for ‘men’ and ‘women’, more data from surrounding regions are included in the calculation. The map shows that in surrounding areas, the diagnosis rates of head and neck cancer were low. This explains why the percentages for men and women are lower.

In summary, more data provide more certainty about an area, and areas can be pulled down in percentage by surrounding areas if those surrounding areas have a lower Incidence. The value in each area is thus a weighted average of its own incidence and the incidences of surrounding areas, with the surrounding areas providing more information (having more weight) when there is more uncertainty in the area in question (which is the case for cancer types with low incidence). Read more to learn more about smoothing.

Who was involved in the writing of the texts?

The texts in the atlas were developed in consultation with experts, including scientific researchers, and when necessary, healthcare professionals and Public Health Services (GGD).

What is the meaning of the text panel that pops up when I click on an area?

Clicking on an area in the Dutch Cancer Atlas prompts a window that shows by what percentage the diagnosis rate in that area differs from the Dutch average. It also indicates whether there is little or sufficient certainty that the area actually differs from the Dutch average. The percentage displayed in the atlas is an estimate. By indicating whether there is little or sufficient certainty that the area deviates, it reflects the level of certainty of the Standardized Incidence Ratio (SIR) of that area. For example, if the display shows '14% above the Dutch average, likely to be a real difference,' this means that the probability that the SIR in this area differs from the Dutch average is high. Conversely, if it says '14% above the Dutch average, unlikely to be a real difference’, this means that the probability that the SIR differs from the Dutch average is low.

In the panel with the statistics, I see the average number of new diagnoses per year. Is this the average for my own region?

No, these statistics represent the average number of diagnoses per year for the entire country of the Netherlands over the period from 2011 to 2022.

Which ages are included in the atlas?

The atlas includes all age groups. In the analyses the estimates are standardized for age by taking age group at diagnosis into account. By adjusting for differences in age distribution between regions, regional differences that are shown in the atlas are not the result of an older or younger population in an area.

Is it possible to filter the results by age in the Atlas?

No, this is not possible. The atlas includes all ages. In the analyses the estimates are standardised for age. By adjusting for differences in age distribution between regions, regional differences that are shown in the atlas are not the result of an older or younger population in an area. However, for scientific research or statistical purposes it is possible to request certain data from the Netherlands Cancer Registry (NCR).

What does ‘unlikely to be a real difference’ mean?

‘Unlikely to be a real difference’ means that the probability that the Standardized Incidence Ratio (SIR) differs from the Dutch average is low.

What does ‘likely to be a real difference’ mean?

‘Likely to be a real difference’ means that the probability that the Standardized Incidence Ratio (SIR) differs from the Dutch average is high.

How many additional diagnoses are there when the diagnosis rate is higher?

The number of additional diagnoses differs per region. It is impossible to know how many diagnoses there are in each region, solely based on the estimated standardized incidence ratios (SIRs). Two regions with the same SIR still can have a different number of additional diagnoses, and two regions with the same number of additional diagnoses can have a different SIR.

Below are two hypothetical and simplified examples of areas with the same SIR, but a different number of additional diagnoses. These examples are solely for illustrative purposes and show an easy way of calculating the SIR. The estimation of SIRs for the atlas is more complex, using not only the expected and observed diagnoses in a particular region, but also information from neighbouring areas (spatial smoothing).

In region 1, we expect 10 diagnoses over the studied period (accounting for the population size and age distribution of region 1). The actual number of diagnoses in this region was 15. So, in this region, there are 5 more patients than we would expect based on the Dutch average.

In region 2, we expect 20 diagnoses over the studied period (accounting for the population size and age distribution of region 2). The actual number of diagnoses in this region was 30. So, in this region, there are 10 more patients than we would expect based on the Dutch average.

In both these areas however, the SIR is 1.50. In the atlas, this would have been shown as '50% above the Dutch average’.

What does 'higher' or 'lower' mean in the Atlas?

For each area, the modelled estimate is reported relative to the Dutch average. The standardized incidence ratio (SIR) is estimated and the value 1 represents the Dutch average. If the SIR is 1.30 this means that the average diagnosis rate in that area is 30% (or 1.3 times) higher than the Dutch average. If an area has an estimated SIR of 0.7, this means that the average diagnosis rate in that area is 30% lower than or (0.7 times) the Dutch average.

However, we do not automatically label an area with an SIR lower than 1 as 'lower' and an area with an SIR higher than 1 as 'higher'. We also consider (un)certainty. We assess how likely it is that the SIR truly deviates from the Dutch average. For further explanation, refer to Statistical Model for Cancer Diagnoses.

Is it possible to extract the underlying data from the Atlas?

No, this is not possible. In the Atlas, the estimated Standardized Incidence Ratios (SIR) are produced using statistical modelling. These estimates show whether the cancer diagnosis rate in an area is higher than, equal to, or lower than the Dutch average diagnosis rate. From these estimates it is not possible to derive the underlying data. However, for scientific research or statistical purposes it is possible to request certain data from the Netherlands Cancer Registry (NCR).

Is cancer mortality included in the Dutch Cancer Atlas?

No, for now we decided to focus only on cancer incidence.

Are data on Socioeconomic Status (SES) included in the Dutch Cancer Atlas?

No, for now, these data are not included. The focus of the atlas is currently only on cancer incidence.

Does the Dutch Cancer Atlas contain information on distance to health care?

No, for now we decided to focus only on cancer incidence.

Which postcode is used in the Dutch Cancer Atlas?

The postcode of the place of residence at the time of diagnosis is registered in the Netherlands Cancer Registry (NCR). This postcode is used in the analyses for the atlas. Usually several decades pass between exposure to risk factors and cancer diagnosis. Therefore, geographic variation of cancer incidence in the Netherlands will mainly be caused by variation in exposure to risk factors in the past. Many people move during their lives, so the location where someone was exposed to a risk factor may differ from the place where their cancer is diagnosed. It was not possible to account for relocations in the analysis.

What is meant by smoothing?

The estimated SIRs (standardized incidence ratios) in the Dutch Cancer Atlas are ‘smoothed’. This basically means that the models that produce the estimates for a particular area not only use data from that area but also use information from neighbouring areas.

By smoothing, the estimated SIRs will be more stable and less affected by random fluctuations. Random fluctuations in cancer incidence have more impact on SIRs in areas with few residents and/or low cancer incidence. For instance, a difference of ‘just’ one cancer diagnosis in an area with low cancer incidence can lead to more extreme SIR estimates than a similar difference in an area with higher cancer incidence.

A hypothetical example: if in a small area 3 cancer diagnoses are observed while the expected count is 2 (SIR = 3/2 = 1.5), the diagnosis rate appears to be 50% higher than expected. However, in a larger area where 30 cancer diagnoses are observed while the expected count is 29 (SIR = 30/29 = 1.03), the incidence is ‘just’ 3% higher than expected. Low populated areas can cause extreme outliers, while differences are sometimes simply due to randomness. Smoothing reduces the extremeness of such outliers and makes the uncertainty of the estimated SIR smaller. Areas where the diagnosis rate truly deviates from the Dutch average will still differ after smoothing. Adjacent areas that deviate due to randomness because of small numbers will appear more similar after smoothing. Smoothing leads to a more stable and realistic depiction of the geographic pattern of cancer incidence. Read more about the statistical methods .Another result of smoothing is that it preserves the privacy of cancer patients. Individual cases cannot be identified from the results in the atlas.

See Figure 7a for a hypothetical example that shows the difference between results that are not 'smoothed' and results that are smoothed.

Are unsmoothed data available for the atlas?

No, the atlas shows only smoothed estimates. There are two reasons for this. First, smoothing the estimates preserves cancer patients’ privacy. Second, smoothing provides more stable and robust estimates of cancer incidence and reduces the effects of random fluctuations in low populated areas. Smoothed estimates are designed to reflect the real differences in cancer diagnosis rates between areas.

How does the atlas deal with statistical uncertainty of the estimates?

The Dutch Cancer Atlas provides information on the certainty of the SIRs (Standardized Incidence Ratios) in several ways. The atlas shows two measures of certainty. The first is the probability that the SIR truly deviates from the Dutch average, and the second is the so-called ‘credible interval ,' which shows the precision of the estimated SIR.

The probability that the SIR truly deviates from the Dutch average is shown in the V-plot and through the use of colour on the map (transparency). The credible interval is represented in the so-called Wave-plot. Read more about statistical (un)certainty of the Cancer Atlas.

How do I switch to a different cancer type for a particular area?

Clicking on a particular area in the Dutch Cancer Atlas will prompt an information panel that shows all different cancer types. By clicking on the box next to a cancer type in this panel, the map will switch to that specific cancer type. This way, you can check all cancer types in a particular area. The colour of the box in the information panel indicates whether the diagnosis rate for that cancer type is lower than (blue), equal to (yellow) or higher than (red) the Dutch average. Within the information panel you can switch to grid view (click on the icon in the top right corner next to the cross). The grid view shows coloured blocks which represent for each cancer type whether the diagnosis rate is lower than, equal to or higher than the Dutch average. Clicking on each block will switch the map to that particular cancer type.

I do not see the full text in 'info on cancer types.'

Either click on the arrow under ‘More information’ or simply scroll down (or up) in the text to see all the text.

Why does the descriptive text in the panel ‘info on cancer types’ not automatically change when I select a different cancer type in the panel on the right?

Unfortunately, this is not possible in the atlas. If you want to see the map for a particular cancer type and read the accompanying descriptive text, you can select a different cancer type in the left panel ‘info on cancer types’.

How was the selection of cancer types determined?

The selection of cancer types was based on whether there were a sufficient number of diagnoses to produce reliable and stable estimates. The cancer types displayed in the Dutch Cancer Atlas are therefore the 24 most common cancer types. The atlas includes only primary tumours and not metastases (secondary cancer sites).

What are the black areas with grey crosses on the map?

In the Dutch Cancer Atlas, there are some missing pieces on the map. These areas are displayed as black regions with light grey crosses. These areas contain either no residents or too few residents to produce reliable estimates. Such areas are for instance the port of Rotterdam and the ‘Maasvlakte’ at the 3-digit postcode level. At the 4-digit postcode level such areas are for instance on the south side of Eindhoven or ‘Schokland’ in the Noordoostpolder.

Is the Dutch Cancer Atlas also accessible to people with colour blindness?

In the Dutch Cancer Atlas, the geographic variation of cancer is represented using a range of colours on the map. The used colour scheme is selected by the creators of the Australian Cancer Atlas . In their colour selection, they took various forms of colour blindness into account. A colour blindness simulator (https://www.color-blindness.com/coblis-color-blindness-simulator/) provided substantial evidence that the chosen colour scheme remained interpretable by most forms of colour blindness, except for monochromatism.

For some types of cancer, the incidence increased over time. Is this visible in the atlas?

The current version of the atlas includes no information about time trends. However, a report from IKNL (Netherlands Comprehensive Cancer Organization) titled Kanker in Nederland. Trends en prognoses tot en met 2032 (Cancer in the Netherlands: Trends and Projections up to 2032) was published at the end of 2022 and contains information about trends on national level (in Dutch).

How am I allowed to use data from the atlas?

The IKNL website's disclaimer explains how you can use information from the atlas, as we consider the Dutch Cancer Atlas to be part of the IKNL website.