The RARECAREnet database on the epidemiology of rare cancers in Europe is drawn from the dataset of EUROCARE-5, the wider collaborative study on cancer patients’ survival in Europe (www.eurocare.it). Overall 94 European population-based cancer registries (CRs) participating in EUROCARE-5 adhered to the RARECAREnet project also. They provided information on cancer patients diagnosed up to 2007 and followed-up for vital status ascertainment to the end of 2008 or later.

The European regions and countries represented in the RARECAREnet project are the following:

1. Northern Europe: Finland, Iceland, Norway
2. United Kingdom and Ireland: England, Northern Ireland, Republic of Ireland, Scotland, Wales
3. Central Europe: Austria, Belgium, France, Germany, Switzerland, The Netherlands
4. Eastern Europe: Bulgaria, Czech Republic, Estonia, Latvia, Lithuania, Poland, Slovakia
5. Southern Europe: Croatia, Italy, Malta, Portugal, Slovenia, Spain

Overall, 27 countries are represented in the project, and among them:

• 19 Countries are covered by national CRs (Austria, Bulgaria, Czech Republic, Croatia, Estonia, England, Finland, Iceland, Northern Ireland, Latvia, Lithuania, Malta, Norway, Republic of Ireland, Slovakia, Slovenia, Scotland, The Netherlands, Wales);
• 8 countries are covered by regional CRs partially representing the population of their country (Belgium, France, Germany, Switzerland, Italy, Portugal, Poland, Spain).

The mean European population covered, over the period 2000-2007, was about 207,942,000, corresponding to 48% of the population of countries participating in RARECAREnet and 46% of the European Union population (excluding Norway, Switzerland, and Iceland, which are not EU members).

Rare cancers are those with an incidence rate of <6 per 100,000 per year in the European population. It is important to stress that rare cancers among the RARECARE list of cancers are identified based on the above criterion, in other words, on the basis of the European population and not of a country-specific population. Thus, rare cancers are always the same in all European countries.

Rare tumour entities, because of their specific problems related to health care organisation and clinical management, are more appropriately defined as a combination of topographical and morphological characteristics, as both defined by the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3).7 Each tumour entity in the ICD-O list has a pathologic basis; however, in order to define clinically distinct diseases, the pathological entities have to be grouped.

The RARECARE list is organised into three tiers, as illustrated, for example, in Table 1 for epithelial tumours of nasal cavity and sinuses.
The bottom tier (tier 3) on the list is the WHO name of individual cancer entities and its corresponding ICD-O-3 morphology and topography codes.  ICD-O-3 entities are grouped into categories (tier 2) of cancers, considered similar from the point of view of clinical management and research. These categories are then further grouped into more general categories of tumours (tier 1), considered to involve the same clinical expertise and patient referral structure.

Tier 2 entities, by definition, include only specific morphologies; thus, rare cancers are identified in this tier. Not Otherwise Specified (NOS) morphology codes (NOS: 8000, 8001 for solid cancers, and 9590, 9591, 9760, 9800, 9801, 9820, 9860 for haematological diseases) are never assigned to tier 2.

According to RARECAREnet analyses, 198 tiers 2 are rare thus commonly it is reported that rare cancers are 198 (➫ cancer list).

In the RARECAREnet list, a tier 1 can be common or rare, but in this on line analyses tool only data on rare cancers are presented, therefore:

• if a tier 1 is common (incidence >6 per 100,000 at EU level), such as epithelial tumours of lung, the common tier 2 entities of this tier and the NOS histotype are excluded from the tier 1 definition (e.g., for the lung: squamous cell carcinoma, adenocarcinoma, and poorly differentiated endocrine carcinoma are excluded from tier 1 together with the NOS histotypes);
• if a tier 1 is rare (incidence rate <6 per 100,000 at EU level), such as epithelial tumours of trachea, all the corresponding tier 2 entities as well as the NOS histotypes are included in the tier 1 definition. The list below identified these types of Tier1 1 as Tier1.

In summary, in this online analyses only rare cancers are displayed.

Incidence – General definition

Incidence Count

The observed number of cases is the number of new cancer patients arising in a given period in a specified population. This information is collected routinely by cancer registries, thus, the term observed refer to the number of new cases observed by CR.

Crude incidence rate

Incidence rate is the incidence count divided by the population at risk during the same period expressed as person-years. The rate provides an approximation of the average risk of developing a cancer. Incidence rate is usually expressed as an annual rate per 100,000 persons at risk (persons/year). Such an incidence rate is called crude and cannot be used to compare different populations.

Age-adjusted incidence rate

Age-adjustement is a technique for “removing” the effects of age from crude incidence rates so as to allow meaningful comparisons across populations with different underlying age structures.

Incidence – RARECAREnet project

Crude incidence in Europe in the period 2000-2007

To calculate crude incidence rates for rare cancers in Europe in the period 2000-2007, the pool of 83 RARECAREnet CRs providing data for all cancer types is considered. The population at risk during the same period is the general population (males and females) in each CR area, expressed as person-years.

Estimated number of new cases in 2013 in EU-28

The new cases in Europe and in EU-28 are obtained by multiplying the age and sex-specific crude incidence in Europe in the period 2000-2007 by the corresponding 2013 population provided by EUROSTAT for Europe (27 Countries represented in RARECAREnet project) and for EU-28, respectively.

Age-adjusted incidence over time in Europe

Age-adjusted incidence in Europe over time is calculated considering the ➫ pool of 42 RARECAREnet CRs with available incidence data for all cancer types between 1995 and 2007. Incidence rates over time are provided in three period of diagnosis: 1995-1998, 1999-2002, 2003-2007. The European standard population is used to derive age-adjusted rates.

Confidence intervals for incidence

The normal approximation is used with the standard errors to obtain confidence intervals for crude incidence rates. For age-adjusted incidence rates, the method described in Tiwari et al. (2006) is used to obtain confidence intervals. For large numbers, the above methods give the same results. Confidence intervals are expressed as 95% Confidence Interval (95% CI).

Survival – General definition

Relative survival

Relative survival is defined as the ratio of the observed survival in a group of cancer patients to the expected survival in a comparable group from the general population with similar age, sex, and other demographic characteristics. Relative survival gives a measure of the "net" survival due only to cancer by removing the effect of mortality due to causes other than cancer. Relative survival is estimated by the Ederer II method from registry-specific life-tables. Life-tables were smoothed with standard demographic methods and checked against published official mortality data.

The cohort/period approach

The cohort approach produces relative survival estimates for a distinct cohort of people diagnosed with cancer during a defined period of time. The period approach derives long-term survival estimates exclusively from the survival experience of patients during a defined calendar period, regardless of the date of diagnosis.

Confidence interval for survival

Standard errors of survival are derived by applying the Greenwood’s formula. Ninety-five per cent confidence intervals were computed through the logarithmic transformation, so that the lower bound was always positive and the upper bound could exceed 100%. However, in these cases the upper bound of the 95% CI has been truncated at 100%.

Survival – RARECAREnet project

Relative survival estimates in 2000-2007

Relative survival estimates are obtained by the cohort approach considering patients diagnosed with a rare cancer in the period 2000-2007 and followed-up to 31st December 2008. Since all patients are included (not only those followed up for 5 years) this method is also called ‘complete’ cohort analysis. All the participating ➫ 94 RARECAREnet CRs were included in this survival analysis.

Estimates are presented by follow-up time (1, 3, and 5 years after diagnosis) with the number of cases included in the analyses in Europe:

• by age (0-14;15-24;25-64;65+), by sex;
• by Country, by European geographic area.

Relative survival over time in Europe

Relative survival over time in Europe is estimated considering the subset of ➫45 RARECAREnet CRs with incidence data available from 1995 to 2007, using the period approach. The survival experience of rare cancers patients during the three calendar periods of diagnosis 1999-2001, 2002-2004, and 2005-2007 is analysed by follow-up time (at 1, 3, and 5 years from diagnosis).

Prevalence – General definition

Prevalence count

Prevalence count is the number of patients alive at a certain date (the prevalence index date), in a defined population, who have been diagnosed with a given cancer. The patients with more than one cancer of different types are counted just once in each single cancer type.

Prevalence proportion

Prevalence proportion is the number of cancer patients alive in a defined population at a given date (prevalence count) divided by the population living at the same date.

Confidence interval for prevalence proportions

The normal approximation is used with the standard errors to obtain confidence intervals for prevalence proportions.

Limited-duration and Complete prevalence

Limited-duration prevalence limits the number of patients to those diagnosed with cancer within a fixed time in the past (i.e., 2, 5, or 20 years) of a prevalence index date. This information can be obtained by CR data. The life status of cases lost to follow-up or censored before the index date can be estimated from the survival probability between the censoring and the index dates, derived from a subset of cancer patients matched by age and cancer type. Complete prevalence count/proportion includes all previously diagnosed patients alive at the prevalence index date, regardless of how long ago the diagnosis was given.
Complete prevalence can be estimated correcting the limited-duration prevalence obtained from the CR data using the completeness index, a correction factor to account for cancer survivors diagnosed before the CR activity started.

Prevalence – RARECAREnet project

Complete prevalence in Europe at 01/01/2008: counts and proportions

To calculate the 15-years prevalence at the index date of 1^st January 2008, 26 CRs which are able to provide cases for the period 1993-2007 are considered. Complete prevalence in Europe is calculated correcting the 15-years prevalence through the completeness indexes obtained in the RARECARE project. Complete prevalence proportions in Europe at 01/01/2008 are estimated dividing complete prevalence counts by the corresponding population at risk covered by the CRs participating in the RARECAREnet project.

Estimated number of prevalent cases in 2008 in EU-28

The estimated number of prevalent rare cancer patients in Europe was obtained multiplying the age and sex-specific prevalence proportions estimated in the RARECAREnet sample by the corresponding population provided by EUROSTAT at 2008 in the EU-28.

Confidence interval for prevalence proportions

The normal approximation is used with the standard errors to obtain confidence intervals for prevalence proportions.