The Northern Ireland Cancer Registry (NICR) was established in 1994 and uses an automated computer system with multiple information sources to collate information on new diagnoses of cancer. Information has been collected for incidence years from 1993 onwards. The three main sources for registration are the Patient Administration System (PAS) used by all the Hospital Trusts, histopathology reports, and death notifications supplied by the General Register Office (GRO). From PAS the Registry obtains demographic information on individual patients along with basic site and behaviour information for each tumour. This information is supplemented by electronic downloads from histopathology and cytopathology laboratories. A major focus of the Registry’s operation is on the verification of the information from a single hospital admission, a single histopathology report or a single death certificate (death-certificate initiated cases). Trained Tumour Verification Officers (TVOs) examine general practitioners’ (GPs) notes for patients who have died from cancer, hospital records for cases identified without histopathology or cytology confirmation and histopathology reports where there is conflicting information or other possible errors. In the event that no further information on death-certificate initiated cases is obtainable the record is included in the Registry but flagged as a death certificate only (DCO) case. These comprised less than 2% of cases in 1995-2007. Follow up of patients is conducted passively by linking cancer incidence data to death certificate information.
In the NICR the information on cancer site received by the Registry has been coded using the Systematized Nomenclature of Medicine (SNOMED) which is used in the UK National Health Service (NHS) and information on cancer morphology has been coded to the second revision of the International Classification of Diseases for Oncology (ICD-O-2) (World Health Organisation, 1990). Cancer site is recoded at the Registry to the tenth revision of the International Classification of Diseases (ICD-10) (World Health Organisation, 1997).
The National Cancer Registry Ireland was established in 1991 and has produced national figures on cancer incidence since 1994. Most registrations are based on “active” data collection, whereby trained Cancer Data Registrars (CDRs), based in hospitals around the country, access a range of data sources to identify all new cancer cases and register all relevant patient, tumour and treatment details. Hospital pathology reports, provided to the Registry shortly after diagnosis, comprise the bulk of information, providing data on approximately 85% of all new cases. Most pathology reports are registered manually by the CDRs but about 10% of pathology reports are now provided in electronic format. Information on non-microscopically diagnosed cases is registered mainly from other hospital sources, principally the Hospital Inpatient Enquiry system (HIPE) as well as records from radiology and oncology departments, medical charts, etc. Most cases (≥95%) are registered in this way. The main non-hospital source of case information is death certificate data. The Registry is provided with all death certificates by the Central Statistics Office (CSO). All cases initially notified by death certificate are followed up with the hospital of death or the certifying doctor and most cases are subsequently found in other data sources. Only a small percentage of cases (<3%) remain classified as notified by death certificate only (DCO). As in the NICR, follow up of patients is passive, where cancer cases are linked to death certificate information provided regularly from the CSO.
Although case data from pathology reports is registered almost immediately after diagnosis, data from other sources can take longer to obtain. Together with essential case checking and data quality assurance, the Registry normally produces definitive statistics for case data a minimum of 18 months to 2 years following the end of year of diagnosis. Currently the completeness of cancer registration for all invasive cancers diagnosed to end 2007 is estimated to be over 96%.
Incident cases are coded according to the third edition of the International Classification of Diseases for Oncology (ICD-O-3) (World Health Organisation, 2000).
Before analysis, data on cases diagnosed 1995-2007 were recoded to the equivalent ICD-10 classification in both registries. Cases were extracted from both registry datasets, based on ICD-10 codes, and data from Northern Ireland (NICR) and Republic of Ireland (National Cancer Registry) were then amalgamated into a single dataset. Table 2.1 lists the 18 cancer sites analysed, with the relevant ICD-10 codes and total number of cancers in RoI and NI. Data from both registries have previously been used in three joint all-Ireland incidence reports—further information on data comparability and quality is provided in Donnelly et al., 2009. Multiple primary cancers were excluded in the calculation of incidence figures, based upon the rules published by the International Agency for Research on Cancer (IARC) (Ferlay et al., 2005). Death certificate only registrations were also excluded. For all cancers, maps are based on the incidence period 1995-2007. For breast and prostate cancers additional analysis was carried out, and maps produced, for the periods 1995-2001 and 2002-2007 separately because of the introduction of mammographic screening in RoI in 2000, and striking temporal trends in both countries in both prostate cancer incidence and prostate-specific antigen testing (Carsin et al., 2010).
Table 2.1 Incident cancers diagnosed 1995-2007 included in this report: Republic of Ireland (RoI) and Northern Ireland (NI)
cancer site | ICD-10 codes | Ireland | RoI | NI | |||
females | males | females | males | females | males | ||
|---|---|---|---|---|---|---|---|
non-melanoma skin cancer | C44 | 49,102 | 55,826 | 34,655 | 40,034 | 14,447 | 15,792 |
breast | C50 | 38,545 | *254 | 25,876 | *181 | 12,669 | *73 |
colorectal | C18-C21 | 16,992 | 21,205 | 11,041 | 14,485 | 5,951 | 6,720 |
lung | C34 | 13,005 | 20,831 | 8,437 | 13,672 | 4,568 | 7,159 |
prostate | C61 | — | 33,144 | — | 24,704 | — | 8,440 |
non-Hodgkin's lymphoma | C82-C85 | 4,605 | 5,094 | 2,917 | 3,441 | 1,688 | 1,653 |
stomach | C16 | 3,616 | 5,748 | 2,353 | 3,818 | 1,263 | 1,930 |
melanoma of the skin | C43 | 5,467 | 3,702 | 3,871 | 2,611 | 1,596 | 1,091 |
bladder | C67 | 2,513 | 6,226 | 1,730 | 4,309 | 783 | 1,917 |
head and neck | C01-C14,C30-C32 | 2,211 | 5,692 | 1,371 | 3,820 | 840 | 1,872 |
leukaemia | C91-C95 | 3,164 | 4,527 | 2,235 | 3,331 | 929 | 1,196 |
pancreas | C25 | 3,533 | 3,502 | 2,499 | 2,454 | 1,034 | 1,048 |
kidney | C64-C65 | 2,450 | 4,032 | 1,603 | 2,785 | 847 | 1,247 |
oesophagus | C15 | 2,370 | 3,907 | 1,592 | 2,627 | 778 | 1,280 |
ovary | C56 | 6,222 | — | 4,149 | — | 2,073 | — |
brain and other central nervous system | C70-C72 | 2,266 | 3,041 | 1,630 | 2,161 | 636 | 880 |
corpus uteri | C54 | 5,237 | — | 3,355 | — | 1,882 | — |
cervix uteri | C53 | 3,758 | — | 2,665 | — | 1,093 | — |
* Since breast cancer in males is rare, most of the analyses in chapter 4 are limited to breast cancer in women.
Apart from its obvious use in mapping, geocoding of cancer cases allows linkage to area-based data, such as population density and measures of socio-economic status (e.g. percentage unemployed). This is described in more detail in sections 2.2.4.2 and 2.2.4.3. This type of information is not, in general, accessible at the level of the individual cancer case in Ireland, and has to be inferred from area-based measures.
NICR routinely collects address information for registered cancers, allowing small geographic areas to be assigned to individual cancer registrations. This is accomplished through an electronic process which uses the postcode that accompanies the majority of NI addresses along with a postcode-to-electoral ward lookup file known as the Central Postcode Directory (CPD). This is maintained by the Northern Ireland Statistics and Research Agency (NISRA) and updated annually (Northern Ireland Statistics and Research Agency, 2010a). Addresses with an unknown, incomplete or invalid postcode cannot be assigned an electoral ward.
The National Cancer Registry attempts to code all addresses of cancer cases to the level of electoral division (ED), the smallest area for which census data can be obtained. Unlike NI, addresses in RoI do not have postcodes. Each address is therefore assigned to an ED by means of matching Registry address information to other data sources. This process of geocoding is carried out for the most part by matching Registry patient address data to the GeoDirectory database which provides a list of official postal addresses and location details for every residential and commercial property in the country (www.geodirectory.ie [1]). Data matching is carried out using software developed by the Registry, with some manual coding for any remaining unmatched records. Additional resources used include address tables from census surveys, supplied by the Central Statistics Office (CSO), as well as manually locating addresses on maps provided by Ordnance Survey Ireland (OSI). Using a combination of these resources, almost all complete patient addresses can be assigned to a particular ED.
For some cases it is impossible to assign an address confidently to a single ED, usually because the address is incomplete or ambiguous (4.4% of all RoI cancers included in this report; Table 2.2). The number of EDs to which the address could potentially belong is usually small (2 or 3) and, for analysis, these cases were assigned at random to one of the possible EDs, with the possibility of assignment weighted by the population of the alternative EDs.
At the end of the geocoding process, a number of registrations in both RoI and NI remained which could not be assigned to any ED/ward (3.3% of all cancers included in this report; 3.6% in RoI and 2.7% in NI) (Table 2.2). For these registrations, a fraction of the cases of each cancer type was allocated in proportion to each ED (RoI) or ward (NI) weighted by population. In NI almost all of these cases were non-melanoma skin cancers, but in RoI all cancer sites had a similar percentage of cases with unknown ED.
Table 2.2 Number and percentage of cases not assigned to an ED or ward, and of cases assigned to multiple EDs
cancer site | cases assigned to more than one ED (RoI only) | cancer cases not assigned to an ED or ward | |||||||
|---|---|---|---|---|---|---|---|---|---|
|
|
| Ireland | RoI | NI | ||||
number | % RoI cases | % all cases | number | % | number | % | number | % | |
non-melanoma skin cancer | 3443 | 4.6% | 3.3% | 4896 | 4.7% | 2803 | 3.8% | 2093 | 6.9% |
breast | 1045 | 4.0% | 2.7% | 948 | 2.4% | 839 | 3.2% | 109 | 0.9% |
colorectal | 1136 | 4.5% | 3.0% | 985 | 2.6% | 885 | 3.5% | 100 | 0.8% |
lung | 783 | 3.5% | 2.3% | 808 | 2.4% | 749 | 3.4% | 59 | 0.5% |
prostate | 1282 | 5.2% | 3.9% | 1128 | 3.4% | 983 | 4.0% | 145 | 1.7% |
non-Hodgkin's lymphoma | 270 | 4.2% | 2.8% | 305 | 3.1% | 254 | 4.0% | 51 | 1.5% |
stomach | 256 | 4.1% | 2.7% | 270 | 2.9% | 250 | 4.1% | 20 | 0.6% |
melanoma of the skin | 280 | 4.3% | 3.1% | 365 | 4.0% | 281 | 4.3% | 84 | 3.1% |
bladder | 248 | 4.1% | 2.8% | 222 | 2.5% | 208 | 3.4% | 14 | 0.5% |
head and neck | 199 | 3.8% | 2.5% | 190 | 2.4% | 168 | 3.2% | 22 | 0.8% |
leukaemia | 258 | 4.6% | 3.4% | 247 | 3.2% | 230 | 4.1% | 17 | 0.8% |
pancreas | 243 | 4.9% | 3.5% | 173 | 2.5% | 164 | 3.3% | 9 | 0.4% |
kidney | 205 | 4.7% | 3.2% | 129 | 2.0% | 120 | 2.7% | 9 | 0.4% |
oesophagus | 203 | 4.8% | 3.2% | 169 | 2.7% | 151 | 3.6% | 18 | 0.9% |
ovary | 183 | 4.4% | 2.9% | 165 | 2.7% | 154 | 3.7% | 11 | 0.5% |
brain and other central nervous system | 168 | 4.4% | 3.2% | 155 | 2.9% | 150 | 4.0% | 5 | 0.3% |
corpus uteri | 129 | 3.8% | 2.5% | 133 | 2.5% | 113 | 3.4% | 20 | 1.1% |
cervix uteri | 75 | 2.8% | 2.0% | 90 | 2.4% | 81 | 3.0% | 9 | 0.8% |
all cancers in this report | 10406 | 4.4% | 3.0% | 11378 | 3.3% | 8583 | 3.6% | 2795 | 2.7% |
A census of population was carried out in NI in 2001, the only census between 1995 and 2007. This census provided population data, broken down by sex and age, for 582 wards in 26 district councils. Population estimates for each year were available by sex and age at district council level. Annual estimates for the wards were derived from these total annual estimates, using the 2001 census as the basis for the splits by ward. The estimates for each year were then averaged to give an estimated average population by ward for the 1995-2007 period. Over this period the wards had an average population of 2,913, ranging from 784 (Bushmills, Moyle) to 9,654 (Botanic, Belfast) (Table 2.3, Figure 2.1).
Table 2.3 Population distribution of NI wards and RoI EDs
number of areas | mean population | standard error of mean | standard deviation | minimum population | 25th percentile | median | 75th percentile | maximum population | |
NI wards | 582 | 2913 | 48 | 1147 | 784 | 2219 | 2618 | 3238 | 9654 |
RoI EDs | 3355 | 1161 | 34 | 1956 | 62 | 309 | 525 | 1146 | 33983 |
Figure 2.1 Population distribution of NI wards and RoI EDs
RoI | |
 |  |
Three censuses were carried out in RoI during the period of this report, in 1996, 2002 and 2006. These censuses provided population data, broken down by sex and age, for 3,422 EDs in 1996 and 2002, and for 3,409 EDs in 2006. Population data were derived from the census small area population statistics (SAPS) files for 1996, 2002 and 2006. Official CSO estimates of the total population split by sex and age (but not by ED) were available for each year from 1995 to 2007. Annual estimates for the EDs were derived from the appropriate census and the CSO total annual estimates–the 1996 census results were used as the basis for the ED populations for 1995, a linear interpolation of the 1996 and 2002 census counts was used for 1997-2001, a linear interpolation of the 2002 and 2006 census counts was used for 2003-2005, and the 2006 census results were used for 2007 estimates. The estimates for each year were then averaged to give an estimated average population by ED for the period 1995-2007.
The average ED population over the period was 1,161; ranging from 62 (Mountstuart, Co. Waterford) to 33,983 (Dundalk Urban, Co. Louth) (Table 2.3, Figure 2.1). Dundalk Urban District comprised a number of EDs in 2006 which were merged for the purposes of this atlas (see below); this merged area was the largest single population unit treated as an ED in RoI. The population of the largest single ED (Blanchardstown-Blakestown) was 23,179.
At each census, the population of a number of EDs was so low that the CSO considered these EDs "confidential", published only total population figures for them, and amalgamated them with one or more neighbouring EDs for the purpose of reporting age-specific population numbers. EDs were considered confidential by the CSO if they included either 15 households or less, or 50 persons or less. There were 12 such confidential EDs in 1996, 19 in 2002 and 32 in 2006. Three of the 2006 confidential EDs had been merged with different EDs in 2002 and so, to create an estimated population for each ED for 1995-2007, any EDs that had been merged during any of these censuses were combined. These are shown in Appendix table A2.1.
The definition of a small number of EDs, and therefore the associated SAPS data, changed between the 1996 and 2002 censuses. These changes consisted of splitting or amalgamation of areas, rather than any movement of boundaries. EDs which had changed in this way were combined for analysis, and the available age and sex distribution similarly combined (Appendix table A2.2). In addition, between 1996 and 2006 there was considerable population growth in a number of towns, many of which consisted of a single ED (urban part), with a surrounding ED (rural part). As the population of these towns increased, they expanded into the rural area, but the ED boundaries remained unchanged. Because of the uncertainty of geocoding of new buildings in these towns, the urban and rural EDs were combined for analysis (Appendix table A2.3). Finally, for the towns of Drogheda, Dundalk and Wexford, population splits were not available for all EDs for all censuses, and the affected EDs were also merged for analysis (Appendix table A2.4). The population of the largest merged ED (Dundalk Urban) was 33,983. This combining of areas gave a final total of 3,355 EDs.
As the formal definition of “urban” areas in Ireland (RoI and NI) does not include many areas at the periphery of towns and cities, urban and rural populations were distinguished by population density (Table 2.4), based on the estimated average number of inhabitants in 1995-2007. Three categories were created for analysis, with the cut-off points (<1 person/hectare, 1-15 persons/hectare, >15 persons/hectare) chosen to give an approximately equal population in each group.
Table 2.4 Distribution of cancer cases and estimated average population in 1995-2007, and number of EDs and wards, by population density tertiles
population density | no. of cancer cases* | estimated average population | % of total population | number of EDs and wards |
<1 person/ha | 121,810 | 2,004,451 | 36% | 2,892 |
1-15 persons/ha | 90,597 | 1,644,792 | 29% | 403 |
>15 persons/ha | 129,380 | 1,940,844 | 35% | 642 |
Total | 341,787 | 5,590,087 | 3,937 |
* All cancers included in this report.
A range of area-based socio-economic measures is available from the population censuses in NI and RoI. However, the majority of these, particularly those relating to occupation and social class, use different definitions in NI and RoI, and are not directly comparable. Three measures were identified as having a degree of compatibility and have been used for analysis in this report:
These socio-economic measures had to be changed from those in the RoI cancer atlas (Carsin et al., 2009), as the necessary information was not available from the 2001 NI census.
Wards and EDs were ranked according to increasing levels of each of these three variables and were divided into population quintiles, (i.e. each quintile contained as close to 20% of the population as possible). The 20% of the population resident in areas with the lowest percentage of, for instance, unemployment, was assigned to quintile 1 while the 20% resident in areas with the highest percentage was assigned to quintile 5. All measures were based upon data for men and women combined from the censuses of 2001 in NI and 2002 in RoI.
Overall, 40% of the NI 16-74 year old population was economically inactive compared to 34% in RoI. Of the economically active population 7% in NI were unemployed in the 2001 census compared to 8% in the 2002 RoI census (Northern Ireland Statistics and Research Agency, 2003; Central Statistics Office, 2003). While, overall, 20% of the population of the island was resident in each unemployment quintile, 30% of the NI population lived in the areas of highest unemployment, compared to 16% of the RoI population (Table 2.5).
Among 16-74 year olds in RoI, 87% did not have a university degree (or academic equivalent) compared to 84% in NI. 25% of the RoI population lived in the areas with the lowest level of tertiary-level education in Ireland, compared to 10% of the NI population.
41% of the NI population aged 75 years and over lived alone, compared to 31% in RoI. 39% of the NI population lived in areas with the highest level of elderly living alone, compared to 12% of the RoI population.
Table 2.5 Population and number of areas (wards and EDs) included in each area-based socio-economic category
| RoI | NI | Ireland | ||||||||
| Quintile Range | Number of areas | Pop* | % of total population | Number of areas | Pop* | % of total population | Number of areas | Pop* | % of total population | |
| Unemployment; % of economically active persons, aged 16-74, who were unemployed | ||||||||||
| Least unemployed (Q1) | 0.0% - 3.5% | 967 | 754,815 | 19% | 114 | 358,345 | 21% | 1,081 | 1,113,160 | 20% |
| Quintile 2 | 3.6% - 4.7% | 695 | 894,496 | 23% | 78 | 225,691 | 13% | 773 | 1,120,187 | 20% |
| Quintile 3 | 4.8% - 6.2% | 672 | 831,031 | 21% | 104 | 283,577 | 17% | 776 | 1,114,608 | 20% |
| Quintile 4 | 6.3% - 8.6% | 593 | 803,616 | 21% | 121 | 320,098 | 19% | 714 | 1,123,713 | 20% |
| Most unemployed (Q5) | 8.7% - 47.3% | 428 | 610,592 | 16% | 165 | 507,827 | 30% | 593 | 1,118,419 | 20% |
| Total | 3,355 | 3,894,549 | 582 | 1,695,538 | 3,937 | 5,590,087 | ||||
| Education; % of persons aged 16-74 without a university degree (or academic equivalent) | ||||||||||
| Least with no degree (Q1) | 44.5% - 81.0% | 239 | 691,541 | 18% | 123 | 425,072 | 25% | 362 | 1,116,612 | 20% |
| Quintile 2 | 81.1% - 86.2% | 319 | 682,943 | 18% | 146 | 434,777 | 26% | 465 | 1,117,720 | 20% |
| Quintile 3 | 86.3% - 89.5% | 557 | 714,096 | 18% | 149 | 402,893 | 24% | 706 | 1,116,989 | 20% |
| Quintile 4 | 89.6% - 92.7% | 978 | 847,567 | 22% | 110 | 271,190 | 16% | 1,088 | 1,118,757 | 20% |
| Most with no degree (Q5) | 92.8% - 100.0% | 1,262 | 958,403 | 25% | 54 | 161,606 | 10% | 1,316 | 1,120,009 | 20% |
| Total | 3,355 | 3,894,549 | 582 | 1,695,538 | 3,937 | 5,590,087 | ||||
| Elderly living alone; % of persons aged 75 and older living alone | ||||||||||
| Least 75+ living alone (Q1) | 0.0% -24.4% | 843 | 1,003,543 | 26% | 37 | 110,246 | 7% | 880 | 1,113,789 | 20% |
| Quintile 2 | 24.5% -30.6% | 681 | 897,794 | 23% | 83 | 222,320 | 13% | 764 | 1,120,114 | 20% |
| Quintile 3 | 30.7% -35.7% | 589 | 796,720 | 20% | 108 | 322,087 | 19% | 697 | 1,118,807 | 20% |
| Quintile 4 | 35.8% -42.5% | 663 | 745,285 | 19% | 134 | 373,596 | 22% | 797 | 1,118,882 | 20% |
| Most 75+ living alone (Q5) | 42.6% -100.0% | 579 | 451,207 | 12% | 220 | 667,288 | 39% | 799 | 1,118,495 | 20% |
| Total | 3,355 | 3,894,549 | 582 | 1,695,538 | 3,937 | 5,590,087 | ||||
* Annual average of combined 1995-2007 population.
The three socio-economic measures and population density had varying degrees of correlation. However while the correlation coefficients between several of the measures were statistically significant, none represented a high level of correlation. The highest correlation was a negative association between education and population density (-0.365) (Table 2.6).
Table 2.6 Correlation coefficients (Spearman’s rank) for ward/ED characteristics
% of economically active persons aged 16-74 who were unemployed | % of persons aged 16-74 without a university degree (or academic equivalent) | % of persons aged 75 and over living alone | |
Population density | 0.231 | -0.365 | 0.112 |
% of economically active persons aged 16-74 who were unemployed | 0.197 | 0.147 | |
% of persons aged 16-74 without a university degree (or academic equivalent) | 0.008 |
Map 2.3 shows Ireland divided into approximate population density tertiles (<1 person/hectare, 1-15 persons/hectare and >15 persons/hectare). As expected, only geographic areas at the centre of large towns and cities, such as Belfast and Dublin, fell into the highest tertile. The majority of wards/EDs in Ireland had a population density of less than 1 person/hectare.
Map 2.4 shows the percentage unemployed in each ED/ward by quintiles. Areas of highest unemployment were found in north and west Belfast, north-west Ireland (including Donegal, Derry and Strabane), the west of Ireland (including Mayo) and parts of Newry & Mourne and Louth.
Low levels of tertiary-level education (as illustrated in Map 2.5) were found in rural parts of RoI, north and west Belfast, north-east Dublin and south-west Dublin. High levels of tertiary education were found in south Belfast and surrounding areas, central and southern Dublin and surrounding areas, and other urban areas and their environs in RoI, such as parts of Cork, Galway and Limerick.
Areas with high proportions of elderly persons (aged 75 and over) living alone were fairly randomly spread across Ireland, as seen in Map 2.6. The proportion was relatively high in Dublin and Belfast city centres, but low in the surrounding areas.
Estimates of cancer incidence rates in 19 (mainly European) countries, of a level of economic development comparable to Ireland, were taken from the GLOBOCAN 2008 dataset (Ferlay et al., 2008). Data for NI is included in that for the UK in this dataset, so the incidence rates shown for both NI and RoI are based on 2005-2007 data from the respective cancer registry. For this reason, the ranking of RoI and NI relative to each other shown in these figures is not always the same as that shown in the summary section, or in the sections for each cancer titled “Small geographic area characteristics and cancer risk”. It should also be noted that, although countries are shown as ranked in descending order of incidence rates, the differences in rates between countries were often quite small and may not be statistically significant.
For some cancers, the definition of cancer site used in GLOBOCAN differed slightly from that used in this atlas. In these cases international comparisons between Ireland and other countries are based upon the GLOBOCAN definition. The exception to this was non-Hodgkin’s lymphoma, where the ICD code C96 was included in the GLOBOCAN definition, but omitted from the NI and RoI figures. A footnote to the comparison graphs is provided to indicate where such differences occur.
Maps 2.3-2.6 Ward/ED characteristics
 |  |
 |  |