All records in NZCR have been linked to the NHI (National Health Index) and have a unique identifier (Health Care User number) assigned.
Ethnicity information must be collected and classified according to the Ministry of Health's Ethnicity Data Protocols for the health and disability sector. In 2009 the way ethnicity was assigned in the NZCR changed with the introduction of an ethnicity algorithm. This algorithm was run retrospectively across all registrations back to 1989.
With this method all records are assigned up to three ethnicity codes. This algorithm searches the NHI, the Mortality Collection and hospital discharge data in the NMDS (National Minimum Data Set) for ethnicity information.
The algorithm applies a 20% threshold for ethnicity data collected in NMDS (that is the ethnicity must be on 20% of records to be used in the algorithm). Events prior to 1989 retain their original ethnicity due to the lack of supplementary information from other data collections.
The ethnicity algorithm is periodically re-run over the cancer registration data in the Cancer Datamart to update and maintain high quality data.
The following geographical codes are assigned to cancer registrations:
- Domicile code - Statistics New Zealand domicile code representing the healthcare user's usual residential address at the time of diagnosis. For domicile codes see the domicile code table.
- Health facility code - code of the health facility where the cancer was diagnosed or treated. See the facility code table on the Ministry of Health's website.
Clinical coding systems
Two clinical coding systems are used by the Cancer Registry:
- The WHO International Statistical Classification of Diseases and Related Health Problems (ICD) codes are used to classify the tumour site (location).
- The WHO International Classification of Diseases for Oncology (ICD-O) is used to classify the tumour morphology (histological type and behaviour code).
ICD coding changes
Table 1: ICD (site) revision changes
|-1967||Mapped to ICD-9-CMA-II||ICD-7|
|2000-2002||All records from 1948 mapped to ICD-10-AM in 2001. Also an ICD-9-CMA-II code is automatically derived and stored with the record.||ICD-10-AM 2nd Edition|
|2003-2008||ICD-10-AM 3rd Edition|
|2009-2013||ICD-10-AM 6th Edition|
|2014-2019||ICD-10-AM 8th Edition|
|2020 +||ICD-10-AM 11th Edition|
The changes and increasing detail over time may make it hard to map data forward (mapping older data to new codes). Mapping the ICD-10-AM data back to ICD-9 CMA-II is useful for looking at time series data.
Morphology coding changes (ICD-O)
The third edition of ICD-O contains a revised morphology section. New classifications were introduced and new codes assigned to accommodate them. This resulted in changes to the coding of cancers diagnosed from 1 January 2003.
For some tumour types, particularly haematological malignancies and ovarian cancer, these changes have affected incidence reporting and registrations from 1 January 2003 onwards are not comparable with registrations prior to this date (see below).
Table 2: ICD-O (morphology) version changes
|Prior to 1995||ICD-O 1st Edition|
|1995 -2002||ICD-O 2nd Edition|
ICD-O 3rd Edition
Table 3: Specific morphology coding changes
|ICD-O 2nd Edition||ICD-O 3rd Edition||Result|
|Ovarian tumours of borderline malignancy||Considered malignant||Considered to be of uncertain behaviour in the third edition and have been excluded from incidence reporting since 2003||Resulted in a slight reduction in the number of cases of ovarian cancer since 2003 compared with previous years.|
|Polycythaemia vera, myelodysplastic syndromes and chronic myeloproliferative disorders||Considered to be of uncertain behaviour||Considered to be malignant||The ICD-10 codes for these new malignancies are in the range D45–D47 and were included for the first time in the 2003 data. A diagnosis of leukaemia in a person already registered with one of the above malignancies may not be counted in incidence statistics because of the way the coding rules for multiple primary tumours are applied (see the next section).|
|Pilocytic astrocytomas||Considered malignant||Considered to be of uncertain behaviour||Excluded from incidence reporting since 2003.|
Table 4: Other coding changes
|Prior to 1 January 2005||From 1 January 2005||Description|
|Superficial transitional cell carcinoma of the bladder||Considered invasive||No longer considered invasive||Coded as in-situ and therefore excluded from incidence reporting. This resulted in a decrease in the number of bladder cancer registrations compared with previous ye|
Some fields may be added, deleted or modified over time. For a list of the fields and any field changes please see the table of available data.
Long-term trend data is unavailable for some fields and some cancer types. This may be due to increasing/decreasing field details and changes to the inclusion/exclusion criteria (eg leukaemia and bladder cancers).
After the passing of the Cancer Registry Act 1993 and Cancer Registry Regulations 1994, data quality and completeness significantly improved.
Due to data quality issues staging information is not considered reliable prior to 1997.
Data quality measures
Routine data assurance measures include:
1. Ensuring completeness of data collection by:
- monitoring reporting from laboratories
- reconciling information from multiple sources (NMDS, Mortality Collection, screening programme and clinician-maintained databases)
- using NHI (National Health Index) for identification purposes.
2. Ensuring accuracy and completeness of coding by:
- electronic data entry validation checks such as morphology/site checks
- periodic manual checks of coding accuracy
- extensive pre-publication data checks (manual and automated)
- engineering the NZCR IT system to deal with the complex rules for recording and reporting multiple primaries, ensuring that duplicate registrations do not occur.
3. Ensuring consistency of coding by:
- having specialist coders
- having an in-house procedure manual
- the use of international standards.