The South Australian Cancer Registry and Data Linkage

1
The South Australian Cancer Registry and Data
Linkage

• Dr Wayne Clapton1, Dr Colin Luke1,2, Kevin
  Priest3, Adrian Heard3, Anh-Minh Thi Nguyen3,
  Graeme Tucker3, Maria Cirillo1, Heather Hall1,
  Mary Merdo1, Teresa Molik1, Maxene Rosenberg1,
  Chris Scott1, Lesley Milliken2
• 1South Australian Cancer Registry, 2 Clinical
  Epidemiology and Hospital-Based Cancer Registries
  Coordination, 3Health Statistics Unit - (all in
  the Epidemiology Branch, SPPH Division, DHS)

2
OVERVIEW

• Introduction
• Some brief points about Cancer
• General Features of a Cancer Registry
• Overview of the SA Cancer Registry
• Current Use of Data Linkage
• Other Potential Uses of Data linkage
• Summary and Conclusions

3
CANCER STATISTICS - SA

• South Australia - 2001 Figures (2003 SA Cancer
  Registry Report)
• - All Cancers, excluding non-melanotic skin
  cancers
• Males
• 4,208 new cancer cases diagnosed (11 per day)
• AS Incidence Rate/100,000 males 352.7
• 1,830 deaths from cancer (5 per day)
• AS Death Rate/100,000 males 138.0
• Females
• 3,492 new cancer cases diagnosed (9 per day)
• AS Incidence Rate/100,000 females 271.2
• 1,442 deaths from cancer (4 per day)
• AS Death Rate/100,000 females 89.6

4
Most Frequently Diagnosed Male Cancers - SA 2001
5
Most Frequently Diagnosed Female Cancers - SA 2001
6
Most Frequent Cancer Deaths in Males - SA 2001
7
Most Frequent Cancer Deaths, Females - SA 2001
8
CANCER STATISTICS - AUS

• Australia - 1999 (AIHW AACR, Cancer in
  Australia 1999, November 2002)
• - All Cancers, excluding non-melanotic skin
  cancers
• Males
• 44,514 new cases diagnosed (122 per day)
• AS Incidence Rate/100,000 males 380.0
• 19,609 deaths from cancer (54 per day)
• AS Death Rate/100,000 males 163.0
• Females
• 37,671 new cases diagnosed (103 per day)
• AS Incidence Rate/100,000 females 291.1
• 15,086 deaths from cancer (41 per day)
• AS Death Rate/100,000 females 102.3

9
CANCER - GENERAL

• Risk of cancer before age 75 years (AIHW AACR,
  2002)
• Males 1 in every 3 males
• Females 1 in every 4 females
• One of the 7 National Health Priority Areas
• Cardiovascular, Cancer, Mental Health, Injury,
  Diabetes Mellitus, Asthma, Arthritis
• Contribution to Total Burden of Disease and
  Injury (by DALY score (Mathers et al., 1999))
• Cardiovascular Disease 20
• Cancer 19
• Mental Disorders 14

10
WHAT IS A CANCER REGISTRY ?

• A perpetually on-going and ever-growing
  repository of data on cancer in the community
• Collects data on ALL cases of cancer in a defined
  population.
• Cancer is a legally mandated notifiable disease.
• A Cancer Registry provides a Service for Public
  Health Monitoring and Surveillance of Cancer in
  the Community.
• It is an Applied Public Health Practice Service
  activity, rather than being research.

11
SA Cancer Registry - Legal Underpinning

• Cancer is a legally mandated notifiable disease
  (without penalty)
• South Australian Health Commission (Cancer)
  Regulations 1991 (No. 73) (South Australian
  Health Commission Act, 1976)
• Hospitals and Pathology Laboratories are obliged
  to report all cases of cancer (except
  non-melanotic skin cancers) within one month of
  discharge and/or diagnosis.

12
SA Cancer Registry - Legal Underpinning - contd

• Conditions relating to Confidentiality and
  Provision of certain information
• Section 42 42a, Public and Environmental Health
  Act, 1987
• Defines confidential information and authorises
  access to it for people employed or engaged by
  the State for the purpose of
• Monitoring public health in the State, or
• Investigating public health problems within the
  State, or
• Assessing and improving the quality of public
  health in the State.
• Authorises disclosure to persons performing
  functions as above or to persons providing
  technical, administrative or secretarial
  assistance to the authorised person.
• Prohibits disclosure of confidential information
  unless the disclosure is made
• In the course of official duties or
• With the consent of the person to whom the
  information relates or
• Is required by a court or tribunal constituted by
  law.

13
What does a Cancer Registry measure?

• By collecting ALL cases of cancer (ie FULL
  ASCERTAINMENT) for a defined population,
• INCIDENCE (new cases in a defined time period)
• MORTALITY (deaths from cancer in the time period)
• By collecting accurate information on times and
  causes of death,
• SURVIVAL RATES can be calculated for various
  forms of cancer.

14
What does a Cancer Registry measure? (contd)

• Incidence, mortality and survival are usually
  STANDARDISED by age and gender to allow
  comparisons to be made between populations and
  time periods with different mixes of age and
  gender proportions. They are expressed usually
  as RATES per 100,000 population.
• Estimates of PREVALENCE (number of existing cases
  in the population at a particular time point or
  over a time period) may be done as well or this
  can be estimated by population sampling (eg,
  Omnibus and/or SERCIS surveys)

15
Some egs of how SA Cancer Registry Information
is used?

• Assessing the burden of cancer, and of its
  various types, in the community at particular
  time points and over time (time series
  monitoring).
• Comparing different population groups and
  different geographical areas.
• Evaluating the effects of interventions at a
  population level (eg, population-based screening
  programs)
• Contributing to research, eg suggesting
  hypotheses for aetiological or interventional
  studies.
• Contributing to health services planning
  activities.
• Contributing to disease prevention and health
  promotion.
• Supporting clinical endeavours and quality
  assurance.

16
An Example of SA Cancer Registry data as early
warning
17
Cancer Registries - Bottom Line

• A good quality Cancer Registry provides accurate,
  complete, timely information about the extent,
  burden and trends over time of cancer in the
  community - so that informed, logical, sensible
  decisions can be made.
• Without such a facility, there is only incomplete
  and inaccurate information and thus conjecture
  and ill-informed poor decision making is the most
  likely outcome.

18
Important Non-technical factors

• Hospitals and Path Labs have a legally mandated
  obligation to report cancer but no obligation
  necessarily to cooperate further in resolving
  problems of data accuracy and completeness.
• Medical practitioners are under no legal
  obligation to report or cooperate with the cancer
  registry (though exchanges of information are
  protected under Section 42a) but do so
  voluntarily to assist with resolution of problems
  on their patients.
• Hospital-based Cancer Registries are owned by the
  clinicians
• ie, bottom-up rather than top-down
  organisation
• Therefore, to do its work, the cancer registry
  relies heavily on goodwill, trust and cooperation
  with numerous individuals and institutions, and
  on wide-ranging networks of contacts built up
  over many years.

19
SACR System Outputs Some Routine interactions

• INTERNATIONAL
• International Agency for Research on Cancer
  (IARC), Lyon, France.
• International Association of Cancer Registries
  (IACR))
• Surveillance, Epidemiology and End Results (SEER)
  Program, USA.
• NATIONAL
• National Cancer Statistics Clearing House
  (NCSCH), Canberra
• Australian Institute of Health and Welfare
  (AIHW), Canberra
• Australasian Association of Cancer Registries
  (AACR)
• Other Australian Cancer Registries
• STATE/LOCAL
• Case notifications to Hospital-Based Cancer
  Registries
• Linkages with
• BreastScreenSA (interval cancer detection)
• SA Cervix Screening Program (Data quality
  assurance)
• Interactions with other cancer data collections
• Supporting Govt Non-Govt programs (eg, The
  Cancer Council SA)

20
SACR System Outputs - Some ad hoc contributions

• National Cancer Survival project
• AIHW/AACR Ductal Carcinoma in situ (DCIS) Report
• National Colorectal Cancer Care Survey
• National Breast Cancer Care Survey
• Health Status Follow-up Studies
• Perceived Cancer Clusters in the community
  (usually in association with some environmental
  concern)
• Metropolitan Services Reviews
• National Public Health Partnerships activities
• Data for Policy Reviews
• Health services utilisation reviews
• National Cancer Control Initiative (NCCI)
  activities
• The Cancer Council Australia activities
• The Cancer Council South Australia activities
• Australasian Association of Cancer Registries
  activities
• Health statistics chartbooks

21
SACR System Outputs - Current

• Annual Report
• Paper form
• Internet form at http//www.dhs.sa.gov.au/pehs/di
  sease-control-stats.htm
• Responses to ad hoc enquiries and requests

22
SACR System Outputs - Proposed

• Annual report mainly Internet /- CD-ROM
• Regular Interim Internet mini-reports
• Interactive enquiries of the database
• Integration with other systems, eg clinical
  information systems, decision support systems,
  health statistics collections, etc
• Defining user needs better
• Secure electronic data exchange with authorised
  agencies

23
SACR Data Capture - Current

• Sources
• Hospitals (paper)
• Pathology Laboratories (paper /- electronic)
• Radiation Oncology Units (paper)
• Australian Bureau of Statistics (electronic)
• Registrar of Births, Deaths and Marriages
  (electronic /- microfiche /- paper)
• Individual Medical Practitioners (paper)
• Interstate Cancer Registries (paper)
• National Death Index (paper /- electronic)

24
SACR Data Capture - Proposed

• Move to secure electronic data exchange from/to
  all sources
• Discs / CDs
• Electronic transmission (modem /- direct lines
  /- secure encrypted e-mail)
• Direct updating from sources databases
• ?Data quality issues
• Paper
• Scan and deal with data electronically via a
  document processing system

25
SACR Data Capture - Non-technical issues

• The legislation defines the format of the
  hospitals Reporting of Cancer form
• Privacy and confidentiality aspects
• Changes to the ways in which hospitals, medical
  practitioners and laboratories operate, eg
• Inpatient vs outpatient care
• Day surgery
• Amalgamation of pathology labs - at the
  corporate and/or test level

26
SACR Analyses methodologies

• Current
• Extraction of a de-identified file for ad hoc
  analyses on statistical packages as are
  appropriate for the task, eg EpiInfo, SAS, Stata,
  Statistica etc
• Batch tables annually from the SACR system
• Proposed
• Enhanced ad hoc enquiry facilities for simple
  analyses
• Better integration of analyses with external
  reporting processes (eg, WA LSHTM/ONS)
• ?User access to de-identified data via the
  Internet or CD-ROM (eg, SEER) to do their own
  analyses

27
SACR Quality assurance

• Cross checks on receipt of all data.
• Chasing information from hospitals, labs and
  medicos where discrepancies are detected.
• Experienced professional coders and a coding
  manual.
• Editing and validation at computer data entry.
• Ascertainment checks against other sources, eg
  hospital discharge diagnoses.
• Cross checks again at analysis stages.
• However, limitations are present and QA
  activities will be extended in any re-vamped
  system.

28
SACR Privacy and confidentiality issues

• Taken very seriously by all concerned
• A privilege and an obligation to operate the
  Cancer Registry
• Australian and State Privacy Legislation
• Other concerns about privacy are prevalent,
  locally and nationally
• A very significant non-technical issue for the
  operation of cancer registries
• The perennial ethical conundrum - the greater
  good of the community vs individual rights

29
Current Use of Data Linkage SA Cancer Reg

• BDM Death Linkage
• AIHW National Death Index
• Hospital Ascertainment Checks
• Potential Duplicates
• BreastScreen SA Interval Cancers
• Cervix Screening Quality Assurance
• Research Projects
• Hospital-Based Cancer Registries

30
Births, Deaths and Marriages - Deaths Linkage

• Cancer Deaths found monthly by manual processing
  of in-coming BDM discs
• Annual match of whole previous years BDM data
  with Ca Reg live cases
• Automatch (name, date of birth, gender)
• Finds any cancer deaths missed on monthly
  processing
• Picks up non-cancer deaths

31
AIHW National Death Index Linkage

• Coded deaths from ABS go to National Death Index
  at AIHW
• NDI matched against Ca Reg live cases from 1977
  to most current complete year
• Finds interstate and non-cancer deaths
• Done annually
• Automatch name, date of birth, gender, date of
  diagnosis, or date of last contact

32
Hospital Ascertainment Checks Linkages

• 6 monthly comparisons of major hospitals cancer
  cases with Ca Reg records
• If Hospital has a cancer case which Ca Reg does
  not, this indicates a missed notification ? it
  can be retrieved
• Locally written program
• Crude match (name, date of birth, gender)

33
Potential Duplicates

• Looking for duplicate cases on Ca Reg
• Match of Ca Reg with itself
• Done intermittently every few years
• Also done on a national basis by AIHW
  (time-consuming for registries)

34
BreastScreen SA Linkage with Ca Reg

• Match BreastScreen SA files with Ca Reg
• Indicates Cancers reported to Ca Reg, but not
  known by BreastScreen SA and vice versa
• ? Interval Cancers arise between screens
• A BreastScreen QA accreditation factor
• Presumes an up-to-date Ca Reg
• Ca Reg menu item in-built Soundex match (every
  six months)

35
SA Cervix Screening Program link with Ca Reg

• Match Cervix Screening files with Ca Reg
• Indicates Cancers reported to Ca Reg, but not
  known by Cervix Screening and vice versa
• A quality assurance check for both
• Presumes an up-to-date Ca Reg
• Done by Cervix Screening - annually

36
Research Studies

• Ethics Committee approved studies only
• Not compromise our legal obligations
• Matching Research data with SACR to
• Identify Cancer Cases
• Confirm a Cancer Diagnosis
• Determine Vital Status
• Usually via Automatch
• Time, effort and cost implications

37
Hospital-Based Cancer Registries

• HBCR/SACR matches for
• Mutual ascertainment case capture
• Research study purposes
• Usually a crude match (registration number, date
  of birth, sex, primary site)

38
Potentially Useful Data Linkages (if available) -
1

• Hospital-based (Clinical) Cancer Registries
• clinical information, grade, stage,
  investigations, treatment
• Hospital Clinical and Administrative Systems
• Hospital experience
• Medicare and/or Health Practitioner Systems
• Community experience
• Burden of Disease Systems

39
Potentially Useful Data Linkages (if available) -
2

• Molecular biology data repositories
  (Bioinformatics)
• Molecular characterisation of tumours, genetic
  associations, treatment targeting
• Research Databases
• Mutual enrichment of information
• Geo-spatial Databases
• Spatial descriptive analyses, aetiological and
  ecological relationships, cluster investigations
  (Keyhole Earthviewer)

40
(No Transcript)
41
(No Transcript)
42
Inhibitions to Data Linkage of SA Ca Reg with
Others

• Legislative
• Eg, SA Cancer Registry cannot share identified
  information with anyone not fulfilling criteria
  of Section 42a
• Moral and Ethical
• Utilitarianism vs Autonomy
• Community Acceptance/Objection

43
Future Role of a Cancer Registry in a linked
world

• Expert collector of quality data
• Everybody should not try to do everything
  themselves wasted effort
• Integration within a broad-based Health
  Information System (eg, HealthConnect)
• Broadband Internet and emerging communications
  technologies (eg wireless)
• Personal Health Identification Number needed

44
Conclusions

• Cancer is a very important health problem
• Cancer registries provide essential
  population-based surveillance information
• Data linkage is integral to Ca Reg work
• (QA, finding deaths, ascertainment, assisting
  others)
• More widespread data linkage and integration
  could enhance information breadth and depth
• Legislative, moral and ethical issues, and
  potential community objections are the most
  important limiting factors

45
Acknowledgements

• The South Australian Cancer Registry is located
  within and is fully funded by the South
  Australian Department of Human Services. It is
  situated within the Epidemiology Branch (Acting
  Director Dr Colin Luke) of the Population Health
  Group (Director Dr Kevin Buckett) of the
  Strategic Planning and Population Health Division
  (Director Dr David Filby). This paper
  acknowledges and thanks the people who do the
  excellent day to day work of the Cancer Registry
  Maria Cirillo, Mary Merdo, Heather Hall, Teresa
  Molik, Maxene Rosenberg, and Chris Scott
  Coordinator of Hospital-based Registries, Lesley
  Milliken Dr Colin Luke (Acting Director,
  Epidemiology Branch and Head, Clinical
  Epidemiology Unit) Health Statistics Unit -
  Kevin Priest (who does most of the data
  matching), Adrian Heard, Graeme Tucker and Anh-
  Minh Thi Nguyen Information Management Services
  Division current and former staff who designed
  and implemented and/or who now maintain the
  Cancer Registries computer systems such as
  Joanne McQuin, Stuart Speight, Alan Boman, Geoff
  Murphy, Andrew Grandison, Jayson Rogers, Som
  Liensavanh, Phil Ancell, Greg Brodie-Tyrrell, Jon
  Holloway, and others. The former Epidemiology
  Branch and Cancer Registry Director A/Prof David
  Roder and A/Prof Kerry Kirke (of the Cancer
  Council South Australia) also are acknowledged
  for their past and/or continuing support of
  Cancer Registry activities.