The DAHNO Audit and the Cancer Registries

1
The DAHNO Audit and the Cancer Registries

• Chris Carrigan, National Coordinator, Cancer
  Registration

2
Cancer Registration

• A standard dataset is collected for all new
  incident cases
• For England, approximately 225,000 new cases are
  registered each year
• Of these, 8000 are head and neck cancers

3
Registry/ DAHNO collaboration

• Key Partnership
• Data Quality Assurance
• internal quality/consistency of data
• produce first output analyses
• Provision of Population Denominators
• Data Analysis
• Collaboration through registries National
  Analysis Group, key staff from South West and
  Oxford

4
Data Quality

• What is Data Quality?
• Quality Control
• Quality Assurance
• Manual vs. Automated

5
Comparisons to DAHNO?
Medical Records
DAHNO System
Online Forms
Manual Collation
DAHNO Database
Automated Collation
Supplier Information Systems
6
Manual vs Automated Assurance

• Manual
• Expensive
• Can be highly detailed and abstract
• Highly skilled
• Interpretation error
• Automated
• Cost efficient
• Requires multiple electronic sources
• Very specific
• Absolute (interpretation)
• What does each method test?

7
Automated Quality Assurance

• Cross-registry collaboration (led by Trent and
  Thames via Lucada)
• Additional validations devised with Lead
  Clinician and Project Manager (cross-checks for
  consistency)
• 100 initially identified, essentials agreed

8
Manual Quality Assurance

• Small scale QA exercise using audit data
• Involves manual re-abstraction of the whole
  dataset by trained registry staff for randomly
  selected cases
• With thanks to Carol Lister at NYCRIS

9
Pilot Data Quality Exercise

• Registries defined methodology
• Cover one whole cancer network
• Pilot cancer network within the NYCRIS area
• 3 Trusts 6 hospitals

10
Method

• A sample of records to be quality assured via
  data re-abstraction
• Time period cases submitted to audit database
  between July-September 2004
• Each Trust asked to identify 50 records submitted
  during time period
• 20 casenotes requested from each of these

11
Method

• Each Trust asked to identify 50 records submitted
  during time period
• 20 casenotes requested from each of these
• Data abstracted onto data collection form by two
  experienced NYCRIS data abstractors
• Results compared with original submissions to
  audit database

12
Problems in Reabstraction (1)

• NHS number missing from most of the sample
  casenotes for one hospital crucial to the
  matching exercise
• Lack of information in casenotes about MDT
  meetings only one hospital had a standard
  recording sheet
• (Sometimes several meetings, but details often
  not in casenotes)

13
Problems in Reabstraction (2)

• Data definitions some just different from
  cancer registry definitions, some open to
  interpretation (even using the manual)
• e.g.
• Co-morbidity
• Watchful waiting

14
Comparison of results
matched data items by Trust
15
Results (cont)

• Postcode different from postcode at diagnosis
• Referral dates problems when referral pathway
  not straightforward

16
Results (cont)

• For one hospital, 13 specialist referral dates
  were missing
• Diagnosis date wide variation in interpretation
• Primary site diagnosis 41 had site recorded,
  when a more definitive site should have been used

17
Results (cont)

• Basis of diagnosis missing for 21 cases
• Confusion between recording cytology and
  histology
• Staging could have been added for 30 cases from
  CT scans or investigation results

18
Results (cont)

• Treatment in some cases, all treatment details
  missing in others only partially completed.
• For one hospital, treatment recorded for all
  cases, where only investigative had been
  performed
• RT details missing for 27 cases
• Death details missing for 37 cases

19
Conclusions (cont)

• A time-consuming exercise (4 weeks of trained
  staff time for 71 cases)
• Significant variations, including
• Important details missed
• Lack of definition (e.g. site)
• Illogical details added

20
Conclusions (cont)

• Balance between Quality Control and Quality
  Assurance
• Not getting it right first time takes time,
  energy, effort and money to discover further down
  the line
• Improve the Quality Control on entry (unpopular)

21
The DAHNO Audit and the Cancer Registries

• Chris Carrigan, National Coordinator, Cancer
  Registration