Automatic Quality Control and Assurance at the Thames Cancer Registry

1
Automatic Quality Control and Assurance at the
Thames Cancer Registry Neil Hanchett Thames
Cancer Registry
2

• What do we mean when we talk about quality at the
  Thames Cancer Registry?
• A few examples
• Completeness of case ascertainment
• Timeliness of case registration
• Completeness of the record for a given patient
• Accuracy of the record for a given patient
• Rigour of analytical methods
• Data describing reports
• Extent and detail of process documentation
• National and International collaboration

This presentation is concerned with automatic
quality control and assurance as practised at the
Thames Cancer Registry. Currently this is
focused on accuracy of our patient records.
3
Why do we differentiate between quality control
and quality assurance? The Thames Cancer Registry
is engaged in the collation of cancer
registration records for patients from a wide
array of different data sources. Our data
sources are extracts from data collected
originally for differing purposes at differing
points along the patient pathway and subsequently.
CRFs
ONS
Collation Process
TCR Integrated Cancer Registration Database
NHS Trusts
Other Registries
4
What do we mean by automatic? Quality checks made
by the Thames Cancer Registry Information System
(TCRIS) rather than manually.
Medical Records
TCRIS
CRFs
Manual Collation
TCR Integrated Cancer Registration Database
Automated Collation
Supplier Information Systems
5

• Automatic Quality Control
• 122 simple (single data item) validations
• One example Is the incoming hospital name in our
  dictionary of standard hospital names?
• Coded in TCRIS as (Dont try to read this)
• CRSV576 Automatic Input - Simple Validation
• (c)ATT Istel
• Itemname hospital (local variable lv76)
• s vpok""
• sql(SELECT HOSPITAL INTO vpok FROM HOSPITAL
  WHERE hospital_namelv76)
• i SQLCODE sql(SELECT HOSPITAL INTO vpok FROM
  HOSPITAL WHERE hospital_codelv76)
• i SQLCODE0(vpok'"") k vpok q
• s vresult0 k vpok
• q
• 134 cross (more than one data item) validations
  

6

• Automatic Quality Control
• Each validation is a discrete, identifiable
  module of code in TCRIS, to aid maintenance.
• The same validations are applied automatically
  to all manual and electronic batch inputs.
• Incoming records that fail a validation, whether
  initially processed automatically or manually,
  must be corrected manually before they can be
  processed further.

7
Automatic Quality Assurance

• There are 60 automatic quality assurance checks
  at TCR.
• 43 checks are run each day across all
  registrations that have been created or had data
  added to them during the previous day.
• Thats approximately 700 to 1000 registrations
  as candidates for each check each day.
• Problems highlighted by these checks are
  examined and corrected manually.
• The daily approach keeps the job manageable.
• The remaining checks are run on demand.
• Examples
• Check that a new NHS number is unique within
  the database.
• Check for cases with an unusual address at
  diagnosis.
• Check for cases with inconsistent behaviour code
  and last morphology digit.
• Check for incomplete death records.

8
Automatic Quality Assurance
Types of quality assurance check
9

• The QA Group
• Set up in August 2001
• Membership is multidisciplinary
• Pam Acworth Senior Data Officer
• Heather Bourne Clinical Coding Specialist
• Neil Hanchett IT and QA Officer
• Jason Hiscox Database Administrator
• Remit includes
• Review of automatic validations
• Validations are time dependent
• Addition of new validations
• Review and refinement of the TCRIS automated
  collation processes
• Patient matching
• Tumour matching
• Registration update
• Data-level liaison with trusts and other
  suppliers, and new supplier and data set
  take-on...

10
Automatic Quality Control and Assurance at the
Thames Cancer Registry Neil Hanchett Thames
Cancer Registry 020 7378 7688 neil.hanchett_at_kcl.a
c.uk