Congenital heart disease update

1

• Congenital heart disease update
• April 2005

2
Congenital heart disease

• All 13 centres in UK submitting data since April
  2000
• UK overall data and centre specific survival data
  for benchmark procedures (surgery cath)
  published in BMJ
• Editorial criticising power to detect poor
  performance

3
Problems with data

• Benchmark procedures
• Reasonable numbers but low risk procedures
• Need to include all procedures
• Still no validated risk adjustment for congenital
  heart disease
• Risk adjustment tools in development, both
  consensus based
• RACHS (US) not impressed, preliminary
  application to CCAD data suggests poor capability
  to detect poor outcomes across all complexities
• Aristotle (EACTS/SCTS/US) simplified and complex
  forms, simple score so far looking weak when used
  to predict real deaths in CCAD, complex form
  needs much larger dataset. Pilots ongoing.

4
Data validation visits

• Continuing, annual for each centre
• Lin Denne volunteer clinician
• Audit staff from other units
• Data quality good overall
• But no improvement in some centres
• Still poor data management support locally
• Getting worse with cuts this year
• Some centres still havent sent data for 03/04
• Target for 03/04 data publication July 05
• Deadline for submission of end of month
• What do we do if the data doesnt come??

5
Data validation visits

• Missing procedures
• 0-80 procedures
• Poor data collection by consultants
• Lack of interest
• Ancillary staff too busy, not capable of coding
• Nobody willing to take action locally
• Missing NHS numbers
• Cant track mortality
• Have to censor these patients from analysis
• Can give very misleading results
• Harley St 42 benchmark ops, 30d FU on 4, 1 dead,
  mortality 20!!
• Remove foreign nationals from analyses?
• Report on status at discharge instead?

6
Data publication

• 2000/01, 01/02 reports on website (02/03 any
  min), open access
• National results for bypass/non bypass,
  individual lesions
• Centre specific 30 day survival for benchmarks
  only, with comparative stats
• Next plan to include 1 year survival
• Suggestion to include centre specific results for
  all lesions but without comparative stats voted
  against at recent contributorsmeeting
• Including by CHF !!
• Website explanation of why data analysis is
  limited

7
Survival Newcombes scoring (99 CIs)

• Criticised in BMJ
• More support for funnel plots (Spiegelhalter,
  2004)

8
Plans

• Morbidity (PCPC score), new post op seizures in
  dataset for 1 year now
• Risk adjustment
• Look at our own 4 years data
• Await Aristotle pilot study result
• Treat with caution huge opportunities for gaming
• Centre specific data for all individual
  procedures ??
• Operator specific data for all procedures ??
• Improved adult congenital data collection
• HES data suggests many non specialists dabbling
• How to get all congenital procedures into CCAD?

9
Web and paper reports.

• CCAD credibility and support from contributors
  and public at risk if we dont provide more
  comprehensive data on line
• Risk of more trouble from Dr FFoster
• Time and expertise within CCAD scarce
• We need simple stats, survival curves, funnel
  plots
• VLADs to come soon?
• Farm out data analysis and presentation to
  Bruces team at UCL?

10
What to do if a centre looks dodgy?

• SCTS BCCA