The Framingham score

1
The Framingham score NZ Guideline adjustments
useful for CVD risk prediction in NZ?
Work in progress Preliminary results
J B Broad, R J Marshall, S Wells, A J Kerr, T
Riddell, R Jackson on behalf of HRC-Predict
Co-Investigators
2
NZ Guidelines 2003
Criteria to identify people for risk assessment
Framingham score
Apply modifications e.g. for certain ethnicities,
family history of very high BP or lipids
If HxCVD, classify as clinically at high risk
NZ risk score
NZ risk group
Management recommendations
3
Who should be treated?...NZ Guidelines 2003
4
Framingham CVD risk equation

• 5200 patients in Framingham, MA
• aged 35-74 years no prior CVD
• recruited in 1960s and 1970s
• followed for 12 years, min. 4 years
• outcomes MI, CVA, CHF, angina, CVD death
• equation gives CVD risk score in n years(also
  equations for CHD, stroke )
• risk factors age, sex, BP, DM, smoking,
  TChol/HDL ratio, ECG-confirmed LVH

5
NZ modifications to Framingham score

• for those of Maori, Pacific or Indian ethnicity
  5 upward adjustment
• if a family history (1st order relative Mlt55,
  Flt65) 5 upward adjustment
• for those very high BP or lipids AND with risk
  score lt15 treat as 15-20
• Others not covered in this presentation

6
Possible problems with risk stratification based
on Framingham score

• For risk-based prevention approach to work
  well, CVD risk prediction must be accurate
• very different US population
• different lifestyles
• based on dated data
• effect of medications

7
Research questions

• Do higher Framingham 5 year CVD risk scores
  predict future CVD events in NZ?
• Do NZ guideline modifications improve prediction
  accuracy?

8
Methods PREDICT-CVD system

• Using data from PREDICT-CVD
• Web based system, integrated with PMS
• Users are GPs practice nurses in primary care
• Provides CVD risk assessment management
• Patient risk assessments based on criteria in NZ
  CVD Guideline
• Data from 2 Auckland PHOs ProCARE, HealthWest
• Risk assessments from 2002 to January 2007

9
PREDICT-CVD MedTechrisk assessment template
Not real patient
10
Linkage of risk to outcomes
GP practice
Demog. risk profile
Risk score management advice
Univ. of Auckland Merge Predict data with NZHIS
outcomes
Predict server NHI Predict data
Encrypted NHI baseline data
eNHI outcomes data
NHI eNHI
NZHIS eNHI used to extract outcomes data
(hospital discharges deaths)
Secure systems, approved by MultiRegion Ethics
Committee
11
30,878 patients as at January 2007
No prior CVD Prior CVD n27,210
(78) n3,668 (12) Age, mean (SD) 54.5 (11.6)
64.8 (11.8) Men, 55.3 58.9 European other,
69.2 67.4 NZ Maori, 9.6 11.3 Pacific,
14.7 16.1 Indian, 3.1 2.9 Other Asian,
3.5 2.2 Type II Diabetes, 14.2 27.9
12
Framingham risk score in first 30,878 patients

13
Results diagnoses in first events, after 1.9
years mean follow-up

• No prior CVD Prior CVD
• n627(53) n561(47)
• Diagnosis
• IHD sudden death 334 303
• CHF 174 195
• Ischaemic stroke or TIA 141 82
• PAD 47 87
• Diagnostic groups not mutually exclusive

14
Results events by risk group, in first 30,878
patients
47
16
26
11
63 of events in 21 rated at high risk
15
Results est. 5-year incidence, by HxCVD
16
Results est. 5-year incidence if no HxCVD, by
age group
17
Results est. 5-year incidence if no HxCVD, by
ethnicity
Note scale change, lines fairly parallel, 3
higher
18
Results est. 5-year incidence if no HxCVD, by
family history
Plot withdrawn. Please contact the author at
j.broad_at_auckland.ac.nz for further information
19
Results est. 5-year incidence if no HxCVD, by
very high BP or lipids modification
Modifying for very high BP or lipids may be
unjustified, but based on only 42 events
20
Discussion

• Risk increases across Framingham groups, observed
  close to predicted, but
• we included pts aged over 75 years, some on BP or
  lipid treatment already
• we did not include ECG-confirmed LVH in score
• NZ-modified score provided to user, changed some
  interventions as a result
• our follow-up is very short, 1.9 years,
  extrapolated to 5 yr probability
• we were unable to include non-hospitalised events
  e.g. CHF, angina

21
Conclusions NZ Guideline modifications to
Framingham

• Justifies NZ Guidelines 5 adjustment for
  ethnicity, seems additive ok
• Modification for family history perhaps not
  additive, may need revision
• Too few events (i.e. little power) yet to
  evaluate modification for very high blood
  pressure or lipids

22
Conclusions Framingham score

• Framingham score HxCVD placed at high risk 63
  of those who had an event
• score less useful for those aged gt75yrs
• NZ needs a more discriminating score to find the
  37 of pts having events when at low risk /or
  more effort to lower population risk
• PREDICT will be able to inform guideline
  development derive scores for NZ

23
Investigators in HRC Project Grant

• Dr Sue Wells Dr Tania Riddell Co-Principal
  Investigators Public Health specialists
• Prof Rod Jackson, Professor of Epidemiology
• Joanna Broad, Epidemiologist
• Dr Andrew Kerr, Cardiologist
• Dr Dale Bramley Dr Sue Crengle
• Dr Tim Kenealy
• Assoc Prof Richard Milne
• Dr Diana North

Work in Progress Preliminary results
24
Acknowledgements

• Many many GPs, practice nurses clinical
  medical nursing specialists
• Primary Healthcare Organisations esp.
  ProCare HealthWest
• Enigma Publishing
• New Zealand Guidelines Group
• National Cardiovascular Advisory Group
• Maori Cardiovascular Group
• Ministry of Health Clinical Services Directorate
• National Heart Foundation
• Diabetes NZ
• Counties Manukau District Health Board esp.
  Chronic Care Management programme, Middlemore
  Hospital Coronary Care Unit, Diabetes Clinic
• MedTech Global Ltd
• Health Research Council funding

HealthWEST
25
(No Transcript)