Vascular Risk Prediction

1
Vascular Risk Prediction
Profiling tools for subjects at risk for a
first, unheralded vascular disease An
update from the Taskforce on Vascular Risk
Prediction
2
Vascular Risk Prediction
Principle Goal of the Taskforce
Improve risk prediction in originally healthy
people, in whom a high risk derived from the
major independent cardiovascular risk factors is
not evident
Michel Romanens, 05.05.2007
3
Vascular Risk Prediction
Guidelines 2005 ....
AGLA / Executive Committee SGK Secondary
prevention Intensification Primary
Prevention Discuss Guidelines of ESC and IAS
Try to reach common guidelines in the next
years
Michel Romanens, 05.05.2007
4
Vascular Risk Prediction
3 Questions
IAS-AGLA or EAS-SCORE or FRAMINGHAM as a first
measure ? PLAQUE or ISCHEMIA or hsCRP to further
stratify risk ? BAYES or ROC or ODDS to estimate
the value of a test ?
Michel Romanens, 05.05.2007
5
Vascular Risk Prediction
3 levels of risk
PROCAM CORDICARE I Germany OLTEN
COHORT Men Only Men and Women Age 35-65 Age
45 N5389 N914 RISK Distribution
Events Distribution Events High 8 33
2 ? Intermediate 15 33 11
? Low 77 33 87 ?
Michel Romanens, 05.05.2007
6
Vascular Risk Prediction
3 possible strategies
A down-calibrate definition of high risk from 20
to 15? general practitionars might be even
more confused than they are already
now B treat virtually everybodies cholesterol in
intermediate risk C further risk stratify with
new tools in intermediate risk
Michel Romanens, 05.05.2007
7
Vascular Risk Prediction
IAS-AGLA or EAS-SCORE / Strategy B
CORIDCARE I Indication for LDL or total
Cholesterol lowering intervention
Michel Romanens, 05.05.2007
8
Vascular Risk Prediction
IAS-AGLA or EAS-SCORE
Strategy B treat them all in intermediate risk
defined by IAS-AGLA IAS-AGLA Intermediate Risk
CORDICARE I 11 of 914 subjects gt 45 years 10
year risk for AMI is about 15 on average Year
2005 410251 subjects in CH gt 40 years at
intermediate risk Of these, 91 would be treated
with statins (CORDICARE I) 61538 AMI would occur
in 10 years (30769 deaths) Statins lower risk by
30 Risk would be reduced from 15 to 11 with
statins Absolute Risk reduction 16800 AMI (8400
deaths) Relative Risk reduction 27 NNT 24
Michel Romanens, 05.05.2007
9
Vascular Risk Prediction
IAS-AGLA or EAS-SCORE
Strategy C further risk stratify intermediate
risk defined by IAS-AGLA Ca-Scoring
? Ultrasound Plaque Imaging ? hsCRP?
Michel Romanens, 05.05.2007
10
Vascular Risk Prediction
Strategies to assess the value of a test
A ROC analysis B ODD ratios, relative risk,
correction for co-factors C Bayes Posttest
Calculator
Michel Romanens, 05.05.2007 /
11
Vascular Risk Prediction
Strategies to assess the value of a test
Michel Romanens, 05.05.2007 /
12
Vascular Risk Prediction
ROC, ODD, or BAYES
Nancy Cook statements Indeed, if improvement
in the ROC statistic was used as the criterion
for model inclusion, then neither LDL-C, HDL-C,
nor total cholesterol would have been included in
risk models after accounting for age, blood
pressure, and smoking. The Bayes information
is related to the posterior probability that the
model is correct, and is a conservative criterion
for model selection.
Michel Romanens, 05.05.2007 /
13
Vascular Risk Prediction
ROC, ODD, or BAYES
Single Risk Factor Testing Significant overlap
between diseased and non diseased
Michel Romanens, 05.05.2007 / Wald NJ, Law MR.
BMJ. 20033261419
14
Vascular Risk Prediction
IAS-AGLA or EAS-SCORE
Strategy C further risk stratify intermediate
risk defined by IAS-AGLA relative risk for
various risk factors in middle aged subjects
Hypertension
LDLgt4.1 Men
LDLgt4.1 Women
Nicotine Men
Nicotine Women
Diabetes Men
Diabetes Women
ARIC Study
Michel Romanens, 05.05.2007 / Hypertension.
200648392-396.
15
Vascular Risk Prediction
Bayes Posttest Risk as a third solution
between ROC and ODDs ?
Look for practicle solutions on www.scopri.ch
SE Sensitivity SP Specificity PV
prevalence or pretest probability PTP posttest
probability PTP pos PV SE / PV SE (1
PV) (1 SP) PTP neg PV (1 SE) / PV (1
SE) SP (1 PV)
Michel Romanens, 05.05.2007
16
Vascular Risk Prediction
Bayes Posttest Risk with 95 CI
Using Coronary calcium percentiles 95 CI by
Newcombe
Michel Romanens, 05.05.2007
17
Vascular Risk Prediction
Bayes Posttest Risk with 95 CI
Michel Romanens, 05.05.2007
18
Vascular Risk Prediction
Bayes Posttest Risk with 95 CI
http//scopri.ch/posttestcalculators1.html
Michel Romanens, 05.05.2007
19
Vascular Risk Prediction
Future themes of the taskforce www.taskforce.athe
rosclerosisimaging.ch
Which emerging imaging test should be ideally
used in Switzerland? Are sequential
posttest-calculations an issue in risk prediction
? Who should do atherosclerosis imaging tests
(accreditation)? Does atherosclerosis imaging
convince patients more than verbal communication,
that a treatment is necessary? For which test
should there be reimbursement and how can that be
achieved? What is the cost-efficiency issues of
emerging imaging tests? What is the risk-benefit
ratio of calcium scoring in intermediate risk
subjects in view of the radiation burden per scan
(1 mSv). ..
Michel Romanens, 05.05.2007
20
Vascular Risk Prediction
An Update of the Taskforce
Taskforce Founding Members in 2005 Darioli
Roger Romanens Michel Scientific
Coordinator Romanens Michel
Michel Romanens, 06.05.2006
21
Vascular Risk Prediction
Summary and Future Directions
Proposition in God we trust, the others must
bring numbers Let us do the COCA study (5 year
outcome of coronary calcifications and carotid
total plaque area from a real life experience)
It is designed to produce numbers on outcome,
since practice based algorithms may be more
useful than population wide epidemiological
testing in the future
Michel Romanens, 05.05.2007
22
Vascular Risk Prediction
Summary and Future Directions
Goal of the taskforce improve risk prediction in
originally healthy subjects, and several
publications are underway . IAS-AGLA or
EAS-SCORE rather IAS-AGLA Calcium scoring
passes the hardliner test (ROC analysis) Bayes
Posttest Risk Calculators are useful Many themes
are open to debate within the Taskforce Check on
www.taskforce.atherosclerosisimaging.ch
Michel Romanens, 05.05.2007
23
Vascular Risk Prediction
The presentation is open for discussion
Michel Romanens, 05.05.2007