A Risk Prediction Model for Recurrent Events in Chronic Coronary Heart Disease: The Heart and Soul Study - PowerPoint PPT Presentation

Loading...

PPT – A Risk Prediction Model for Recurrent Events in Chronic Coronary Heart Disease: The Heart and Soul Study PowerPoint presentation | free to download - id: 5ba1a1-NzFjN



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

A Risk Prediction Model for Recurrent Events in Chronic Coronary Heart Disease: The Heart and Soul Study

Description:

Title: Complications of STEMI Author: Information Systems Last modified by: Ivy Ku Created Date: 12/31/2010 5:54:25 PM Document presentation format – PowerPoint PPT presentation

Number of Views:488
Avg rating:3.0/5.0
Slides: 48
Provided by: informat405
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: A Risk Prediction Model for Recurrent Events in Chronic Coronary Heart Disease: The Heart and Soul Study


1
A Risk Prediction Model for Recurrent Events in
Chronic Coronary Heart Disease The Heart and
Soul Study
  • Ivy Ku, Eric Vittinghoff, Kirsten
    Bibbins-Domingo, Michael Shlipak, Mary Whooley
  • January 14, 2011

2
Background and Significance
  • 1 in 3 Americans live with cardiovascular disease
  • With advances in therapies, patients live longer
    with CHD
  • Prognosis varies widely
  • Risk stratification integral to patient
    management

3
Risk Prediction is Useful
4
Risk Prediction in Primary Prevention
  • 10-year risk of incident coronary heart disease
    (CHD)
  • Guides cholesterol and BP treatment in primary
    prevention

5
Risk Prediction in ACS
6
Predictors of worse outcomes in stable CHD
  • Biomarkers CRP, BNP, hs-troponin

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Risk Prediction in Stable CHD
  • Clinically useful, up to date, simple, integrated
    risk scores lacking
  • HERS, LIPID, Framingham severe limitations
  • Furthermore, long-term risk in CHD has not been
    well-characterized and quantified
  • CHF not included in CHD risk prediction

12
Project Aims
  • To develop a clinical prediction model and point
    score for 5-year risk of recurrent CV events in
    stable CHD
  • To quantify and categorize the range of long-term
    risk in stable CHD

13
Methods
  • The Heart and Soul Study
  • Cohort study of 1024 subjects with stable CHD
    enrolled 2000-02
  • Effect of psychosocial factors on prognosis in
    stable CHD
  • Thorough phenotyping of baseline condition,
    biomarkers, echo, stress
  • Mean 6 years follow-up, gt 400 CV events

14
Population
  • SF bay area
  • VA, UCSF, CHN clinics
  • Inclusion hx MI, revascularization, angiographic
    CAD, abnormal stress test
  • Exclusion MI within 6 mo, unable to walk 1
    block, moving away within 3 years

15
Methods
  • 2 Cox models
  • Dichotomized predictors
  • Continuous predictors
  • Composite outcome time to MI, CVA, CHF
    hospitalization, or CV death
  • Use baseline survival function, relative hazards
    to calculate 5-year risk

16
Coding of Predictors
  • Selected functional form of continuous predictors
    using AIC
  • categorical (quantiles, clinical cutpoints)
  • linear
  • 3, 4, 5 knot restricted cubic splines
  • Steyerberg recommends doing this a priori if
    possible, to avoid over-fitting
  • Cross-validation can also be used

17
Model selection
  • Need to maximize the signal without over-fitting
  • Three main strategies
  • Outcome-free data reduction use the literature,
    expert opinion, practical considerations to
    eliminate candidate predictors without looking at
    the outcome
  • Parsimony select highly significant predictors
  • Cross-validation (CV) mimics external validation

18
Our implementation
  • Outcome free data reduction eliminated 18 of 36
    candidate predictors on the basis of expert
    judgment, practical considerations
  • Parsimony cut 4 more using backward selection
  • Cross-validation 10-fold CV of C-index for
    1,000 candidate models
  • Final decision between top candidates again
    considered clinical convenience and face validity

19
How cross-validation works
  • Divide sample into 5-10 subsets
  • For each subset
  • set aside, fit model to remaining subsets
  • calculate predictions for set-aside subset
  • Estimate prediction error using quasi- external
    predictions for all observations
  • Repeat 20 times and average results
  • repetition needed to reduce noise

20
C-index
  • A measure of model discrimination
  • Extension of C-statistic, area under ROC curve to
    survival models
  • Estimates probability that in a randomly selected
    pair of observations, the earlier failure has the
    higher predicted risk
  • Naïve C-index is optimistic cross-validation
    reduces the optimism

21
Selecting Point Score Model
  • Cross-validation involves five steps for each
    candidate point score model
  • fit model using binary predictors only
  • round coefficients to obtain point scores
  • refit model using calculated point scores as sole
    (continuous) predictor
  • save predictions from the refitted model
  • use predictions to calculate CV C-index

22
Shrinkage using calibration slope
  • Cross-validation to get calibration slope
  • calculate xb for omitted subsets
  • re-fit model using xb as the sole predictor
  • coefficient for xb lt1.0 signals over-fitting
  • Use slope to improve calibration
  • shrink coefficients by calibration slope (i.e.,
    the coefficient for xb in the refitted model)
  • pulls in extreme high and low predictions
  • does not affect discrimination

23
Model Performance
  • Discrimination C-index
  • Net reclassification improvement (NRI)
  • continuous vs point score models
  • continuous model vs Framingham
  • Calibration goodness-of-fit test, visual
    inspection, calibration slope

24
External Model Validation
  • Cross-validation is strictly internal
  • reduces over-fitting
  • but does not protect against predictor effects
    that differ across populations
  • Plan external validation in separate cohort
  • recommended by Altman and Royston, often demanded
    by reviewers

25
Results
26
(No Transcript)
27
Included vs Excluded
Included (n912) Excluded (n108) P value
Outcome, 27 32 0.23
Follow-up time, yrs 5.8 (5.6-5.9) 5.6 (5.0-6.2) 0.47
Age, yrs 67 11 68 11 0.36
History of CHF 17 22 0.27
smoker 20 18 0.69
LVEF, 62 10 61 9 0.60
UACR, mg/g 8.7 (5.1-17.9) 7.9 (2.2-11.6) 0.18
BNP, pg/mL 173 (74-452) 222 (89-532) 0.20
28
Included vs Excluded
Included (n912) Included (n912) Excluded (n108) Excluded (n108) P value between HR
HR (CI) P HR (CI) P
Age 1.04 (1.03-1.06) lt 0.001 1.01 (0.98-1.05) 0.43 0.10
Hx CHF 2.27 (1.72-3.0) lt 0.001 2.3 (1.12-4.72) 0.02 0.90
smoker 1.17 (0.86-1.59) 0.32 1.66 (0.75-3.67) 0.21 0.39
LVEF 2.74 (2.03-3.71) lt 0.001 4.96 (2.15-11.47) lt 0.001 0.18
BNP 4.9 (3.81-6.31) lt 0.001 8.4 (3.45-20.45) lt 0.001 0.22
29
Functional Form
  • Determined by AIC
  • Age linear
  • LVEF dichotomized at 50
  • UACR, BNP, BMI, CRP 3-knot restricted cubic
    splines

30
Backward Selection
  • Eliminated 4 weakest predictors (pgt0.5)
  • HDL, LDL, hx MI, HTN
  • Top 4 predictors were always the same by all
    exploratory methods
  • Age, EF, BNP, UACR
  • Remaining 10 candidates
  • Gender, BMI, smoker, diabetes, CRP, CKD,
    troponin, hx CHF, med nonadherence, physical
    inactivity

31
Screening models using CV
  • Base model age, LVEF, BNP, UACR
  • Screened all 5 to 11-predictor models using 20
    repetitions of 10-fold cross-validated C-index
  • Targeting 5 to 7 predictor range, for
    practicality
  • Done for both point score and continuous models

32
Top Models
33
Final Model
  • Age, LVEF, BNP, UACR, smoker
  • Point score
  • Naïve C-index 0.742
  • CV C-index 0.736
  • Continuous model
  • Naïve C-index 0.768
  • CV C-index 0.763

34
Final Model with Dichotomized Predictors
35
Point score
  • Age 65 1
  • Smoker 1
  • LVEF lt 50 2
  • BNP gt 500 3
  • UACR 30 3

36
(No Transcript)
37
(No Transcript)
38
Continuous Model
39
Calibration Continuous Model
Pseudo-Hosmer-Lemeshow goodness-of-fit test p
0.94 Cross-validated calibration slope 0.94
40
Calibration with shrinkage
41
NRI with FHS model
Cases Cases Cases Cases Cases Cases
FHS Adding HS variables Adding HS variables Adding HS variables Adding HS variables Adding HS variables
FHS 0-10 10-20 20-50 50 Total
0-10 3 1 1 0 5
10-20 6 3 13 6 28
20-50 3 25 61 72 161
50 0 0 13 36 49
Total 12 32 88 114 243
93 cases moved up 47 cases moved down 46 net
cases 46 / 243 18.9, p lt 0.001 329
non-cases moved down 82 non-cases moved up 247
net non-cases 247 / 661 37.4 p lt 0.001 Net
reclassification 56.3, p lt 0.001
Non-Cases Non-Cases Non-Cases Non-Cases Non-Cases Non-Cases
FHS Adding HS variables Adding HS variables Adding HS variables Adding HS variables Adding HS variables
FHS 0-10 10-20 20-50 50 Total
0-10 25 5 2 0 32
10-20 109 49 25 5 188
20-50 33 166 164 45 408
50 0 4 17 12 33
Total 167 224 208 62 661
42
NRI comparing point to cont.
Cases Cases Cases Cases Cases Cases
Point score Continuous model Continuous model Continuous model Continuous model Continuous model
Point score 0-10 10-20 20-50 50 Total
0-10 8 2 1 0 11
10-20 12 35 32 1 80
20-50 0 8 52 18 78
50 0 0 9 68 77
Total 20 45 94 87 246
54 cases moved up 29 cases moved down 25 net
cases 25 / 246 10.2, p 0.006 153
non-cases moved down 94 non-cases moved up 59 net
non-cases 59 / 670 8.8 p 0.002 Net
reclassification 19, p lt 0.001
Non-Cases Non-Cases Non-Cases Non-Cases Non-Cases Non-Cases
Point score Continuous model Continuous model Continuous model Continuous model Continuous model
Point score 0-10 10-20 20-50 50 Total
0-10 146 24 1 0 171
10-20 126 157 56 0 339
20-50 3 18 100 13 134
50 0 0 6 20 26
Total 275 199 163 33 670
43
Summary of results
  • Our model had good discrimination (CV C-statistic
    0.76), and had 56 net reclassification vs
    framingham secondary events model
  • Many traditional risk factors (HTN, lipids,
    obesity) were not significant predictors

44
Limitations
  • Population (VA men, CHN, urban)
  • No external validation yet

45
Conclusion
  • Developed a risk model with 5 predictors
  • Can stratify 5-year recurrent CV event risk in
    stable CHD

46
External Validation
  • PEACE cohort
  • Clinical trial of trandolapril vs placebo in
    low-risk stable CAD
  • 3600 subjects with biomarkers
  • Patients were less sick, excluded EFlt40
  • 1996-2000

47
References
  • Steyerberg E. Clinical Prediction Models A
    practical approach to development, validation and
    updating. Springer, NY 2009.
  • Lloyd-Jones D. Cardiovascular risk prediction
    Basic concepts, current status, and future
    directions. Circ 2010 121 1768-77.
  • Morrow D. Cardiovascular risk prediction in
    patients with stable and unstable coronary heart
    disease. Circ 2010 121 2681-91.
  • DAgostino R. Primary and subsequent coronary
    risk appraisal new results from the Framingham
    study. AHJ 2000 139 272-81.
  • Altman DG, Royston P. What do we mean by
    validating a prognostic model? Stat Med,
    200019453-473.
About PowerShow.com