Contemporary Predictors of Procedural Mortality Among Patients Undergoing PCI: Results From National Cardiovascular Data Registry (NCDR) Eric D Peterson, David Dai, Elizabeth R DeLong, Sunil V Rao, Matthew T Roe, Kalon K L Ho, Mandeep Singh, John S Rumsfeld, Richard E Shaw, William S Weintraub, Ralph G Brindis, John A Spertus on behalf of the NCDR Registry Participants
19
Embed
Contemporary Predictors of Procedural Mortality Among Patients Undergoing PCI: Results From National Cardiovascular Data Registry (NCDR) Eric D Peterson,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Contemporary Predictors of Procedural Mortality Among Patients Undergoing
PCI: Results From National Cardiovascular
Data Registry (NCDR)
Eric D Peterson, David Dai, Elizabeth R DeLong, Sunil V Rao,Matthew T Roe, Kalon K L Ho, Mandeep Singh, John S Rumsfeld, Richard E Shaw, William S Weintraub, Ralph G Brindis, John A
Spertus on behalf of the NCDR Registry Participants
Disclosure Statement
Research support provided by the American College of Cardiology National Cardiovascular Data Registry
Background
• Prior models have estimated risks of percutaneous coronary intervention (PCI) mortality. Yet, these have limitations:– Representing older patient populations– Selected centers or regions– Limited number of procedures– Incomplete clinical or angiographic detail– Failed to develop user-friendly formats
Objectives
We propose to use the ACC-NCDR PCI database,the nation’s largest interventional data repository to:
• Develop a contemporary PCI mortality risk model for those receiving primary PCI and elective procedures
• Internally validate the NCDR PCI model– overall and among selected sub-populations
• Determine the incremental prognostic value of angiographic factors beyond pre-cath risk factors
• Develop user-friendly PCI risk score for bedside risk estimation
Methods: Populations
• Development Sample – 302,958 PCI at 470 participating NCDR centers– Jan, 2004 – March, 2006 (v3 data only)– Divided into development (60%) and 1st validation
(40%)
• 2nd Validation Sample– 285,440 PCI at 608 participating NCDR centers– April, 2006 – April, 2007 (v3 data only)
• Exclusions– Not first PCI (57%)– Transfer out (1%)– Missing 2+ candidate variables (<1%)
Methods (2)
• Model Development – Multivariate logistic regression– Backward selection from 34 candidate variables
including demographics, risk factors, cardiac status, cath lab visit, and PCI procedures
• Model Validation– Model discrimination measured using c-index
(ranging from 0.5 for chance discrimination to 1.0 for perfect prediction)
– Model calibration examined by comparing average observed and predicted values within risk groups
Patient Clinical Characteristics (1)
Development (181,775)
1st validation (121,183)
2nd validation (285,440)
Age 63.9±12.1 63.9±12.1 64.1±12.1
Female 33.4% 33.3% 33.3%
Caucasian 87.2% 87.1% 85.6%
BMI (kg/m2) 29.6±6.3 29.7±6.3 29.8±6.3
Prior MI (>7days)
29.1% 29.1% 27.3%
Prior CHF 10.1% 10.0% 9.9%
Diabetes
– Non-insulin 21.5% 21.7% 22.3%
– Insulin 10.0% 10.0% 10.3%
Patient Clinical Characteristics (2)
Development (181,775)
1st validation (121,183)
2nd validation (285,440)
Mean GFR* (+/-SD) 73.6±30.5 73.5±29.0 73.2±28.1
Dialysis Dependent 1.6% 1.5% 1.5%
Cerebral Vascular Disease
10.9% 11.1% 11.1%
Peripheral Vascular Disease
11.7% 11.7% 11.9%
Chronic Lung Disease 16.0% 16.0% 15.8%
Prior PCI 35.1% 35.4% 36.6%
NYHA Class IV 18.3% 18.3% 18.8%
Cardiogenic Shock 1.9% 1.8% 1.7%* Glomerular Filtration Rate (GRF) assessed by MDRD formula
Procedural CharacteristicsDevelopment (181,775)
1st validation (121,183)
2nd validation (285,440)
LVEF 52.7±13 52.7±13 52.7±13
PCI Status
- Elective 49.3% 49.3% 50.2%
- Urgent 36.1% 35.6% 34.7%
- Emergency 14.4% 14.5% 15.0%
- Salvage 0.2% 0.2% 0.2%
Highest Risk Lesion
-pLAD 18.2% 18.2% 18.2%
-Left Main 1.7% 1.8% 1.8%
-Total Occlusion
11.0% 10.7% 14.9%
Multivessel PCI 14.0% 13.9% 14.1%
In Hospital Mortality
1.2%
4.8%
0.7%
1.3%
4.8%
0.7%
1.2%
4.7%
0.6%
0%
1%
2%
3%
4%
5%
6%
Overall STEMI Other
Mo
rta
lity
Ra
te
Develop Sample 1st Validation 2nd Validation
Full and Pre-Cath Risk Models
Full Model † Precath Simple Model
Label O.R. 95% CI Wald Chi-Sq O.R. 95% CI Wald
Chi-Sq
Cardiogenic Shock 8.35 7.40 9.44 1168.
3 12.19 10.86
13.68
1804.7
GFR for STEMI ‡ 0.77 0.74 0.80 181.9 0.77 0.75 0.78 377.6
† Full model includes: Prior PCI, PreOp IABP, Ejection Fraction, Coronary Lesion >= 50%: Subacute Thrombosis, Total pre-procedure Occlusion, Diabetes treatment, SCAI Lesion Class 2 or 3, BMI for STEMI/non STEMI, Prior Dialysis for STEMI/non STEMI , Highest risk status for non-STEMI, NYHA Class IV for NSTEMI‡ Per 10 unit increase. § Versus Elective
PCI Risk Score System
Age <60 ≥60,<70 ≥70,<80 ≥80
0 4 8 14Cardiogenic Shock No Yes
0 25Prior CHF No Yes
0 5PVD No Yes
0 5CLD No Yes
0 4GFR <30 30-60 60-90 >90
18 10 6 0NYHA Class 4 No Yes
0 4PCI Status (STEMI) Elective Urgent Emergent Salvage
12 15 20 38PCI Status (Other) Elective Urgent Emergent Salvage
• Voluntary participation • Limited auditing of data source• Only internal validation of model• No core lab angiographic readings• No data on functional status• Outcomes limited in in-hospital
mortality
Conclusions
• We developed a highly predictive contemporary mortality model for PCI
• Angiographic variables add modestly to risk assessment
• Model accurate in – Overall, <75 >75, women vs men, DM vs not, – STEMI vs other – Low vs high risk groups