Background

1. Outcomes for Patients with ST-Elevation Myocardial Infarction in Hospitals With and Without Onsite Coronary Artery Bypass Graft Surgery: The New York State Experience Hannan EL, Zhong Y, Racz M, Jacobs AK, Walford G, Cozzens K, Holmes DR, Jones RH, Hibberd M, Doran D, Whalen D, King SB III. Circulation: Cardiovascular Interventions, published ahead of print 11/10/09.

2. Background • The benefit of primary percutaneous coronary interventions (P-PCI) for patients with ST-elevation myocardial infarction (STEMI) has been well-documented. • However, controversy still exists as to whether PCI should be expanded to hospitals without coronary artery bypass graft (CABG) surgery. • Elective PCI at hospitals without onsite CABG surgery is currently not recommended in the most recent ACC/AHA guidelines

3. Purpose of Study • To compare long- and short-term outcomes for STEMI patients in NY hospitals without CABG surgery backup (P-PCI centers) to those in hospitals with backup (called Full Service (FS) centers).

4. Methods • Patients and Outcomes: STEMI Patients Undergoing PCI • Observational study of 1735 patients who were discharged after PCI for STEMI between 1/1/03 and 12/31/06 in P-PCI centers were propensity-matched to 8817 patients in full service (FS) centers to obtain 1729 pairs of patients with very similar propensity scores. • These patients were followed through the end of 2006 and outcomes of patients treated in P-PCI centers were compared with outcomes of patients in FS centers.

5. Methods, Cont’d • Patients and Outcomes: STEMI Patients Undergoing PCI • Outcomes that were examined included in-hospital mortality and need for CABG surgery, and three-year mortality, repeat target vessel PCI and subsequent revascularization.

6. Methods, Cont’d • Patients and Outcomes: STEMI Patients Not Undergoing PCI • Patients: Observational study comparing all STEMI patients presenting at P-PCI centers not undergoing PCI (34.3% of all pts.) and all STEMI patients presenting at FS centers not undergoing PCI (30.3% of all pts.)

7. Methods, Cont’d • Patients and Outcomes: STEMI Patients Not Undergoing PCI • Outcome: The outcome used was risk-adjusted in-hospital mortality using the variables used in the CMS AMI reports for the risk-adjustment process (no out-of-hospital mortality was available for these pts).

8. Results • Variables used in the propensity matching included: • Age, sex, race • Ejection fraction, congestive heart failure, previous AMI, shock, hemodynamic instability • Several comorbidities (renal failure, COPD, diabetes, carotid/cerebrovascular disease, peripheral vascular disease, ventricular arrhythmia • Previous revascularization • Anatomic group (no. of vessels diseased and presence absence of proximal LAD disease)

9. Results • Prior to propensity matching, patients inFS centers were sicker (lower ejection fractions, higher comorbidity rates, more likely to be hemodynamically unstable or in shock, more likely to have congestive heart failure). • After propensity matching, there were no differences based on an examination of % standardized differences in risk factor prevalences.

10. Results: Short-Term Outcomes for Pts with PCI • For patients undergoing PCI, there were no differences for in-hospital/30-day mortality (2.3% for P-PCI centers vs. 1.9% for FS centers (P=0.40)), emergency CABG surgery immediately following PCI (0.06% vs. 0.35%, P=0.06). • However, P-PCI centers had a lower same/next day CABG rate (0.23% vs. 0.69%, P=0.046).

11. Results: Short-Term Outcomes for Pts without PCI • A higher percentage of STEMI patients arriving at P-PCI centers did not undergo PCI (34.3% vs. 30.3%) • These patients had significantly higher mortality rates (28.5% vs. 22.3%, adjusted OR =1.38, 95% CI (1.10, 1.77)). • After removing the patients who died within 2 hours of arriving at the hospital, P-PCI center patients still had higher mortality rates (adjusted OR= 1.39, 95% CI (1.10, 1.77)).

12. Results: Longer-Term Outcomes for Pts with PCI • For patients undergoing PCI, there were no differences in three-year mortality (6.8% vs. 6.9%, P=0.63) or subsequent revascularization (23.4% vs. 20.7%, P=0.11). • However, P-PCI centers had higher repeat target vessel PCI rates (12.4% vs. 9.2%, P=0.0005).

13. Mortality Rates and P-values for Matched STEMI Patients Undergoing Primary PCI in P-PCI Centers and Full Service Centers: New York, Jan. 2003 - Dec. 2006

14. Caveats • Operators in New York P-PCI centers were required to perform at least 200 PCIs in the past 3 years, 75 PCIs per year and 11 P-PCIs per year on a regular basis. Also, P-PCI centers were required to ensure 24/7/365 coverage for P-PCI, maintain a volume of 36 P-PCIs per year, and maintain an active affiliation with a high-volume FS center. Hence, P-PCI center results in less restrictive settings could be worse. • The study is observational and therefore subject to possible selection bias. However, propensity matching was used to adjust for baseline differences. Also, it would be impractical to conduct an RCT for this type of study given that many patients are transported by ambulance.

15. Caveats, Cont’d • It is possible thatP-PCI centers were at a disadvantage because they were in a startup period during which teams were becoming accustomed to one another and to a new procedure being performed in the hospital. • To test for this bias, we repeated the analyses after excluding the first year of operation, and found that the results were essentially the same (the significant differences remained).

16. Conclusions • Summary: No differences between P-PCI centers and FS centers were found in in-hospital/30 day mortality, the need for emergency surgery, three-year mortality or subsequent revascularization • Summary: However, P-PCI centers had higher repeat TV PCI rates and higher mortality rates for patients not undergoing PCI. • Conclusion: P-PCI centers should be monitored closely, including the monitoring of STEMI patients who do not undergo PCI.