Cardiovascular Surgery Practice Patterns and Clinical Outcomes After Hybrid Coronary Revascularization in the United States An Analysis From the Society of Thoracic Surgeons Adult Cardiac Database Ralf E. Harskamp, MD; J. Matthew Brennan, MD, MPH; Ying Xian, MD, PhD; Michael E. Halkos, MD, MS; John D. Puskas, MD, MS; Vinod H. Thourani, MD; James S. Gammie, MD; Bradley S. Taylor, MD; Robbert J. de Winter, MD, PhD; Sunghee Kim, PhD; Sean O’Brien, PhD; Eric D. Peterson, MD, MPH; Jeffrey G. Gaea, MD Background— Hybrid coronary revascularization (HCR) involves a combination of surgical and percutaneous techniques, which in selected patients may present an alternative to conventional coronary artery bypass grafting (CABG).

Methods and Results— Patients were included who underwent HCR (staged/concurrent) or isolated CABG in the Society of Thoracic Surgeons Adult Cardiac Surgery Database (July 2011 to March 2013). HCR represented 0.48% (n=950; staged=809, concurrent=141) of the total CABG volume (n= 198 622) during the study period, and was performed in onethird of participating centers (n=361). Patients who underwent HCR had higher cardiovascular risk profiles in comparison with patients undergoing CABG. In comparison with CABG, median sternotomy (98.5% for CABG, 61.1% for staged HCR, and 52.5% for concurrent HCR), direct vision harvesting (98.9%, 66.0%, and 68.1%) and cardiopulmonary bypass (83.4%, 45%, and 36.9%) were less frequently used for staged and concurrent HCR, whereas robotic assistance (0.7%, 33.0%, and 30.5%) was more common. After adjustment, no differences were observed for the composite of in-hospital mortality and major morbidity (odds ratio, 0.93; 95% confidence interval, 0.75-1.16; P=0.53 for staged HCR, and odds ratio, 0.94; 95% confidence interval, 0.56-1.56; P=0.80 for concurrent HCR in comparison with CABG). There was no statistically significant association between operative mortality and either treatment group (odds ratio, 0.74; 95% confidence interval, 0.42-1.30; P=0.29 for staged HCR, and odds ratio, 2.26; 95% confidence interval, 0.99-5.17; P=0.053 for concurrent HCR in comparison with CABG). Conclusion— HCR, either as a staged or concurrent procedure, is performed in one-third of US hospitals and is reserved for a highly selected patient population. Although HCR may appear to be an equally safe alternative for CABG surgery, further randomized study is warranted. (Circulation. 2014;130:872-879.)

Key Words: coronary artery bypass ■ coronary disease ■ stents ■ thoracic surgery

ybrid coronary revascularization (HCR), which involves the combined use of percutaneous and surgical tech­ niques, has emerged as an alternative to conventional coro­ nary artery bypass grafting (CABG) for selected patients with multivessel coronary disease.1-3 In most cases, HCR involves a surgical procedure in which the internal mammary artery (IMA) is grafted to the left anterior descending (LAD) coronary artery, preceded or followed by percutaneous coro­ nary intervention (PCI) of non-LAD coronary lesions, either

performed in 1 setting (concurrent) or as a staged procedure.4 A prerequisite for HCR is that the patient has LAD anatomy that is eligible for surgical revascularization, but also has nonLAD lesions that are amenable for PCI. The concept of HCR stems from the hypothesis that (1) bypass grafting of the LAD with an IMA graft is superior to coronary stenting, and (2) PCI with the latest drug-eluting stents is equal or even superior to other bypass grafts used for non-LAD disease.5-6 Advances in surgical techniques also allowed IMA-to-LAD grafting to be performed by using less invasive techniques than conven­ tional CABG. Although several reports suggested a reduc­ tion in perioperative morbidity, length of intensive care and

H

Editorial see p 869 Clinical Perspective on p 879

Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz. Received February 13, 2014; accepted June 20, 2014. From the Duke Clinical Research Institute and Duke University Medical Center, Durham NC (R.E.H., J.M.B., Y.X., S.K., S.O'B., E.D.R, J.G.G.); Academic Medical Center of the University of Amsterdam, Amsterdam, Netherlands (R.E.H., R.J.d.W.); Cardiothoracic Surgery Clinical Research Unit, Division of Cardiothoracic Surgery, Emory University School of Medicine, Atlanta, GA (M.E.H., V.H.T); Department of Cardiothoracic Surgery, Mount Sinai Beth Israel, New York, NY (J.D.P.); and Heart Center of the University of Maryland Medical Center. Baltimore, MD (J.S.G., B.S.T.). Correspondence to Ralf E Harskamp, MD, Duke Clinical Research Institute, 2400 Pratt St, Durham, NC 27705. Email [email protected] © 2014 American Heart Association, Inc. Circulation is available at http://circ.ahajournaIs.org

DOI: 10.1161/CIRCULATIONAHA.114.009479

872

Harskamp et al hospital stay, with equal left IMA patency, these studies have generally been limited to a small number of high-volume cen­ ters with experienced operators.3,4,7 It is unknown the degree to which HCR is used in routine clinical practice in the United States, and the clinical characteristics or in-hospital outcomes of patients who undergo HCR, as well. To address this paucity of information, we performed a con­ temporary examination of HCR among isolated CABG pro­ cedures by using data from the Society of Thoracic Surgeons (STS) national registry. The objectives of this study were 3-fold: (1) to describe the incidence and variation of HCR, and HCR characteristics among US centers, as well; (2) to assess and compare clinical and operative characteristics between HCR and conventional CABG; (3) to describe and compare in-hospital outcomes between HCR and conventional CABG.

Methods Study Population From the STS adult cardiac database version 2.73 we included patients that were enrolled from July 2011 through March 2013. From this starting population, we included patients who underwent a hybrid procedure or isolated CABG, and applied a number of exclu­ sion criteria, which are listed in Figure 1.

Procedural Definitions and Outcomes In our study, HCR was considered a planned procedure that included the use of an IMA graft and coronary stent placement that could either be performed concurrently or as a 2-stage procedure performed dur­ ing the same hospitalization. Other variable definitions followed the standards of the Society of Thoracic Surgeons Adult Cardiac Surgery Database data version 2.73 (http://www.sts.org). The primary out­ come was in-hospital mortality or major morbidity, a composite of (peri)operative mortality, the need for reoperation (for bleeding/ tamponade, valvular dysfunction, graft occlusion, or other cardiac reasons), stroke, renal failure, mediastinitis, or prolonged ventilation. The secondary outcomes were the individual components of the com­ posite end point, and postoperative length of hospital stay.

( n = 2 5 1 , 1 2 2 / s ite s : 1 ,0 6 0 ) -

M is s in g d a ta o n u s e o f HCR ( n = 4 ,0 8 3 / s ite s = 8 )

-

P r io r c o n g e n it a l s u r g e r y (n = 2 4 4 , s ite s = 0 )

-

S in g le v e s s e l d is e a s e ( n = l l , 7 8 5 , s ite s = 0 )

-

PCI f a i lu r e o r c o m p lic a tio n ( n = 4 ,4 4 2 , s it e s = l)

-

S T E M I c a s e s ( n = 1 2 ,5 7 4 , s ite s = 0 )

-

M is s in g d a ta o n p la n n e d HC R ( n = 7 8 , s ite s = 0 )

-

U n p la n n e d HCR ( n = 4 4 3 , s ite s = 0 ) N o I M A u s e d ( n = l l , 1 8 6 , s ite s = 0 )

-

N o c o r o n a r y s t e n t p la c e d ( n = 6 0 1 , s ite s = 0 )

-

C a r d io g e n ic s h o c k ( n = l , 9 7 8 , s it e s = l)

-

E m e r g e n t/s a lv a g e p r o c e d u r e ( n = 4 ,5 5 9 , s ite s = 0 )

-

R e s u s c ita tio n ( n = 5 2 7 , s ite s = 0 )

S tu d y p o p u la t io n ( n = 1 9 8 , 6 2 2 / s i t e s : 1 ,0 5 0 )

C o n v e n tio n a l CABG (n=197,672 / sites; 1,050)

S taged HCR (n=809, sites: 322)

873

Statistical Analysis The incidence of HCR in the study population was presented as a median, interquartile range, minimum and maximum. Variation in the incidence among US hospitals was assessed by histograms, sorted from largest to smallest percentage of HCR use, to graphically dis­ play hospital variation. Temporal trends of hospital rates of HCR among the study population were calculated by 3-month intervals during the study period. Clinical and procedural characteristics were displayed in descrip­ tive tables, in which continuous variables are summarized as median and interquartile range, and categorical variables as percentage and frequency counts between parentheses. P values for comparing the distribution of variables among the comparison groups was computed with the x2 test for categorical variables and the Kruskal-Wallis test for continuous variables. In-hospital outcomes including the composite end point of mortal­ ity and major morbidity were presented as odds ratios (ORs) and 95% confidence intervals (CIs) and P values with adjusting for potential confounders using logistic regression with the generalized estimat­ ing equations (to account for the correlation in the same site) for the composite end point and for operative mortality. Potential confound­ ers were based on previously developed and validated STS CABG mortality model.8 The following variables were included in the adjustment model: CABG volume, age, race, sex, diabetes mellitus, hypertension, cere­ brovascular disease, peripheral vascular disease, dialysis, immuno­ compromised status, creatinine, chronic lung disease, body surface area, ejection fraction, congestive heart failure, unstable angina, previous myocardial infarction, recent myocardial infarction (