All patients who underwent isolated, on-pump CABG at the Queen Elizabeth II (QEII) Health Sciences Center from 1995 until 2003, in Halifax, Canada, were identified using the Maritime Heart Center (MHC) database. From this group, only those patients who underwent CABG using TAG or LIMA+SVG were included for consideration in the final analysis. Patients were excluded from the final analysis if they underwent single-vessel bypass surgery or if they required an emergent or emergent salvage procedure. In both groups only patients with a LIMA to the left anterior descending artery were included.
All interventions were performed via a midline sternotomy, and cardiopulmonary bypass was instituted in a standardized manner for all cases. Briefly, body temperature during the procedure was allowed to drift to 32°C. Intermittent cold cardioplegia solution was delivered antegrade via the aortic root unless otherwise indicated. Arterial conduits were harvested with minimal trauma (LIMA and RIMA were not skeletonized) and were treated with either a papaverine solution or a nitroglycerine/calcium channel blocker (verapamil) solution prior to their use.
TAG was defined as the use of any arterial conduit (LIMA, RIMA, RA or GEPA), either alone or in combination, without concomitant use of SVG. The choice of conduit and the manner in which the grafts were constructed, including whether or not grafts were constructed in a composite T- or Y- fashion, proximal aorto-coronary anastomoses were performed or sequential anastomoses were used, was based entirely on surgeon preference rather than on any fixed criteria such as territory to graft or degree of target vessel stenosis. LIMA+SVG, on the other hand, was defined as any case in which the LIMA was used for the purposes of a single bypass to the left anterior coronary artery (LAD) and the remaining bypasses were carried out using SVG. In these cases, SVG were constructed as either a series of sequential anastomoses or as single bypasses and were anastomosed proximally to the aorta.
All patients received intravenous nitroglycerine infusions for the first 24 hours upon return from the operating room unless hypotensive (systolic blood pressure < 90 mm Hg). Oral nifedipine (Adalat 10 mg by mouth 4 times a day or Adalat extended release (XL) 20 to 30 mg by mouth daily; Bayer Inc., Toronto, Ontario, Canada) was prescribed for all patients receiving a radial artery beginning on day 1 post-operatively for a period of 3 to 6 months. Other routine post-operative medications included daily aspirin as well as resumption of cholesterol lowering agents, β-blockers and angiotensin converting enzyme inhibitors as appropriate.
The MHC database captures detailed information on a wide range of pre-operative, intra-operative, and in-hospital post-operative variables including post-operative complications and in-hospital mortality for all patients undergoing cardiac surgery at the QEII Health Sciences Center in Nova Scotia, Canada. In order to gather information regarding long-term outcomes, the MHC database was linked to the Canadian Institute for Health Information (CIHI) Discharge Abstract Database and the Nova Scotia Vital Statistics database. The CIHI Discharge Abstract Database is a national database that contains extensive data for each inpatient and outpatient hospital visit in Nova Scotia and enables us to track all readmissions to hospital. The Nova Scotia Vital Statistics database collects information on all births and deaths occurring within the province of Nova Scotia.
Pre-operative variables of interest included age (age ≥ 70 vs. age < 70), gender, body mass index or BMI (BMI > 25 kg/m2 vs. BMI = 25 kg/m2), smoking history, diabetes, hypercholesterolemia, renal insufficiency (pre-operative serum creatinine of ≥ 176 μmol/L), hypertension, peripheral and/or cerebrovascular disease, left ventricular ejection fraction or EF (EF < 40% vs. EF ≥ 40%), recent myocardial infarction (MI) defined as the occurrence of an MI in the 21 days prior to surgery, pre-operative inotropes/intra-aortic balloon pump (IABP) use, New York Heart Association (NYHA) functional classification (NYHA class IV vs. NYHA classes I – III), urgency status (operation performed within 24 hours from the time of referral vs. operation performed at a time interval of greater than 24 hours from the time of referral), prior percutaneous coronary intervention (PCI) and number of diseased vessels (triple-vessel or left main disease vs. single- or double-vessel disease). As mentioned earlier, patients undergoing emergent or emergent salvage procedures were excluded. Intra-operative variables of interest included number of distal anastomoses, cross clamp time and total bypass time. The medium-term outcomes of interest included all-cause mortality following discharge from hospital, readmission to hospital for any cardiac cause as defined by the following codes from the ninth revision of the International Classification of Disease, Clinical Modification (ICD-9-CM) (20): 410 (acute myocardial infarction), 411 (unstable angina), 412 (old myocardial infarction), 413 (angina pectoris), 414 (other forms of chronic IHD), 426 (conduction disorders), 427 (cardiac dysrhythmias), 428 (heart failure), 429 (ill-defined descriptions and complications of heart disease), coronary re-intervention (PCI or CABG) and a composite outcome defined as all-cause mortality following discharge from hospital or readmission to hospital for any cardiac cause.
Univariate comparisons between cases performed using TAG and cases performed using LIMA+SVG were carried out based on pre-, intra-, and post-operative variables including rates of in-hospital mortality and long-term adverse events using Chi-square tests for dichotomous variables and two-tailed t-tests for continuous variables. Unadjusted rates of the long-term composite outcome were then compared between TAG and LIMA+SVG cases using Kaplan Meier survival plots and the log-rank test. A fully-adjusted Cox proportional hazard model was created to determine the effect of TAG on the long-term outcome of interest after adjusting for differences between patients in clinical presentation. Pre-operative variables regardless of whether or not they differed between TAG and LIMA+SVG were included in the model building process, and a non-parsimonious model was created that retained all variables. The following variables were used in the model: age groups, COPD, gender, diabetes, preop renal insufficiency, cerebrovascular disease, peripheral vascular disease, EF groups, congestive heart failure, urgency status, number of diseased vessels, group assignment (TAG vs LIMA+SVG) and propensity analysis score (p score). A propensity score was calculated for each patient using the predicted probability of being in the TAG group. A multinomial logistic regression model was used to predict the probability of TAG group assignment (p score) after adjusting for all relevant preoperative patient characteristics including era (1995–2000 vs 2000–2005). Statistical significance was set at p < 0.05. All statistical analyses were performed using the SAS software package version 8.2 (SAS, Cary, North Carolina).