Hemoglobin A1c (HbA1c) levels are used as a measure of glycemic control, with greater levels indicating poorer control and a greater risk of death. However, recent studies have found a U -shaped association between the HbA1c levels and mortality in patients with heart failure, with the lowest risk of death associated with elevated HbA1c levels, usually >7%. Cardiac surgery is frequently used to mitigate the signs and symptoms of heart failure. The purpose of the present study was to determine the association between HbA1c levels and late mortality in cardiac surgery patients with and without heart failure. Patients with and without New York Heart Association class III or IV heart failure were divided into quartiles according to the preoperative HbA1c level. Mortality was determined for each group and compared using chi-square tests and Cox modeling. Of the 311 patients with heart failure, 65 (21%) were dead at follow-up compared to 57 of 669 patients (9%) without heart failure (p <0.001). After adjusting for confounders, the patients without heart failure and with HbA1c ≤5.7% had the lowest risk of death. In patients with preoperative heart failure, we found a U -shaped association between HbA1c levels and late mortality, with those patients with HbA1c levels of 5.8% to 6.2% having the lowest risk of death. HbA1c levels ≤5.7% and ≥7.2% were associated with statistically significant greater risks of death. In conclusion, we found in patients with heart failure that the lowest risk of death was associated with HbA1c levels of 5.8% to 6.2%.
Heart failure (HF) due to ischemic coronary artery disease in both diabetic and nondiabetic patients is frequently treated with coronary artery bypass grafting (CABG) in an attempt to correct the ischemic cause of their HF or at least to prevent worsening of their HF. To date, we are not aware of any studies elucidating the role of preoperative glycemic control on the outcomes in patients with HF who underwent CABG. Therefore, the purpose of the present study was to characterize the relation between hemoglobin A1c (HbA1c) and mortality in patients with and without HF who underwent CABG.
Methods
The institutional review board of Mercy St. Vincent Medical Center approved the present study and waived informed consent. All patients underwent CABG surgery from August 1, 2007 to July 30, 2010, with the internal mammary artery as the primary graft vessel. Radial artery and saphenous vein grafts were used as clinically indicated. The goal was complete revascularization. Additionally, all patients were discharged with aspirin, a β blocker, an angiotensin-converting enzyme inhibitor, statin, and clopidogrel, unless contraindicated. Patients with elevated HbA1c were given diet counseling and activity and weight loss instructions. If necessary, their diabetic medications were adjusted, and follow-up for future care was arranged.
The measurement of HbA1c is part of the routine preoperative laboratory evaluation of all patients undergoing cardiac surgery under our care. All HbA1c levels were measured within 7 days before surgery, with 92% measured the morning of surgery. The laboratory warehouse with the HbA1c levels was combined with the institutional cardiac surgery database to identify patients undergoing CABG surgery with or without concomitant valvular surgery. The database is routinely updated using the Social Security Death Index and other sources to record the date of late mortalities. The present analysis reflects all deaths through December 1, 2011. To identify possible U – or J -shaped relations between mortality and HbA1c levels, we separated patients into quartiles according to their HbA1c level (<5.8%, 5.8% to 6.2%, 6.3% to 7.1%, and >7.1%). We defined HF as New York Heart Association class III or IV and stratified the analysis to compare patients with New York Heart Association class III or IV (HF-yes group) to patients with lesser New York Heart Association classes or no HF (HF-no group). Comparisons were made using Student’s t test, the chi-square test, or Fisher’s exact test, as appropriate, and p <0.05 was used to denote statistical significance. Cox proportional hazard models, with all variables listed in Table 1 , were used for multivariate analyses, with p <0.05 and confidence intervals that excluded 1 to denote statistical significance. Separate Cox models were constructed for all CABG patients (including patients with valvular or other surgical procedures) and for isolated CABG patients without concomitant procedures stratified into HF-yes and HF-no groups. Predictive discrimination of the Cox models is presented as Harrell’s C and its SD as the modified Kendall τ. Adjusted Cox survival curves were created using a method of direct adjustment of the observed conditional probability of survival at each event. All statistics were analyzed using SPSS, version 19.0 (IBM, Armonk, New York).
