Body mass index and postoperative mortality in patients undergoing coronary artery bypass graft surgery plus valve replacement: a retrospective cohort study

Study design and patient population

This retrospective cohort study assessed the relationship between BMI and postoperative death during mid-term observation in patients undergoing valve replacement plus CABG. The target independent variable was the BMI in the preoperative baseline data, and the dependent variable was postoperative death. We included all patients who underwent CABG combined with valve replacement surgery corresponding to the International Classification of Disease, 9th Revision, Clinical Modification of Operations and Procedures (ICD-9-CM) 36.1 (bypass anastomosis for heart revascularization) in combination with 35.2 (replacement of heart valve) between January 2017 and April 2021. All eligible patients from The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine Electronic Information Center, were screened by a research nurse and a doctor of clinical medicine.

This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of the University of Science and Technology of China and was conducted in accordance with the Declaration of Helsinki (approval number: 2021-RE-046, approval date: 24/05/2020). All participants provided written informed consent. All data were anonymized to protect the patients’ privacy. All research was conducted in accordance with relevant guidelines and regulations.

The inclusion criteria for this study were as follows: (1) Patients with a degree of stenosis greater than 70% and the need to undergo coronary artery bypass surgery to diagnose coronary heart disease and (2) availability of the patient’s complete medical record data. The exclusion criteria were as follows: (1) Patients who underwent coronary artery bypass surgery but did not have valves with moderate or higher valvular stenosis or insufficiency and did not undergo simultaneous valve replacement. (2) Patients undergoing tricuspid valve replacement. (3) Uncontrolled bleeding during the operation, death during the operation, or rapid death due to bleeding after the operation, such as in delayed left ventricular rupture after the operation.

Definitions, endpoints, and follow-up

Baseline characteristics, including comorbidities, were obtained from Medicine Electronic Information Center, which includes ICD codes for all patients. The primary endpoint was all-cause mortality after surgery and follow-up. Vital status and date of death were telephonically acquired by a follow-up nurse until 31 February 2022. For patients who could not be reached, the household registration management department of the local police station was contacted by a doctor of clinical medicine to determine whether the patient was still alive.

The height and weight of each patient were measured upon admission. BMI was calculated as BMI = weight (kg) ÷ height² (m²) and treated as a continuous variable. For the definition of postoperative death, we used the definition by the American Thoracic Society (ATS), that is, death during the same hospital stay after surgery or within 30 days postoperatively at any time and regardless of location (Shahian et al., 2009). In this study, we included the following data: (1) demographic data; (2) variables that may affect BMI or postoperative mortality according to existing literature; and (3) variables that may affect postoperative mortality based on our clinical experience. The following variables were used to construct a calibration model: age, sex, surgical history, pulmonary arterial hypertension, smoking status, diabetes mellitus, left ventricular ejection fraction (EF), B-type natriuretic peptide (BNP) level, extracorporeal circulation time (pump time), red blood cell (RBC) count, operation time, chronic renal disease, and a history of cerebrovascular disease. Each adjusted variable was changed to a categorical variable based on clinical guidance significance, previous literature, or the balance of data between groups.

Treatment protocol

All patients underwent standardized surgical procedures. A brief description of the procedure is as follows: for patients undergoing mitral valve replacement plus CABG, the heart continued to function normally after establishment of cardiopulmonary bypass. Subsequently, the aorta was blocked, myocardial protective fluid was infused, and the mitral valve was replaced. For patients undergoing simple aortic valve replacement and double valve replacement, anastomosis of the distal coronary artery was first performed. The aorta was then incised, and myocardial protection solution was infused through the left and right coronary sinuses and bridging vessels, after which valve replacement was subsequently performed. After valve replacement, the proximal anastomosis of the bridging vessel was completed, and the heart resumed beating. All patients returned to the Cardiac Surgery Intensive Care Unit after the procedure for monitoring and treatment.

Statistical analysis

Continuous variables with a normal distribution are presented as mean ± standard deviation, and categorical variables are expressed as frequencies or as percentages. The chi-squared or Fisher’s exact tests were used to assess categorical variables. Multiple logistic regression models will be used to explore the outcomes of interest and adjusted for statistically significant variables in univariate analyses. We performed nonlinear regression analysis to fit the clinical results with 18 and 25 as thresholds, and used a two-stage linear regression model and logarithmic likelihood ratio to test the 30-day mortality and threshold effect. All analyses were performed with the statistical software packages in R (R Core Team, 2020) and Empower Stats (http://www.empowerstats.com; X&Y Solutions, Inc, Boston, MA, USA). P < 0.05 (two-sided) was deemed statistically significant.

Characteristics of the study population

A total of 1,976 patients met the inclusion criteria, and 202 patients were selected for final analysis based on the exclusion criteria (Fig. 1). The baseline characteristics of the patients according to their BMI values are listed in Table 1. We grouped the participants as follows: BMI ≥25 kg/m²; BMI ≥18, <25 kg/m² and BMI <18 kg/m² based on the results of nonlinearity. The difference between the groups was statistically significant (P < 0.01). There were no statistical differences between the groups in terms of sex, surgical history, pulmonary arterial hypertension, smoking, diabetes mellitus, EF, BNP, extracorporeal circulation time (pump time), RBC count, operation time, chronic renal disease, or history of cerebrovascular disease.

Univariate analysis

The results of univariate analyses are listed in Table 2. The results of univariate analysis showed that BMI was significantly associated with postoperative mortality (p < 0.05). Without adjusting for other variables, per univariate analysis, male patients, those aged >60 years, smokers, those with diabetes mellitus, those with an EF of ≥55%, and those with longer pump time tended to have a higher risk of postoperative mortality but did not reach significance level (Table 2).

Results of crude and adjusted models

To investigate the relationship between BMI and mortality, a multivariate analysis was conducted. We built different models by adjusting for different confounding factors. Model I adjusted for the baseline information of patients, including age, sex, and smoking, and Model II added all confounding factors, including age, sex, smoking, diabetes mellitus, extracorporeal circulation time (pump time), history of cerebrovascular disease, chronic renal disease, surgical history, operation time, pulmonary arterial hypertension, EF, and RBC count. The results are shown in Table 3. In the unadjusted model (crude model), patients with BMI ≥25 kg/m² had a 4.8-times higher risk of death than patients with BMI ≥18, <25 kg/m². In the fully adjusted model, patients with BMI ≥25 kg/m² had an 11.5-times higher risk of death than patients with BMI <25 kg/m².

Results of nonlinearity

We analyzed the relationship between BMI and 30-day mortality after smooth curve fitting (Fig. 2). The results showed that before the inflection point, the 30-day mortality rate of patients decreased with an increase in BMI, and after the inflection point was reached, the 30-day mortality rate increased with an increase in BMI. Using threshold effect analysis, the inflection points of the 30-day mortality after surgery were determined to be 18 and 25 kg/m² (Table 4). The log-likelihood ratio test value was less than 0.001.