Did blood transfusion increase mortality in patients with diabetes undergoing isolated coronary artery bypass graft surgery? A propensity score-matched analysis of 816 patients.

BACKGROUND: The aim of this study was to compare clinical outcomes of blood transfusion in patients with diabetes mellitus undergoing isolated on-pump coronary artery bypass grafting.
METHODS: The medical records of a total of 1,912 patients (1,300 males, 612 females; mean age 60.7±10.0) with diabetes who underwent isolated on-pump coronary artery bypass grafting between January 1999 and June 2019 were retrospectively analyzed. The patients were divided into two groups as patients with and without blood transfusions. The mortality rates were compared between the two groups.
RESULTS: The mortality rate was 14 times higher in the patients receiving blood transfusion (odds ratio: 14.80; 95% confidence interval 5.05 to 43.34; p<0.001). However, in the multivariate logistic regression analysis, there were no statistically significant difference in mortality between the patient groups, when diabetes mellitus was a covariate factor (Odds ratio: 8.34; 95% confidence interval 3.94 to 17.66 vs. odds ratio 8.36; 95% confidence interval 3.95 to 17.70).
CONCLUSION: The propensity score-matched analysis of patients with diabetes showed that clinical outcomes were more severely affected by blood transfusion.

Introduction

Diabetes mellitus increased the risk of cardiovascular disease with insulin resistance, inflammation and endothelial dysfunction.[1] The risk of cardiovascular diseases is 1.6 to 2.6 times higher in patients with diabetes, particularly among younger age and in women.[2] After cardiac surgery, the incidences of morbidity and mortality in these patients are higher than those without diabetes.[3,4]

Blood transfusion are associated with postoperative wound infections, pneumonia, renal dysfunction, multiple organ failure, and increased hospital stay.[5] Additionally, perioperative blood transfusions are known to be associated with an increased morbidity and mortality after cardiac surgery.[6,7]

In the present study, we aimed to compare clinical outcomes of blood transfusion in patients with diabetes mellitus who underwent isolated on-pump coronary artery bypass grafting (CABG).

Patients and Methods

Between January 1999 and June 2019, a total of 6,148 patients underwent isolated on-pump CABG in Department of Cardiovascular Surgery of Acıbadem Mehmet Ali Aydınlar University. Of these, 1,912 patients (1,300 males, 612 females; mean age 60.7±10.0 year) with diabetes were included in this retrospective study. Patients who underwent off-pump CABG or concomitant surgery were excluded. Patients who had revision for bleeding were also excluded to eliminate the negative outcomes caused by bleeding with hemodynamic instability or blood transfusions. Study flow chart is shown in Figure 1. The medical records were retrieved from the electronic registry and hospital database. A written informed consent was obtained from each patient. The study protocol was approved by the Institutional Review Board of Acıbadem Mehmet Ali Aydınlar University, ATADEK. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Figure 1

Study flowchart.

The patients were divided into two groups as those with (n=515, 26.94%) and without (n=1,397, 73.06%) blood transfusions. Demographic data, EuroSCORE scores, ejection fraction (EF), hematocrit, creatinine levels, use of medication, and the presence of comorbidities were analyzed. The durations of cross-clamp (CC) and cardiopulmonary bypass (CPB) during surgery, durations of endotracheal intubation and intensive care unit (ICU) stay, and morbidity, mortality rates were evaluated.

All operations were performed by a single surgical team. All patients received a balanced anesthesia using clinical protocols. Standard CPB was established and antegrade cold blood cardioplegia was used for myocardial protection. Blood transfusion was not driven by a numerical trigger value alone, but by a restrictive red blood cell (RBC) transfusion policy based on hematocrit levels and hemodynamic parameters. If the hematocrit value was below 17% during the hypothermic period of CPB and below 20% after CPB, RBCs were transfused. The patients receiving at least one unit of RBC were included in the blood transfusion group. All patients were transferred to the ICU after surgery.

Statistical analysis Statistical analysis was performed using the SPSS version 10 software (SPSS Inc., Chicago, IL, USA). Data were presented in mean ± standard deviation (SD), median (min-max) or number and percentage. Univariate comparisons were made using the chi-square (χ2) test or Fisher"s exact test for categorical variables, and the t-test was used for continuous variables. Transfused patients were matched with non-transfused patients using the propensity score matching analysis to control for the imbalance between the groups. The propensity score was estimated using a regression model. Variables with a p v alue o f <0.1 w ere e ntered into the logistic regression analysis. A multinomial logistic regression analysis was used to examine the relationships between diabetes mellitus, blood transfusion, and mortality. A two-sided p value of <0.05 was considered statistically significant.

Results

Baseline demographic and clinical characteristics of the patients are presented in Table 1. There was no statistically significant difference between the groups in terms of age, sex, body mass index, and preand intraoperative values after the propensity score matching.

Table 1

Baseline demographic and clinical characteristics of patients with and without blood transfusion

			Before propensity matching							After propensity matching
	No transfusion (n=1,397)			Blood transfusion (n=515)				No transfusion (n=408)			Blood transfusion (n=408)
	n	%	Mean±SD	n	%	Mean±SD	p	n	%	Mean±SD	n	%	Mean±SD	p
Age (year)			61.8±8.9			64.3±8.3	<0.001			64.6±8.8			64.1±8.3	0.430
Body mass index (kg/m2)			29.7±9.8			29.7±24.1	0.980			30.5±15.0			30.0±27.0	0.770
Sex							<0.001							0.720
Female	378			234				177			182
Male	1019			281				231			226
EuroSCORE logistics (%)			3.6±5.5			5.4±8.6	<0.001			4.7±6.4			4.4±5.3	0.390
NYHA Class 3 and 4 (%)	192	13.7		111	19.6		0.004	75	18.4		70	17.2		0.710
Previous cardiac surgery (%)	19	1.3		22	4.2		<0.001	12	2.9		10	2.5		0.670
EF <30 (%)	35	2.5		25	4.9		<0.001	17	4.2		15	3.7		0.850
Hypertension (%)	1022	73.2		379	73.6		0.840	317	77.7		299	73.3		0.140
Hypercholesterolemia (%)	871	62.3		334	64.9		0.330	254	62.3		274	67.2		0.160
COPD (%)	95	6.8		60	11.7		0.001	45	11.0		44	10.8		1.000
Smoking (%)
Smokers	305	21.8		77	15		< 0.001	86	21.1		60	14.7		0.050
Former smokers	583	41.7		189	36.7			140	34.3		154	37.7
Never smoked	509	36.4		259	48.3			182	44.6		194	47.5
Medications
Beta blockers (%)	751	53.8		288	55.9		0.390	225	55.1		235	57.6		0.520
Calcium canal blockers (%)	283	20.3		101	19.6		0.650	95	23.3		91	22.3		0.800
ACE inhibitors (%)	470	33.6		173	33.6		0.980	140	34.3		139	34.1		1.000
Aspirin (%)	848	60.7		292	56.7		0.115	222	54.4		233	57.1		0.480
Clopidogrel (%)	31	2.2		6	1.2		0.130	7	1.7		4	1.0		0.540
Elective surgery (%)	1270	90.9		457	88.7		0.070	372	91.2		371	90.9		1.000
Preoperative hematocrit level (%)			40.3±4.5			38.0±4.7	< 0.001			38.3±4.8			38.3±4.4	0.940
Preoperative creatinine level (mg/dL)			0.9±0.4			1.0±0.5	< 0.001			0.9±0.4			1.0±0.5	0.300
CC time (min)			39.8±16.2			40.2±16.9	0.680			38.9±16.2			40.3±17.7	0.260
CPB time (min)			68.9±24.8			71.4±29.0	0.060			67.0±23.7			70.1±27.2	0.080
Number of distal anastomoses			3.3±1.0			3.4±0.9	0.040			3.3±1.0			3.5±0.9	0.003
SD: Standard deviation; NYHA: New York Heart Association; EF: Ejection fraction; COPD: Chronic obstructive pulmonary disease; ACE: Angiotensin-converting enzyme; CC: Cross-clamp; CPB: Cardiopulmonary bypass.

Before propensity score matching, the length of ICU stay, infection rates (odds ratio [OR]: 3.22; 95% confidence interval [CI]: 1.95 to 5.32; p<0.001), hospital stay, and mortality (OR: 14.80; 95% CI: 5.05 to 43.34; p<0.001) were significantly higher in the blood transfusion group (Table 2). After propensity score matching, the length of ICU stay was significantly higher in the blood transfusion group than the non-transfusion group; however, there was no significant difference in the duration of hospitalization between the two groups. Additionally, the infection rates were higher in patients who received blood transfusions than those who did not receive blood transfusions (OR: 2.35; 95% CI: 1.14 to 4.85; p=0.017).

Table 2

Postoperative data of patients with and without blood transfusion

			Before propensity matching				p			After propensity matching				p
	No transfusion (n=1,397)			Blood transfusion (n=515)				No transfusion (n=408)			Blood transfusion (n=408)
	n	%	Mean±SD	n	%	Mean±SD		n	%	Mean±SD	n	%	Mean±SD
Intubation time (h)Chest tube output (mL) Blood transfusion unit ICU duration (h)New onset stroke (%)			7.0±13.8  495.8±213.90.0±0.0  21.6±13.1			11.7±43.0  742.5±545.42.2±2.2  38.4±58.3	<0.001 <0.001 <0.001 <0.001			8.2±22.2  462.4±201.70.0±0.0  22.5±15.1			9.3±33.5  659.8±372.31.9±1.6  31.2±54.7	0.610 <0.001 <0.001 0.002
Transient	7	0.5		5	1.0		0.030	2	0.5		3	0.7		0.540
Permanent	0	0.0		2	0.4			0	0.0		1	0.2
Postoperative AF (%)	199	14.2		115	22.3		<0.001	67	16.4		84	20.6		0.140
New onset dialysis (%)
Transient	2	0.1		6	1.2		0.001	2	0.5		2	0.5		0.600
Permanent	1	0.1		3	0.6			0	0.0		1	0.2
Pulmonary complications (%)	5	0.3		5	0.9		0.130	3	0.7		3	0.7		0.220
Sternal dehiscence (%)	16	1.1		5	1.0		1.000	6	1.5		4	1.0		0.750
Overall infections (%)	30	2.1		34	6.6		<0.001	11	2.7		25	6.1		0.020
Vasoactive agent infusion >4 h (%)	143	10.2		103	20.0		<0.001	51	12.5		68	16.7		0.280
Discharge hematocrit level (%)			28.3±4.0			27.0±3.5	<0.001			27.8±3.9			27.0±3.4	<0.001
Discharge creatinine level (mg/dL)			0.9±0.4			1.1±0.8	<0.001			0.9±0.5			1.0±0.6	0.210
Hospital duration (days)			6.4±4.6			7.4±6.8	<0.001			6.7±4.7			7.0±4.8	0.320
ICU readmission (%)	32	2.3		31	6.0		<0.001	11	2.7		23	5.6		0.030
Reintubation (%)	11	0.8		20	3.9		<0.001	3	0.7		12	2.9		0.019
Hospital readmission (%)	45	3.2		26	5.0		0.070	13	3.2		20	4.9		0.210
Mortality (%)	4	0.3		21	4.1		<0.001	2	0.5		8	2.0		0.050
SD: Standard deviation; ICU: Intensive care unit; AF: Atrial fibrillation.

Furthermore, the mortality rates were four times higher in the patients who received blood transfusions than those who did not receive blood transfusions (OR: 4.06; 95% CI: 0.85 to 19.23; p=0.05). However, the multivariate logistic regression analysis revealed no statistically difference in mortality between the groups, when diabetes mellitus was a covariate factor (OR: 8.34; 95% CI: 3.94 to 17.66 vs. OR: 8.36; 95% CI: 3.95 to 17.70).

Discussion

The higher incidence of mortality in patients with diabetes after cardiac surgery was reported in previous studies.[8-11] In a meta-analysis, Zhang et al.[11] found that the incidence of morbidity and mortality was higher in patients with diabetes than those without after CABG.

Blood transfusions have been shown to be associated with a high incidence of infection, transfusion-related lung injury, pneumonia, sternal infections, leg wound infections, circulatory overload, low cardiac output syndrome, renal dysfunction, atrial fibrillation, stroke, and short and long-term mortality rates.[12-14] The independent risk factors for blood transfusions are age, female sex, low body surface area, low EF (<35%), emergency operation, anemia, redo cardiac surgery, use of extracorporeal circulation, prolonged bypass time, and re-exploration for any reason.[15,16] Blood transfusions are also associated with worse survival and increased risk factors, leading to prolonged hospital stay after cardiac surgery.[17-19]

Hemorheological alterations of the storage of RBCs may disturb the microcirculation.[20] Additionally, endothelial dysfunction causes microvascular complications and disturbs the microcirculation in patients with diabetes.[21] A randomized-controlled trial showed that correcting CPB-induced dilutional anemia with blood transfusions in patients with diabetes undergoing CABG increased the risk of renal injury due to the microcirculatory derangements caused by the transfusions.[22] Additionally, this patient population is at an increased risk of renal injury compared to those without diabetes due to possible end-organ damages.[9] Although blood transfusions and diabetes mellitus are associated with a high incidence of renal injury, in our study, there were no statistically significant differences in discharge creatinine levels or the number of new-onset dialysis between the patients with diabetes who received blood transfusion and who did not.

In their study, Vranken et al.[23] reported that the rate of infections increased in female patients, smokers, and patients with advanced age, diabetes mellitus, obesity, chronic obstructive pulmonary disease, low EF, prolonged CPB time, and perioperative administration of inotropes after cardiac surgery. They also reported that the number of blood transfusions was associated with infections in patients after cardiac surgery. Blood transfusions are a predominant factor for all types of postoperative infections. Likosky et al.[24] observed that the incidence of pneumonia increased for every unit of transfused RBCs after cardiac surgery. Additionally, the impaired chemotaxis and phagocytosis of neutrophils increase the risk of infections in patients with diabetes.[11] In our study, there were no significant differences in pulmonary complications, such as pneumonia, between the groups; however, infections were higher in the patients receiving blood transfusion.

A multi-center study in Europe reported that at least one unit of RBCs and a transfusion rate of 40.2% was recorded in patients who underwent CABG surgery.[8] In our study, a blood transfusion rate of 25.7% was recorded in all isolated CABG patients. Additionally, the length of ICU stay, chest-tube drainage, postoperative atrial fibrillation, new-onset dialysis and stroke, hospital readmission, ICU readmission, and mortality rates in the patients who received blood transfusions were higher than those who did not in our study.

In addition, in the current study, the patients were matched for propensity scores using perioperative risk assessments. Based on propensity score-matched analyses, the ICU stay, ICU readmission, reintubation, and mortality rates were statistically higher in the patients with diabetes who received blood transfusions compared to the others.

One of the strengths of the study was that the patient sample was from a single center, and all operations were done by the same surgeons and anesthesiologists using similar practices for blood transfusion. However, the main limitation includes its retrospective design.

In conclusion, propensity score-matched analysis of the patients with diabetes shows worse clinical outcomes due to blood transfusion, although mortality remains comparable between those who receive blood transfusion and those who do not. Nonetheless, further, large-scale, prospective studies are needed to confirm these findings.