Validation of a Predictive Scoring System for Deep Sternal Wound Infection after Bilateral Internal Thoracic Artery Grafting in a Cohort of French Patients

Abstract

Background:

The Gatti score is a weighted scoring system based on risk factors for deep sternal wound infection (DSWI) that was created in an Italian center to predict DSWI risk after bilateral internal thoracic artery (BITA) grafting. No external evaluation based on validation samples derived from other surgical centers has been performed. The aim of this study is to perform this validation.

Patients and Methods:

During 2015, BITA grafts were used as skeletonized conduits in all 255 consecutive patients with multi-vessel coronary disease who underwent isolated coronary bypass surgery at the Department of Thoracic and Cardio-Vascular Surgery, University Hospital Jean Minjoz, Besançon, France. Baseline characteristics, operative data, and immediate outcomes of every patient were collected prospectively. A DSWI risk score was assigned to each patient pre-operatively. The discrimination power of both models, pre-operative and combined, of the Gatti score was assessed with the calculation of the area under the receiver operating characteristic curve.

Results:

Fourteen (5.5%) patients had DSWI. Major differences both as the baseline characteristics of patients and surgical techniques were found between this series and the original series from which the Gatti score was derived. The area under the receiver operating characteristic curve was 0.78 (95% confidence interval: 0.64–0.92) for the pre-operative model and 0.84 (95% confidence interval: 0.69–0.98) for the combined model.

Conclusions:

The Gatti score has proven to be effective even in a cohort of French patients despite major differences from the original Italian series. Multi-center validation studies must be performed before introducing the score into clinical practice

Despite long-term survival benefits [1,2], the increased risk of sternal complications, primarily surgical site infection, prevents widespread use of bilateral internal thoracic artery (BITA) grafting for myocardial re-vascularization [3,4]. The Gatti score is the first scoring system created specifically to predict deep sternal wound infection (DSWI) risk after BITA grafting [5]. It arises from a risk factor analysis performed on a series of nearly 3,000 consecutive patients (the original series) who underwent coronary artery bypass grafting (CABG) using skeletonized BITA grafts in an Italian institution where BITA grafting is performed on a routine basis (more than 70% of cases of isolated CABG). The Gatti score outperformed the existing scoring systems for sternal wound infection after median sternotomy [5 –7]. A simulation study to investigate the accuracy of bootstrap estimates of optimism (internal validation) has been performed successfully [5]. In addition, at Gatti's institution, the Gatti score has proven to be effective in predicting DSWI after BITA grafting even in a new prospective series of patients in whom the score had been calculated pre-operatively (external validation) [8].

To date, there has been no evaluation of the precision of performance estimates of the Gatti score in independent validation samples derived from other surgical centers. The aim of the present study was to perform this second external validation by testing the score in a new prospective series of CABG patients operated on in a French cardiac surgical unit with great experience in the use of BITA grafting (the French series).

Patients and Methods

Between January 1 and December 31, 2015, 255 consecutive patients with multi-vessel coronary disease underwent isolated CABG at the Department of Thoracic and Cardio-Vascular Surgery of the University Hospital Jean Minjoz of Besançon, France. Bilateral internal thoracic artery grafts alone were used to perform myocardial re-vascularization in all patients (Table 1).

Table 1.

Pre-Operative Patient Characteristics and Risk Profiles: The Original versus the French Series

Characteristics	Original series (n = 2,936)^a	French series (n = 255)	p
Age, years	66.3 ± 9	69.7 ± 10	<0.0001
70–80	1054 (35.9)	104 (40.8)	0.12
>80	130 (4.4)	40 (15.7)	<0.0001
Female gender	460 (15.7)	52 (20.4)	0.049
Hypertension	2,109 (71.8)	192 (75.3)	0.24
Former smoker	654 (22.3)	53 (20.8)	0.58
Current smoker	112 (3.8)	52 (20.4)	<0.0001
Hyperlipidemia	2,620 (89.2)	110 (43.1)	<0.0001
BMI, kg/m²	27.2 ± 3.6	27.3 ± 4.9	0.68
>30	586 (20)	60 (23.5)	0.17
Diabetes mellitus on oral hypoglycemic agents	652 (22.2)	73 (28.6)	0.019
Diabetes mellitus on insulin	213 (7.3)	24 (9.4)	0.16
Poor glycemic control	137 (4.7)^b	71 (27.8)^c	<0.0001
Poor mobility^d	14 (0.5)	24 (9.4)	<0.0001
Chronic lung disease^d	134 (4.6)	10 (3.9)	0.64
GFR, mL/min^e	78.5 ± 27.7	71 ± 23.4	<0.0001
50–85^d	1,515 (51.6)	155 (60.8)	0.0049
<50^d	365 (12.4)	43 (16.9)	0.042
Long-term dialysis	35 (1.2)	3 (1.2)	1
Extracardiac arteriopathy^d	192 (6.5)	47 (18.4)	<0.0001
Congestive heart failure	135 (4.6)	10 (3.9)	0.62
Unstable angina	860 (29.3)	85 (33.3)	0.17
Recent myocardial infarction^d	366 (12.5)	14 (5.5)	<0.0001
Left main disease	1,062 (36.2)	63 (24.7)	0.0002
Coronary artery disease			<0.0001
One-vessel	14 (0.5)	4 (1.6)
Two-vessel	375 (12.8)	79 (31)
Three-vessel	2,547 (86.8)	172 (67.5)
LVEF, %	55.2 ± 10.3	—	—
30–50^d	710 (24.2)	39 (15.3)	0.0013
<30^d	85 (2.9)	19 (7.5)	<0.0001
Previous cardiac operation	31 (1.1)	1 (0.4)	0.51
Critical state^d	161 (5.5)	1 (0.4)	<0.0001
Urgent surgical priority^d	1,750 (59.6)	84 (32.9)	<0.0001
Emergency^d	53 (1.8)	6 (2.4)	0.47
Expected operative risk (by EuroSCORE II), %^d	3.6 ± 5.1	2.6 ± 3	0.0021

Gatti et al. [5].

Defined as basal serum glucose >200 mg/dL at three consecutive measurements.

Defined as glycated hemoglobin ≥6.5%.

Nashef et al. [9].

The creatinine clearance rate, calculated according to the Cockcroft-Gault formula, was used for approximating the GFR.

BMI = body mass index; EuroSCORE = European System for Cardiac Operative Risk Evaluation; GFR = glomerular filtration rate; LVEF = left ventricular ejection fraction.

Unless otherwise stated, definitions of pre-operative clinical variables were those used for the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) [9]. Poor pre-operative glycemic control was defined as glycated hemoglobin 6.5% or greater [8]. The risk profile for each patient was calculated according to EuroSCORE II [9]. Post-operative myocardial infarction was defined according to the definition criteria for type V myocardial infarction by Moussa et al. [10]. Acute kidney injury was defined according to the Kidney Disease: Improving Global Outcomes (KIDGO) criteria [11]. The U.S. Centers for Disease Control and Prevention classification of the surgical site infections was adopted to define sternal wound infection [12]. For the purposes of this study, deep incisional infection and mediastinitis were considered to be DSWI.

All peri-operative data were recorded prospectively for every patient in a computerized data registry. For each patient, the score of the pre-operative and the combined model of the Gatti score [5] were calculated immediately before operation and recorded in a dedicated file; the score of the combined model was completed at hospital discharge. Post-discharge surveillance of the surgical sites was performed in a dedicated surgical outpatient unit. All patients with a surgical site complication were referred to this outpatient unit, at any time after hospital discharge. Approval to conduct the study was obtained from the Hospital Ethics Committee; the need for patients to provide their individual written consent was waived.

Surgery

Surgery was carried out via a median sternotomy either with cardiopulmonary bypass, with or without cross-clamping the aorta or off-pump technique. When a period of myocardial ischemia was used, myocardial protection was usually achieved with multi-dose tepid (28°C) blood cardioplegia delivered in antegrade mode (Tables 2 and 3) [13 –17].

Table 2.

Operative Data: The original versus the French Series

Characteristics	Original series (n = 2,936)^a	French series (n = 255)	p
Number of coronary anastomoses	3.8 ± 1.1	3 ± 1	<0.0001
With BITA graft	2.8 ± 0.8	3 ± 1	0.0002
Use of saphenous vein grafts	2,085 (71)	0	<0.0001
Surgical technique			<0.0001
Off-pump	168 (5.7)	65 (25.5)
Beating heart on-pump	54 (1.8)	8 (3.1)
On-pump	2,714 (92.4)	182 (71.4)
Aortic cross-clamp time (min)^b	79.3 ± 26.2	61.3 ± 21	<0.0001
CPB time (min)^c	101 ± 34.9	72.9 ± 24.2	<0.0001

Gatti et al. [5].

Calculated in the patients who underwent aortic cross-clamping.

Calculated in the patients who underwent CPB.

BITA = bilateral internal thoracic artery; CPB = cardiopulmonary bypass.

Table 3.

Measures Taken to Reduce the Risk of Sternal Wound Infection: The Original versus the French Series

Measure	Original series (n = 2,936)^a	French series (n = 255)
Study period	1999–2013, 15 y	2015, 12 mo
Selection criteria of the patients	All patients with multi-vessel coronary disease who required left-sided myocardial re-vascularization were candidates for BITA grafting; the sole exceptions being the cases in which one or both ITAs were unsuitable as coronary grafts, or when there was an unexpected operative finding of severe cardiac dysfunction or when rapid worsening of hemodynamics because of ischemia required immediate institution of CPB. Actually, there have been even some cases where a second ITA graft was harvested during CPB	All patients with multi-vessel coronary disease had BITA grafting; the sole exceptions being the cases in which one or both ITAs were unsuitable as coronary grafts, Actually, there have been even some cases in which a second ITA graft was harvested during CPB
Rate of BITA use	2,936/4,160 patients, 70.6%	255/255 patients, 100%
Use of prophylactic antibiotics	A first-generation cephalosporin (cefazolin) was chosen. Vancomycin was used if there was a severe allergy to β-lactam antibiotics, or in the event of mediastinal re-exploration; in the last case, the addition of an aminoglycoside was considered	A second-generation cephalosporin (cefuroxime) was chosen. Vancomycin was used if there was a severe allergy to β-lactam antibiotics. In the event of mediastinal re-exploration, the same protocol was used
Skin preparation	Careful skin preparation was performed with alcohol iodine solution. Chlorhexidine-alcohol was used only for patients with iodine allergy. A microbial sealant that immobilizes bacteria was adopted	A careful skin preparation was performed with povidone-iodine. Chlorhexidine-alcohol was used only for patients with iodine allergy. A microbial sealant that immobilizes bacteria was adopted
ITA harvesting technique	Both ITAs were harvested as skeletonized conduits with low-intensity cautery and bipolar coagulation forceps, extending distally just to include either the superior epigastric or the musculophrenic artery	Both ITAs were harvested as skeletonized conduits with low-intensity cautery; the collaterals were divided between vascular clips. Distal bifurcation was spared, bilaterally
Use of off-pump and beating heart on-pump techniques	Off-pump and beating heart on-pump techniques were adopted only in the presence of a diffusely atherosclerotic ascending aorta (by intra-operative EAS^b)	The use either of off-pump or on-pump technique depended on the surgeon's choice. Some surgeons did exclusively off-pump and others did exclusively on-pump, except for diffusely atherosclerotic ascending aorta, where off-pump (or beating heart on-pump) technique was used
Sternal closure	Standard single-loop sternal wiring technique was used as a sternal closure method until 2009. Since 2010, the Erdinc double-loop sternal wiring technique^c was adopted systematically. Bone wax was forbidden.^d Skin staples were used	All patients were closed by the Sutherland sternal wiring technique.^e Bone wax was used depending on the surgeon's preference.^d A subcuticular suture was used
Management of hyperglycemia	All patients with diabetes mellitus were treated during operation and then in ICU with a continuous intravenous insulin infusion in order to maintain serum glucose <180–200 mg/dL^f	All patients with diabetes mellitus were treated during operation and then in ICU with a continuous intravenous insulin infusion in order to maintain serum glucose <180–200 mg/dL^f
Other	Both ITAs were used as in situ grafts when possible (based on the double-source concept). The right ITA was preferentially directed to the left anterior descending coronary artery, and the left ITA to the posterolateral cardiac wall. Sometimes, the right ITA was taken down and used as a free graft from either the in situ left ITA (Y-graft) or (rarely) the proximal (aortic) end of a saphenous vein graft. The ante-aortic crossover right ITA bypass graft was protected by means of a pedicled flap taken from the thymic remnants.^g Additional coronary bypasses, usually for the right coronary artery, were performed with saphenous vein grafts	In every patient the right ITA was taken down and used as the second branch of a Y-graft with the in situ left ITA. The BITA grafting alone technique was used

Gatti et al. [5].

Zingone et al. [13].

Erdinc et al. [14].

Neither platelet gel nor topical antibiotics were used.

Sutherland et al. [15].

Furnary et al. [16].

Gatti et al. [17].

BITA = bilateral internal thoracic artery; CPB = cardiopulmonary bypass; EAS = epiaortic ultrasonography scan; ICU = intensive care unit; ITA = internal thoracic artery

Statistical methods

The data were expressed as the number of patients with the percentage in brackets, or the mean ± standard deviation. Pre-operative clinical patient characteristics, operative data, and post-operative complications were compared using the χ² or the Fisher exact test for dichotomous variables and the Student t test or the Mann-Whitney U test for continuous variables, as appropriate. For both models of the Gatti score, the goodness-of-fit was assessed with the Hosmer-Lemeshow test; the discrimination power was assessed with the receiver operating characteristic (ROC) curve and the calculation of the area under the receiver operating characteristic curve (AUCROC). According to arbitrary guidelines, the accuracy of prediction was defined as low (AUCROC: 0.5–0.7), moderate (AUC: 0.7–0.9), and high (AUCROC: 0.9–1) [18]. The pre-operative and the combined model were compared using the method of DeLong et al. [19]. A p value <0.05 was considered significant. For both models of the Gatti score, the Youden index was measured and a list of threshold values (criteria) corresponding with the coordinates of the ROC curves, with associated sensitivity and specificity was created. For equal sensitivity/specificity the threshold values with the highest specificity/sensibility were given. Finally, the data of the sample were used to create a large set (1,000 iterations) of new bootstrap samples, and the bias-corrected and accelerated bootstrap adjusted for possible bias and skewness in the bootstrap distribution [20]. The expected risk of DSWI according to the Gatti score (low: <10%; moderate: 10%–20%; high: >20%) was compared with the actual rate of DSWI in the French series. Analyses were performed with the SPSS program for Windows, version 13.0 (SPSS, Inc., Chicago, IL).

Results

The French series: Immediate outcomes

There were no hospital deaths. The peri-operative complications that occurred are listed in Table 4. Fourteen (5.5%) patients had DSWI.

Table 4.

The French Series: Peri-Operative Complications

Complication	Patients (n = 255)
Prolonged (>48 h) mechanical ventilation	43 (16.9)
Atrial fibrillation, new-onset	44 (17.3)
Myocardial infarction^a	51 (20)
Acute kidney injury^b	6 (2.4)
Multiple blood transfusion (>2 RBCs)	15 (5.9)
Mediastinal re-exploration	3 (1.2)
SSWI^c	16 (6.3)
DSWI^c	14 (5.5)

Most myocardial infarctions were defined only according to cardiac troponin I elevation to ≥70 × the local laboratory ULN (Moussa et al. [10]).

Acute Kidney Injury Work Group: Kidney Disease: Improving Global Outcomes (KDIGO) [11].

Mangram et al. [12].

DSWI = deep sternal wound infection; RBCs = packed red blood cells; SSWI = superficial sternal wound infection; ULN = upper limits of normal.

The original versus the French series

During the corresponding study periods, the rate of BITA use in the French series was greater than in the original series (100% versus 70.6%; p < 0.0001) [5]. Major differences were found between the original and the French series as the pre-operative characteristics of patients. Overall, the patients' risk profiles were higher in the original series (mean EuroSCORE II: 3.6% ± 5.1% versus 2.6% ± 3%, p = 0.0021; Table 1). In the French series, there was a lower number of coronary anastomoses and both the aortic cross-clamp and cardiopulmonary bypass times were shorter (p < 0.0001). Between the two series, there were also major differences as the rates of use of different surgical techniques and the measures that had been taken to reduce the risk of sternal wound infection (Tables 2 and 3). On the basis of the Gatti score estimates [5], there was no difference as the DSWI risk between the two series of patients (p = 0.64 for the pre-operative model and p = 0.85 for the combined model; Table 4). The actual rates of DSWI in the French and in the original series were 5.5% and 4.4% [5], respectively (p = 0.42).

The French validation of the Gatti score

The corresponding rates of the Gatti score risk factors for DSWI in the original and in the French series are reported in Table 5 [5]. The Gatti score goodness-of-fit was satisfactory both for the pre-operative (p = 0.59) and the combined model (p = 0.82). The AUCROC was 0.78 (95% confidence interval: 0.64–0.92) for the pre-operative model and 0.84 (95% confidence interval: 0.69–0.98) for the combined model (Fig. 1). The Youden index was 0.51 (associated criterion: >148 points) for the pre-operative model and 0.69 (associated criterion: >101 points) for the combined model. The sensibility and the specificity for different threshold values were calculated (Table 6). There was a good correspondence between the expected risk of DSWI according to the Gatti score and the actual rate of DSWI in the French series both for low and high DSWI risk; for middle risk, the Gatti score underestimated the actual risk (Fig. 2).

FIG. 1.

The Gatti score. The pre-operative and the combined model in the French series (AUCROC = 0.78, 95% CI: 0.64–0.92 versus AUCROC = 0.84, 95% CI: 0.69–0.98; p = 0.057). AUC = area under the receiver operating characteristics curve; CI = confidence interval.

FIG. 2.

The actual rate of DSWI in the French series and the expected DSWI risk according to the Gatti score. DSWI = deep sternal wound infection; SWI = sternal wound infection.

Table 5.

The Original versus the French Series: The Rates of the Gatti Score Risk Factors for Deep Sternal Wound Infections ^a

Risk factor	Original series (n = 2,872)^a	French series (n = 239)	p
Female gender	441 (15.4)	47 (19.7)	0.078
BMI >30 kg/m²	572 (19.9)	55 (23)	0.25
Diabetes mellitus on oral hypoglycaemic agents	638 (22.2)	65 (27.2)	0.077
Diabetes mellitus on insulin	204 (7.1)	22 (9.2)	0.23
Poor glycemic control	135 (4.7)^b	67 (28)^c	<0.0001
Chronic lung disease^d	128 (4.5)	9 (3.8)	0.62
Long-term dialysis	34 (1.2)	3 (1.3)	0.76
Congestive heart failure	141 (4.9)	9 (3.8)	0.43
Urgent surgical priority^d	1,711 (59.6)	75 (31.4)	<0.0001
Use of chlorhexidine-alcohol	56 (1.9)	0	0.02
Diffusely atherosclerotic ascending aorta	220 (7.7)^e	24 (10)^f	0.19
Post-operative
Low cardiac output	25 (0.9)^g	0^h	0.25
Multiple blood transfusion (>2 RBCs)	315 (11)	14 (5.9)	0.014
Mediastinal re-exploration	136 (4.7)	3 (1.3)	0.012
The Gatti score^a
The pre-operative model (points)	79.3 ± 65.1	81.4 ± 75.9	0.64
The combined model (points)	59.4 ± 48.1	58.8 ± 53.2	0.85

Gatti et al. [5].

Defined as basal serum glucose >200 mg/dL at three consecutive measurements.

Defined as glycated hemoglobin ≥6.5%.

Moussa et al. [10].

By intra-operative EAS (Zingone et al. [13]).

By intra-operative palpation.

Defined as three consecutive cardiac index measurements <2 L/min/m² despite adequate pre-load, afterload, and inotropic support or IABP (Gatti et al. [5]).

Defined according to clinical evaluation.

BMI = body mass index; DSWI = deep sternal wound infection; EAS = epiaortic ultrasonography scan; IABP = intra-aortic balloon pumping; RBCs = packed red blood cells.

Table 6.

Performance of the Gatti Score in the French Series

The pre-operative model
Estimated specificity at fixed sensitivity				Estimated sensitivity at fixed specificity
Sensitivity (%)	Specificity (%)	95% CI ^a (%)	Criterion (points)	Specificity (%)	Sensitivity (%)	95% CI ^a (%)	Criterion (points)
80	58.1	25–89.8	>97.8	80	64.3	35.7–78.6	>143.3
90	36.8	20.9–74.1	>39	90	55.4	28.6–85.7	>174.7
95	25.3	17.7–41.6	>0	95	42.9	11.6–71.4	>199.4
97.5	25.3	17.7–37.6	>0	97.5	14.3	0–28.6	>242

The combined model
Estimated specificity at fixed sensitivity				Estimated specificity at fixed sensitivity
Sensitivity (%)	Specificity (%)	95% CI ^a (%)	Criterion (points)	Specificity (%)	Sensitivity (%)	95% CI ^a (%)	Criterion (points)
80	83.6	23.6–94.7	>101.8	80	85.7	50–100	>99.9
90	34	17.9–88.9	>25.6	90	64.3	29.1–85.7	>126.5
95	22.2	15.1–37.4	>0	95	42.9	7.1–64.3	>141.5
97.5	22.2	15.1–33.4	>0	97.5	28.6	0–57.1	>167.7

Efron [20].

CI = confidence interval.

Discussion

Several models have been devised to predict the risk of developing sternal complications after median sternotomy [6,7]. However, the predictive power of these models are low when they are tested in BITA patients [5]. Actually, these predictive models arose from cohorts of patients undergoing various surgical procedures or pre-selected series of CABG patients in which most of the patients have received single internal thoracic artery (and saphenous vein) grafts [6,7].

In 2015, the Gatti score was created specifically to predict DSWI risk after BITA grafting to reduce the rate of DSWI after CABG without giving up too much in terms of long-term survival benefits derived from the use of BITA grafts [1,2,5]. The score has proven to work well even when it has been tested on an independent validation sample of prospectively enrolled patients derived from the institution where it had been created [8]. In the present study, the Gatti score was tested in a prospective series of 255 patients operated on at the Department of Thoracic and Cardio-Vascular Surgery of the University Hospital of Besançon, France.

There were major differences between the original and the French series in terms of baseline characteristics of patients, the surgical techniques that were adopted, operative data, and post-operative complications. Overall, the French patients were older and had more frequent comorbidities (diabetes mellitus, impairment of mobility secondary to neurologic dysfunction, renal failure, and extracardiac arteriopathy) whereas in the original series there was a stronger ischemic heart disease (higher rates of recent myocardial infarction, left main and three-vessel coronary disease, moderate left ventricular dysfunction, critical state, and urgent surgical priority). In the original series, BITA grafts were used (preferably as in situ conduits) to obtain a left-sided myocardial re-vascularization; the off-pump technique was adopted infrequently and only in the presence of a diffusely atherosclerotic ascending aorta [5,13]. In the French series, the BITA grafting alone with Y-graft was used; in the absence of a diffusely atherosclerotic ascending aorta, either the off- or on-pump technique was adopted according to the surgeon's preference. In both series, prophylactic antibiotics were used and special attention was placed on the skin preparation, internal thoracic artery harvesting technique (both internal thoracic arteries were harvested as skeletonized conduits with low-intensity cautery), sternal reconstruction [14,15], and peri-operative management of hyperglycemia in patients with diabetes mellitus [16]. The longer mean aortic cross-clamp and cardiopulmonary bypass times of the original series were partially due to a higher mean number of coronary anastomoses.

In both series, DSWI was a frequent post-operative complication. It was much more frequent than reported in cohorts of patients who have received single ITA (and saphenous vein) grafts. It was more frequent than reported in BITA patients by other authors [1 –4]. In our opinion, the high rate of sternal wound infections of the two series was caused by the use of BITA grafting without any pre-operative selection of candidates. Conversely, this proves that no substantial reduction of the DSWI risk can be achieved without performing a pre-operative selection of patients. The Gatti score could be a reasonable way to perform this selection [5]. The expected risk of DSWI according to the Gatti score was the same in both series (for both models of the score) despite there were in the original series higher rates of urgent surgical priority, multiple blood transfusions, and mediastinal re-exploration. In fact, both the prevalence of female gender and the rate of diabetes mellitus on oral hypoglycemic agents were greater in the French series, even though the differences were not quite significant. In addition, among the French patients, there was also a superior rate of poor pre-operative glycemic control with respect to the original series, however, the definition of poor glycemic control that was adopted was not the same in both groups.

The goodness-of-fit and the discrimination power of both models of the Gatti score were satisfactory and there was a good correspondence between the expected DSWI risk (by the Gatti score) and the actual rate of DSWI, at least for the low- and high-risk patients. Despite the major differences between the two series including pre-operative characteristics, surgical techniques, and post-operative complications, the Gatti score has proven to be effective in predicting DSWI after routine BITA grafting also in a French validation sample. This could mean that the score works well even in populations of patients with different rates of comorbidities from other institutions. This hypothesis has to be verified by new larger multi-center studies before introducing the Gatti score into clinical practice.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Hemo

, Mohr

, Uretzky

, et al. Long-term outcomes of patients with diabetes receiving bilateral internal thoracic artery grafts. J Thorac Cardiovasc Surg, 2013; 146:586–592.

Dorman

, Kurlansky

, Traad

, et al. Bilateral internal mammary artery grafting enhances survival in diabetic patients: A 30-year follow-up of propensity score-matched cohorts. Circulation, 2012; 126:2935–2942.

Taggart

, Altman

, Gray

, et al. Randomized trial to compare bilateral vs. single internal mammary coronary artery bypass grafting: 1-year results of the Arterial Revascularisation Trial (ART). Eur Heart J, 2010; 31:2470–2481.

Sá

, Ferraz

, Escobar

, et al. Patency of skeletonized versus pedicled internal thoracic artery in coronary bypass graft surgery: A systematic review, meta-analysis and metaregression. Int J Surg, 2014; 12:666–672.

Gatti

, Dell'Angela

, Barbati

, et al. A predictive scoring system for deep sternal wound infection after bilateral internal thoracic artery grafting. Eur J Cardiothorac Surg, 2016; 49:910–917.

Paul

, Raz

, Leibovici

, et al. Sternal wound infection after coronary artery bypass graft surgery: Validation of existing risk scores. J Thorac Cardiovasc Surg, 2007; 133:397–403.

Buja

, Zampieron

, Cavalet

, et al. An update review on risk factors and scales for prediction of deep sternal wound infections. Int Wound J, 2012; 9:372–386.

Gatti

, Barbati

, Luzzati

, et al. Prospective validation of a predictive scoring system for deep sternal wound infection after routine bilateral internal thoracic artery grafting. Interact Cardiovasc Thorac Surg, 2016; 2:606–611.

Nashef

, Roques

, Sharples

, et al. EuroSCORE II. Eur J Cardiothorac Surg, 2012; 41:1–12.

10.

Moussa

, Klein

, Shah

, et al. Consideration of a new definition of clinically relevant myocardial infarction after coronary revascularization: An expert consensus document from the Society for Cardiovascular Angiography and Interventions (SCAI). J Am Coll Cardiol, 2013; 62:1563–1570.

11.

Acute Kidney Injury Work Group: Kidney Disease: Improving Global Outcomes (KDIGO). Clinical practice guideline for acute kidney injury. Kidney Int, 2012; 2:1–138.

12.

Mangram

, Horan

, Pearson

, et al. Guideline for prevention of surgical site infection, 1999. Hospital Infection Control Practices Advisory Committee. Infect Control Hosp Epidemiol, 1999; 20:250–278.

13.

Zingone

, Rauber

, Gatti

, et al. The impact of epiaortic ultrasonographic scanning on the risk of perioperative stroke. Eur J Cardiothorac Surg, 2006; 29:720–728.

14.

Erdinc

, Ocal

, Sezer

, et al. Double-loop sternal wiring technique. Ann Thorac Surg, 1996; 62:320–321.

15.

Sutherland

, Martinez

, Guynes

. A rapid, secure method of sternal closure. Cardiovasc Dis, 1981; 8:54–55.

16.

Furnary

, Zerr

, Grunkemeier

, Starr

. Continuous intravenous insulin infusion reduces the incidence of deep sternal wound infection in diabetic patients after cardiac surgical procedures. Ann Thorac Surg, 1999; 67:352–360.

17.

Gatti

, Pappalardo

, Gon

, Zingone

. Protecting the crossover right internal thoracic artery bypass graft with a pedicled thymus flap. Ann Thorac Surg, 2006; 82:1919–1921.

18.

Swets

. Measuring the accuracy of diagnostic systems. Science, 1988; 240:1285–1293.

19.

DeLong

, DeLong

, Clarke-Pearson

. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics, 1988; 44:837–845.

20.

Efron

. Better bootstrap confidence intervals. J Am Stat Assoc, 1987; 82:171–185.