Incidence and healing times of postoperative sternal wound infections: a retrospective observational single-centre study

Abstract

Objectives: Analyses of incidence and time required to heal sternal wound infections after heart surgery performed via a median sternotomy between 2020 and 2022. Results: Superficial wound infections (SWI) were five times more common (2.7%) than mediastinitis (0.5%) among 2693 patients. The median time between the operation and diagnosis of SWI was 26 (interquartile range [IQR] 15–33) days vs. 16 (IQR 9–25) days for mediastinitis (p = .12). Gram-negative bacteria caused 44% of the 85 infections. Sternal wound infection correlated to higher body mass index, female sex, smoking, diabetes mellitus, previous myocardial infarction, coronary artery bypass grafting, use of internal mammary graft, and re-entry for postoperative bleeding. Eight of 59 patients (13.6%) with sternal wound infections had bilateral mammary grafts, compared to 102 of 1191 patients (8.6%) without wound infections (p = .28). Negative pressure wound therapy was always used to treat mediastinitis and applied in 63% of patients with SWI. Two of 13 patients with mediastinitis (15%) and none of 72 patients with SWI died within 90 days after the operation. The median time until the wound healed was 1.9 (IQR 1.3–3.7) months after SWI vs. 1.7 (IQR 1.3–5.3) months after mediastinitis (p = .63). Six patients (7%) required longer than one year to treat the infection. Conclusions: Postoperative sternal wound infections usually appeared several weeks after surgery and were associated with factors as high body mass index, diabetes mellitus and coronary artery bypass. SWI were more common than mediastinitis and often required negative pressure wound therapy and similar treatment time as mediastinitis.

Keywords:

• Superficial sternal wound infection
• mediastinitis
• negative pressure wound therapy

Introduction

Sternal wound infection after cardiac surgery may be a devastating and potentially life-threatening complication causing suffering for the patient, can require intravenous antibiotic treatment, repeat surgical intervention, a long hospital stay, or sick-leave, and is associated with excessive costs. The number of wound infections reported to the National Swedish Cardiac Surgery Registry in the National Swedish Web-system for Enhancement and Development of Evidence-based Care in Heart Disease Evaluated According to Recommended Therapies (SWEDEHEART) registry (www.ucr.uu.se/swedeheart) is based only on events occurring during the initial hospital stay, and therefore, underestimates the true incidence of postoperative infections. Most patients are discharged from the hospital within a week of heart surgery, and wound infections usually appear later. Many reports have addressed the incidence and methods of treating sternal wound infections; however, no study has focused on the time required to heal the infection [1–10].

This study aimed to assess the incidence of mediastinitis and superficial sternal wound infections after cardiac surgery and to analyse the time until the infections healed.

Methods

Patients

We included adult patients with sternal wound infections verified by positive cultures of microorganisms and identified up to three months after heart surgery performed via median sternotomy between January 2020 and December 2022 at the Karolinska University Hospital in Stockholm, Sweden. Information on all eligible patients was obtained from the SWEDEHEART registry, and the study was approved by the Swedish Ethical Review Authority (DNR. 2022-05561-02).

At discharge, all patients were routinely instructed to contact our outpatient clinic with any signs of wound infection. Any early wound infection was recorded at an outpatient clinic visit approximately six weeks after the operation, when all wounds were inspected. Some patients were contacted by telephone by an experienced nurse and readmitted for examination if the wound did not heal. All patient charts were retrospectively examined for a minimum of three months after the operation to detect any wound complications. The patients were followed-up until 31 March 2023.

Perioperative management

Before surgery, all patients were screened for ongoing infections using laboratory tests. A dentist was consulted to perform a scan for dental infections in the patients scheduled for prosthetic valve surgery. The night before and the morning of surgery the patients showered with chlorhexidine soap. Any hair was shaved on the breast and legs the day of surgery a few hours before the operation. The skin of the anterior chest area was scrubbed with chlorhexidine alcohol (Fresenius KABI, Uppsala, SE) and covered with a plastic film. Intravenous antibiotic prophylaxis with isoxazolyl penicillin (2 g) and one dose of teicoplanin (600 mg or 400 mg if the patient’s body weight was less than 80 kg) was started 30 min before skin incision. Five doses of isoxazolyl penicillin were administered over 24 h. Bone wax for haemostasis was generally applied to the surface of the spongiosa after the sternotomy. No local antibiotic sponges were used when the sternotomy was closed using multiple wires, two subcutaneous suture layers, and continuous intracutaneous sutures. A bandage (Mepilex Border, Mölnlycke, Göteborg, SE) was used to cover the wound and was left for seven days or changed earlier in case of bleeding. Perioperative glucose levels were monitored in all patients and subcutaneous short-acting insulin injections were administered if the levels exceeded 10 mmol/L [11]. All surgical sites were inspected before hospital discharge.

Definitions and treatment

Information on age, sex, and diagnosis was based on data reported in the SWEDEHEART register. Chronic obstructive pulmonary disease was defined as the need for continuous medication with bronchodilators or steroids for lung diseases. Diabetes was defined as the requirement for treatment with insulin, oral hyperglycaemic drugs, or a special diet in patients with hyperglycaemia or elevated glycosylated haemoglobin 1. Creatinine clearance was estimated using the Cockcroft-Gault equation [12]. Myocardial infarction was defined as having two of three indications: clinical presentation, electrocardiography, or elevated biomarkers [13]. Emergent surgery was performed within 24 h of the decision to operate. Surgical risk was assessed using the European System for Cardiac Operative Risk Evaluation Score (EuroSCORE) II [14].

Superficial wound infection (SWI) involved the subcutaneous tissue but did not require reopening of the sternotomy. SWI sometimes had an insidious onset, with a few centimetres of infected skin opening and purulent secretion. If the infected wound could be probed subcutaneously, it was opened and revised under anaesthesia. All infections were treated with antibiotics. Mediastinitis commonly presented with increased pain, fever, purulent drainage, and sternal dehiscence. An infection was classified as mediastinitis if microorganisms were detected in mediastinal cultures during surgical exploration. Mediastinitis was treated by reopening the entire sternotomy, debriding the devitalised tissue, irrigating the wound, and applying negative pressure wound therapy, which was changed every fourth day until cultures were negative. The sternum was sutured again with wires, and the skin was closed. A surgical site infection was considered healed when the wound had completely closed.

Statistical methods

Continuous variables were presented as arithmetic means and one standard deviation or non-normal distribution with median and interquartile range (IQR). A Student’s t test was used to compare continuous variables between the two groups. A Mann–Whitney U test was used for skewed distributions. The chi-square method with Yate’s correction for continuity was used to analyse the proportions of categorical variables. A p-value less than .05 was considered statistically significant. Calculations were performed using STATISTICA 13 software (Stat Soft, Dell).

Results

A minimally invasive approach via a 5 cm right thoracotomy was used to repair mitral and tricuspid valve lesions and atrial defects in 432 of 3125 patients (13%) between 2020 and 2022. There were two cases of SWIs among these patients. Thirty-three patients who had aortic valve surgery via ministernotomy were not included in this analyses. In total, 2693 patients who underwent full median sternotomy were included in the analysis of sternal wound infections. Of these, 96 patients (3.6%) experienced sternal wound complications. Five patients underwent reoperation for postoperative sternal dehiscence without clinical signs of infection or negative mediastinal cultures. Six patients were excluded from wound infection analysis who presented with 2–10 cm skin openings in the sternal wound, suggesting infection, but repeated cultures were negative, and the wounds healed within two months.

During three months of follow-up, 85 of 2693 patients (3.2%) developed a sternal wound infection. There were five times more SWI (n = 72, 2.7%) than mediastinitis (n = 13, 0.5%) among the 2693 patients. The incidence of infection did not decrease during the three years studied (Table 1). The median time from the date of surgery to diagnosis was 26 (IQR 15–33) days for SWI and 16 (IQR 9–25) days for mediastinitis (p = .12) (Figure 1). In 27 patients (32%) the infection was diagnosed more than 30 d after the operation.

Figure 1. Time to diagnosis of mediastinitis and superficial wound infection (SWI) within three months after open-heart surgery.

Table 1. Mediastinitis and superficial wound infections during 2020–2022.

Year of surgery	2020	2021	2022	Total
No. patients with sternotomy	889	904	900	2693
Superficial wound infection (%)	23 (2.6)	20 (2.2)	29 (3.3)	72 (27)
Mediastinitis (%)	3 (0.3)	4 (0.4)	6 (0.7)	13 (0.5)
Total no. infections (%)	26 (2.9)	24 (2.6)	35 (4,0)	85 (3.2)

Patients with a sternal wound infection had a higher body mass index, were more often female, were smokers, had diabetes mellitus and prior myocardial infarctions than those without a wound infection (Table 2). Fourteen of the 207 patients (6.8%) with diabetes mellitus and body mass index ≥30 kg/m² suffered from a sternal wound infection. There was no increased risk of wound infection in patients with valve replacement for endocarditis and ongoing intravenous antibiotic treatment after the operation (1/150, 0.7%). Age, EuroSCORE risk score, emergent surgery, and previous cardiac surgery did not correlate with the risk of wound infection.

Table 2. Baseline characteristics of patients with and without sternal wound infection after heart surgery from 2020–2022.

	No infection		Mediastinitis		Superficial infection
	(n = 2608)		(n = 13)		(n = 72)		p-value*
	Mean	SD	Mean	SD	Mean	SD
Age (yrs)	629	12.6	64.6	10.3	62.2	8.6	.85
Body mass index (kg/m^2)	27.2	4.7	29.6	4.7	28.6	4.8	<.001
Creatinine clearance (mL/min/1.73m^2)	96.4	35.4	94.6	43.0	98.4	36.2	.73
	Median	25^th–75^th percentiles	Median	25th–75th percentiles	Median	25th–75th percentiles
EuroSCORE II	4.0	2.1–7.4	2.6	1.6–5.8	4.0	2.1–5.3	.07
	n	%	n	%	n	%
Female	600	23.0	3	23.1	25	34.7	.02
Present smoker	334	12.8	3	23.1	14	19.7	.01
Hypertension	1713	65.7	9	69.2	46	63.9	.94
All diabetes	557	21.4	7	53.8	24	33.3	.001
Insulin treatment	229	8.8	5	38.5	9	12.5	.02
Obstructive pulmonary disease	253	9.7	1	7.7	7	9.9	.81
Peripheral vascular disease	158	6.1	2	15.4	7	9.9	.25
Previous myocardial infarction	580	22.2	6	46.2	26	36.7	<.001
Endocarditis	150	5.7	0		1	1.4	.19
Emergent surgery	174	6.7	0		5	6.9	.93
Urgent surgery	962	36.9	8	61.5	19	26.4	.40
Previous cardiac surgery	281	10.8	0		8	11.3	.95
Ventricular ejection fraction ≤30%	156	6.0	0		1	1.4	.10
Pulmonary pressure >30 mmHg	580	22.2	3	3	18	25.4	.17

IQR: Interquartile range; EuroSCORE: European System for Cardiac Operative Risk Evaluation Score II [14].

* Infection vs. no infection.

Coronary artery bypass, use of an internal mammary graft, and re-entry for early postoperative bleeding were more common in patients with sternal wound infections than in those without wound infections (Table 3). Eight of 59 patients (13.6%) with sternal wound infections had bilateral mammary grafts, compared to 102 of 1191 patients (8.6%) without wound infections (p = .28). The mammary artery pedicle was skeletonised in all but one of these patients.

Table 3. Perioperative data in patients with and without sternal wound infection from 2020–2022.

	No infection		Mediastinitis		Superficial infection
	(n = 2608)		(n = 13)		(n = 72)
	Mean	SD	Mean	SD	Mean	SD	p-value*
Extra-corporeal circulation (min)	93.6	46.8	93.7	54.7	86.1	40.7	.23
Aortic occlusion time (min)	68.1	36.4	66.3	52.8	63.7	36.4	.34
	n	%	n	%	n	%
Coronary artery bypass	1191	45.6	11	84.6	48	66.7	<.001
Internal mammary artery graft	1151	44.1	10	76.9	46	63.9	.001
Bilateral mammary grafts	102	8.6**	1	9.1**	7	14.6	.28**
Valve surgery	1285	49.3	4	30.8	24	33.8	.08
Postoperative bleeding	130	5.0	2	15.4	8	11.3	.01
Postoperative stroke	64	2.5	0		1	1.4	.86
Postoperative dialysis	90	3.4	1	7.8	2	2.8	.98
Mechanical circulatory support	10	0.4	0		1	1.4	.79
Initial respirator treatment >2 d	70	2.7	1	7.8	4	5.6	.07

*Infection vs. no infection; **percent of coronary artery bypass procedures.

Outcomes

Gram-negative bacteria were detected in the sternal wounds of 37 of the 85 patients (44%), and 31 of these infections (84%) occurred within 30 d (Table 4 and Supplementary Figure). Gram-negative bacteria caused eight of the 13 mediastinitis (62%) cases and 29 of 72 SWI (40%) cases. Two patients (15%) with mediastinitis died at 35 and 90 d, after the operation from organ failure and ongoing infection. None of the patients with SWI died within 90 d after surgery.

Table 4. Primary organisms detected in 85 sternal wounds.

Cultured organism	n
Staphylococcus aureus	21
Coagulase-negative Staphylococcus	7
Staphylococcus epidermidis	2
Normal skin flora	9
Cutibacterium acnes	5
Corynebacterium striatum	2
Enterobacter	7
Proteus mirabilis	4
Klebsiella oxytoca	12
Serratia marcescens	5
Escherichia coli	5
Morganella morganii	1
Citrobacter koseri	1
Pseudomonas aeruginosa	2
Candida albicans	2

Negative pressure wound therapy was always used in case of mediastinitis and applied in 45 of the 72% of patients (63%) with SWI, some of whom could be managed at our outpatient clinic. The median time until SWI healed was 1.9 (IQR 1.3–3.7) months compared to 1.7 (IQR 1.3–5.3) months after mediastinitis (p = 0.63). Among the 45 subjects with SWI treated with negative pressure wound therapy the wound healed after a median of 2.2 (IQR 1.7–4.4) months compared a median of 1.3 (IQR 0.6–2.3) months in those treated only with redressing of the wound (p = .002). Six patients (7%), two with mediastinitis and four with SWI, required longer than one year of treatment until the infection healed (Figure 2). Computed tomography in these patients with a longstanding infection showed subcutaneous fistula but no certain osteomyelitis in the sternum. Muscle flaps were used in two patients

Figure 2. Time to diagnosis of mediastinitis and superficial wound infection (SWI) after open-heart surgery in relation to time until the wound healed.

Discussion

We assume that our combined prospective recording of infections occurring within six weeks and retrospective examination of each patient’s chart for up to three months after the operation reflected the total incidence of sternal wound infections at a rate of 3.2%. Our rate of infection did not decline during the three-year study period. In accordance with our observations SWI is more frequent than mediastinitis, with a reported incidences in the range 1.2–6.7% [3,7,8] whereas the incidence of mediastinitis ranges from 0.15 to 2.7% [2–8,10]. We did not strictly apply the classification of surgical site infections according to the Centers for Disease Control and Prevention [15]. We included infections that were diagnosed more than one month after the operation and classified infections as superficial also in cases with sternal wires in the wound. Our subjects with SWI were ambulant and managed at the outpatient clinic. A clinical course different from cases with mediastinitis who all required repeat surgery during anaesthesia, open surgical debridement of the mediastinum and treatment in hospital until the sternum could be sutured. However, SWI was not a benign infection and often required negative pressure wound therapy and as long treatment time as mediastinitis. Particularly SWI extending down to the sternum required a long time to heal. One reason is that a subcutaneous infection down to the sternal bone usually does not heal until the exposed sternal wires can be removed several months after the operation when the sternotomy is stable and healed.

Our findings agree with other reports of multiple patient factors correlated with an increased risk of wound infection, such as female sex, smoking history, high body mass index, diabetes mellitus, peripheral vascular disease, renal failure, chronic obstructive pulmonary disease, urgency of surgery, and perioperative factors such as use of bone wax, bilateral mammary artery grafts, suture technique, and reopening of the sternum [3,7–10,16–19]. Smoking impairs lung function and skin circulation and may increase the risk of wound infection [15]. Diabetes and hyperglycaemia are associated with impaired leukocyte function and are known to increase the risk of surgical site infection [20]. Blood glucose was measured a minimum of four times daily in case of postoperative hyperglycaemia but despite repeated doses of short acting insulin we often failed to obtain blood glucose levels at 8–10 mmol/L. Despite the use of skeletonised grafts, the infection rate was high when both mammary arteries were used.

Similar to our findings skin flora as coagulase-negative staphylococci and Staphylococcus aureus are the most common causes of sternal wound infections [3,4,9,18,19]. In other reports Gram-negative bacteria account for approximately 25–40% of sternal wound infections[3,4,19]. The high proportion of early occurring Gram-negative bacteria among our infections indicates possible contamination with bowel microorganisms. One hypothesis is that perioperative prophylaxis with both beta-lactam antibiotics and glycopeptide antibiotics, such as teicoplanin, increases the colonisation of Gram-negative bacteria due to the more extended effect on the skin flora. Even with teicoplanin added to isoxazolyl penicillin we had Gram-positive infections. Gram-negative infections cannot be treated by teicoplanin, that can be nephrotoxic. Cepahlosporins are usually used as antibiotic prophylaxis in patients undergoing cardiac surgery [21]

We relied on negative pressure wound therapy to clean the infection; however, in some studies, the use of muscle flaps to cover infected wounds has been shown to shorten the hospital stay [1].

Postoperative wound infections occurred despite extensive preventive efforts, meticulous perioperative routines, and antibiotic prophylaxis. Possible contributing causes were a mix of patient and perioperative factors. We are currently revising our preoperative scrubbing routines, type of antibiotic prophylaxis, monitoring of postoperative blood glucose levels, and the management of surgical wounds. The goal is to eliminate nosocomial wound infections that cause suffering and require significant resources for treatment.

Limitations

We are certain that all severe sternal wound infections were included but additional minor infections may have been missed in our survey. Patients may have moved to other regions and experienced minor wound infections that healed without informing the study team.

Conclusions

Postoperative sternal wound infections usually appeared several weeks after surgery and were associated with factors as high body mass index, diabetes mellitus and coronary artery bypass. SWI were more common than mediastinitis, often required negative pressure wound therapy and as long treatment time as mediastinitis.

Ethical approval

The study was approved by Swedish Ethical Review Authority (DNR. 2022-05561-02). Informed consent was obtained from operated patients to analyze and publish anonymized data.

Authors contributions

The authors TI and MD have contributed equally to this work and participated in design, acquisition of data, analyses drafting and revising the manuscript. AB and SB contributed to analyses, drafting and revision of the manuscript. All authors have read and approved the final manuscript.

Data availability statement

Due to the study’s ethical consent data are not available for sharing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Magnus Dalén was supported by a research grant from Fredrik Lundberg.