Does routine topical antimicrobial administration prevent sternal wound infection after cardiac surgery?

Summary

A best evidence topic in cardiac surgery was written according to a structured protocol. The question addressed was ‘Does routine topical antimicrobial administration prevent sternal wound infection (SWI) after cardiac surgery? Altogether >238 papers were found using the reported search, of which 11 represented the best evidence to answer the clinical question. The authors, journal, date and country of publication, patient group studied, study type, relevant outcomes and results of these papers are tabulated. Several different antimicrobial agents, dosages and application protocols were found in the literature. Regarding topical vancomycin use, a meta-analysis by Kowalewski et al. demonstrated a 76% risk reduction in any SWI. Collagen-gentamicin sponge application was associated with a 38% risk reduction in SWI in another meta-analysis by Kowalewski et al., which included 4 randomized control trials and >23 000 patients. Lower evidence observational studies found benefit in the use of different regimes, including: combination of vancomycin paste and subcutaneous gentamycin; combined cefazoline and gentamicin spray; isolated cefazolin; bacitracin ointment; and rifampicin irrigation. We conclude that, in light of the body of evidence available, topical antibiotic application prevents SWI, including both superficial and deep SWI. The strongest evidence, derived from 2 meta-analyses, is related to the use of gentamicin-collagen sponges and topical vancomycin. Heterogeneity throughout studies regarding antibiotic agents, dosages, application protocols and SWI definition makes providing general recommendations challenging.

A best evidence topic was constructed according to a structured protocol.

INTRODUCTION

A best evidence topic was constructed according to a structured protocol. This is fully described in the ICVTS [1].

THREE-PART QUESTION

In [patients undergoing cardiac surgery] does [routine use of topical antimicrobials] prevent [sternal wound infection]?

CLINICAL SCENARIO

During sternal closure, after routine aortic valve replacement for aortic valve stenosis in a 75-year-old male patient; the surgical assistant asks the scrub nurse for diluted vancomycin and applies it to the wound. The anaesthesiologist challenges this practice, suggesting that topical antimicrobials make no difference to wound infection rates. The resident on the case, unconvinced of the best practice decides to review the literature for evidence.

SEARCH STRATEGY

A comprehensive literature review was performed by searching MEDLINE from 1965 to July 2020 using the PubMed interface. The following strategy was used: ((sternotomy[Mesh Terms]) OR (‘cardiac surgery’) OR (‘sternal wound’)) AND ((anti-bacterial agents[MeSH Terms]) OR (antibiotics) OR (antimicrobials)) AND ((topical) OR (local)). The eligible papers were in English.

SEARCH OUTCOME

A total of 238 papers were found using the reported search. Initial screening identified 18 relevant papers. After screening the references for additional relevant papers, 1 was added to the review. Two of the 19 relevant papers are meta-analysis, which in their turn include 8 studies individually identified in the search. For review purposes, the pooled data from the meta-analysis were considered and are presented below. The final 11 papers that provided the best evidence to answer the question are presented in Table 1.

Table 1:

Best evidence papers

Author, date, journal and country	Patient group	Outcomes	Key results (topical antibiotic versus control)	Comments
Study type (level of evidence)
Interventions with vancomycin
Kowalewsky et al. (2017), J Thorac Cardiovasc Surg, Poland/Italy [2] Meta-analysis (level II)	Includes 1 RCT and 3 observational studies n = 20 039 patients Control: 13 196 Vancomycin: 6843 Heterogeneous application protocols ranging from 250 mg to 10 g of topical vancomycin	Any SWI	RR: 0.24 (95% CI: 0.05–0.91) P = 0.04	Heterogeneity in protocols, follow-up and outcome definition
		SSWI	RR: 0.2 (95% CI: 0.01–6.07) P = 0.36
		DSWI	RR: 0.24 (95% CI: 0.06–0.99) P = 0.05
Interventions with gentamicin
Kowalewsky et al. (2015), J Thorac Cardiovasc Surg, Poland/Germany [3] Meta-analysis (level I)	Includes 4 RCT and 10 observational studies n = 23 102 patients Control: 16 247 Gentamicin-collagen spong: 6855 Heterogeneous application protocols including retrosternal and/or along the sternum	Any SWI	RR: 0.62 (95% CI: 0.47–0.81) P < 0.001	Heterogeneity in protocols, follow-up and outcome definition
		SSWI	RR: 0.60 (95% CI: 0.43–0.83) P = 0.002
		DSWI	RR: 0.62 (95% CI: 0.44–0.87) P = 0.006
		Mediastinitis	RR: 0.64 (95% CI: 0.45–0.91) P = 0.01
Sahim (2018), Arch Med Sci Atheroscler Dis, Turkey [4] Retrospective study (level III)	n = 381 Control: 125 Gentamicin-collagen sponge (above sternal wires): 79 Gentamicin spray (before soft tissue closure): 177	Any SWI within 2 months	Control: 20.8% Gentamicin-collagen sponge: 11.6% Gentamicin spray: 5.0% P = 0.04	Small cohort Multivariate analysis of both protocols is not presented by the authors The gentamicin spray group only presented a statistically significant difference (versus control) for DSWI
		SSWI within 2 months	Control: 9.5% Gentamicin-collagen sponge: 7.7% Gentamicin spray: 2.5% P = 0.009
		DSWI within 2 months (CDC definition) [13]	Control: 11.1% Gentamicin-collagen sponge: 3.9% Gentamicin spray: 2.5% P = 0.011
Waldow et al. (2018), J Hosp Infect, Germany [5] Prospective non-randomized study (level III)	n = 2340 patients Control: 613 Gentamicin sponge (retrosternal): 559 Chlorhexidine disinfection: 611 Chlorhexidine disinfection + gentamicin sponge: 557	DSWI within 30 days (Similar to CDC definition [13])	Univariate analysis: Control: 5.06% Gentamicin sponge: 2.15% Chlorhexidine disinfection: 2.13% Chlorhexidine disinfection + gentamicin sponge: 1.8% P = 0.002	Authors compared not only gentamicin sponge use but also chlorhexidine disinfection (versus alcohol disinfection) which was also independently associated with lower rates of DSWI Deep dehiscence without infection did not differ statistically between the groups
			Regression analysis: Gentamicin sponge use: OR 1.98 (1.40–2.56 95% CI); P = 0.013
Interventions combining vancomycin and gentamicin
Andreas et al. (2017), Interact Cardiovasc Thoracic Surg, Austria [6] Retrospective study (level III)	n = 1851 patients Control: 919 Vancomycin paste + gentamicin (applied on the sternal wires): 932	DSWI (El Oakley class 2B) [14]	1.4% vs 3.7% P = 0.002 Binary regression analysis: OR = 0.3 (95% CI 0.18–0.54)	Additional changes in the sternal wiring procedure were part of the novel protocol and may induce bias Follow-up time not clear
		Any sternal wound infection or dehiscence	2.0% vs 5.8% P < 0.001
Interventions with other antimicrobial agents
Osawa et al. (2016), Gen Thorac Cardiovasc Surg, Japan [7] Retrospective study (level III)	n = 6960 patients Control: 686 Cefazolin + gentamicin spray (sprayed throughout the duration of the procedure): 6274	DSWI within 30 days (CDC definition) [13]	0.46% vs 1.7% P < 0.001	Although a multicentric study the control cases were all from same centre and a different timeframe which may induce bias
Ali et al. (2020), Heart Lung Circ, Australia [8] Retrospective study (level III)	n = 1902 patients Cefazolin group (applied to the sternum and subcutaneous tissue): 315 Control group: 1587	DSWI (CDC definition) [13]	0% vs 1.7% P = 0.022	The incidence of DSWI in patients not exposed to cefazolin and the results of the multivariate analysis regarding antimicrobial usage are not presented in the manuscript Follow-up time not clear Only 1 of 5 surgeons used topical cefazolin presenting important bias
Chan et al. (2017), Ann Thorac Surg, USA [9] Retrospective study (level III)	n = 1495 CABG or valve surgery patients All patients received topical bacitracin after incision closure	SSWI within 6 weeks	0.27%	The frequency of outcomes was compared with the predicted rates calculated by STS National Database risk calculator
		DSWI within 6 weeks (CDC definition) [13]	0% Expected: 0.29% (IQR: 0.19–0.46%)
MacIver et al. (2006), Heart Surg Forum, USA [10] Retrospective study (level III)	n = 2455 patients Control: 1036 Bacitracin (after incision closure): 1419	Mediastinitis requiring surgical intervention	0.2% vs 1.2% P < 0.01	Higher prevalence of risk factors for mediastinitis in the bacitracin group
Aygün et al. (2014), Cardiovasc J Afr, Turkey [11] Prospective non-randomized study (level III)	n = 159 CABG diabetic patients Control: 73 Rifamicine (applied to mediastinum, sternum and suprasternal tissues): 78	SSWI within 30 days	0% vs 1.4% P = 0.303	Method of evaluation and distinction of SSWI and DSWI lacking Small number of patients
		DSWI within 30 days	0% vs 0% P = 1
De Santo et al. (2020), Sci Rep, Italy [12] Retrospective study (level III)	n = 965 CABG patients Control: 448 Rifampicin (applied to mediastinum, sternum and suprasternal tissues): 517 PSM (1:1): 363	DSWI (CDC definition) [13]	0.2% vs 2.5% P = 0.002 PSM analysis: 0.3% vs 2.1% P = 0.04	The treatment group was compared with historic controls Follow-up time not clear

CABG: coronary artery bypass graft; CDC: Center for Disease Control and Prevention; CI: confidence interval; DSWI: deep sternal wound infection; IQR: interquartile range; NR: not reported; OR: odds ratio; PSM: propensity score matched; RCT: randomized control trial; RR: risk ratio; SSWI: superficial sternal wound infection; STS: Society of Thoracic Surgeons; SWI: sternal wound infection.

RESULTS

Interventions with vancomycin

Kowalewski et al. [2] conducted a meta-analysis (n = 20 039) evaluating the effectiveness of topical vancomycin in reducing sternal wound infection (SWI). The authors conclude that topically administered vancomycin was associated with a 76% risk reduction in any SWI [risk ratio (RR): 0.24; 95% confidence interval (CI): 0.06–0.91; P = 0.04; I² = 70%] and a 76% risk reduction in deep sternal wound infection (DSWI) (RR: 0.24; 95% CI: 0.06–0.99; P= 0.05; I² = 58%). However, 4 studies included in the meta-analysis present heterogeneous application protocols (ranging from 250 mg to 10 g of vancomycin), time of follow-up (ranging from 1 to 12 months after index surgery) and outcome definition.

Interventions with gentamicin

Kowalewski et al. [3] performed a meta-analysis (n = 23 102) including 4 randomized controlled trials and 10 observational studies regarding the effect of gentamicin-collagen sponges on SWI. The sub-analysis of the 4 randomized controlled trials demonstrated a 39% risk reduction in SWI in the treatment group (RR: 0.61; 95% CI: 0.39–0.98; P = 0.04; I² = 71%). Overall analysis of all studies demonstrated a 38% risk reduction in SWI with gentamicin-collagen sponges (RR: 0.62; 95% CI: 0.47–0.81; P < 0.001; I² = 63%). Similar results were obtained for DSWI, superficial sternal wound infection (SSWI) and mediastinitis with a risk reduction in the overall analysis of 38% (RR: 0.62; 95% CI: 0.44–0.87; P = 0.006; I² = 34%), 40% (RR: 0.60; 95% CI: 0.43–0.83; P = 0.002; I² = 66%) and 36% (RR: 0.64; 95% CI: 0.45–0.91; P = 0.01; I² = 16%). The authors alert to the fact that this benefit might be attenuated in patients receiving bilateral internal mammary artery (BITA) grafts as a regression of logRR against the percentage of patients receiving BITA demonstrated a significant positive correlation (Pslope = 0.05) between the risk of SWI and DSWI (but not mediastinitis or SSWI). Heterogeneous definition of SWI, follow-up (ranging from perioperative/30–90 days postoperatively) and placement technique (retrosternal and/or along the sternum) was noted between the 14 studies.

Sahim [4] retrospectively reported the incidence of DSWI and SSWI in 381 CABG patients using 2 different strategies: a gentamicin-collagen sponge applied directly above the sternal wires after sternal closure and direct spraying of a 10-ml solution containing 80 mg of gentamicin before soft tissue closure. The gentamicin-collagen group presented statistically significant lower rates of any SWI, SSWI and DSWI. The gentamicin spray group presented only statistically lower rates of DSWI when compared to the control group (3.9% vs 11.1%). This study is limited by the small number of patients included and no presentation of baseline patient characteristics between the groups.

Waldow et al. [5] investigated (n = 2340) the effect of topical gentamicin sponge and chlorhexidine skin disinfection (versus alcohol solution) in sternal wound complications in the first 30 days after the index procedure. The authors compared 4 different groups: the control group, receiving neither intervention; the gentamicin sponge only group; the chlorhexidine skin disinfection only group; and the combined strategy group, with patients receiving both interventions. Both strategies were independently associated with the reduced risk of DSWI (OR: 1.820; 95% CI: 1.219–2.422; P = 0.026 for chlorhexidine disinfection and OR: 1.980; 95% CI: 1.399–2.560; P = 0.013 for gentamicin sponge). The combination of both strategies presented the lowest rate of DSWI (1.8%; P = 0.002). Rates of deep dehiscence without infection were not reduced by either strategy.

Interventions combining vancomycin and gentamicin

Andreas et al. [6] evaluated (n = 1851) the impact of a novel closure protocol in the incidence of SWI against historical controls. The novel protocol included applying 3 g of vancomycin paste prior to sternal wiring and 160 mg of gentamicin on the sternal wires prior to subcutaneous closure. The authors identified several risk factors for DSWI (including mammary artery harvesting, diabetes and peripheral vascular disease) and a protective effect of the novel closuring protocol that consisted of the aforementioned topical antibiotics as well as the usage of 6 figure-of-8 wires for sternal closure in all patients, if appropriate; and bone wax removal before the application of vancomycin paste.

Interventions with other antimicrobial agents

Osawa et al. [7] retrospectively investigated (n = 6960) the effect of spraying a solution of 1 g of cefazolin, 40 mg of gentamicin and 40 ml of saline in several occasions throughout the duration of the procedure. The rate of DSWI was significantly reduced with this protocol (0.46% vs 1.7%, P < 0.001). This study, however, includes 5 different centres, one of which acting as the control group with no baseline characteristics available for comparison, which may render interpretation of the results troublesome. The authors further mention that diabetes mellitus was not associated with higher rates of DSWI (0.32% vs 0.46%; RR: 0.69; 95% CI: 0.15–3.12; P = 1.0) in 3019 patients sprayed with the cefazolin–gentamicin combination.

Ali et al. [8] presented a retrospective analysis (n = 1902) on risk factors for developing DSWI. In their article, the authors mention that 1 out of 5 surgeons in their department routinely uses cefazolin (1 g in 300 cc of normal saline to the surface of the sternum and overlying subcutaneous tissue) and that none of his patients developed DSWI (n = 315; P = 0.022). As the treatment group consisted only of patients from the same surgeon and no baseline characteristics between the 2 groups are presented, bias should be taken into consideration when interpreting these results.

Chan et al. [9] published a retrospective cohort of 1495 patients in which bacitracin was applied in all. Bacitracin application was done after skin closure but before removal of sterile drapes and then covered with sterile gauze, which was only removed on postoperative day 2. This cohort of patients presented no DSWI within 6 weeks of index cardiac surgery. The expected DSWI rate for this cohort according to the Society Thoracic Surgeons National Database risk calculator was 0.29% (Interquartile range: 0.19–0.46%).

MacIver et al. [10] compared (n = 2455) the application of bacitracin ointment to the sternotomy incision after closure with historic controls and found a 6-fold reduction in mediastinitis requiring reintervention (P < 0.01).

Aygün et al. [11] prospectively evaluated (n = 159) the effects of irrigating the mediastinum, sternum and suprasternal tissues using 250 mg of rifamycin diluted with 10 ml of isotonic solution in diabetic patients undergoing on-pump coronary revascularization. This is the only of the 11 papers considered that exclusively evaluated diabetic patients. No statistically significant difference was found regarding SSWI and DSWI 30-day rates. However, interpretation of the results must take into consideration the small cohort presented.

In a similar fashion, De Santo et al. [12] compared (n = 965) irrigation of the mediastinum, sternum and suprasternal tissues with 600 mg rifampicin prior to closure, with historic controls. The authors found a significant reduction in DSWI rates in the overall series (0.2% vs 2.5%; P = 0.002) and in the propensity matched cohort (0.3% vs 2.1%; P = 0.04). These findings must be interpreted with caution, as the comparison is made with historical controls. This and other studies that use a historical control group are susceptible to bias due to other time-dependent interventions other than topical antimicrobial use.

CLINICAL BOTTOM LINE

The body of evidence available suggests that topical antibiotic application reduces the incidence of SWI, including both SSWI and DSWI. The strongest evidence, derived from 2 meta-analyses, is related to the use of gentamicin-collagen sponges and topical vancomycin. Heterogeneity throughout studies regarding antimicrobial agents, dosages, application protocols and SWI definition makes providing general recommendations challenging.

Author contributions

Pedro Lamares Magro: Conceptualization; Data curation; Investigation; Methodology; Writing—original draft; Writing—review & editing. Miguel Sousa Uva: Supervision; Validation; Writing—review & editing.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Jason M. Ali and the other, anonymous reviewer(s) for their contribution to the peer review process of this article.