Antibiotic prophylaxis for prevention of postoperative wound infection in adults undergoing open elective inguinal or femoral hernia repair

Abstract

Background

Inguinal or femoral hernia is a tissue protrusion in the groin region and has a cumulative incidence of 27% in adult men and of 3% in adult women. As most hernias become symptomatic over time, groin hernia repair is one of the most frequently performed surgical procedures worldwide. This type of surgery is considered 'clean' surgery with wound infection rates expected to be lower than 5%. For clean surgical procedures, antibiotic prophylaxis is not generally recommended. However after the introduction of mesh‐based hernia repair and the publication of studies that have high wound infection rates the debate as to whether antibiotic prophylaxis is required to prevent postoperative wound infections started again.

Objectives

To determine the effectiveness of antibiotic prophylaxis in reducing postoperative (superficial and deep) wound infections in elective open inguinal and femoral hernia repair.

Search methods

We searched several electronic databases: Cochrane Registry of Studies Online, MEDLINE Ovid, Embase Ovid, Scopus and Science Citation Index (search performed on 12 November 2019). We also searched two trial registers and the reference list of included studies.

Selection criteria

We included randomised controlled trials comparing any type of antibiotic prophylaxis versus placebo or no treatment for preventing postoperative wound infections in adults undergoing inguinal or femoral open hernia repair surgery (tissue repair and mesh repair).

Data collection and analysis

Two review authors independently selected studies, extracted data and assessed risk of bias. We separately analysed results for two different surgical methods (herniorrhaphy and hernioplasty). Several studies revealed infection rates that were higher than the expected 5% for clean surgery and we therefore divided studies into two subgroups: high infection risk environments (≥ 5% infection rate); and low infection risk environments (< 5% infection rate). We performed meta‐analyses with random‐effects models. We analysed three outcomes: superficial surgical site infections (SSSI); deep surgical site infections (DSSI); and all postoperative wound infections (SSSI + DSSI).

Main results

In this review update we identified and included 10 new studies. In total, we included 27 studies with 8308 participants in this review.

It is uncertain whether antibiotic prophylaxis as compared to placebo (or no treatment) prevents all types of postoperative wound infections after herniorrhaphy surgery (risk ratio (RR) 0.86, 95% confidence interval (CI) 0.56 to 1.33; 5 studies, 1865 participants; very low quality evidence). Subgroup analysis did not change these results. We could not perform meta‐analyses for SSSI or DSSI as these outcomes were not reported separately.

Twenty‐two studies related to hernioplasty surgery (total of 6443 participants) and we analysed three outcomes: SSSI; DSSI; SSSI + DSSI.

Within the low infection risk environment subgroup, antibiotic prophylaxis as compared to placebo probably makes little or no difference for the outcomes 'prevention of all wound infections' (RR 0.71, 95% CI 0.44 to 1.14; moderate‐quality evidence) and 'prevention of SSSI' (RR 0.71, 95% CI 0.44 to 1.17, moderate‐quality evidence). Within the high infection risk environment subgroup it is uncertain whether antibiotic prophylaxis reduces all types of wound infections (RR 0.58, 95% CI 0.43 to 0.77, very low quality evidence) or SSSI (RR 0.56, 95% CI 0.41 to 0.77, very low quality evidence). When combining participants from both subgroups, antibiotic prophylaxis as compared to placebo probably reduces the risk of all types of wound infections (RR 0.61, 95% CI 0.48 to 0.78) and SSSI (RR 0.60, 95% CI 0.46 to 0.78; moderate‐quality evidence).

Antibiotic prophylaxis as compared to placebo probably makes little or no difference in reducing the risk of postoperative DSSI (RR 0.65, 95% CI 0.26 to 1.65; moderate‐quality evidence), both in a low infection risk environment (RR 0.67, 95% CI 0.11 to 4.13; moderate‐quality evidence) and in the high infection risk environment (RR 0.64, 95% CI 0.22 to 1.89; low‐quality evidence).

Authors' conclusions

Evidence of very low quality shows that it is uncertain whether antibiotic prophylaxis reduces the risk of postoperative wound infections after herniorrhaphy surgery. Evidence of moderate quality shows that antibiotic prophylaxis probably makes little or no difference in preventing wound infections (i.e. all wound infections, SSSI or DSSI) after hernioplasty surgery in a low infection risk environment. In a high‐risk environment, evidence of very low quality shows it is uncertain whether antibiotic prophylaxis reduces all wound infections and SSSI after hernioplasty surgery. Evidence of low quality shows that antibiotic prophylaxis in a high‐risk environment may have little or no difference in reducing the risk of DSSI.

Plain language summary

Can antibiotics as compared to placebo prevent postoperative wound infections after open groin hernia repair surgery?

Background

Groin hernia is a weakness in the abdominal wall in the groin area, through which soft tissue or organs can protrude. Groin hernias occur often and therefore groin hernia repair is one of the most frequently performed surgical operations worldwide. It is considered a 'clean' surgical technique with low postoperative wound infections rates and administration of antibiotics to patients undergoing open hernia repair surgery is therefore not generally recommended. Up to the 1990s, suture‐based hernia repair (herniorrhaphy) was the method of choice. From then onwards hernia repair with a synthetic mesh (hernioplasty) gained increasing popularity and the debate as to whether antibiotics are needed to prevent postoperative wound infections started again.

Investigation

We searched the literature (12 November 2019) for randomised controlled trials comparing antibiotics versus placebo to prevent wound infections after open groin hernia repair surgery. We included both suture‐based and mesh‐type surgical methods. We divided infections into superficial and deep wound infections. Several studies revealed infection rates that were higher than the expected 5% for clean surgery. Therefore, we divided studies into a group with low infection rates (less than 5%) and one with high infection rates (more than 5%).

Study characteristics and key results

We identified five suture‐based surgery studies and 22 mesh‐type surgery studies. The suture‐based studies were of very low methodological quality. The mesh‐type surgery studies were of low to moderate methodological quality.

This review shows that antibiotics do not prevent the occurrence of any type of wound infections after suture‐based hernia repair. For mesh‐type hernia repair in a low infection risk environment, antibiotics probably make little to no difference in prevention of postoperative superficial wound infections. However, in a high infection risk environment it is uncertain whether antibiotics reduce the risk of superficial wound infection occurrence.

For deep wound infections, we show that antibiotics probably make little or no difference in reducing the risk in both a low and high infection risk environment.

Quality of the evidence

Evidence of very low quality shows that it is uncertain whether antibiotics reduce the risk of postoperative wound infections after suture‐based hernia repair. Evidence of moderate quality shows that that antibiotics probably make little or no difference in preventing superficial or deep wound infections after mesh‐type hernia repair in a low infection risk environment. Evidence of (very) low quality shows that it is uncertain whether antibiotics reduce the risk of superficial wound infections, and antibiotics have little or no effect on deep wound infections after mesh‐type hernia repair in a high infection risk environment.

Summary of findings

Summary of findings 1. Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open inguinal or femoral herniorrhaphy surgery.

Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open elective inguinal or femoral herniorrhaphy hernia repair
Patient or population: prevention of postoperative wound infection in adults undergoing open elective inguinal or femoral herniorrhaphy hernia repair Setting: Intervention: antibiotic prophylaxis Comparison: placebo
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with antibiotic prophylaxis
All wound infections (SSSI + DSSI)	Study population		RR 0.86 (0.56 to 1.33)	1865 (5 RCTs)	⊕⊝⊝⊝ VERY LOW 1 2
	49 per 1000	42 per 1000 (27 to 65)
All wound infections (SSSI + DSSI) ‐ low infection risk environment	Study population		RR 0.63 (0.28 to 1.41)	1302 (4 RCTs)	⊕⊝⊝⊝ VERY LOW 2 3
	32 per 1000	20 per 1000 (9 to 45)
All wound infections (SSSI + DSSI) ‐ high infection risk environment	Study population		RR 0.99 (0.58 to 1.68)	563 (1 RCT)	⊕⊝⊝⊝ VERY LOW 2 4
	89 per 1000	88 per 1000 (52 to 150)
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ 3 of 5 studies had high risk of bias for outcome or attrition bias.

² The confidence interval of the pooled effect size estimate includes both benefit and harm. Also, the optimal information size was not reached and sample size was small. Downgrade −2

³ All studies have unclear or high risk of bias for selection bias, detection bias or attrition bias

⁴ The risk of bias for selection bias and detection bias is unclear and attrition bias is high for this study

Summary of findings 2. Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open inguinal or femoral hernioplasty surgery.

Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open elective inguinal or femoral hernioplasty hernia repair
Patient or population: prevention of postoperative wound infection in adults undergoing open elective inguinal or femoral hernioplasty hernia repair Setting: Intervention: antibiotic prophylaxis Comparison: placebo
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with antibiotic prophylaxis
All wound infections (SSSI + DSSI)	Study population		RR 0.61 (0.48 to 0.78)	6443 (22 RCTs)	⊕⊕⊕⊝ MODERATE 1 2 3 4
	55 per 1000	33 per 1000 (26 to 43)
All wound infections (SSSI + DSSI) ‐ low infection risk environment	Study population		RR 0.71 (0.44 to 1.14)	3100 (9 RCTs)	⊕⊕⊕⊝ MODERATE 2 3 5 6
	26 per 1000	18 per 1000 (11 to 30)
All wound infections (SSSI + DSSI) ‐ high infection risk environment	Study population		RR 0.58 (0.43 to 0.77)	3343 (13 RCTs)	⊕⊝⊝⊝ VERY LOW 1 2 3 6 7
	85 per 1000	49 per 1000 (37 to 65)
SSSI	Study population		RR 0.60 (0.46 to 0.78)	6263 (21 RCTs)	⊕⊕⊕⊝ MODERATE 1 2 3 4
	50 per 1000	30 per 1000 (23 to 39)
SSSI ‐ low infection risk environment	Study population		RR 0.71 (0.44 to 1.17)	3100 (9 RCTs)	⊕⊕⊕⊝ MODERATE 2 3 5 6
	24 per 1000	17 per 1000 (11 to 28)
SSSI ‐ high infection risk environment	Study population		RR 0.56 (0.41 to 0.77)	3163 (12 RCTs)	⊕⊝⊝⊝ VERY LOW 1 2 3 6 7
	79 per 1000	44 per 1000 (32 to 61)
DSSI	Study population		RR 0.65 (0.26 to 1.65)	4185 (12 RCTs)	⊕⊕⊕⊝ MODERATE 1 2 3 4
	6 per 1000	4 per 1000 (2 to 10)
DSSI ‐ low infection risk environment	Study population		RR 0.67 (0.11 to 4.13)	1488 (3 RCTs)	⊕⊕⊕⊝ MODERATE 2 3 5 6
	4 per 1000	3 per 1000 (0 to 17)
DSSI ‐ high infection risk environment	Study population		RR 0.64 (0.22 to 1.89)	2697 (9 RCTs)	⊕⊕⊝⊝ LOW 1 2 3 6
	7 per 1000	4 per 1000 (2 to 13)
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ Excluding the studies with unclear and/or high risk of bias from the meta‐analysis affects the summary RR. Downgrade −1.

² Chi² test P > 0.05, I² < 60%, all confidence intervals of pooled effect estimates overlap. Clinical heterogeneity is limited as in similar patients are included, although different types and doses of antibiotics were applied. Evidence was not downgraded for clinical heterogeneity.

³ There are no indirect comparisons or surrogate outcomes used in the studies

⁴ The confidence interval of the pooled effect size includes both benefit and harm. Optimal information size was not reached, but evidence was not downgraded as the sample size is large.

⁵ Excluding the studies with unclear or high risk of bias hardly affects the summary RR, therefore not downgraded.

⁶ The confidence interval of the pooled effect size includes both benefit and harm. Optimal information size was not reached and sample size is small, therefore downgrade −1.

⁷ Asymmetrical forest plot reveals a under representation of studies that favour the control treatment.

Background

Description of the condition

An inguinal hernia is a viscera or adipose tissue protrusion through the inguinal or femoral canal. The lifelong cumulative incidence of inguinal and femoral hernia repair in adults is 27% to 42.5% for men and 3% to 5.8% for women, and increases with age (HerniaSurge Group 2018; Fitzgibbons 2015).

Groin hernias are inguinal or femoral; inguinal hernias are either direct or indirect or a combination. Direct and indirect hernias both protrude above the inguinal ligament; a direct hernia protrudes medially to the inferior epigastric vessels, whereas an indirect hernia protrudes laterally (Gilbert 1989; Zollinger 2003; Fitzgibbons 2015). A femoral hernia protrudes below the inguinal ligament and medially to the femoral vessels (Fitzgibbons 2015). In day‐to‐day surgical practice a classification system of inguinal hernia types is seldom used other than the types described here (HerniaSurge Group 2018). Inguinal or femoral hernias can affect one or both sides of the groin, resulting in a unilateral or bilateral hernia respectively (Fitzgibbons 2015).

In the late 19th century Bassini described for the first time the pathophysiology of the groin hernia, namely a weakness of the posterior inguinal wall (Bassini 1890). He developed a hernia repair technique involving suturing the weakness of the posterior wall (herniorrhaphy) (Bassini 1890). His repair led to tension, however, and was associated with high recurrence rates and postoperative pain. Therefore, from the mid‐20^th century onwards surgeons began to experiment with synthetic meshes (Kurzer 1998). From the mid‐ and late‐1980s tension‐free, mesh‐based (hernioplasty) methodologies were introduced for recurrent hernia and rapidly increased in popularity, as recurrence rates were lower, operation time was shorter, recovery time faster and surgery could be performed under local anaesthesia in a day care setting (Kurzer 1998; Robbins 1998). In 1989 Lichtenstein described the first large series (1000 participants) of mesh‐based repair in primary hernia (Lichtenstein 1989).

Inguinal and femoral hernia repair is worldwide one of the most commonly performed surgical operations, with more than 20 million procedures annually, making up approximately a third of total surgical interventions (Cainzos 1990; Kingsnorth 2003; Rutkow 2003; Rodriguez 2005). A minority of patients are asymptomatic, but even a watch‐and‐wait approach in this group results in surgery in approximately 70% within seven years (Fitzgibbons 2013). In many countries hernioplasty probably remains the first choice and most frequently applied surgical method in a majority of cases, although accurate and recent data are not available (HerniaSurge Group 2018). Inguinal and femoral hernia repairs are considered clean surgical procedures with a low risk (< 5%) of postoperative wound infection (Ortega 2012;Cai 2018). However, when infectious complications do occur following inguinal or femoral hernia repair, they can be a risk factor for developing a recurrent hernia (Finan 2005). Development of a deep‐mesh infection is rare, but is considered a serious complication as it usually requires mesh removal. Most studies report that (late) removal of the infected mesh usually does not result in hernia recurrence (Mann 1998; Fawole 2006; Johanet 2011; Rehman 2012).

postoperative wound infections can be divided into superficial surgical site infections (SSSI) and deep surgical site infections (DSSI) (Mangram 1999; Mandell 2010). SSSI is defined by the Centers for Disease Control and Prevention (CDC) as an infection occurring within 30 days after the operation and affecting only the skin or subcutaneous tissue at the incision site. Additionally, at least one of the following characteristics should apply: (1) the infection is characterised by purulent drainage; (2) positive microbiological culture can be grown from fluids or tissue obtained from the incision; (3) pain, tenderness, swelling, redness or heat in the wound area occurs and re‐opening of the wound by a surgeon is needed; (4) wound infection is diagnosed based on clinical experience of a surgeon of physician (Mangram 1999). DSSI is defined by the CDC as infection within one year after the operation if an implant is in place and the infection appears to be related to the operation. DSSI involves deep soft tissues (such as muscle or fascial layers). Additionally one of the following characteristics needs to be present: (1) purulent drainage from the deep incision; (2) spontaneous dehiscence or the wound is re‐opened by a surgeon for reasons of fever (> 38 °C), localised pain or tenderness; (3) an abscess or infection is found on direct or radiologic or histopathologic examination; (4) wound infection is diagnosed based on clinical experience of a surgeon of physician (Mangram 1999).

Wound infections are caused by bacterial contamination during surgery or the postoperative period (Mangram 1999). Upon microbiological culture of the wound exudates, most often normal nasal or skin bacteria such as Staphylococcus aureus or Staphylococcus epidermidis bacterial strains are detected (Mangram 1999; Mandell 2010). Wound Infection may result from a number of factors both intrinsic and extrinsic to the patient (Mangram 1999). Several factors that may increase the infection rate after hernia repair have been analysed (Berard 1964; Haley 1985; Wittmann 1995; Porcu 1996; Finan 2005). Although many of these intrinsic factors (such as the overall clinical status of the patient) cannot be modified, the external ones (such as operation room ventilation, pre‐operative shaving, length of operation time and experience of the surgeon) can certainly be influenced. In particular the factors related to aseptic conditions, surgical technique and perioperative care can be easily influenced (Mangram 1999). However, even under the most scrupulous aseptic conditions and with a careful technique, postoperative wound infection still presents a very serious problem. Antibiotic prophylaxis is therefore often used to decrease the risk of wound infections during surgical procedures (Mozillo 1988; Codina 1999; Heineck 1999; Mangram 1999; Mandell 2010; Bratzler 2013).

Description of the intervention

Antibiotic prophylaxis as an intervention to prevent surgical site infections and other infectious complications after surgery is recommended, especially in 'clean‐contaminated', 'contaminated' and 'dirty‐infected' surgical procedures (Mangram 1999; Bratzler 2013; Berrios‐Torres 2017). For 'clean' surgery, such as inguinal and femoral hernia repair, antibiotic prophylaxis is not generally recommended (Mangram 1999; Berrios‐Torres 2017). Antibiotic prophylaxis is recommended, however, when prosthetic material is being used or when risk factors are present (Condon 1991; Page 1993; Woods 1998; Mangram 1999; Simons 2009).

The most common surgical site infection pathogens after clean surgical procedures are Staphylococcus aureus or coagulase‐negative Staphylococci (S. epidermidis) bacteria (Mangram 1999; Mandell 2010). S. epidermidis bacteria belong to commensal skin bacterial flora and S. aureus bacteria are often present on (nasal and respiratory tract) mucosal tissues as well as on skin (Mandell 2010). To a lesser extent bacteria belonging to the commensal intestinal bacterial flora, such as Escherichia coli, are detected in wound infection exudates (Mangram 1999; Mandell 2010; Bratzler 2013). Therefore antibiotics such as cefazolin or related cephalosporin class of antibiotics that kill these bacterial species are recommended for hernia repair (Mangram 1999; Mandell 2010; Bratzler 2013).

Severe adverse reactions to beta‐lactam antibiotics are relatively rare (Lagace‐Wiens 2012), but side effects such as diarrhoea, nausea/vomiting, rash, fatigue, itching and urticaria have more frequently been reported (Lagace‐Wiens 2012). The cost of antibiotic prophylaxis is low, but much concern has been raised about the development of antibiotic‐resistant bacteria when antibiotics are used too frequently or inappropriately (Mangram 1999; Mandell 2010; Chokshi 2019).

How the intervention might work

Different types of antibiotics, bactericidal and bacteriostatic, have different mechanisms of action but they respectively kill bacteria by interfering with bacterial cell wall synthesis or by preventing cell division by inhibiting protein synthesis (Mandell 2010).

The type of antibiotics used for prophylaxis during 'clean' surgical procedures belong to the beta‐lactam bacteriocidal antibiotics type and more specifically to the cephalosporin‐group (i.e. cefuroxime and cefazolin) or the penicillin‐group (i.e. penicillin) antibiotics (Mandell 2010). These antibiotics are effective against bacterial species that are most often detected in infected surgical wounds(S. aureus, S. epidermidis or E. coli). Also, with a half‐lifetime of 1 to 2 hours, a single dose of these antibiotics will supply a therapeutic level until approximately 3 to 7 hours after wound closure. (Mangram 1999; Bratzler 2013).

The use of antibiotic prophylaxis in a perioperative setting will reduce endogenous patient bacteria and bacteria that colonise patients' exposed tissues during surgery. Additionally, antibiotics could kill bacteria present on the implanted prosthetic material during hernioplasty surgery (Mandell 2010). Depending on the overall aseptic operative (room) conditions, poor conditions are likely to pose a higher risk for patients to develop surgical site infections (Mangram 1999; Mandell 2010). Altogether, antibiotic prophylaxis might prevent surgical site infections and other infectious complications.

Why it is important to do this review

Inguinal and femoral hernia repair are considered to be 'clean' surgical procedures, with an estimated postoperative wound infection rate of 1% to 5% (Mangram 1999; Ortega 2012; Bratzler 2013). Some studies have even estimated that the rate of postoperative infections should not be greater than 2% (Condon 1991; Page 1993; Rutkow 1993; Dellinger 1994; Woods 1998). In line with this, several randomised studies and systematic reviews reported wound infection rates after open hernia repair surgery ranging from 1.4% to 4.1% (Platt 1990; Morales 2000; Aufenacker 2004; Jain 2008; Cai 2018; Finch 2019). However, some randomised studies report wound infection rates exceeding 5% wound infection incidence with infection rates ranging up to 18.1% (Ullah 2013).

For suture‐based inguinal and femoral hernia repair, antibiotic prophylaxis is not generally recommended in the absence of risk factors (Mangram 1999; Berrios‐Torres 2017). Antibiotic prophylaxis is recommended for hernia repair involving the use of prosthetic material or when risk factors are present (Condon 1991; Page 1993; Woods 1998; Mangram 1999; Simons 2009; Berrios‐Torres 2017). There is an ongoing debate whether antibiotic prophylaxis is of added value for preventing infectious complications when prosthetic materials are implanted during 'clean' surgical procedures (Stephenson 2003; Biswas 2005; Aufenacker 2006; Fawole 2006). Controversy especially arises when wound infection rates exceed the expected figures (Bailey 1992; Ranaboldo 1993; Holmes 1994; Wittmann 1995; Leaper 1998). Many parameters during surgery, such as length of operation, aseptic conditions, operation room ventilation, expertise of the surgeon etc., can affect postoperative wound infection rates. The effect of one of these surgery‐related parameters, or the combined effect of more than one, results in the infection pressure for a specific clinical centre or study. The wound infection incidence in the control group is a reflection of the effect of single parameters combined and can therefore be regarded as a composite parameter for overall infection pressure. To address the issue of wound infection rates below or exceeding the expected 5% infection incidence for clean surgery, we stratified studies into high infection risk environments or low infection risk environments based on the wound infection incidence in the control group.

As with any medical drug treatment, antibiotic prophylaxis has side effects, including adverse reactions and allergic reactions. Moreover, the use of antibiotics can result in the development of antibiotic‐resistant bacterial strains (Mandell 2010; Chokshi 2019). By reducing the use of unnecessary antibiotic prophylaxis, adverse events and development of antibiotic‐resistant microbiological strains can be reduced.

Objectives

To determine the effectiveness of antibiotic prophylaxis in reducing postoperative (superficial and deep) wound infections in elective open inguinal and femoral hernia repair.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) comparing antibiotic prophylaxis to placebo or no prophylaxis for open inguinal or femoral hernia repair surgery.

Types of participants

Adult male or female participants undergoing open elective inguinal or femoral hernia repair, with or without the use of prosthetic material. Studies for all types of inguinal or femoral hernia (unilateral, bilateral, primary or recurrent) were eligible for inclusion. We excluded studies addressing laparoscopic inguinal or femoral hernia repair surgery from this review.

Types of interventions

Studies comparing antibiotic prophylaxis of any type with placebo or no treatment in participants undergoing open inguinal or femoral hernia repair surgery. We also included studies that compared antibiotic prophylaxis against placebo for so‐called 'clean wound' surgery in general, if data specifically for inguinal or femoral hernia could be extracted. We excluded studies in which antibiotic prophylaxis was compared with another antibiotic regimen, combinations of antibiotics, or studies comparing different doses or administration routes of antibiotics. We did not include studies using antiseptics of the incision area instead of antibiotic prophylaxis in the review.

Types of outcome measures

Primary outcomes

Wound infection after inguinal or femoral repair surgery. We classified wound infections as superficial surgical site infections (SSSI), deep surgical site infections (DSSI) or all wound infections (a sum total of SSSI and DSSI). SSSI can occur up to 30 days postoperatively and DSSI can occur up to one year postoperatively (Mangram 1999). We included studies reporting wound infections as defined by Centers for Disease Control and Prevention (CDC) (Horan 1992; Berrios‐Torres 2017); or as defined specifically by authors (definitions including parameters such as discharge of pus from the wound; a wound that was opened and not closed; spreading erythema indicative of cellulitis). We excluded other postoperative infections (i.e. pulmonary, urinary tract and other infections) from analysis.

Secondary outcomes

Any adverse event attributable to the use of antibiotics, including allergic reactions of participants. We accepted any definition of adverse event by the primary authors. We also analysed the reporting of development of antibiotic resistance due to antibiotic prophylaxis.

Search methods for identification of studies

Electronic searches

For this update we searched the following electronic databases, with no restriction on language or publication date, on 12 November 2019.

1. Cochrane Registry of Studies Online (CRSO.cochrane.org) (Appendix 1)

2. MEDLINE Ovid (1946 to 12 November 2019) (Appendix 2)

3. Embase Ovid (1946 to 12 November 2019) (Appendix 3)

4. Scopus (2011 to 2019) (Appendix 4)

5. Science Citation Index (Web of Science) (2011 to 2019) (Appendix 5)

The MEDLINE search strategy that was used in previous versions of this review can be found in Appendix 6. In the previous updates, the searches were performed until June 2009 and October 2011.

Searching other resources

We contacted the Dutch inguinal hernia guideline panel of medical specialists for information about any additional completed studies. We checked the reference lists of the included studies to identify further studies.

We searched prospective trial registers with the following search terms: (inguinal hernia OR groin hernia) AND (antibiotic OR antibiotics) (search date 12 November 2019).

1. ClinicalTrials.gov

2. World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/en)

Data collection and analysis

Selection of studies

Two review authors (CO and CvH) independently assessed all titles and abstracts that were retrieved in the literature search in Covidence software. We included studies when study design, intervention and type of participants met the inclusion criteria for this review. We resolved disagreements on the inclusion or exclusion of studies by discussion.

Data extraction and management

Two authors (CO and FJSM) developed a digital data extraction form, which was piloted by two authors (CO and CvH) on three studies to ensure that data extraction was performed in a consistent manner. Data extraction was performed by pairs of authors (from CO, CvH and FJSM) independently. We resolved disagreements by discussion or by involving another author (RS) as arbiter. When we considered a publication to be eligible but data in the publication were incomplete, we contacted the principal author to obtain the necessary information. We subsequently entered extracted data into Review Manager 5 (RevMan 5) software (Review Manager 2014).

We extracted the following data.

1. General study information, including first author and year of publication.

2. Study methods, including study design, country and enrolment period of the participants.

3. Participant characteristics, including age and sex distribution, body mass index (BMI), American Society of Anesthesiologists (ASA) status of the participants, inclusion and exclusion criteria, and specific information on the surgical or hospitalisation procedure (including method of repair, skin antiseptic, operative time, type of anaesthesia, drain use, surgeons' expertise, hospitalisation method and length of stay).

4. Intervention information, including type of antibiotic (or placebo), dose, number of doses, administration route, and administration timing.

5. Outcome information, including the wound infection definition that was applied, duration of follow‐up, number of participants with a wound infection per total number of participants in both the intervention (antibiotic prophylaxis) and control (placebo) group for any superficial or deep surgical wound infection (SSSI, DSSI).

6. Other relevant information, including details on which bacterial species were cultured from the infection sites or adverse effects.

Assessment of risk of bias in included studies

Pairs of review authors (CO and FJSM or CO and CvH) independently assessed the risk of bias for each included study. We resolved disagreements by discussion or by involvement of an independent author (RS) as arbiter. We assessed all included studies for risk of bias based on the criteria as described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (version 5.2.0) (Higgins 2017).

We assessed the following risk of bias domains: random sequence generation (selection bias); allocation concealment (selection bias); blinding of participants and personnel (performance bias); blinding of outcome assessment (detection bias); incomplete outcome data (attrition bias); selective reporting (reporting bias); and other bias. In addition to the assessment criteria described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (version 5.2.0) (Higgins 2017), we judged random sequence generation as low risk of bias when block randomisation was reported, as block randomisation indicates that authors have considered the importance of randomisation to prevent selection bias. We judged studies that reported that allocation concealment involved (numbered) sealed envelopes as unclear bias as the information on sequentially numbered envelopes and opaqueness was not reported. We judged studies that were terminated early after interim analysis of the results as high risk of bias for the domain 'other bias'.

We considered a study at low risk of bias if we assessed the domains 'selection bias', 'performance bias' and 'detection bias' as low risk and none of the other domains was at high risk. We contacted authors when there were any questions regarding risk of bias in their publication. If we did not succeed in retrieving additional information from study authors we had to make our final judgement of methodological quality on the basis of the available information.

Measures of treatment effect

We analysed the data using Cochrane Review Manager 5 software (Review Manager 2014). We used a Mantel‐Haenszel random‐effects model to compute risk ratios (RRs) with 95% confidence intervals (CIs) for dichotomous outcomes.

Unit of analysis issues

We used the number of individual participants as the unit of analysis. All trials were parallel in design. We combined participants from both intervention groups for meta‐analyses that included studies that compared more than one intervention with the same control group.

Dealing with missing data

For each study we extracted the number of participants originally allocated to each treatment group and performed intention‐to‐treat analysis, or available‐case analysis when participants were lost to follow‐up.

Assessment of heterogeneity

We assessed clinical diversity among the included studies by analysing differences in participants and interventions, as well as outcome measurements and length of follow‐up period. We first performed a visual inspection of forest plots. Then we used both the Chi² test for heterogeneity and the I² statistic to test for heterogeneity between study results. We set the P value for the Chi² test at a conservative 0.1; and we considered I² values of 50% to 90% as indicating substantial heterogeneity.

Assessment of reporting biases

We obtained fewer than 10 studies for the meta‐analysis of herniorrhaphy repair studies, so assessment of reporting bias by funnel plot analysis was not possible. We performed a funnel plot analysis for hernioplasty groin hernia repair studies for the outcomes 'all wound infection' (SSSI + DSSI) and for 'superficial surgical site infections' (SSSI) for both infection risk subgroups.

Data synthesis

We pooled dichotomous outcomes from studies by meta‐analysis and used a random‐effects model to analyse data. We gathered results from herniorrhaphy studies separately from hernioplasty studies. We used Review Manager 5 to generate forest plots and additionally summarised results narratively (Review Manager 2014). We used GRADEpro software to summarise findings and to present the quality of evidence in the 'Summary of findings' tables (GRADE pro GDT).

Subgroup analysis and investigation of heterogeneity

We performed separate analyses for the various categories of surgery that were performed (i.e. herniorrhaphies or hernioplasties). Inguinal or femoral hernia repair is considered a clean surgical procedure with wound infection rates of 5% or lower (Mangram 1999; Ortega 2012). A number of studies reported wound infection rates that exceeded the generally accepted rate. To address this issue, we stratified studies into high infection risk environments (≥ 5% infection incidence) or low infection risk environments (< 5% infection incidence) based on the wound infection incidence in the control group. Many parameters during surgery, such as length of operation, aseptic conditions, operation room ventilation, expertise of the surgeon etc. can affect postoperative wound infection rates. The effect of one or a combination of these single surgical‐related parameters results in the observed infection pressure for a specific clinical centre or study. The wound infection incidence in the control group is a reflection of the effect of single parameters combined and can therefore be regarded as a composite parameter for overall infection pressure.

Sensitivity analysis

We performed sensitivity analysis by excluding studies at high risk of bias or studies with more than 10% missing outcome data from the meta‐analysis.

Results

Description of studies

Results of the search

We identified 430 studies by searching the primary electronic databases (140 in MEDLINE, 78 in Embase and 85 in the Cochrane Registry of Studies Online; the remaining studies in Scopus and Web of Science). After removal of duplicates, 285 studies were left. We screened the titles and abstracts of these studies, and selected 46 for full‐text review. Reasons for excluding studies are provided In Characteristics of excluded studies. We finally included 27 studies, reported in 28 publications, that met the inclusion criteria for both the qualitative and quantitative synthesis (meta‐analysis). Lazorthes 1992 was reported in two publications (one in French and one in English), hence we discarded Lazorthes 1993. We present the PRISMA study flow chart in Figure 1.

1.

Study flow diagram.

Seventeen RCTs were included in the meta‐analyses in the previous version of this review (Sanchez‐Manuel 2012).

We searched two prospective trial registers and retrieved five study protocols corresponding to four publications (Ergul 2011; Rahmani 2012; Mazaki 2013; Wang 2013). Additionally, we retrieved one relevant study protocol of which no publication was found. We have added this study to the Characteristics of studies awaiting classification.

Included studies

We present detailed information of the 27 included studies in Characteristics of included studies. Of those, 24 were published in English; two in Spanish (Morales 2000; Oteiza 2004); and one in Farsi (Rahmani 2012).

Five RCTs used herniorrhaphy as surgical procedure (Evans 1973; Andersen 1980; Platt 1990; Lazorthes 1992; Taylor 1997); and 22 RCTs used hernioplasty as surgical procedure for inguinal or femoral hernia repair. The studies using herniorrhaphy surgical procedure were published in the period 1973 to 1997, while the studies applying hernioplasty surgical procedures were published in the period from 2004 to 2015.

We included a total of 8308 participants in this systematic review, of whom 1865 underwent herniorrhaphy surgery. Within this group, 922 participants received antibiotic prophylaxis and 943 received either no antibiotics or placebo. In the hernioplasty group 6443 participants were included; 3399 received antibiotic prophylaxis and 3044 received either no antibiotics or placebo.

Treatment characteristics of the studies

The treatment characteristics of the included herniorrhaphy studies are described in Table 3 and those of the included hernioplasty studies in Table 4. Further details are described in Characteristics of included studies.

1. Overview of treatment characteristics of the included herniorrhaphy studies.

	Antibiotic	Dose (g)	Administration route	Follow‐up period	Operative time (minutes); (mean ± SD or median (range))	Infection risk environment (Infection percentage in control group)	Remark
Andersen 1980	AMP	1	in fascia	1 m, 3 m, 6 m, 12 m	not reported	Low (4.0%)	Several surgery types performed
Evans 1973	CLR	1 (3×)	intravenous (1x)/ intramuscular (2x)	1 m	not reported	Low (4.1%)	Several surgery types performed
Lazorthes 1992	CAM	0.75	subcutaneous (added to local anaesthesia)	1 m	not reported	Low (4.6%)
Platt 1990	CON	1	intravenous	1 w, 4 to 6 w	PG: 75 ± 32 / CG: 75 ± 30	Low (1.9%)	Several surgery types performed
Taylor 1997	AMC	1.2	intravenous	4 to 6 w	not reported	High (8.9%)

PG = prophylactic group, CG = control group

AMC = Amoxicillin‐clavoulanic acid, AMP = ampicillin, CAM = cefamandole, CON = cefonicid, CLR = cephaloridine (Abbreviations according to EUCAST System for Antimicrobial Abbreviations)

w = week, m = month, y = year

Infection risk environment (L = low; ≤ 5% infections in control group for all wound infections (SSSI + DSSI), H = high; > 5% infections in control group for all wound infections (SSSI + DSSI) ).

2. Overview of treatment characteristics of the included hernioplasty studies.

	Antibiotic	Dose (g)	Administration route	Follow‐up period	Operative time (minutes); (mean ± SD or median (range))	Infection risk environment (Infection percentage in control group)
Al‐Fatah 2011	AMC	1.2	intravenous	1 w/1 m	PG: 45 (20 to 90); CG:45 (20 to 80)	High (5.0%)
Aufenacker 2004	CUR	1.5	intravenous	1 w, 2 w, 3 m	PG: 40 (IQR 30 to 50); CG 40 (IQR 28 to 51)	Low (1.8%)
Bidhur 2013	CUR	1.5	intravenous	1 w, 1 m	51.3 ± 9.8 (range 40 to 75 min) (overall population) /Time<50 min: PG: n = 19 (63%); CG: n = 15 (50%)	Low (3.3%)
Celdran 2004	CZO	1	not reported	1 w, 1 m, 3 m, 6 m, 1 y, 2 y	PG: 65 ± 23; CG: 64 ± 14	High (8.2%)
Ergul 2011	CZO	1	intravenous	1 to 6 d, 1 m	PG: 60 (35 to 160); CG: 60 (40 to 135)	High (7.0%)
Goyal 2011	AMC	1.2	intravenous	1 w	not reported	Low (3.0%)
Ijaz 2010	CZO	1	intravenous	1 w, 2 w, 1 m	not reported	High (10.0%)
Jain 2008	AMC	1.2	intravenous	1 w, 2 w, 1 m, 1 y	PG: 56.33 ± 11.67; CG: 60.33 ± 6.81	Low (1.7%)
Kochhar 2014	AMC	1.2	intravenous	1 w, 1 m	not reported	Low (4.7%)
Mazaki 2013	CZO	1	intravenous	1 w, 1 m, 3 m	PG: 66.3 ± 25.4; CG: 65.2 ± 27.1	High (13.0%)
Morales 2000	CZO (or ERY)	2 (or 1)	intravenous	1 w, 1 m, 1 y	not reported	Low (2.1%)
Oteiza 2004	AMC	2	intravenous	1 w, 1 m	PG and CG: Mean 40 min	Low (0%)
Othman 2011	AMC	1.2	intravenous	1 w, 1 m	PG 38.8 ± 10.8; CG 40.9 ± 11.1	High (12,5%)
Perez 2005	CZO	1	intravenous	1 w, 2 w, 1 m	PG 52.18 ± 16.4; CG 54.07 ± 15.3	Low (3.9%)
Rahmani 2012	CLT	1	intravenous	12 w	not reported	High (6.4%)
Razack 2015	CZO	1	intravenous	1 w, 1 m	PG: 53.54 ± 15.82; CG 52.60 ± 15.28	High (9.3%)
Shankar 2010	CZO	1	intravenous	1 w, 1 m	PG 53.54 ± 15.82; CG 52.60 ± 15.28	High (10.5%)
Thakur 2010	CZO	1.5	intravenous	1w,1m	PG: 79.3% (n = 23) had the total duration of surgery > 1 hour; CG: 92.3% (n = 24) had the total duration of surgery > 1 hour.	High (15.4%)
Tzovaras 2007	AML	1.2	intravenous	1 w, 1 m	PG: 45 (20 to 90); CG: 45 (20 to 80)	Low (4.8%)
Ullah 2013	AMC	1	not reported	2 d, 2 w	not reported	High (18.1%)
Wang 2013	CZO or LEV	1 or 0.2	intravenous	1 w, 2 w, 3 w, 1 m	not reported	High (5.1%)
Yerdel 2001	AMS	1.5	intravenous	1 w, 4 to 6 w, 6 m, 1 y	PG: 64.18 ± 22.8; CG: 62.78 ± 19.3	High (9.0%)

PG = prophylactic group, CG = control group

AMC = Amoxicillin‐clavoulanic acid, AML = ampicillin‐clavulanic acid, AMS = ampicillin‐sulbactam, CLT = cephalotin, CUR = cefuroxime, CZO = cefazolin, ERY = erythromycin, LEV = Levofloxacin (Abbreviations according to EUCAST System for Antimicrobial Abbreviations)

w = week, m = month, y = year

Infection risk environment (L = low; ≤ 5% infections in control group for all wound infections (SSSI + DSSI), H = high; > 5% infections in control group for all wound infections (SSSI + DSSI) ).

In the herniorrhaphy studies, all studies used different antibiotics as prophylaxis (Table 3). In the hernioplasty studies (Table 4) the two antibiotics that were most frequently applied were amoxicillin‐clavulanic acid (AMC; Oteiza 2004; Jain 2008; Al‐Fatah 2011; Goyal 2011; Othman 2011; Ullah 2013; Kochhar 2014); and cefazolin (CZO; Morales 2000; Celdran 2004; Perez 2005; Ijaz 2010; Shankar 2010; Thakur 2010; Ergul 2011; Mazaki 2013; Wang 2013; Razack 2015). Other applied antibiotics are summarised in Table 4. One study compared two different antibiotics against placebo (Wang 2013). We combined results from both antibiotic prophylaxis intervention groups in the meta‐analysis.

All four studies administered antibiotic prophylaxis intravenously. Andersen 1980 applied antibiotics subfascially before closure; Evans 1973 applied the first dose of antibiotics intravenously, but the second and third dose intramuscularly; Lazorthes 1992 added antibiotics to local anaesthesia solution; and Ullah 2013 did not specify the route of administration.

The follow‐up period in the herniorrhaphy studies ranged from one week in Platt 1990 to four to six weeks (Evans 1973; Andersen 1980; Platt 1990; Lazorthes 1992; Taylor 1997). Only one study had a follow‐up of one year to investigate hernia recurrences (Andersen 1980). The follow‐up period reported in the hernioplasty studies ranged from several days or one week postoperatively to two to six weeks. A limited number of studies had follow‐up time points at three months (Aufenacker 2004; Celdran 2004; Mazaki 2013), six months (Yerdel 2001; Celdran 2004) or one year (Morales 2000; Celdran 2004; Jain 2008). Goyal 2011 was the only study with a single follow‐up time point at one week postoperatively.

Four herniorrhaphy studies did not report the length of operative time (Evans 1973; Andersen 1980; Lazorthes 1992; Taylor 1997); nor did seven hernioplasty studies (Morales 2000; Ijaz 2010; Goyal 2011; Rahmani 2012; Ullah 2013; Wang 2013; Kochhar 2014). The remaining herniorrhaphy study reported a mean operation length of more than one hour (Platt 1990). Eleven hernioplasty studies reported a mean (or median) length of operative time equal to or less than one hour (Aufenacker 2004; Oteiza 2004; Perez 2005; Tzovaras 2007; Jain 2008; Shankar 2010; Al‐Fatah 2011; Ergul 2011; Othman 2011; Bidhur 2013; Razack 2015); and three studies reported a mean (or median) of more than one hour (Yerdel 2001; Celdran 2004; Mazaki 2013), with Thakur 2010 reporting that 79% to 92% of the operations had a total duration of more than one hour.

Three studies investigated the effectiveness of antibiotic prophylaxis for the prevention of wound infections not only in inguinal or femoral hernia repair, but also in several other types of surgical procedures: cholecystectomy (Andersen 1980); clean and contaminated surgery (Evans 1973); and breast surgery (Platt 1990).

Wound infection rates in the control group (without antibiotic prophylaxis) ranged widely, from 0% to 18.1%. Nine studies were considered low infection risk environments (Morales 2000; Aufenacker 2004; Oteiza 2004; Perez 2005; Tzovaras 2007; Jain 2008; Goyal 2011; Bidhur 2013; Kochhar 2014); and 13 studies were considered high‐risk environments (Yerdel 2001; Celdran 2004; Ijaz 2010; Shankar 2010; Thakur 2010; Al‐Fatah 2011; Ergul 2011; Othman 2011; Rahmani 2012; Mazaki 2013; Ullah 2013; Wang 2013; Razack 2015).

Participant characteristics

The participant characteristics of the included herniorrhaphy studies are described in Table 5 and of the included hernioplasty studies in Table 6. Further details are described in Characteristics of included studies.

3. Overview of participant characteristics of the included herniorrhaphy studies.

	% male	Age (years) in mean ± SD or median (range)	Participants age < 18 y included	Inguinal (I) or femoral (F) hernia	ASA class	BMI mean ± SD or median (range)
Andersen 1980	PG: 72; CG:73	PG: 54 to 56 years (range 21 to 82); CG: 55 years (range 20 to 82)	No	I + F	Not reported	Not reported
Evans 1973	not reported	not reported	not reported	I + F	Not reported	Not reported
Lazorthes 1992	PG: 90; CG: 88	PG: 62 (11 to 90)/ CG: 70 (16 to 92)	Yes	I + F	Not reported	Not reported
Platt 1990	PG: 91; CG: 89	PG: 51.0 ± 17.0/ CG: 49.8 ± 17.6	No	I + F	Not reported	PG 24.6 ± 3.2; CG 24.7 ± 3.2
Taylor 1997	PG: 95; CG: 95	PG: 56.7 ± 17.4 / CG: 56.6 ± 16.5	No	I + F	Not reported	Not reported

PG = prophylactic group, CG = control group

4. Overview of participant characteristics of the included hernioplasty studies.

	% male	Age (years) in mean ± SD or median (range)	Participants age < 18 y included	Inguinal (I) or femoral (F) hernia	ASA class	BMI mean ± SD or median (range)
Al‐Fatah 2011	PG: 94; CG: 96	PG: 63 (17 to 87); CG: 63 (15 to 90)	Yes	I	I; II; III	PG: 26 (18 to 34); CG: 26 (20 to 33)
Aufenacker 2004	PG: 96; CG: 97	PG: 58.3 ± 12.9; CG: 58.2 ± 13.2	No	I	Not reported	Not reported
Bidhur 2013	PG: 100; CG: 97	38.5 ± 17.7 (range: 19 to 90 years) (overall study population)	No	I	Not reported	Not reported
Celdran 2004	PG: 94; CG: 86	PG: 58 ± 13; CG: 58 ± 17	No	I	I; II	PG: 26.1 ± 5/ CG: 26.2 ± 5
Ergul 2011	PG: 95; CG: 89	PG: 48 ± 17; CG: 50 ± 15	No	I	I; II; III; IV	Not reported
Goyal 2011	NR	range 11 to 90 (overall study population)	Yes	I	Not reported	Not reported
Ijaz 2010	99 (overall)	PG: 44.06; CG: 44.84	No	I	Not reported	Not reported
Jain 2008	PG: 100; CG:100	PG: 41.28 ± 11.49; CG: 40.2 ± 9.84	No	I	I/ II	Not reported
Kochhar 2014	96 (overall)	PG: 37.42 ± 9.9; CG: 37.42 ± 11.5	Yes	I	Not reported	Not reported
Mazaki 2013	PG: 89/ CG: 94	PG: 69 (57 to 76); CG: 72 (60 to 77)	No	I	I; II; III	PG: 23.0 ± 2.8 PG; CG: 22.7 ± 3.0
Morales 2000	89 (overall)	54.2 (17 to 87) (overall study population)	No	I+F	I; II; III	Not reported
Oteiza 2004	PG: 89; CG: 82	PG: 58 (22 to 91)/ CG: 56.2 (17 to 88)	Yes	I+F	I; II; III	Not reported
Othman 2011	PG: 96; CG: 100	PG: 43.4 ± 19.8/ CG: 44.5 ± 20.5	No	I	Not reported	Not reported
Perez 2005	PG: 98; CG: 98	PG: 61.37 ± 13.2/ CG: 60.8 ± 14.5	No	I	I; II	Not reported
Rahmani 2012	PG: 45; CG: 41	25 to 84 (overall population)	No	I	not reported	Not reported
Razack 2015	PG: 99; CG: 100	PG: 42.44 ± 15.61 / CG: 45.56 ± 15.43	Yes	I	I; II	Not reported
Shankar 2010	PG: 99; CG: 98	PG: 44.44 ± 15.59 / CG: 45.56 ± 16.43	Yes	I	I; II	Not reported
Thakur 2010	PG: 100; CG: 100	Not reported	No	I	Not reported	Not reported
Tzovaras 2007	PG: 94; CG: 94	PG: 63 (17 to 87) / CG: 63 (15 to 90)	Yes	I	I; II; III	PG: 26 (18 to 34); CG: 26 (20 to 33)
Ullah 2013	PG: 100; CG: 100	PG: 54.33 ± 11.77 / CG: 52.58 ± 11.80	No	I	Not reported	Not reported
Wang 2013	PG group 1: 90; PG group 2: 91; CG: 90	PG group 1: 53 ± 16.9; PG group 2: 54.6 ± 16.2/ CG: 56 ± 17.3	Yes	I	I, ≥ I	18 to 32 (overall)
Yerdel 2001	PG: 90; CG: 95	PG: 55.57 ± 15.1; CG: 55.78 ± 13.8	No	I	I; II	PG: 24.95 ± 2.6; CG: 25.02 ± 3.0

PG = prophylactic group, CG = control group

Most study participants were male (ranging from 72% to 99%), with three studies including only males (Jain 2008; Thakur 2010; Ullah 2013). The only exception was the study of Rahmani 2012, that included 41% to 45% males.

Mean (or median) age in the herniorrhaphy studies ranged from 50 to 51 years (Platt 1990) and from 62 to 70 years (Lazorthes 1992). Age of the participants was not separately reported for the hernia repair group in one study (Evans 1973). In the hernioplasty studies the mean or median age ranged from younger than 50 years (Jain 2008; Ijaz 2010; Shankar 2010; Ergul 2011; Othman 2011; Bidhur 2013; Kochhar 2014; Razack 2015) up to 72 years (Mazaki 2013). Thakur 2010 did not report any age characteristics of participants. Nine studies also included participants aged younger than 18 years (Lazorthes 1992; Oteiza 2004; Tzovaras 2007; Shankar 2010; Al‐Fatah 2011; Goyal 2011; Wang 2013; Kochhar 2014; Razack 2015).

Known risk factors for wound infection development (Finan 2005; Mandell 2010), such as high American Society of Anesthesiologists (ASA) score or high body mass index (BMI) of participants were not consistently reported. Herniorrhaphy studies did not report these characteristics, except the study of Platt and colleagues that reported the BMI of the participants (Platt 1990).

Of the hernioplasty studies, nine did not report any ASA score information (Aufenacker 2004; Ijaz 2010; Thakur 2010; Goyal 2011; Othman 2011; Rahmani 2012; Bidhur 2013; Ullah 2013; Kochhar 2014); and 16 did not report BMI information (Morales 2000; Aufenacker 2004; Oteiza 2004; Perez 2005; Jain 2008; Ijaz 2010; Shankar 2010; Thakur 2010; Ergul 2011; Goyal 2011; Othman 2011; Rahmani 2012; Bidhur 2013; Ullah 2013; Kochhar 2014; Razack 2015). Most studies included participants with a low (I/II) ASA score (Yerdel 2001; Celdran 2004; Perez 2005; Jain 2008; Shankar 2010; Razack 2015), few studies also included ASA III score participants (Morales 2000; Tzovaras 2007; Al‐Fatah 2011; Mazaki 2013) or ASA IV score participants (Ergul 2011). Wang 2013 did not specifically address the score, but reported that participants with an ASA score of greater than 1 were included in their study.

The BMI in the studies ranged from 23 kg/m² to 26 kg/m² (Yerdel 2001; Celdran 2004; Tzovaras 2007; Al‐Fatah 2011; Mazaki 2013), with the exception of Wang 2013 in which a range of 18 kg/m² to 32 kg/m² was reported.

Excluded studies

Studies were excluded for the following reasons (see also Characteristics of excluded studies).

1. Studies focused on clean surgical techniques including inguinal or femoral hernia pathology, but data for this subgroup of patients could not be collected (Esposito 2006; Mehrabi Bahar 2015).

2. Study design not eligible: Gierhake 1975, Taylor 1996, Barreca 2000, Sanchez‐Manuel 2003, Aufenacker 2006, Pessaux 2006 (prognostic study based on three randomised studies that compared antibiotics prophylaxis versus another antibiotic prophylaxis regime), Sanabria 2007, Leon 2011 (non‐randomised study), Li 2012; Sanchez‐Manuel 2012, Yin 2012 (systematic reviews), Mazaki 2013b, Erdas 2016, Boonchan 2017.

3. Comparator not eligible (antibiotics prophylaxis versus antibiotics prophylaxis with another regime): Pessaux 2006, Bhuiyan 2017.

4. Article in French (Lazorthes 1993) of similar study described previously in English (Lazorthes 1992)

5. The results section and large text fragments of the study of Ahmed 2014 were similar to those of the study of Goyal 2011. We therefore considered that Ahmed 2014 was plagiarised and excluded it from analysis.

Risk of bias in included studies

A summary of the risk of bias is presented in Figure 2. Detailed information on risk of bias assessment, including substantiation for all risk of bias categories for each individual study can be found in the 'Risk of bias' tables of the Characteristics of included studies.

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Six studies provided no (or insufficient) information on random sequence generation (Andersen 1980; Lazorthes 1992; Ijaz 2010; Shankar 2010; Rahmani 2012; Razack 2015) and we therefore judged them at 'unclear risk of bias'. Allocation concealment was insufficiently reported in the majority (19 of 27) of studies and we therefore scored them as 'unclear risk of bias'. Eight studies scored 'low risk of bias' for both random sequence generation and allocation concealment (Morales 2000; Yerdel 2001; Aufenacker 2004; Perez 2005; Jain 2008; Othman 2011; Mazaki 2013; Wang 2013).

Blinding

Twelve studies reported the blinding for participants and personnel (Platt 1990; Morales 2000; Yerdel 2001; Aufenacker 2004; Perez 2005; Tzovaras 2007; Jain 2008; Al‐Fatah 2011; Ergul 2011; Othman 2011; Mazaki 2013; Wang 2013). The other 15 studies provided no or insufficient information on participant or personnel blinding, or both. However, we scored all studies as 'low risk of bias' as we judged that participant and personnel blinding in this setting was unlikely to contribute to performance bias.

In contrast to participant and personnel blinding, the blinding of outcome assessment is pivotal in this setting as there is a subjective element in the assessment of wound infection. Most studies refer to Centers for Disease Control and Prevention (CDC) wound infection criteria or alternatively specifically describe wound infection criteria. Ten studies scored an 'unclear risk of bias' for detection bias (Andersen 1980; Lazorthes 1992; Taylor 1997; Ijaz 2010; Thakur 2010; Goyal 2011; Othman 2011; Rahmani 2012; Bidhur 2013; Ullah 2013), as these studies did not provide (specific) information about the person who performed outcome assessment. The study by Evans and colleagues scored 'high risk of bias' as surgeons who performed procedures were also involved in outcome assessment while blinding of personnel was insufficiently reported (Evans 1973).

Incomplete outcome data

There were no missing data in 13 studies (Evans 1973; Celdran 2004; Perez 2005; Jain 2008; Ijaz 2010; Thakur 2010; Al‐Fatah 2011; Ergul 2011; Goyal 2011; Othman 2011; Rahmani 2012; Bidhur 2013; Ullah 2013). The other studies reported low numbers of missing data, and missing data were also balanced between treatment groups (Andersen 1980; Lazorthes 1992; Morales 2000; Yerdel 2001; Aufenacker 2004; Oteiza 2004; Tzovaras 2007; Mazaki 2013; Wang 2013; Kochhar 2014). These studies performed available‐case analysis on their data.

We scored five studies as 'high risk' for attrition bias. In the study by Evans and colleagues, it was reported that patients died from unexplained reasons after surgery and were excluded from analysis (Evans 1973). In the other four studies high numbers of missing data were reported ranging from 7.7% (Platt 1990), 9% (Taylor 1997), 10% (Razack 2015), up to 26% (Shankar 2010), and missing data were not balanced between treatment groups (Razack 2015).

Selective reporting

Protocol registration was not reported in the herniorrhaphy studies. The reported outcome in the herniorrhaphy studies was wound infection, without further distinction between superficial or deep surgical site infections.

Three hernioplasty studies prospectively registered a study protocol and we scored them at 'low risk of bias' (Ergul 2011; Mazaki 2013; Wang 2013). One study registered the study protocol after the study was conducted and we therefore scored it at 'high risk of reporting bias' (Rahmani 2012). No other hernioplasty studies registered a study protocol. Most studies did, however, provide data for expected outcomes (superficial and deep surgical site infection: SSSI and DSSI). Eleven studies reported both outcomes separately (Yerdel 2001; Aufenacker 2004; Celdran 2004; Perez 2005; Shankar 2010; Thakur 2010; Ergul 2011; Othman 2011; Rahmani 2012; Mazaki 2013; Wang 2013); and nine studies reported only SSSI (Oteiza 2004; Tzovaras 2007; Jain 2008; Ijaz 2010; Al‐Fatah 2011; Goyal 2011; Bidhur 2013; Ullah 2013; Kochhar 2014). We rated these studies as 'low risk of bias'.

Two hernioplasty studies did not distinguish between SSSI and DSSI outcomes and therefore meta‐analysis for either outcome was not possible (Morales 2000; Razack 2015). For this reason we scored these studies as 'unclear risk of bias'.

Other potential sources of bias

Three studies were terminated early after interim analysis of the results (Yerdel 2001; Celdran 2004; Mazaki 2013). As early study termination could have introduced risk of bias for increased benefit in the antibiotic prophylaxis group, these three studies were scored 'high risk' for 'other bias'.

Effects of interventions

See: Table 1; Table 2

1. Primary outcomes

1.1 Herniorrhaphy repair

The overall wound infection incidence for the herniorrhaphy studies is 4.3%, with 3.8% in the antibiotic prophylaxis group and 4.9% in the control group. The incidence of wound infections in the control groups in individual herniorrhaphy repair surgery studies ranged from 0% to 8.9%. In the low infection risk environment studies the wound infection incidence ranged from 0% to 4.6% and in the high infection risk environment the infection incidence was 8.9%.

All five studies reported the number of participants who developed a wound infection of any type after herniorrhaphy surgery for inguinal or femoral hernia repair (Figure 3) (Evans 1973; Andersen 1980; Platt 1990; Lazorthes 1992; Taylor 1997). No distinction was made between superficial or deep surgical site infections and therefore we could not analyse these outcomes separately. Overall, 35 participants in the antibiotic prophylaxis group developed a wound infection and 25 participants in the placebo/control group. The summary risk ratio (RR) for wound infection of antibiotic prophylaxis compared with placebo or no prophylaxis was 0.86 (95% confidence interval (CI) 0.56 to 1.33). There were no signs of heterogeneity (Chi² P = 0.41 and I² = 0%). For a mean background risk in the control group of 4.9%, the number needed to treat for an additional beneficial outcome (NNTB) is 146. In the low risk environment group (< 5% infections in the control group; four studies with 1302 participants) the RR for any wound infection was 0.63 (95% CI 0.28 to 1.41) and in the high risk environment group (one study with 563 participants) the RR was 0.99 (95% CI 0.58 to 1.68). The P value of the test for subgroup differences was 0.85.

3.

Forest plot of comparison antibiotic prophylaxis against placebo for preventing all types of wound infection (SSSI and DSSI) after herniorrhaphy surgery.

All herniorrhaphy studies scored at least two 'unclear' or 'high' risk of bias items and therefore we could not perform a sensitivity analysis.

1.2 Hernioplasty repair

1.2.1 All wound infections (SSSI + DSSI)

Twenty‐two studies reported on all wound infections (SSSI + DSSI) in all risk environments with a total of 109 wound infections in the antibiotic prophylaxis group and 166 wound infections in the control group (Figure 4). The average wound infection incidence is 4.2%, with 3.2% in the antibiotic prophylaxis group and 5.5% in the control group. The incidence of wound infections in the control groups of the individual hernioplasty repair surgery studies ranged from 0% to 18.1%. In the low infection risk environment studies the wound infection incidence in the control groups ranged from 0% to 4.8% and in the high infection risk environment the infection incidence in the control groups ranged from 5.0% to 18.1%.

4.

Forest plot of comparison antibiotic prophylaxis against placebo for preventing all types of wound infection (SSSI and DSSI) after hernioplasty surgery.

The summary RR for wound infection in the low infection risk environment subgroup was 0.71 (95% CI 0.44 to 1.14), with no indication of heterogeneity (Chi² P = 0.98 and I² = 0%). For a mean background risk in the low infection risk environment control group of 2.6%, the number needed to treat (NNT) to prevent one case of any type of wound infection is 133. The summary RR for wound infections in the high risk environment was 0.58 (95% CI 0.43 to 0.77) and again there was no indication of heterogeneity (Chi² P = 0.46 and I² = 0%). For a mean background risk in the high infection risk environment control group of 8.5%, the NNT to prevent one case of infection is 29. The test for subgroup differences was not significant (Chi² = 0.46), but clinical relevance justified dividing the studies into infection risk subgroups for meta‐analysis. The summary RR for all wound infections in all risk environments was 0.61 (95% CI 0.41 to 0.78) and there were no indications of heterogeneity between studies (Chi² P = 0.85 and I² = 0%).

For sensitivity analysis, we excluded all studies with 'unclear' risk of bias for the domains 'selection bias', 'performance bias' or 'detection bias' and we excluded studies if we scored any of the domains as 'high risk of bias'. After exclusion, only five studies remained in the meta‐analysis (Morales 2000; Aufenacker 2004; Perez 2005; Jain 2008; Wang 2013). The summary RR for all wound infections in all risk environments changed to 0.80 (95% CI 0.53 to 1.21). In the low infection risk environment subgroup, for sensitivity analysis four studies were eligible for inclusion in the meta‐analysis (Morales 2000; Aufenacker 2004; Perez 2005; Jain 2008). This resulted in an RR of 0.78 (95% CI 0.42 to 1.45). For the high infection risk environment subgroup, sensitivity analysis resulted in an RR of 0.82 (95% CI 0.47 to 1.41).

Funnel plot analysis showed that the low infection risk studies are symmetrically distributed (data not shown). For the high infection risk studies the asymmetrical funnel plot reveals an under‐representation of studies that favour the control treatment.

1.2.2 Superficial surgical site infections (SSSI)

Twenty‐one studies reported on superficial wound infections (SSSI) with a total of 94 superficial wound infections in the antibiotic prophylaxis group and 147 superficial wound infections in the control group (Figure 5).

5.

Forest plot of comparison antibiotic prophylaxis against placebo for preventing superficial wound infection (SSSI) after hernioplasty surgery.

The average superficial wound infection incidence is 3.8%, with 2.8% in the antibiotics prophylaxis group and 5.0% in the control group. The incidence of superficial wound infections in the control groups of the individual hernioplasty repair surgery studies ranged from 0% to 18.1%. In the low infection risk environment studies the wound infection incidence in the control groups ranged from 0% to 4.8% and in the high infection risk environment the infection incidence in the control groups ranged from 5.0 to 18.1%.

In the low infection risk environment studies the summary RR for SSSI was 0.71 (95% CI 0.44 to 1.17), with no indication for heterogeneity (Chi² P = 0.96 and I² = 0%). For a mean background risk in the low infection risk environment control group of 2.4%, the NNT to prevent one case of SSSI is 144. For the high infection risk environment studies the summary RR for SSSI was 0.56 (95% CI 0.41 to 0.77). Also for this subgroup there is no indication of heterogeneity (Chi² P = 0.51 and I² = 0%). For a mean background risk in the high infection risk environment control group of 7.9%, the NNT to prevent one case of SSSI is 29. Again, the test for subgroup differences was not significant (Chi² = 0.42), but also in this case clinical relevance of dividing the studies into infection risk subgroups justified subgroup analysis. The summary RR for SSSI in all risk environments was 0.60 (95% CI 0.46 to 0.78), with no indication of heterogeneity (Chi² P = 0.87 and I² = 0%).

For sensitivity analysis, we excluded all studies with 'unclear' risk of bias for the domains 'selection bias', 'performance bias' or 'detection bias' and we excluded studies if any of the domains was scored 'high' risk of bias. After exclusion, only five studies remained in the meta‐analysis (Morales 2000; Aufenacker 2004; Perez 2005; Jain 2008; Wang 2013). The summary RR for superficial wound infections changed to 0.79 (95% CI 0.51 to 1.21). In the low infection risk environment subgroup, for sensitivity analysis four studies were eligible for inclusion in the meta‐analysis (Morales 2000; Aufenacker 2004; Perez 2005; Jain 2008). This resulted in an RR of 0.80 (95% CI 0.41 to 1.56). For the high infection risk environment subgroup, sensitivity analysis resulted in an RR of 0.78 (95% CI 0.44 to 1.37).

1.2.3 Deep surgical site infections (DSSI)

Twelve studies reported on deep wound infections with a total of eight deep wound infections in the antibiotic prophylaxis group and 11 deep wound infections in the control group (Figure 6).

6.

Forest plot of comparison antibiotic prophylaxis against placebo for preventing deep wound infections (DSSI) after hernioplasty surgery.

The overall deep wound infection incidence was 0.45%, with 0.3% in the antibiotic prophylaxis group and 0.6% in the control group. The incidence of deep wound infections in the control groups of the individual hernioplasty repair surgery studies ranged from 0% to 3.8%. In the low infection risk environment studies the wound infection incidence in the control groups ranged from 0% to 0.6% and in the high infection risk environment the infection incidence in the control groups ranged from 0% to 3.8%.

In the low infection risk environment subgroup the RR for DSSI was 0.67 (95% CI 0.11 to 4.13), with no indication of heterogeneity (Chi² P = 0.71 and I² = 0%). For a mean background risk in the low infection risk environment control group of 0.4%, the NNT to prevent one case of DSSI is 758. In the high infection risk environment subgroup, the RR for DSSI was 0.64 (95% CI 0.22 to 1.89), with no indication of heterogeneity (Chi² P = 0.92 and I² = 0%). For a mean background risk in the high infection risk environment control group of 0.7%, the NNT to prevent one case of DSSI is 397. Similarly as for superficial wound infections, the test for subgroup differences was not significant (Chi² = 0.96). Also in this case, clinical relevance of dividing the studies into infection risk subgroups justified subgroup analysis. The summary RR for DSSI in all risk environments was 0.65 (95% CI 0.26 to 1.65), with no indication for heterogeneity between studies (Chi² P = 0.98 and I² = 0%).

For sensitivity analysis, we excluded all studies with 'unclear' risk of bias for the domains 'selection bias', 'performance bias' or 'detection bias' and we excluded studies if any of the domains was scored 'high' risk of bias. After exclusion, four studies remained in the meta‐analysis (Aufenacker 2004; Perez 2005; Jain 2008; Wang 2013). The summary RR changed to 0.93 (95% CI 0.23 to 3.82). For the low infection risk environment subgroup, all studies could be included in the sensitivity analysis and therefore the RR did not change. For the high infection risk environment subgroup, the RR changed to 1.53 (95% CI 0.16 to 14.67).

2. Secondary Outcomes

2.1 Adverse events and development of antibiotic resistance

The outcome 'development of antibiotic resistance' was not addressed in any of the included studies. Only one study reported that emergence of antibiotic‐resistant organisms was not detected in any of the participants (Evans 1973).

Adverse events were sparsely addressed. Six studies reported that no adverse events — or more specifically that no allergic reactions — were observed (Andersen 1980; Platt 1990; Lazorthes 1992; Jain 2008; Ijaz 2010; Mazaki 2013). These six studies plus a further 11 studies excluded participants allergic to antibiotics from the trial (Andersen 1980; Platt 1990; Lazorthes 1992; Taylor 1997; Yerdel 2001; Aufenacker 2004; Oteiza 2004; Perez 2005; Tzovaras 2007; Jain 2008; Ijaz 2010; Al‐Fatah 2011; Ergul 2011; Bidhur 2013; Mazaki 2013; Kochhar 2014; Razack 2015). Two studies documented adverse effects: one study reported, without further explanation, that few patients (n = 6; 0.8%) failed to tolerate antibiotic prophylaxis (Wang 2013); the other study reported an allergic reaction (most likely due to anaesthetics) in a participant who belonged to the control group (Ergul 2011).

Discussion

This review summarises the available evidence from RCTs assessing the effect of antibiotic prophylaxis on the prevention of wound infections after open inguinal or femoral hernia repair in adults. Twenty‐seven trials with a total of 8308 participants were included in this review. Five trials (with 1865 participants) used a non‐mesh repair technique (herniorrhaphy) and 22 trials (with 6443 participants) used a mesh repair technique (hernioplasty).

Summary of main results

For prevention of wound infections in herniorrhaphy surgery, we identified and included five studies in this review. Meta‐analysis (Figure 3, Table 1) demonstrated that a beneficial effect of antibiotic prophylaxis on the prevention of wound infections could neither be confirmed nor refuted as compared to placebo (or no treatment). For a mean background risk in the control group of 4.9%, the number needed to treat (NNT) to prevent one case of infection is 146. Subgroup analysis did not alter the result of the meta‐analysis and we could not perform a sensitivity analysis.

For prevention of wound infections in hernioplasty surgery, we identified and included 22 studies in this review. We present meta‐analyses in Figure 4, Figure 5, Figure 6 and in the Table 2.

The incidence of wound infections in the control groups of the individual hernioplasty repair surgery studies ranged widely from 0% to 18.1%. The average superficial wound infection incidence is 4.2%, with 3.2% in the antibiotics prophylaxis group and 5.5% in the control group. In the low infection risk environment studies the wound infection incidence in the control groups ranged from 0% to 4.8% and in the high infection risk environment the infection incidence in the control groups ranged from 5.0% to 18.1%. The average deep wound infection (DSSI) rate in the low infection risk environment was 0.34% and in the high infection risk environment was 0.52%. This is consistent with reported rates in the literature (Mann 1998; Robbins 1998; Stephenson 2003; Fawole 2006).

In the low infection risk environment a beneficial effect of antibiotic prophylaxis on the prevention of all wound infections (SSSI + DSSI) as well as on superficial wound infections could neither be confirmed nor refuted. The anticipated absolute effect in a low‐risk environment shows that the incidence of 26 wound infections (SSSI + DSSI) per 1000 participants in the placebo (or no treatment) group is lowered to 18 wound infections (SSSI + DSSI) per 1000 participants in the antibiotic prophylaxis group (Table 2). Thus, for a mean background risk in the low infection risk environment control group of 2.6%, the NNT to prevent one case of any type of wound infection is 133. Similar results are observed for superficial wound infections (Table 2). For a mean background risk in the low infection risk environment control group of 2.4%, the NNT to prevent one case of SSSI is 144.

In contrast, a beneficial effect of antibiotic prophylaxis on prevention of all types of wound infection (SSSI + DSSI) as well as of superficial wound infections was demonstrated in the high infection risk environment. This result reflects the importance of taking aseptic operation (room) conditions and wound infection background risk into account when making a clinical decision on antibiotic prophylaxis administration during open hernioplasty surgery. The anticipated absolute effect in a low‐risk environment shows that the incidence of 85 wound infections (SSSI + DSSI) per 1000 participants in the placebo (or no treatment) group is lowered to 49 wound infections (SSSI + DSSI) per 1000 participants in the antibiotic prophylaxis group (Table 2). Thus, for a mean background risk in the high infection risk environment control group of 8.5%, the NNT to prevent one case of any type of wound infection is 29. Similar results were observed for superficial wound infections (Table 2). For a mean background risk in the high infection risk environment control group of 7.9%, the NNT to prevent one case of SSSI is 29. Upon combination of the two subgroups, a beneficial effect of antibiotic prophylaxis as compared to placebo (or no treatment) was still detected. However, upon exclusion of the trials at high risk of bias, few trials could be included into the meta‐analysis and the beneficial effect of antibiotic prophylaxis could no longer be established.

The prevention of deep wound infections (DSSI) by antibiotic prophylaxis could neither be confirmed nor refuted as compared to placebo (or no treatment). Subgroup analysis or sensitivity analysis did not change these results. The anticipated absolute effect in the low infection risk environment shows that the incidence of four deep wound infections per 1000 participants in the placebo (or no treatment) group is lowered to three deep wound infections per 1000 participants in the antibiotic prophylaxis group (Table 2). Hence, for a mean background risk in the low infection risk environment control group of 0.4%, the NNT to prevent one case of DSSI is 758. The anticipated absolute effect in the high infection risk environment shows that the incidence of seven deep wound infections per 1000 participants in the placebo (or no treatment) group is lowered to four deep wound infections per 1000 participants in the antibiotic prophylaxis group (Table 2). Hence, for a mean background risk in the high infection risk environment control group of 0.7%, the NNT to prevent one case of DSSI is 397.

Overall completeness and applicability of evidence

For the effect of antibiotic prophylaxis on the prevention of wound infections after open inguinal or femoral hernia repair in adults, 27 placebo‐controlled studies with 8308 participants are currently available. These include studies published over a period of more than 40 years. Hernia repair method, operation room facilities and applied antibiotics prophylaxis regimes etc. have changed over time. This results in clinical heterogeneity. To address this issue, we have distinguished in our meta‐analysis between two hernia repair methods (i.e. herniorrhaphy and hernioplasty). By doing so, the herniorrhaphy studies with an older publication date were separated from the more recently published hernioplasty studies, thereby reducing clinical heterogeneity between studies. Given the strong improvement of the operation room facilities over the years, it is questionable whether the herniorrhaphy study results can be extrapolated to the current situation. On the other hand, herniorrhaphy surgery is nowadays not performed to a great extent and therefore presented results in this review may not be relevant to current clinical practice.

Study design and treatment characteristics

The number of participants in seven studies were small (≤ 100 participants) (Evans 1973; Celdran 2004; Ijaz 2010; Thakur 2010; Al‐Fatah 2011; Othman 2011; Bidhur 2013), which made interpretation of the results of these single studies difficult. The herniorrhaphy studies were few and total number of participants in the meta‐analysis did not approach optimal information size. In contrast, we included a large number of hernioplasty studies (and participants) into meta‐analyses in this review and optimal information was nearly reached for the outcomes 'all wound infections' and 'superficial wound infections'.

The subject of antibiotic prophylaxis is a complicated issue; only doses that exceed the therapeutic level are effective. This is dependent on many factors including the (number of) dose(s), route and timing of administration, type of antibiotic, and presence of antibiotic‐resistant bacterial species in the geographical area of the study. The included studies varied in the type of administered antibiotics, the frequency and timing of administration (see also Characteristics of included studies; Table 3; Table 4). Despite the variation in antibiotic prophylaxis regimes, heterogeneity analysis revealed low levels of heterogeneity between studies. Moreover, large overlap of the 95% CIs of the various RRs was observed. Altogether, this indicates the absence of important differences between antibiotic regimens. This is supported by the results of a recent systematic review investigating the effects of different antibiotic types on wound infection prevention. Boonchan 2017 analysed beta‐lactam inhibitors and first‐generation cephalosporins and other antibiotics for wound infection prevention. They demonstrated a non‐significant difference between antibiotics classes in wound infection prevention effectiveness. However, it is unclear whether herniorrhaphy and hernioplasty require the same antibiotic approach, given the many differences between the surgical methods.

Aseptic operation (room) conditions, length of the surgical procedure, drain usage and skin preparation (including shaving) are considered risk factors for wound infection development (Mangram 1999); these factors could have contributed to the large differences in baseline wound infection percentages we observed between studies. Hernia surgery is considered a clean surgical procedure with wound infection percentages in low‐risk patients and short surgical procedures (< 1 hour) of below 2% (Mangram 1999). Baseline wound infection percentages as high as 18.1% that we observed do not fall within the expected wound infection percentage range of 0% to 2%. For this reason, we performed subgroup analysis for low and high infection risk environments. These analyses revealed that antibiotic prophylaxis has different preventive effects in different environments, geographical locations and clinical settings/policies.

Participants characteristics and outcome assessment

Overall, participant characteristics and wound infection risk factors (i.e. age, BMI or ASA‐status) were not consistently or specifically reported in the included studies (Mangram 1999). When they were reported, participants vary widely in risk factors. This makes data interpretation difficult. Nevertheless, most studies excluded patients with co‐morbidities (such as diabetes mellitus, malignancy, HIV) and high ASA status and therefore the results of this review seem to be applicable to low‐risk patients. It is currently unclear which patient characteristics are significant to distinguish a low‐risk from a high‐risk patient.

Previous versions of the review made no distinction between superficial and deep wound infections. This distinction is relevant to both the patient and clinical practice. Here, we did analyse superficial and deep wound infections separately and therefore results are more applicable to clinical practice. It should be noted that several studies made no clear distinction between SSSI and DSSI. These studies were excluded from meta‐analysis for the separate outcomes.

Accurate assessment of wound infections is challenging, as there is a subjective component in the diagnosis of a wound infection. New tools and higher quality standards are continually being developed to more accurately assess surgical site infection (Bluebelle Study Group 2018; GlobalSurg Collaborative 2018). Studies included in this review applied many different wound infection assessment tools and definitions which are not consistent with the latest high‐quality standards. Additionally, dichotomisation of any outcome inherently results in a loss of information. The outcome 'presence of wound infection' versus 'absence of wound infection' (SSSI or DSSI) does not contain information on the severity of the infection. This makes the comparison between included studies with different outcome assessment tools difficult. Nevertheless, treatment and control groups within studies can be compared as they were assessed in a similar manner. Altogether, the data should be interpreted with caution as outcome assessment did not meet the high current standards and outcomes are assessed with different measures.

The follow‐up periods varied between studies, but most studies reported various short follow‐up periods of several days to one month postoperatively. Superficial wound infection occurs by definition within one month postoperatively (Mangram 1999). For this reason, we pooled all these studies in the meta‐analyses and are confident that no or an insignificantly small number of infections were missed. In contrast, deep wound infections can very occasionally occur up to years postoperatively, especially in low infection risk environments when multifilament meshes are implanted. Most hernioplasty studies had shorter follow‐up periods than one year. We are not, therefore, completely confident that all late‐onset deep wound infections were detected. Due to the low incidence of late‐onset deep wound infections, very large trials are required to provide a sufficient number of events to address this question, especially when taking into account that deep wound infections usually require either surgical drainage or mesh removal (in case of multifilament meshes). Therefore deep wound infections are considered as a serious complication (Mann 1998; Fawole 2006).

Quality of the evidence

The overall methodological quality ranged from very low to moderate. Quality of the evidence was very low for the herniorrhaphy studies for the outcome 'all wound infections' (see Table 1). Quality of the evidence in hernioplasty surgery was very low to moderate for the outcome 'all wound infections' and the outcome 'superficial wound infections' (see Table 2). Quality of the evidence for the outcome 'deep wound infections' in hernioplasty surgery was low to moderate (see Table 2).

Wound infection is a binomial outcome measurement with a low incidence. To obtain an optimal information size (number of infection events) for meta‐analysis of the pooled results, a large participant population (> 2000 participants per treatment arm) is required. For several meta‐analyses performed in this review this optimal information size was not reached and therefore quality of the evidence was downgraded for imprecision (herniorrhaphy studies, hernioplasty studies for all outcomes in subgroup analyses).

Besides downgrading for imprecision, we also downgraded for risk of bias. A large number of studies did not describe the randomisation or allocation concealment procedures in great detail, which resulted in an unclear risk of selection bias. There is a subjective component in the clinical judgement of wound infections, despite the fact that wound infections are defined in great detail by CDC guidelines. Blinding of outcome assessment is therefore important. Several studies (all herniorrhaphy and seven hernioplasty) did not describe blinding of outcome assessment and were scored 'unclear risk of detection bias'. One herniorrhaphy study was scored 'high risk of detection bias', as surgeons who were not blinded for the treatment also performed outcome assessment.

Five studies scored 'high risk of bias' for attrition bias (two herniorrhaphy and three hernioplasty studies) as the number of missing data (participants lost to follow‐up or participants excluded from the analysis) was high. The proportion of missing outcomes compared with the observed event risk was large enough to have induced clinically relevant bias in the intervention effect estimate.

One study scored 'high risk of selective reporting bias' as the study protocol was registered after the trial was terminated. It should be noted that many studies did not register their study protocol in a prospective trial register. Since most of these studies did report on all expected outcomes, however, we scored these studies as 'low risk of bias' for selective reporting.

Finally, three hernioplasty studies scored 'high risk of other bias' as these studies were terminated after interim analysis of the results. This may have resulted in an increased benefit for the antibiotics prophylaxis group. Overall, studies within the subgroup 'low infection risk environment' had scored more items at low risk of bias than the studies in the subgroup 'high infections risk environment'.

Potential biases in the review process

We searched several electronic databases and prospective trial registers, performed citation searching and we contacted experts in the field to obtain all relevant studies for this review. Nevertheless, it is always possible that we missed relevant studies or data. Two authors independently performed data extraction and quality assessment of the studies to minimise bias in the review process. For several studies we could not obtain all the data required to make judgements for all risk of bias from the publication. In those cases we tried to contact study authors to retrieve additional information. Unfortunately, very few authors responded to our request and the information provided did not affect our initial risk of bias judgement.

For some meta‐analyses (i.e. herniorrhaphy studies and DSSI outcome) we did not perform a publication bias (funnel plot) analysis, because too few studies (< 10 studies) were available for these analyses.

Reporting of adverse effects was scarce. Seventeen studies excluded participants with an allergy to antibiotics; this particular adverse event was therefore not detected in those studies. The outcome 'development of antibiotic‐resistant pathogens' was only addressed in one study. The short administration period of antibiotic prophylaxis during surgery is not likely to result in the development of antibiotic‐resistant bacterial strains. It would, however, be of interest if study authors had reported the antibiotic resistance profile of the bacterial strains that were present in wound infections. This information gives an estimate of the incidence of antibiotic‐resistant strains causing postoperative wound infections. Also it could provide an explanation for the ineffectiveness of the applied antibiotic prophylaxis in the patients with wound infections.

Agreements and disagreements with other studies or reviews

A number of (systematic) reviews and clinical guidelines reported on antibiotic prophylaxis intervention compared to placebo (or no treatment) treatment for prevention of wound infections in open inguinal or femoral hernioplasty surgery. Depending on the publication date of review a smaller or more comprehensive amount of studies was included in the meta‐analysis. Moreover, most reviews did not perform a meta‐analysis of separate high and low infection risk subgroups. No systematic reviews with a meta‐analysis of herniorrhaphy surgery were found.

Biswas 2005 included four RCTs and a few retrospective studies and summarised the data narratively. They concluded that antibiotic prophylaxis is required for hernioplasty repair. Reviews of later publication date revealed a protective effect of antibiotic prophylaxis, similar to the results reported in this review. Sanabria 2007 (meta‐analysis of six studies), Yin 2012 (meta‐analysis of nine studies) and Li 2012 (meta‐analysis of six studies) revealed a protective effect of antibiotic prophylaxis on postoperative wound infection incidence after hernioplasty surgery. These reviews included studies of low and high infection risk environment and expressed the pooled effect size as odds ratio (OR). They did not analyse high and low infection risk studies separately.

Similar to the results presented in this review, some systematic reviews analysed superficial and deep wound infections in separate meta‐analyses. In their analysis, Aufenacker 2006 (meta‐analysis of six studies) was unable to show a significant reduction of the OR for wound infections in the antibiotic prophylaxis treatment group. The systematic reviews of Mazaki 2013 and Erdas 2016 included a larger number of studies (respectively 12 and 16 studies) and were able to reveal a protective effect, expressed as pooled OR, of antibiotic prophylaxis on superficial wound infections, but not on deep wound infections.

International guidelines included 12 and 16 RCTs in their meta‐analysis (Miserez 2014; HerniaSurge Group 2018). Separate meta‐analyses for a low and high infection risk environment were performed. The definition of low infection risk (< 5% infection risk) and high infection risk (> 5% infection risk) is similar to the definition used in this review. Both guidelines reveal a beneficial effect, expressed as OR, of antibiotic prophylaxis on wound infection prevention in a high infection risk environment, but not in a low infection risk environment. Miserez 2014 presented a separate analysis for deep wound infections. In their analysis no significant benefit of antibiotic prophylaxis on wound infection prevention was observed.

Some reviews applied patient inclusion criteria more strictly. Our meta‐analysis included eight studies that also included a small number of adolescent participants (Oteiza 2004; Tzovaras 2007; Shankar 2010; Al‐Fatah 2011; Goyal 2011; Wang 2013; Kochhar 2014; Razack 2015). Excluding these studies from our meta‐analysis did not affect the summary RR to a great extent (RR for all wound infections in all risk environments changed to 0.5 (95% CI 0.35 to 0.72). Therefore we believe that including a small number of participants aged younger than 18 years has not affected our results.

Altogether, the results presented in these (systematic) reviews are in line with the results presented here. The reviews described above and previous versions of this review used the OR as effect estimate. In the current version of this review we have presented the results as RR, which is a more appropriate effect measure for presenting the results of prospective studies (Altman 1991). For rare outcome events, however, such as postoperative wound infections in inguinal or femoral hernia surgery, the odds ratio (OR) and risk ratio ((RR) is approximately the same (Altman 1991). Therefore the results between the previous and current version of this review have only slightly changed, despite the fact that more studies were included in the meta‐analysis. In the previous version of this review the OR (fixed‐effect model) for all wound infections was 0.56 (95% CI 0.38 to 0.81). The RR (random‐effects model) presented here for all wound infections in all risk environments is 0.61 (95% CI 0.48 to 0.78). For comparison we have also computed the OR. Here, the OR for all wound infections in all risk environments in hernioplasty surgery studies is 0.59 (95% CI 0.46 to 0.77). In the current version of this review, more outcomes (all wound infections, superficial and deep wound infections) and subgroups (low and high infection risk environment) are reported than in the previous versions. Altogether, this review presents a more comprehensive analysis of more outcomes and subgroups than the above‐mentioned systematic reviews and guidelines.

Authors' conclusions

Implications for practice.

Administration of antibiotic prophylaxis for elective open inguinal or femoral hernia repair cannot be universally recommended. For open herniorrhaphy surgery we are uncertain whether antibiotic prophylaxis prevents the development of all types of postoperative wound infections. For open hernioplasty surgery, it is uncertain whether antibiotic prophylaxis reduces all types and superficial wound infections in a high infection risk environment. In a low infection risk environment antibiotic prophylaxis makes little or no difference in reducing these type of wound infections.Regarding deep wound infections, antibiotic prophylaxis probably makes little or no difference in preventing the development of these type of wound infections after open hernioplasty surgery.

Implications for research.

This review and meta‐analysis included a large number of studies for the outcome 'superficial wound infections'. Unfortunately, the follow‐up period of these studies was too limited to address the development of the rare phenomenon of late‐onset deep wound infections. Currently, the number of events for the outcome 'deep infections' is too small to draw firm conclusions. It could be beneficial for future studies to include high numbers of participants (> 1000 participants per study arm) and to extend follow‐up periods to two years postoperatively in order to detect sufficient numbers of deep wound infections, which have an incidence lower than 1%. Currently monofilament meshes are usually implanted, however, and these meshes do not require removal upon deep wound infection. The clinical relevance of such studies might therefore be limited. In their review Sanabria and colleagues calculated that larger trials (1600 to 3000 participants) are required to obtain enough wound infection events to make a 50% decrease in antibiotic prophylaxis versus placebo (depending on the background rate of wound infections) (Sanabria 2007). To more adequately address the outcome of 'wound infections', it would be highly beneficial if future studies would apply higher quality standards for outcome assessment. Standards such as Bluebelle Wound healing Questionnaire or the Clavien‐Dindo classification system provide more insight into severity of postoperative wound complications and healing (Dindo 2004; Bluebelle Study Group 2018).

To be able to gain more insight into the patient population for which the results reported here are most applicable, future studies should report in more detail relevant baseline characteristics of risk factors of participants and operation characteristics (ASA‐status, BMI, age, co‐morbidities, operative time length and skin preparation etc.). Identification of risk factors for postoperative wound infection after open inguinal or femoral hernioplasty surgery would be useful to identify those groups of patients that may benefit from antibiotic prophylaxis.

It would be of interest if study authors for future studies would report on the antibiotic resistance profile of the bacterial strains that were present in wound infections. This information gives an estimate of the incidence of antibiotic‐resistant strains causing postoperative wound infections. Also it provides a possible explanation for the ineffectiveness of the administered antibiotic prophylaxis. Finally, a cost‐effectiveness analysis to evaluate the advantages of antibiotic prophylaxis versus the cost of prolonged hospital stay due to wound infection is needed to appropriately appraise the economic implications. The cost of development of antibiotic resistant bacteria should also be taken into account in such analyses.

What's new

Date	Event	Description
13 October 2020	Amended	Minor errors corrected.

History

Protocol first published: Issue 3, 2002
Review first published: Issue 2, 2003

Date	Event	Description
16 December 2019	New citation required and conclusions have changed	Ten new trials were included in this update. Conclusions have changed.
3 December 2019	New search has been performed	Ten new trials were included in this update. Conclusions have changed.
4 January 2012	New citation required but conclusions have not changed	Four new trials added
25 October 2011	New search has been performed	Four new trials included. Conclusions not changed.
28 June 2009	New citation required but conclusions have not changed	One new trial included. Conclusions not changed.
8 August 2006	Amended	Administration of antibiotic prophylaxis for elective inguinal hernia repair cannot be universally recommended. Nevertheless, its administration cannot either be recommended against when high rates of wound infection are observed.
17 March 2004	New search has been performed	Administration of antibiotic prophylaxis for elective inguinal hernia repair cannot be firmly recommended or discarded.

Appendices

Appendix 1. Cochrane Register search strategy

#1	(inguinal hernia:EH or 'femoral hernia':EH or "Hernia, Inguinal":MH or "Hernia, Femoral":MH) OR (((inguin* or groin* or bladder or femoral or crural) adj7 herni*) or femorocele or lichtenstein OR bassini OR shouldice):ti,ab,kw	1793
#2	('antibiotic agent' OR 'antibiotic prophylaxis'):EH OR (Anti‐Bacterial Agents or Antibiotic Prophylaxis):MH OR (antibiotic* or ((antibacterial or anti‐bacterial) adj (compound* or agent*)) or prophylaxis):ti,ab,kw	34797
#3	#1 AND #2	72

Appendix 2. MEDLINE search strategy

1	exp Hernia, Inguinal/ or exp Hernia, Femoral/ or (((inguin* or groin* or bladder or femoral or crural) adj7 herni*) or femorocele or lichtenstein or bassini or shouldice).ti,ab,kf.	18246
2	exp Anti‐Bacterial Agents/ or Antibiotic Prophylaxis/ or (antibiotic* or ((antibacterial or anti‐bacterial) adj (compound* or agent*)) or prophylaxis or anti‐microb* or antimicrob* or antiinfecti* or anti‐infecti*).ti,ab,kf.	910751
3	((randomized controlled trial or controlled clinical trial).pt. or drug therapy.fs. or (random* or placebo or randomly or trial or groups).ab.) not (exp animals/ not humans/)	3757793
4	1 and 2 and 3	133

Appendix 3. Embase search strategy

1	(Randomized controlled trial/ not ((exp experimental organism/ or animal tissue/ or animal cell/ or exp animal disease/ or exp carnivore disease/ or exp bird/ or exp experimental animal welfare/ or exp animal husbandry/ or animal behavior/ or exp animal cell culture/ or exp mammalian disease/ or exp mammal/ or exp marine species/ or nonhuman/ or animal.hw.) not human/)) or ((((((((((((Controlled clinical study/ or Random$.ti,ab. or randomization/ or intermethod comparison/ or placebo.ti,ab. or (compare or compared or comparison).ti. or ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab. or (open adj label).ti,ab. or ((double or single or doubly or singly) adj (blind or blinded or blindly)).ti,ab. or double blind procedure/ or parallel group$1.ti,ab. or (crossover or cross over).ti,ab. or ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)).ti,ab. or (assigned or allocated).ti,ab. or (controlled adj7 (study or design or trial)).ti,ab. or (volunteer or volunteers).ti,ab. or human experiment/ or trial.ti.) not Randomized controlled trial/ not (random$ adj sampl$ adj7 ("cross section$" or questionnaire$1 or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.)) or Cross‐sectional study/) not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or randomi?ed controlled.ti,ab. or control group$1.ti,ab.)) or (((case adj control$) and random$) not randomi?ed controlled).ti,ab. or (Systematic review not (trial or study)).ti. or (nonrandom$ not random$).ti,ab. or "Random field$".ti,ab. or (random cluster adj3 sampl$).ti,ab. or (review.ab. and review.pt.)) not trial.ti.) or "we searched".ab.) and (review.ti. or review.pt.)) or "update review".ab. or (databases adj4 searched).ab. or (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti.) and animal experiment/) or Animal experiment/) not (human experiment/ or human/))	2448495
2	exp 'inguinal hernia'/ or exp 'femoral hernia'/ or ((inguin* or groin* or bladder or femoral or crural) adj7 herni*).ti,ab,hw. or femorocele.ti,ab,hw. or lichtenstein.ti,ab,hw. or bassini.ti,ab,hw. or shouldice.ti,ab,hw.	23865
3	exp 'antibiotic agent'/ or exp 'antibiotic prophylaxis'/ or antibiotic*.ti,ab,hw. or ((antibacterial or 'anti bacterial') adj (compound* or agent*)).ti,ab,hw. or prophylaxis.ti,ab,hw. or 'anti microb*'.ti,ab,hw. or antimicrob*.ti,ab,hw. or antiinfecti*.ti,ab,hw. or 'anti infecti*'.ti,ab,hw.	1755258
4	2 and 3	1186
5	1 and 4	83
6	(elsevier or canadian or embase).cr.	24995844
7	5 and 6	74

Appendix 4. Scopus search strategy

antibiotic prophylaxis (title, abs, key) AND inguinal hernia repair (title, abs, key) AND >2010 (pubyear)

Appendix 5. Web of Science search strategy

#1	TI=(antibiotic)
#2	TI=(antimicrob*)
#3	TI=(anti infecti*)
#4	TI=(antiinfecti*)
#5	#1 OR #2 OR #3 OR #4
#6	TI=(prophyla*)
#7	TI=(prevent*)
#8	#6 OR #7
#9	#8 AND #5
#10	TI=(clean)
#11	TI=(surgery)
#12	TI=(tech*)
#13	TI=(proced*)
#14	#13 OR #12 OR #11
#15	#14 AND #10
#16	TI=(herni*)
#17	TI=(wound infection)
#18	#17 AND #15
#19	#18 OR #16 OR #15
#20	#19 AND #9

period: 2011‐2018

Appendix 6. Medline search strategy previous versions of this review

#1	antibiotic* or antimicrob* or anti infecti* or antiinfecti*
#2	prophyla* or prevent*
#3	#1 AND #2
#4	clean and (surgery or tech* or proced*)
#5	herni*
#6	(wound infection) and #4
#7	#3 and (#4 or #5 or #6)

Data and analyses

Comparison 1. Wound infections herniorrhaphy.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 All wound infections (SSSI+DSSI)	5	1865	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.56, 1.33]
1.1.1 Low infection risk environment	4	1302	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.28, 1.41]
1.1.2 High infection risk environment	1	563	Risk Ratio (M‐H, Random, 95% CI)	0.99 [0.58, 1.68]

1.1. Analysis.

Comparison 1: Wound infections herniorrhaphy, Outcome 1: All wound infections (SSSI+DSSI)

Comparison 2. Wound infections hernioplasty.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 All wound infections (SSSI+DSSI)	22	6443	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.48, 0.78]
2.1.1 low infection risk environment	9	3100	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.44, 1.14]
2.1.2 high infection risk environment	13	3343	Risk Ratio (M‐H, Random, 95% CI)	0.58 [0.43, 0.77]
2.2 Superficial Surgical Site Infections (SSSI)	21	6263	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.46, 0.78]
2.2.1 low infection risk environment	9	3100	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.44, 1.17]
2.2.2 high infection risk environment	12	3163	Risk Ratio (M‐H, Random, 95% CI)	0.56 [0.41, 0.77]
2.3 Deep Surgical Site Infections (DSSI)	12	4185	Risk Ratio (M‐H, Random, 95% CI)	0.65 [0.26, 1.65]
2.3.1 low infection risk environment	3	1488	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.11, 4.13]
2.3.2 high infection risk environment	9	2697	Risk Ratio (M‐H, Random, 95% CI)	0.64 [0.22, 1.89]

2.1. Analysis.

Comparison 2: Wound infections hernioplasty, Outcome 1: All wound infections (SSSI+DSSI)

2.2. Analysis.

Comparison 2: Wound infections hernioplasty, Outcome 2: Superficial Surgical Site Infections (SSSI)

2.3. Analysis.

Comparison 2: Wound infections hernioplasty, Outcome 3: Deep Surgical Site Infections (DSSI)

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Al‐Fatah 2011.

Study characteristics
Methods	Double‐blind RCT Enrolment period: not reported Country: Egypt Single centre study
Participants	N: 200 participants Age: (years; median (range)) PG: 63 (17 to 87); CG: 63 (15 to 90) % Men: PG: 94%; CG: 96%. Inclusion criteria: elective open repair of an inguinal hernia. Age 15 to 80 years old Exclusion criteria: allergy to penicillin, antibiotic intake within the last 5 days before the operation, bilateral hernia repair at the same operation, pregnancy or lactation, existing indication for antibiotic prophylaxis (i.e. valvular heart disease) and known renal or liver impairment (potentially immunocompromised patients) BMI (median (range)): PG: 26 (18 to 34); CG: 26 (20 to 33) Skin antiseptic: povidone‐iodine Operative time (minutes; median (range)): PG 45 (20 to 90); CG 45 (20 to 80) Method of repair: tension‐free polypropylene mesh technique (Lichtenstein technique, using a polypropylene mesh) (hernioplasty) Drain use: PG: n = 3 (3%); CG: n = 4 (4%) ASA class: PG: ASAI‐n = 75 (75%); ASAII n = 22 (22%); ASAIII n = 3 (3%); CG: ASAI n = 71 (71%); ASAII n = 27 (27%); ASAIII n = 2 (2%) Anaesthesia method: spinal Hospitalisation method: inpatient Length of hospital stay: 1 day (patients were discharged 1 day after surgery) Surgeon: a single surgeon
Interventions	Antibiotic prophylactic group (PG): N: 100 patients antibiotic: amoxicillin‐clavulanic acid dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: 30 min before operation Control group (CG): N: 100 patients substance: equal volume of sterile saline
Outcomes	Wound infection definition: 1) purulent discharge (or serosanguineous with positive culture) or 2) spreading erythema indicative of cellulitis or 3) wound breakdown/dehiscence with clinical evidence of infection. Follow‐up: 1 week and 1 month post operation SSSI N infected/N patients (%) PG: 3/100 (3%); CG: 5/100 (5%) DSSI: not reported (reported: "not necessary to remove any mesh") Adverse effects: not reported
Notes	Staphylococcus aureus, Staphylococcus epidermidis and Streptococcus haemolyticus bacteria were isolated from the wound infections Participants aged < 18 years old were included in this study No funding reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “ …list with code numbers generated by the Arcus Quiqstat randomisation program”
Allocation concealment (selection bias)	Unclear risk	Not specified whether envelopes were sequentially numbered and opaque. Quote: “Patients were randomised …. using numbered sealed envelopes” and “The list with code numbers generated by the Arcus Quiqstat randomisation program was kept by a secretary who was not involved in the treatment of the patients at any stage and was opened at the end of the trial for analysis of the results.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “The agent (antibiotic or placebo) was prepared by the anaesthetic nurse and neither the surgeon or the patient was aware of the randomisation arm”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “The list with code numbers generated by the Arcus Quiqstat randomisation program was kept by a secretary who was not involved in the treatment of the patients at any stage and was opened at the end of the trial for analysis of the results.” Quote: “They were followed up as outpatients at 1 week and 1 month after the operation by the surgeon.” ; “…neither the surgeon or the patient was aware of the randomisation arm”
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data.
Selective reporting (reporting bias)	Low risk	The study protocol was not available but report includes all expected outcomes, including specification of SSSI.
Other bias	Low risk	The study appears to be free of other sources of bias.

Andersen 1980.

Study characteristics
Methods	Triple‐blind RCT Country: Denmark Enrolment period: April 1973 to October 1975 Single centre study
Participants	N: 285 participants Age (years; mean (range)): PG: 54 to 56 years (range 21 to 82); CG: 55 years (range 20 to 82) % Men: PG: 72%; CG: 73% Inclusion criteria: direct hernia Exclusion criteria: allergy to penicillin BMI: not reported. Skin antiseptic: not reported. Operative time: not reported. Method of repair: Mc Vay technique (herniorrhaphy) Drain use: removed at postoperative day 4 ASA class: not reported Anaesthesia method: not reported Hospitalisation method: inpatient Length of hospital stay: not reported Surgeon: not reported
Interventions	Prophylactic group (PG): N: 135 Group 1: (n = 59) Ampicillin sodium powder 1 g in the subfascial layers and silk sutures in the fascia and peritoneum Group 2: (n = 76) As Group 1 but polyglycolic acid instead silk sutures Timing of administration: before closure fascial Control group (CG): N: 150 Group 3: (n = 86) No antibiotic therapy and silk sutures Group 4: (n = 64) No antibiotic therapy and polyglycolic acid sutures
Outcomes	Wound infection definition: collection of pus in the wound that requires revision Follow‐up: 1, 3, 6, 12 months SSSI: N infected/N patients (%) PG: 5/135 (3.7%) (Group 1: 1/59; Group 2: 4/76); CG: 6/150 (4%) (Group 3: 3/86; Group 4: 3/64 DSSI: not reported Adverse effects: "none of the patients had an allergic reaction to ampicillin"
Notes	The study was realised in participants who underwent various techniques other that hernia repair, but it was possible to extract the information relating to the patients underwent to hernia repair. Funding was not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information available on how "triple‐blind random allocation design" was performed Quote: “patients … who had a direct hernia or gallstones in a gallbladder (shown not to be infected on macroscopic examination) were assigned at the end of the operation to one of the following four regimens: one gram of ampicillin powder … or no antibiotic therapy”
Allocation concealment (selection bias)	Unclear risk	Quote: "...were assigned at the end of the operation to one of the following four regimens:..."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available
Incomplete outcome data (attrition bias) All outcomes	Low risk	2 patients (out of total of 400 patients) were lost in follow‐up. Available case analysis was performed. The proportion of missing outcomes was considered not enough to have had a clinically relevant impact on the intervention effect estimate.
Selective reporting (reporting bias)	Low risk	Study protocol was not registered, but report contains all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Aufenacker 2004.

Study characteristics
Methods	Double‐blind RCT Country: the Netherlands Enrolment period: November 1998 to May 2003 Multicentre (4) study
Participants	N: 1040 participants Age: (years; mean ± SD) PG: 58.3 ± 12.9; CG: 58.2 ± 13.2 % Men: PG: 95.6%; CG: 97.0% Inclusion criteria: primary uni‐ or bilateral inguinal hernia and an indication for Lichtenstein hernia repair Exclusion criteria: age under 35, the need for antibiotics for a different reason, immunosuppressive disease (diabetes mellitus, malignancy, HIV) or medication (glucocorticoid therapy), allergy to the given antibiotic, recurrent hernia, or the inability to get an informed consent. BMI: not reported Skin antiseptic: PG: povidone iodine 98% of patients; CG: povidone iodine 98.4% of patients Operative time (minutes, median (IQR)) PG: 40 (30 to 50) / CG 40 (28 to 51) Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty). Drain use: PG: n = 11 (2.2%); CG: n = 4 (0,8%) ASA class: not reported Anaesthesia method: PG: local n = 10 (2%); spinal n = 180 (35.8%); general n = 311 (61%); unknown n = 2 (0.4%); CG: local n = 7 (1.4%); spinal n = 191 (37.8%); general n = 330 (60%); unknown n = 4 (0.8%). Hospitalisation method: PG: outpatients n = 231 (46.1%); CG: outpatients n = 232 (45.9%) Length of hospital stay: not reported Surgeon: PG: resident n = 212 (42.1%), certified surgeon n = 291 (57.9%); CG resident n = 225 (44.6%), certified surgeon n = 280 (55.4%)
Interventions	Prophylactic group (PG): N: 503 antibiotic: cefuroxime dose: 1500 mg number of doses: 1 route of administration: intravenous timing of administration: induction of anaesthesia. The exact timing of administering was not standardised, thereby copying daily practice. Control group (CG): N: 505 substance: equal volume of sterile saline.
Outcomes	Wound infection definition: CDC criteria. Follow‐up: 1, 2, 12 weeks SSSI N infected/N patients (%) PG: 7/503 (1.4%); CG: 7/505 (1.4%) DSSI N infected/N patients (%) PG: 1/503 (0.2%); CG 2/505 (0.4%) Adverse effects: not reported
Notes	Staphylococcus aureus, Enterococcus faecalis, Corynebacterium, Streptococcus group G/ Aspergillus fumigatus and skin bacteria were isolated from the wound infection Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “….according to a computer generated list in blocks of 10 patients with stratification for each hospital.”
Allocation concealment (selection bias)	Low risk	Quote: “The patients were double‐blinded randomly assigned to either intravenous placebo or antibiotic prophylaxis. A pharmacist carried out randomization….” and “A pharmacist prepared the trial medication under laminar airflow condition, and it was packed in nontransparent material to exclude optical differences. The anesthesiologist administered the trial medication at the induction of anesthesia.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “A pharmacist prepared the trial medication under laminar airflow condition, and it was packed in nontransparent material to exclude optical differences. The anesthesiologist administered the trial medication at the induction of anesthesia.”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "In most cases, the surgeon who performed the operation did not perform the follow‐up. In case of missing observations, the patients were contacted and a standardized telephone interview was performed." In the few cases that the surgeon who performed the operation also performed the follow‐up, it should be kept in mind that Quote: " trial medication ...was packed in nontransparent material to exclude optical differences" and that therefore the surgeon was unaware whether participant was part of intervention or control group.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data were balanced between intervention groups. PG n = 17 (17/520 = 3.2%) and CG n = 15 (15/520 = 2.9%). The proportion of missing outcomes was considered not enough to have had a clinically relevant impact on the intervention effect estimate Available case analysis was performed 1 death in placebo‐group due to operation‐related complication (Quote: "One patient died of pulmonary complications and a bleeding gastric ulcer".)
Selective reporting (reporting bias)	Low risk	Study protocol was not registered, but report contains all expected outcomes, including specification of SSSI and DSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Bidhur 2013.

Study characteristics
Methods	RCT Country: Nepal Enrolment period: April 2008 to June 2009 Single centre study.
Participants	N: 60 participants Age (years; mean ± SD): overall population: 38.5 ± 17.7 (range: 19 to 90 years); age < 30 years PG: n = 12 (40%); CG: n = 11 (36.7%) % Men: PG: 100%; CG: 96.7% Inclusion criteria: age > 18 years, primary inguinal hernia Exclusion criteria: recurrent, strangulated, bilateral inguinal hernia, femoral hernia, age < 18 years, allergy to Cefuroxime, systemic or advanced diseases like diabetes mellitus, liver or renal impairment, patients on steroids or antibiotic within a week before surgery, or pregnant or lactating women. BMI: not reported Skin antiseptic: not reported Operative time (minutes, mean ± SD): overall population: 51.3 ± 9.8 (range 40 to 75 min); time < 50 min: PG: n = 19 (63%); CG: n = 15 (50%) Method of repair: Lichtenstein technique (hernioplasty) with monofilament polypropylene mesh (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia method: local (n = 42) or spinal (regional, n = 18) Hospitalisation method: outpatient (for local anaesthesia) or inpatient (for spinal anaesthesia) Length of hospital stay: 1 day for spinal anaesthesia, day surgery for local anaesthesia Surgeon: qualified surgeon or resident under supervision of faculty (not specified per treatment group)
Interventions	Prophylactic group (PG): N: 30 antibiotic: cefuroxime dose: 1500 mg number of doses: 1 route of administration: intravenous timing of administration: induction of anaesthesia (spinal anaesthesia) or just before incision (local anaesthesia). Control group (CG): N: 30 substance: equal volume of sterile saline.
Outcomes	Wound infection definition: CDC criteria. Follow‐up: 7 to 9 days, 28 to 42 days post operation SSSI N infected/N patients (%) PG: 0/30 (0%); CG: 1/30 (3.3%) DSSI: not reported Adverse effects: not reported
Notes	Reported is: no conflict of interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Randomization was done as listed by randomization technique in Microsoft Excel Programme”
Allocation concealment (selection bias)	Unclear risk	No information available.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data.
Selective reporting (reporting bias)	Low risk	Study protocol was not registered, but report contains all expected outcomes, including specification of SSSI.
Other bias	Low risk	Study appears to be free from other sources of bias.

Celdran 2004.

Study characteristics
Methods	Double‐blind RCT Enrolment period: October 1997 ‐ not reported Country: Spain Single centre study
Participants	N: 99 participants Age (years; mean ± SD): PG: 58 ± 13; CG: 58 ± 17 % Men: PG 94% and CG 86% Inclusion criteria: ASA I/II patients eligible for open mesh inguinal hernia repair Exclusion criteria: not specified BMI (mean ± SD): PG: 26.1 ± 5 and CG: 26.2 ± 5 Skin antiseptic: povidone‐iodine Operative time (minutes, mean ± SD): PG 65 ± 23; CG 64 ± 14 Method of repair: Lichtenstein or Celdran mesh technique (hernioplasty) Drain use: not reported ASA class: I or II Anaesthesia method: local and sedation Hospitalisation method: autonomous ambulatory surgery unit (outpatient). Length of hospital stay: < 1 day Surgeon: PG: staff n = 37 (74%) resident n = 13 (26%); CG: staff n = 38 (78%) resident n = 11 (22%); (all patients were operated on by the senior author or by residents whom he assisted)
Interventions	Prophylactic group (PG): N: 50 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: parenteral (not specified) timing of administration: 30 minutes before the incision Control group (CG): N: 49 substance: placebo (not specified)
Outcomes	Wound infection definition: CDC criteria Follow‐up: 1 week, 1, 3, 6 months, 1 and 2 years SSSI N infected/N patients (%) PG: 0/50 (0%); CG: 4/49 (8.2%) DSSI N infected/N patients (%) PG: 0/50 (0%); CG: 0/49 (0%) (reported: "not necessary to remove any mesh") Adverse effects: not reported
Notes	Staphylococcus aureus and Klebsiella oxytoca bacteria were isolated from the wound infections Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A list of random numbers was generated to assign the treatment."
Allocation concealment (selection bias)	Unclear risk	Insufficient information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "All patients were examined 1 week, 1, 3, and 6 months, and 1 and 2 years postoperatively by trained impartial surgeon."
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data.
Selective reporting (reporting bias)	Low risk	The study protocol was not registered, but report includes all expected outcomes, including specification of SSSI and DSSI
Other bias	High risk	Study was terminated early after interim analysis, and therefore there might be risk of bias for increased benefit. Quote: "Although it was designed to include a higher number of patients, the interim analysis, performed after the first 91 patients had been included, recommended ending the study due to ethical reasons."

Ergul 2011.

Study characteristics
Methods	Double‐blind RCT Enrolment period: July 2008 to October 2010 Country: Turkey Single centre
Participants	N: 200 participants Age (years; mean ± SD): PG: 48 ± 17/ CG: 50 ± 15 % Men: PG: 95%; CG: 89% Inclusion criteria: primary unilateral direct and indirect inguinal hernias. Exclusion criteria: age under 18, recurrent hernias, simultaneous bilateral repairs, incarcerated or strangulated hernias requiring emergency repair, coagulation disorder or anticoagulant medication (i.e. acetylsalicylic acid, clopidogrel, warfarin), history of allergy, sensitivity, or anaphylaxis to cephalosporin antibiotics, antibiotic therapy within 72 h before operation, cardiac valvular problems that require specific perioperative antibiotic regimen, presence of infection at the time of operation, pregnancy or lactation, immunosuppressive diseases (i.e. newly diagnosed or uncontrolled diabetes mellitus, malignancy, HIV), glucocorticoid medication, giant scrotal hernias, prosthetic valves or joints, drain usage, and patients did not accept the registry and randomisation BMI: not reported Skin antiseptic: povidone iodine Operative time (minutes, median (min‐max): PG: 60 (35 to 160); CG: 60 (40 to 135) Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty) Drain use: not reported ASA class: PG: ASAI n = 57 (57%); ASAII n = 35 (35%); ASAIII‐IV n = 8 (8%); CG: ASAI n = 54 (54%); ASAII n = 36 (36%); ASAIII‐IV n = 10 (10%) Anaesthesia method: PG: general n = 48 (48%), regional n = 39 (39%), local n = 12 (12%); CG: general n = 53 (53%), regional n = 36 (36%), local n = 11 (11%) Hospitalisation method: inpatient Length of hospital stay (days, median (min‐max)): PG: 1 (1 to 1); CG: 1 (1 to 4) Surgeon: supervised residents or staff surgeons
Interventions	Prophylactic group (PG): N: 100 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: when the patient entered the operating room or, at the latest, at the induction of anaesthesia. Control group (CG): N: 100 substance: equal volume of sterile saline.
Outcomes	Wound infection definition: CDC criteria Follow‐up: POD 1 and 3; 3, 5, 6 and 30 days after discharge SSSI N infected/N patients (%): PG: 5/100 (5%); CG: 7/100 (7%) DSSI N infected/N patients (%): PG: 0/100 (0%); CG: 0/100 (0%) Adverse effects: perioperative allergic reaction in 1 patient, probably due to allergy to anaesthesia as patient belonged to control group.
Notes	Staphylococcus aureus bacteria were isolated from the wound infections Reported: no conflict of interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Patients were randomized by use of sealed envelopes which included equal numbers of patients to be randomized either to the control arm or to the antibiotic prophylaxis arm.”
Allocation concealment (selection bias)	Unclear risk	Not specified whether envelopes were sequentially numbered and opaque. Quote: “Patients were randomized by use of sealed envelopes which included equal numbers of patients to be randomized either to the control arm or to the antibiotic prophylaxis arm.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: The anaesthesiologist administered the trial medication (antibiotic or sterile saline in coded syringes)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The surgeon who performed the follow‐up frequently was not the surgeon who performed the operation."
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data. Quote: “No patient was lost to follow‐up.”
Selective reporting (reporting bias)	Low risk	Quote: "The trial was registered on http://www.controlled‐trials.com (ISRCTN85660082)" and all of the prespecified outcomes that were of interest have been reported in the prespecified way.
Other bias	Low risk	The study appears to be free of other sources of bias

Evans 1973.

Study characteristics
Methods	RCT Enrolment period: not reported Country: UK Single centre
Participants	N: 97 participants Age: not reported % Men: not reported Inclusion criteria: Several types of clean or contaminated surgeries, age > 16 years Exclusion criteria: not specified BMI: not reported Skin antiseptic: not reported Operative time: not reported Method of repair: not reported, herniorrhaphy Drain use: not reported ASA class: not reported Anaesthesia method: not reported Hospitalisation method: not reported (most likely inpatient) Length of hospital stay: not reported Surgeon: not reported
Interventions	Prophylactic group (PG): N: 48 antibiotic: cephaloridine dose: 1 g number of doses: multiple (3) route of administration: intravenous at induction and intramuscular after. timing of administration: induction and at 5 and 12 h later. Control group (CG): N: 49 substance: nothing
Outcomes	Wound infection definition: presence of pus which either discharged or needed to be released from the wound Follow‐up: 4 weeks SSSI N infected/N patients (%): PG: 1/48 (2.1%); CG: 2/49 (4.1%) DSSI not reported Adverse effects: reported: "the administration of cephalordine did not result in any change in the type of organism or curiously in the emergence of cephaloridine‐resistant organisms"
Notes	The study was realised in patients who underwent types of surgery other than hernia repair, but it is possible to extract the information on wound infection related to the patients who underwent hernia repair. Antibiotics (Cephaloridine) were supplied by Glaxo laboratories Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “The method of random selection has been, as before, by the toss of a coin”
Allocation concealment (selection bias)	Unclear risk	Insufficient information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Insufficient information available, but performance bias is unlikely in this setting Quote: "To avoid bias the details were entered not in the patients' case notes but on punch cards which were kept separately and analysed manually."
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: “All patients were seen by one of us for at least 4 weeks.”
Incomplete outcome data (attrition bias) All outcomes	High risk	Patients died from unexplained reasons after surgery and were excluded from analysis. Quote: “Thirty‐two patients (out of total 762 patients) who died within 2 weeks of operation in whom the presence or absence of infection was uncertain have been excluded from the study, …”
Selective reporting (reporting bias)	Low risk	The study protocol was not registered, but report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Goyal 2011.

Study characteristics
Methods	RCT Enrolment period: 1 year (not specified) Country: India Single centre
Participants	N: 200 participants Age (years): range 11 to 90 yrs, with 70.5% of the patients in the 31 to 70 years age group (overall population) % Men: not reported; both male and female were included Inclusion criteria: any type of primary inguinal hernia. Exclusion criteria: patients with complicated, strangulated hernia, those having local skin infection, systemic infection, diabetes or history of antibiotic use within previous week BMI: not reported Skin antiseptic: 10% povidone iodine Operative time: not reported Method of repair: Lichtenstein technique (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia method: not reported Hospitalisation: inpatient Length of hospital stay: 1 day Surgeon: not reported
Interventions	Prophylactic group (PG): N: 100 antibiotic: amoxicillin‐clavulanate dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: induction Control group (CG): N: 100 substance: equal volume of sterile saline
Outcomes	Wound infection definition: serous discharge, seroma formation, erythema and stitch abscess. Follow‐up: 8 days SSSI N infected/N patients (%) PG: 1/100 (1%); CG: 3/100 (3%) (SSSI cases had erythema or stitch abscess, cases with serous discharge were considered not having wound infection) DSSI: not reported (reported: "not necessary to remove any mesh") Adverse effects: not reported
Notes	Patients aged < 18 years included Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Patients were randomized in two groups by random number table method, Group 1 as cases, Group 2 as controls”
Allocation concealment (selection bias)	Unclear risk	Insufficient information available.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Insufficient information available, but performance bias is unlikely in this setting Quote: "Group 1 was given IV injection of 1,2 gr amoxicillin‐clavulanate in 20 ml saline at the time of induction, while the other group was given 20 ml of sterile saline as placebo”
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	The study protocol was not registered, but report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free of other sources of bias.

Ijaz 2010.

Study characteristics
Methods	RCT Enrolment period: January 2007 to December 2007 Country: Pakistan Single centre
Participants	N: 100 participants Age (years; mean) PG: 44.06; CG: 44.84 % Men: 99% (overall study population) Inclusion criteria: elective inguinal hernia, age > 20 years Exclusion criteria: patients with obstructed, strangulated or recurrent hernias, patients with immunosuppressive disease or on medication (steroids) and patients with a debilitating disease like chronic liver, renal or cardiac impairment. Also patients allergic to the given antibiotic or taking some antibiotic 7 days prior to surgery for any reason. BMI: not reported Skin antiseptic: not reported Operative time: not reported Method of repair: Lichtenstein technique (hernioplasty); polypropylene mesh Drain use: not reported ASA class: not reported Anaesthesia method: spinal Hospitalisation method: inpatient Length of hospital stay: 2 days (discharge on second postoperative day) Surgeon: senior registrar
Interventions	Prophylactic group (PG): N: 50 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: 30 minutes before surgery Control group (CG): N: 50 substance: distilled water
Outcomes	Wound infection definition: CDC criteria Follow‐up: 7, 14, 30 days post operation SSSI N infected/N patients (%): PG: 2/50 (4%); CG: 5/50 (10%) DSSI: not reported Adverse effects: "None of the patients developed adverse reactions related to antibiotics prophylaxis"
Notes	Staphylococcus aureus and Staphylococcus epidermidis bacteria were isolated from the wound infections Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information available on how randomisation was performed. Quote: "All included patients were randomized into two groups A and B and prepared for surgery". Quote: “Simple random sampling technique was applied”
Allocation concealment (selection bias)	Unclear risk	No information available.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	The study protocol was not registered, but report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free from other sources of bias

Jain 2008.

Study characteristics
Methods	Double‐blind RCT Enrolment period: 1 year (not specified) Country: India Single centre
Participants	N: 120 participants Age (years; mean ± SD): PG: 41.28 ± 11.49; CG: 40.2 ± 9.84 % Men: 100% (both groups) Inclusion criteria: primary, unilateral inguinal hernia (direct and indirect) Exclusion criteria: recurrent, bilateral, irreducible or strangulated hernia; patients with systemic and advanced diseases (e.g. liver failure, chronic renal failure, diabetes); immuno‐compromised patients; patients with ASA scores higher than 11; patients receiving steroids for any reason; patients younger than 18 or older than 60; patients with local skin infections or disease at the site of incision; and patients allergic to antibiotics or who had received antibiotics less than a week before surgery. BMI: not reported Skin antiseptic: povidone iodine and alcohol Operative time (minutes, mean ± SD): PG: 56.33 ± 11.67; CG: 60.33 ± 6.81 Method of repair: prolene hernia system (PHS) mesh repair with prolene mesh (hernioplasty). Drain use: no drains were used ASA class: PG: ASAI n = 52 (87%), ASAII n = 8 (13%); CG: ASAI n = 49 (82%), ASAII n = 11 (18%) Anaesthesia method: local or spinal Hospitalisation method: inpatient Length of hospital stay: 1 day Surgeon: not reported
Interventions	Prophylactic group (PG): N: 60 antibiotic: amoxicillin‐clavulanic acid dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: before incision Control group (CG): N: 60 substance: sterile saline
Outcomes	Wound infection definition: CDC criteria Follow‐up: till discharge, 7 to 9 days, 2 and 4 weeks, 1 year SSSI N infected/N patients (%): PG: 1/60 (1.7%); CG: 1/60 (1.7%) DSSI N infected/N patients (%): PG: 0/60 (0%); CG: 0/60 (0%) Adverse effects: "None of the patients developed adverse reactions related to antibiotics prophylaxis"
Notes	Staphylococcus aureus bacteria were isolated from the wound infections Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Randomization was performed by a computer‐generated code…”
Allocation concealment (selection bias)	Low risk	Quote: "Randomization was performed by a computer‐generated code by a junior resident who was not involved in the surgery, data compilation or patient follow‐up."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “The same resident also prepared the antibiotic or the placebo syringes containing normal saline.”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “In order to remove any personal bias, the surgeon who performed the operation was not allowed to follow up their patient.”
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	No study protocol available, but the published report includes all expected outcomes, including specification of SSSI and DSSI.
Other bias	Low risk	The study appears to be free of other sources of bias

Kochhar 2014.

Study characteristics
Methods	Double‐blind RCT Enrolment period: August 2007 to December 2012 Country: India Single centre
Participants	N: 217 participants Age (years; mean ± SD): PG: 37.42 ± 9.9; CG: 37.42 ± 11.5 % Men: 95.7% (overall study population) Inclusion criteria: unilateral or bilateral inguinal hernia, age 15 to 70 years Exclusion criteria: Patients allergic to the given antibiotic, recurrent hernia, strangulated hernia, pregnancy or lactation, an immune compromised state, antibiotic treatment within last 5 days, existing indication for antibiotic prophylaxis (i.e. valvular heart disease) BMI: not reported Skin antiseptic: not reported Operative time: not reported Method of repair: Lichtenstein technique with monofilament polypropylene mesh (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia method: not reported Hospitalisation method: inpatient Length of hospital stay: 1 day Surgeon: not reported
Interventions	Prophylactic group (PG): N: 109 antibiotic: amoxicillin clavulanic acid dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: just before operation Control group (CG): N = 108 substance: equal volume of sterile saline
Outcomes	Wound infection definition: ASEPSIS criteria purulent discharge, erythema, wound breakdown) Follow‐up: 1 and 4 weeks SSSI N infected/N patients (%): PG: 4/106 (3.8%); CG: 5/106 (4.7%) DSSI: not reported (reported: "not necessary to remove any mesh") Adverse effects: not reported
Notes	Participants < 18 years old were included in this study Staphylococcus aureus, E. coli, Klepsiella pneumoniae bacteria were isolated from the wound infections Funding: nil.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “All patients were randomized to either intravenous placebo (control group) or antibiotic prophylaxis (study group) by block randomization”. No information on how randomisation was accomplished. However, the block randomisation method indicates that the randomisation methodology has been performed carefully to minimise bias.
Allocation concealment (selection bias)	Unclear risk	No information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “Surgeon who was not involved in surgery followed the case after 1 and 4 weeks post‐operatively”
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data balanced in numbers across intervention groups, with similar reason for missing data. The proportion of missing outcomes was considered not enough to have had a clinically relevant impact on the intervention effect estimate. Quote: “Out of the total study population 3 patients from the study and 2 patients from the control group were excluded postoperatively for development of URI and thus requirement of antibiotics.” Available‐case analysis was performed.
Selective reporting (reporting bias)	Low risk	The study protocol was not available but report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Lazorthes 1992.

Study characteristics
Methods	Double‐blind RCT Enrolment period: October 1987 to March 1989 Country: France Single centre
Participants	N: 324 participants Age (years; mean (range)): PG: 62 (11 to 90); CG: 70 (16 to 92) % Men: PG: 90%; CG: 88% Inclusion criteria: primary inguinal hernia Exclusion criteria: allergy to antibiotics, recurrent hernia, anaesthesia other than local (epidural or general) BMI: not reported Skin antiseptic: not reported Operative time: not reported Method of repair: not reported (herniorrhaphy) Drain use: not reported ASA class: not reported Anaesthesia method: local Hospitalisation method: inpatient Length of hospital stay: not specified, hospital stay was prolonged due to infection Surgeon: not reported
Interventions	Prophylactic group (PG): N: 162 antibiotic: cefamandole dose: 750 mg number of doses: 1 route of administration: added to local anaesthesia solution (lidocaine) timing of administration: during anaesthesia/ surgery Control group (CG): N: 162 substance: no antibiotics, only local anaesthesia (lidocaine) solution
Outcomes	Wound infection definition: wounds with discharge with positive microbiological culture Follow‐up: till 1 month SSSI N infected/N patients (%): PG: 0/155 (0%)/ CG: 7/153 (4.6%) DSSI: not reported Adverse effects: "No instances of allergy"
Notes	Participants < 18 years old were included in this study Staphylococcus aureus, S. epidermidis, S. albans, Streptococcus sanguis, Streptococcus alpha and Escherichia coli bacteria were cultured from the wound infections Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information available Quote: “Two groups of 162 patients were randomly allotted to receive ...”
Allocation concealment (selection bias)	Unclear risk	No information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Insufficient information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information available. Quote: “A questionnaire was given to the patient upon discharge and the patient was asked to return it one month postoperatively to ensure that there was no late wound abscess. In the instance of wound discharge, the patient was examined by the surgeon to ascertain its true nature.”
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data are balanced in numbers across intervention groups (PG n = 7 (4.3%); CG n = 9 (5.5%)). The proportion of missing outcomes was considered not enough to have had a clinically relevant impact on the intervention effect estimate Available case analysis was performed.
Selective reporting (reporting bias)	Low risk	No study protocol available, but the published report includes all expected outcomes, including specification of SSSI.
Other bias	Low risk	The study appears to be free of other sources of bias

Mazaki 2013.

Study characteristics
Methods	Double‐blind RCT Enrolment period: July 2007 to December 2011 Country: Japan Single centre
Participants	N: 200 participants Age (years; median( IQR)): PG: 69 (57 to 76 yrs); CG: 72 (60 to 77 yrs) % Men: PG: 89%; CG: 94% Inclusion criteria: primary unilateral or bilateral inguinal hernia Exclusion criteria: outpatient status; day surgery; age < 18 years; recurrent hernia; incarcerated or strangulated hernia requiring emergency hernia repair; pregnancy or lactation; earlier history of allergy, sensitivity, or anaphylaxis to beta‐lactam or cephalosporin antibiotics; antibiotic therapy < 48 hours before surgery; presence of an infection at the time of surgery; cardiac valvular problem; increased risk for infection secondary to a coexisting medical condition; immunosuppression (e.g. human immunodeficiency virus infection, malignancy, or chemotherapy); American Society of Anesthesiologists (ASA) grade > IV; and refusal to participate in the study. BMI: PG: 23.0 ± 2.8 PG; CG: 22.7 ± 3.0 Skin antiseptic: 10% povidone iodine Operative time (minutes): PG: 66.3 ± 25.4; CG: 65.2 ± 27.1 Method of repair: mesh‐plug technique with polypropylene mesh (hernioplasty) Drain use: no drains were placed ASA class: PG: ASAI n = 52 (52%), ASAII n = 43 (43%), ASAIII n = 5 (5%); CG: ASAI n = 37 (37%), ASAII n = 58 (58%), ASAIII n = 5 (5%) Anaesthesia method: general or local Hospitalisation method: inpatient Length of hospital stay (days, median (IQR)) : PG 3 (2 to 3); CG 3 (2 to 3) Surgeon: PG: staff n = 16 (16%), resident n = 84 (84%); CG: staff n = 17 (17%), resident n = 83 (83%) (in total 33 surgeons or residents)
Interventions	Prophylactic group (PG): N: 100 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: intravenous (CG: continuous IV, PG: single administration) timing of administration: 30 minutes before incision Control group (CG): N: 100 substance: equal volume sterile saline
Outcomes	Wound infection definition: CDC criteria Follow‐up: 7 to 8 days, 1 month and 3 months post operation SSSI N infected/N patients (%): PG: 2/100 (2%); CG: 13/100 (13%) DSSI: N infected/N patients (%): PG: 0/100 (0%); CG: 0/100 (0%) Adverse effects: "No adverse events were seen nor mortality"
Notes	Reported: No conflict of interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “…..computer‐generated list in blocks of 50 patients."
Allocation concealment (selection bias)	Low risk	Quote: “.... a pharmacist carried out randomization” Quote: "All surgeons and other staff members were blinded to randomization and to patients' details"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “All surgeons and other staff members were blinded to randomization and to patients”. Quote: "Patients were randomly assigned on admission in a double‐blinded manner to either the antibiotic prophylaxis group or the placebo group."
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “All wounds were carefully examined by 2 certified surgeons, who did not perform the operations”
Incomplete outcome data (attrition bias) All outcomes	Low risk	Flow chart of participants reveals missing outcome data is low (n = 1 in antibiotic prophylaxis group and n = 3 in placebo group), with similar reasons for missing data (other surgical technique applied) Also intention‐to‐treat analysis was applied.
Selective reporting (reporting bias)	Low risk	The protocol was registered (NCT00636831) and study outcomes that are of interest in the review have been reported in the prespecified way.
Other bias	High risk	Study was terminated early after interim analysis, and therefore there might be risk of bias for increased benefit in the antibiotics prophylaxis group.

Morales 2000.

Study characteristics
Methods	Double‐blind RCT Enrolment period: June 1994 to June 1997 Country: Spain/Ibiza Multicentre (3)
Participants	N: 554 participants Age (years; mean (range)): 54.2 (17 to 87yrs) (overall study population) % Men: 89.9% (overall study population) Inclusion criteria: inguinal or femoral hernia repair with synthetic mesh Exclusion criteria: emergency surgery, cytostatic or corticosteroid therapy, HIV/AIDS or neoplastic disease, non‐compliance of the allocation concealment, no fulfilment of the time of follow‐up according to the protocol, drainage placement. BMI: not reported Skin antiseptic: 10% povidone iodine Operative time: not reported (in patients with infections the mean operation time was 34 minutes and in patients without infections the mean operation time was 36.5 minutes) Method of repair: polypropylene mesh repair (hernioplasty) Drain use: patients with drains were excluded ASA class: ASAI n = 224 (42.7%), ASAII n = 219 (41.8%), ASAIII n = 78 (14.9%) (overall study population, not specified per treatment group) Anaesthesia method: not reported Hospitalisation method: inpatient and outpatient Length of hospital stay: 0 to 1 day Surgeon: only staff, no residents
Interventions	Prophylactic group (PG): N: 237 antibiotic: cefazolin or Erythromycin (in case of allergy to cefazolin) dose: 2 g (1 g for erythromycin) number of doses: 1 route of administration: intravenous timing of administration: at induction Control group (CG): N: 287 substance: physiological serum
Outcomes	Wound infection definition: presence of 1 of the following criteria: 1) skin erythema > 2 cm on both sides of the incision; 2) discharge of purulent material from the wound; 3) discharge not purulent with microbiological positive culture; 4) wound had to be re‐opened by the surgeon Follow‐up: 7 days, 1 month, 12 months post operation. SSSI N infected/N patients (%): PG: 4/233 (1.6%); CG: 6/281 (2.1%) DSSI N infected/N patients (%): 4 patients (all with previous SSI, not specified in which treatment group) Adverse effects: not reported
Notes	S. aureus, E. Coli + Proteus mirabilis were isolated from the wound infections Spanish manuscript Funding was not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Marked cards Quote: "una classificaccion de los pacientos de forma aleatona mediante la extraccion de tarjetas marcades como "SF" (profilaxis) o "no" (placebo) (= a classification of the patients in a random way by extracting cards marked as "SF" (prophylaxis) or "no" (placebo))"
Allocation concealment (selection bias)	Low risk	Quote: "se designo un coordinator del estudio en cada centro respondable de la recigida de datos y de velar por el cumplimiento de los criterios de inclusion de los pacientes, asi como de la aleatorizacion, estraficacion y aleatoriedad del estudio, y de la confidencialidad de los resultados (= a study coordinator was designated in each center responsible for the data collection and to ensure compliance with the inclusion criteria of the patients, as well as the randomization, and the confidentiality of the results)"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "La administración de la solución (antibiótico o placebo) fue realizada por la enfermera circulante, no informando del contenido de la solución a ninguno de los dos cirujanos implicados en la intervención." (= "The administration of the solution (antibiotic or placebo) was performed by the nurse, not informing the content of the solution to any of the two surgeons involved in the intervention.")
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "En el seguimiento postoperatorio, los cirujanos, las enfermeras de planta y la de consulta externa desconocían la solución administrada al paciente." and “control en la consulta externa de cirurgia (= "In the postoperative follow‐up, the surgeons, the plant nurses and the outpatient nurses were unaware of the solution administered to the patient" and "control in the outpatient clinic of surgery")
Incomplete outcome data (attrition bias) All outcomes	Low risk	In the antibiotic prophylaxis group 4 patients were lost to follow‐up and in the control group 6 patients were lost to follow‐up. Missing data is balanced in numbers across intervention groups. Also the proportion of missing outcomes compared with the observed event risk is not enough to have had a clinically relevant impact on the intervention effect estimate. Available case analysis of the data was performed.
Selective reporting (reporting bias)	Unclear risk	No reporting that study protocol was registered. Authors failed to report number of cases with DSSI and mesh removal surgery per treatment group and therefore meta‐analysis for this outcome is not possible.
Other bias	Low risk	The study appears to be free of other sources of bias

Oteiza 2004.

Study characteristics
Methods	RCT Country: Spain Enrolment period: March 2001 to March 2002 Single centre
Participants	N: 250 participants Age (years; mean (range)): PG: 58 (22 to 91); CG: 56.2 (17 to 88) % Men: PG: 89%; CG: 82% Inclusion criteria: unilateral direct and indirect inguinal hernia Exclusion criteria: recurrent or bilateral hernia, allergy to penicillin, ASA IV and those who needed hospitalisation. BMI: not reported Skin antiseptic: not reported Operative time: (mean) 40 min in both groups Method of repair: tension free (Lichtenstein, or Plug‐Stein) with polypropylene mesh (hernioplasty) Drain use: not reported ASA class: PG: ASAI n = 30 (24%), ASAII n = 72 (58%), ASAIII n = 22 (18%); CG: ASAI n = 35 (28%), ASAII n = 70 (57%), ASAIII n = 18 (15%) Anaesthesia: local with sedation or spinal Hospitalisation: outpatient Length of hospital stay: no admission Surgeon: 2 surgeons
Interventions	Prophylactic group (PG): N: 125 antibiotic: amoxicillin‐clavulanate dose: 2 g number of doses: 1 route of administration: intravenous timing of administration: 15 to 30 minutes before incision Control group (CG): N: 125 substance: nothing
Outcomes	Wound infection definition: discharge of purulent material for the wound, discharge of purulent material for the wound, the surgeon felt that would need to be open the wound to drain a possible infection. Follow‐up: 1 week till 1 month post operation SSSI N infected/N patients (%): PG: 1/124 (0.8%); CG: 0/123 (0%) DSSI: not reported Adverse effects: not reported
Notes	Spanish manuscript Participants < 18 years old were included in the study S. epidermidis bacteria were cultured from the wound infection Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote:“…. mediante una tabla de numeros Aleatorios generade por ordenador” (= “using a table of numbers randomly generated by computer ")
Allocation concealment (selection bias)	Unclear risk	No information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Patient self‐reporting. Quote: “antes de la intervencion los patientes fueron advisadosde la necesidad de acudir a la consulta si tenian fiebre (temperature mayor de 38 degrees Celcius), tumefaccion local and/or supuracion de la herida.(= Before the intervention the patients were advised of the need to go to the clinic if they had a fever (temperature greater than 38 degrees Celcius), local swelling and / or suppuration of the wound.)”
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data analysed as “available case analysis” (PG n = 1; CG n = 2). Missing outcome data balanced in numbers across intervention groups. Also the proportion of missing outcomes compared with the observed event risk is not enough to have had a clinically relevant impact on the intervention effect estimate. Quote: “tres pacientes precisaron ingreso despues la cirurgia y fueron excluidos (= “three patients required admission after the surgery and were excluded”)
Selective reporting (reporting bias)	Low risk	No study protocol available, but the published report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	Study seems to be free of other sources of bias

Othman 2011.

Study characteristics
Methods	Double‐blind RCT Country: Saudi Arabia Enrolment period: July 2006 to April 2010 Single centre
Participants	N: 98 participants Age (years; mean ± SD): PG: 43.4 ± 19.8; CG: 44.5 ± 20.5 % Men: PG 96%; CG 100% Inclusion criteria: unilateral or bilateral inguinal hernia Exclusion criteria: strangulated or obstructed hernia, patients on steroid treatment or chemotherapy or those who received antibiotics in the past 48 hours and patients with valvular heart disease or prosthetic heart valves BMI: not reported Skin antiseptic: 10% povidone iodine Operative time: PG 38.8 ± 10.8; CG 40.9 ± 11.1 Method of repair: Lichtenstein repair with polypropylene mesh (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia method: general or spinal Hospitalisation method: inpatient Length of hospital stay (days, mean ± SD): PG: 1.3 ± 0.463; CG: 1.25 ± 0.438 Surgeon: not reported
Interventions	Prophylactic group (PG): N: 50 antibiotic: amoxicillin‐clavulanic (Augmentin) dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: 30 minutes before induction Control group (CG): N: 48 substance: equal volume of sterile saline
Outcomes	Wound infection definition: CDC and APIC criteria Follow‐up: follow‐up every other day till removal of staples up to 1 month post operation SSSI N infected/N patients (%): PG: 3/50 (6%); CG: 5/48 (10.4%) DSSI N infected/N patients (%): PG: 1/50 (2%); CG: 1/48 (2.1%) Adverse effects: not reported
Notes	S. aureus, Streptococcus pyogenes,Pseudomonas aeruginosa bacteria were isolated from the wound infections Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Patients were randomly divided into two groups using a computer randomization program”
Allocation concealment (selection bias)	Low risk	Quote: “Randomization and preparation of drug and placebo were controlled by a surgery clinic nurse without the previous knowledge of the patient or surgeon.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Randomization and preparation of drug and placebo were controlled by a surgery clinic nurse without the previous knowledge of the patient or surgeon."
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information available
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of SSSI and DSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Perez 2005.

Study characteristics
Methods	Double‐blind RCT Country: Republic of the Philippines Enrolment period: January 2000 to December 2002 single centre
Participants	N: 360 participants Age (years; mean ± SD): PG: 61.37 ± 13.2; CG: 60.8 ± 14.5 % Men: 98% (both groups) Inclusion criteria: primary unilateral (direct and indirect) inguinal hernia. Exclusion criteria: pregnant or lactating women; those with earlier history of allergy, sensitivity, or anaphylaxis to beta‐lactam or cephalosporin antibiotics; antibiotic therapy within 48 hours before operation; presence of infection at the time of operation; patients with cardiac valvular problems; patients with prosthetic valves or joints; patients determined to be at increased risk of infection secondary to a coexisting medical condition; and patients with ASA class more than II. BMI: not reported Skin antiseptic: povidone iodine Operative time (minutes) (mean ± SD): PG 52.18 ± 16.4; CG 54.07 ± 15.3 Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty) Drain use: no drains were used ASA class: PG ASAI n = 146 (81%); ASAII n = 34 (19%); CG ASAI n = 138 (77%); ASAII n = 42 (23%) Anaesthesia method: regional Hospitalisation method: not reported Length of hospital stay: not reported Surgeon: senior surgical residents or consultants
Interventions	Prophylactic group (PG): N: 180 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: before incision Control group (CG): N: 180 substance: equal volume of sterile saline
Outcomes	Wound infection definition: CDC criteria Follow‐up: at discharge, 7 days post operation, 2 and 4 weeks post discharge SSSI N infected/N patients (%): PG: 3/180 (1.7%); CG: 6/180 (3.4%) DSSI N infected/N patients (%): PG: 1/180 (0.6%); CG: 1/180 (0.6%) Adverse effects: not reported
Notes	S. aureus, S. epidermidis were cultured from SSSI wound infections and S. aureus and Pseudomonas sp bacteria were cultured from infected meshes Reported: no competing interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Allocation was done with simple randomization using a computer‐generated table of random numbers.”
Allocation concealment (selection bias)	Low risk	Quote: “Randomization and preparation of drug and placebo were controlled by a surgery clinic nurse without the previous knowledge of the patient or surgeon.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Senior surgical residents or consultants, blinded to the study group, performed all operations.”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “All wounds were inspected before discharge, and all incisions were carefully reexamined by an independent surgeon blinded to the study during first followup visit…”
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data. Quote “The 10 patients who were excluded for failure to report for the last follow‐up were contacted and none had developed infections.”
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of SSSI and DSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Platt 1990.

Study characteristics
Methods	Double‐blind RCT Country: USA Enrolment period: 1 April 1985 to 30 September 1987 Multicentre (6)
Participants	N: 612 participants Age: (years; mean ± SD): PG: 51.0 ± 17.0; CG: 49.8 ± 17.6 % Men: PG 91%; CG 89% Inclusion criteria: unilateral or bilateral inguinal or femoral hernia. Exclusion criteria: operation in which insertion of prosthetic material was expected, people not able to speak English, living further away than 35 miles form the hospital, pregnant or lactating women; those with earlier history of allergy to beta‐lactam antibiotics; antibiotic therapy within 1 week before operation; presence of infection at the time of operation. BMI (mean ± SD): PG 24.6 ± 3.2; CG 24.7 ± 3.2 Skin antiseptic: PG: povidone iodine n = 235 (78%), alcohol or soap n = 62 (21%), other n = 4 (1%); CG povidone iodine n = 241 (77%), alcohol or soap n = 60 (19%), other n = 10 (3%) Operative time (mean ± SD; in minutes): PG 75 ± 32; CG 75 ± 30 Method of repair: herniorrhaphy Drain use: PG: n = 0 (0%); CG: n = 4 (1%) Anaesthesia method: PG: general n = 147 (49%), local n = 154 (51%); CG: general n = 154 (50%), local n = 157 (50%) ASA class: not reported Hospitalisation method: in and outpatient Length of hospital stay: PG: outpatient n = 58 (19%); admitted or discharged on day of surgery n = 167 (56%); admitted before day of surgery or discharged > 1 day after n = 76 (25%); CG: outpatient n = 58 (19%); admitted or discharged on day of surgery n = 171 (55%); admitted before day of surgery or discharged > 1 day after n = 82 (26%) Surgeon: not reported
Interventions	Prophylactic group (PG): N: 301 antibiotic: cefonicid dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: half hour before intervention (max 90 minutes before surgery) Control group (CG): N: 311 substance: 50 ml mixture of glycine, mannitol and riboflavin in 5% dextrose in water
Outcomes	Wound infection definition: erythema and drainage, or purulent discharge or a wound that was opened and not reclosed Follow‐up: to 1 week, 4 to 6 weeks post operation SSSI + DSSI N infected/N patients (%): PG: 4/301 (1.3%); CG: 6/311 (1.9%) Adverse effects: "no serious adverse events were attributable to treatment with Cefonid or placebo"
Notes	In this study were included patients with other types of surgery (mastectomy, lumpectomy, excisional breast biopsy, axillary‐node dissection for breast cancer or reduction mammoplasty. For meta‐analysis, only patients who underwent hernia repair were included. S. aureus bacteria were cultured from infected wounds. Authors (Wikler, Moonsammy, Jarosz) are employees of and stockholders in SmithKline Beecham Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “ patients….were randomly assigned separately in blocks of 10 to receive cefonicid or placebo” No information on how randomisation was accomplished. However, the block randomisation method indicates that the randomisation methodology has been performed carefully to minimise bias.
Allocation concealment (selection bias)	Unclear risk	Quote: “the treatment codes were not known by anyone at the participating centers, unless the sealed, opaque label attached to each vial was opened. " Not specified whether envelopes were sequentially numbered.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: ““the treatment codes were not known by anyone at the participating centers, unless the sealed, opaque label attached to each vial was opened.……Investigators were required to return these labels intact or to indicate the reason for opening them.” Quote: “none of the personnel at the data processing or coordinating center knew the treatment codes, and the codes were not revealed to …medical personnel…” Quote: “…the codes (i.e. treatment codes) were not revealed to the patients..”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “…drug assignments were not known during any follow‐up evaluations…for suspected wound infections”
Incomplete outcome data (attrition bias) All outcomes	High risk	Missing outcome data (exclusion of patients from analysis after randomisation) were reported and are balanced across treatment (antibiotic prophylaxis) and placebo group. However, numbers of missing outcome data are high (PG n = 50 and CG n = 51 from n = 1319 randomised patients for all different types of surgical procedures analysed in this study). These data are excluded from data‐analysis and available‐case analysis was performed. The proportion of missing outcomes compared with the observed event risk is enough to have induced clinically relevant bias in the intervention effect estimate.
Selective reporting (reporting bias)	Low risk	The study protocol was not available but report includes expected outcome wound infection.
Other bias	Low risk	The study appears to be free of other sources of bias

Rahmani 2012.

Study characteristics
Methods	Double‐blind RCT Country: Iran Enrolment period: 2010 to 2011 Multicentre (2)
Participants	N: 282 participants Age: 25 to 84 years old (not specified per treatment group; insignificant difference between treatment groups (p = 0.23) % Men: PG 44.7%; CG 40.8% Inclusion criteria: primary unilateral inguinal hernias. Exclusion criteria: emergency repair of hernias (incarcerated‐strangulated hernia), allergy to cefalotin or penicillin, recurrent hernia, taking antibiotics because of other situations like infectious disease or cardiac valve disease, patients with deficiency of immune system because of disease (diabetes, HIV, malignancies) or taking medications (immunosuppressant, corticosteroids), not receiving patient consent for joining the study BMI: not specified per treatment group; insignificant difference between treatment groups (P = 0.66) Skin antiseptic: not reported Operative time: not specified per treatment group, insignificant difference between treatment groups (P = 0.12) Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty) Drain use: PG: n = 11; CG n = 7 ASA class: not reported Anaesthesia method: not reported Hospitalisation method: outpatient Length of hospital stay: not reported Surgeon: (Senior) surgeon or resident
Interventions	Prophylactic group (PG): N: 141 antibiotic: cephalotin ((keflin) dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: 30 minutes before incision Control group (CG): N: 141 substance: equal volume of distilled water
Outcomes	Wound infection definition: CDC criteria. Follow‐up: 1, 2, 12 weeks post operation SSSI N infected/N patients (%): PG: 4/141 (2.8%); CG: 8/141 (5.7%) DSSI N infected/N patients (%): PG: 0/141 (0%); CG: 1/141 (0.7%) Adverse effects: not reported
Notes	The most prevalent bacterial species cultured form wound infections was S. aureus (50% of cases) Article in Farsi Funding: Mazandaran University of Medical Sciences
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: “This is a double‐blind randomised trial …” but no further information on how randomisation was performed was provided.
Allocation concealment (selection bias)	Unclear risk	No information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “This is a double‐blind randomised trail …” No further information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	High risk	The protocol was registered after conducting the study (irct.ir/trial/1359).
Other bias	Low risk	The study appears to be free of other sources of bias

Razack 2015.

Study characteristics
Methods	Double‐blind RCT Country: India Enrolment period: June 2013 to February 2015 single centre
Participants	N: 200 participants Age: (years; mean ± SD) PG: 42.44 ± 15.61; CG: 45.56 ± 15.43 % Men: PG 98.9%; CG 100% Inclusion criteria: primary unilateral or bilateral hernia repaired with Lichtenstein repair, age 15 to 70 years. Exclusion criteria: patients allergic to the given antibiotic, strangulated hernia, recurrent hernia, diabetes mellitus, pregnancy or lactation, immunocompromised state (i.e. HIV, malignancy, steroid medication), antibiotic treatment within last 5 days for a different reason or existing indication (i.e. valvular heart disease) BMI: not reported Skin antiseptic: povidone iodine Operative time(mean ± SD; in minutes): PG: 53.54 ± 15.82; CG: 52.60 ± 15.28 Method of repair: Lichtenstein technique (hernioplasty) with monofilament polypropylene mesh. Drain use: not reported ASA: PG: ASAI n = 88 (93.6%), ASAII n = 6 (6.8%); CG: ASAI n = 80 (93%), ASAII n = 6 (6.9%) Anaesthesia method: not reported Hospitalisation method: inpatient Length of hospital stay (mean ± SD, days): PG: 3.9 ± 3.33; CG 4.1 ± 2.07 (pre‐operative stay in days); patients were discharged next day post operation Surgeon: not reported
Interventions	Prophylactic group (PG): N: 94 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration:intravenous timing of administration: just before incision Control group (CG): N: 86 substance: equal volume of sterile saline
Outcomes	Wound infection definition: CDC criteria/ASEPSIS criteria Follow‐up: 2 days, 1 week, 4 weeks post operation SSSI+DSSI N infected/N patients (%) PG: 7/94 (7.4%); CG: 8/86 (9.3%) DSSI: 1 case (not specified in which group) ((reported: "one mesh was removed") Adverse effects: not reported
Notes	participants < 18 years old were included in this study Staphylococcus aureus, Streptococci, Klebsiella pneumonia, E. coli, enterobacteria were isolated from the wound infections Reported: no financial or competing interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information available how randomisation was accomplished. Quote: “200 patients were randomized into antibiotic group and control group by sealed envelope method on the day before the surgery”
Allocation concealment (selection bias)	Unclear risk	Quote: “200 patients were randomized into antibiotic group and control group by sealed envelope method on the day before the surgery” Not specified whether envelopes were sequentially numbered and opaque.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “Surgeon who was not involved in surgery followed the case after 1 week and 4 weeks, postoperatively”
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: “The study was concluded in Feb 2015, by then, out of 200 patients who had entered the study, 180 patients had completed one month follow up.” “Among the 200 patients with one month follow up, 94 were in the antibiotic group and 86 were in the control group.” Unclear why not all randomised patients were analysed. The missing outcome data were not balanced in numbers across intervention groups (PG n = 6 (6%); CG n = 14 (14%)) and the proportion of missing outcomes compared with the observed event risk is enough to have induced clinically relevant bias in the intervention effect estimate. Available case analysis was performed.
Selective reporting (reporting bias)	Unclear risk	No reporting that study protocol was registered. Authors failed to report the number of cases with SSSI or DSSI per treatment group and therefore meta‐analysis for these subgroups is not possible. Quote: "No significant difference was found between the study groups on analyzing the sub types of infection".
Other bias	Low risk	Study appears to be free of other sources of bias

Shankar 2010.

Study characteristics
Methods	Double‐blind RCT Country: India Enrolment period: November 2006 to June 2008 Single centre
Participants	N: 450 participants Age: (years; mean ± SD) PG: 44.44 ± 15.59; CG: 45.56 ± 16.43 % Men: PG: 99.4%; CG: 98.1% Inclusion criteria: primary unilateral or bilateral uncomplicated inguinal hernia with polypropylene mesh repair Exclusion criteria: recurrent hernia, immunosuppressive disease, diabetes mellitus BMI: not reported Skin antiseptic: ioprep (povidone iodine) Operative time (mean ± SD; minutes): PG 53.54 ± 15.82; CG 52.60 ± 15.28 Method of repair: standard tension free repair using polypropylene mesh (hernioplasty) Drain use: not reported ASA class: PG: ASAI n = 157 (91.3%), ASAII n = 15 (8.7%); CG: ASAI n = 151 (93.2%), ASAII n = 11 (6.8%) Anaesthesia: PG: SA n = 153 (89%), GA n = 5 (2.9%), LA n = 13 (7.6%), epidural n = 1 (0.6%); CG SA n = 142 (87.7%), GA n = 7 (4.3%), LA n = 13 (8.0%) Hospitalisation: inpatient Length of hospital stay: preoperative stay (mean days ± SD): PG: 4.38 ± 3.11; CG: 3.93 ± 3.0 Surgeon: PG: consultant 14.5%, resident 85.5%; CG: staff 8.1%, resident 92%
Interventions	Prophylactic group (PG): N: 225 antibiotic: cefazolin dose: 1 g number of doses: 1 route of administration: intravenous timing of administration: induction Control group (CG): N: 225 substance: normal saline
Outcomes	Wound infection definition: CDC criteria Follow‐up: 2, 7 to 10 and 30 days post operation SSSI N infected/N patients (%): PG:11/172 (7%); CG: 16/162 (10.5%) DSSI N infected/N patients (%): PG 1/172 (0.6%); CG 1/162 (0.6%) Adverse effects: not reported
Notes	Participants < 18 years old were included in this study Staphylococcus aureus, Streptococci, Klebsiella pneumonia, E. coli, enterobacteria were isolated from the wound infections Reported: no conflict of interest. Funding by the institute.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: “After informed consent, 450 patients were randomized into antibiotic group and control group by sealed envelope method on the day before surgery” Not specified how randomisation was established
Allocation concealment (selection bias)	Unclear risk	Quote: “After informed consent, 450 patients were randomized into antibiotic group and control group by sealed envelope method on the day before surgery” Not specified whether envelopes were sequentially numbered and opaque.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quoted: "Follow‐up was done by residents who where blinded to the drug used."
Incomplete outcome data (attrition bias) All outcomes	High risk	Missing outcome data (exclusion of patients from analysis after randomisation) were reported and are balanced across treatment groups. However, numbers of missing outcome data are very high (PG n = 53 and CG n = 63 from n = 450 randomised (24% to 26%)). These missing data were excluded from data analysis and available‐case analysis was performed. The proportion of missing outcomes compared with the observed event risk is enough to have induced clinically relevant bias in the intervention effect estimate.
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of SSSI and DSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Taylor 1997.

Study characteristics
Methods	Double‐blind RCT Country: UK Enrolment period: September 1993 to January 1995 Multicentre (6)
Participants	N: 619 participants Age: (years; mean ± SD) PG: 56.7 ± 17.4; CG: 56.6 ± 16.5 % Men: PG 94.7%; CG 94.6% Inclusion criteria: inguinal or femoral hernia, age > 18 years Exclusion criteria: hypersensitivity to penicillin or cephalosporin, had taken antimicrobial agents within the preceding 72 h, bilateral hernia, pregnant or lactating, renal or liver impairment, had previous been entered into the study, or patients for whom antibiotic prophylaxis was indicated for some other reason such as valvular disease. BMI: not reported Skin antiseptic: betadine or chlorhexidine Operative time: not reported Method of repair: herniorrhaphy (Bassini or Shouldice or darn, n = 508), hernioplasty (Lichtenstein, n = 33), unknown n = 22 Drain use: 4 patients in each group ASA class: not reported Anaesthesia method: local, spinal, general Hospitalisation method: inpatient Length of hospital stay: not reported Surgeon: Consultant, (senior) registrar/Staff or senior house officer
Interventions	Prophylactic group (PG): N: 283 antibiotic: co‐amoxicillin clavulanic acid dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: induction or start surgery Control group (CG): N: 280 substance: equal volume of sterile saline
Outcomes	Wound infection definition: as a purulent wound discharge or spreading erythema indicative of cellulitis, wound breakdown, or dehiscence with clinical evidence of infection. Follow‐up: 4 to 6 weeks SSSI + DSSI N infected/N patients (%): PG: 25/283 (8.8%); CG: 25/280 (8.9%) DSSI: not reported Adverse effects: not reported
Notes	S. aureus, S. albus, Streptococci (beta‐hemolytic, anaerobic, unspecified), Coliforms, Enterobacter aerogens bacteria were cultured from the wound infections Funding, supply of co‐amoxiclav antibiotics and statistical analysis assistance by SmithKline Beecham Pharmaceuticals Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “randomization was from a computer‐generated code in blocks of four.”
Allocation concealment (selection bias)	Unclear risk	Insufficient information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Unclear whether general practitioner or district nurses were blinded to treatment and therefore it is unclear whether outcome assessment was properly blinded. Quote: “Patients were given a diary card on which their general practitioner or district nurse was asked to record any suggestion of a wound or other infection…”
Incomplete outcome data (attrition bias) All outcomes	High risk	Missing outcome data (exclusion of patients from analysis after randomisation) were reported and seem to be balanced across treatment groups. However, numbers of missing outcome data are considerablely high (n = 56 from n = 619 randomised (9%)). These data are excluded from data analysis and available‐case analysis was performed. The proportion of missing outcomes compared with the observed event risk is enough to have induced clinically relevant bias in the intervention effect estimate.
Selective reporting (reporting bias)	Low risk	The study protocol was not available but report includes expected outcome wound infection.
Other bias	Low risk	The study appears to be free of other sources of bias

Thakur 2010.

Study characteristics
Methods	Double‐blind RCT Country: India Enrolment period: not reported Single centre
Participants	N: 55 participants Age: not specified (all adult) % Men: 100% Inclusion criteria: unilateral or bilateral elective inguinal hernia Exclusion criteria: femoral, strangulated or obstructed hernias; steroid therapy or chemotherapy; had received antibiotics in the past 48 hours; valvular heart disease or prosthetic heart valves or joints. BMI: not reported Skin antiseptic: 10% povidone iodine Operative time: PG 79.3% (n = 23) had the total duration of surgery > 1 hour; CG 92.3% (n = 24) had the total duration of surgery > 1 hour. Method of repair: Lichtenstein technique using monofilament polypropylene mesh (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia: not reported Hospitalisation: not reported Length of hospital stay: not reported Surgeon: not reported
Interventions	Prophylactic group (PG): N: 29 antibiotic: cefuroxime sodium dose: 1,5 g number of doses: 1 route of administration: intravenous timing of administration: 30 minutes before induction Control group (CG): N: 26 substance: equal volume sterile saline
Outcomes	Wound infection definition: Center of Diseases Control criteria (as described by Teresa and colleagues 1992) Follow‐up: discharge, 1 week, 1 month SSSI N infected/N patients (%): PG: 3/29 (10.3%); CG: 3/26 (11.5%) DSSI N infected/N patients (%): PG: 0/29 (0%); CG: 1/26 (3.8%) Adverse effects: not reported
Notes	S. aureus bacteria were cultured from the wound infections Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	No information on how randomisation was accomplished. However, the block randomisation method indicates that the randomisation methodology has been performed carefully to minimise bias. Quote: “Randomization was accomplished by block randomization”.
Allocation concealment (selection bias)	Unclear risk	No information available
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information available, but performance bias is unlikely in this setting
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of SSSI and DSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Tzovaras 2007.

Study characteristics
Methods	triple blind RCT Country: Greece Enrolment period: January 2000 and June 2004 Single centre
Participants	N: 386 participants Age (years; median (range)): PG: 63 (17 to 87); CG: 63 (15 to 90) % Men: PG 94.2%; CG 93.7% Inclusion criteria: open inguinal hernia repair with mesh technique, age > 15 years Exclusion criteria: allergy to penicillin, antibiotic intake within the last 5 days before the operation, bilateral hernia repair at the same operation, pregnancy or lactation, existing indication for antibiotics prophylaxis (i.e. valvular heart disease) and known renal or liver impairment (potentially immunocompromised patients) BMI (median (range)): PG: 26 (18 to 34); CG: 26 (20 to 33) Skin antiseptic: povidone iodine Operative time (median (range)): PG: 45 (20 to 90); CG: 45 (20 to 80) Method of repair: tension free repair with plain or pre‐shaped polypropylene mesh (hernioplasty) Drain use: PG n = 7 (3.7%); CG n = 8 (4.2%) ASA class: PG: ASAI n = 137 (72%), ASAII n = 47 25%), ASAIII n = 6 (3%); CG: ASAI n = 132 (70%), ASAII n = 53 (28%), ASAIII n = 4 (2%) Anaesthesia: PG: general n = 11 (6%), spinal/regional n = 16 (8%), local n = 163 (86%) /CG: general n = 9 (5%), spinal/regional n = 14 (7%), local n = 166 (88%) Hospitalisation: inpatient Length of hospital stay: The patients were discharged (if they agreed) on the evening of the procedure when this was performed under local anaesthesia, or the day after the operation. Surgeon: consultant surgeon or surgical trainee under consultant’s assistance and supervision
Interventions	Prophylactic group (PG): N: 193 antibiotic: ampicillin + clavoulanic acid dose: 1.2 g number of doses: 1 route of administration: intravenous timing of administration: induction. Control group (CG): N: 193 substance: equal volume sterile saline
Outcomes	Wound infection definition: purulent discharge (or serosanguineous with positive culture) or spreading erythema indicative of cellulitis or wound breakdown/dehiscence with clinical evidence of infection Follow‐up: 1 week, 1 month after operation SSSI N infected/N patients (%): PG: 5/190 (2.6%); CG: 9/189 (4.8%) DSSI N infected/N patients: not reported (mesh removal was not required in any case of wound infection) Adverse effects: not reported
Notes	Participants < 18 years old were included in this study S. aureus, S. epidermidis, Streptococcus beta‐haemolytic bacteria were cultured from wound infections Reported is: no conflict of interest or bias
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “The list with code numbers generated by the Arcus Quistat software program was kept by a secretary not involved in the treatment of the patients at any stage and was opened at the end of the trial for analysis of the results”
Allocation concealment (selection bias)	Unclear risk	Not specified whether envelopes were sequentially numbered and opaque. Quote: “Patients were randomized ….using numbered sealed envelopes”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “the agent (antibiotic of placebo)was prepared by the anesthetic nurse and neither the surgeon or the patient was aware of the randomization arm”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “They were followed‐up as outpatients at 1 week and 1 month after the operation by one of the authors who was not involved in surgery and therefore was blind to the randomization arm”
Incomplete outcome data (attrition bias) All outcomes	Low risk	A small number of patients is lost‐in‐follow‐up (PG n = 3 (1.6%); CG n = (2.1%)). Missing data were balanced between treatment groups and reasons for missing outcome data unlikely to be related to the true outcome. The proportion of missing outcomes compared with the observed event risk is not enough to have had a clinically relevant impact on the intervention effect estimate. Available case analysis was performed. Quote: “Three group A patients and four group B patients were excluded because they had inadvertent medical administration (n = 3), were lost‐to‐follow‐up (n = 2), were found to have unexpected femoral hernia (n = 1) or had hernia repair without mesh insertion (n = 1)”
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of SSSI.
Other bias	Low risk	The study appears to be free of other sources of bias

Ullah 2013.

Study characteristics
Methods	single blind RCT Country: Pakistan Enrolment period: January to December 2012 Single centre
Participants	N: 166 participants Age: (years; mean ± SD): PG: 54.33 ± 11.77; CG: 52.58 ± 11.80 % Men: 100% Inclusion criteria: unilateral inguinal hernia, male patients, age > 18 years Exclusion criteria: diabetes (fasting glucose > 126 mg/dl), history of intake of steroids in last 2 weeks, obstructed/strangulated or recurrent hernia, chronic liver disease, body mass index ≥ 29 and impaired renal function BMI: not reported Skin antiseptic: not reported Operative time: not reported Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty) Drain use: not reported ASA class: not reported Anaesthesia: not reported Hospitalisation: inpatient Length of hospital stay: 2 days Surgeon: a single general surgeon
Interventions	Prophylactic group (PG): N: 83 antibiotic: co‐amoxiclav dose: 1 g number of doses: 1 route of administration: not reported timing of administration: 1 hour before intervention Control group (CG): N: 83 substance: placebo (not specified)
Outcomes	Wound infection definition: not reported Follow‐up: 2 days, 14 days after operation after operation SSSI N infected/N patients (%): PG: 6/83 (7.2%); CG: 15/83 (18.1%) DSSI: not reported Adverse effects: not reported
Notes	Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	No information on how randomisation was accomplished. However, the block randomisation method indicates that the randomisation methodology has been performed carefully to minimise bias. Quote: “The patients diagnosed as inguinal hernia were taken through OPD, and were randomly placed into 2 groups using block method of randomisation (block size of 4).”
Allocation concealment (selection bias)	Unclear risk	No information available.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information on patient blinding available, but performance bias is unlikely in this setting Quote: “All patients in both groups were put on OT list for the next OT day after following routine and standard preoperative preparations kept uniform for all participants.”
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information available.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data.
Selective reporting (reporting bias)	Low risk	The study protocol was not available but report includes all expected outcomes, including specification of SSSI
Other bias	Low risk	The study appears to be free of other sources of bias

Wang 2013.

Study characteristics
Methods	Double‐blind RCT Country: China Enrolment period: January 2010 to July 2011 Multicentre (6)
Participants	N: 1200 participants Age (years; mean ± SD): PG Group 1: 53 ± 16.9; PG Group 2: 54.6 ± 16.2; CG: 56 ± 17.3 % Men: PG Group 1: 90%; PG Group 2: 91%; CG: 90% Inclusion criteria: primary inguinal hernia, aged 16 to 81 years old, no infection or haematological system disease before surgery, normal liver and kidney function, no pre‐operative valvular heart disease, Body Mass Index (BMI) 18 to 32 kg/m² Exclusion criteria: failed to meet 1 of the inclusion criteria above; malignant tumour; history of immune system diseases; using or had used corticosteroids or other immunosuppressants less than 1 month before surgery, using or had used antibiotics less than 3 days before surgery; pregnancy or lactation. BMI: Range 18 to 32, not reported per treatment group Skin antiseptic: Povidone iodine (PI) or chlorhexidine acetate (CA) PG Group 1: PI n = 292 (76%) and CA n = 91 (24%); PG group 2: PI n = 303 (79%) CA n = 82 (21%); CG: PI n = 290 (74%) CA n = 102 (26%) Operative time: not reported Method of repair: standard open tension free inguinal hernia repair with mesh plug Drain use: not reported ASA class: ASA ≥ 1: PG group 1: 77.55%, PG group 2: 84.31%; CG 85.46% Anaesthesia method: PG group 1 epidural n = 270 (71%) general n = 35 (9%), local n = 78 (20%); PG group 2 epidural n = 326 (85%), general n = 23 (6%), local n = 36 (9%)/ CG epidural n = 202 (51%), general n = 66 (17%), local n = 124 (32%) Hospitalisation method: inpatient Length of stay (mean ± SD): PG group 1: 3.23 ± 0.54; PG group 2: 3.03 ± 0.81; CG 3.12 ± 0.76 Surgeon: not reported
Interventions	Prophylactic group (PG): N: 800 (Group 1: 400; Group 2: 400) antibiotic: cefazolin (Group 1); levofloxacin (Group 2) dose: group 1: 1 g; group 2: 200 mg number of doses: 1 route of administration: intravenous timing of administration: 30 to 60 minutes before operation Control group (CG): N: 400 substance: equal volume of sterile saline
Outcomes	Wound infection definition: CDC criteria. Follow‐up: 1 week, 2 weeks, 3 weeks up to 30 days after operation. SSSI N infected/N patients (%): PG: (group 1 + group 2) = 29 (13 + 16)/768 (3,8%); CG: 19/392 (4.8%) DSSI N infected/N patients (%): PG (group 1 + group 2) = 3 (2 + 1)/768 (0.4%); CG: 1/392 (0.3%) (reported: "one mesh was removed") Adverse effects: 1 participant in the cefazolin treatment group and 5 participants in the levofloxacin treatment group failed to tolerate antibiotics
Notes	Participants age < 18 years old were included in this study Mainly Staphylococcus aureus, Staphylococcus epidermidis (skin) and Escherichia coli, Enterococci, Enterococcus faecalis (gut) bacteria were cultured from infections. Also Acinobacter baumanii, Candida tropicalis, Enterobacter cloacae, Haemolytic streptococcus, Klebsiella pneumoniae, Candida albicans, MRSA, Pseudomonas aeruginosa, bacteria were cultured from infections (incl. wound, urinary tract and pulmonary and mesh infections) Reported is: no conflict of interest
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Allocation was accomplished ….with a computer randomization program (Randomization Adviser 1.0)”
Allocation concealment (selection bias)	Low risk	Quote: “Allocation was accomplished by a resident who was unaware of this trial and was not involved in surgery, data collection, follow‐up or the analysis of results with a computer randomization program (Randomization Adviser 1.0)”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “Drugs and placebo were prepared by a surgery nurse without previous knowledge of this trial.” Quote: “On all patients a standard open tension‐free inguinal hernia repair operation was performed by surgeons‐in‐charge (blinded) using a polypropylene mesh”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “The main endpoint was the amount of post‐operation infection patients evaluated by surgeons and nurses (blinded) within 30 days after operation.”
Incomplete outcome data (attrition bias) All outcomes	Low risk	A small number of participants was lost‐in‐follow‐up (PG n = 32 (4%); CG n = 8 (2%)). Missing data were not balanced across intervention groups, but the proportion of missing outcomes compared with the observed event risk is not enough to have had a clinically relevant impact on the intervention effect estimate. Available‐case analysis was performed.
Selective reporting (reporting bias)	Low risk	The protocol was registered (NCT 01802606) and study outcomes that are of interest in the review have been reported in the prespecified way.
Other bias	Low risk	The study appears to be free of other sources of bias

Yerdel 2001.

Study characteristics
Methods	Triple‐blind RCT Country: Turkey Enrolment period: between September 1996 and July 1998 Single centre
Participants	N: 280 participants Age: (years; mean ± SD) PG: 55.57 ± 15.1; CG: 55.78 ± 13.8 % Men: PG: 90%; CG: 95% Inclusion criteria: primary, unilateral inguinal hernia elective repair Exclusion criteria: recurrent, irreducible, strangulated, bilateral or femoral hernias; giant scrotal hernias with massive defects were also excluded, systemic or advanced disease (e.g. diabetes, liver or renal impairment), ASA score more than II, patients receiving steroids for any reason, allergy to antibiotics, intake of antibiotics less than a week before surgery, and pregnant or lactating women. BMI: PG: 24.95 ± 2.6; CG: 25.02 ± 3.0 Skin antiseptic: povidone iodine Operative time (mean ± SD; minutes): PG: 64.18 ± 22.8; CG: 62.78 ± 19.3 Method of repair: Lichtenstein technique with polypropylene mesh (hernioplasty). Drain use: PG: n = 29 (21%); CG: n = 31 (23%) ASA class: PG: ASAI n = 97 (71%), ASAII n = 39 (29%); CG: ASAI n = 101 (76%), ASAII n = 32 (24%) Anaesthesia method: PG local n = 56 (41%), spinal n = 54 (40%), general n = 26 (19%); CG local n = 55 (41%), spinal n = 63 (47%), general n = 15 (11%) Hospitalisation: inpatient Length of hospital stay: not reported Surgeon: not reported
Interventions	Prophylactic group (PG): N: 140 antibiotic: ampicillin‐sulbactam dose: 1.5 g (1g + 0.5g) number of doses: 1 route of administration: intravenous timing of administration: before incision Control group (CG): N: 140 substance: equal volume sterile saline
Outcomes	Wound infection definition: CDC criteria Follow‐up: discharge, 7 to 9 days and 4 to 6 weeks, 6 months, 12 months post operation SSSI N infected/N patients (%): PG: 0/136 (0%); CG: 9/133 (6.8%) DSSI N infected/N patients (%): PG: 1/136 (0.7%); CG: 3/133 (2.3%) (reported: "three meshes were removed") Adverse effects: not reported
Notes	S. aureus and S. epidermidis bacteria were cultured from wound infections. Funding is not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Randomisation was accomplished by a computer‐generated code…”
Allocation concealment (selection bias)	Low risk	Quote: “Randomisation was accomplished …by a resident….. He was unaware of the research in progress and was never involved in surgery, data collection or patient follow‐up”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “…by a resident who also prepared the sealed antibiotic or placebo syringes. He was unaware of the research in progress …”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “A surgeon masked to the randomization and patient details evaluated the wound during each follow‐up.”
Incomplete outcome data (attrition bias) All outcomes	Low risk	A small number of patients were lost to follow‐up (PG n = 4 (2.9%); CG n = 7 (5%)). Missing data were not balanced across intervention groups, but the proportion of missing outcomes compared with the observed event risk is not enough to have had a clinically relevant impact on the intervention effect estimate. Available‐case analysis was performed. Quote: “All patients lost‐to‐follow‐up were contacted by phone and none had any wound problems during the year after surgery”
Selective reporting (reporting bias)	Low risk	The study protocol is not available but report includes all expected outcomes, including specification of both SSSI and DSSI
Other bias	High risk	Study was terminated early after interim analysis, and therefore there might be risk of bias for increased benefit of in the antibiotics prophylaxis group. Quote: "Because of the high rate of wound infections, the code was broken after the discharge of patient 280 (140 patients in the placebo and 140 patients in the antibiotic group). At that point, the results revealed an extreme discrepancy in favour of antibiotics prophylaxis and therefore the study was prematurely stopped before reaching the previously established sample size of 334"

PG = prophylaxis group, CG = control (placebo) group, SD= standard deviation

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Ahmed 2014	Considered plagiarism of published report of Goyal 2011. Results section and large text fragments of Ahmed 2014 were similar to Goyal et al.. The study by Goyal and colleagues was published in 2011 and the study by Ahmed and colleagues was published in 2014, therefore Ahmed 2014 was considered plagiarism and excluded from analysis.
Aufenacker 2006	Wrong study design. This is a systematic review and meta‐analysis.
Barreca 2000	Wrong study design; non‐randomised study.
Bhuiyan 2017	Wrong comparator. Different antibiotic prophylaxis regimes were compared for prevention of wound infection during inguinal hernia repair. This is an abstract for the ASGBI conference.
Boonchan 2017	Wrong study design. This is a systematic review and meta‐analysis.
Erdas 2016	Wrong study design. This is a systematic review and meta‐analysis. Cross‐referencing resulted in 2 additional RCTs that were included in this review and meta‐analysis.
Esposito 2006	Incomplete data. Study focused on clean surgical techniques including hernia repair, but data extraction for this subgroup of participants was not possible
Gierhake 1975	Wrong study design; non‐randomised study.
Lazorthes 1993	Duplicate publication in French of similar study population of Lazorthes 1992
Leon 2011	Wrong study design; non‐randomised study.
Li 2012	Wrong study design. This is a systematic review and meta‐analysis.
Mazaki 2013b	Wrong study design. This is a systematic review and meta‐analysis.
Mehrabi Bahar 2015	Incomplete data. Study focused on clean surgical techniques including hernia repair, but data extraction for this subgroup of participants was not possible.
Pessaux 2006	Prognostic study based on 3 randomised studies that compared antibiotics prophylaxis versus another antibiotics prophylaxis regime.
Sanabria 2007	Wrong study design. This is a systematic review and meta‐analysis.
Sanchez‐Manuel 2003	Wrong study design. This is a systematic review and meta‐analysis.
Sanchez‐Manuel 2012	Wrong study design. This is a systematic review and meta‐analysis.
Taylor 1996	Wrong study design; non‐randomised study.
Yin 2012	Wrong study design. This is a systematic review and meta‐analysis.

Characteristics of studies awaiting classification [ordered by study ID]

Barkhordary 2012.

Methods	Double‐blind RCT
Participants	elective hernioplasty with polypropylene mesh (n = 100 adult male participants)
Interventions	1 g cefazolin (IV) versus sterile saline (placebo)
Outcomes	surgical site infections, follow‐up at 1 week, 6 to 8 weeks and 12 weeks postoperatively
Notes	IRCT2012090210719N1; registered while recruiting

Differences between protocol and review

This is a comprehensive update from previous versions of this review, with new insights regarding data‐extraction, 'Risk of bias' assessment, methodological analysis etc. All identified studies from the previous updates were reassessed according to new Cochrane Handbook for Systematic Reviews of Interventions and Methodological Expectations of Cochrane Intervention Reviews (MECIR) guidelines (Higgins 2016; Higgins 2017).

The outcome 'wound infections' from previous versions of this review was divided into superficial and deep surgical site infections according to the classification of the latest CDC criteria. The outcome 'all wound infections (SSSI + DSSI)' is similar to the outcome 'wound infections' in previous versions of this review.

A new subgroup analysis was performed for studies in a low infection risk environment (< 5% SSSI infection in the placebo group) and a high infection risk environment (≥ 5% SSSI in the placebo group).

Nine studies included participants aged younger than 18 years (Lazorthes 1992; Oteiza 2004; Tzovaras 2007; Shankar 2010; Al‐Fatah 2011; Goyal 2011; Wang 2013; Kochhar 2014; Razack 2015). Four of these studies were included in the previous version of this review (Lazorthes 1992; Oteiza 2004; Tzovaras 2007; Shankar 2010).

Contributions of authors

Amending the protocol, selecting studies, assessing risk of bias, extracting data, performing analysis (including application of GRADE criteria) and writing the draft review: Claudia Orelio, Coen van Hessen, Franscisco Javier Sanchez‐Manuel, Rob Scholten

Revision of the draft review, critical interpretation of the data: Theodorus Aufenacker

All authors approved the final version of the review and have agreed to be accountable for all aspects of the review.

Sources of support

Internal sources

• No sources of support supplied

External sources

• Knowledge Institute of Medical Specialists (KIMS), Utrecht, Netherlands

  The Dutch Knowledge Institute of Medical Specialists commissioned and financially supported a series of data‐syntheses for the Dutch Inguinal Hernia guideline group. Antibiotic prophylaxis for prevention of wound infection was one of the addressed research questions. Identification of several new RCTs as compared to the 2012 version of this review resulted in this new update.

Declarations of interest

Claudia Orelio: None known.

Coen van Hessen: None known.

Francisco Javier Sanchez‐Manuel: None known.

Theodorus Aufenacker: None known.

Rob Scholten: None known.

Edited (no change to conclusions)