Profilaxis antibiótica para la prevención de la infección de heridas posoperatorias en adultos sometidos a reparación abierta electiva de hernia inguinal o femoral
Resumen
Antecedentes
La hernia inguinal o femoral es una protuberancia tisular en la región de la ingle que tiene una incidencia acumulada del 27% en los hombres adultos y del 3% en las mujeres adultas. Debido a que la mayoría de las hernias se vuelven sintomáticas con el tiempo, la reparación de la hernia inguinal es uno de los procedimientos quirúrgicos realizados con mayor frecuencia en todo el mundo. Este tipo de cirugía se considera una cirugía «limpia», y se espera que las tasas de infección de la herida sean inferiores al 5%. Por lo general, para procedimientos quirúrgicos limpios no se recomienda la profilaxis con antibióticos. Sin embargo, después de la introducción de la reparación de la hernia con malla y la publicación de estudios que presentan tasas altas de infección de las heridas, volvió a comenzar el debate sobre la necesidad de profilaxis con antibióticos para prevenir las infecciones posoperatorias de las heridas.
Objetivos
Determinar la efectividad de la profilaxis con antibióticos para reducir las infecciones (superficiales y profundas) posoperatorias de las heridas en la reparación abierta electiva de hernia inguinal y femoral.
Métodos de búsqueda
Se realizaron búsquedas en varias bases de datos electrónicas: Registro Cochrane de Estudios en Línea (Cochrane Registry of Studies Online), MEDLINE Ovid, Embase Ovid, Scopus y Science Citation Index (búsqueda realizada el 12 de noviembre 2019). También se realizaron búsquedas en dos registros de ensayos y en la lista de referencias de los estudios incluidos.
Criterios de selección
Se incluyeron ensayos controlados aleatorizados que comparaban cualquier tipo de profilaxis con antibióticos versus placebo o ningún tratamiento para la prevención de las infecciones posoperatorias de la herida en adultos sometidos a una cirugía de reparación abierta de hernia inguinal o femoral (reparación con tejido y reparación con malla).
Obtención y análisis de los datos
Dos autores de la revisión seleccionaron los estudios, extrajeron los datos y evaluaron el riesgo de sesgo de forma independiente. Se analizaron por separado los resultados de dos métodos quirúrgicos diferentes (herniorrafia y hernioplastia). Varios estudios revelaron tasas de infección superiores al 5% esperado para la cirugía limpia y, por lo tanto, los estudios se dividieron en dos subgrupos: entornos de alto riesgo de infección (tasa de infección ≥ 5%); y entornos de bajo riesgo de infección (tasa de infección < 5%). Se realizaron metanálisis con modelos de efectos aleatorios. Se analizaron tres resultados: infecciones superficiales del sitio quirúrgico (SSSI, por sus siglas en inglés); infecciones profundas del sitio quirúrgico (DSSI, por sus siglas en inglés); y todas las infecciones posoperatorias de la herida (SSSI + DSSI).
Resultados principales
En esta actualización de la revisión se identificaron e incluyeron 10 estudios nuevos. En total, se incluyeron 27 estudios (con 8308 participantes) en esta revisión.
No se sabe con certeza si la profilaxis con antibióticos en comparación con placebo (o ningún tratamiento) previene todos los tipos de infecciones posoperatorias de la herida después de la cirugía de herniorrafia (riesgos relativos [RR] 0,86; intervalo de confianza [IC] del 95%: 0,56 a 1,33; cinco estudios, 1865 participantes; evidencia de calidad muy baja). El análisis de subgrupos no cambió estos resultados. No se pudieron realizar metanálisis para la SSSI o la DSSI, ya que estos resultados no se informaron por separado.
Veintidós estudios estaban relacionados con la cirugía de hernioplastia (con un total de 6443 participantes) y se analizaron tres resultados: SSSI; DSSI; SSSI + DSSI.
Dentro del subgrupo de entorno de bajo riesgo de infección, la profilaxis con antibióticos en comparación con placebo probablemente logra poca o ninguna diferencia para los resultados «prevención de todas las infecciones de la herida» (RR 0,71; IC del 95%: 0,44 a 1,14; evidencia de calidad moderada) y «prevención de la ISSI» (RR 0,71; IC del 95%: 0,44 a 1,17; evidencia de calidad moderada). Dentro del subgrupo de entorno de alto riesgo de infección no se conoce si la profilaxis con antibióticos reduce todos los tipos de infecciones de las heridas (RR 0,58; IC del 95%: 0,43 a 0,77; evidencia de calidad muy baja) o la SSSI (RR 0,56; IC del 95%: 0,41 a 0,77; evidencia de calidad muy baja). Cuando se combinan participantes de ambos subgrupos, la profilaxis con antibióticos, en comparación con placebo, probablemente reduce el riesgo de todos los tipos de infecciones de las heridas (RR 0,61; IC del 95%: 0,48 a 0,78) y de SSSI (RR 0,60; IC del 95%: 0,46 a 0,78; evidencia de calidad moderada).
La profilaxis con antibióticos en comparación con placebo probablemente logra poca o ninguna diferencia en la reducción del riesgo de DSSI posoperatoria (RR 0,65; IC del 95%: 0,26 a 1,65; evidencia de calidad moderada), tanto en un entorno de bajo riesgo de infección (RR 0,67; IC del 95%: 0,11 a 4,13; evidencia de calidad moderada) como en un entorno de alto riesgo de infección (RR 0,64; IC del 95%: 0,22 a 1,89; evidencia de calidad baja).
Conclusiones de los autores
La evidencia de calidad muy baja muestra que no se conoce si la profilaxis con antibióticos reduce el riesgo de infecciones posoperatorias de la herida después de una cirugía de herniorrafia. La evidencia de calidad moderada muestra que la profilaxis con antibióticos probablemente logra poca o ninguna diferencia en la prevención de infecciones de la herida (es decir, todas las infecciones de la herida, SSSI o DSSI) después de una cirugía de hernioplastia en un entorno de bajo riesgo de infección. En un entorno de riesgo alto, la evidencia de calidad muy baja muestra que no se conoce si la profilaxis con antibióticos reduce las infecciones de la herida y la SSSI después de una cirugía de hernioplastia. La evidencia de baja calidad observó que la profilaxis con antibióticos en un entorno de riesgo alto podría suponer poca o ninguna diferencia en la reducción del riesgo de DSSI.
PICO
Resumen en términos sencillos
¿En comparación con placebo, los antibióticos pueden prevenir las infecciones posoperatorias de la herida después de una cirugía abierta de reparación de una hernia inguinal?
Antecedentes
La hernia inguinal es una debilidad en la pared abdominal en la zona de la ingle, a través de la cual pueden sobresalir los tejidos blandos o los órganos. Las hernias inguinales se producen con frecuencia y, por lo tanto, la reparación de las hernias inguinales es uno de los procedimientos quirúrgicos más frecuentes en todo el mundo. Se considera una técnica quirúrgica «limpia» con tasas bajas de infección de las heridas posoperatorias y, por lo tanto, en general no se recomienda la administración de antibióticos a los pacientes sometidos a una cirugía abierta de reparación de hernia. Hasta la década de 1990 la reparación de hernia con suturas (herniorrafia) era el método de elección. A partir de entonces, la reparación de la hernia con una malla sintética (hernioplastia) fue ganando cada vez más popularidad y se volvió a abrir el debate sobre la necesidad de antibióticos para prevenir las infecciones de las heridas posoperatorias.
Investigación
Se realizaron búsquedas en la literatura (12 de noviembre 2019) de ensayos controlados aleatorizados que compararan antibióticos versus placebo para prevenir las infecciones de la herida después de la cirugía abierta de reparación de la hernia inguinal. Se incluyeron tanto los métodos quirúrgicos con suturas como los de malla. Se dividieron las infecciones en infecciones superficiales y profundas de la herida. Varios estudios revelaron que las tasas de infección fueron mayores al 5% esperado para la cirugía limpia. Por lo tanto, los estudios se dividieron en un grupo con tasas bajas de infección (menos del 5%) y otro con tasas altas de infección (más del 5%).
Características de los estudios y resultados clave
Se identificaron cinco estudios de la cirugía con suturas y 22 estudios de la cirugía con malla. Los estudios de la cirugía con suturas fueron de calidad metodológica muy baja. Los estudios de la cirugía con malla fueron de calidad metodológica baja a moderada.
Esta revisión muestra que los antibióticos no previenen la aparición de ningún tipo de infecciones de la herida después de la reparación de la hernia con suturas. En el caso de la reparación de una hernia con malla en un entorno donde el riesgo de infección es bajo, es probable que los antibióticos logren poca o ninguna diferencia en la prevención de las infecciones de las heridas posoperatorias superficiales. Sin embargo, en un entorno donde el riesgo de infección es alto no se sabe con certeza si los antibióticos reducen el riesgo de que se produzcan infecciones superficiales de las heridas.
En el caso de las infecciones profundas de las heridas, se observó que los antibióticos probablemente logran poca o ninguna diferencia en la reducción del riesgo tanto en entornos en los que el riesgo de infección es bajo como donde es alto.
Calidad de la evidencia
La evidencia de calidad muy baja muestra que no se conoce si los antibióticos reducen el riesgo de infecciones posoperatorias de la herida después de la reparación de la hernia con suturas. La evidencia de calidad moderada muestra que los antibióticos probablemente logran poca o ninguna diferencia en la prevención de las infecciones superficiales o profundas de la herida después de la reparación de una hernia con malla en un entorno de bajo riesgo de infección. La evidencia de calidad (muy) baja muestra que no se sabe si los antibióticos reducen el riesgo de infecciones superficiales de la herida,y que los antibióticos tiene poco o ningún efecto en infecciones profundas de la herida después de la reparación de una hernia con malla en un entorno de alto riesgo de infección.
Authors' conclusions
Summary of findings
| 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 | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with antibiotic prophylaxis | |||||
| All wound infections (SSSI + DSSI) | Study population | RR 0.86 | 1865 | ⊕⊝⊝⊝ | ||
| 49 per 1000 | 42 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ low infection risk environment | Study population | RR 0.63 | 1302 | ⊕⊝⊝⊝ | ||
| 32 per 1000 | 20 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ high infection risk environment | Study population | RR 0.99 | 563 | ⊕⊝⊝⊝ | ||
| 89 per 1000 | 88 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 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 | ||||||
| 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 | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with antibiotic prophylaxis | |||||
| All wound infections (SSSI + DSSI) | Study population | RR 0.61 | 6443 | ⊕⊕⊕⊝ | ||
| 55 per 1000 | 33 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ low infection risk environment | Study population | RR 0.71 | 3100 | ⊕⊕⊕⊝ | ||
| 26 per 1000 | 18 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ high infection risk environment | Study population | RR 0.58 | 3343 | ⊕⊝⊝⊝ | ||
| 85 per 1000 | 49 per 1000 | |||||
| SSSI | Study population | RR 0.60 | 6263 | ⊕⊕⊕⊝ | ||
| 50 per 1000 | 30 per 1000 | |||||
| SSSI ‐ low infection risk environment | Study population | RR 0.71 | 3100 | ⊕⊕⊕⊝ | ||
| 24 per 1000 | 17 per 1000 | |||||
| SSSI ‐ high infection risk environment | Study population | RR 0.56 | 3163 | ⊕⊝⊝⊝ | ||
| 79 per 1000 | 44 per 1000 | |||||
| DSSI | Study population | RR 0.65 | 4185 | ⊕⊕⊕⊝ | ||
| 6 per 1000 | 4 per 1000 | |||||
| DSSI ‐ low infection risk environment | Study population | RR 0.67 | 1488 | ⊕⊕⊕⊝ | ||
| 4 per 1000 | 3 per 1000 | |||||
| DSSI ‐ high infection risk environment | Study population | RR 0.64 | 2697 | ⊕⊕⊝⊝ | ||
| 7 per 1000 | 4 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 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‐20th 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.
-
Cochrane Registry of Studies Online (CRSO.cochrane.org) (Appendix 1)
-
MEDLINE Ovid (1946 to 12 November 2019) (Appendix 2)
-
Embase Ovid (1946 to 12 November 2019) (Appendix 3)
-
Scopus (2011 to 2019) (Appendix 4)
-
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).
-
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.
-
General study information, including first author and year of publication.
-
Study methods, including study design, country and enrolment period of the participants.
-
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).
-
Intervention information, including type of antibiotic (or placebo), dose, number of doses, administration route, and administration timing.
-
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).
-
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.
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 1 and those of the included hernioplasty studies in Table 2. Further details are described in Characteristics of included 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 | |
| AMP | 1 | in fascia | 1 m, 3 m, 6 m, 12 m | not reported | Low (4.0%) | Several surgery types performed | |
| CLR | 1 (3×) | intravenous (1x)/ intramuscular (2x) | 1 m | not reported | Low (4.1%) | Several surgery types performed | |
| CAM | 0.75 | subcutaneous (added to local anaesthesia) | 1 m | not reported | Low (4.6%) | ||
| CON | 1 | intravenous | 1 w, 4 to 6 w | PG: 75 ± 32 / CG: 75 ± 30 | Low (1.9%) | Several surgery types performed | |
| 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) ).
| Antibiotic | Dose (g) | Administration route | Follow‐up period | Operative time (minutes); (mean ± SD or median (range)) | Infection risk environment (Infection percentage in control group) | |
| AMC | 1.2 | intravenous | 1 w/1 m | PG: 45 (20 to 90); CG:45 (20 to 80) | High (5.0%) | |
| 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%) | |
| 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%) | |
| 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%) | |
| CZO | 1 | intravenous | 1 to 6 d, 1 m | PG: 60 (35 to 160); CG: 60 (40 to 135) | High (7.0%) | |
| AMC | 1.2 | intravenous | 1 w | not reported | Low (3.0%) | |
| CZO | 1 | intravenous | 1 w, 2 w, 1 m | not reported | High (10.0%) | |
| 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%) | |
| AMC | 1.2 | intravenous | 1 w, 1 m | not reported | Low (4.7%) | |
| CZO | 1 | intravenous | 1 w, 1 m, 3 m | PG: 66.3 ± 25.4; CG: 65.2 ± 27.1 | High (13.0%) | |
| CZO (or ERY) | 2 (or 1) | intravenous | 1 w, 1 m, 1 y | not reported | Low (2.1%) | |
| AMC | 2 | intravenous | 1 w, 1 m | PG and CG: Mean 40 min | Low (0%) | |
| AMC | 1.2 | intravenous | 1 w, 1 m | PG 38.8 ± 10.8; CG 40.9 ± 11.1 | High (12,5%) | |
| CZO | 1 | intravenous | 1 w, 2 w, 1 m | PG 52.18 ± 16.4; CG 54.07 ± 15.3 | Low (3.9%) | |
| CLT | 1 | intravenous | 12 w | not reported | High (6.4%) | |
| CZO | 1 | intravenous | 1 w, 1 m | PG: 53.54 ± 15.82; CG 52.60 ± 15.28 | High (9.3%) | |
| CZO | 1 | intravenous | 1 w, 1 m | PG 53.54 ± 15.82; CG 52.60 ± 15.28 | High (10.5%) | |
| 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%) | |
| AML | 1.2 | intravenous | 1 w, 1 m | PG: 45 (20 to 90); CG: 45 (20 to 80) | Low (4.8%) | |
| AMC | 1 | not reported | 2 d, 2 w | not reported | High (18.1%) | |
| CZO or LEV | 1 or 0.2 | intravenous | 1 w, 2 w, 3 w, 1 m | not reported | High (5.1%) | |
| 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 1). In the hernioplasty studies (Table 2) 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 2. 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 3 and of the included hernioplasty studies in Table 4. Further details are described in Characteristics of included 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) | |
| 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 | |
| not reported | not reported | not reported | I + F | Not reported | Not reported | |
| PG: 90; CG: 88 | PG: 62 (11 to 90)/ CG: 70 (16 to 92) | Yes | I + F | Not reported | Not reported | |
| 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 | |
| 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
| % 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) | |
| 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) | |
| PG: 96; CG: 97 | PG: 58.3 ± 12.9; CG: 58.2 ± 13.2 | No | I | Not reported | Not reported | |
| PG: 100; CG: 97 | 38.5 ± 17.7 (range: 19 to 90 years) (overall study population) | No | I | Not reported | Not reported | |
| PG: 94; CG: 86 | PG: 58 ± 13; CG: 58 ± 17 | No | I | I; II | PG: 26.1 ± 5/ CG: 26.2 ± 5 | |
| PG: 95; CG: 89 | PG: 48 ± 17; CG: 50 ± 15 | No | I | I; II; III; IV | Not reported | |
| NR | range 11 to 90 (overall study population) | Yes | I | Not reported | Not reported | |
| 99 (overall) | PG: 44.06; CG: 44.84 | No | I | Not reported | Not reported | |
| PG: 100; CG:100 | PG: 41.28 ± 11.49; CG: 40.2 ± 9.84 | No | I | I/ II | Not reported | |
| 96 (overall) | PG: 37.42 ± 9.9; CG: 37.42 ± 11.5 | Yes | I | Not reported | Not reported | |
| 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 | |
| 89 (overall) | 54.2 (17 to 87) (overall study population) | No | I+F | I; II; III | Not reported | |
| PG: 89; CG: 82 | PG: 58 (22 to 91)/ CG: 56.2 (17 to 88) | Yes | I+F | I; II; III | Not reported | |
| PG: 96; CG: 100 | PG: 43.4 ± 19.8/ CG: 44.5 ± 20.5 | No | I | Not reported | Not reported | |
| PG: 98; CG: 98 | PG: 61.37 ± 13.2/ CG: 60.8 ± 14.5 | No | I | I; II | Not reported | |
| PG: 45; CG: 41 | 25 to 84 (overall population) | No | I | not reported | Not reported | |
| PG: 99; CG: 100 | PG: 42.44 ± 15.61 / CG: 45.56 ± 15.43 | Yes | I | I; II | Not reported | |
| PG: 99; CG: 98 | PG: 44.44 ± 15.59 / CG: 45.56 ± 16.43 | Yes | I | I; II | Not reported | |
| PG: 100; CG: 100 | Not reported | No | I | Not reported | Not reported | |
| 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) | |
| PG: 100; CG: 100 | PG: 54.33 ± 11.77 / CG: 52.58 ± 11.80 | No | I | Not reported | Not reported | |
| 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) | |
| 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).
-
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).
-
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.
-
Comparator not eligible (antibiotics prophylaxis versus antibiotics prophylaxis with another regime): Pessaux 2006, Bhuiyan 2017.
-
Article in French (Lazorthes 1993) of similar study described previously in English (Lazorthes 1992)
-
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.
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: Summary of findings 1 Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open inguinal or femoral herniorrhaphy surgery; Summary of findings 2 Antibiotic prophylaxis compared to placebo for prevention of postoperative wound infection in adults undergoing open inguinal or femoral hernioplasty surgery
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.
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%.
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).
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).
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, summary of findings 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 summary of findings 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 (summary of findings 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 (summary of findings 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 (summary of findings 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 (summary of findings 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 (summary of findings 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 (summary of findings 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 1; Table 2). 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 summary of findings 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 summary of findings Table 2). Quality of the evidence for the outcome 'deep wound infections' in hernioplasty surgery was low to moderate (see summary of findings 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.
Study flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Forest plot of comparison antibiotic prophylaxis against placebo for preventing all types of wound infection (SSSI and DSSI) after herniorrhaphy surgery.
Forest plot of comparison antibiotic prophylaxis against placebo for preventing all types of wound infection (SSSI and DSSI) after hernioplasty surgery.
Forest plot of comparison antibiotic prophylaxis against placebo for preventing superficial wound infection (SSSI) after hernioplasty surgery.
Forest plot of comparison antibiotic prophylaxis against placebo for preventing deep wound infections (DSSI) after hernioplasty surgery.
Comparison 1: Wound infections herniorrhaphy, Outcome 1: All wound infections (SSSI+DSSI)
Comparison 2: Wound infections hernioplasty, Outcome 1: All wound infections (SSSI+DSSI)
Comparison 2: Wound infections hernioplasty, Outcome 2: Superficial Surgical Site Infections (SSSI)
Comparison 2: Wound infections hernioplasty, Outcome 3: Deep Surgical Site Infections (DSSI)
| 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 | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with antibiotic prophylaxis | |||||
| All wound infections (SSSI + DSSI) | Study population | RR 0.86 | 1865 | ⊕⊝⊝⊝ | ||
| 49 per 1000 | 42 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ low infection risk environment | Study population | RR 0.63 | 1302 | ⊕⊝⊝⊝ | ||
| 32 per 1000 | 20 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ high infection risk environment | Study population | RR 0.99 | 563 | ⊕⊝⊝⊝ | ||
| 89 per 1000 | 88 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 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 | ||||||
| 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 | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with antibiotic prophylaxis | |||||
| All wound infections (SSSI + DSSI) | Study population | RR 0.61 | 6443 | ⊕⊕⊕⊝ | ||
| 55 per 1000 | 33 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ low infection risk environment | Study population | RR 0.71 | 3100 | ⊕⊕⊕⊝ | ||
| 26 per 1000 | 18 per 1000 | |||||
| All wound infections (SSSI + DSSI) ‐ high infection risk environment | Study population | RR 0.58 | 3343 | ⊕⊝⊝⊝ | ||
| 85 per 1000 | 49 per 1000 | |||||
| SSSI | Study population | RR 0.60 | 6263 | ⊕⊕⊕⊝ | ||
| 50 per 1000 | 30 per 1000 | |||||
| SSSI ‐ low infection risk environment | Study population | RR 0.71 | 3100 | ⊕⊕⊕⊝ | ||
| 24 per 1000 | 17 per 1000 | |||||
| SSSI ‐ high infection risk environment | Study population | RR 0.56 | 3163 | ⊕⊝⊝⊝ | ||
| 79 per 1000 | 44 per 1000 | |||||
| DSSI | Study population | RR 0.65 | 4185 | ⊕⊕⊕⊝ | ||
| 6 per 1000 | 4 per 1000 | |||||
| DSSI ‐ low infection risk environment | Study population | RR 0.67 | 1488 | ⊕⊕⊕⊝ | ||
| 4 per 1000 | 3 per 1000 | |||||
| DSSI ‐ high infection risk environment | Study population | RR 0.64 | 2697 | ⊕⊕⊝⊝ | ||
| 7 per 1000 | 4 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 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. | ||||||
| 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 | |
| AMP | 1 | in fascia | 1 m, 3 m, 6 m, 12 m | not reported | Low (4.0%) | Several surgery types performed | |
| CLR | 1 (3×) | intravenous (1x)/ intramuscular (2x) | 1 m | not reported | Low (4.1%) | Several surgery types performed | |
| CAM | 0.75 | subcutaneous (added to local anaesthesia) | 1 m | not reported | Low (4.6%) | ||
| CON | 1 | intravenous | 1 w, 4 to 6 w | PG: 75 ± 32 / CG: 75 ± 30 | Low (1.9%) | Several surgery types performed | |
| 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) ). | |||||||
| Antibiotic | Dose (g) | Administration route | Follow‐up period | Operative time (minutes); (mean ± SD or median (range)) | Infection risk environment (Infection percentage in control group) | |
| AMC | 1.2 | intravenous | 1 w/1 m | PG: 45 (20 to 90); CG:45 (20 to 80) | High (5.0%) | |
| 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%) | |
| 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%) | |
| 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%) | |
| CZO | 1 | intravenous | 1 to 6 d, 1 m | PG: 60 (35 to 160); CG: 60 (40 to 135) | High (7.0%) | |
| AMC | 1.2 | intravenous | 1 w | not reported | Low (3.0%) | |
| CZO | 1 | intravenous | 1 w, 2 w, 1 m | not reported | High (10.0%) | |
| 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%) | |
| AMC | 1.2 | intravenous | 1 w, 1 m | not reported | Low (4.7%) | |
| CZO | 1 | intravenous | 1 w, 1 m, 3 m | PG: 66.3 ± 25.4; CG: 65.2 ± 27.1 | High (13.0%) | |
| CZO (or ERY) | 2 (or 1) | intravenous | 1 w, 1 m, 1 y | not reported | Low (2.1%) | |
| AMC | 2 | intravenous | 1 w, 1 m | PG and CG: Mean 40 min | Low (0%) | |
| AMC | 1.2 | intravenous | 1 w, 1 m | PG 38.8 ± 10.8; CG 40.9 ± 11.1 | High (12,5%) | |
| CZO | 1 | intravenous | 1 w, 2 w, 1 m | PG 52.18 ± 16.4; CG 54.07 ± 15.3 | Low (3.9%) | |
| CLT | 1 | intravenous | 12 w | not reported | High (6.4%) | |
| CZO | 1 | intravenous | 1 w, 1 m | PG: 53.54 ± 15.82; CG 52.60 ± 15.28 | High (9.3%) | |
| CZO | 1 | intravenous | 1 w, 1 m | PG 53.54 ± 15.82; CG 52.60 ± 15.28 | High (10.5%) | |
| 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%) | |
| AML | 1.2 | intravenous | 1 w, 1 m | PG: 45 (20 to 90); CG: 45 (20 to 80) | Low (4.8%) | |
| AMC | 1 | not reported | 2 d, 2 w | not reported | High (18.1%) | |
| CZO or LEV | 1 or 0.2 | intravenous | 1 w, 2 w, 3 w, 1 m | not reported | High (5.1%) | |
| 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) ). | ||||||
| % 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) | |
| 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 | |
| not reported | not reported | not reported | I + F | Not reported | Not reported | |
| PG: 90; CG: 88 | PG: 62 (11 to 90)/ CG: 70 (16 to 92) | Yes | I + F | Not reported | Not reported | |
| 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 | |
| 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 | ||||||
| % 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) | |
| 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) | |
| PG: 96; CG: 97 | PG: 58.3 ± 12.9; CG: 58.2 ± 13.2 | No | I | Not reported | Not reported | |
| PG: 100; CG: 97 | 38.5 ± 17.7 (range: 19 to 90 years) (overall study population) | No | I | Not reported | Not reported | |
| PG: 94; CG: 86 | PG: 58 ± 13; CG: 58 ± 17 | No | I | I; II | PG: 26.1 ± 5/ CG: 26.2 ± 5 | |
| PG: 95; CG: 89 | PG: 48 ± 17; CG: 50 ± 15 | No | I | I; II; III; IV | Not reported | |
| NR | range 11 to 90 (overall study population) | Yes | I | Not reported | Not reported | |
| 99 (overall) | PG: 44.06; CG: 44.84 | No | I | Not reported | Not reported | |
| PG: 100; CG:100 | PG: 41.28 ± 11.49; CG: 40.2 ± 9.84 | No | I | I/ II | Not reported | |
| 96 (overall) | PG: 37.42 ± 9.9; CG: 37.42 ± 11.5 | Yes | I | Not reported | Not reported | |
| 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 | |
| 89 (overall) | 54.2 (17 to 87) (overall study population) | No | I+F | I; II; III | Not reported | |
| PG: 89; CG: 82 | PG: 58 (22 to 91)/ CG: 56.2 (17 to 88) | Yes | I+F | I; II; III | Not reported | |
| PG: 96; CG: 100 | PG: 43.4 ± 19.8/ CG: 44.5 ± 20.5 | No | I | Not reported | Not reported | |
| PG: 98; CG: 98 | PG: 61.37 ± 13.2/ CG: 60.8 ± 14.5 | No | I | I; II | Not reported | |
| PG: 45; CG: 41 | 25 to 84 (overall population) | No | I | not reported | Not reported | |
| PG: 99; CG: 100 | PG: 42.44 ± 15.61 / CG: 45.56 ± 15.43 | Yes | I | I; II | Not reported | |
| PG: 99; CG: 98 | PG: 44.44 ± 15.59 / CG: 45.56 ± 16.43 | Yes | I | I; II | Not reported | |
| PG: 100; CG: 100 | Not reported | No | I | Not reported | Not reported | |
| 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) | |
| PG: 100; CG: 100 | PG: 54.33 ± 11.77 / CG: 52.58 ± 11.80 | No | I | Not reported | Not reported | |
| 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) | |
| 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 | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 All wound infections (SSSI+DSSI) Show forest plot | 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] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 All wound infections (SSSI+DSSI) Show forest plot | 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) Show forest plot | 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) Show forest plot | 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] |
