Meta-Analysis of Acute Systemic Antimicrobial Prophylaxis (ASAP) for Open Extremity Fracture Management: Subgroup Analysis and Methodological Guidelines

1. Introduction

Infection rates after open extremity fracture range from 1.8% in Gustilo-Anderson Type I to 40.5% in Type IIIB.[1,2,3,4] Surgical site Infections (SSI) require multifaceted management with the potential for complications that result in a significant clinical burden to patients and high cost to the healthcare system.[5] SSI alone may increase healthcare costs by 1.2 to 6-fold.[6]

Open fracture SSI rates have remained high and relatively unchanged for decades despite the widespread use of accelerated prophylactic antibiosis.[7] Acute systemic antimicrobial prophylaxis (ASAP) regimens have evolved and frequently change depending on geography, hospital system standards, or surgeon practice. There is also concern that traditional and even current antibiotic regimens are not accounting for the modern open fracture bioburden.[8]

Antibiotic stewardship in open fractures has primarily been guided by Gustilo and Anderson’s studies[3,9,10,11,12,13] conducted in 1976, in which they showed that prophylactic antibiotics such as penicillin, streptomycin, chloramphenicol, oxacillin, and kanamycin reduced deep infections from 10% to 5%.[12,13] In 1990, Gustilo et al. recommended two grams of cefazolin or cefamandole upon admission and one gram three times daily for up to 72 hours for Type I injuries. Further recommendations for Gustilo-Anderson (GA) Type II and III open fractures included the addition of 1.5 mg/kg of tobramycin upon admission and 3-5 mg/kg daily for 72 hours.[10]

The Surgical Infection Society (SIS) and the Eastern Association for the Surgery of Trauma (EAST) have also developed literature-based guidelines for ASAP regimens. In 2006, SIS recommended first-generation cephalosporin for 24-48 hours perioperatively for Type I open fractures and 48 hours for Type II and III open fractures. Additionally, the SIS guidelines reported a lack of evidence for gram-negative prophylaxis. A few years later, the EAST guidelines recommended a first-generation cephalosporin for all GA types for no longer than 24 hours after soft tissue coverage for type I and II open fractures and 72 hours for type III open fractures. Additionally, when indicated, an aminoglycoside for gram-negative coverage for GA type III fractures and high-dose penicillin for Clostridium spp. should be added. However, EAST guidelines discouraged using Fluoroquinolones for their non-superiority compared to the cephalosporin/aminoglycoside regimen.[14,15]

Despite these guidelines, there continues to be a lack of standardized ASAP regimens in the open fracture setting. Our meta-analysis aims to delineate which regimen is most effective in preventing SSI and to propose a minimum reporting standard for future studies to minimize evidence heterogeneity.

2. Materials and Methods:

2.1. Protocol

The study protocol was designed and conducted according to the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) checklist and the Cochrane Handbook for Systematic Reviews of Interventions guidelines.[16,17]

2.2. Search Methodology

A detailed, comprehensive literature search for all relevant studies was performed using several online databases on October 21, 2022: PubMed, MEDLINE via Ovid, Web of Science, Cochrane database, and Science-Direct. The search strategy combined subject headings and free text words such as “open fracture” and “antibiotics” in Ovid MEDLINE, topic searching in Web of Science, and free text words in the other databases. The references of relevant studies (reviews and meta-analyses) were manually examined for any that met our inclusion criteria to ensure the inclusion of all potentially eligible studies.

2.3. Study Selection and Eligibility Criteria

Two independent reviewers scoped the studies’ titles, abstracts, and full-text articles that were in English for their eligibility PICO criteria:

2.4. Population, Intervention, Comparison, and Outcome (PICO) Criteria

• Population: An eligible study should have at least five subjects with Gustilo-Anderson (GA) Type I, II, or III open extremity fractures.
• Intervention: Any of the following acute systemic antimicrobial prophylaxis (ASAP) protocols.
  • Cefazolin alone (Cef)
  • Cefazolin + Gentamicin (Cef+Gent)
  • Tobramycin + Cefazolin (Tobra+Cef)
  • Piperacillin + Tazobactam (Pip+Tazo)
  • Ampicillin + Sulbactam (Amp+Sulb)
  • Cefazolin + Ciprofloxacin (Cef+Cipro)
  • Flucloxacillin + Benzylpenicillin (Flu+Benz)
  • Two doses of Cefazolin + Gentamicin (2Cef+Gent)
  • Cefamandole + Gentamicin (Cefam+Gent)
  • Ciprofloxacin (Cipro)
  • Clindamycin (Clinda)
  • Cloxacillin (Clox)
  • Flucloxacillin (Flu)
  • Cephradine (Ceph)
• Comparison: Each intervention was compared to each other, and the following groups were specified:
  • Single dose of Cefazolin plus Gentamicin (Cef+Gent) versus:

    ➢

    Cefazolin alone (Cef)

    ➢

    Piperacillin plus Tazobactam (Pip+Tazo)

    ➢

    Ampicillin plus Sulbactam (Amp+Sulb)

    ➢

    Cefazolin plus Ciprofloxacin (Cef+Cipro)

    ➢

    Flucloxacillin plus Benzylpenicillin (Flu+Benz)

    ➢

    Two doses Cefazolin plus Gentamicin (2Cef+Gent)

  • Cefazolin alone versus Piperacillin plus Tazobactam (Cef vs Pip+Tazo)
  • Cefazolin plus Clindamycin versus Piperacillin plus Tazobactam (Cef/Clinda vs Pip+Tazo)
  • Clindamycin versus Cloxacillin (Clinda vs Clox)
  • Penicillin versus Cephradine (Pen vs Ceph)
  • Cefamandole plus Gentamicin versus Ciprofloxacin (Cefa+Gent vs Cipro)
• Outcome: Outcome analyses of the included studies must report the post-fixation incidence of surgical site infection (SSI) and acute kidney injury (AKI). The length of hospital stay was considered as our secondary outcome.

2.5. Inclusion Criteria

Inclusion criteria included randomized controlled trials (RCTs), non-randomized prospective, and retrospective observational cohort studies published in English comparing different ASAP protocols following an open fracture.

2.6. Exclusion Criteria

Exclusion criteria included language other than English, article unavailable, or duplicate data/study. Some articles were excluded by reviewing the inclusion criteria in the title or abstract. All other studies required full-text review to determine relevance. Disagreements were resolved by direct communication between the reviewers.

2.7. Data Extraction

The reviewers extracted the required information from all eligible studies included. A data collection spreadsheet was established to analyze quantitative and qualitative data. The data was extracted from clinical studies using the following data variables: 1) demographics and characteristics (author, study design, year of publication, country of study, patient number, age, and female sex), 2) Gustilo-Anderson fracture type, and 3) antibiotic regimens used.

2.8. Quality Assessment of Included Studies (MINORS)

The methodological index for nonrandomized studies (MINORS) was used to assess the quality of the included studies. Full-text versions of included nonrandomized studies were reviewed for methodological quality analysis using the MINORS score. Twelve items were scored as “0” (not reported), “1” (reported but inadequate), or “2” (reported and adequate). Two reviewers independently assessed the quality of the included studies.[18]

2.9. Risk of Bias Assessment of Randomized Controlled Trials

Cochrane Collaboration’s assessment tool for risk of bias from the Cochrane Handbook for systematic reviews was used to assess included RCTs for methodological bias.[17] The following items were considered as ‘low risk,’ ‘high risk,’ or ‘unclear risk’ of bias: (1) random sequence generation (selection bias), (2) allocation concealment (selection bias), (3) blinding of participants and personnel (performance bias), (4) blinding of outcome assessment (detection bias), (5) incomplete outcome data (attrition bias), (6) selective reporting addressed (reporting bias).

2.10. Synthesis of Results

GraphPad Prism (version 9.4.1) was used to calculate the descriptive analysis and perform a linear regression analysis between the publication year and the MINORS score.[19] In cases where the standard deviation was not reported, we calculated the mean, sample size, and range as described in previous literature.[20,21] Continuous variables were presented as mean ± standard deviation (SD), while categorical variables were reported as percentages. Using standard meta-analytic techniques, the authors calculated overall effect size estimates and forest plots based on all studies that provided sufficient data. To estimate the heterogeneity across studies, the I² values were calculated. If the homogeneity of studies was identified with an I² <50%, we used the fixed effects model to estimate the overall effects using the RevMan 5.3 software.[22]

3. Results

3.1. Literature Search and Study Selection

Upon the conclusion of our search strategy, there were 5,300 articles. After duplicates were removed, 144 records of full-text articles were screened for eligibility to the inclusion criteria. After screening, 4,189 studies were excluded for not meeting our PICO criteria. After eligibility screening, 17 clinical studies met the inclusion criteria for a qualitative descriptive and quantitative meta-analysis (Supplemental Figure S1).[23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39]

3.2. Characteristics of Included Studies

From 1988 to 2022, 17 studies (4 RCTs, 3 prospective studies, and 10 retrospective studies) were analyzed. These studies included 2,031 patients (70% males), with an average age of 40.0 ± 5.9 years old, ranging from 30-50.7 years old. The average follow-up duration was 9.12 ± 6.40 months, (1.5 – 24 months.)[28,38] 14 unique antibiotic regimens were analyzed in the 17 studies. The average infection rate across all studies was 18.0% (SD=18.0%, range [0-70%]). The feature diagnosis of the CDC definition of surgical site infection (SSI) is found in Supplemental Table S1. The number of patients and the associated dose and duration of antibiotic regimens are outlined in Table 1. There were 482 patients included in studies that did not stratify open fractures and antibiotic regimens by type (Table 2).[23,33,39] 88% of studies reported the duration of antibiotic type, ranging from 1 to 12.7 days.[23,24,25,26,28,29,30,31,32,34,35,36,37,38,39] 88% of the studies reported the fracture location either partially or completely (Table S3). [23,24,25,26,27,28,29,30,31,32,33,36,37,38,39] 35% of studies reported the fixation method in total.[24,30,32,33,36,38] 47% of the studies did not report their fixation method completely (Supplemental Table S2).[23,25,26,27,29,31,37,39] 75% of the studies were performed at a Level 1 Trauma Center; 33% and 8% were at academic and pediatric Level 1 trauma centers, respectively.[24,26,27,28,30,31,32,33,34,35,36,39] 29% of the studies did not report the study center.[23,25,29,37,38]

3.3. Geographical Distribution of the Included Studies

The 17 studies included in this meta-analysis were conducted in 7 countries. 65% of the studies were performed in the United States,[25,26,27,28,30,31,32,33,34,35,36] 11% were performed in Japan,[37,38], and 24% were completed in Finland,[39] the United Kingdom,[23] Germany,[24] and Iran,[29] respectively (Table 2, Supplemental Figure S2A).

3.4. Risk of Bias Assessment of Randomized Control Trials

All four RCTs were analyzed [23,33,36,39], each providing a brief or detailed discussion of the random sequence generation methodology. Sorger et al. did not report allocation concealment.[36] Three RCTs reported allocation concealment in detail.[23,33,39] Two RCTs could not blind the participants or personnel.[23,36] Three could not blind the outcome assessment.[23,36,39] All four RCTs had incomplete outcome data and demonstrated high bias in selective reporting (Supplemental Figure S2B&D).

3.5. Methodological Index for Nonrandomized Studies (MINORS)

There was no significant correlation between publication year and MINORS total score (r²=0.2020, p=0.1233, Supplemental Figure S2C). Any deduction in the MINORS score was given due to methodological limitations, such as using mostly retrospective comparative studies (Supplemental Figure S2C). The average MINORS score was 17.23 ± 2.89 out of 24, ranging from 10 to 20 (Supplemental Table S1). The following parameters were most likely to receive a low score: “a prospective collection of data,” “unbiased assessment of the study endpoint,” “follow-up period appropriate to the aim of the study,” and “prospective calculation of the study size.”

3.6. Synthesis of Results

1. 

Surgical Site Infection (SSI)

➢

Type I Gustilo-Anderson Open Fractures

The pooled results of 3 studies (169 patients) utilizing 3 ASAP protocols (Clinda vs. Cef, Clinda vs. Cloxa, and Cipro vs. Cefam+Gent) showed no significant difference between groups with regard to SSI (OR = 2.24, p=0.52, Figure 1).[28,33,39]

➢

Type II Gustilo-Anderson Open Fractures

The pooled results of 4 studies (429 patients) utilizing 6 ASAP protocols (Cef + Gent vs. Cef, Cef + Gent vs. Pip+Tazo, Pip+Tazo vs. Cef alone OR Pip+Tazo vs. Cef+Clinda, 2Cef+Gent vs. 1Cef+Gent, and Clinda vs. Clox) demonstrated no significant differences between other groups. (OR=1.19, p=0.52, Figure 2).[26,30,36,39]

➢

Type III Gustilo-Anderson Open Fractures

Our pooled results of 11 studies (1,232 patients) using 12 ASAP protocols demonstrated no significant difference in the rate of SSI between the following regimens including: (Cef+Gent vs. Cef, Cef+Gent vs. Pip+Tazo, Cef+Gent vs. Amp+Sulb, Cef+Gent vs. Cef+Cipro, Cef+Gent vs. Flu+Benz, and 2Cef+Gent vs. Cef+Gent) (OR=1.07, p=0.68). However, Flu+Benz had a lower SSI rate than Cef+Gent (OR=2.49, p=0.05, Figure 3).[24,25,26,27,29,31,32,34,36,37,38]

➢

Open fracture, regardless of the type.

The pooled data from three studies (426 patients) investigating infection rate after open fracture without stratification by type demonstrated a lower infection rate with the use of Cloxacillin compared to Clindamycin (OR=0.43, p=0.03, Figure 4).[23,33,39]

2. 

Acute Kidney Injury (AKI)

➢

Type III Gustilo-Anderson Open Fractures

The results based on a pooled sample of four studies and five comparisons (471 patients) showed no significant difference in AKI with the use of Cef+Gent vs. Cef, Cef+Gent vs. Pip+Tazo, or Cef+Gent vs. Amp+Sulb in Type III open fractures (OR=0.74, p=0.43, Figure 5A).[24,26,32,36]

➢

Type II & III Gustilo-Anderson Open Fractures

The pooled results of two studies and four comparisons (367 patients) demonstrated no significant difference in the rate of AKI in GA Type II or III fractures with the use of Cef+Gent vs. Pip+Tazo, Cef+Gent vs. Cef, Cef vs Pip+Tazo, or Cef+Gent vs. 2Cef+Gent (OR=1.14, p=0.66, Figure 5B).[26,36]

3. 

Length of Hospital Stay (LOHS)

➢

Type II Gustilo-Anderson Open Fractures

Our pooled results of two studies and three comparisons (230 patients) demonstrated a four-day longer LOHS with the use of Cef+Gent compared to Cef monotherapy (OR=4.00, p<0.00001, Figure 6A).[26,30]

➢

Type III Gustilo-Anderson Open Fractures

When comparing different antibiotics and the length of hospital stay, our results from five studies and six comparisons (576 patients) showed no significant difference in LOHS comparing Cef+Gent to Pip+Tazo (OR=-2.39, p=0.25).[24,25,26,31,35] However, on subgroup analysis, there was around a four-day increased hospital stay using Cef+Gent compared to Cefazolin alone (OR=3.78, p<0.00001), Figure 6B.

4. Discussion

The risks of open fractures are well-known, with infection causing significant morbidity to patients and increasing health care costs. Despite efforts to reduce infection rates, SSI remains high in this patient population. To our knowledge, this is the largest meta-analysis on this topic to date, including 17 clinical studies with 2,031 patients comparing prophylactic antibiotic regimens in open fractures. This meta-analysis has identified poor and inconsistent reporting across previous studies, which contributes to the lack of standardized initial open fracture management protocols across practices and hospital systems.

Our meta-analysis demonstrated that Cef+Gent, Cef, Pip/Tazo, and Amp/Sulb antibiotics did not affect AKI in open fracture patients without pre-existing renal problems in GA Type II or III fractures. Cef+Gent was associated with an increased length of hospital stay in Type II and III fractures when compared to Cef alone. This supports the use of Cefazolin alone for Type II open fractures as infection rates were not significantly different with the addition of Gentamicin. The original GA paper’s conclusions on infection rates with Cef+Gent in Type III fractures were supported. However, a stratified meta-analysis of articles describing GA Type III fractures demonstrated that Flu+Benz had a lower infection rate than Cef+Gent. Overall, no recommendations can be made regarding the most effective antibiotic regimen for open fractures, given the inconsistencies between studies and overall poor-quality evidence within studies. Treatment should be tailored to the patient’s needs, as they vary based on severity of injury, contamination of wound, bacterial resistance, and regional/seasonal factors.[40,41,42]

Given the continued high infection rate in open fractures despite intravenous antibiosis, the orthopedic trauma community is increasingly using local intrawound antibiotic therapies. Soft tissue trauma around the open fracture site can limit the effectiveness of prophylactic systemic antibiotics, leading to the investigation of local antibiotics as an adjunct.[41] Local antibiotic powder is safe and affordable for open fractures, but its efficacy and impact on antibiotic resistance require further study.[42] O’Toole et al. found that applying intrawound vancomycin powder before closure reduced gram-positive surgical site infections in high-risk tibial plateau and pilon fractures. Yet, there are no large randomized controlled trials demonstrating the efficacy of powered in open fractures, but there have been studies supporting its use. [43] Future research should explore these methods and collaborate with systemic antibiotic use to establish new guidelines. [39,44]

This meta-analysis was constrained by the available evidence of the included studies. The heterogeneous protocols for the dosages and durations of described antibiotic regimens are evident in Table 2. Pooled subgroup analysis of GA Type I fractures failed to show significant differences in infection rates despite Vasenius et al. reporting a lower infection rate with Clindamycin compared to Cloxacillin and Iobst et al. reporting no infections in pediatric GA Type I fractures.[28,39] Insufficient data prevented analysis of AKI rates in GA Type I or II fractures alone or when combined with GA Type III fractures. Comparison of LOHS between GA Type I, II, and III fractures was also limited by available data. Limited data on BMI, comorbidities, number of irrigation and debridement procedures, mortality, and readmission also precluded subgroup analysis of this data. Recent studies lacked stratification by GA type and antibiotic regimen.[44]

4.1. Recommendations and Guidelines for Future Trials

After a review of the current literature, standardized methods are needed to compare antibiotic regimens in open fracture management. This review provides key information for future trials analyzing outcomes (Figure 7). Future trials should report categorical variables with mean, range, and standard deviation.

4.2. Study Design

When reporting data for randomized controlled trials and observational studies, the CONSORT and STROBE checklist items must be implemented and adhered to for increasing the quality of articles on this topic and maintain validity and reproducibility.

4.3. Patient’s Characteristics

Patient matching of demographics is essential for reporting differences. Comorbidities like substance abuse, diabetes, hypertension, and immunosuppression increase infection risk. Osteoporotic and immunosuppressive drugs can impair bone healing. Previous orthopedic surgery may increase wound healing issues. We recommend stratifying groups by antibiotic regimen, fracture type, age, sex, BMI, income, comorbidities, previous fractures/surgery, substance abuse, and immunosuppressive medication use.[45,46,47,48,49]

4.4. Fracture Characteristics

A thorough history and identification of the injury mechanism determines treatment aggressiveness. Fracture characteristics such as location and AO/OTA classification should be reported for reproducibility. Stratifying patients by antibiotic and GA type is essential. The GA type is an independent predictive risk factor for infection.[33,47] Fracture bioburden varies by location and season, as shown in a study by Sagi et al. Reporting the geographic region and month of injury is recommended.[50]

4.5. Treatment Characteristics

4.5.1. Antibiotics

Details of antibiotics used, including type, frequency, duration, and time of treatment initiation is recommended. The timing of antibiotics within three hours of injury and the longer duration of antibiotics have been debated regarding their effect on reducing SSI.[33,51] While a recent meta-analysis found no benefit in using antibiotics after 72 hours, a multicenter retrospective cohort study suggested that prolonged antibiotic duration may benefit high-contamination open fractures.[51,52,53,54] The use of intrawound therapy (antibiotic beads, powder) should also be reported, as this local strategy of additional prophylaxis is being commonly adopted in the trauma community.

4.5.2. Irrigation & Debridement

Conflicting evidence exists regarding the time to debridement for open fractures.[55,56] Type III open fractures with delayed time to debridement greater than 8 hours have an increased risk of infection.[55] Irrigation and debridement procedures, the type of irrigation used, the amount of solution used, and the method have been shown to affect infection rates. Alternative irrigation solutions may damage tissue.[57] Reporting the time to debridement and the type of irrigation used is recommended for future trials.[28,55,56,57]

4.5.3. Would Closure

Wound closure for open fractures varies by GA type and closure method, as shown by one of our analyzed studies.[25] Delayed wound closure, external fixation, free flap, and bone graft use should be reported and stratified by fracture type. Negative pressure wound therapy, which reduced infection rates, should also be noted, along with the timing of wound closure.[58]

4.5.4. Post-operative Assessment

Aminoglycosides are commonly used in contaminated fractures but have potential consequences such as AKI. Conflicting studies have reported on this effect, so we recommend reporting it as well as any pre-existing renal impairment.[13,27,29,59] Post-operative antibiotics should be reported to accurately determine infection rates and treatment reproducibility, with a minimum follow-up of one year to evaluate nonunion and its definition.[60] MRI should be performed and reported for suspected infection and septic nonunion and to rule out osteomyelitis.

5. Conclusion

This meta-analysis is unable to definitively recommend one prophylactic antibiotic regimen over another with confidence to reduce open fracture-related infection. We offer methodological guidelines for future trials that will allow robust evidence for future studies describing antibiotic prophylaxis for open fracture management.