The effect of early broad-spectrum versus delayed narrow-spectrum antibiotic therapy on the primary cure rate of acute infection after osteosynthesis

Fracture-related infection (FRI) is a serious problem after operative fracture management. It is accompanied by increased morbidity, prolonged hospital stays, an increased number of operations and thereby a significant increase in healthcare costs [1]. FRI has an incidence of approximately 1–2% in closed fractures and increases to up to 50% in severe open fractures [2‐4].

Treatment regimens for prosthetic joint infections (PJI) are becoming increasingly well defined, but PJI differs significantly from FRI in terms of the soft tissue problems and the need for fracture stability that accompany it, and evidence-based guidelines for FRI are currently insufficient [5]. Unfortunately, infection near orthopaedic implants seems to be highly complex and poorly described. This has resulted in a diverse collection of treatment protocols, with recommendations only for specific circumstances. Although aggressive surgical debridement seems to be embedded in most approaches, a consensus on standardised antibiotics or the retention of orthopaedic implants has yet to be reached [3, 6‐8].

In a previous study, the effect of an aggressive standardised approach to FRI, which consisted of immediate surgical debridement with hardware retention and combination antibiotics with rifampicin, produced promising results. High overall success rates were achieved, but the implications of empirical antibiotics in the context of emerging resistance remain the subject of debate [9].

To contribute to the knowledge gap on this subject, we compared two standardised regimens that shared the philosophy of a uniform approach to all acute FRI, including all anatomic sites and fixation types. Both consisted of immediate aggressive debridement for source control, retention of orthopaedic implants and early soft tissue coverage, but differed in their antibiotic treatment guidelines. The aim was to analyse the effect of the regimens in terms of success rates, microbiological aspects and emerging resistance.

The aim of this study was to compare the effect of two standardised treatment regimens for (sub)acute FRI that were similar in their philosophy for aggressive surgical debridement and implant retention, but differed in their use of antibiotics. The immediate administration of an antibiotic directed against the most common pathogens in combination with rifampicin seemed to contribute to a higher rate of success. In addition, an early change to oral antibiotics did not lead to worse outcomes in terms of success or antibiotic resistance in protocol A. However, as these elements are totally integrated in the protocols, their individual effects cannot be distinguished based on this data. Despite better logistics and earlier soft tissue coverage in protocol B, the primary success rate was significantly higher in protocol A (A = 72% and B = 47%), which could be attributed to the treatment protocol as a whole and the liberal use of rifampicin (adj-OR 3.45, CI 1.13–10.52). As the antibiotic strategy was the most prominent difference between the two protocols in the logistic regression analysis, this supports the argument for the early use of empirical antibiotics in the treatment of acute FRI.

As evidenced in our study, the majority of bone and joint infections are caused by gram-positive bacteria, with S. aureus as the most common [3, 9]. However, in the case of the exogenous infection of a fracture site, for example in open fractures or perioperative contamination, CoNS are more common. This is thought to be due to contamination by skin flora and the increased pathogenicity of low-virulent microorganisms caused by the presence of the implant material [13, 14] and/or the decreased microperfusion of injured tissue. In our study, a relatively high number of patients with polymicrobial cultures were observed [3]. This could be due not only to the relatively high number of complicated fractures, but also to the contamination of superficial wound swabs. Therefore, for future microbiological identification, only deep tissue samples are recommended [15]. Nonetheless, we studied two protocols that represented clinical practice, and as the scope of this study was the effect of the treatment rather than the microbiological identification, we chose not to exclude swab samples from this study. Biofilm-forming bacteria are a common cause of FRI and require special attention in terms of a treatment approach [8]. For example, planktonic S. aureus can begin to form a biofilm within 24 h. In a biofilm, bacteria form an extracellular matrix—which hinders antibiotic penetration—and enter a stationary growth phase, resulting in decreased antibiotic susceptibility [16]. We attempted to exclude infections with mature biofilms by limiting the study to acute infections (< 3 weeks). However, it is possible that, as more delayed infections were included in protocol B, dormant infections led to the unnoticed maturation of biofilm before symptoms occurred. If this is the case, it could have contributed to the lower primary success rates in protocol B. It has been recognised that the successful treatment of mature biofilm infections consists not only of antimicrobial, but also of a surgical approach to reduce bacterial load, for which implant replacement is sometimes necessary [13]. Biofilms in their early stages (< 10 days) are significantly less resilient to antibiotic eradication than older biofilms [17]. The appropriate antibiotic treatment has been demonstrated to be able to eradicate young biofilms entirely without the need for mechanical reduction of the biofilm [18‐20]. The use of local antibiotics permits immediate availability and may thereby contribute to the eradication or destabilisation of young biofilms [21]. Thus, with a vigorous antibiotic approach directly following initial debridement, one might maintain control of the local situation. As the majority of implant-related infections are caused by Staphylococci, empirical therapy must possess antimicrobial properties for at least gram-positive bacteria.

Rifampicin is an example of an antibiotic with favourable pharmacokinetic properties for biofilm infections because of its high bioavailability and capacity to affect bacteria in the stationary phase [22]. A disadvantage of rifampicin use is the quick emergence of (cross-)resistance, mainly when it is administered as monotherapy or in cases with high bacterial load [23, 24]. In this study, we found one new case of rifampicin resistance. This low rate of emergence of new resistance is in line with the literature, as rifampicin was not administered as monotherapy in either of the protocols, and thorough surgical debridement minimalised bacterial load prior to administration. Although we believe that the addition of rifampicin to the antibiotic therapy makes a valuable contribution, perhaps stricter indications should be maintained.

Amoxicillin/clavulanic acid resistance was found most frequently in the patients who had been treated according to protocol B, in line with the most frequent administration of this drug in the corresponding protocol. Resistance emerged twice in Escherichia coli. With high rates of the prescription of antibiotics in primary care facilities, the resistance of bacteria is a growing problem [25]. Although the Netherlands has relatively low bacterial resistance, broad-spectrum penicillin, like amoxicillin/clavulanic acid, is prescribed most often and correlates with higher resistance rates [25]. Studies have demonstrated that amoxicillin/clavulanic acid resistance in E. coli has risen to 14% in the past decades [26]. In addition, resistance has been observed in commensal skin flora, such as wild-type S. aureus and epidermidis, which have 5% and 14% resistance against rifampicin, respectively [27]. Taking this into account, we are unable to claim with certainty whether our findings demonstrate the emergence of new resistance or merely reflect the selection of already present resistant mutants. Furthermore, in 40% of the cases, there was no information on follow-up cultures (20% missing, 20% negative cultures), and thus definitive statements on the emergence of resistance could not be made.

Both protocols were intended to achieve early surgical debridement and wound closure at the initial fracture fixation. Several studies have described the benefits of primary closure in fracture treatment. Both Scharfenberger et al. [28] and Jenkinson et al. [29] showed that there are lower infection and non-union rates in primary closures after open fractures than after delayed closures. In our study, we did not find any advantages of primary closure. The high numbers of primary closures, as intended in both protocols, may have hindered statistical significance in this cohort.

When primary closure was feasible, a drain was left in protocol B. This has been considered in the literature to have both advantages (fluid and hematoma drainage) and disadvantages (a conduit to the external environment) [30]. Parker et al. [31] showed in a meta-analysis that there is no benefit to closed suction drainage after orthopaedic surgery in terms of infection, hematoma, dehiscence or re-operations, while they demonstrated that there is a higher requirement for blood transfusion when drains are applied [31]. When primary closure could not be achieved, direct flap coverage was preferred in protocol B and vacuum dressing in protocol A, with secondary wound healing or delayed flap coverage in extensive soft tissue damage. Direct flap coverage of severe soft tissue injuries that accompany fractures can be executed safely, whereas delayed (> 72 h) free flap transplantation may be associated with more wound complications [32, 33]. Conversely, others have advocated that staged soft tissue management minimises swelling and thereby reduces post-reconstructive wound complications [34]. Vacuum dressing may contribute to reduction in the need for flap transfer, size and reoperation, but uncertainty exists as to whether the beneficial effects extend beyond the first week after initiation [35]. In this cohort, we could not confirm nor counter these findings.

There are some limitations to this study. Since this was a retrospective cohort, although patients’ baseline characteristics were similar, fracture treatment characteristics were not standardised. This resulted in differences in, e.g., free flap transplantations (timing/frequency) and time until definitive fixation that could not be analysed separately because of their small numbers and integration in the rest of the protocol. This also applies to the analysis of the independence of separate elements of antibiotic treatment. The major differences between the two protocols were the timing of the start and the choice of antibiotics. Both, protocol A, as a whole, and the rifampicin treatment, as a separate element, were predicted to have higher success rates. However, due to the limited number of failures in this study and the incorporation of the rifampicin in protocol A, the aforementioned elements could not be analysed separately. Second, not all patients had routine follow-up cultures. Although all clinically relevant, redebridement or protracted infection follow-up tissue samples were obtained, additional samples from all treated patients could contribute to further insight into resistance patterns. Last, the follow-up time was significantly longer in protocol B. Although this is mostly explained by the prolonged inclusion period and longer time-until-success, it may lead to higher relapse rates and thereby result in lower success rates for protocol B. In addition to methodological limitations, since the execution of this study, a consensus on the definition of FRI has been reached and published [15]. Although this study did not adhere to the strict criteria formulated in the consensus paper, it reflects clinical practice at the time in question, and its findings are still valuable.

In conclusion, overall success rates were high in this setting, where source control and soft tissue coverage were the basis of both treatment protocols. Direct empirical combination antibiotic therapy that consisted of vancomycin and rifampicin yielded higher success rates than delayed narrow-spectrum antibiotics. It is also evident that the former treatment can be used safely, as both protocols had low rates of emerging bacterial resistance. Early combination antibiotics that are directed at the most common pathogens should be considered as the standard of care in acute FRIs.