Introduction

Open fractures have an increased risk of infection and non-union. These complications may result in amputation and septic shock. The most severe cases, with significant soft tissue injury, need both osteosynthesis of the bone and a plastic surgical procedure, in the form of a free flap, to restore the soft tissue. Furthermore, the fragile soft tissue mantle in the distal tibia and the lack of reliable local flaps in this area is a challenge for orthopaedic and plastic surgeons. The ultimate goals of the treatment are to avoid amputation and infection, restore soft tissue cover and achieve union of the fracture (Figs. 1, 2 and 3).

Fig. 1
figure 1

A free fibula graft with muscle and skin from the right leg of the patient, is transferred to the left side, where the patient sustained an open distal tibial fracture, with substantial bone loss

Full size image
Fig. 2
figure 2

The fibula is exposed

Full size image
Fig. 3
figure 3

6 months after. Donor and the graft site. The patient is walking unaided, with no pain

Full size image

The literature remains inconclusive on the topic of antibiotic treatment and timing of soft tissue cover, probably due to the relatively small number of patients in each centre with this condition [1–5, 7–9]. Furthermore, hospital logistics may delay the most optimal course. The delay in time to skin cover is probably rooted in a lack of consensus on timing, different approaches to the treatment of severely injured patients with other life or limb-threatening injuries and lack of capacity. In our hospital, the delay in flap coverage was rooted in a capacity problem; typically, an elective tumour patient operation must be cancelled for the microsurgery team to operate on an open fracture patient.

The purpose of our study was to investigate the determining factors that reduce the risk of amputation, infection and nonunion and to identify relevant first-line antibiotics. We believe that our study is unique in its combination of data on microbiology and timing of cover of open fractures. These aspects have not previously been discussed in the same context, although they are closely associated.

Methods

This study was conducted at the Department of Orthopaedic Surgery and Trauma and at the Department of Plastic Surgery, Copenhagen University Hospital, Rigshospitalet, Denmark. Rigshospitalet is a referral centre for fractures with soft tissue loss and has a catchment population of 1.7 million.

The study included all patients with open fractures of the tibia, covered with free vascularized flaps at our institution from January 2002 to June 2013. Patients with initially closed fractures and patients with chronic osteomyelitis were excluded. The patients included in the study were identified from our database of all microsurgical procedures conducted by the Department of Plastic Surgery during the period.

We retrospectively collected data from patient records (history, tobacco use, fracture type, fracture union, timing of surgical procedures, flap failure, infection, amputation) and from our local microbiological database (samples, species, antibiotics, susceptibility patterns) and from the microsurgical database (flap types, timing).

Injury type was recorded according to Müller-OTA fracture classification and the Gustilo-Anderson soft tissue damage classification [1, 12]. Initial wound treatment was classified as “Open” when negative pressure wound therapy (NPWT), or any other type of open dressing was used. “Closed” wound treatment denoted cases where primary suturing of the wound proceeded to wound breakdown and necrosis. We defined infection when CRP and/or white cell count was elevated in combination with pus, discharge or wound breakdown, provided it was related to the initial lesion, including the flap. Superficial signs of infection and external fixator pin tract infections were excluded. Positive cultures or blood tests without clinical signs of infection were not included.

Union of the fracture was evaluated radiographically and we defined non-union when less than three out of four cortices had bridging callus in anteroposterior and lateral views, one year or later, after initial surgery. High-energy trauma was defined as: polytrauma in general, including falls from a height of ≥2.5 m, motor vehicle or motorcycle accidents, bicycle accidents, pedestrians being hit by any of the above and crushing injuries. Low-energy trauma was defined as fall from standing height or up to < 2.5 m or other low-impact injuries.

The bacterial species isolated from the wounds and their susceptibility patterns were defined with respect to time from injury. We included sample cultures between the second and 30th day after injury. Cultures from initial wound revisions were not included. This avoided the early wound contamination period (which has previously been shown to have poor correlation with later infection pathogens) [22, 23]. The samples included were biopsies harvested from deep tissue during surgical wound revisions of patients that were clinically infected. Blood-, pin site- and catheter cultures were not included. Identical results were counted only once. The susceptibility of identified microorganisms to relevant antibiotics was tested by disc diffusion. We included only positive samples that were fully susceptible to the antibiotic tested.

We used Fischer’s Exact test for dichotomous variables and set the level of significance at p = 0.05. We calculated the relative risk ratio for each outcome measure related to the energy of initial trauma. Clinical follow-up was a minimum of one year for all patients.

Results

From January 2002 until June 2013, 56 patients received a free vascularized flap to cover an open fracture of the tibia at our institution. Of these, 11 had insufficient or irretrievable patient records, leaving 45 patients to be included in the study. The patients with irretrievable records were all from the period (2002–2005), prior to the introduction of electronic patient records.

The study group consisted of 13 women and 32 men. The average age was 42 years (range 16–71, SD 18). Gustilo Anderson type IIIB fractures accounted for 26 (57 %) of the lesions, and six (13 %) were Gustilo Anderson IIIC. Thirty-one patients (67 %) sustained high-energy trauma. There were 15 smokers (33 %). One patient had bilateral fractures.

All patients were initially debrided with copious lavage. The average time to first debridement was 6.8 hours (range one to 26, SD 6.2). After debridement 31 patients (67 %) with 32 fractures continued with open wound care, typically with negative pressure wound therapy (NPWT). The remainder (15) had closed wound therapy with sterile dressings and suturing of the wound, that later went on to wound breakdown.

Sixteen patients were primarily stabilized with internal fixation (plates, nail or screws). The remaining 29 patients were treated with temporary external fixation, which was converted to internal fixation in combination with the free flap.

The average time to flap cover was 16 days (range two to 54 days, SD 13, excluding one outlier at 450 days). The free flaps consisted of 24 latissimus dorsi flaps, 13 gracilis flaps, three vascularized free fibulas, five anterolateral thigh flaps (ALT) and a single radial forearm flap. One patient had flaps to both tibias (patient no. 7) (see Table 1).

Table 1 Patient demographics
Full size table

Infection

Twenty-two fractures (48 %) became infected at an average of 21 days from the initial trauma (range four to 83 days, SD 21 days, excluding an outlier at 360 days). In the group receiving flap cover before day seven (early cover), five out of 18 became infected (27 %), and in the group of patients receiving the flap after day seven (late cover), 17 out of 28 became infected (60 %). The difference between infection rates in the two groups was statistically significant (p < 0.04).

Non-union

Nineteen (41 %) fractures were not united one year after osteosynthesis. Non-union occurred in ten out of 16 patients in the smoking group (63 %), compared to nine out of 30 patients (30 %) in the non-smoking group. The difference between non-union rates in the smoking and the non-smoking group was almost significant (p < 0.058).

Limb salvage

In four patients (9 %), continuing infection required treatment with a below the knee amputation. Two of these had an infected non-union. Mean time to amputation was 17.2 months (0.4, 14, 14 and 40 months). The association between amputation and infection was statistically significant (p < 0.04).

Flap failure

Seven patients (19 %) sustained partial or complete loss of the free flap, resulting in a secondary procedure. None of these patients were amputated and all of them underwent either successful repair or replacement of their flaps. Flap failure was significantly associated with smoking, with five out of seven (71 %) flap failures occurring in the smoking group (p < 0.04).

Injury severity

All four patients who were amputated were in the high-energy trauma group. Seventeen of 22 infected patients (77 %) were in the high-energy group. Furthermore, six out of seven (86 %) flap failures and 14 of 19 (74 %) non-union cases were in the high-energy group. When comparing high-energy trauma with low-energy trauma, the relative risk ratios for amputation, flap failure, infection and non-union were 3.8, 2.9, 1.6 and 1.4, respectively.

Culture results

We isolated 43 different bacterial species in 22 infected patients from day two to 30. Six of the infections were monomicrobial, nine had two different bacteria and the rest were polymicrobial. Seven bacteria accounted for 75 % of the infections, enterococcus species and coagulase negative staphylococcus (CoNs) being the most frequent. The patterns of sensitivity are seen in Table 2.

Table 2 The number of cultures fra infected wounds and their sensitivity pattern
Full size table

Discussion

The importance of timing of cover in open fractures has been investigated by a number of authors, most notably Godina, who was the first to report the importance of early skin cover to reduce the risk of infection [19]. Later, a number of other observers have come to similar conclusions, but many other aspects of trauma care may also play a role in preventing infection and securing union in these injuries.

Alleuyrand et al. found that patients receiving flap cover before day seven had a better outcome in terms of flap failure and infection, even when controlling for known risk factors such as severity of trauma [2]. Choudy et al. also found a higher non-union rate and infection rate in patients with flap cover after day seven [20].

Gopal et al. and Sinclair et al. reported series of open tibial fractures with very early skin cover (before day three) and definitive stabilization; 90–95 % of these patients had successful flap cover, with no infection, union of the fracture and excellent outcome without pain or walking disability [4, 5]. Such results are exceptional. In another series, infection rates, flap failure rates and non-union rates exceed 30–50 %. Other authors, in line with the guidelines of the British Orthopaedic Association, have reached similar conclusions, albeit at various breakpoints [2–5, 7–11].

Our study sample size did not permit a multivariate analysis of all possible confounders, but it confirmed unequivocally that patients covered before day seven had a significantly lower infection and non-union rate, irrespective of trauma degree.

These results should encourage surgeons to strive for an ortho-plastic service enabling rapid free-flap cover and definitive stabilization within one week after trauma. We accept that none of these studies are randomized trials of early and late cover, which is a general weakness of the literature.

In our study, flap failure was not a predictor of amputation. This is an important point, also observed by Choudry et al., illustrating that a flap revision or a second flap can often allow limb salvage [20]. At our institution, local muscle flaps are not used for immediate soft tissue cover after lower extremity trauma due to high complication and revision rates [3, 18, 20]. Choudry et al. also found that cover later than one week using soleus pedicled flaps for open tibia fractures resulted in higher nonunion rates, higher flap failure rates and more infection when compared to free muscle flaps [20]. Use of tobacco was a significant predictor of flap failure, a well known problem in plastic surgery, also described by Christy et al., [17]. Hence, smokers with complex injuries should be counseled on quitting smoking.

Patzakis and Wilkins (in 1989) were among the first to observe that immediate antibiotic prophylaxis in patients with open fractures is the single most important factor that will reduce the risk of infections [6]. Furthermore, grade III open fractures in need of tissue cover pose a problem for the clinician. The wound may be open for several days allowing colonization and adherence of selected bacteria that are resistant to the antibiotics given. In line with these observations, it has been shown that cultures obtained at initial debridements correlate poorly with later infections, which is why we only included cultures from patients that were clinically infected, and not cultures from day zero to two [22, 23]. Thus, antibiotic treatment should be broad, target both Gram positive and negative organisms, and the risk of generating resistance should be small [13–16]. The risk can be further reduced by using antibiotics that are renally excreted with minor impact on the normal flora, as proposed by Sullivan et al. [21]. Also, reduced selection for resistant pathogens can be expected due to the reduced time to soft tissue coverage, and the resulting decreased period with need for antibiotic treatment.

Gopal et al., in common with Pollack et al., have proposed the use of Cefuroxime and metronidazole for open tibia fractures. This was the combination of antibiotics used at our institution, but in only 12 of 43 (28 %) cases would these antibiotics have been effective against the bacteria cultured from our patients before flap cover [3, 4].

As depicted in Table 3, vancomycin, which is bacteriocidal, was effective against 29 of 43 isolated cultures and was active against all Gram-positive bacteria identified in the study.

Table 3 Outcome of patient demographics and complications
Full size table

Meropenem was effective against 24 of 43 organisms, with particular effect against the miscellaneous group, enterobacteriacae and other Gram-negative rods, enterococcus and anaerobes. Gentamicin covered 15 of 43 organisms, but none of the important enterococcus species.

Linezolid also covered the Gram-positive organisms in our samples, and has good penetration into tissues, but is only licensed for a limited period of time and is very costly.

Based on these results, we suggest a combination of vancomycin and meropenem as first line antibiotic prophylaxis. In combination, these antibiotics seldom lead to resistance, are generally well tolerated, and supplement each other well. They are both mainly renally excreted. In this series, vancomycin and meropenem would have covered 40 of 43 (93 %) organisms cultured. This has also been demonstrated in a series of 166 patients with chronic osteomyelitis occurring mainly after fracture with internal fixation, in which Sheehy et al. recommended vancomycin and meropenem for empirical initial treatment of the organisms identified at excision of the bone infection [13].

The patterns of resistance may differ geographically and should also be considered in a regional context. We are aware that prophylaxis with broad-spectrum antibiotics could result in unwanted resistance patterns, but this problem should be seen in the light of a very small number of patients presenting with open fractures with compromised soft tissue. However, short duration treatment with effective antibiotic regimes should also minimize the development of resistance and prevent later infection that will inevitably require much longer antibiotic therapy with risks for resistance.

Conclusion

We conclude that a delay in soft tissue cover beyond day seven from the initial trauma is associated with an increased infection and non-union rate. Smoking markedly increases the risk of non-union and flap failure. High-energy trauma increases the relative risk of flap failure, infection, non-union and amputation.

We also conclude that currently proposed antibiotics have limited effect on bacteria infecting grade 3 open fractures.

We have changed the standard antibiotic prophylaxis at our institution to vancomycin and meropenem, thus improving the expected coverage of organisms from 28 to 93 %.