The main finding of this study is that external fixation of proximal tibia fractures is associated with a higher rate of superficial infections, but has a lower incidence of heterotopic ossifications compared to internal stabilization using locking plates. Furthermore, bony consolidation occurred slightly earlier in the plate fixation group. The likelihood to develop a non-union significantly depended on fracture displacement and classification and was associated with deep infections.
In agreement with our data regarding earlier healing following internal stabilization, Krupp et al. described a decreased time to union after plate fixation of bicondylar tibia fractures reporting an average of 5.9 months compared to 7.4 months for external fixation [
13]. Although a recent meta-analysis comparing both treatment methods for tibial plateau fractures found no significant difference regarding the time to union [
7], the recorded data document a similar tendency (17.73 ± 4.87 after external fixation vs 15.64 ± 4.36 weeks following plate fixation). This study included 22 case series and a retrospective cohort study, indicating the lack of direct comparing trials and a low summary evidence level. Similarly, a case series published by Cole et al. investigating the time to union in proximal tibia fractures following internal fixation using the Less Invasive Stabilization System (LISS) reported a time to full weight bearing without pain after an average of 12.6 weeks [
11]. Clinically, the reason might also be associated to a more courageous decision regarding weight bearing, when an internal stabilizer is in place to support the healed bone. The incidences of non-union were similar in both groups and were associated with fracture displacement and classification, and the occurrence of deep infections. This is supported by the literature, showing no significant differences in the rates of non-unions [
7,
13,
14]. A meta-analysis by Bhandari et al., examining extra-articular tibial fractures, described a trend towards higher non-union rates after external fixation; however, this study questioned its conclusions itself because of its low evidence grade [
16]. Our reported rates of non-union are in line with earlier studies, reporting a rate of non-unions up to 13% [
5‐
7,
11,
13,
15,
17,
18]. The fact that superficial infections are more common following external fixation corresponds well with the literature. A meta-analysis by Metcalfe et al., comparing 6 retrospective cohort studies, reported an odds ratio of 2.96, when external was compared to internal fixation with plating [
19]. Case series, investigating complications in fractures treated by Ilizarov frame fixation published by Keitgthley et al. and El-Sayed et al., showed superficial infection rates of 51.3 and 41.8%, respectively, which is similar to the 40.4% rate seen in this study [
5,
20]. Like our results, heterotopic ossifications occurred also more frequently in other series following plate fixation compared to external fixation [
7]. However, our reported rates were higher than other incidences, stating an average of only 1.23% [
7]. This might be caused by a subjective evaluation and lacking standards. A study published by Krupp et al. reported higher rates ranging at 7.14% after plate fixation, which also was higher than following external fixation [
13]. Considering the results for joint mobility, the statistically significant difference seemed not to limit function. Extensive soft tissue and bone damage often accompany high-energy injuries of proximal tibia fractures, increasing the likelihood of infectious complications and affecting the treatment algorithm [
3,
18]. Indeed, deep infections are a severe problem and have often been reported as relatively frequent following plate fixation in proximal tibia fractures. A RCT by the Canadian Orthopedic Trauma Society reported 17%, which was often used to argue for external fixation [
13,
21]. However, the study by Krupp et al. describes a higher deep infection rate following external fixation ranging at 13% [
13]. Besides this, the meta-analyses by Yu et al., Metcalfe et al., and Bhandari et al. support the results found in our study and reported comparable rates for both methods [
7,
13,
16,
19]. The data described in the literature support the concept that an operative method should be practiced by a surgeon and a team, which is used to do this. This is probably sometimes more important than the method itself and might explain the large discrepancies of published data. All steps from indication to aftercare require a certain competence, which is connected to experience and daily practice. Therefore, this study had a two-center-two-method design, considering the necessity of routine for each type of treatment. Furthermore, this study presents to our knowledge the highest number of patients in one trial published to date. Corresponding with our data, the range of motion and knee stiffness has been reported to be similar after fixation with one of the two examined methods [
7,
19,
21]. However, Krupp et al. reported a trend towards higher rates of extension deficits and knee stiffness following external fixation, but this was not conclusive [
13]. Similarly, Conserva et al. reported problems in 7.3% of tibia fractures that underwent framing compared to 0% treated by plate fixation [
22]. Unfortunately, this study included also type Schatzker IV and V fractures; so, it remains unclear how applicable this observation is regarding fractures with a complete metaphyseal component. Although the incidences of deep venous thrombosis, reoperations, compartment syndromes, and peroneal nerve paresis are decisive for the single affected patient, they are overall not very frequent and were observed in both groups with the same frequency. This conforms with the data published [
7,
19]. Metaphyseal-diaphyseal misalignments have been identified as one of the decisive parameters determining the clinical outcome. Corresponding with other studies, no significant differences between the two groups could be found in the presented study [
14,
16,
21]. However, as described in the meta-analysis by Yu et al. [
7], it is important to define how the measurements were done methodologically. This might also explain the differences in various studies reporting rates ranging for both groups between 3.3 and 43% [
13,
14,
21]. Although this study focused on the stabilization of the metaphyseal component of the proximal tibia fractures, the incidence of posttraumatic osteoarthritis has been included since the correct alignment is not only decisive for the clinical outcome [
23] but also for the development of osteoarthritis. Interestingly, the analysis by Yu et al. [
7] indicated a higher percentage with arthritis in the external fixation group. However, other studies found no significant difference. One reason, which needs certainly to be considered in this background, is the follow-up time. Only the study by Jansen et al. reported a mean follow-up period of 67 months; usually, the time frame was between 6 and 24 months [
13,
17,
19,
21,
24]. However, the posttraumatic osteoarthritis does not seem to be decisive for short-term patient reported outcome measures [
25], and might be more important, when long-term results are compared. In contrast, knee instability immediately affects the clinical satisfaction and has been reported to be more frequent following external fixation compared to plating [
7]. Our data did not indicate a difference just as described by the case series published by El-Sayed et al. [
5], investigating Schatzker VI fractures, and the study by Krupp et al., which reported rates between 3.6 and 3.3%, respectively [
13].
Limitations of the study are the sample size and the retrospective study design, which encounter the risk to miss significances and incomplete documentation. Moreover, rehabilitation and physiotherapy were assumed to follow standard guidelines, but this was partially not individually documented. Some outcome parameters are based on measurements in X-rays; however, the technical adjustments such as size and overview differed between the various pictures. To minimize the resulting bias, two examiners have contributed to the evaluation being especially critical when the values ranged around the set cutting values. Furthermore, the follow-up period and the outcome parameters were limited, lacking patient-related outcome measures.
Naturally, these limitations could be avoided when applying a different study design. Considering the limited case load of these injuries, a register or a multicenter setup would be preferable.