The most important finding in this study was that, in both unicondylar (Schatzker I-IV) and bicondylar fractures (Schatzker V-VI), the Ilizarov fixation allowed early weight-bearing without jeopardising the fracture stability and healing.
Maripuri et al.  claimed that the Schatzker classification was superior to the AO  and the Hohl and Moore  classification in terms of both inter-observer reliability and intra-observer reproducibility. However, they also concluded that none of the classifications was able fully to describe all fracture types. In the present study, the Schatzker classification was used to differentiate between two biomechanically different fracture subsets, one with continuity between a part of the articular surfaces and the diaphysis (I-IV types) and one without such continuity (V-VI types). Most unicondylar tibial fractures are caused by a forced varus or valgus load. In bicondylar tibial fractures, there is also an axial load resulting in a combination of depression of the articular surface, metaphyseal crush and shearing of one or both condyles. Vertical displacement is possible because there is no shaft below the fragment, which creates a shear vector. With the “olive wires” in the Ilizarov ring fixator; these forces are counteracted, holding the condyles together, which creates a relatively stable joint surface configuration that can be fixed to the tibia distally of the fracture. The distinction between uni- and bicondylar fracture is important, because, in fracture types I-IV, there is a risk of dislocation of the fractured part of the articular surface relative to the diaphysis when loaded. Because of the discontinuity between the articular fragments and the diaphysis in the V-VI fractures, compressive forces will not normally increase the risk of displacement of the articular surfaces.
As expected, the operating time was longer for the more complex fractures. In spite of this, the operating time in the present study compares favourably with that of Lee et al.  who operated on thirty-six tibial plateau fractures using the less invasive stabilisation system (LISS); their mean operation time was 150 minutes. Pre-assembling the frame could reduce the time in the operating room but one important advantage of the Ilizarov technique is that it is an essentially closed method and if the surgical time is extended, the risk of wound contamination is low when compared with open plating of the tibial plateau .
The pain subsided rapidly and did not constitute a problem after the first 24 hours post-operatively. We have not found any report of a need for post-operative analgesia in these types of fractures, but the amount of analgesics in the PCA pump corresponds to that in patients with total knee arthroplasties in our hospital data base.
The reported incidence of joint capsule, ligament and meniscal injuries is high. Colletti et al.  analysed MRI findings in 29 tibial plateau fractures and found associated collateral ligament injuries in 55%, lateral meniscal tears in 45%, anterior cruciate ligament injuries in 41%, posterior cruciate ligament injuries in 28%, and medial meniscal tears in 21%. Gardner et al.  found that only 1% of tibial plateau fractures showed a complete absence of soft-tissue injuries, evaluated by MRI. These injuries can also be diagnosed arthroscopically . However, even if recommended by some authors [32–35] there is no support for this in randomised trials . The percutaneous treatment of fractures of the tibial plateau can be performed using arthroscopy or fluoroscopy to control the reduction of the joint surface. Lobenhoffer et al.  were not able to demonstrate any significant benefit from arthroscopy compared with fluoroscopic reduction in 168 patients with tibial plateau fractures. Ohdera et al.  found no significant difference between arthroscopic management of tibial plateau fractures compared with the open reduction method in terms of duration of operation, post-operative flexion, and clinical results in 28 patients. The arthroscopic procedure was only recommended in selected tibial plateau fractures. In the present series, it was possible to achieve an acceptable reduction according to the criteria formulated by Rasmussen  in most patients without the use of arthroscopy.
In a retrospective study, Park et al.  found a low rate (1.6%) of compartment syndromes requiring fasciotomy for proximal tibia fractures. However, in more complex fractures, the risk of compartment syndrome is considerably higher. For Schatzker type VI fractures, Stark et al.  found an overall risk of 27%, as well as a difference depending on whether or not the medial plateau was dislocated, 53 and 18% respectively. The incidence of compartment syndrome in the severe fractures (V and VI) in the present series was comparatively low; 2/19 patients; however the observed compartment syndromes were interpreted as a direct result of the fracture and the soft-tissue injury and not of the operation. In spite of the Ilizarov technique is beneficial with respect to soft-tissue injury, minimizing the risk of developing compartment syndrome; the frame should not prevent this salvage procedure when necessary.
Some studies support the staged protocol for proximal tibial fractures, especially if high energy fractures are present [42–45]. The Ilizarov method gives the advantage, independently of fracture pattern, to operate on all patients without delay. In this way, we were able to avoid disturbing the healing process with other further interventions to the soft-tissues which may delay rehabilitation.
Most treatment methods do not allow full weight-bearing in intra-articular proximal tibial fractures . The mobilisation and degree of weight-bearing that is allowed is determined by the fracture displacement, method of treatment, and quality of aftercare [47, 48]. In the present study, all the patients were allowed unrestricted weight-bearing without any signs of the reduction being compromised.
In earlier series, the infection rate after treating tibial plateau fractures with ORIF, varies from 6% to 87.5% [49–51]. The use of bilateral incisions and the reduction of the size of the implants have reduced this rate to 3 –8.4% [52–54]. Despite using a generally recommended staged protocol, Egol et al.  reported a deep wound infection rate of 5%. When comparing external devices in different locations, Parameswaran et al.  reported that ring fixators had the lowest incidence of infection. Using the Ilizarov technique, Catagni et al.  did not observe any deep infections in a series of 59 patients with Schatzker V-VI fractures. In the present series, the majority of observed infections were easy-to-treat superficial “pin-site” infections. Only two patients had “pin-tract” infections, and they could be treated without compromising the fixation or fracture healing.
The use of autogenous iliac crest bone grafts is associated with risk of increased morbidity from the donor site [57, 58]. Good results have been reported in previous studies using bone graft substitutes in terms of the prevention of redislocation of the articular surface in tibial plateau fractures [59, 60]. Beuerlein and McKee  found several studies reporting that calcium sulphate is an effective safe void-filler in bone defects after impacted fractures have been reduced. There is also evidence that the bioresorbable calcium phosphate is a better choice than autogeneous iliac bone grafts for the treatment of subarticular defects associated with unstable tibial plateau fractures [62, 63]. At the one-year control, we were unable to detect any subsidence of the graft, which can be regarded as being at least partly integrated in all patients.
Conventional radiographs alone are not able to define union in internally fixed fractures with sufficient accuracy to enable their use as end-points of fracture healing. Generally, deciding when a fracture can be regarded as “healed” is difficult. In a recent study, Corrales et al.  reported a lack of consensus with regard to the definition of fracture healing. The surgeon’s ability to judge fracture union using chronological radiographs following internal fixation is estimated to be correct in approximately 70% . The use of traditional external fixation methods, such as manual testing of fracture stability and/or pain response to weight-loading with the frame disassembled, can be added to the evaluation of the radiological healing. These tests could therefore be used to assess whether the fracture has healed sufficiently to allow the safe removal of the fixator and full, unprotected weight-bearing. Using these criteria, we had no refractures or increased deformities.
Several authors have discussed the degree of dislocation that can be accepted with remaining good knee function. The long-term results reported by Rasmussen  and Lansinger et al.  showed that a residual depression of up to 10 mm could be accepted if the knee was stable. In a 5-year follow-up on 109 fractures, Lucht and Pilgaard  reported that the functional outcome with a depression of <10 mm was acceptable. In terms of articular depression the recommended “acceptable” dislocation varies between 2 and 10 mm . Marsh et al.  pointed out that the scientific basis for the different recommendations is generally weak. Giannoudis et al.  found that, in tibial plateau fractures, articular incongruities appear to be well tolerated. In addition to the articular depression, Rasmussen also found that instability and residual joint malalignment with varus and valgus angulations over 10° affected the outcome adversely. The residual displacements observed in the present series are within these limits in all but three patients. No one of the three patients with asymptomatic knee laxity had a valgus plateau tilt exceeding five degrees.
Knee stiffness is a common problem after tibial plateau fracture surgery  Gaston et al.  reported that, at one year, patients with tibial plateau fractures still ran a risk of 20% risk of knee stiffness, defined as flexion of less than 100° and an extension deficit of less than 5°. However, good results have been achieved with hybrid or ring fixators [72, 73] and the results in the present study compare favorably with these. Even in the complex fractures requiring a hinged extension to the femur, only four of 15 patients had knee flexion of 90° or less.
It has been shown that individuals with proximal tibial fractures have substantial residual limb-specific and general health deficits even at two years of follow-up . When we started the present study self-appraisal scores were rarely used in fracture patients. In recent years there has been an increasing interest in the patients opinion about the outcome. However there is no consensus of which score to use. As showed in Tables 7 and 8 different self-appraisal scores are being used or scores that are a mixture of the patients and the surgeons opinion. A previous evaluation of NHP scores in a prospective trial designed to study the effect of Ilizarov reconstruction of post-traumatic lower-limb deformities on general health status showed improvements equal to or better than the improvements reported for other orthopaedic procedures, including total joint arthroplasty . The patients self-appraisals used in the present series (NHP, EQ5D, Pain-VAS, Satisfaction-VAS, KOOS) showed that the Ilizarov fixator was well tolerated and the overall restoration of function was good. Some residual pain was still present at the one-year control, which most probably reflects the severe nature of these fractures more than treatment failure.
Despite successful treatment and improvement in their outcomes, the KOOS subscores showed the lowest values for Sports and QOL activities, which is probably due to the fact that patients studied earlier with this score are commonly younger and more active than the patients enrolled in the present study. In two recently published studies KOOS have been used in the follow-up after intramedullary nailing of tibial shaft fractures and operated patellar fractures showing comparable results as in the present study [75, 76]. Apart from this, patients with fractures type I-IV had results similar to patients after ACL reconstruction  and also the patients with type V-VI had acceptable results.