Intra-articular fractures of the tibial plafond (pilon fractures) are among the most challenging of orthopaedic problems. Open reduction and rigid internal fixation5,12,18,22 replaced the traditional non-operative treatment6,11,16 of these injuries in the late 1960's and 1970's. The optimum treatment of this fracture remains controversial, as retrospective reviews of the results of open reduction and internal fixation have demonstrated high rates of complications. External fixation with limited internal fixation2 has gained some popularity since the late 1980's.
As far as we know, all previous studies of the operative treatment of these fractures had the inherent flaws of retrospective analyses and lacked concurrent controls. The purpose of this randomized, prospective study was to compare the rate of complications, the radiographic results, and the functional results after treatment of displaced fractures of the tibial plafond with open reduction and internal fixation with those variables after treatment of such fractures with external fixation and limited internal fixation.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Department of Orthopaedics and Rehabilitation, Vanderbilt University Medical Center, Nashville, Tennessee 37232-2550. Please address requests for reprints to Dr. Schwartz.
‡Orthopaedic Center of the Rockies, 2500 East Prospect Road, Fort Collins, Colorado 80525.
All adult patients who were to receive operative treatment of a displaced fracture of the tibial plafond at Vanderbilt University Medical Center or an affiliated Veterans Administration and county hospital in 1990, 1991, or 1992 were entered into the study. The patients were randomly selected for one of two operative procedures: open reduction and internal fixation or external fixation and limited internal fixation. The six attending orthopaedic surgeons who performed the operations had been assigned to a treatment group according to their expertise or to their preference with regard to fixation. Each patient was managed by the one of the six surgeons who was on call when the patient was seen in the emergency room. The treatment of the fracture was therefore subject to the patient's informed consent and the ethical constraints of the surgeon. This method (the randomized-surgeon design)20 minimizes selection bias in randomized clinical trials. The two surgeons who performed open reduction and internal fixation were fellowship-trained trauma surgeons.
All of the patients included in the study had sustained an intra-articular fracture of the tibial plafond, which was classified with use of the system of Rüedi and Allgöwer17,22. The indications for an operation included an open fracture and unacceptable alignment of the fracture (defined as a joint space or incongruity of the articular surface of more than two millimeters) or malreduction (greater than 10 degrees in any plane), or both, of the tibia or the fibula. A closed reduction was attempted for every fracture.
Fifty-three patients who had a fracture of the tibial plafond were identified during the period of the study. Patients who had a simple fracture of the malleolus were not included. The criteria for exclusion included an acceptable reduction of the fracture (seven patients, all of whom had a type-I fracture), severe osteoporosis (one patient), an inability to walk (one patient), or a neuropathic joint (one patient). One additional patient was excluded because of a transfemoral amputation secondary to a compartment syndrome that had developed in the thigh after open reduction and internal fixation. One patient was excluded because of inadequate follow-up; no complications had developed by the time of the latest examination (thirteen months postoperatively). Two patients who had external fixation for a severely comminuted fracture were managed with a primary arthrodesis and were therefore not included in the study. Thus, a total of fourteen patients were excluded from the cohort, leaving thirty-nine patients for analysis.
Open reduction and rigid internal fixation of both the tibia and the fibula was done through two separate incisions in eighteen patients, and an additional patient had open reduction and internal fixation of the tibia only through one incision (the fibula was intact) (Group I). Twenty patients were managed with external fixation, with or without limited internal fixation (Group II) (Table I). Twenty-seven patients were male and twelve were female. The ages ranged from sixteen to sixty-seven years. The average age was thirty-nine years in Group I and thirty-eight years in Group II. Of the thirty-nine fractures, twenty-nine (74 per cent) were closed and ten (26 per cent) were open. There were three open fractures (one grade I and two grade II)9,10 in Group I and seven (one grade I, two grade II, and four grade III) in Group II. According to the system of Rüedi and Allgöwer17,22, eight fractures (21 per cent) were type I, fourteen (36 per cent) were type II, and seventeen (44 per cent) were type III. There were four type-I, ten type-II, and five type-III fractures in Group I and four type-I, four type-II, and twelve type-III fractures in Group II. Twenty-one patients (54 per cent) had other associated injuries. Fifteen fractures (38 per cent) had been sustained in a motor-vehicle accident, and twenty-four (62 per cent) had resulted from a fall or a jump from a height.
Plain anteroposterior and lateral radiographs were made preoperatively and postoperatively. All patients who had an open fracture had débridement, followed by immediate stabilization of the fracture, at an average of three hours (range, two to five hours) after the injury. The closed fractures were treated with reduction and application of a splint, followed by operative treatment within forty-eight hours unless severe swelling or fracture blisters were present or there were medical contraindications. If the operation was delayed for more than forty-eight hours, the patient was placed in skeletal traction or was elevated on a Bohler-Braun frame. The average time from the injury to the operative fixation of the closed fractures over-all was five days (range, three hours to seventeen days).
Preoperatively and postoperatively, antibiotics were administered parenterally to all patients. The patients who had a closed fracture received cephalexin, one gram every eight hours for forty-eight to seventy-two hours before and after the operation; an aminoglycoside (gentamicin) was added to this regimen for the patients who had an open fracture.
Two surgeons managed all nineteen of the patients in Group I. Eighteen of these patients had an operation that involved two separate incisions to stabilize both the tibia and the fibula (Figs. 1-A, 1-B, 1-C, 1-D through 1-E). The fracture of the fibula was reduced through a lateral incision and was stabilized with a plate in seventeen patients and with an intramedullary rod in one. The tibial fracture was exposed through an anteromedial incision, with at least a seven-centimeter skin bridge from the posterolateral incision. The nineteenth patient had an intact fibula, and only one incision was made, for fixation of the tibia. After open reduction of the distal articular surface of the tibia and inspection of the talar dome, a buttress plate was applied to stabilize the fracture. The type of buttress plate varied according to the type of fracture and was selected by the attending surgeon. A 3.5 and a 4.5-millimeter dynamic compression plate was used in six and three patients, respectively; a cloverleaf plate, in eight; and a mini-fragment T-plate, in two. A femoral distractor was used to apply distraction intraoperatively and to assist with achieving an appropriate length for reduction and with visualization of the joint. Bone-grafting was performed at the time of fixation in ten patients in this group. Seven patients in whom the wound could not be closed primarily were returned to the operating room within seventy-two hours for repeat irrigation and débridement or a soft-tissue coverage procedure, or both. This was not recorded as a complication. Postoperatively, the lower extremity was immobilized in a plaster splint or cast for two to three weeks until the swelling had resolved and the soft tissues had healed. Active and passive range-of-motion exercises were then initiated with the patient wearing a hinged brace, and the patient walked with toe-touch weight-bearing for an additional eight weeks.
The twenty patients in Group II had limited internal fixation combined with external fixation; an Orthofix fixator (EBI Medical, Parsippany, New Jersey) was used for eleven patients, and a Synthes AO fixator (Paoli, Pennsylvania) was used for nine (Figs. 2-A, 2-B, 2-C through 2-D). Four attending surgeons managed the patients in this group. In eighteen patients, an unstable fracture of the fibula was fixed with a plate; however, no tibia was fixed with a plate. Two patients had an intact fibula. Reduction of the articular surface was performed through a small (less than two-centimeter-long) anteromedial incision in each patient. Any joint-surface fragments were disimpacted and reduced with use of a 3.5-millimeter interfragmentary screw or a percutaneous cannulated screw in all but three patients. For external fixation, two proximal and two distal 5.0-millimeter pins were used. Distal fixation was accomplished with use of one talar pin and one calcaneal pin in eighteen patients and with two calcaneal pins in two patients.
Care of the pin site was begun immediately postoperatively and consisted of cleaning with hydrogen peroxide three times per day. The patients were also advised to maintain toe-touch weight-bearing for ten to twelve weeks. The external fixator was kept in place for an average of ten weeks (range, six to fourteen weeks) and was removed after there was radiographic evidence of healing callus. Bone-grafting was not done at the time of the initial fixation in any patient in this group; at an average of six weeks (range, four to eight weeks) after the initial treatment, bone from the iliac crest was grafted to fill a metaphyseal defect and promote fracture-healing in ten patients.
A major complication was defined as an infection that necessitated operative treatment, breakdown of the wound that necessitated a soft-tissue coverage procedure, a neurovascular injury, failure of fixation, malunion (more than 10 degrees in any plane), non-union, or amputation. Non-union was defined as failure of clinical and radiographic union more than six months after the injury.
Patients returned for follow-up visits at least every three months for the first year and every six months thereafter. The average duration of follow-up was thirty-nine months (range, twenty-five to fifty-one months). A clinical score19 was derived with use of a questionnaire with which the patient evaluated pain and the functional outcome and the surgeon evaluated gait and the range of motion (Table II). Serial radiographs were made during follow-up visits at the clinic and were evaluated for the development and progression of post-traumatic osteoarthrosis. Degenerative changes, if present, were rated as mild (one to two millimeters of narrowing of the joint space), moderate (more than two millimeters of narrowing with or without periarticular cysts or osteophytes), or severe (complete loss of the joint space with osteophytes or cysts). At the time of the latest follow-up, the grade of osteoarthrotic changes and the clinical score were determined by one of us (B. W.). Statistical analysis with use of the Student t test was employed to compare the average clinical scores in the two groups. The radiographic scores for the two groups were compared with use of the Mann-Whitney U test for non-parametric ordinal variables. The complications in the two groups were compared with use of chi-square analysis.
Complications
Group I (open reduction and internal fixation): In seven patients, fifteen major complications necessitated twenty-eight additional operations. The complications included breakdown of the wound necessitating coverage with a free flap (six patients); recurrent infection or osteomyelitis, or both, that developed after a soft-tissue coverage procedure and the parenteral administration of antibiotics (five patients); infection around the hardware (one patient); and amputation (three patients). Of the six patients who had breakdown of the wound, four had had a closed fracture and two had had an open fracture (one grade I and one grade II). Breakdown of the wound occurred at an average of six weeks (range, four to thirteen weeks) postoperatively. The patient who had infection around the hardware (with no breakdown of the wound) had a grade-II open fracture9. Three of the four recurrent infections were due to Staphylococcus aureus that was resistant to methicillin. The fourth infection was caused by Enterococcus and was associated with a non-union. Despite an initial good result after treatment with an Ilizarov device, recurrent osteomyelitis developed with a draining sinus and the patient chose to have a below-the-knee amputation twenty months after the initial treatment for the closed fracture (Figs. 3-A, 3-B, 3-C, 3-D through 3-E). This patient was an otherwise healthy man and had no apparent risk factors for infection. Two other patients had a below-the-knee amputation, at fourteen and twenty-four months after the initial fracture. Both patients had had a wound slough after a closed, comminuted fracture; treatment had included débridement, prolonged antibiotic therapy, and soft-tissue coverage with a free flap. In both patients, osteomyelitis caused by methicillin-resistant Staphylococcus aureus developed, and they chose to have an amputation rather than additional attempts at limb salvage. All of the other patients in whom a complication occurred had healing of the fracture and resolution of the infection, if one had been present, by the time of the most recent follow-up.
Group II (external fixation): There were four major complications in four patients, necessitating five additional operations. Diminished plantar sensation and mild reflex sympathetic dystrophy developed in the foot of one patient postoperatively. Exploration of the posterior tibial nerve at the tarsal tunnel sixteen months after the injury revealed scarring around the nerve secondary to a partial nerve injury caused by placement of a calcaneal fixation pin. At the time of the latest follow-up, the dystrophy was gradually resolving with full return of the function of the posterior tibial nerve. In another patient, who had a grade-III open fracture, the wound became infected and the skin graft failed two weeks postoperatively. The patient was managed successfully with a temporalis-parietal free flap. One patient had loss of the reduction four weeks postoperatively; the fracture healed in satisfactory alignment after use of an Ilizarov device. The fourth complication occurred in a patient who had a closed fracture; a pin track became infected with methicillin-resistant Staphylococcus aureus and infection subsequently developed in the ankle joint. The patient was managed successfully with two irrigations and débridements. A spontaneous fusion of the tibiotalar joint occurred, as two-thirds of the articular surface of the tibial plafond had been lost at the time of the injury.
Of the nineteen complications in the two groups, thirteen had occurred within four months after the injury and all involved a soft-tissue infection or a wound breakdown. In one patient in Group I, the hardware was removed because of infection eleven months after the injury, and an additional three patients in this group had a below-the-knee amputation at fourteen, twenty, and twenty-four months after the injury because of recurrent breakdown of the wound, purulent drainage, and osteomyelitis.
Adequacy of the Reduction and Rates of Union
For both groups, postoperative radiographs were evaluated for the adequacy of the reduction of the joint. In Group I, eight (three type-I and five type-II) fractures had an anatomical reduction. A one-millimeter incongruency was seen after the reduction of five (one type-I, two type-II, and two type-III) fractures, and a two-millimeter step-off or gap in the articular surface was seen after the reduction of six (three type-II and three type-III) fractures. All of the patients in Group I had an anatomical reduction or healing with less than 5 degrees of angular malalignment.
In Group II, four type-I fractures had an anatomical reduction. A one-millimeter incongruency was seen after the reduction of five (two type-II and three type-III) fractures, and a two or three-millimeter incongruency was seen after the reduction of ten (two type-II and eight type-III) fractures. Four (one type-II and three type-III) fractures healed in malalignment, with 5 to 10 degrees of recurvatum (two fractures) or valgus (two fractures). Substantial metaphyseal comminution or bone loss, or both, was associated with all of these fractures. The remaining fractures healed in anatomical alignment or less than 5 degrees of angulation.
The fractures in Group I healed at an average of fourteen weeks (range, ten to forty-two weeks) after the operation. Ten patients had bone-grafting at the time of the initial operation or after delayed closure of an open wound. One patient had bone-grafting at twelve weeks because of delayed union.
The fractures in Group II healed at an average of fifteen weeks (range, eight to forty weeks) after the operation. None of these patients had had bone-grafting at the time of the initial treatment of the fracture; however, eight patients had bone-grafting, at an average of six weeks (range, four to eight weeks) after the initial treatment, to fill a metaphyseal defect and promote fracture-healing. Bone-grafting was done sixteen weeks postoperatively in an additional two patients because of delayed union of a metaphyseal fracture.
Clinical Score
A clinical score could be determined at the time of the most recent follow-up for fifteen patients in Group I and for nineteen in Group II; the average scores were 61 and 73 points, respectively (Table I). With the numbers available, the difference between these scores was not significant (p = 0.6).
The clinical score did not correspond well with the type of fracture17,22. For type-I fractures, the lowest score was 33 points (in the patient in whom the hardware was removed because of an infection); this score was less than the average score for the patients who had a type-II or III fracture. Alternatively, one patient who had a type-III fracture and radiographic evidence of severe osteoarthrosis had a score of 95 points, which was greater than the average score for the patients who had a type-I fracture. A general trend, however, was found when the average scores associated with each type of fracture were compared. The patients who had a type-I fracture had a greater average score (80 points) than the patients who had a more comminuted fracture (average score, 65 points for the patients who had a type-II fracture and 64 points for those who had a type-III fracture).
Only one patient (in Group I) had a late, elective arthrodesis. However, at the time of the most recent follow-up, four patients (all of whom had moderate or severe osteoarthrosis) from each group were considering an arthrodesis. One of these patients (in Group I) was also considering an amputation.
Radiographic Score
All of the patients who had a type-II or III fracture, regardless of the treatment, had some degree of narrowing of the joint space on the latest follow-up radiographs. In Group I, the osteoarthrotic changes were mild in six patients, moderate in four, and severe in four; one patient had no changes. Of the remaining patients in Group I, three had a well healed residual limb after an amputation and one had a solid fusion of the ankle joint; the data for these patients were not included in the radiographic evaluation. Similar results were found in Group II: one patient had no changes, six patients had mild changes, eight had moderate changes, and four had severe changes. With the numbers available, the difference between the two groups was not significant (Mann-Whitney U test for non-parametric ordinal variables). One patient in Group II had a spontaneous fusion of the ankle joint, which was solid at the time of the most recent follow-up. All of the patients who had severe osteoarthrotic changes had progressive narrowing of the joint space on radiographs, beginning one year after the injury.
In 1969, Rüedi and Allgöwer22 reported a 74 per cent rate of good or excellent functional results after the treatment of eighty-four consecutive pilon fractures with open reduction and internal fixation. They advocated that surgeons follow four ASIF principles in the treatment of these fractures: (1) restoration of the length of the fibula, (2) reconstruction of the articular surface of the tibia, (3) grafting with cancellous bone to fill metaphyseal defects, and (4) stabilization of the medial portion of the tibia with a buttress plate17,22. Many authors have reported good results, with anatomical reduction and stable fixation of these fractures, after treatment according to these basic principles5,12,15,18.
Many of the fractures in the series of Rüedi and Allgöwer22 had occurred during skiing accidents and involved a rotational mechanism of injury. A rotational injury usually results in large metaphyseal fragments with minimum impaction or comminution of the joint. In contrast, in large urban trauma centers, the most common cause of fractures of the tibial plafond is axial compression (dorsiflexion), which results in severe comminution of the joint with impacted fragments and injury of the articular cartilage. In addition, soft-tissue trauma usually is much less severe with a rotational fracture than with an axial-compression injury12. As a result, the axial-compression injury is more challenging to treat12,21, with a greater likelihood of soft-tissue complications postoperatively and a poorer clinical result secondary to post-traumatic osteoarthrosis.
The results of open reduction and internal fixation of fractures of the tibial plafond have not been excellent or good at many trauma centers. The thin, traumatized soft-tissue envelope about the ankle and the complex pathoanatomy of this fracture can lead to numerous complications. In various clinical series5,12,16,18,26, the reported frequencies of osteomyelitis, amputation, post-traumatic osteoarthrosis of the ankle, and non-union have been as high as 20, 6, 54, and 18 per cent, respectively. In addition, the rates of wound breakdown and deep infection have been reported to be as high as 100 per cent after open reduction and internal fixation of severely comminuted fractures of the tibial plafond7. In a recent retrospective study of the operative treatment of these fractures at our institution, major complications developed in twenty-one of fifty-two patients who had had open reduction13. Excellent results with few complications after treatment with external fixation have recently been reported1,4,14,24.
Delicate handling of soft-tissues and meticulous débridement of wounds in open fractures have been advocated to minimize soft-tissue complications3,15,21. The timing of the operation also is an important factor3; an operation that is performed in the presence of severe intradermal edema or fracture blisters may increase the risk of wound tension, leading to sloughing of the skin and tissue necrosis, with subsequent infection. Even with at least a seven-centimeter skin bridge between the medial and lateral incisions, skin slough and wound breakdown may be inevitable because of the initial soft-tissue injury. Excessive skin tension at the time of closure of the wound may also be problematic when tissue edema is present. Even after swelling of the soft tissues has subsided, there may be increased wound tension as a result of the medial skin flap being stretched over a buttress plate on the tibia.
The use of a plate for fixation of a fracture of the tibial plafond should not be universally condemned. Some soft-tissue problems and complications may be avoided with use of a lower profile, less bulky buttress plate. In addition, open reduction and internal fixation may be an excellent option for this fracture when the fibula is intact; because a lateral incision is unnecessary, the need to create a vulnerable skin bridge between two incisions is avoided. Furthermore, open reduction and internal fixation can usually be performed safely when there is little soft-tissue injury or compromise, as is often seen after a rotational injury.
Several investigators have returned to the principles of Scheck, who advocated reconstruction of the joint surface with limited open reduction and emphasized that little soft-tissue stripping is needed for this technique. In two reports by Bone et al.3,4, the result was good or excellent for eleven of sixteen patients who had been managed with external fixation combined with limited internal fixation; there were no infections or clinically important complications related to the wound. Recently, Bonar and Marsh1 reported on twenty-one patients in whom a severe fracture of the tibial plafond had been treated with unilateral external fixation. There were no soft-tissue complications and osteomyelitis did not develop in any patient. The early functional results were promising. Those authors also demonstrated good functional results with few complications after use of an articulated fixator14. The results of our study are similar to these findings; we demonstrated a substantially lower rate of soft-tissue complications after the use of external fixation, even for severely comminuted and open fractures.
When patients who had an arthrodesis, a spontaneous fusion, or an amputation were excluded, there was some loss of motion of the tibiotalar joint (compared with the contralateral extremity) in all but eight patients (four in each group). It is important to note that our study protocol did not include release of the articulated hinge of the fixator to allow early motion at the ankle joint, as has been reported in other studies of external fixation1,14,24. Therefore, we could not determine whether early motion in the fixator may have improved motion for the patients in Group II. (We now include release of the articulated hinge of the DynaFix fixator [EBI Medical].) In their long-term follow-up study, Rüedi and Allgöwer noted that 48 per cent of their patients had some loss of motion compared with that of the contralateral extremity21,23. Bourne et al. reported that 81 per cent (twenty-five) of their thirty-one patients had loss of motion after open reduction and internal fixation of a type-II or III fracture. Bone et al.4 reported that of twenty patients who had external fixation of an open fracture of the tibial plafond fourteen had loss of motion at the time of the most recent follow-up. The findings in the current study seem to be similar to the results of these latter two studies: almost 80 per cent of our patients had some loss of tibiotalar motion.
A loss of motion, however, does not always correspond with a poor outcome. Rüedi and Allgöwer23 noted that many of their patients seemed to adapt to their status after the injury, despite a major loss of motion. Similarly, most patients in the current series did not consider the loss of motion to be a disabling factor. It should be noted, however, that limitation of ankle motion may hamper the specific activities of certain patients, such as very active athletes or dancers.
In all but two patients, there was some loss of the joint space and radiographic changes consistent with post-traumatic osteoarthrosis at the time of the latest follow-up. Mild, moderate, or severe osteoarthrotic changes developed in all of the patients who had a type-II or III fracture in both groups. This suggests that the development of post-traumatic osteoarthrosis after a fracture of the tibial plafond may be a result of severe damage to the articular cartilage at the time of the original injury and is independent of the type of treatment or the accuracy of the articular reconstruction. Lower rates of osteoarthrosis have been found in previous studies5,12,16,18,21,26; however, osteoarthrosis has been poorly defined in these studies, which may have resulted in the underestimation of true rates.
The most serious osteoarthrosis in our patients appeared within the first twelve to twenty-four months after the injury. Rüedi and Allgöwer also reported that osteoarthrosis usually developed within one to two years after the injury23. In their series, however, 22 per cent of the patients had a slight improvement in function nine years after open reduction and internal fixation of a fracture of the tibial plafond; 10 per cent had a deterioration in functional status. In the current study, no patient had radiographic or clinical evidence of decreased osteoarthrosis or improved function. There was, however, deterioration of function in both groups, as demonstrated by the number of patients who were considering an arthrodesis (eight patients) or who had already had one (one patient).
Another important finding was that the radiographic severity of osteoarthrosis did not always correspond well with the subjective clinical result. For example, nine patients (three in Group I and six in Group II) who had moderate or severe osteoarthrotic changes on the radiographs had a clinical score of more than 70 points. Similarly, in a ten-year follow-up study of patients who had had open reduction and internal fixation of a fracture of the tibial plafond, Etter and Ganz noted that severe osteoarthrosis did not correspond with a poor subjective or objective result. This is relevant because the decision about an arthrodesis should be based on a consideration of pain, disability, functional limitation, and over-all dissatisfaction, rather than on the radiographic appearance of the joint after trauma.
Because of the substantially greater number of complications after open reduction and internal fixation in our series, with no difference in the long-term clinical outcome, we concluded that limited internal fixation combined with use of an external fixator is an equally effective and safer method of treatment for most fractures of the tibial plafond.