Historically, complication rates between 9% and 60% [14, 25, 28, 33, 35, 41], nonunion rates between 5% and 37% [1, 9, 35, 55], and limitations of the activities of daily life or occupation in 33% to 50% of patients [1, 17, 18, 25] have been reported after ankle arthrodesis. These studies looked at historical techniques and thus included a high proportion of patients who had been treated with external fixation, were fused in equinus or varus malalignment, and those who underwent simultaneous fusion of adjacent joints. More recent studies that employed screw fixation and strived for fusion in neutral position of the ankle reported substantially better functional outcomes and union rates between 87% and 100% [5–7, 16, 19, 30, 42, 46–48, 53].

Over the last 20 years, total ankle replacement has evolved as a treatment alternative to arthrodesis in end-stage ankle arthritis. While comparative studies are still missing, the complication rates and medium-term functional outcomes of both procedures seem similar [20, 44]. While aseptic loosening is the most frequent reason for failure after total ankle replacement [20, 44], the development or progression of arthritis at the talonavicular and subtalar joints remains a major concern after ankle arthrodesis [4, 6, 17, 44, 55]. Long-term studies with followup of more than 20 years revealed a high percentage (ie, 47% to 100%) of arthritis at the adjacent joints [11, 17, 51]. However, only few studies reported the amount of preexisting arthritis of these joints [5, 27, 43, 48] and the clinical relevance of radiographic signs of arthritis is not entirely clear [4, 37, 51].

All patients were evaluated by one of the authors (TE) pre- and postoperatively using the American Orthopaedic Foot and Ankle Society (AOFAS) ankle/hindfoot scale [29], and weight-bearing anteroposterior (with both feet in 20° of internal rotation) and lateral radiographs of both ankles. To evaluate the amount of axial deviation, a hindfoot alignment view [10] of the lower leg was obtained and the hindfoot axis with respect to the tibia measured according to Saltzman and el-Khoury [40]. The degree of arthritis at the ankle, subtalar, and talonavicular joints was documented by one of us (TE) using the radiographic scale provided by Bargon and Henkemeyer [3] (Table 1). Data on interobserver variability of this radiographic scale are not available. We performed CT scanning of the ankle and/or subtalar joint in all patients with radiographic evidence of subtalar or talonavicular arthritis at the time of initial presentation because this could influence the management plan, indicating necessity to fuse adjacent joints. We obtained MRI in cases of suspected necrosis after talar or pilon fractures to determine the subsequent need of bone grafting. In 35 patients with malunited malleolar fractures and only mild signs of posttraumatic arthritis (Bargon grade 0 to I), we performed an initial arthroscopy or arthrotomy of the ankle to evaluate the quality of weight-bearing cartilage. In 29 of these 35 patients viable cartilage was still present, therefore peritalar osteotomies or anatomic repair of a chronic unstable distal tibiofibular syndesmosis were preferred as joint-preserving procedures [31, 36, 54]. These patients are not included in the present study.

All patients underwent an ankle arthrodesis with débridement and fusion using the four-screw fusion technique, first described by the senior author (HZ [55]). The patient was positioned supine with a thigh tourniquet that was inflated to between 200 and 300 mm Hg depending on the individual weight. We approached the ankle midline through a 6- to 8-cm-long longitudinal incision close to the medial border of the anterior tibial tendon (Fig. 1A–B). Care was taken to avoid injury to the medial dorsal cutaneous nerve, which may cross over the distal incision medially. We incised the proximal extensor retinaculum in a Z-shaped manner to allow easier reapproximation at the end of the operation. The anterior tibial artery and veins and the deep peroneal nerve were held away laterally with the extensor hallucis longus and extensor digitorum longus tendons (Fig. 2A–B). The anterior joint capsule and any existing osteophytes were resected. A generous débridement and removal of all sclerotic and nonviable bone was performed with osteotomes, curettes, and a rongeur. A midsized laminar spreader was placed between tibia and talus first laterally, then medially to allow complete débridement of the posterior aspect of the ankle, as well as the medial and lateral gutters. To determine adequate joint resection, we deflated the tourniquet at that time to ensure that a bleeding bone surface was exposed. Defects in 38 of 94 cases were grafted with autologous bone, especially in case of varus malposition of the talus or in the presence of avascular necrosis at the talar dome or lateral tibial plafond. A monocortical bone block was taken from the ipsilateral iliac crest and additional cancellous bone chips were interposed (Fig. 3A–B). In 21 cases of shortening of the distal tibia of more than 5 mm after correcting frontal malalignment, a shortening osteotomy of the distal fibula was performed to avoid calcaneofibular impingement. In 10 patients with marked equinus of more than 10° we performed percutaneous lengthening of the Achilles tendon to achieve correction. A roll of sheets below the distal tibia served as a fulcrum for forced reduction of the foot with respect to the tibial axis to avoid anteversion of the talus. The ankle was fixed temporarily with a 2.5-mm Kirschner wire (Fig. 4A–B). Proper positioning of the center of the talar body below the longitudinal axis of the tibia in both planes, correction of equines, and frontal (varus/valgus) malalignment were assessed fluoroscopically. The arthrodesis was fixed with four 6.5-mm cancellous lag screws as described previously [15, 55]. Two screws were inserted parallel from the anterior aspect of the distal tibia into the body of the talus (Fig. 5A–D). The third, mechanically most important, screw was inserted through a posteromedial stab incision approximately 3 cm proximal to the tip of the medial malleolus. The screw was inserted into the anterolateral portion of the talar head. The fourth screw was inserted percutaneously from the posterolateral aspect of the distal fibula approximately 1.5 cm proximal to the tip of the lateral malleolus into the dorsal portion of the talar body. In the presence of greater defects or osteopenic bone, fully threaded 6.5-mm screws were used. If a manifest syndesmotic instability was present at the time of fusion, a fifth screw was introduced as a tibiofibular set screw approximately 1.5 cm above and parallel to the tibial plafond. In four patients with a known or suspected allergy to components of surgical steel (cobalt, chromium, or nickel), we used titanium screws. Correct screw length and realignment of the ankle position was assessed by intraoperative anteroposterior and lateral radiographs of the ankle (Fig. 6A–D). The wound was closed in layers over a suction drain with reapproximation of the extensor retinaculum.

The postoperative management aimed at early functional rehabilitation. The suction drain was typically removed after 2 days. Active and passive motion of the subtalar and midtarsal joints was begun on the second postoperative day while temporarily removing the cast. A split short-leg cast was applied for the period of wound healing only, typically 5 to 9 days. Following that time, 60 of 92 patients were treated postoperatively with an arthrodesis boot with rigid medial and lateral struts and a flexible sole (Variostabil®, Adidas, Herzogenaurach, Germany) allowing early full weight bearing as tolerated, usually after 2 to 3 weeks; otherwise, partial weightbearing of 20 kg was recommended for 6 to 8 weeks, the longer period being necessary after bone grafting. Ten patients were treated with partial weightbearing of 20 kg in their regular shoes. Twenty-two patients were placed in a short-leg cast for 6 weeks with partial weight bearing because of osteopenic bone quality or extensive bone grafting. Weightbearing was increased gradually after removal of the cast. All patients were given daily injections of low-molecular-weight heparin (eg, dalteparin, enoxaparin) until full weight bearing was achieved.

Patients were routinely seen at 6 weeks, and at 3, 6, and 12 months postoperatively for a complete physical examination, to assess bony union and determine weight-bearing status individually as detailed above. Union was verified by obliteration of the joint space and trabecular continuity from the tibia to the talus on lateral and anteroposterior weight-bearing radiographs of the ankle.

Nonunion was defined as radiographic evidence of incomplete fusion more than 9 months after surgery. All postoperative complications were noted. We defined major complications as deep soft tissue infection, osteomyelitis, and malunion or nonunion. Minor complications included superficial wound dehiscence, wound edge necrosis, or hematoma without signs of infection.

Union was achieved in 93 of the 94 patients (99%). One patient with nonunion had repeat arthrodesis with bone grafting 9 months after surgery and healed uneventfully.

At final followup at an average of 5.9 years, the mean AOFAS ankle/hindfoot score [29] had increased to 85 from a mean preoperative value of 36. Sixty-nine of 72 (96%) patients stated that given the same circumstances as before the arthrodesis they would undergo surgery again.

The mean sagittal arc of motion at the midtarsal (Chopart) joint as measured radiographically was 24° (standard deviation 16°).

At last followup, radiographic signs of arthritis had developed or progressed in 25 of 72 patients (35%) at the subtalar joint and in 13 of 72 patients (18%) at the talonavicular joint. Among those, secondary arthritis at the subtalar joint had developed during the followup period in 12 patients (17%) and in eight patients (11%) at the talonavicular joint. Progression of preexisting arthritis by at least one grade on the Bargon scale [3] was noted in 13 of 43 patients (30%) at the subtalar joint and five of 26 patients (19%) at the talonavicular joint. None of these patients had fusion of either of these joints.

Readers should be aware of the limitations of our study. First is the medium-term followup. While a longer followup period would have been desirable, long-term studies necessarily include many patients treated with historical methods that may lead to less satisfactory results [11, 17]. Although there is evidence from one longitudinal study, that long-term outcome (23 years) does not differ from medium-term outcome (7 years) in the same patient population [51], arthritis at the adjacent joints may develop or further progress over time. Second is the loss to followup in 22% of patients (20 of 92). Loss to followup is unavoidable with many patients being referred to our center from distant locations. Nonetheless, we report one of the largest cohorts in the literature treated with a single method of fixation (Table 2).

With the reported technique, using an anterior midline approach and a four-screw technique with two screws introduced percutaneously, only minor complications occurred in 12% patients. All went on to uneventful healing. The still frequently cited historic complication rates after ankle arthrodesis have varied considerably between 9% and 60%. They stem from earlier studies in the 1960s and ’70s that reported ankle fusion with external fixation and included infected cases at the time of fusion [25, 28, 33, 35, 41]. Recent studies with open and percutaneous techniques of internal fixation report substantially lower complication rates that compare with our study (Table 2).

Union rate with our technique was 99% which is in keeping with the recent literature with union rates ranging from 87% to 100% (Table 2). We agree with other authors that judicious resection of remaining cartilage, sclerotic and nonviable bone while preserving the bone contours results in a highly stable fusion and avoids rotational malalignment [27, 34, 47, 55].

Patients with ankle arthritis usually seek care for pain and disability. Sixty-nine of our 72 patients (96%) stated that given the same circumstances as preoperatively they would undergo surgery again. The functional results as measured with the AOFAS score improved substantially as compared to the preoperative values and compare well to the recent literature (Table 2). Similarly, other authors have reported that between 88% and 98% of patients would undergo ankle arthrodesis again [11, 27]. This probably reflects the patients’ subjective relief of long-standing pain and disability despite residual functional restrictions. This effect has also been observed in corrective fusion of the subtalar joint [39].

Residual function after ankle fusion greatly depends on compensatory motion of the adjacent joints. In the present study, the total sagittal motion of the foot at the midtarsal (Chopart) joint averaged 24°. Several studies have described residual sagittal motion between 13° and 24° over the medium to long term [1, 5, 32, 34, 46] while others have noted stiffness at the subtalar and midtarsal joints, at least in a proportion of patients, after ankle arthrodesis [4, 6, 7, 25, 46]. We concur with Takakura et al. [46] that shorter periods of immobilization (about 6 weeks) after ankle arthrodesis will lead to increased range of motion at the subtalar and midtarsal joints as compared with longer immobilization times of 12 to 15 weeks [5, 6, 42]. The arthrodesis boot used in our study protects the ankle against varus and valgus stress with rigid struts while allowing some sagittal motion of the midfoot with partial or full weight bearing because of the flexible sole. Early weight bearing after 2 to 3 weeks as tolerated in compliant patients without extensive bone grafting and with normal bone quality does not prevent a high rate of union [8]. In addition, we advocate early active and passive motion of the midtarsal and subtalar joints.

Ankle fusion puts greater strain on the adjacent joints which may lead to the development or progression of arthritis [7, 26, 32]. During the followup period, we observed development of arthritis at the subtalar joint in 16.6% and at the talonavicular joint in 11.1% of patients and progression of preexisting arthritis 30% and 19%, respectively. Our observations confirm those of Sheridan et al. [43] who reported a high percentage (96%) of subtalar and midfoot arthritis was preexisting due to ankle malfunction in end-stage arthritis. Progression of preexisting arthritis has been reported in between 0 and 70% [5, 27, 30, 48]. Several long-term studies have reported high overall rates of subtalar arthritis of between 47% and 100% [4, 6, 11, 17]. All of these studies included a substantial portion of ankles that were fused in either varus/valgus or equinus. There is consensus in the literature that a neutral position in the sagittal plane and neutral to slight valgus in the coronal plane is essential for a satisfactory functional outcome and puts less strain on adjacent joints including the knee [1, 7, 32, 52]. Conti and Walter observed a correlation of arthritic changes at the midfoot and fusion in plantar flexion [13].

The clinical relevance of radiographic signs of arthritis at the subtalar and talonavicular joints is not well-established [37]. While some authors [6, 17] noted a worse clinical outcome in the presence of subtalar arthritis, patient satisfaction was still high in these studies. Trichard et al. [51] reported less than 1/3, Kennedy et al. [27] found one in seven, and Ben Amor et al. [4] found none of them clinically relevant. Consequently, only a small proportion of the patients with radiographic subtalar arthritis [27, 51]—and none of our patients—had an additional fusion of either of these joints.

Several clinical studies have reported a correlation between radiographic alignment of the fused ankle and functional outcome [1, 5, 6, 13, 35, 46]. We believe that the favorable outcomes in our study are partly attributable to alignment. The one patient with residual equinus of 10° had a low outcome score and another patient with loss of correction was relatively pain-free probably because of his neuropathy. We presumed the best position for ankle arthrodesis was neutral flexion of the foot in the sagittal plane, neutral to slight valgus (5°) in the coronal plane, and 5° to 10° external rotation in the horizontal plane with the talus centered exactly below the tibia [7, 22, 32, 48]. These recommendations from clinical outcome studies could be substantiated by a wealth of studies employing further measurements like gait analyses and pressure measurements of the subtalar joint [7, 15, 34, 47, 52, 55]. It is assumed that less favorable outcomes and high rates of subtalar arthritis in historical and long-term studies result from fusion in malalignment [1, 4–7, 11, 17, 32].