587045
research-article2015
FAIXXX10.1177/1071100715587045Foot & Ankle InternationalChalayon et al
Article
Factors Affecting the Outcomes of Uncomplicated Primary Open Ankle Arthrodesis
Foot & Ankle International® 1–10 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1071100715587045 fai.sagepub.com
Ornusa Chalayon, MD1, Bibo Wang, MD1,2, Brad Blankenhorn, MD3, J. Benjamin Jackson III, MD4, Timothy Beals, MD1, Florian Nickisch, MD1, and Charles L. Saltzman, MD1
Abstract Background: The objective of this study was to identify factors influencing operative outcomes in straightforward, uncomplicated open ankle fusions. Methods: We reviewed all primary open ankle fusions conducted at 1 institution over an 11-year period to identify straightforward, uncomplicated open ankle fusions. Inclusion required a minimum of 6 months follow-up. Patients were excluded for neuropathic arthropathy, insensate limb, failed total ankle replacement, simultaneous arthrodesis of the subtalar joint, or fusions performed within 1 year of injury to salvage failed fixation and painful function due to (1) open fractures, (2) segmental bone loss greater than 1 cm, (3) infection, or (4) talar body fractures. The primary outcome variable was radiographic union at 6 months. Other operative complications were analyzed as secondary outcomes. Five hundred twenty-eight ankle fusion surgeries were performed on 440 patients at 1 institution during the study period. Two hundred fifteen surgeries met inclusion/exclusion eligibility criteria for uncomplicated open ankle fusions. Results: The overall union rate was 91%. In this cohort of uncomplicated open ankle fusions, bivariate analysis over a broad range of potential factors and further focused multivariate analysis found that nonunion was more than 3 times more likely to occur after previous subtalar fusion, and 2 times more likely to occur in patients with preoperative varus ankle alignment. The rate of reoperation was 19%, with nonunion revision as the leading reason, followed by hardware removal and incision and drainage for presumed infection. Diabetes was not a significant risk factor of either deep or superficial infection. Conclusion: Open ankle fusion failed in 9% of uncomplicated ankles with arthritis. Patients who had an open ankle fusion done after previous subtalar joint fusion, as well as those who had preoperative varus ankle alignment, had a significantly higher rate of nonunion. Level of Evidence: Level III, retrospective comparative study. Keywords: ankle arthrodesis, ankle fusion, ankle nonunion, subtalar fusion
Introduction Ankle arthrodesis, or fusion, is a long-standing successful operative treatment for end stage ankle arthritis. Patients typically have excellent pain relief with minimal functional impairment after an ankle fusion.13,26 Although an ankle arthrodesis is an excellent procedure, complications such as nonunion, malunion, wound complication, or infection can arise.7,9,23,25 Ankle arthritis has many different manifestations, and therefore, many different techniques have been developed to perform an arthrodesis. Techniques vary in their operative approach (lateral, anterior, posterior, arthroscopic, or miniopen) and the means of bony fixation and compression
(screw, plate, or external fixator). Ultimately, the choice of operative approach and fixation is individualized to the 1
University Orthopaedic Center, University of Utah, Salt Lake City, Utah, USA 2 Shanghai Institute of Traumatology and Orthopaedics, Orthopaedic Department, Shanghai Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China 3 University of New Mexico, Albuquerque, New Mexico, USA 4 University of South Carolina, Columbia, South Carolina, USA Corresponding Author: Charles L. Saltzman, MD, University Orthopaedic Center, University of Utah, 590 Wakara Way, Salt Lake City, UT 84108, USA. Email: [email protected]
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
2
Foot & Ankle International
patient. In addition to the operative technique, other patientrelated factors, such as smoking, neuropathy, skin conditions, diabetes, previous or current infection, underlying cause of arthritis, and previous or concurrent adjacent joint fusions, can influence the complication rate after an ankle fusion. Several previous retrospective studies have investigated ankle fusions to determine risk factors for nonunion and other complication rates. Mann and Rongstad19 investigated 81 ankle fusions performed through a lateral approach. This study included primary fusions as well as revision procedures and showed a nonunion rate of 12%. The amount of medial malleolar resection was identified as a risk factor for developing a nonunion; however, no other risk factors were analyzed. In another study, Perlman and Thordarson23 identified a history of open trauma as a significant risk factor for the development of a nonunion in a study looking at 67 ankle fusions stabilized with cancellous screws. A history of diabetes, smoking, alcohol use, psychiatric disorders, and illegal drug use trended toward significance but did not reach statistical significance. A study by Dohm et al7 compared different operative techniques for ankle fusion. In this study, the authors compared 37 ankle fusions completed with one of the following techniques: Charnley external fixation, T-plate fixation, crossed screws, and a Royal Air Force procedure (ankle fusion using a long fibular onlay autograft stabilized with several screws from the fibula to the tibia). They concluded that rigid internal fixation with a T-plate was the best method for obtaining primary union. It is difficult to correctly identify risk factors for complications from these studies since complications for ankle fusions are relatively rare and all of these studies included a relatively small sample size. Diagnosis of a nonunion after an attempted ankle fusion can be difficult to define. A nonunion can be determined based on clinical evaluation including the need for revision surgery, plain radiographs, or computed tomography (CT) scans. Several studies have compared the ability of plain radiographs and CT scans to diagnose nonunions after hindfoot fusion.5,10,16 These studies showed that CT scans can be more accurate at diagnosing a nonunion than plain radiographs and that 25% to 49% of osseous bridging on CT scan across a fusion site is enough to result in an improvement in clinical outcome. The most common perioperative complication of open ankle arthrodesis is nonunion. The implications for the patient with nonunion after an attempted ankle fusion are always substantial, often requiring revision surgery, autologous bone grafting, and a second prolonged period of nonweightbearing, immobilization, and loss of work.8 To better understand the risk factors related to this, we undertook the current retrospective investigation of uncomplicated, primary open ankle arthrodesis. The goal of this study was to develop an understanding of the rate and risk factors of nonunion in uncomplicated, open ankle fusions.
Table 1. Number of Cases Excluded From Study by Exclusion Criteria. Exclusion Criteria Arthroscopic fusion Neuropathic arthropathy Failed total ankle replacement Simultaneous subtalar fusion Fusions done less than 1 year of trauma with infection, segmental bone loss, failed fixation of talar body fractures Revision surgeries Insufficient images
No. of Cases 69 46 8 39 19 40 92
Methods This retrospective study reviewed all primary open ankle fusions conducted at 1 institution between March 30, 2002, and January 14, 2013. After the protocol was approved by the institutional review board (University of Utah ID 00058900), the electronic charts of the patients who had a billing code of open ankle arthrodesis were reviewed. Patients were excluded if they had any of the following potential complicating factors: neuropathic arthropathy, failed total ankle replacement, simultaneous subtalar arthrodesis, arthroscopic fusion, revision ankle fusion, ankle fusions done less than 1 year after trauma with persistent infection, segmental bone loss greater than 1 cm, or failed fixation of talar body fractures. Only open ankle fusions with a minimum 6 months of follow-up X-rays were included. There were 528 ankle fusion surgeries performed on 440 patients. Within this population, 215 open ankle fusion surgeries in 209 patients matched the inclusion and exclusion criteria and remained for data analysis (Table 1). The primary outcome variable was a presence or absence of radiographic union by 6 months after surgery, defined as bridging trabeculae between the tibia and the talus apparent on 2 orthogonal views agreed on by 2 independent orthopedic surgeons. In addition, we reviewed the written report of a fellowship-trained musculoskeletal radiologist. When there was disagreement between any of the 3 assessors, we considered the ankle not to be fused. To ensure the correct outcome of successful fusion, the charts of all patients were reviewed until the most recent clinic visit. The secondary outcomes of interest were postoperative complications. Independent variables collected on each patient were age, sex, weight, height, diagnosis of diabetes, cardiovascular disease (including hypertension), history of smoking, illicit drug use, alcohol consumption, etiology of arthritis, preoperative ankle condition (bone loss, osteoporosis, infection, presence of a subtalar fusion, and clinical ankle alignment), and the technique used for fusion.
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
3
Chalayon et al
Figure 1. Measurement of tibial-calcaneus axis angle: angle between tibial axis (a line through the center of 2 circles drawn 8 cm and 13 cm above tibial plafond) and calcaneal axis (a line parallel to the lateral wall of calcaneus) using the hindfoot alignment view. Positive value determines valgus alignment.
In addition to the clinical observation of the ankle alignment, the preoperative radiographs were measured for (1) tibial-calcaneus axis angle, (2) apparent moment arm, and (3) talar tilting angle. The tibial-calcaneal axis angle was measured on the hindfoot alignment view (Figure 1). This is an angle between the tibial axis (a line through the center of 2 circles drawn 8 cm and 13 cm above tibial plafond) and the calcaneal axis (a line parallel to the lateral wall of calcaneus).27 The apparent moment arm was also measured using the hindfoot alignment view, by drawing a line from the lowest point of the calcaneus perpendicular to the axis of the tibia as previously described24 (Figure 2). The talar tilting angle was calculated from the difference between the medial distal tibial angle (MDTA)28 and the angle between the tibial axis and the superior surface of the talar dome measured on the ankle mortise view (Figure 3).
Statistical Analysis The aim of the statistical analysis was to identify which factors were predictive of nonunion and, in a separate analysis, which factors were predictive of a complication of treatment, such as infection. Each predictor variable was first analyzed individually. Significant variables associated with outcomes of interest were then tested using a multivariable mixed effects Poisson regression model. The mixed effects modeling approach correctly adjusts the standard errors in the model for lack of independence by having 2 surgeries performed on some patients, which happened in 6 patients. All surgeries were primary surgeries, so there was no need to account for prior surgeries on the same ankle. In the nonunion outcome model, given the large number of predictor variables with a relatively low number of events
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
4
Foot & Ankle International
Figure 2. Measurement of apparent moment arm as described by Saltzman and el-Khoury.24 On the hindfoot alignment view, a line was drawn from the lowest point of the calcaneus perpendicular to the tibial axis. Positive values were defined as valgus alignment.
(20 nonunions), an interactive forward variable selection was done, with a restrictive inclusion criteria, to help mitigate potential overfitting. “Overfitting” is the situation of identifying unreliable associations from having too many predictor variables for the number of nonunion events in the outcome variable. In the univariable analysis, there were 4 risk factors with P ≤ .06. One of these, underweight body mass index (BMI), had only 1 observation in that category, making it an unreliable predictor variable, so it was not considered for inclusion in the multivariable model. Given that there were few events for most complications, complications are reported descriptively. For postoperative infections, however, there were enough occurrences to fit a univariable mixed effects Poisson model, but not enough for a multivariable model, without introducing overfitting. All reported P values are for a 2-sided comparison. Stata statistical software, version 12.1 (StataCorp, College Station, TX, USA), was used for the statistical analysis.
Results With the total number of 209 patients, the characteristics of the studied group are summarized in Table 2. When evaluated at 6 months postoperatively, 20 of 215 surgeries (9%) had an ankle nonunion (Table 3). Each factor was analyzed for a risk of nonunion, as shown in Table 4. Based on the univariate analysis, the predictor variables selected for the multivariable analysis were illicit drug use, preoperative varus ankle alignment, and history of subtalar fusion. The associated risks for nonunion for these factors are shown in Table 5. In patients who previously underwent subtalar fusion, nonunion was more than 3 times more likely to occur (risk ratio [RR] = 3.89; 95% confidence interval [CI], 1.25-12.15; P = .02). The incidence was doubled in patients with preoperative varus ankle (RR = 2.61; 95% CI, 1.05-6.52; P = .04). Illicit drug use was found to not significantly increase risk of nonunion (RR = 3.91; 95% CI, 0.88-17.37; P = .07), although the numbers of patients in this subgroup were small.
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
5
Chalayon et al
Figure 3. Measurement of talar tilting angle obtained from the ankle mortise view by first drawing the medial distal tibial angle between the axis and distal articular surface of the tibia and then measuring the angle between the tibial axis and the superior surface of the talar dome. The talar tilting ankle was calculated as the difference between these 2 angles. Positive value determines valgus alignment.
Complications Nineteen percent of the surgeries (40 of 215) had postoperative complications that caused unplanned return to the operating room (Table 6). The most common operations were revision of fusion (16 cases). Among these cases, 15 were reoperated because of nonunion, and 1 case was revised soon after the initial surgery due to malalignment. The next most common cause of reoperation was painful hardware requiring removal (12 cases). The third most common operation was irrigation and debridement for presumed infection (11 cases), of which 5 already were suspected to have an infection preoperatively. One patient on anticoagulation sustained a fall, resulting in a hematoma that required operative evacuation. Regarding an association between postoperative infection and diabetes mellitus (DM), among 26 patients with DM, 2 (8%) were suspected to have a deep infection that was treated operatively. The rate of superficial infection that resolved with outpatient care was 15% (4 of 26 diabetic patients). Diabetes was not a significant risk factor for
either deep or superficial infection, with a relative risk of 1.62 (95% CI, 0.35-7.48; P = .54) and 1.12 (95% CI, 0.393.20; P = .84), for deep and superficial infection, respectively. For presumed infection treated operatively, smoking had a relative risk of 11.36 (95% CI, 0.73-177.32) with a marginal significance of P = .08. Smoking also elevated the risk of ultrasound-proven venous thromboembolism—5 times the risk when compared to the controls (95% CI, 0.92-31.16; P = .04)
Discussion In this case series of uncomplicated primary open ankle arthrodesis, 20 of 215 surgeries (9%) had no evidence of union 6 months postoperatively. For the purposes of this analysis, they were considered to have a nonunion. We chose 6 months as the critical time-point because in our clinical practice, patients with symptomatic ankle fusions without any evidence of bony bridging are offered the option of a revision fusion procedure at 6 months. However, if the nonunion was diagnosed after 6 months,
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
6
Foot & Ankle International
Table 2. Patient Characteristics for the Study Population of 209 Patients. No. Age Sex Male Female Diabetes Cardiovascular disease and hypertension Smoker Alcohol use Illicit drug use Weight, kg BMI Underweight (BMI < 18.5) Normal (BMI 18.5-25) Overweight (BMI 25.0-29.9) Obese (BMI 30-34.9) Morbidly obese (BMI ≥ 35) Etiology Nontrauma Posttraumatic ASA class 1 2 3
%
Mean ± SD
209
56 ± 14 (min 18, max 88)
112 97 26 101
54 46 12 48
45 76 6 204
22 36 3
90 ± 21 (min 46, max 168)
1 46 55 55 47
1 y after trauma) 154 6 Early fusion (occurs within 1 y after trauma) Weight, per kg increase BMI Underweight (BMI < 18.5) 1 Normal (BMI 18.5-25) 47 Overweight (BMI 25.0-29.9) 56 Obese (BMI 30-34.9) 58 Morbidly obese (BMI ≥ 35) 48 ASA class, per score increase Preop. bone loss 53 Preop. osteoporosis 64 Preop. infection 8 Previous subtalar fusion 15 Preop. alignment (clinical) Neutral 79 70 Varusa 55 Valgusa Unknown 11 Moment arm (min/max), mm Varus (–86.3/–10) 37 Neutral (–10/10) 46 Valgus (10/47) 31 Tibial-calcaneus axis angle (min/max), degrees Varus (–54/–10) 54 Neutral 47 Valgus (10/31) 13 Talar tilting angle (min/max), degrees Varus (–77/–10) 40 Neutral (–10/10) 132 Valgus (10/37) 30 Absolute value of talar tilting Angle, degrees < 10 132 10-19.9 34 > 20 36 Operative factors Distal syndesmosis fusion included 153
No. With Nonunion (%)
Risk Ratio
95% CI
P Value
8 (8) 2 (8) 9 (9) 5 (11) 6 (8) 2 (33)
1.00 0.75 0.79 0.89 1.21 0.75 5.31
0.97-1.03 0.29-1.91 0.17-3.63 0.37-2.15 0.42-3.53 0.29-1.96 0.91-31.00
.96 .54 .76 .79 .72 .56 .06
8 (15) 12 (8)
[Referent] 0.52
0.21-1.26
.15
11 (7) 1 (17)
[Referent] 2.33 1.00
0.30-18.08 0.98-1.02
.42 .94
1 (100) 4 (9) 4 (7) 5 (9) 5 (10)
11.75 [Referent] 0.84 1.01 1.22 1.08 1.31 1.93 1.36 4.18
0.21-3.36 0.27-3.77 0.33-4.56 0.46-2.51 0.50-3.41 0.80-4.66 0.18-10.17 1.19-14.65
3 (4) 10 (14) 6 (11) 1 (9)
[Referent] 4.22 3.10
1.11-16.02 0.74-12.98
2 (5) 5 (11) 5 (16)
0.50 [Referent] 1.48
5 (9) 4 (9) 3 (23)
1.18 [Referent] 11.29
7 (18) 10 (8) 2 (7)
2.31 [Referent] 0.88
0.19-4.02
.09 .87
10 (8) 5 (15) 4 (11)
[Referent] 1.94 1.47
0.66-5.68 0.46-4.68
.23 .52
12 (8)
0.61
0.25-1.49
.28
6 (11) 9 (14) 1 (13) 4 (27)
1.31-105.14
0.10-2.56 0.43-5.13 0.09-15.30 0.32-396.03 0.88-6.07
.03 .80 .98 .76 .87 .58 .14 .76 .03 .03 .12 .40 .53 .90 .18
(continued)
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
8
Foot & Ankle International
Table 4. (continued) Risk Factor Approach Open lateral malleolus fused as onlay Open lateral malleolus removed Anterior approach Posterior approach Mini open Implant Screws without plate Lateral plate Anterior plate Posterior plate External fixator Bone graft/substitute None Any bone graft
No. of Surgeries
No. With Nonunion (%)
Risk Ratio
95% CI
P Value
141 10 47 10 7
11 (8) 2 (20) 4 (9) 1 (10) 2 (29)
[Referent] 2.56 1.09 1.28 3.66
0.57-11.56 0.35-3.43 0.17-9.93 0.81-16.52
.22 .88 .81 .09
147 7 47 10 8
14 (10) 0 (0) 3 (6) 1 (10) 2 (25)
[Referent] 0.62 1.06 2.82
0.18-2.11 0.14-8.00 0.65-12.22
.44 .95 .17
8 207
1 (13) 19 (9)
1.41 [Referent]
0.16-12.11
b
.75
Abbreviations: ASA class, American Society of Anesthesiologists Physical Status classification; BMI, body mass index; CI, confidence interval; Preop., preoperative. a Risk ratio calculated after elimination of patients with “unknown” alignment. b All patients receiving these treatments had successful union, so risk ratio could not be estimated.
Table 5. Variables Found to be Highly Associated With Nonunion of Fusion by the Univariate Analysis (P < .20) Tested Using a Multivariable Mixed Effect Poisson Regression Model.a Risk Factor
Risk Ratio
Illicit drug use Preoperative varus ankle alignment Previous subtalar fusion
95% CI
3.91 2.61 3.89
P Value
0.88-17.37 1.05-6.52 1.25-12.15
.07 .04 .02
Abbreviation: CI, confidence interval. a Preoperative varus alignment and previous subtalar fusion remained as significant factors (P < .05) that increased the risk of nonunion, whereas illicit drug use was not significant.
Table 6. Postoperative Complications of 215 Cases of Open Ankle Arthrodesis. Complication Return to the operating room Incision and drainage for presumed infection Revision of fusion Hardware removal Other wound complication Wound complication treated in clinic Venous thromboembolism Amputation Malalignment Fracture Nerve injury
present, which can potentially increase the relative risks of identified risk factors larger than they actually were. We did not include the clinical description of the amount of pain that patients were having at 6 months to corroborate the
No. of Cases
% of 215 Surgeries
11 16 12 1 30 5 1 7 2 3
5 7 6
research-article2015
FAIXXX10.1177/1071100715587045Foot & Ankle InternationalChalayon et al
Article
Factors Affecting the Outcomes of Uncomplicated Primary Open Ankle Arthrodesis
Foot & Ankle International® 1–10 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1071100715587045 fai.sagepub.com
Ornusa Chalayon, MD1, Bibo Wang, MD1,2, Brad Blankenhorn, MD3, J. Benjamin Jackson III, MD4, Timothy Beals, MD1, Florian Nickisch, MD1, and Charles L. Saltzman, MD1
Abstract Background: The objective of this study was to identify factors influencing operative outcomes in straightforward, uncomplicated open ankle fusions. Methods: We reviewed all primary open ankle fusions conducted at 1 institution over an 11-year period to identify straightforward, uncomplicated open ankle fusions. Inclusion required a minimum of 6 months follow-up. Patients were excluded for neuropathic arthropathy, insensate limb, failed total ankle replacement, simultaneous arthrodesis of the subtalar joint, or fusions performed within 1 year of injury to salvage failed fixation and painful function due to (1) open fractures, (2) segmental bone loss greater than 1 cm, (3) infection, or (4) talar body fractures. The primary outcome variable was radiographic union at 6 months. Other operative complications were analyzed as secondary outcomes. Five hundred twenty-eight ankle fusion surgeries were performed on 440 patients at 1 institution during the study period. Two hundred fifteen surgeries met inclusion/exclusion eligibility criteria for uncomplicated open ankle fusions. Results: The overall union rate was 91%. In this cohort of uncomplicated open ankle fusions, bivariate analysis over a broad range of potential factors and further focused multivariate analysis found that nonunion was more than 3 times more likely to occur after previous subtalar fusion, and 2 times more likely to occur in patients with preoperative varus ankle alignment. The rate of reoperation was 19%, with nonunion revision as the leading reason, followed by hardware removal and incision and drainage for presumed infection. Diabetes was not a significant risk factor of either deep or superficial infection. Conclusion: Open ankle fusion failed in 9% of uncomplicated ankles with arthritis. Patients who had an open ankle fusion done after previous subtalar joint fusion, as well as those who had preoperative varus ankle alignment, had a significantly higher rate of nonunion. Level of Evidence: Level III, retrospective comparative study. Keywords: ankle arthrodesis, ankle fusion, ankle nonunion, subtalar fusion
Introduction Ankle arthrodesis, or fusion, is a long-standing successful operative treatment for end stage ankle arthritis. Patients typically have excellent pain relief with minimal functional impairment after an ankle fusion.13,26 Although an ankle arthrodesis is an excellent procedure, complications such as nonunion, malunion, wound complication, or infection can arise.7,9,23,25 Ankle arthritis has many different manifestations, and therefore, many different techniques have been developed to perform an arthrodesis. Techniques vary in their operative approach (lateral, anterior, posterior, arthroscopic, or miniopen) and the means of bony fixation and compression
(screw, plate, or external fixator). Ultimately, the choice of operative approach and fixation is individualized to the 1
University Orthopaedic Center, University of Utah, Salt Lake City, Utah, USA 2 Shanghai Institute of Traumatology and Orthopaedics, Orthopaedic Department, Shanghai Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China 3 University of New Mexico, Albuquerque, New Mexico, USA 4 University of South Carolina, Columbia, South Carolina, USA Corresponding Author: Charles L. Saltzman, MD, University Orthopaedic Center, University of Utah, 590 Wakara Way, Salt Lake City, UT 84108, USA. Email: [email protected]
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
2
Foot & Ankle International
patient. In addition to the operative technique, other patientrelated factors, such as smoking, neuropathy, skin conditions, diabetes, previous or current infection, underlying cause of arthritis, and previous or concurrent adjacent joint fusions, can influence the complication rate after an ankle fusion. Several previous retrospective studies have investigated ankle fusions to determine risk factors for nonunion and other complication rates. Mann and Rongstad19 investigated 81 ankle fusions performed through a lateral approach. This study included primary fusions as well as revision procedures and showed a nonunion rate of 12%. The amount of medial malleolar resection was identified as a risk factor for developing a nonunion; however, no other risk factors were analyzed. In another study, Perlman and Thordarson23 identified a history of open trauma as a significant risk factor for the development of a nonunion in a study looking at 67 ankle fusions stabilized with cancellous screws. A history of diabetes, smoking, alcohol use, psychiatric disorders, and illegal drug use trended toward significance but did not reach statistical significance. A study by Dohm et al7 compared different operative techniques for ankle fusion. In this study, the authors compared 37 ankle fusions completed with one of the following techniques: Charnley external fixation, T-plate fixation, crossed screws, and a Royal Air Force procedure (ankle fusion using a long fibular onlay autograft stabilized with several screws from the fibula to the tibia). They concluded that rigid internal fixation with a T-plate was the best method for obtaining primary union. It is difficult to correctly identify risk factors for complications from these studies since complications for ankle fusions are relatively rare and all of these studies included a relatively small sample size. Diagnosis of a nonunion after an attempted ankle fusion can be difficult to define. A nonunion can be determined based on clinical evaluation including the need for revision surgery, plain radiographs, or computed tomography (CT) scans. Several studies have compared the ability of plain radiographs and CT scans to diagnose nonunions after hindfoot fusion.5,10,16 These studies showed that CT scans can be more accurate at diagnosing a nonunion than plain radiographs and that 25% to 49% of osseous bridging on CT scan across a fusion site is enough to result in an improvement in clinical outcome. The most common perioperative complication of open ankle arthrodesis is nonunion. The implications for the patient with nonunion after an attempted ankle fusion are always substantial, often requiring revision surgery, autologous bone grafting, and a second prolonged period of nonweightbearing, immobilization, and loss of work.8 To better understand the risk factors related to this, we undertook the current retrospective investigation of uncomplicated, primary open ankle arthrodesis. The goal of this study was to develop an understanding of the rate and risk factors of nonunion in uncomplicated, open ankle fusions.
Table 1. Number of Cases Excluded From Study by Exclusion Criteria. Exclusion Criteria Arthroscopic fusion Neuropathic arthropathy Failed total ankle replacement Simultaneous subtalar fusion Fusions done less than 1 year of trauma with infection, segmental bone loss, failed fixation of talar body fractures Revision surgeries Insufficient images
No. of Cases 69 46 8 39 19 40 92
Methods This retrospective study reviewed all primary open ankle fusions conducted at 1 institution between March 30, 2002, and January 14, 2013. After the protocol was approved by the institutional review board (University of Utah ID 00058900), the electronic charts of the patients who had a billing code of open ankle arthrodesis were reviewed. Patients were excluded if they had any of the following potential complicating factors: neuropathic arthropathy, failed total ankle replacement, simultaneous subtalar arthrodesis, arthroscopic fusion, revision ankle fusion, ankle fusions done less than 1 year after trauma with persistent infection, segmental bone loss greater than 1 cm, or failed fixation of talar body fractures. Only open ankle fusions with a minimum 6 months of follow-up X-rays were included. There were 528 ankle fusion surgeries performed on 440 patients. Within this population, 215 open ankle fusion surgeries in 209 patients matched the inclusion and exclusion criteria and remained for data analysis (Table 1). The primary outcome variable was a presence or absence of radiographic union by 6 months after surgery, defined as bridging trabeculae between the tibia and the talus apparent on 2 orthogonal views agreed on by 2 independent orthopedic surgeons. In addition, we reviewed the written report of a fellowship-trained musculoskeletal radiologist. When there was disagreement between any of the 3 assessors, we considered the ankle not to be fused. To ensure the correct outcome of successful fusion, the charts of all patients were reviewed until the most recent clinic visit. The secondary outcomes of interest were postoperative complications. Independent variables collected on each patient were age, sex, weight, height, diagnosis of diabetes, cardiovascular disease (including hypertension), history of smoking, illicit drug use, alcohol consumption, etiology of arthritis, preoperative ankle condition (bone loss, osteoporosis, infection, presence of a subtalar fusion, and clinical ankle alignment), and the technique used for fusion.
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
3
Chalayon et al
Figure 1. Measurement of tibial-calcaneus axis angle: angle between tibial axis (a line through the center of 2 circles drawn 8 cm and 13 cm above tibial plafond) and calcaneal axis (a line parallel to the lateral wall of calcaneus) using the hindfoot alignment view. Positive value determines valgus alignment.
In addition to the clinical observation of the ankle alignment, the preoperative radiographs were measured for (1) tibial-calcaneus axis angle, (2) apparent moment arm, and (3) talar tilting angle. The tibial-calcaneal axis angle was measured on the hindfoot alignment view (Figure 1). This is an angle between the tibial axis (a line through the center of 2 circles drawn 8 cm and 13 cm above tibial plafond) and the calcaneal axis (a line parallel to the lateral wall of calcaneus).27 The apparent moment arm was also measured using the hindfoot alignment view, by drawing a line from the lowest point of the calcaneus perpendicular to the axis of the tibia as previously described24 (Figure 2). The talar tilting angle was calculated from the difference between the medial distal tibial angle (MDTA)28 and the angle between the tibial axis and the superior surface of the talar dome measured on the ankle mortise view (Figure 3).
Statistical Analysis The aim of the statistical analysis was to identify which factors were predictive of nonunion and, in a separate analysis, which factors were predictive of a complication of treatment, such as infection. Each predictor variable was first analyzed individually. Significant variables associated with outcomes of interest were then tested using a multivariable mixed effects Poisson regression model. The mixed effects modeling approach correctly adjusts the standard errors in the model for lack of independence by having 2 surgeries performed on some patients, which happened in 6 patients. All surgeries were primary surgeries, so there was no need to account for prior surgeries on the same ankle. In the nonunion outcome model, given the large number of predictor variables with a relatively low number of events
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
4
Foot & Ankle International
Figure 2. Measurement of apparent moment arm as described by Saltzman and el-Khoury.24 On the hindfoot alignment view, a line was drawn from the lowest point of the calcaneus perpendicular to the tibial axis. Positive values were defined as valgus alignment.
(20 nonunions), an interactive forward variable selection was done, with a restrictive inclusion criteria, to help mitigate potential overfitting. “Overfitting” is the situation of identifying unreliable associations from having too many predictor variables for the number of nonunion events in the outcome variable. In the univariable analysis, there were 4 risk factors with P ≤ .06. One of these, underweight body mass index (BMI), had only 1 observation in that category, making it an unreliable predictor variable, so it was not considered for inclusion in the multivariable model. Given that there were few events for most complications, complications are reported descriptively. For postoperative infections, however, there were enough occurrences to fit a univariable mixed effects Poisson model, but not enough for a multivariable model, without introducing overfitting. All reported P values are for a 2-sided comparison. Stata statistical software, version 12.1 (StataCorp, College Station, TX, USA), was used for the statistical analysis.
Results With the total number of 209 patients, the characteristics of the studied group are summarized in Table 2. When evaluated at 6 months postoperatively, 20 of 215 surgeries (9%) had an ankle nonunion (Table 3). Each factor was analyzed for a risk of nonunion, as shown in Table 4. Based on the univariate analysis, the predictor variables selected for the multivariable analysis were illicit drug use, preoperative varus ankle alignment, and history of subtalar fusion. The associated risks for nonunion for these factors are shown in Table 5. In patients who previously underwent subtalar fusion, nonunion was more than 3 times more likely to occur (risk ratio [RR] = 3.89; 95% confidence interval [CI], 1.25-12.15; P = .02). The incidence was doubled in patients with preoperative varus ankle (RR = 2.61; 95% CI, 1.05-6.52; P = .04). Illicit drug use was found to not significantly increase risk of nonunion (RR = 3.91; 95% CI, 0.88-17.37; P = .07), although the numbers of patients in this subgroup were small.
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
5
Chalayon et al
Figure 3. Measurement of talar tilting angle obtained from the ankle mortise view by first drawing the medial distal tibial angle between the axis and distal articular surface of the tibia and then measuring the angle between the tibial axis and the superior surface of the talar dome. The talar tilting ankle was calculated as the difference between these 2 angles. Positive value determines valgus alignment.
Complications Nineteen percent of the surgeries (40 of 215) had postoperative complications that caused unplanned return to the operating room (Table 6). The most common operations were revision of fusion (16 cases). Among these cases, 15 were reoperated because of nonunion, and 1 case was revised soon after the initial surgery due to malalignment. The next most common cause of reoperation was painful hardware requiring removal (12 cases). The third most common operation was irrigation and debridement for presumed infection (11 cases), of which 5 already were suspected to have an infection preoperatively. One patient on anticoagulation sustained a fall, resulting in a hematoma that required operative evacuation. Regarding an association between postoperative infection and diabetes mellitus (DM), among 26 patients with DM, 2 (8%) were suspected to have a deep infection that was treated operatively. The rate of superficial infection that resolved with outpatient care was 15% (4 of 26 diabetic patients). Diabetes was not a significant risk factor for
either deep or superficial infection, with a relative risk of 1.62 (95% CI, 0.35-7.48; P = .54) and 1.12 (95% CI, 0.393.20; P = .84), for deep and superficial infection, respectively. For presumed infection treated operatively, smoking had a relative risk of 11.36 (95% CI, 0.73-177.32) with a marginal significance of P = .08. Smoking also elevated the risk of ultrasound-proven venous thromboembolism—5 times the risk when compared to the controls (95% CI, 0.92-31.16; P = .04)
Discussion In this case series of uncomplicated primary open ankle arthrodesis, 20 of 215 surgeries (9%) had no evidence of union 6 months postoperatively. For the purposes of this analysis, they were considered to have a nonunion. We chose 6 months as the critical time-point because in our clinical practice, patients with symptomatic ankle fusions without any evidence of bony bridging are offered the option of a revision fusion procedure at 6 months. However, if the nonunion was diagnosed after 6 months,
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
6
Foot & Ankle International
Table 2. Patient Characteristics for the Study Population of 209 Patients. No. Age Sex Male Female Diabetes Cardiovascular disease and hypertension Smoker Alcohol use Illicit drug use Weight, kg BMI Underweight (BMI < 18.5) Normal (BMI 18.5-25) Overweight (BMI 25.0-29.9) Obese (BMI 30-34.9) Morbidly obese (BMI ≥ 35) Etiology Nontrauma Posttraumatic ASA class 1 2 3
%
Mean ± SD
209
56 ± 14 (min 18, max 88)
112 97 26 101
54 46 12 48
45 76 6 204
22 36 3
90 ± 21 (min 46, max 168)
1 46 55 55 47
1 y after trauma) 154 6 Early fusion (occurs within 1 y after trauma) Weight, per kg increase BMI Underweight (BMI < 18.5) 1 Normal (BMI 18.5-25) 47 Overweight (BMI 25.0-29.9) 56 Obese (BMI 30-34.9) 58 Morbidly obese (BMI ≥ 35) 48 ASA class, per score increase Preop. bone loss 53 Preop. osteoporosis 64 Preop. infection 8 Previous subtalar fusion 15 Preop. alignment (clinical) Neutral 79 70 Varusa 55 Valgusa Unknown 11 Moment arm (min/max), mm Varus (–86.3/–10) 37 Neutral (–10/10) 46 Valgus (10/47) 31 Tibial-calcaneus axis angle (min/max), degrees Varus (–54/–10) 54 Neutral 47 Valgus (10/31) 13 Talar tilting angle (min/max), degrees Varus (–77/–10) 40 Neutral (–10/10) 132 Valgus (10/37) 30 Absolute value of talar tilting Angle, degrees < 10 132 10-19.9 34 > 20 36 Operative factors Distal syndesmosis fusion included 153
No. With Nonunion (%)
Risk Ratio
95% CI
P Value
8 (8) 2 (8) 9 (9) 5 (11) 6 (8) 2 (33)
1.00 0.75 0.79 0.89 1.21 0.75 5.31
0.97-1.03 0.29-1.91 0.17-3.63 0.37-2.15 0.42-3.53 0.29-1.96 0.91-31.00
.96 .54 .76 .79 .72 .56 .06
8 (15) 12 (8)
[Referent] 0.52
0.21-1.26
.15
11 (7) 1 (17)
[Referent] 2.33 1.00
0.30-18.08 0.98-1.02
.42 .94
1 (100) 4 (9) 4 (7) 5 (9) 5 (10)
11.75 [Referent] 0.84 1.01 1.22 1.08 1.31 1.93 1.36 4.18
0.21-3.36 0.27-3.77 0.33-4.56 0.46-2.51 0.50-3.41 0.80-4.66 0.18-10.17 1.19-14.65
3 (4) 10 (14) 6 (11) 1 (9)
[Referent] 4.22 3.10
1.11-16.02 0.74-12.98
2 (5) 5 (11) 5 (16)
0.50 [Referent] 1.48
5 (9) 4 (9) 3 (23)
1.18 [Referent] 11.29
7 (18) 10 (8) 2 (7)
2.31 [Referent] 0.88
0.19-4.02
.09 .87
10 (8) 5 (15) 4 (11)
[Referent] 1.94 1.47
0.66-5.68 0.46-4.68
.23 .52
12 (8)
0.61
0.25-1.49
.28
6 (11) 9 (14) 1 (13) 4 (27)
1.31-105.14
0.10-2.56 0.43-5.13 0.09-15.30 0.32-396.03 0.88-6.07
.03 .80 .98 .76 .87 .58 .14 .76 .03 .03 .12 .40 .53 .90 .18
(continued)
Downloaded from fai.sagepub.com at University of Sussex Library on August 22, 2015
8
Foot & Ankle International
Table 4. (continued) Risk Factor Approach Open lateral malleolus fused as onlay Open lateral malleolus removed Anterior approach Posterior approach Mini open Implant Screws without plate Lateral plate Anterior plate Posterior plate External fixator Bone graft/substitute None Any bone graft
No. of Surgeries
No. With Nonunion (%)
Risk Ratio
95% CI
P Value
141 10 47 10 7
11 (8) 2 (20) 4 (9) 1 (10) 2 (29)
[Referent] 2.56 1.09 1.28 3.66
0.57-11.56 0.35-3.43 0.17-9.93 0.81-16.52
.22 .88 .81 .09
147 7 47 10 8
14 (10) 0 (0) 3 (6) 1 (10) 2 (25)
[Referent] 0.62 1.06 2.82
0.18-2.11 0.14-8.00 0.65-12.22
.44 .95 .17
8 207
1 (13) 19 (9)
1.41 [Referent]
0.16-12.11
b
.75
Abbreviations: ASA class, American Society of Anesthesiologists Physical Status classification; BMI, body mass index; CI, confidence interval; Preop., preoperative. a Risk ratio calculated after elimination of patients with “unknown” alignment. b All patients receiving these treatments had successful union, so risk ratio could not be estimated.
Table 5. Variables Found to be Highly Associated With Nonunion of Fusion by the Univariate Analysis (P < .20) Tested Using a Multivariable Mixed Effect Poisson Regression Model.a Risk Factor
Risk Ratio
Illicit drug use Preoperative varus ankle alignment Previous subtalar fusion
95% CI
3.91 2.61 3.89
P Value
0.88-17.37 1.05-6.52 1.25-12.15
.07 .04 .02
Abbreviation: CI, confidence interval. a Preoperative varus alignment and previous subtalar fusion remained as significant factors (P < .05) that increased the risk of nonunion, whereas illicit drug use was not significant.
Table 6. Postoperative Complications of 215 Cases of Open Ankle Arthrodesis. Complication Return to the operating room Incision and drainage for presumed infection Revision of fusion Hardware removal Other wound complication Wound complication treated in clinic Venous thromboembolism Amputation Malalignment Fracture Nerve injury
present, which can potentially increase the relative risks of identified risk factors larger than they actually were. We did not include the clinical description of the amount of pain that patients were having at 6 months to corroborate the
No. of Cases
% of 215 Surgeries
11 16 12 1 30 5 1 7 2 3
5 7 6
