ORIGINAL ARTICLE
Standard Perioperative Imaging Modalities Are Unreliable in Assessing Articular Congruity of Ankle Fractures Matthew R. Garner, MD,* Peter D. Fabricant, MD, MPH,* Patrick C. Schottel, MD,* Marschall B. Berkes, MD,* Andre D. Shaffer, MD,* Amelia Ni, BA,† and Dean G. Lorich, MD*†
Objectives: To determine the sensitivity, specificity, and interobserver and intraobserver reliabilities of intraoperative fluoroscopy and postoperative plain radiographs (XR) in the assessment of articular congruency after open reduction and internal fixation (ORIF) of ankle fractures involving the tibial plafond.
Design: Retrospective cohort. Setting: Academic level 1 trauma center. Patients/Participants: One hundred five patients treated surgically for rotational ankle fractures. Intervention: ORIF. Main Outcome Measurements: Sensitivity, specificity, and interobserver and intraobserver reliabilities of fluoroscopy and plain radiographs when compared with computed tomography imaging. Results: The sensitivities of fluoroscopy and XR were 21% and 36%, respectively. Specificities were 95% (fluoroscopy) and 89% (XR). Fluoroscopy interobserver reliability was k = 0.15, and mean intraobserver reliability was k = 0.32. XR interobserver and mean intraobserver reliabilities were k = 0.30 and k = 0.59. Conclusions: Although results show acceptable specificity, the reliability and sensitivity of both intraoperative fluoroscopy and postoperative XR in the assessment of ankle articular congruency are low. This calls into question available literature correlating clinical results with articular reduction. During ORIF of an intra-articular ankle fracture, surgeons should be highly critical of fluoroscopic imaging that seems adequately reduced and direct visualization of the articular surface should be used as a reduction aid if possible. Furthermore, in the postoperative period, axial imaging may be warranted in patients who have poor clinical outcomes despite apparent anatomic articular reduction to evaluate for occult joint incongruence. Key Words: ankle fracture, surgical fixation of ankle fracture, ankle fracture imaging, trauma (J Orthop Trauma 2015;29:e161–e165) Accepted for publication July 25, 2014. From the *Hospital for Special Surgery, New York, NY; and †Department of Orthopaedics, Weill Cornell Medical College, New York, NY. The authors report no conflict of interest. Presented in part at the Annual Meeting of the American Academy of Orthopaedic Surgeons, March 20–23, 2013, Chicago, IL. Reprints: Matthew R. Garner, MD, Hospital for Special Surgery, 535 E. 70th St, New York, NY 10021 (e-mail: [email protected]). Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
J Orthop Trauma Volume 29, Number 4, April 2015
INTRODUCTION The ankle joint is the most commonly injured weightbearing joint, and the incidence of ankle trauma continues to increase.1–3 The annual incidence of ankle fractures is approximately 122 per 100,000 people4 or 187 per 100,000 fractures each year in the United States.5 The tibial plafond is forced to withstand loads 1.5–5.5 times body weight, and anatomic alignment after fracture has been shown to be critical in restoring contact area and pressures and preventing posttraumatic arthritis.6–12 A more recent study has correlated articular incongruity after surgical fixation of ankle fractures with poor short-term functional outcomes.13 Although various indications for operative fixation may exist, intraoperative and postoperative ankle fracture reduction and articular congruency are routinely assessed with fluoroscopy and plain radiographs (XR). To our knowledge, there exists no literature addressing the accuracy of articular reduction as measured by standard intraoperative and postoperative imaging modalities. The purpose of this cohort study was to determine the sensitivity and specificity and the interobserver and intraobserver reliabilities of intraoperative fluoroscopy and plain radiographs in the assessment of articular congruency in operatively treated ankle fractures when compared with computed tomography (CT) as the gold standard.
PATIENTS AND METHODS All patients with ankle fractures treated operatively at our level 1 trauma center were enrolled into a prospective institutional review board–approved registry beginning in 2003. Each patient was treated by the senior surgeon and underwent standard 3-view (Anterior-posterior, lateral, and mortise) intraoperative fluoroscopy and postoperative CT scan of the involved ankle on postoperative day 1. Out-ofplaster plain radiographs were also obtained for each patient at the initial postoperative visit 2 weeks after surgery. The current study is a retrospective analysis of this prospectively collected data, including patients from the initiation of data collection until the end of 2011 who sustained rotational ankle fractures with involvement of medial malleolus and/or posterior malleolus (OTA classification 43-B2, 43-B3, and 44 excluding 44-A1 and 44B-114). Of the 628 patients within the registry, 454 had osseous injuries limited to the lateral malleolus only, leaving 174 for potential inclusion. Each patient underwent open reduction and internal fixation (ORIF) of the fibula and posterior malleolus (when necessary) via www.jorthotrauma.com |
e161
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
Garner et al
a posterolateral approach to the ankle. The fibula was most commonly plated from a posterior antiglide position. If a posterior malleolus fracture was present, this was typically fixed after the fibula through the same posterolateral incision. If a medial malleolus fracture was present, this was fixed with an anteromedial or direct medial incision after fibular and posterior malleolar fixation. A standardized “spectral” CT to reduce metal artifact was performed on postoperative day 1. Axial CT acquisition was performed at 0.625 mm slice thickness and 0.312 mm intervals. Sagittal and coronal reformats were then performed. Initial postoperative radiographs were obtained after 2 weeks of splint immobilization and were taken out of plaster at the time of the first follow-up visit. Those without a full complement of images (n = 67) and patients requiring staged internal fixation (n = 2) were excluded. Articular congruency was determined based on the criteria established by Berkes et al13: (1) articular step-off . 2 mm, (2) intra-articular loose body, and (3) articular gap . 2 mm. Three independent observers, including 2 senior residents and the senior attending surgeon, reviewed a randomized series of both intraoperative fluoroscopic images and initial out-of-plaster postoperative radiographs at 2 separate time points that were a minimum of 2 weeks apart. Each patient was determined to have a congruent or incongruent joint based on the above listed criteria. CT imaging was independently interpreted by the 2 senior residents. Disagreement regarding articular congruency was settled with an additional review by the senior attending surgeon who was blinded to the readings of other reviewers and CT results. Statistical analyses were performed by a member of the research team with advanced training in epidemiology and biostatistics using SPSS Version 19.0 (IBM Software, Armonk, NY). Consensus fluoroscopic and plain radiographic assessments were determined for each patient and used to calculate sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) when compared with CT imaging. If there was any disagreement in radiographic interpretation, the images were evaluated as a group to provide a consensus. Interobserver reliability was evaluated using a customized SPSS macro that calculated Fleiss’ kappa; intraobserver reliability was assessed using Cohen’s kappa.
RESULTS
One hundred five patients were included in the final analysis. Sixty-three (60%) were females, and 42 (40%) were males. The average age at the time of surgery was 50.5 years (range: 16–88). Twenty-six (24.8%) had a posterior malleolar fracture without a medial malleolar fracture, 35 (33.3%) had a medial malleolar fracture without a posterior malleolar fracture, and 44 (41.9%) had both medial and posterior malleolar fractures. Seventy-three (69.5%) sustained supination external rotation injuries, 31 (29.5%) sustained pronation external injuries, and 1 (95%) was classified as a pronation abduction injury. On CT imaging obtained on postoperative day 1, 25/ 105 (23.8%) patients were noted to have an incongruency of the distal tibial articular surface. Six (24%) were because of the presence of an intra-articular loose body, 7 (28%) were
e162
| www.jorthotrauma.com
because of a gap at the articular surface of .2 mm, and 12 (48%) were because of a step-off of .2 mm at the fracture site. Only 1 of 6 (17%) loose body was recognized by any of the 3 reviewers on either fluoroscopy or plain radiographs. Reliability, sensitivity, and specificity results are displayed in Table 1. Intraoperative fluoroscopy had low sensitivity (21%) but high specificity (95%), corresponding to an NPV of 56% and a PPV of 80%, respectively. According to the Landis criteria,15 interobserver agreement was slight (k = 0.15) and mean intraobserver agreement was fair (k = 0.32). With regard to postoperative plain radiographs, sensitivity was 36% and specificity was 89%, corresponding to an NPV of 50% and PPV of 82%. Interobserver reliability was fair (k = 0.30), and mean intraobserver reliability was moderate (k = 0.59). Accuracy of fluoroscopy and plain radiography for both medial and posterior malleolar reductions can also be seen in Table 1. No patient underwent a second procedure for malreduction seen on postoperative CT.
DISCUSSION It is well known that ankle fractures can lead to the development of posttraumatic osteoarthritis of the ankle, with most studies showing that severity of injury also plays a role in a patient’s clinical outcome.6–10,12,13 Lubbeke et al7 reviewed 373 ankle fractures: 37 of 102 (36.3%) available for follow-up demonstrated advanced signs of osteoarthritis. They concluded that Weber C fractures, medial malleolar fractures, and fracture–dislocations were all risk factors for the development of osteoarthritis.7 Beris et al8 retrospectively reviewed 144 patients and found that patients suffering bimalleolar fractures, a dislocation at the time of injury, or malpositioning of a large articular fragment had worse clinical outcomes. Furthermore, malreduction has been shown to alter contact stresses between the tibial plafond and talar dome as shown by Ramsey and Hamilton10 in their 1976 study. Using a cadaveric model, they were able to establish that 1 mm of lateral translation accounts for a 42% decrease in tibiotalar contact area, which subsequently leads to abnormal stresses
TABLE 1. Sensitivity, Specificity, NPV, PPV, Interobserver Reliability, Mean Intraobserver Reliability, and Accuracy for Intraoperative Fluoroscopy and Postoperative Plain Radiography in the Assessment of Ankle Joint Congruency Indicate Deficiencies in These Diagnostic Modalities Sensitivity Specificity NPV PPV Interobserver reliability Intraobserver reliability* Accuracy (%) Medial malleolus Posterior malleolus
Fluoroscopy
Plain Radiography
21% 95% 56% 80% 0.15 0.3
36% 89% 50% 82% 0.30 0.59
85.4 84.6
86.7 80.0
*Mean of 3 observers.
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
across the joint and presumably arthritis. More recently, a clinical study was able to correlate postoperative ankle articular incongruity with statistically worse foot and ankle outcome scores at a mean of 21 months, indicating that early functional outcomes are also worse in patients with articular incongruency.13 Although multiple authors have been able to show that malreduction is correlated with increased contact stresses, the development of osteoarthritis, and poor clinical outcomes, little attention has been paid to the reliability and imaging modalities that are used to assess reduction. Fluoroscopy has been shown to be less accurate when compared with plain radiographs16; yet, it allows easy intraoperative assessment of fractures and fixation with minimal added operative time. Plain radiographs, which are inexpensive and readily accessible, are the standard instruments for assessment of fracture reduction and healing postoperatively. However, to our knowledge, the sensitivity, specificity, and reliability with regard to the assessment of articular congruity in the ankle have yet to be determined. The goal of this study was to determine these values using axial CT with coronal and sagittal reformats as the gold standard. The results of the current study cast significant doubt on the ability of both fluoroscopy and postoperative plain radiographs to accurately assess ankle articular congruity after ORIF of ankle fractures that directly involve the articular surface. Sensitivities of 21% for fluoroscopy and 36% for plain radiographs are unacceptable for modalities that are considered to be standard of care worldwide. Based on these numbers, only 21%–36% of patients with articular incongruency are actually identified using fluoroscopy or plain radiographs. Furthermore, neither imaging modality was found to have better than “moderate” interobserver or intraobserver reliability (Table 1), indicating that assessment of articular reduction varies significantly between practitioners. Lastly, although limited by sample size, only 1 of 6 (17%) intraarticular loose body was visualized on standard imaging modalities. These results mimic those of Ebraheim et al17 who showed the inadequacy of anterior–posterior radiographs in the assessment of injuries to the tibial plafond. Figure 1 demonstrates a patient with fluoroscopic and plain radiographs that demonstrate a congruent joint surface but greater than 2 mm of articular step-off on sagittal reformats of the postoperative CT scan. As can be seen in Table 1, specificity and PPV were greater than 0.8 in all cases, indicating that if an articular congruency is seen on either fluoroscopy or plain radiographs it is very likely to be present on axial imaging as well. NPV was moderate for both modalities (fluoroscopy = 0.56, XR = 50%). Therefore, a negative result may be incorrect up to 50% of the time. Our results highlight the inadequacy of intraoperative fluoroscopy and postoperative plain radiography and call into question the ability to assess articular reduction without axial imaging or, at a minimum, a more vigilant and detailed assessment of available imaging modalities. Furthermore, with regard to existing ankle fracture research, adequacy of articular reduction is a common outcome variable and the current study indicates that these imaging techniques are neither sensitive nor reliable. To our knowledge, only one Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Ankle Imaging Is Unreliable
study has correlated patient outcomes with articular incongruency confirmed on axial imaging.13 As mentioned previously, this study shows a clinically significant decrease in FAOS scores at 21 months in patients with postoperative articular incongruency. Future investigation using articular reduction as a variable should strongly consider advanced axial imaging to confirm reduction before data analysis. Despite the results of this study, we are unable to advocate for an alternative to fluoroscopy and plain radiography at this time. Ankle arthroscopy after ORIF is one possible method to assess the articular reduction at the time of surgery. However, arthroscopy would lengthen the duration of the case, increase the complexity of operative equipment needed, add to operative costs, and is not the standard of care in our institution. Additionally, the ability of arthroscopy to assess an ankle articular malreduction has never been compared with CT and remains an area of possible future study. Although CT scans are obtained on all postoperative patients at our center to assess articular incongruency, this is not a routine practice at most institutions, and this does not provide real-time information at the time of surgical fixation. The limitations of postoperative axial imaging are highlighted by the fact that, despite an articular incongruency rate of 23.8% (25/105), these radiographic results did not have any impact on clinical decision making. In the current study, no patient underwent revision surgery. At the time of postoperative CT scan, we had no reasons to expect an improved ability to adequately detect and correct a malreduction at the time of a second surgery if it could not be identified at the time of the index procedure. Although intraoperative CT scanning provides real-time data, it is not feasible at many centers. An area of future study may include the addition of oblique imaging of the ankle as a supplement to the standard 3 views. Ideally, future investigations will compare other intraoperative imaging techniques to assess their sensitivity, specificity, and reliability. Franke et al18 have recently advocated the use of intraoperative 3-dimensional fluoroscopy to assess syndesmosis reduction, which has been shown to be superior to 2-dimensional fluoroscopy and similar to CT when used as an intraoperative assessment of articular reduction and implant position.19,20 These findings, however, may be questioned given the recent finding of Davidovich et al.21 In an international multicenter trial, they were unable to show a benefit to the use of intraoperative CT scan with regard to syndesmotic reduction after fixation with transsyndesmotic screws. Other limitations to advance imaging techniques may include cost, operative time, and radiation dose. Each scan of an intraoperative O-arm cone-beam CT (Medtronic, Minneapolis, MN) takes approximately 2 minutes and gives off 7.9 mGy of radiation (equivalent to 5.6 radiographs).22 This study has limitations. Because the majority of the fractures in this cohort were read as “reduced” by both the diagnostic modality in question (eg, XR, fluoroscopy) and the gold standard (CT scan), there were disproportionate counts in the 2 · 2 tables. Therefore, although this study analyzed a large cohort of fracture patients, further subgroup analysis by fracture location or pattern was impossible because of limited number of “nonreduced” fractures. It is therefore possible that the results of this study are driven primarily by the www.jorthotrauma.com |
e163
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Garner et al
J Orthop Trauma Volume 29, Number 4, April 2015
FIGURE 1. A, Injury, (B) fluoroscopic, (C) postoperative XR, and (D) postoperative CT scan of a patient revealed a congruent ankle joint on both intraoperative fluoroscopy and postoperative XR but incongruent on postoperative CT.
inaccurate interpretation of either the medial malleolus or, more likely, the posterior malleolus reduction because of fibular hardware obstruction. However, the accuracy for both medial and posterior malleolar interpretation is similar for both modalities indicating in both cases that interpretation of fluoroscopy or plain radiographs is equally challenging (Table 1). Furthermore, there is a theoretical risk of fracture displacement postoperatively or between the time of CT scan and initial out-of-plaster radiographs, which may alter the comparison of plain radiography with CT. All patients were splinted postoperatively and immobilized until their first follow-up appointment 2 weeks after surgery. During this time, they were not permitted to bear weight. At this visit, the plaster splint and sutures were removed, followed by out-of-plaster radiographs. We believe that this period of
immobility not only limits the possibility of fragment displacement but also improves the quality of the plain radiographs by having it performed without a plaster splint obstructing osseous detail. With this in mind, however, future studies may consider obtaining all imaging on the same day to limit the possibility of fracture fragment motion between studies. In conclusion, we have shown that intraoperative fluoroscopy and postoperative plain radiographs have low sensitivity when assessing articular congruency of ankle fractures involving the medial and/or posterior malleolus. Based on these results, we believe that surgeons should rely on direct visualization of the joint surface in addition to intraoperative imaging when fracture type and surgical approach permit to allow an additional assessment of reduction. We also recommend a low threshold
e164
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
| www.jorthotrauma.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
for obtaining 3-dimensional imaging in postoperative patients to explain an atypical postoperative course as they may have a malreduced articular surface that cannot be seen on plain radiographs. Furthermore, we encourage future studies to assess the utility of intraoperative 3-dimensional imaging and those correlating outcomes with joint congruency to consider the use of axial imaging to allow a more accurate assessment of the articular surface. REFERENCES 1. Hughes JL, Weber H, Willenegger H, et al. Evaluation of ankle fractures: non-operative and operative treatment. Clin Orthop Relat Res. 1979;138: 111–119. 2. Kannus P, Parkkari J, Niemi S, et al. Epidemiology of osteoporotic ankle fractures in elderly persons in Finland. Ann Intern Med. 1996; 125:975–978. 3. Phillips WA, Schwartz HS, Keller CS, et al. A prospective, randomized study of the management of severe ankle fractures. J Bone Joint Surg Am. 1985;67:67–78. 4. Court-Brown CM, McBirnie J, Wilson G. Adult ankle fractures—an increasing problem? Acta Orthop Scand. 1998;69:43–47. 5. Daly PJ, Fitzgerald RH Jr, Melton LJ, et al. Epidemiology of ankle fractures in Rochester, Minnesota. Acta Orthop Scand. 1987;58: 539–544. 6. Joy G, Patzakis MJ, Harvey JP Jr. Precise evaluation of the reduction of severe ankle fractures. J Bone Joint Surg Am. 1974;56:979–993. 7. Lubbeke A, Salvo D, Stern R, et al. Risk factors for post-traumatic osteoarthritis of the ankle: an eighteen year follow-up study. Int Orthop. 2012;36:1403–1410. 8. Beris AE, Kabbani KT, Xenakis TA, et al. Surgical treatment of malleolar fractures. A review of 144 patients. Clin Orthop Relat Res. 1997;341: 90–98. 9. Yablon IG, Heller FG, Shouse L. The key role of the lateral malleolus in displaced fractures of the ankle. J Bone Joint Surg Am. 1977;59: 169–173.
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Ankle Imaging Is Unreliable
10. Ramsey PL, Hamilton W. Changes in tibiotalar area of contact caused by lateral talar shift. J Bone Joint Surg Am. 1976;58:356–357. 11. Lindsjo U. Operative treatment of ankle fracture-dislocations. A followup study of 306/321 consecutive cases. Clin Orthop Relat Res. 1985;199: 28–38. 12. Thordarson DB, Motamed S, Hedman T, et al. The effect of fibular malreduction on contact pressures in an ankle fracture malunion model. J Bone Joint Surg Am. 1997;79:1809–1815. 13. Berkes MB, Little MT, Lazaro LE, et al. Articular congruity is associated with short-term clinical outcomes of operatively treated SER IV ankle fractures. J Bone Joint Surg Am. 2013;95:1769–1775. 14. Marsh JL, Slongo TF, Agel J, et al. Fracture and dislocation classification compendium—2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma. 2007;21(suppl): S1–S133. 15. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. 16. Capo JT, Kinchelow T, Orillaza NS, et al. Accuracy of fluoroscopy in closed reduction and percutaneous fixation of simulated Bennett’s fracture. J Hand Surg Am. 2009;34:637–641. 17. Ebraheim N, Sabry FF, Mehalik JN. Intraoperative imaging of the tibial plafond fracture: a potential pitfall. Foot Ankle Int. 2000;21:67–72. 18. Franke J, von Recum J, Suda AJ, et al. Intraoperative three-dimensional imaging in the treatment of acute unstable syndesmotic injuries. J Bone Joint Surg Am. 2012;94:1386–1390. 19. Beerekamp MS, Sulkers GS, Ubbink DT, et al. Accuracy and consequences of 3D-fluoroscopy in upper and lower extremity fracture treatment: a systematic review. Eur J Radiol. 2012;81:4019–4028. 20. Atesok K, Finkelstein J, Khoury A, et al. The use of intraoperative threedimensional imaging (ISO-C-3D) in fixation of intraarticular fractures. Injury. 2007;38:1163–1169. 21. Davidovitch RI, Weil Y, Karia R, et al. Intraoperative syndesmotic reduction: three-dimensional versus standard fluoroscopic imaging. J Bone Joint Surg Am. 2013;95:1838–1843. 22. Hsu AR, Gross CE, Lee S. Intraoperative O-arm computed tomography evaluation of syndesmotic reduction: case report. Foot Ankle Int. 2013; 34:753–759.
www.jorthotrauma.com |
e165
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Standard Perioperative Imaging Modalities Are Unreliable in Assessing Articular Congruity of Ankle Fractures Matthew R. Garner, MD,* Peter D. Fabricant, MD, MPH,* Patrick C. Schottel, MD,* Marschall B. Berkes, MD,* Andre D. Shaffer, MD,* Amelia Ni, BA,† and Dean G. Lorich, MD*†
Objectives: To determine the sensitivity, specificity, and interobserver and intraobserver reliabilities of intraoperative fluoroscopy and postoperative plain radiographs (XR) in the assessment of articular congruency after open reduction and internal fixation (ORIF) of ankle fractures involving the tibial plafond.
Design: Retrospective cohort. Setting: Academic level 1 trauma center. Patients/Participants: One hundred five patients treated surgically for rotational ankle fractures. Intervention: ORIF. Main Outcome Measurements: Sensitivity, specificity, and interobserver and intraobserver reliabilities of fluoroscopy and plain radiographs when compared with computed tomography imaging. Results: The sensitivities of fluoroscopy and XR were 21% and 36%, respectively. Specificities were 95% (fluoroscopy) and 89% (XR). Fluoroscopy interobserver reliability was k = 0.15, and mean intraobserver reliability was k = 0.32. XR interobserver and mean intraobserver reliabilities were k = 0.30 and k = 0.59. Conclusions: Although results show acceptable specificity, the reliability and sensitivity of both intraoperative fluoroscopy and postoperative XR in the assessment of ankle articular congruency are low. This calls into question available literature correlating clinical results with articular reduction. During ORIF of an intra-articular ankle fracture, surgeons should be highly critical of fluoroscopic imaging that seems adequately reduced and direct visualization of the articular surface should be used as a reduction aid if possible. Furthermore, in the postoperative period, axial imaging may be warranted in patients who have poor clinical outcomes despite apparent anatomic articular reduction to evaluate for occult joint incongruence. Key Words: ankle fracture, surgical fixation of ankle fracture, ankle fracture imaging, trauma (J Orthop Trauma 2015;29:e161–e165) Accepted for publication July 25, 2014. From the *Hospital for Special Surgery, New York, NY; and †Department of Orthopaedics, Weill Cornell Medical College, New York, NY. The authors report no conflict of interest. Presented in part at the Annual Meeting of the American Academy of Orthopaedic Surgeons, March 20–23, 2013, Chicago, IL. Reprints: Matthew R. Garner, MD, Hospital for Special Surgery, 535 E. 70th St, New York, NY 10021 (e-mail: [email protected]). Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
J Orthop Trauma Volume 29, Number 4, April 2015
INTRODUCTION The ankle joint is the most commonly injured weightbearing joint, and the incidence of ankle trauma continues to increase.1–3 The annual incidence of ankle fractures is approximately 122 per 100,000 people4 or 187 per 100,000 fractures each year in the United States.5 The tibial plafond is forced to withstand loads 1.5–5.5 times body weight, and anatomic alignment after fracture has been shown to be critical in restoring contact area and pressures and preventing posttraumatic arthritis.6–12 A more recent study has correlated articular incongruity after surgical fixation of ankle fractures with poor short-term functional outcomes.13 Although various indications for operative fixation may exist, intraoperative and postoperative ankle fracture reduction and articular congruency are routinely assessed with fluoroscopy and plain radiographs (XR). To our knowledge, there exists no literature addressing the accuracy of articular reduction as measured by standard intraoperative and postoperative imaging modalities. The purpose of this cohort study was to determine the sensitivity and specificity and the interobserver and intraobserver reliabilities of intraoperative fluoroscopy and plain radiographs in the assessment of articular congruency in operatively treated ankle fractures when compared with computed tomography (CT) as the gold standard.
PATIENTS AND METHODS All patients with ankle fractures treated operatively at our level 1 trauma center were enrolled into a prospective institutional review board–approved registry beginning in 2003. Each patient was treated by the senior surgeon and underwent standard 3-view (Anterior-posterior, lateral, and mortise) intraoperative fluoroscopy and postoperative CT scan of the involved ankle on postoperative day 1. Out-ofplaster plain radiographs were also obtained for each patient at the initial postoperative visit 2 weeks after surgery. The current study is a retrospective analysis of this prospectively collected data, including patients from the initiation of data collection until the end of 2011 who sustained rotational ankle fractures with involvement of medial malleolus and/or posterior malleolus (OTA classification 43-B2, 43-B3, and 44 excluding 44-A1 and 44B-114). Of the 628 patients within the registry, 454 had osseous injuries limited to the lateral malleolus only, leaving 174 for potential inclusion. Each patient underwent open reduction and internal fixation (ORIF) of the fibula and posterior malleolus (when necessary) via www.jorthotrauma.com |
e161
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
Garner et al
a posterolateral approach to the ankle. The fibula was most commonly plated from a posterior antiglide position. If a posterior malleolus fracture was present, this was typically fixed after the fibula through the same posterolateral incision. If a medial malleolus fracture was present, this was fixed with an anteromedial or direct medial incision after fibular and posterior malleolar fixation. A standardized “spectral” CT to reduce metal artifact was performed on postoperative day 1. Axial CT acquisition was performed at 0.625 mm slice thickness and 0.312 mm intervals. Sagittal and coronal reformats were then performed. Initial postoperative radiographs were obtained after 2 weeks of splint immobilization and were taken out of plaster at the time of the first follow-up visit. Those without a full complement of images (n = 67) and patients requiring staged internal fixation (n = 2) were excluded. Articular congruency was determined based on the criteria established by Berkes et al13: (1) articular step-off . 2 mm, (2) intra-articular loose body, and (3) articular gap . 2 mm. Three independent observers, including 2 senior residents and the senior attending surgeon, reviewed a randomized series of both intraoperative fluoroscopic images and initial out-of-plaster postoperative radiographs at 2 separate time points that were a minimum of 2 weeks apart. Each patient was determined to have a congruent or incongruent joint based on the above listed criteria. CT imaging was independently interpreted by the 2 senior residents. Disagreement regarding articular congruency was settled with an additional review by the senior attending surgeon who was blinded to the readings of other reviewers and CT results. Statistical analyses were performed by a member of the research team with advanced training in epidemiology and biostatistics using SPSS Version 19.0 (IBM Software, Armonk, NY). Consensus fluoroscopic and plain radiographic assessments were determined for each patient and used to calculate sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) when compared with CT imaging. If there was any disagreement in radiographic interpretation, the images were evaluated as a group to provide a consensus. Interobserver reliability was evaluated using a customized SPSS macro that calculated Fleiss’ kappa; intraobserver reliability was assessed using Cohen’s kappa.
RESULTS
One hundred five patients were included in the final analysis. Sixty-three (60%) were females, and 42 (40%) were males. The average age at the time of surgery was 50.5 years (range: 16–88). Twenty-six (24.8%) had a posterior malleolar fracture without a medial malleolar fracture, 35 (33.3%) had a medial malleolar fracture without a posterior malleolar fracture, and 44 (41.9%) had both medial and posterior malleolar fractures. Seventy-three (69.5%) sustained supination external rotation injuries, 31 (29.5%) sustained pronation external injuries, and 1 (95%) was classified as a pronation abduction injury. On CT imaging obtained on postoperative day 1, 25/ 105 (23.8%) patients were noted to have an incongruency of the distal tibial articular surface. Six (24%) were because of the presence of an intra-articular loose body, 7 (28%) were
e162
| www.jorthotrauma.com
because of a gap at the articular surface of .2 mm, and 12 (48%) were because of a step-off of .2 mm at the fracture site. Only 1 of 6 (17%) loose body was recognized by any of the 3 reviewers on either fluoroscopy or plain radiographs. Reliability, sensitivity, and specificity results are displayed in Table 1. Intraoperative fluoroscopy had low sensitivity (21%) but high specificity (95%), corresponding to an NPV of 56% and a PPV of 80%, respectively. According to the Landis criteria,15 interobserver agreement was slight (k = 0.15) and mean intraobserver agreement was fair (k = 0.32). With regard to postoperative plain radiographs, sensitivity was 36% and specificity was 89%, corresponding to an NPV of 50% and PPV of 82%. Interobserver reliability was fair (k = 0.30), and mean intraobserver reliability was moderate (k = 0.59). Accuracy of fluoroscopy and plain radiography for both medial and posterior malleolar reductions can also be seen in Table 1. No patient underwent a second procedure for malreduction seen on postoperative CT.
DISCUSSION It is well known that ankle fractures can lead to the development of posttraumatic osteoarthritis of the ankle, with most studies showing that severity of injury also plays a role in a patient’s clinical outcome.6–10,12,13 Lubbeke et al7 reviewed 373 ankle fractures: 37 of 102 (36.3%) available for follow-up demonstrated advanced signs of osteoarthritis. They concluded that Weber C fractures, medial malleolar fractures, and fracture–dislocations were all risk factors for the development of osteoarthritis.7 Beris et al8 retrospectively reviewed 144 patients and found that patients suffering bimalleolar fractures, a dislocation at the time of injury, or malpositioning of a large articular fragment had worse clinical outcomes. Furthermore, malreduction has been shown to alter contact stresses between the tibial plafond and talar dome as shown by Ramsey and Hamilton10 in their 1976 study. Using a cadaveric model, they were able to establish that 1 mm of lateral translation accounts for a 42% decrease in tibiotalar contact area, which subsequently leads to abnormal stresses
TABLE 1. Sensitivity, Specificity, NPV, PPV, Interobserver Reliability, Mean Intraobserver Reliability, and Accuracy for Intraoperative Fluoroscopy and Postoperative Plain Radiography in the Assessment of Ankle Joint Congruency Indicate Deficiencies in These Diagnostic Modalities Sensitivity Specificity NPV PPV Interobserver reliability Intraobserver reliability* Accuracy (%) Medial malleolus Posterior malleolus
Fluoroscopy
Plain Radiography
21% 95% 56% 80% 0.15 0.3
36% 89% 50% 82% 0.30 0.59
85.4 84.6
86.7 80.0
*Mean of 3 observers.
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
across the joint and presumably arthritis. More recently, a clinical study was able to correlate postoperative ankle articular incongruity with statistically worse foot and ankle outcome scores at a mean of 21 months, indicating that early functional outcomes are also worse in patients with articular incongruency.13 Although multiple authors have been able to show that malreduction is correlated with increased contact stresses, the development of osteoarthritis, and poor clinical outcomes, little attention has been paid to the reliability and imaging modalities that are used to assess reduction. Fluoroscopy has been shown to be less accurate when compared with plain radiographs16; yet, it allows easy intraoperative assessment of fractures and fixation with minimal added operative time. Plain radiographs, which are inexpensive and readily accessible, are the standard instruments for assessment of fracture reduction and healing postoperatively. However, to our knowledge, the sensitivity, specificity, and reliability with regard to the assessment of articular congruity in the ankle have yet to be determined. The goal of this study was to determine these values using axial CT with coronal and sagittal reformats as the gold standard. The results of the current study cast significant doubt on the ability of both fluoroscopy and postoperative plain radiographs to accurately assess ankle articular congruity after ORIF of ankle fractures that directly involve the articular surface. Sensitivities of 21% for fluoroscopy and 36% for plain radiographs are unacceptable for modalities that are considered to be standard of care worldwide. Based on these numbers, only 21%–36% of patients with articular incongruency are actually identified using fluoroscopy or plain radiographs. Furthermore, neither imaging modality was found to have better than “moderate” interobserver or intraobserver reliability (Table 1), indicating that assessment of articular reduction varies significantly between practitioners. Lastly, although limited by sample size, only 1 of 6 (17%) intraarticular loose body was visualized on standard imaging modalities. These results mimic those of Ebraheim et al17 who showed the inadequacy of anterior–posterior radiographs in the assessment of injuries to the tibial plafond. Figure 1 demonstrates a patient with fluoroscopic and plain radiographs that demonstrate a congruent joint surface but greater than 2 mm of articular step-off on sagittal reformats of the postoperative CT scan. As can be seen in Table 1, specificity and PPV were greater than 0.8 in all cases, indicating that if an articular congruency is seen on either fluoroscopy or plain radiographs it is very likely to be present on axial imaging as well. NPV was moderate for both modalities (fluoroscopy = 0.56, XR = 50%). Therefore, a negative result may be incorrect up to 50% of the time. Our results highlight the inadequacy of intraoperative fluoroscopy and postoperative plain radiography and call into question the ability to assess articular reduction without axial imaging or, at a minimum, a more vigilant and detailed assessment of available imaging modalities. Furthermore, with regard to existing ankle fracture research, adequacy of articular reduction is a common outcome variable and the current study indicates that these imaging techniques are neither sensitive nor reliable. To our knowledge, only one Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Ankle Imaging Is Unreliable
study has correlated patient outcomes with articular incongruency confirmed on axial imaging.13 As mentioned previously, this study shows a clinically significant decrease in FAOS scores at 21 months in patients with postoperative articular incongruency. Future investigation using articular reduction as a variable should strongly consider advanced axial imaging to confirm reduction before data analysis. Despite the results of this study, we are unable to advocate for an alternative to fluoroscopy and plain radiography at this time. Ankle arthroscopy after ORIF is one possible method to assess the articular reduction at the time of surgery. However, arthroscopy would lengthen the duration of the case, increase the complexity of operative equipment needed, add to operative costs, and is not the standard of care in our institution. Additionally, the ability of arthroscopy to assess an ankle articular malreduction has never been compared with CT and remains an area of possible future study. Although CT scans are obtained on all postoperative patients at our center to assess articular incongruency, this is not a routine practice at most institutions, and this does not provide real-time information at the time of surgical fixation. The limitations of postoperative axial imaging are highlighted by the fact that, despite an articular incongruency rate of 23.8% (25/105), these radiographic results did not have any impact on clinical decision making. In the current study, no patient underwent revision surgery. At the time of postoperative CT scan, we had no reasons to expect an improved ability to adequately detect and correct a malreduction at the time of a second surgery if it could not be identified at the time of the index procedure. Although intraoperative CT scanning provides real-time data, it is not feasible at many centers. An area of future study may include the addition of oblique imaging of the ankle as a supplement to the standard 3 views. Ideally, future investigations will compare other intraoperative imaging techniques to assess their sensitivity, specificity, and reliability. Franke et al18 have recently advocated the use of intraoperative 3-dimensional fluoroscopy to assess syndesmosis reduction, which has been shown to be superior to 2-dimensional fluoroscopy and similar to CT when used as an intraoperative assessment of articular reduction and implant position.19,20 These findings, however, may be questioned given the recent finding of Davidovich et al.21 In an international multicenter trial, they were unable to show a benefit to the use of intraoperative CT scan with regard to syndesmotic reduction after fixation with transsyndesmotic screws. Other limitations to advance imaging techniques may include cost, operative time, and radiation dose. Each scan of an intraoperative O-arm cone-beam CT (Medtronic, Minneapolis, MN) takes approximately 2 minutes and gives off 7.9 mGy of radiation (equivalent to 5.6 radiographs).22 This study has limitations. Because the majority of the fractures in this cohort were read as “reduced” by both the diagnostic modality in question (eg, XR, fluoroscopy) and the gold standard (CT scan), there were disproportionate counts in the 2 · 2 tables. Therefore, although this study analyzed a large cohort of fracture patients, further subgroup analysis by fracture location or pattern was impossible because of limited number of “nonreduced” fractures. It is therefore possible that the results of this study are driven primarily by the www.jorthotrauma.com |
e163
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Garner et al
J Orthop Trauma Volume 29, Number 4, April 2015
FIGURE 1. A, Injury, (B) fluoroscopic, (C) postoperative XR, and (D) postoperative CT scan of a patient revealed a congruent ankle joint on both intraoperative fluoroscopy and postoperative XR but incongruent on postoperative CT.
inaccurate interpretation of either the medial malleolus or, more likely, the posterior malleolus reduction because of fibular hardware obstruction. However, the accuracy for both medial and posterior malleolar interpretation is similar for both modalities indicating in both cases that interpretation of fluoroscopy or plain radiographs is equally challenging (Table 1). Furthermore, there is a theoretical risk of fracture displacement postoperatively or between the time of CT scan and initial out-of-plaster radiographs, which may alter the comparison of plain radiography with CT. All patients were splinted postoperatively and immobilized until their first follow-up appointment 2 weeks after surgery. During this time, they were not permitted to bear weight. At this visit, the plaster splint and sutures were removed, followed by out-of-plaster radiographs. We believe that this period of
immobility not only limits the possibility of fragment displacement but also improves the quality of the plain radiographs by having it performed without a plaster splint obstructing osseous detail. With this in mind, however, future studies may consider obtaining all imaging on the same day to limit the possibility of fracture fragment motion between studies. In conclusion, we have shown that intraoperative fluoroscopy and postoperative plain radiographs have low sensitivity when assessing articular congruency of ankle fractures involving the medial and/or posterior malleolus. Based on these results, we believe that surgeons should rely on direct visualization of the joint surface in addition to intraoperative imaging when fracture type and surgical approach permit to allow an additional assessment of reduction. We also recommend a low threshold
e164
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
| www.jorthotrauma.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
J Orthop Trauma Volume 29, Number 4, April 2015
for obtaining 3-dimensional imaging in postoperative patients to explain an atypical postoperative course as they may have a malreduced articular surface that cannot be seen on plain radiographs. Furthermore, we encourage future studies to assess the utility of intraoperative 3-dimensional imaging and those correlating outcomes with joint congruency to consider the use of axial imaging to allow a more accurate assessment of the articular surface. REFERENCES 1. Hughes JL, Weber H, Willenegger H, et al. Evaluation of ankle fractures: non-operative and operative treatment. Clin Orthop Relat Res. 1979;138: 111–119. 2. Kannus P, Parkkari J, Niemi S, et al. Epidemiology of osteoporotic ankle fractures in elderly persons in Finland. Ann Intern Med. 1996; 125:975–978. 3. Phillips WA, Schwartz HS, Keller CS, et al. A prospective, randomized study of the management of severe ankle fractures. J Bone Joint Surg Am. 1985;67:67–78. 4. Court-Brown CM, McBirnie J, Wilson G. Adult ankle fractures—an increasing problem? Acta Orthop Scand. 1998;69:43–47. 5. Daly PJ, Fitzgerald RH Jr, Melton LJ, et al. Epidemiology of ankle fractures in Rochester, Minnesota. Acta Orthop Scand. 1987;58: 539–544. 6. Joy G, Patzakis MJ, Harvey JP Jr. Precise evaluation of the reduction of severe ankle fractures. J Bone Joint Surg Am. 1974;56:979–993. 7. Lubbeke A, Salvo D, Stern R, et al. Risk factors for post-traumatic osteoarthritis of the ankle: an eighteen year follow-up study. Int Orthop. 2012;36:1403–1410. 8. Beris AE, Kabbani KT, Xenakis TA, et al. Surgical treatment of malleolar fractures. A review of 144 patients. Clin Orthop Relat Res. 1997;341: 90–98. 9. Yablon IG, Heller FG, Shouse L. The key role of the lateral malleolus in displaced fractures of the ankle. J Bone Joint Surg Am. 1977;59: 169–173.
Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
Ankle Imaging Is Unreliable
10. Ramsey PL, Hamilton W. Changes in tibiotalar area of contact caused by lateral talar shift. J Bone Joint Surg Am. 1976;58:356–357. 11. Lindsjo U. Operative treatment of ankle fracture-dislocations. A followup study of 306/321 consecutive cases. Clin Orthop Relat Res. 1985;199: 28–38. 12. Thordarson DB, Motamed S, Hedman T, et al. The effect of fibular malreduction on contact pressures in an ankle fracture malunion model. J Bone Joint Surg Am. 1997;79:1809–1815. 13. Berkes MB, Little MT, Lazaro LE, et al. Articular congruity is associated with short-term clinical outcomes of operatively treated SER IV ankle fractures. J Bone Joint Surg Am. 2013;95:1769–1775. 14. Marsh JL, Slongo TF, Agel J, et al. Fracture and dislocation classification compendium—2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma. 2007;21(suppl): S1–S133. 15. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. 16. Capo JT, Kinchelow T, Orillaza NS, et al. Accuracy of fluoroscopy in closed reduction and percutaneous fixation of simulated Bennett’s fracture. J Hand Surg Am. 2009;34:637–641. 17. Ebraheim N, Sabry FF, Mehalik JN. Intraoperative imaging of the tibial plafond fracture: a potential pitfall. Foot Ankle Int. 2000;21:67–72. 18. Franke J, von Recum J, Suda AJ, et al. Intraoperative three-dimensional imaging in the treatment of acute unstable syndesmotic injuries. J Bone Joint Surg Am. 2012;94:1386–1390. 19. Beerekamp MS, Sulkers GS, Ubbink DT, et al. Accuracy and consequences of 3D-fluoroscopy in upper and lower extremity fracture treatment: a systematic review. Eur J Radiol. 2012;81:4019–4028. 20. Atesok K, Finkelstein J, Khoury A, et al. The use of intraoperative threedimensional imaging (ISO-C-3D) in fixation of intraarticular fractures. Injury. 2007;38:1163–1169. 21. Davidovitch RI, Weil Y, Karia R, et al. Intraoperative syndesmotic reduction: three-dimensional versus standard fluoroscopic imaging. J Bone Joint Surg Am. 2013;95:1838–1843. 22. Hsu AR, Gross CE, Lee S. Intraoperative O-arm computed tomography evaluation of syndesmotic reduction: case report. Foot Ankle Int. 2013; 34:753–759.
www.jorthotrauma.com |
e165
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
