SECTION II: ORIGINAL ARTICLES
Injuries of the distal tibiofibular syndesmosis occur either isolated, or in combination with ankle fractures. If the syndesmosis is completely disrupted, diastasis can be seen on plain AP ankle or mortise radiographs.10 More subtle changes in syndesmotic width however, often are not appreciated and cannot be quantified reliably.9
Several parameters to assess ankle and syndesmotic integrity, measured on AP and mortise views of the ankle, have been described.6,7,11–13 These measurements often are used as tools in treatment decision-making. The tibiofibular clear space is described as the distance between the posterolateral border, the anterolateral border or the incisura fibularis of the tibia, and the medial border of the fibula.6,8,11–13 The tibiofibular overlap is measured as the horizontal distance between the medial border of the fibula and the lateral border of the anterior tibial tubercle.6,11–13 The medial clear space is described as the widest distance between the medial border of the talus and the lateral border of the medial malleolus on the AP view.7,8,13 Medial clear space is said not to exceed the superior clear space, which is measured between the talar dome and the tibial plafond.7 However, there is no scientific validation for this statement.
Because it is not known yet which parameter should be used to determine syndesmotic and ankle integrity, no consensus exists in the literature regarding how to measure the aforementioned parameters, and reproducibility of those parameters has not been assessed.
The current study examined the aforementioned parameters, measured in every way described in the literature and in various radiographic positions to determine the effect of incorrect positioning of the ankle on the parameters, to determine which has the highest reproducibility and should be used to determine syndesmotic and ankle integrity.
MATERIALS AND METHODS
For this study 20 plastinated human cadaver lower legs (10 right, 10 left) were used. Plastination is a process during which biologic tissues are impregnated by curable polymers resulting in dry, odorless, and durable specimens with intact gross and microscopic anatomy.14
Each specimen was mounted on a testing device that allowed full rotation of the leg around its longitudinal axis, while the ankle was kept in a slightly plantar flexed position comparable to common clinical practice, and to avoid widening of the mortise caused by forced dorsiflexion. With the use of a malleolar clamp with two levels, the axis of the tibia and the bimalleolar axis were fixed in the 0 position in the horizontal plane (Fig 1). This 0 position, made with the beam perpendicular to the line centered through the malleoli, correlates with the mortise view on ankle radiographs, whereas the 15° external rotation position correlates with the AP view.4,10 Twelve radiographs were made by rotating the leg accurately from the 0 position, in increments of 5°, to 30° external, and to 25° internal rotation, using an electronic inclinometer (Cybex EDI 320, Lumex Inc, Ronkokoma, NY) attached to the testing device. Because most ankle radiographs are made with the subject bearing no weight, no loads were applied to the tibial plateau. Distances were measured 1 cm above, and parallel to, the tibial plafond for each of the following parameters: the distance between the medial side of the fibula and the anterior tibial tubercle, the distance between the medial side of the fibula and the posterior tibial tubercle, the distance between the floor of the incisura fibularis and the anterior tubercle of the tibia, the distance between the floor of the incisura fibularis and the posterior tubercle, the width of the fibula, and the distance between the medial side of the fibula and the floor of the incisura fibularis. All projections of the tubercles medial to the fibula were given a negative value. The superior tibiotalar clear space was measured as the distance between the tibial plafond and the highest point of the medial talar dome; the medial tibiotalar clear space was measured 0.5 cm beneath it on a line parallel to the superior talar joint surface (Fig 2). The latter was measured this way as it was hypothesized more standardized readings would be given than with measuring the maximum width, as suggested by Leeds and Ehrlich.8 On two occasions three observers measured these parameters using the same ruler with 1-mm discrimination. To determine mean and range, 8640 readings were made for six of the parameters, and 960 each for medial clear space and superior clear space, as the latter two could not be measured in extreme internal and external rotation. To determine reproducibility of placement of the specimens, three investigators each did positioning five times. This multiple (replicate) positioning increased the total amount of readings to 15,360.
The results were analyzed using mixed-model ANOVA (PROC MIXED of SAS version 6.12 SAS, Cary, NC). Variability in the readings was thought to derive from ankle, positioning (nested within ankle: ankle*position), interobserver, and intraobserver and residual (mainly consisting of intraobserver and pure error).
Each of these sources was a certain proportion of the total variance, and together they added up to the total variance of the readings. If there were no replicate positionings per ankle, the positioning component was added and therefore was included in the ankle component whereas the total variance did not change. This caused the ankle component in the total variance to become too great and reproducibility too small. Therefore, total variance was reduced to a between-ankle and a within-ankle component using replicate positioning. The within-ankle component was additionally reduced to a positioning component (necessary for taking a radiograph) and a reading component. The reading component may consist of interobserver and intraobserver variability. These variabilities can be distinguished from one another if there are several observers and several replicate readings by each observer. In this way more realistic intraclass correlation coefficients and reproducibilities can be calculated. The intraclass correlation coefficient measures how accurately ankles can be distinguished from each other using the readings. The intraclass correlation coefficient is defined as the between-ankle variance (ankle component) as a proportion of the total variance, therefore it is a dimensionless number between 0 and 1 (the nearer to 1, the better). Reproducibility measures the maximum absolute difference between two replicate readings (taken under reproducible conditions) that can be attributed to chance with 95% probability, (the maximum absolute difference that is exceeded with a probability of 5%). It is defined as 1.96 times the square root of twice the within-ankle variance (the sum of positioning, interobserver and intraobserver components). It has the same dimension (mm) as the readings and can be interpreted clinically (the smaller, the better).
Tibiofibular overlap of either the anterior or the posterior tubercle and the fibula was positive or 0 in every radiograph. The value of all syndesmotic parameters changed with rotation (Table 1). The distance between the medial side of the fibula and the posterior tibial tubercle was smallest in the maximum external rotation position, and increased with internal rotation. The contrary was found for the distance between the medial side of the fibula and the anterior tibial tubercle, which increased with external rotation. The distances between the floor of the incisura fibularis and the posterior tibial tubercle and between the floor of the incisura fibularis and the anterior tibial tubercle showed similar trends. This is a result of the anterior tubercle projecting greatest in external rotation and the posterior tubercle projecting greatest in internal rotation. The mean distance between the medial side of the fibula and the floor of the incisura fibularis in every position ranged from 2.2–2.5 mm. Unfortunately, the range per position was too great to consider this distance a useful parameter.
The value of the medial clear space and superior clear space changed considerably in the different positions of rotation (Table 2), but in any position of rotation the medial clear space was smaller than or equal to the superior clear space in all but two of 960 readings.
When the positioning component was not taken into account, reliable measurements (intraclass correlation coefficient ≥ 0.7) could be made for the distances between the medial side of the fibula and the anterior tibial tubercle, between the floor of the incisura fibularis and the anterior tibial tubercle and for the width of the fibula in every position from 30° to 5° external rotation. The distance between the medial side of the fibula and the posterior tibial tubercle could only be measured reliably in extreme internal rotation. In the AP position the distances between the medial side of the fibula and the anterior tibial tubercle, between the floor of the incisura fibularis and the anterior tibial tubercle, between the medial side of the fibula and the floor of the incisura fibularis, and the width of the fibula could be reliably measured. On the mortise view the width of the fibula was the only parameter that could be measured reliably. However, when the positioning component was taken into account, the width of the fibula remained the only parameter that could be measured reliably (Tables 3 and 4).
The reproducibilities of all parameters in all positions of rotation (Tables 5 and 6), with the exception of fibular width, neared or exceeded the mean values of those parameters (Tables 1 and 2). Therefore it is evident that none of the measurements can be used in clinical practice. The width of the fibula, the only parameter that could be measured reproducibly, has no clinical relevance.
Our aims in this study were to find the optimal ankle position and parameter to assess syndesmotic integrity radiographically, as those measurements might be influenced by rotation and no consensus exists how to measure tibiofibular clear space and overlap. We radiographed 20 ankles in 12 different positions of rotation and measured tibiofibular overlap and clear space in every way described in the literature. This was followed by repeated positioning, radiography, and measurement of the parameters to assess reproducibility. This study showed that, even when done in optimal laboratory conditions, no reproducible ankle positioning is possible. Therefore the reliability of syndesmotic measurements in repeated ankle radiographs is questionable. This study was done using 20 plastinated cadaveric specimens. Plastination does not change macroscopic or microscopic anatomy, so results from this study should be comparable to results of unloaded ankle radiographs in the clinical setting.
Three observations important for clinical practice were made. First the overlap of the fibula with the tibia with respect to either the anterior (the distance between the medial side of the fibula and the anterior tibial tubercle) or the posterior tibial tubercle (the distance between the medial side of the fibula and the posterior tibial tubercle) was positive or 0 in every radiograph. This finding suggests that absence of overlap is abnormal, and indicates syndesmotic injury. This is in accordance with findings of Pneumaticos et al,12 but now assessed in a larger series with repeated positioning, radiography, and measuring. However, bilateral nontraumatic absence of tibiofibular overlap can be seen in some people. Therefore unilateral absence of tibiofibular overlap after ankle injury should be considered as syndesmotic diastasis.
Second, the value of the medial clear space was smaller than or equal to the superior clear space in all but two of 960 measurements. These two exceptions are most likely attributable to measurement errors, because they were measured differently by the aberrant observer on the other occasion. The absolute values of medial clear space and superior clear space changed considerably in the different positions of rotation. This observation is in contrast with findings of a less extensive study of Joy et al,7 who stated that variations in rotation between 5° and 20° rotation do not alter measurements of the medial clear space or the superior clear space. Furthermore our study shows the combined measurement of medial clear space and superior clear space to be useful, as the medial clear space should not exceed the superior clear space in any projection of the nonweightbearing ankle. If the medial clear space exceeds the superior clear space it is indicative of deltoid ligament injury, which is not uncommon with syndesmotic injury. Finally, quantitative measurement of all syndesmotic parameters in repeated ankle radiographs may not be as useful as previously reported by other authors.6,12 In our opinion the reason for this different finding is that none of the other authors took results of repeated ankle radiographs into account. Pneumaticos et al12 stated that the tibiofibular clear space (measured as the distance between the medial side of the fibula and the posterior tibial tubercle) did not change significantly with rotation and therefore is reproducible and reliable in evaluating the integrity of the distal tibiofibular joint. Those authors however, did not address interobserver reliability and repeated ankle radiography. The current study shows that reliable measurements (intraclass correlation coefficient ≥ 0.7) can be made for the distance between the medial side of the fibula and the anterior tibial tubercle, the floor of the incisura fibularis and the anterior tibial tubercle and the width of the fibula in several positions, but that these measurements are not reliable for repeated ankle radiography. The finding of Pneumaticos et al12 that the tibiofibular overlap (measured as the distance between the medial side of the fibula and the anterior tibial tubercle) never has a negative value is confirmed by our study.
Brage et al1 showed excellent reliability for certain parameters measured on ankle radiographs, with reliability increasing with experience. The interobserver intraclass correlation coefficients from the current single radiograph measurements in our study were better than their results for the distance between the medial side of the fibula and the anterior tibial tubercle (their syndesmosis B or the tibiofibular overlap) in positions of 5° to 30° external rotation, comparable for the distance between the medial side of the fibula and the posterior tibial tubercle (their syndesmosis A or the tibiofibular clear space) on the AP view, and worse for the distance between the medial side of the fibula and the anterior tibial tubercle (their syndesmosis C) on the mortise view. The latter probably is attributable to the fact that in our study the ankles were accurately placed, so that a more precise mortise view was accomplished, with superimposition of the anterior and posterior tubercles, which made discrimination of the tubercles more difficult. Furthermore, the study by Brage et al1 evaluated reliability for single extremity–single occasion radiographs without accounting for the effect of repositioning of the ankle. This essentially is different from the current study and from clinical practice. Also, serial radiographs are mandatory to evaluate a patient during treatment, or to compare the injured with the noninjured extremity.
Measurements of the distance between the medial side of the fibula and the anterior tibial tubercle (mortise view, 4.2 mm; AP view, 8.7 mm) were close to the values which Harper and Keller6 found for the tibiofibular overlap (mortise view, 4.2 mm; AP view, 9.4mm) in these positions. There is a bigger difference between the distance between the medial side of the fibula and the floor of the incisura fibularis and their tibiofibular clear space. Harper and Keller6 reported that measurement of the tibiofibular clear space could exclude syndesmotic diastasis, based on the radiographic evaluation of 12 accurately placed fresh cadaveric lower extremities. This only is true, however, if accurate positioning has been done. Therefore, Harper5 later stated that precise evaluation of the tibiofibular relationship should be done by CT to avoid misinterpretation of the tibiofibular width because of rotational deformities or translations of the fibula. In agreement with her, it was suspected that rotation attributable to positioning errors could be responsible for even weaker reproducibilities in repeated radiographs. This was confirmed with the calculation of more realistic intraclass correlation coefficients and reproducibility based on the replicate positioning.
Computed tomography scans can show the tibial tubercles the incisura fibularis and tibiofibula, as has been shown in cadaveric studies describing the normal aspect of the incisura fibularis,2,5 but criteria how to distinguish the normal from the injured syndesmosis with CT have not been described, and reproducibility of CT for this purpose has not been assessed. Ebraheim et al3 correlated radiographic and CT findings in seven patients with low fibular fractures, but they state that with CT the syndesmosis was found to be disrupted without describing the CT criteria for disruption. Furthermore this investigation may be difficult to access in many trauma units, or be beyond the economic resources available, whereas plain radiographs are universally available and relatively inexpensive.
Based on our observations, it is evident that no optimal radiographic parameter exists to assess syndesmotic integrity, because all parameters are dependent on the position of rotation. As precise positioning of the ankle is not possible, the aforementioned parameters are not reliable in repeated ankle radiography. Two parameters were found to be of some use. These are the tibiofibular overlap and the combination of the medial and superior clear space. Unilateral absence of tibiofibular overlap should raise the suspicion of a syndesmotic injury, and a medial clear space exceeding the superior clear space on a nonweightbearing radiograph should raise the suspicion of deltoid ligament injury in any projection of a normal ankle. Quantitative measurement of all other syndesmotic parameters in repeated ankle radiographs may be of little value.
We thank Wibeke van Leeuwen for technical assistance in radiography.
1. Brage ME, Bennett CR, Whitehurst JB, Getty PJ, Toledano A: Observer reliability in ankle radiographic measurements. Foot Ankle Int 18:324–329, 1997.
2. Ebraheim NA, Lu J, Yang H, Rollins J: The fibular incisure of the tibia on CT scan: A cadaver study. Foot Ankle Int 19:318–321, 1998.
3. Ebraheim NA, Hossein E, Padanilam T: Syndesmotic disruption in low fibular fractures associated with deltoid ligament injury. Clin Orthop 409:260–267, 2003.
4. Goergen TG, Danzig LA, Resnick D, Owen CA: Roentgenographic evaluation of the tibiotalar joint. J Bone Joint Surg 59A:874–877, 1977.
5. Harper MC: An anatomic and radiographic investigation of the tibiofibular clear space. Foot Ankle 14:455–458, 1993.
6. Harper MC, Keller TS: A radiographic evaluation of the tibiofibular syndesmosis. Foot Ankle 10:156–160, 1989.
7. Joy G, Patzakis MJ, Harvey Jr JP: Precise evaluation of the reduction of severe ankle fractures. J Bone Joint Surg 56A:979–993, 1974.
8. Leeds HC, Ehrlich MG: Instability of the distal tibiofibular syndesmosis after bimalleolar and trimalleolar ankle fractures. J Bone Joint Surg 66A:490–503, 1984.
9. McDade WC: Treatment of ankle fractures. Instr Course Lect 24:251–293, 1975.
10. Pavlov H, Burke M, Giesa M, Seager KR, White ET: Orthopaedist’s Guide to Plain Film Imaging. Stuttgart, Thieme 182–187, 1999.
11. Pettrone FA, Gail M, Pee D, Fitzpatrick T, Van Herpe LB: Quantitative criteria for prediction of the results after displaced fracture of the ankle. J Bone Joint Surg 65A:66–77, 1983.
12. Pneumaticos SG, Noble PC, Chatziioannou SN, Trevino SG: The effects of rotation on radiographic evaluation of the tibiofibular syndesmosis. Foot Ankle Int 23:107–111, 2002.
13. Sclafani SJ: Ligamentous injury of the lower tibiofibular syndesmosis: Radiographic evidence. Radiology 175:21–27, 1985.
© 2004 Lippincott Williams & Wilkins, Inc.
14. Von Hagens G, Tiedemann K, Kriz W: The current potential of plastination. Anat Embryol 175:411–421, 1987.