Abstract
Background. The majority of retained syndesmotic screws will either loosen or break once the patient resumes weight-bearing. While evidence is limited, anecdotal experience suggests that intraosseous screw breakage may be problematic for some patients due to painful bony erosion. This study seeks to identify the incidence of intraosseous screw breakage, variables that may predict intraosseous screw breakage, and whether intraosseous screw breakage is associated with higher rates of implant removal secondary to pain. Methods. Five hundred thirty-one patients undergoing syndesmotic stabilization were screened, of which 43 patients (with 58 screws) experiencing postoperative screw breakage met inclusion criteria. Patient charts were retrospectively reviewed for demographic data, comorbidities, time to screw breakage, location of screw breakage, and implant removal. Several radiographic parameters were evaluated for their potential to influence the site of screw breakage. Results. Intraosseous screw breakage occurred in 32 patients (74.4%). Screw breakage occurred exclusively in the tibiofibular clear space in the remaining 11 instances (25.6%). Intraosseous screw breakage was significantly associated with eventual implant removal after breakage (P = .034). Screws placed further from the tibiotalar joint were at less risk for intraosseous breakage (odds ratio 0.818, P = .002). Screws placed at a threshold height of 20 mm or greater were more likely to break in the clear space (odds ratio 12.1, P = .002). Conclusion. Syndesmotic screw breakage may be more problematic than previously described. Intraosseous breakage was associated with higher rates of implant removal secondary to pain in this study. Placement of screws 20 mm or higher from the tibiotalar joint may decrease risk of intraosseous breakage.
When retained, the majority of syndesmotic screws will loosen or break once the patient resumes weight-bearing.
Acute instability of the tibiofibular syndesmosis is common, occurring in up to 23% of patients presenting with ankle fractures.1,2 Despite the rising popularity of suture button devices, surgical stabilization is most commonly achieved with fully threaded screws.3-6 Several controversies including screw size, number of screws, and number of cortices engaged have been previously investigated in the literature. 4 Screw retention versus routine removal remains highly variable in clinical practice despite multiple investigations demonstrating no clinical, radiographic, or functional benefit to screw removal.6,7-10
When retained, the majority of syndesmotic screws will loosen or break once the patient resumes weight-bearing. 11 This occurs due to normal, physiologic motion between the tibia and fibula, and appears to be of little clinical consequence with some studies suggesting equivalent or improved outcomes when compared to well-fixed screws.12,13 While evidence is limited, anecdotal experience has raised concerns that intraosseous screw breakage, as defined as breakage within either the tibia or the fibula, may be problematic for some patients due to painful bony erosion caused by the retained screw in association with fibular kinematics. 14 While commonly encountered clinically, to the best of our knowledge there are no reports or previous investigations in the current literature examining location of screw breakage.
At present, the clinical implications of intraosseous screw breakage (as opposed to breakage in the clear space between the tibia and fibula) have not been examined in the literature. Furthermore, no previous investigation has identified the subset of patients at risk for intraosseous screw breakage nor the factors predisposing to its occurrence. As such, the purposes of this study were 3-fold: (1) to determine the incidence of intraosseous screw breakage, (2) to identify clinical and/or radiographic variables that may predict intraosseous screw breakage, and (3) to determine if intraosseous screw breakage is associated with higher rates of implant removal secondary to pain. We hypothesized that intraosseous screw breakage would be associated with increased rates of implant removal secondary to pain and closer proximity of screw placement relative to the plafond would predispose to intraosseous screw breakage.
Materials and Methods
Following institutional review board approval, we performed a retrospective review of all patients undergoing syndesmotic stabilization at a level 1 academic institution between January 10, 2011, and November 13, 2018. Patients were identified by database search for CPT (Current Procedural Terminology) code 27829 (open treatment of distal tibiofibular joint disruption) and subsequently screened. Inclusion criteria for the final study population were (1) patient age of 18 years or older and (2) evidence of syndesmosis screw breakage in the postoperative period. Exclusion criteria were (1) fixation with suture-button devices, (2) syndesmosis screw breakage intraoperatively at the time of planned removal, (3) syndesmosis screw placement as supplementary fibula pro-tibia fixation (in the absence of syndesmotic injury), (4) treatment of injuries other than acute, primary syndesmotic disruption (eg, nonunion, revision surgery, tibial plafond [pilon] fractures), and (5) fixation with R3lease Screws (Paragon 28, Englewood, CO), which are designed to specifically break within the tibiofibular clear space.
Patient records were reviewed for demographic data, comorbidities, time to full weight-bearing, time to screw breakage, location of screw breakage as determined on the mortise radiograph, and implant removal if applicable. Ankle fractures were classified per the Lauge-Hansen classification system, and the details of syndesmosis fixation including screw size, number of screws, and number of cortices engaged were obtained from the surgical record and intraoperative fluoroscopic imaging. Other surgical details including type of fibular fixation when applicable and the presence or absence of medial malleolar and/or posterior malleolar fixation were also collected.
Multiple radiographic parameters were also evaluated for their potential to influence the site of screw breakage. Measurements were completed digitally utilizing our institution’s picture archiving and communication system (PACS) and performed for intact screws on mortise radiographs obtained at each patient’s first postoperative visit. These included (1) the distance of each screw head from the tibial plafond and (2) the angle of each screw relative to the tibiotalar joint. Imaging review and measurements were completed by 3 investigators. Examples of each measurement are depicted in Figures 1A and B.
Measurements were completed for intact screws on mortise radiographs obtained at each patient’s first postoperative visit. These included (A) the distance of each screw head from the tibial plafond and (B) the angle of each screw relative to the tibiotalar joint.
Patients were treated by a total of 6 staff orthopaedic surgeons well-versed in ankle fracture care. Commonly utilized clinical indices to diagnose syndesmotic injury, such as radiographic evidence of syndesmotic disruption and/or intraoperative stress testing (ie, cotton test), were used. All surgical decisions, including quantity of syndesmotic screws and number of cortices transfixed, were made at the discretion of the treating surgeon at the time of surgery. All patients were treated with a standardized postoperative protocol consisting of 6 weeks of restricted weight-bearing, followed by physical therapy, transition to a walking boot, and eventual progression to full, unrestricted weight-bearing by 12 weeks.
Statistics
Statistical analyses were completed in SPSS version 23 (IBM Corporation, Armonk, NY). For the purpose of our analysis, we defined 2 groups for breakage site to which patients were assigned: (1) Clear Space (CS), in which the failure site for all broken screws was in the tibiofibular clear space (Figure 2A), and (2) Intraosseous (IO), in which one or more screws failed within the tibia, fibula, or both (Figure 2B). Fisher’s exact test was utilized to identify differences between groups assuming nonnormal distribution. Binomial logistic regression was used to determine the ability of several independent variables to predict breakage site, intraosseous versus clear space. To facilitate regression analysis, each broken screw was considered individually with the dependent variable being the site of breakage—clear space versus intraosseous (Table 1). The senior author, who has a declared potential conflict of interest, was not involved in any aspect of data collection, interpretation, and analysis.
(A) Mortise radiograph demonstrating screw break within the tibiofibular clear space. (B) Mortise radiograph demonstrating intraosseous screw breakage occurring in the tibia and fibula.
Comparison of Groups.
WBAT = Weight-bearing as tolerated.
Results
Overall, 531 consecutive patients undergoing surgical stabilization of the syndesmosis during the designated study period were screened. Fifty-six patients experienced syndesmotic screw breakage in the postoperative period of which 13 patients were ultimately excluded including 3 patients treated with Paragon 28 R3lease Screws (Figure 3). The final study population comprised 43 patients (58 screws), including 21 males and 22 females, with average patient age of 40.1 years. The average body mass index (BMI) was 30.3, and 3 patients were active smokers.
Flow diagram demonstrating patient selection process.
The majority of patients were treated for supination/external-rotation (SER) fractures (62.8%). Eleven patients (25.6%) were treated for pronation pattern fractures, and 5 patients were treated for Maisonneuve injuries (11.6%). Stabilization of the syndesmosis was most commonly achieved with 2, tri-cortical 3.5-mm screws. Patients were advanced to full weight-bearing at an average of 2.5 months postoperatively (range 1.4-4.1 months) and screw breakage was detected an average of 8.4 months (range 1.6-35.9 months) after surgery.
In 11 instances (25.6%), screw breakage occurred exclusively in the tibiofibular clear space. In the remaining 32 cases (74.4%), at least one screw broke within the tibia, fibula, or both locations. Nineteen patients (59.4%) in the IO (intraosseous) group as compared to 2 patients (18.2%) in the CS (clear space) group ultimately underwent implant removal after breakage at an average of 16.1 months postoperatively (Table 1). Per clinical documentation, pain was the indication for surgery for all but one patient in the IO group for which removal was indicated for “prominent hardware.” Intraosseous screw breakage was significantly associated with eventual implant removal (P = .034). No association was observed between implant removal and the presence of fibular fixation, medial malleolar fixation, or posterior malleolar fixation (Table 2).
Bivariate Analysis for Implant Removal.
In total, 58 broken screws were included in regression analysis. Thirteen patients were treated with a single syndesmotic screw. The majority of screws were 3.5 mm cortical screws (94.8%), and the average screw length was 47.2 mm (range 34-60 mm). Of 58 screws, 21 broke in the clear space and 37 at an intraosseous site (21 fibula, 16 tibia). Of each of the independent variables included in our regression model, only screw height from the tibial plafond significantly influenced breakage location. Per regression analysis, screws placed further from the tibiotalar joint were at less risk for intraosseous breakage (odds ratio [OR] 0.818, P = .002; Table 3). Similarly, screws placed at a threshold height of 20 mm or greater were more likely to break in the clear space (OR 12.1, P = .002; Table 4). A significant association between breakage location and other surgical variables including screw diameter, length, and angulation was not observed (Tables 3 and 4). While the groups were otherwise similar for the variables examined, there was a statistical difference in age between the CS versus IO cohort (mean 51.1 vs 36.3 years old, respectively, P = .009).
Logistic Regression Model 1.
Abbreviation: CI, confidence interval.
Logistic Regression Model 2.
Abbreviation: CI, confidence interval.
Discussion
While syndesmotic screw breakage has been largely considered inconsequential, many surgeons have anecdotally noted poor outcomes in certain patients. Furthermore, subsequent screw removal can be technically challenging, in particular with broken tricortical screws.15,16 Because no previous investigation has critically examined and differentiated outcomes based on the location of breakage, we performed the current investigation. We observed a relatively high rate of intraosseous breakage in patients with retained screws and a significant association between intraosseous breakage and eventual implant removal for pain in this subset of patients. Additionally, we found that more proximal positioning to the tibiotalar joint to be protective against intraosseous screw breakage.
Multiple studies have indicated equivalent or improved outcomes in patients with broken or loose screws suggesting no benefit to routine removal. This, however, does not sufficiently address a small subset of patients that may eventually require screw removal for pain or other limitation after breakage. In fact, no study to date has examined rates of screw removal after breakage or evaluated the potential influence of breakage location on patient outcomes. Boyle et al conducted a prospective, randomized study comparing screw retention to removal with 12-month follow-up. 17 This study, which was powered to detect a 10-point difference in Olerud-Molander Ankle Scores (OMAS), found no significant difference in OMAS or secondary outcomes measures. The retained screw group, however, was heterogeneous including patients with well-fixed screws, loose screws, and broken screws. Hamid et al performed a retrospective review of 52 patients comparing outcomes between patients with intact screws (n = 27), broken screws (n = 10), and those who underwent planned screw removal (n = 15). 12 Once again, the location of screw breakage was not evaluated. Additionally, only 52 of 142 patients (37%) completed outcomes evaluations and just 10 patients with broken screws were evaluated at an unspecified period of follow-up. Given the small number of patients, the effect of an even smaller subset of patients experiencing pain from suboptimal screw breakage may be concealed.
Intraosseous screw breakage may be a problematic, but underreported clinical entity given the relatively high rate of patients who required subsequent removal in our cohort. In support of our hypothesis, we observed a significant association between screw breakage within the tibia and/or fibula and subsequent implant removal. Only 2 of 11 patients in which screws broke exclusively within the tibiofibular clear space underwent removal. Bivariate analysis comparing patients who retained their broken screws and those who underwent removal found no difference in rates of removal on the basis of other surgical variables including fibular or medial malleolar fixation. Additionally, there was no evidence of nonunion, malunion, ankle arthritis, talar osteochondral lesions, or nonanatomic alignment of the mortise visualized on plain radiographs for the 21 patients undergoing screw removal. Given the lack of other known concomitant pathology, the pain at the syndesmosis may reasonably be attributed to IO screw breakage.
Screw removal after breakage is challenging and not without risks.15-18 As such, a means of identifying screws at risk for intraosseous breakage may be of significant value to surgeons. We constructed a logistic regression model evaluating several clinical and radiographic parameters for their ability to influence breakage site. Our findings suggest that screws placed in close proximity to the tibial plafond are at increased risk for intraosseous breakage. Furthermore, screws placed 20 mm or higher from the tibial plafond were significantly less likely to break within the tibia or fibula. While a biomechanical study would be required to fully elucidate why this was so, we postulate that the close anatomic proximity of the fibula to the tibia near the joint mainly accounts for intraosseous breakage. Screw angulation relative to the tibiotalar joint, in either the sagittal or coronal planes, did not influence the site of screw breakage.
The impact of screw position relative to the plafond has been previously evaluated in biomechanical and clinical studies. Earlier recommendations have been made for screw placement 20 mm to 50 mm above the plafond, in part, to avoid violation of the incisura, which has been estimated to extend 20 mm proximally from the tibiotalar joint.19-21 Sproule et al recommended placement of the syndesmosis screw 1 cm above the incisura to avoid localized pain and calcification from trans-syndesmotic placement should the screw break. 22 McBryde et al observed superior performance of screws placed 20 mm above the plafond in comparison to screws placed at 35 mm in a cadaveric mode. 19 Conversely, Verim et al found the lowest stress and degree of syndesmotic widening to occur when screws were placed between 30 and 40 mm above the plafond in a 3-dimensional finite-element model. 20 A retrospective evaluation by Kukreti et al found no difference in clinical or radiographic outcomes between syndesmotic screws placed within 20 mm of the tibial plafond and suprasyndesmotic screws placed 20 to 50 mm above the plafond. 23 A similar study by Schepers et al arrived at similar conclusions. 24 Notably, no comment was made on screw breakage in either of these studies.
While the groups were otherwise similar for the variables examined, there was a statistical difference in age between the CS versus IO cohort (mean 51.1 vs 36.3 years old, respectively, P = .009). The significance of this finding is unclear. While we did not specifically examine osteoporosis, one would not expect a significant difference in the prevalence of osteoporosis between the cohorts given (1) the relatively younger mean age of both cohorts and (2) the lower incidence of osteoporosis in these age groups. 25 Other factors not examined in this investigation, such as differences in activity level or occupation, may have been contributory.
This study has several limitations. Despite screening a large number of patients, only 43 patients with 58 broken screws ultimately met inclusion criteria, presenting some concern for selection bias. However, we examined data screened from 531 patients who underwent syndesmotic stabilization at a busy level I trauma center over an 8-year time period. Future investigations may be conducted via a multicenter approach to increase numbers. Second, distinction between clear space and intraosseous screw breakage was made on static radiographs. Significant care was taken during the imaging review process to ensure accurate distinction and measurements. Analysis of additional radiographs obtained during the follow-up period was performed in the few cases where the distinction was not absolutely clear on initial imaging. Although postoperative CT imaging after screw breakage would have allowed for even better delineation and measurement of location, this was not available for most patients. Third, there was no standardization of surgical technique nor postoperative protocols given the retrospective nature of this study. While syndesmotic stabilization with screws is a relatively uniform procedure, we cannot rule out the effects of slight variations in technique nor their influence on our findings. Additionally, most patients treated for syndesmotic injuries at our institution initiate range of motion exercises at 2 weeks postoperatively with the resumption of full weight-bearing at 10 to 12 weeks. Deviations from this protocol, patient noncompliance, and other immeasurable and unexamined factors could have influenced our findings. Similarly, the effect of potential syndesmotic malreduction on screw breakage is unknown and was not investigated. Furthermore, it is important to note that the cohort of patients who underwent screw removal was treated by 6 different surgeons. Therefore, it is unlikely that any surgeon-specific preferences led to higher removal rates seen in the IO group especially given that broken syndesmotic screws (in an otherwise asymptomatic patient) are generally not removed at our institution.
The 2 limitations that require additional attention are the following: (1) no patient-based outcome measures were collected which makes comparison to previously published reports difficult and (2) potential confounders. While multiple areas of the medical records (clinic notes, operative notes, etc) were carefully reviewed to ensure that pain was the documented indication for syndesmotic screw removal, visual analog score (VAS) data were generally poorly documented. This precluded any meaningful statistical comparisons utilizing VAS, which would have provided more direct evidence that IO screw breakage is problematic.
Regardless of an investigation’s design, there is always a risk of unexamined confounding variables influencing the primary outcome. While postoperative computed tomography or magnetic resonance imaging may have elucidated other causes of pain in our cohort, there was no evidence of nonunion, malunion, ankle arthritis, talar osteochondral lesions, or nonanatomic alignment of the mortise visualized on plain radiographs for the 21 patients who underwent removal of syndesmotic fixation. Given the lack of other known concomitant pathology, the pain at the syndesmosis was reasonably attributable to IO screw breakage. However, a more judicious use of advanced imaging postoperatively may have elucidated other pathology.
Conclusion
Syndesmotic screw breakage may be more problematic than previously described. Intraosseous breakage may be associated with higher rates of implant removal secondary to pain. Placement of screws 20 mm or higher from the tibiotalar joint may decrease risk of intraosseous breakage and limit need for implant removal in the postoperative period.
Footnotes
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr Kwon reports personal fees from Paragon 28, outside the submitted work; and Dr Kwon had developed a syndesmotic implant with Paragon 28 for which he obtains royalties. This investigation did not involve this implant, nor did it investigate it. Patients who had this implant placed were excluded from the study. All data analysis and interpretation were done by others than Dr Kwon. The remaining authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ethical Approval
Beth Israel Deaconess Medical Center institutional review board approval was submitted and Category 4 exemption was granted.
Informed Consent
Informed consent was not sought for the present study.
Trial Registration
Not applicable.
