Abstract
Background
Ankle arthrodesis (AA) is a common treatment for end-stage ankle arthritis, chronic instability, and degenerative deformity. Although minimally invasive arthroscopic techniques may reduce soft tissue disruption, postoperative pain, and related morbidity, open techniques may be beneficial for treatment of patients with aberrant anatomy, insufficient bone stock, or complex deformity. This study aimed to determine whether arthroscopic AA is associated with lower rates of adverse events, pseudarthrosis, and health care use compared with open AA techniques at short-term and long-term intervals.
Methods:
We conducted a retrospective analysis using the TriNetX research network. Patients undergoing AA were identified using Current Procedural Terminology (CPT) codes for arthroscopic (CPT 29899, n = 879) and open (CPT 27870, n = 10 604) procedures. Two cohorts were defined and propensity score–matched on age, sex, race, body mass index, nicotine dependence, chronic kidney disease, and type 2 diabetes mellitus (n = 873 each). Outcomes were evaluated within 30 days, 90 days, and 2 years.
Results:
The arthroscopic AA cohort experienced significantly lower rates of any adverse event, infection, and hospital admission within the 30-day and 90-day outcome windows. Arthroscopic AA was associated with fewer emergency department visits and wound dehiscence within 90 days of surgery. A diagnosed pseudarthrosis within 2 years was more common in the open arthrodesis cohort. Rates of short-term myocardial infarction, cerebral infarct, transfusion, pulmonary embolism, and hematoma did not differ.
Conclusion:
Arthroscopic AA was associated with significantly lower rates of medical complications at the short-term intervals, in addition to lower rates of nonunion within 2 years. Although observational, propensity-matched data are consistent with fewer short-term medical complications and lower 2-year nonunion after arthroscopic AA, the results should be interpreted with caution because of the inability to assess the degree of coronal or sagittal plane deformity in the included cases.
Level of Evidence:
Level III, retrospective cohort study.
Introduction
Osteoarthritis (OA) of the ankle primarily results from post-traumatic etiology and represents a major cause of pain and disability worldwide.2,19,24,25,27 Although total ankle arthroplasty (TAA) continues to be refined as a motion-sparing treatment option for ankle OA,3,4,7,8,16,30 ankle arthrodesis (AA) remains a mainstay of surgical care for the condition, particularly in patients who are deemed poor candidates for TAA. 14 AA is a successful procedure, subjectively improving a patient’s pain perception and objectively increasing physical function status and achieving fusion rates between 80% and 100% across the literature.11,13
First described in 1879 by Edward Albert, AA techniques have evolved substantially throughout the following decades. 6 External fixation devices were first employed to achieve ankle fusion in patients with significant congenital- or rheumatoid-derived deformity. As internal fixation implantation concepts and devices evolved, external fixation fell out of favor for healthier patients without active infection, and internal fixation for AA became commonplace. 6 More recently, arthroscopic AA was described by Schneider in 1983 during a time period where many open procedures were readily being adapted to more minimally invasive, arthroscopy-based techniques. 20 The primary goal of AA procedures is to correct the gradual, degenerative deformity of the ankle joint. Typically, the hindfoot falls into a varus deformity, and imperfect correction may result in a malunion. 7
Arthroscopic AA has compared favorably to traditional, open techniques for AA, particularly with regard to decreased blood loss, shorter hospital stays, and lower incidence of wound complications.1,10,15,18,22 However, some investigators assert that these findings may be overstated, given the heterogeneity in the cohorts of patients undergoing either open or arthroscopic AA, as hindfoot deformity is difficult to assess visually and radiographically.5,26 To date, no study has leveraged an international, multicenter database to compare the complication and nonunion rates following open AA and arthroscopic AA. Thus, the purpose of this study was to use the TriNetX Research Database (Cambridge, MA) to compare these 2 techniques. Our hypothesis was that arthroscopic AA would be associated with fewer complications, decreased readmission rate and emergency department (ED) visits, and decreased nonunion rates compared with open AA.
Methods
We used TriNetX, a health research platform that allows access to deidentified electronic medical records from international health care organizations. The query was run on the TriNetX Research network, which contains data from 105 health care organizations and more than 130 million patients. The query was performed on March 14, 2025. This study did not involve individually identifiable patient data, so it was exempt from institutional review board approval.
Two cohorts were evaluated for this study. The arthroscopic AA cohort was defined as patients who underwent arthroscopic AA and was identified using Current Procedural Terminology (CPT) code 29899. The open AA cohort included patients who underwent open AA and was identified by CPT code 27870. Patients with incomplete records and those whose procedures occurred prior to 2010 were excluded from the analysis. The variables selected for propensity score matching were age, sex, race, body mass index, type 2 diabetes mellitus, nicotine dependence, chronic kidney disease, and osteoporosis. After matching, there were 873 matched patients in each group.
Postoperative outcomes were identified using the International Classification of Diseases, Tenth Revision (ICD-10) and CPT codes listed in Appendix 1. All complications were assessed within 30 days, 90 days, and 2 years of arthrodesis surgery. Our primary outcomes were rates of any adverse event, ED visits, and hospital readmission within 30 days and 90 days, along with 2-year nonunion. We also queried for short-term rates of surgical site infection (SSI), wound dehiscence, pulmonary embolism (PE), deep vein thrombosis (DVT), myocardial infarction (MI), cerebral infarction, blood transfusion, hematoma, sepsis, and venous thromboembolism (DVT and/or PE). AAE was defined as any instance of death, dehiscence, PE, DVT, MI, cerebral infarction, transfusion, hematoma, SSI, or sepsis within a given postoperative period. Within the TriNetX database, death is defined by a change in a patient’s demographics status to deceased within their health record. All diagnoses are based on billing codes from the patients’ health records without granularity on the extent or diagnosis criteria.
The baseline demographics and comorbidities of the 2 cohorts were assessed using t tests and χ2 tests. Statistical analyses of baseline demographics and postoperative outcomes were conducted within TriNetX, and outcomes are presented as rates and odds ratios (ORs) with P values and 95% CIs. Statistical significance was assessed at P = .05. Given the number of outcomes and time windows evaluated, the P values reported should be considered exploratory, and we encourage readers to focus on the effect sizes and 95% CIs.
Results
In our analysis, 10 604 patients underwent open AA, and 879 patients underwent arthroscopic AA. After matching and excluding patients with incomplete records, there were 873 patients in each cohort. Prior to matching, the open AA cohort was older (56.5 ± 14.5 vs 54.4 ± 15.2 years; P < .001) and suffered from a greater comorbidity load, including higher rates of type 2 diabetes mellitus and chronic kidney disease (P < .001). Sex, race, body mass index, and nicotine dependence were similar between the groups. After matching, no statistically significant differences in demographics and comorbidities were observed between the cohorts, using a significance threshold of a standardized mean difference (SMD) of 0.1 (Table 1).
Arthroscopic and Open Fixation Cohort Demographics Before and After 1:1 Propensity Score Matching.
At 30 days post-surgery, the arthroscopic AA group had a lower rate of overall adverse events compared with the open group (2.3% vs 6.8%, P < .001; OR 0.32, 95% CI 0.19-0.54). Arthroscopic AA was associated with lower rates of hospital admission (4.5% vs 8.6%, P < .001). There were fewer SSIs in the arthroscopic AA group (1.1% vs 2.4%, P = .046). There were no differences in DVT, PE, MI, or sepsis rates within 30 days (Table 2).
Thirty-Day Complication Rates, Risk Analysis P Values, and Odds Ratios with 95% CIs for Arthroscopic vs Open Ankle Arthrodesis Patients After 1:1 Propensity Score Matching.
Abbreviation: N/A, not applicable.
Any adverse event: death, dehiscence, deep vein thrombosis, pulmonary embolism, myocardial infarction, cerebral infarct, transfusion, hematoma, surgical site infection, sepsis.
Statistical significance unknown, as the number of instances between 1 and 10 is not specified within TriNetX.
Statistically significant, P < .05.
Within 90 days, all primary outcomes favored arthroscopic AA. There were lower rates of AAE (4.4% vs 10.7%, P < .001; OR 0.38, 95% CI 0.26-0.56), ED visits (7.1% vs 11.6%, P = .001), and hospitalization (5.7% vs 11.8%, P < .001). Wound dehiscence (1.5% vs 4.5%, P < .001) and surgical site infection (1.3% vs 4.9%, P < .001) were less frequent following arthroscopic AA (Table 3).
Ninety-Day Complication Rates, Risk Analysis P Values, and Odds Ratios With 95% CIs for Arthroscopic vs Open Ankle Arthrodesis Patients After 1:1 Propensity Score Matching.
Any adverse event: death, dehiscence, deep vein thrombosis, pulmonary embolism, myocardial infarction, cerebral infarct, transfusion, hematoma, surgical site infection, sepsis.
Statistically significant, P < .05.
Arthroscopic surgery demonstrated lower rates of diagnosed nonunion within 2 years of surgery (20.3% vs 25.8%, P = .006) (Table 4).
Two-Year Pseudarthrosis Rates, Risk Analysis P Values, and Odds Ratios With 95% CIs for Arthroscopic vs Open Ankle Arthrodesis Patients After 1:1 Propensity Score Matching.
Statistically significant, P < .05.
Discussion
To our literature review, this study represents the first instance where a large, multicenter database was leveraged to assess variable outcomes between open and arthroscopic AA. We found that arthroscopic AA was associated with lower rates of AAE, wound dehiscence, SSI, ED visits, and hospital admission at the 90-day follow-up marks. Moreover, despite several authors pointing to concerns regarding incomplete joint preparation when AA is carried out through arthroscopic methods, we determined that long-term pseudarthrosis or nonunion rates were higher in the open AA cohort compared to the arthroscopic AA group. These findings are consistent with a favorable short-term safety profile and lower 2-year nonunion after arthroscopic AA but should be interpreted as exploratory given residual confounding and database coding constraints.
A growing body of literature exists comparing arthroscopic AA to open AA. Many of these studies are single-center, retrospective reviews.1,10,12,15,18,22,28 Projects that are multicenter in nature are limited by small cohort sizes. 23 Nevertheless, these studies provide an evidentiary basis on which comparisons of these techniques can be made. In one of the largest sample size comparative series to date, Abuhantash et al present outcomes on 223 patients who had undergone arthroscopic AA and 128 patients who had undergone open AA. The patients undergoing open AA had a higher grade of arthritis preoperatively compared with the arthroscopic group. In a mixed effects model analysis, no differences in postoperative outcome scores or risk of revision due to malunion or nonunion were noted. However, instances of wound-healing issues and deep infection did not occur in the arthroscopic AA group but did occur in 4% of the patients in the open AA group. 1
In one of the few multicenter studies on the topic, Townshend et al 23 present a series of 30 open and 30 arthroscopic AA performed across 2 separate institutions. At 2-year followup, there was statistically significantly shorter hospital stays and greater improvements in the Ankle Osteoarthritis Scale score in the arthroscopic AA cohort. Differing from other reports, complications and surgical time were similar between the 2 groups. 23 In further support of arthroscopic AA, a systematic review and meta-analysis published in 2023 provides some of the strongest data to date comparing open vs arthroscopic AA. 9 Despite limitations in the meta-analytic process, stronger, more generalizable results can be generated by combining the smaller sample size studies from single institutions into one large cohort. In their meta-analysis including 13 studies and 994 patients, Lorente et al 9 found a statistically significantly lower rate of complications and a shorter hospital length stay in the arthroscopic AA group. Nonunion rate and total operative time did not differ between the groups.
Smaller-sample-size, single-institution studies can provide more granular data, given the ability to access each patient’s individual chart once the study has been approved by the local institutional review board. However, larger, multicenter database studies can also provide a useful tool in assessing variable outcomes between arthroscopic vs open AA, despite limitations inherent to the use of a deidentified patient cohort such as the one present in this study. Our study allows for a large sample size and for the discussion of results that may prove more generalizable, because of the inclusion of numerous centers across the country, as opposed to isolated, single-institution data.
The evidence presented thus far in this manuscript has seemingly presented data strongly supporting the use of arthroscopic AA over open AA. However, this body of work should be interpreted with appropriate caution. This remains particularly true given the absence of rigorous randomized controlled trials comparing the 2 techniques. Moreover, several authors have argued that much of the literature surrounding arthroscopic AA has failed to consider the fact that the osseous pathology of patients undergoing arthroscopic AA may differ substantially from those who received open AA. Wang et al 26 assert that more complex deformities are often addressed through open AA. Thus, when comparisons are made between open AA and arthroscopic AA in simple deformity correction cases, the performance gap between these procedures narrows. 26 Achieving homogenous groups or matched sampling whereby patients with similar coronal and sagittal plane deformity, medical comorbidities, and bone quality are compared is imperative to definitively demonstrate superiority of one technique over the other.
Conversely, perhaps open and arthroscopic AA are most optimally employed on a case-by-case basis. Open AA may allow for improved deformity correction. Significant varus and valgus angulation can be corrected more easily through direct visualization and preferential resection of greater amounts of either the medial or lateral tibial plafond. 29 Medial malleolar or fibular osteotomy can be used in open cases where optimal alignment cannot be achieved through joint preparation alone. 21 In other circumstances, arthroscopic AA may be the optimal technique. In situations where infection risk and wound healing are of tantamount concern, arthroscopic AA may prove advantageous over open AA. Patients at high risk for longer durations of hospitalization and venous thromboembolic complications may also benefit from arthroscopic AA over open AA. Surgeons who are able to direct their choice of technique to the specific patient at hand may best be able to maximize patient outcomes and return to activity, while limiting complications.
Limitations
As a primary limitation of any study reliant on a national or international multicenter database, the reliability of the presented results is subject to coding errors made by the providers themselves or by the regulatory body of the database in question, without clinical or radiographic confirmation. 17 As a second limitation, the TriNetX National Database does not allow for the granularity of data that would be required to fully assess the indications for AA in the sampled patients. For example, patients with greater degrees of coronal or sagittal plane deformity or more complex pathology may have received open AA rather than arthroscopic AA, which may have influenced the differences in postoperative complications. Such patients may be at a higher risk for wound healing complications and nonunion. As astutely demonstrated by Wang and colleagues and discussed in detail previously in this manuscript, much of the literature comparing open AA to arthroscopic AA likely suffers from this same limitation. 26 TriNetX does not define diagnosis criteria, so there is no way to determine whether a diagnosis such as nonunion was arrived at from a clinical, radiographic, or intraoperative assessment or whether it could have been miscoded in the postoperative period. The absence of postoperative revision data limits the ability to assess the clinical significance of any coded nonunions. There were no available codes for revision ankle arthrodesis, which prevented an analysis of the number of nonunions, among other complications, requiring revision surgery. Similarly, malunion could not be reliably captured. There was also no access to the operative notes or procedural specifics, preventing a comparison of the techniques, intraoperative deformity findings, or matching between the cohorts. Additionally, the database does not provide exact counts for complications with instances between 1 and 10, preventing a comparison of complications with few instances; nor does it allow for stratification of procedures by inpatient or outpatient setting, which may prevent this analysis from accurately comparing patients’ comorbidity load and negating differences in institutional practices for admitting patients. The lack of coding specificity and knowledge of surgeon or institutional practices may have influenced the distinction in readmission rate. Additionally, we could not account for calendar-year effects (surgical era, perioperative protocols, enhanced recovery pathways) or site clustering, which may differentially influence outcomes across institutions and over time. Furthermore, arthroscopic arthrodesis patients were identified using an unlisted arthroscopy CPT code, which introduces the possibility of procedure misclassification and may have influenced the observed comparisons between cohorts. Finally, because multiple outcomes were assessed across several postoperative time windows, the potential for type I error must be acknowledged. For this reason, the reported P values should be regarded as exploratory, and we encourage readers to focus on the effect sizes and 95% CIs when interpreting our findings. Despite the limitations of the present study, the utility of this work as the first of its kind to our knowledge to use a multicenter database to explore outcomes following open AA vs arthroscopic AA remains intact. For these reasons, these data may be best applied as a preliminary comparison of the techniques and their respective medical sequelae after surgery, rather than definitive evidence of superiority. A more thorough analysis of the postoperative nonunion rates and necessity for revision likely requires access to more granular data that allows stratification by coronal plane deformity.
Conclusion
In the first national or international multicenter database study evaluating differential outcomes of open AA vs arthroscopic AA, we found that arthroscopic AA was associated with lower rates of wound dehiscence, SSI, hospital admission, and ED visits at short-term follow-up intervals. Pseudarthrosis or nonunion rates were higher in the open AA cohort compared with the arthroscopic AA group. Although these data support of the use of arthroscopic AA, the results should be interpreted with caution because of the inability to assess the degree of coronal or sagittal plane deformity in the included cases.
Supplemental Material
sj-pdf-1-fao-10.1177_24730114251386025 – Supplemental material for Arthroscopic Ankle Arthrodesis vs Open Ankle Arthrodesis: A Propensity-Matched, Retrospective Database Analysis of Medical Complications and 2-Year Nonunion Rates
Supplemental material, sj-pdf-1-fao-10.1177_24730114251386025 for Arthroscopic Ankle Arthrodesis vs Open Ankle Arthrodesis: A Propensity-Matched, Retrospective Database Analysis of Medical Complications and 2-Year Nonunion Rates by Albert T. Anastasio, Nicholas R. Kiritsis, Isabel R. Shaffrey, Francois Lintz, Canon Cornelius, Nacime S. Mansur, John Dankert, Conor O’Neill and Cesar de Cesar Netto in Foot & Ankle Orthopaedics
Footnotes
Appendix
Queried Complications and Their Respective ICD-10 Diagnosis Codes.
| Complication | ICD-10 Diagnosis Code(s) |
|---|---|
| Dehiscence | T81.3 |
| Pulmonary embolism (PE) | I26 |
| Acute myocardial infarction (MI) | I21 |
| Cerebral infarct | I63 |
| Transfusion | 30233N1 (ICD-10-PCS) a |
| Deep vein thrombosis (DVT) | I82.40 |
| Surgical site infection (SSI) | T81.4 |
| Hematoma | L76.22 |
| Sepsis | A41.9 |
| Venous thromboembolism (VTE) | I82.40, I26 |
| Any adverse event b | T81.3, I26, I21, I63, 30233N1, I82.40, L76.22, T81.4, A41.9 |
| Pseudarthrosis after fusion or arthrodesis | M96.0 |
Abbreviations: ICD-10, International Classification of Diseases, Tenth Revision; ICD-10-PCS, ICD-10 Procedure Coding System.
Any adverse event: death, dehiscence, deep vein thrombosis, pulmonary embolism, myocardial infarction, cerebral infarct, transfusion, hematoma, surgical site infection, sepsis.
Author Note
Investigation performed at Wake Forest University School of Medicine, Winston-Salem, North Carolina, USA.
ORCID iDs
Ethical Considerations
Ethical approval was not sought for the present study because the TriNetX database uses only deidentified patient information without access to individual, identifiable details.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Albert T. Anastasio, MD, reports disclosures related to manuscript of consulting fees: QPIX solutions. Francois Lintz, MD, reports disclosures related to manuscript of royalties: Newclip Technics, Novastep; shareholder: Paragon28, CurvebeamAI; consultant: Paragon28, Newclip Technics, CurvebeamAI; patent: CurvebeamAI and general disclosures of EFAS: committee member; FFAS: board member; AOFAS: youth committee international embassador; iWBCTs: past president; Foot Ankle Clinics: editorial board; FAI/FAO: reviewer; FAS: reviewer. Canon Cornelius, MD, reports disclosures related to manuscript of consulting fees: QPIX solutions. Nacime S. Mansur, MD, reports general disclosures of board or committee member: Brazilian Orthopedic Foot and Ankle Society, American Orthopaedic Foot & Ankle Society. Cesar de Cesar Netto, MD, PhD, reports disclosures related to manuscript of paid consultant / royalties / medical advisory member: Paragon 28; paid consultant / stock options: CurveBeam; paid consultant / royalties: Medartis; paid consultant: Ossio; paid consultant / royalties: Zimmer-Biomet; paid consultant / royalties: Artelon; paid consultant: Stryker; paid consultant: Exactech; paid consultant / royalties: Extremity Medical; paid consultant / stock options: Tayco Brace; and general disclosures of president: International WBCT Society; editor-in-chief: Foot and Ankle Clinics; committee member: AOFAS. Disclosure forms for all authors are available online.
Data Availability Statement
The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
