Ankle arthritis is a complex condition traditionally managed with arthrodesis, often as a result of traumatic, rheumatologic, or structural etiologies. While ankle fusion remains a reliable surgical option for select patients, it is not without drawbacks. Long-term complications include progressive adjacent joint arthrosis, persistent pain, and suboptimal patient-reported outcomes. For end-stage arthritis of adjacent joints, these challenges frequently necessitate additional interventions, highlighting the need for alternative approaches, such as total ankle arthroplasty (TAA). Advancements in total ankle arthroplasty have demonstrated comparable complication rates to arthrodesis, while offering patients an alternative that spares joint motion.

Revision surgeries for failed ankle fusions, such as tibiotalocalcaneal arthrodesis, carry substantial risks, including nonunion and amputation. With this in mind, the process of converting a previously performed ankle fusion to a TAA has emerged as a promising alternative, with well documented functional and clinical outcomes. First described in 2004 by Greisberg, this procedure remains technically demanding, requiring meticulous patient selection and adjunctive techniques to mitigate risks.

Despite its challenges, a systematic review of the current literature reveals encouraging outcomes, including enhanced range of motion and patient satisfaction. The purpose of this study was to provide a comprehensive evaluation of ankle arthrodesis takedown and conversion to TAA. By synthesizing data on patient outcomes, and complication rates, we aim to provide evidence-based insights to guide clinical decision-making and advance treatment strategies for patients with previous ankle fusion.

A formal systematic review was performed using the Preferred Reporting Items for Systematic reviews and Meta-analysis guidelines. A broad search of online databases google scholar, Cochrane Library, and PubMed was conducted, incorporating the following search terms: ankle replacement OR ankle arthrodesis OR takedown total. Each search term was modified to reflect the different databases. Studies and references of selected articles were screened by the primary author (JK). Any potential articles in question were reviewed by the secondary author (CM) and persistent disagreements were reviewed by the senior author (GB). The earliest documentation of this procedure was performed in July 2004 and thus our search was contained between that time point and July 2024. Articles were included if they discussed the types of implants used, presence of adjacent joint arthrodesis, pre- and intra-operative procedures, post-operative protocols, complications, AOFAS scores, and VAS pain scores. Studies were excluded if they did not explore these outcome measures, were outside of our search timeline, were published in non-peer reviewed journals, in vivo or cadaveric studies, or duplicates. The included studies were assessed for study characteristics and outcomes using a data extraction form. When information was missing it was listed as not recorded.

The primary clinical outcome measures included AOFAS functional scores, VAS pain scores, and patient satisfaction, while secondary outcomes included complications such as intraoperative fractures, flexor hallucis longus tendon injuries, delayed wound healing, symptomatic arthrofibrosis, and treatment failures. For continuous outcomes (AOFAS and VAS scores), effect sizes were calculated using mean differences and their corresponding 95% confidence intervals (CIs). For other measures, effect sizes were estimated by calculating the prevalence of each outcome. Freeman-Tukey transformation was used in estimating the prevalence. Heterogeneity among the studies was evaluated using the I² statistic alongside the chi-square test. A threshold of I² ≥ 50% and a chi-square p-value < 0.05 indicated significant heterogeneity, prompting the use of a random-effects model (DerSimonian–Laird method) to calculate the pooled effect sizes for AOFAS scores, VAS scores, intraoperative fractures, delayed wound healing, and treatment failures. In cases where these criteria were not met, a fixed-effects model was employed for the analysis of satisfaction, tendon injuries (specifically flexor hallucis longus), and symptomatic arthrofibrosis.

Meta-regression analyses were conducted to identify potential sources of heterogeneity, by fitting individual models for sample size, publication year, participant age, follow-up duration, and type of implant (Implants included in this analysis were Hintegra, Agility, Inbone, Salto, Infinity, and STAR). Due to the limited number of studies reporting tendon injuries, regression analysis could not be performed. Subgroup analyses were carried out to assess the moderating effects of these sources on the effect sizes, with continuous measures dichotomized at the mean. Statistical significance was set at p < 0.05. Sensitivity analyses were performed to test the robustness of the findings by sequentially excluding each study and recalculating the pooled estimates. Publication bias was evaluated using funnel plots and Egger’s test. All statistical analyses were conducted using Stata version 18.5, with effect-sizes considered statistically significant at p < 0.05.