Response to the Letter to the Editor regarding the meta-analytic evidence on age-stratified ACL repair outcomes, with a commentary on reporting beyond the evidence hierarchy

Abstract

We would like to thank Rao and Yang for their interest in our article, ‘Anterior cruciate ligament primary repair revision rates are increased in skeletally mature patients under the age of 21 compared to reconstruction, while adults (>21 years) show no significant difference: a systematic review and meta-analysis’ [19]. We appreciate their recognition of our methodology and the clinical relevance of our findings.

While we welcome their critique, many of the points raised in their letter to the editor [32] reflect limitations we acknowledged and addressed in our manuscript. Some concerns appear to rise from incorrect interpretations or may not fully consider aspects of our analysis. In this response, we aim to clarify key points and further explain our rationale in a constructive scientific exchange.

We acknowledge the general limitations of meta-analyses, especially in a field like anterior cruciate ligament (ACL) repair where clinical adoption is evolving and high-level evidence remains limited. Accordingly, sound study design and transparent reporting are essential. Our study aimed to navigate the challenges of synthesising data from a developing field and acknowledged the predominance of retrospective designs and demographic heterogeneity. To address these limitations, we: (1) restricted inclusion to Level I–III for greater evidence homogeneity; (2) contacted all study authors for unpublished demographic and outcome data, improving data quality and enabling subgroup analyses and (3) performed an extensive risk-of-bias analysis to support a balanced discussion of current strengths and limitations. Despite these efforts, as stated in our manuscript, selection and publication bias remain possible. However, a systematic review by two independent reviewers ensured comprehensive inclusion of eligible recent studies. In response to the suggestion that ‘future meta-analyses should expand the literature search, including more high-quality prospective studies’, we would have welcomed references to specific overlooked studies to improve future analyses and the ongoing scientific dialogue.

Regarding the critique that the meta-analysis did not adequately explore heterogeneity and bias, and tools such as Egger's test [7] should have been included, we refer readers to our results section ‘Level of Evidence and Risk of Bias Assessment’, specifically figs. 6–7 and tabs. 9–10, which include funnel plots, Egger's test, and assessments using both RoB 2 [23] and MINORS tools [22]. The risk of bias analysis was even performed by two evaluators and to assess reliability of the rating the inter-rater reliability was calculated and presented as excellent for both the MINORS and Rob 2 assessments, with 0.8 and 0.9, respectively. We believe the authors have missed this in our study.

Rao and Yang also state that ‘the interpretation of certain unexpected findings lacks depth’. However, the higher revision rates in patients ≤21 years were not unexpected. This threshold was previously defined by Vermeijden et al. [31] and confirmed by Ferreira et al. [8] in matched cohort studies comparing ACL repair and reconstruction. Similarly, elevated reoperation rates due to removal of hardware have been reported for ACL repair using the dynamic intraligamentary stabilization (DIS) implant [2, 10, 12, 14], while ACL primary repair (ACLPR) and repair with the bridge enhanced ACL restoration (BEAR) scaffold have generally shown lower rates [6, 8, 9, 13, 17]. Notably, no prior review has stratified revision and reoperation rates by age and surgical technique across ACLPR, DIS, and BEAR. This stratified, technique-specific approach constitutes the novelty of our work. Furthermore, the comment of lack of depth in our analysis seems surprising as our meta-analysis has uniquely contacted the authors of the included studies, obtained full datasets when available, and performed rigorous meta-analyses beyond the standard meta-analyses published in the orthopaedic literature.

More specifically on the authors' comment regarding the higher reoperation rates following DIS, van der List et al. [25] suggest possible confounders such as selection bias and the surgeon's learning curve. While we agree these may contribute, the learning curve is a known limitation of all new surgical methods, including ACLPR and BEAR. Crucially, strong evidence links the DIS implant's design to higher reoperation risk [2, 10, 12, 14]. Moreover, the release of a second-generation, smaller implant by the manufacturer supports the view that implant-specific issues significantly contributed to these outcomes.

We also appreciate the mention of biological and behavioural factors in younger patients—such as more advanced ACL degeneration, higher postoperative activity, and metabolic differences. While important, these factors were outside our study's scope and remain inconsistently reported in the literature. We agree this is a key area for future research to improve treatment for this high-risk population.

Lastly, regarding the LTE's repeated call for higher-level evidence: while Level I studies are critical for validation, surgical innovation typically starts with lower levels of evidence and matures over time, an evolution supported by the IDEAL framework [16]. Since 2008, our group has published over 70 studies on ACL repair, contributing short- [31], mid-term [3], and soon first long-term data, while continuously refining indications and techniques [4, 5, 18, 24, 26-29]. We argue that research value lies not just in study level, but in methodological rigour, transparency, and clinical relevance. As van der List et al. [25] noted, only 32% of meta-analyses are limited to Level I–II studies. Rather than dismissing retrospective research, we advocate for its thoughtful inclusion when conducted and reported with care. Dismissing data solely by evidence level oversimplifies early-phase research's role in innovation. It should be emphasised that we cannot solely rely on Level I evidence. Members of our group are involved in designing or participating in Level I studies on this topic [11, 30], but these rigorous scientific studies take time and may not always reflect clinical practice [1, 15, 20, 21], and it is important to improve our patient selection. Our group was the first to recognise that isolated primary ACLPR in young patients has a high risk of failure rate [31], which was later confirmed by others, and such findings are critical to improve our outcomes [8]. Solely relying on randomized crontrolled trials (RCTs) will delay reporting such findings, and studies such as this meta-analysis help us to improve outcomes for our patients and redefine indications for such RCTs.

In conclusion, we value this scientific discussion and the insights shared by Rao and Yang. We hope this dialogue promotes continued improvements in the transparency and quality of meta-analyses. Crucially, we believe that the hierarchy of evidence should not be confused with a hierarchy of truth. Study design and methodology should align with the research question, and data must be critically analysed and interpreted. These principles, we believe, are the true foundation of rigorous and impactful clinical science, laying the groundwork for disruptive scientific research.

Gregory S. DiFelice receives royalties, owns stock and is a paid consultant for Zimmer Biomet, and receives royalties from Arthrex. He received stock options, provides consulting services and participates in funded research with Miach Orthopaedics. Gregory S. DiFelice received stock options and provides consulting services for OSSIO Inc. The remaining authors declare no conflicts of interest.