Article Text

Download PDFPDF

Return-to-sport testing following ACL reconstruction revisited
  1. Kate E Webster1,
  2. Timothy E Hewett2
  1. 1 School Of Allied Health, Human Services and Sport, La Trobe University, Melbourne, VIC, Australia
  2. 2 Hewett Consulting, Rochester and Minneapolis, MN, USA
  1. Correspondence to Kate E Webster, School of Allied Health, Human Services and Sport, La Trobe University, Bundoora, VIC 3086, Australia; k.webster{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In our recent systematic review and meta-analysis,1 one of the questions posed was: Is passing return-to-sport (RTS) test batteries associated with reduced rates of subsequent knee injury?1 The aim of this editorial is to comment on debate2 about the methodology used to avoid further confusion with regard to relevant data and conclusions that can be drawn from the existent literature. Concerns raised centred on two aspects: appraisal of risk-of-bias and pooling data from heterogenous samples, both of which impact on study selection. These are discussed separately.

Risk-of-bias (quality assessment)

For risk-of-bias assessment, two tools provided by the National Institutes of Health were used. While one could argue the semantics of using the terminology ‘risk-of-bias’ versus ‘quality assessment’—we acknowledge that we used both—the content of the assessment was clearly in line with what Capin et al 2 suggest should be included in such an assessment. Full details of the assessment can be accessed in the online supplementary appendix to our article. It is possible that this appendix has been overlooked.

It must be understood that the two studies excluded by Capin et al 2 satisfied more criteria than some studies which were not excluded in their editorial. To understand this discrepancy, one would need to know the process Capin et al 2 used to determine risk-of-bias in their re-analysis. Unfortunately, such details were not provided, and therefore, one cannot tell whether their re-analysis is any less subject to risk-of-bias than the review that they criticised.

Pooling data from heterogenous samples

In our meta-analysis, all eligible studies were considered, as the intent was to provide an overview of all available evidence. As such, there was variation in, for example, patient demographics and the RTS tests used.1 It is reasonable to think that this would be the case. This was fully acknowledged and the potential limitations and implications were discussed.1

Regarding the re-analysis performed by Capin et al,2 the three included studies are equally heterogenous. In both, our original analysis and the re-analysis, the study by Kyritsis et al 3 makes a large contribution to the weighting of the meta-analyses for ACL graft rupture, 63%1 and 83%,2 respectively. Therefore, regardless of which data set one reviews, this single study conducted in ‘elite’ male athletes largely drives this result.

Study selection

We set strict inclusion/exclusion criteria regarding minimum requirements for RTS testing protocols as well as the patient groups in whom these assessments were conducted. All studies that met the predetermined criteria and had the required data were included. Study selection was, therefore, the result of the application of a rigorous systematic strategy in which all potential studies were equally assessed by the same criteria. We are highly uncertain about the validity of a re-analysis that removes studies, apparently ad hoc, with the application of different inclusion/exclusion criteria (some of which cannot be directly applied to the text of the published studies) without giving all potential studies that result from a systematic search strategy equal chance of having the same criteria applied. This could be viewed as ‘cherry-picking’ of studies and data sets.

The two studies4 5 Capin et al 2 excluded, they cited as not designed to answer our research question, which is incorrect. While one could argue subtleties, both contained data that were pertinent to the question regarding second injury risk. The inclusion of the study by Sousa et al 4 has drawn specific criticism. Here, we make two points:

First, concern was raised about potential differences in sports exposure between patients who passed 6-month RTS testing and those who failed. We acknowledged this fact, as did the original study authors. However, this limitation is not unique to Sousa et al.4 For most eligible studies, it was unclear whether patients who failed RTS testing took the advice to wait.

Second, a close look at the Kaplan-Meier survival curves in Sousa et al 4 shows that there are a few early second ACL injuries in that cohort. As such, the impact of any potential difference in early exposure would be minimal. Given that no study formally documented sports exposure, it was not possible to apply uniform inclusion/exclusion criteria related to this parameter across studies.

We also note that CIs reported by Capin et al 2 for our analysis of any knee injury are incorrect. In figure 2, we reported a risk ratio of 0.28 (0.07 to 1.21).1 Although this correctly equates to a 72% decrease in risk, the CI indicates that this could be from a 93% reduction to a 21% increase, not the 6%–96% lower risk Capin et al 2 reported.


Two studies3 4 have shown significant results with regard to passing RTS testing and subsequent ACL injury and their inclusion in any meta-analysis dominates the overall result. It is unclear how generalisable either of these studies is. Until further data are available, caution should be exercised when using information from RTS testing to provide advice to patients regarding their risk for subsequent injury.



  • Correction notice This article has been corrected since it published Online First. The affiliation details have been updated.

  • Contributors Both the authors contributed to the reporting for this manuscript, and read and approved the final submitted manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.