Return-to-Play Practices Following Hamstring Injury: A Worldwide Survey of 131 Premier League Football Teams

Between 24th-October-2017 and 20th-March-2018 (2017–2018 season), 310 professional football teams from 34 premier-leagues worldwide were invited to participate. The purpose and procedure of the online survey was explained, and a web-link provided. We requested the survey be completed by the person/s of the science and sports medicine team responsible for the design and implementation of the RTP programme. Institutional ethical review board approval was granted by Edinburgh Napier University (SAS/00014). Confidentiality and anonymity were detailed before consenting to participate.

We sent a maximum of three reminder emails over a 6-week period from the first email invitation. A follow-up email was also sent if there were missing data. If the question/s remained unanswered, it was excluded from analysis.

The survey design and construction followed recommendations on the design and development of surveys [11]. The survey underwent 3 rounds of piloting (for content validity and usability) with 12 experienced applied researchers/practitioners working in professional football (but not from any teams invited to participate). Twelve modifications resulted: four items deleted and eight added. The survey was originally developed in English and translated (using cross-cultural adaptation process recommended by World Health Organisation (WHO) [12] into French, Spanish, German, Italian, Portuguese (+Brazilian Portuguese) and Japanese. The survey was administered online (Novi Survey, http://​novisurvey.​net).

Respondents were asked to consider their RTP practices during the previous season for a typical football-related hamstring muscle injury (time-loss 18 days) [13] when answering all questions in the survey. There were 29 questions (10 closed, 19 open) (Appendix 1 in Supplementary material) organised into four sections, which were adapted (by the steering committee and through the piloting process) for use in football but based on a RTP continuum model:

Each section comprised four parts *(except RTPerf, which only considered parts 1 and 2):

The survey closed on 31st-April-2018. Raw data were exported to Microsoft Excel. To ensure content analysis accuracy, native speakers skilled in translation verified the translation accuracy of answers to open-ended questions where necessary. We used a cross-sectional design and analysed results descriptively according to the checklist for reporting results of internet e-surveys (CHERRIES) [14].

To evaluate the importance of specific criteria, and corresponding test/tool for clearance to the next RTP phase, we assigned rankings [15‐18]. For each continuum phase, respondents specified and ranked in order of importance (1st–3rd) the criteria they considered to determine RTP progression. For each phase, criteria ranked in 1st, 2nd and 3rd position were reported as a frequency (%) of total responses.

To analyse the open-ended questions, we used inductive content analysis [19] following a three-stage process [20‐22]. We treated survey answers as standalone meaning units, unless they contained more than one self-definable point, in which case, each meaning unit was considered and separated. Responses with insufficient information were excluded. For each section of the survey, meaning units generated from responses pertaining to each question were listed, before being compared for similarities and organised into raw data themes. Raw data themes were grouped for each question into larger and more general themes/categories in a higher order concept [21]. We continued refining the data until theoretical saturation [23].

To enhance our confidence in interpreting the data, two independent authors (GD and AM) read the lists of meaning units at least twice [24]. They discussed meaning units, categories and themes at each stage to reach a consensus regarding data accuracy and clarity. Sample data sets were re-examined by a third independent researcher, blind to the research aims, to audit the assigned categories and themes to ensure they accurately reflected the standalone meaning units [25].

In total, 124/131 premier-league teams (95%) reported following a return-to-play continuum model. Of the 124, 27 (21%) teams did not report following a ‘return-to-performance’ phase (RTPerf).

For both RTRun and RTTrain phases, all teams used a criterion-based approach. At RTPlay 7 (5% of 131) teams reported they did not use specific criteria to determine a player’s clearance. This increased to 27 (21%) teams at RTPerf (Fig. 1). Table 2 provides an overview of the specific criteria used by teams and the level of importance given to guide progression at each phase of the continuum.

We included 378 out of 393 (96%) responses, i.e. 131 responses × 3 main RTP phases. Across each phase, the response rate of teams was 130/131 (99%); 128/131 (98%) and 120/131 (92%) for RTRun, RTTrain and RTPlay, respectively. When returning to RTRun, a frequency of 100% was reported by 68 (52%) teams (i.e. all intended criteria were met before the player was cleared to progress by 68 teams). By comparison, 55 teams at RTTrain and 36 at RTPlay reported with 100% frequency to always successfully meeting the criteria set. The frequency range (%) with which teams successfully reported to achieving all of the intended criteria is displayed in Fig. 2.

We analysed 389 out of 393 (99%) responses. Per phase, 131/131 (100%) teams responded for both RTRun and RTTrain while 127/131 (97%) answered at RTPlay. Overall, 105 (80%) teams use a shared decision-making approach involving at least two people. Table 3 represents the contribution of key staff members to decision-making based on the position (i.e. medical or science) of the practitioner who completed the survey (Table 3).

Challenges relating to team hierarchy (e.g. pressure from management) were regarded the most likely to influence practitioner decision making (27% of challenges cited) (Table 4). As a player transitioned to RTPlay, match related factors were more prominent. We excluded 132 responses (100 due to a blank response; 32 due to an error in the cross-cultural translation in the Spanish version that was not picked up during final piloting—the question could have been misinterpreted). Twelve teams stated that challenges were not applicable as every player must have met all criteria before being cleared.

Based on our sample of premier-league teams worldwide, the majority (124; 95%) assessed criteria over a continuum to guide RTP following hamstring injury. Of 124 teams, 102 (78%) reported assessing criteria at the four specified phases; RTRun, RTTrain, RTPlay and RTPerf. Of the remaining 29 teams, 22 (17%) implemented a criteria-based approach at RTRun, RTTrain and RTPlay, but not RTPerf. Unfortunately, the teams did not provide sufficient details for us to confidently report why this was the case; however, of the minimal feedback we did receive, it was specified that they believed the RTPlay phase should be where the player is also considered to be back to full performance.

Seven (5%) teams did not follow a RTP continuum and did not explain why. Our findings provide preliminary support (at least in our sample) that general research recommendations and practice align in that the majority of team practitioners view RTP from the point of injury until at least returning to play and most through until returning to desired performance. Our RTP continuum differs from the one specified in the 2016 consensus statement. In particular, we had specified an additional phase early in rehabilitation (RTRun). Football (and sport in general) and research are constantly evolving, and the application of a continuum framework within and between sports may need to be adapted to the specific needs of those monitoring and controlling the overall RTP process. Therefore, models such as the RTP continuum may need to be adaptable to suit these needs and research should consider this also.

Team practitioners used a combination of clinical, functional and psychological criteria to guide RTP following a hamstring muscle injury. Multifactorial and criteria-based rehabilitation programmes are advocated in research to support RTP decision-making [26‐28]. Such a criteria-based decision approach provides practitioners with an individualized approach to RTP that integrates quantifiable assessment (objective and subjective) to systematically progress rehabilitation. Criteria-based approaches may reduce re-injury risk and improve player performance and availability of footballers [26, 29]. In our survey, we asked respondents to specify their top three most important criteria used at each of the RTP phases (Table 2) with the aim of uncovering some consistently used criteria, metrics and thresholds that could inform current practice and guide future research.

A shared decision-making approach was used by 80% of premier-league teams surveyed. This is an encouraging finding as low-quality internal communication may be associated with (re)injury rates and reduced player availability [17, 51, 52]. Only eight (6%) teams reported using isolated decision-making across all continuum phases. Eighteen (14%) teams used a combination of isolated and shared approaches to guide rehabilitation progression.

Medical staff (club doctors and physiotherapists) were most frequently consulted throughout the decision-making process. Traditionally regarded as the gatekeepers of the RTP decision, medical staff clearly hold a prominent role within the decision-making practices of clubs. In 96 teams (73%), medical staff were the lead practitioner responsible for the RTP programme. Across each phase of the RTP continuum, ≥ 87% of teams consulted with at least one medical practitioner (Table 3).

While medical staff involvement in decision-making across all RTP continuum phases was reported by both medical and science practitioners surveyed (Table 3), their perceptions as to how other stakeholder groups are involved in decision-making throughout RTP differed. Specifically, medical staff reported less involvement of science and coaching staff across all phases of the continuum and for players at RTTrain and RTPlay compared to when science staff answered the survey. We cannot answer why this is, as potential bias for respondents placing greater emphasis on the involvement of their own discipline should then have also been evident in the responses of science staff, yet this was not the case. Our results raise an important question about how staff are actually involved in the RTP continuum process. Despite an initial encouraging finding that the RTP decision-making is shared among stakeholders, the inconsistency found in the composition raises some potential concerns about the specific dynamics of the communication among staff.

Premature RTP has been suggested as a possible risk factor for re-injury [41, 53‐55]. Throughout the RTP continuum, surveyed practitioners highlighted encountering various challenges capable of influencing their decision-making (Table 4). When progressing through the RTP continuum following hamstring injury, team practitioners reported that there were occasions when the player did not meet all of criteria set (Fig. 2). However, these occasions were not common. Typically, teams met the criteria they set ≥ 90% of the time, yet, the variations demonstrate the reality of the practical setting where it is not possible to achieve this all of the time.

Each injury case must be assessed individually, based on a risk assessment. Accordingly, the risk associated with accelerating a player’s RTP to ensure availability for a decisive fixture may be more readily accepted in the case of the key 1st team player as opposed to the promising youth team prospect—who might be afforded a longer RTP timeframe to reduce reinjury risk. While surveyed teams predominantly displayed a high degree of success in achieving criteria, this finding reflects only one muscle-group (hamstring). Therefore, we do not know if this is representative of rehabilitation across other muscle-groups or injury types.

An inherent limitation of survey-based research is its lack of external validity owing to low response rates. One hundred and thirty-one (42%) of 310 invited teams completed the survey. Accordingly, caution should be exercised when interpreting or generalizing these results, as the extent to which they characterise the perceptions and practices of the non-responding teams is unclear. How these findings extend to other levels of competition (professional vs. amateur), genders, different age groups (senior-level vs. academy-level) and other muscle-groups or injury-types is also unknown and warrants consideration in future research. Representing current opinion (level 5 evidence), we acknowledge our findings may change with emerging evidence and paradigm shifts. Therefore, the perceptions and practices of practitioners should be re-evaluated in the future, based on new research recommendations. While sampled clubs appear to display a high degree of success in meeting their outlined criteria, a perceived limitation (although not a specific focus of our survey) could be that we did not ask practitioners to elaborate on instances where RTP was accelerated without achieving criteria. It is not known if, in these instances, re-injury occurrences predominantly occurred. We also acknowledge that survey responses correspond only to the perceptions and practices of science and medical practitioners responsible for the return-to-play programme. It is possible that responses could vary according to the position of the stakeholder surveyed while the perceptions of other key stakeholders’ groups involved in decision-making (e.g. managers, players) were not considered. We could not compare cultural differences as participating clubs from different confederations/leagues were not equally represented. Further investigation adopting techniques capable of facilitating a more comprehensive picture (e.g. qualitative focus groups, individual interviews etc) of how specific metrics and thresholds inform return-to-play decision-making is required.