Introduction

In comparison to the first anterior cruciate ligament (ACL)-rupture risk, the risk of suffering from a subsequent injury after (ACL) reconstruction increases at least tenfold [1, 2]. More detailed, a cumulated (ipsi- and contralateral, re-injury & graft failure) recurrence risk of 10–25% is reported in the literature [1, 2]. Such secondary injuries often occur during the return to sports (RTS) process [1].

It is a major goal of the RTS-process to lead athletes back to activity, training, and competition without exposing him/her to an excessively high risk for a subsequent rupture [3]. Important criteria to be fulfilled for an RTS release, beyond morphological graft healing and psychosocial readiness, is the restoration of neuromuscular and motor function injury [4,5,6]. This function is mostly assessed by a combination of simple clinical tests, dynamic strength, and hop/jump tests. Functional deficits or limb asymmetries, where the affected leg’s performance is compared to the putatively unaffected contralateral leg’s performance, seems to be predictive for a second ACL injury [4,5,6]. Improving or even restoring these functional abilities vice versa decreases the deficit; in this scenario, that leads to a decrease in the subsequent injury risk [7].

The biology of graft healing and maturation is of great importance during the continuum of rehabilitation, RTS, and re-injury prevention. Based on the individual development in biological healing, functional skills, and psychological readiness, the time slot before RTS is variable [8]. Although it is not possible to define fixed time points at which a certain goal or functional ability should be reached, time (both before and after the reconstruction) is, nevertheless, one factor to consider after ACL reconstructions [9]. It is, for example, likely that the graft healing will take more time than the time until RTS success [10].

Numerous graft types in ACL reconstruction are adopted. In a recently published survey, 90.4% of the interviewed surgeons preferentially used hamstring tendon autografts for most ACL reconstruction followed by bone-patellar tendon-bone grafts [11, 12]. Hamstrings tendon autograft is thus, in particular in Europe, the—by far—most used graft type. A comparably new graft type is based on quadriceps tendons. Quadriceps tendon autografts show comparable clinical and functional outcomes and graft survival rate like other autografts, but significantly less harvest site pain when compared to patella autografts [13]. Further, quadriceps tendon autografts may lead to better functional outcome scores when compared with hamstrings autograft [13].

Conclusively, a multitude of individual and spatiotemporal factors interact during the rehabilitation, RTS, and re-injury prevention processes after ACL reconstruction. Beyond those highlighted so far, age, sex/gender, pain intensity/perception during performance, and concomitant knee injuries like Meniscus and collateral ligament injuries, or even unhappy triad must also be taken into account when the neuromuscular function after ACL reconstruction is rated [14,15,16].

Considering the multitude of factors with the aim to derive individual courses of functional abilities and the impact of major contributors to these abilities is helpful for the for the management of deficit-oriented function-based rehabilitation strategies after ACL reconstructions [17, 18]. Aiming to provide such a more general view on the course of functional abilities after ACL reconstruction, the purpose of our cohort study was to evaluate the contribution of time between injury and surgery, time since reconstruction, age, gender, pain, graft type, and concomitant injuries as isolated and interactive contributors to inertial sensor-assessed motor function after ACL reconstructions in a multiple linear mixed model regression approach. We hypothesized that numerous of the potential contributors interact with each other in their way to interactively impact on different functional abilities.

Methods

Study design, ethics, and informed consent

In this multicenter cohort study, all methods were performed in accordance with the relevant guideline. Data were extracted from a data registry. The registry is the nationwide database from an enterprise (OPED GmbH, Valley, Germany). The register was initiated in January 2018, data from initiation until October 2020 were included in this analyses. As all data was retrieved completely anonymized from a registry, ethical approval is not relevant for this type of analysis.

Informed consent was obtained from all participants and (below 16 years of age) from their legal guardian. All data were assessed as a part of the functional assessment during the rehabilitation after ACL-reconstruction, no measurement or measures for study purposes were additionally undertaken.

The database consists of prospectively assessed multiple, in particular functional, measurements. More detailed, the measurements included in the study were used to assess the participants function during their formal medically prescribed standard rehabilitation process.

Inclusion and exclusion criteria

The data from all database patients (children, adolescent, adult males, females, and diverse) with an acute unilateral ACL rupture with or without concomitant ipsilateral knee injuries (meniscal tear, lateral ligament involvement, unhappy triad) and having passed an arthroscopically applied, anatomical reconstruction was included. Main exclusion criteria were bilateral lower limb injuries, other major injuries than ACL tears with exception of secondary knee injuries, pregnancy, and severe diseases potentially affecting motor function.

Independent outcome variables.

The following potential outcome modifiers were extracted for each participant and at each of the repeated measurement: age range (0–15, 16–20, 21–25, 26–30, 31–40, 41–50, and above 50 years)), gender/sex (male, female, divers/unknown), time since reconstruction [days], time between injury and reconstruction [days], concomitant intra-articular injuries (isolated ACL tear, meniscal tear, lateral ligament, unhappy triad), graft type (hamstrings, patellar, or quadriceps tendon autograft), and pain intensity during the measurement (visual analogue scale 0–10 cm).

Dependent outcomes: functional tests

All functional tests were performed from experienced personnel (athletic trainer, physiotherapists, sports medicine or orthopedic physicians, or sports therapists). A standard operating procedure and test manual is used to perform the standardized test battery. The functional tests display increased chaos, starting from high control angle reproduction tasks to high chaos such as speedy jump [19]. The test were selected by the same experience assessors based on standard principles of function (and time-) based rehabilitations after ACL reconstructions [18]. Details on the outcomes, the underlying function, the tool used, the conduction, and the testing criteria are displayed in Table 1.

Table 1 Overview of the functional outcomes. Each functional ability is described by the corresponding outcome and tool used to assess the ability, its test quality criteria, conduction and the (positive) decision criteria
Full size table

Outcomes assessment

All joint position sense, vertical jumps and hops measurement, side hops, and the speedy jump outcomes were assessed using a single inertial sensor (Orthelligent Pro, OPED GmbH, Valley, Germany). The non-invasive external three-dimensional wireless sensor was positioned at the highest circumference of the lower leg using an elastic band; the sensor itself was placed on the tibia. Using inertial sensor techniques may detect re-injury associated movements behavior more adequate than the “classic” outcomes of functional return to sport testing [29].

The sensor consists of a 9-axis MEMS MotionTracking device (TDK InvenSense, Chūō, Tokio, Japan); with 3 accelerometers (measurement range ± 2 g to ± 16 g), a 3-axis gyroscope (± 250 to ± 2000 degrees per second), and a 3-axis magnetometer. The device was zeroed prior to each measurement.

Sample rate was (accelerometer) 4.5 kHz to (Gyroscope) 9.0 kHz. The data was down-sampled (4:1) and filtered for the further analysis. A low pass and Kalman filter was applied.

The tool, all test settings and outcomes, and the setup has been validated against a gold-standard movement assessment system and was found to be valid in terms of or the objective assessment of movements of the lower limb [30].

Statistical analysis

The statistical analyses were performed blinded to the data retrieval from the database.

Range data plausibility check was undertaken for all independent and dependent outcomes; the data were cleared accordingly.

Repeated measures linear mixed models (multilevel analysis) investigated the impact and interaction of the individual (random effects) predictors (level 2), and of time (level 1) on the functional outcomes’ values. 2-LL estimates were adopted to build the models. Due to the considerable amount of missing information on the graft type, these analyses were, partially, contributed separately. In detail, the analyses which were once modelled including and once without the graft type are highlighted with an empty row (in the tables) between the graft type and the other contributors. The size of the estimates highlight the size of the effect of the independent on the dependent variable (always in the units used), the direction is indicated by the leading sign (minus indicates a negative association).

All analyses were performed in SPSS version 25 (IBM Corporation, New York, NY, USA), an alpha-error of 5% was considered as a relevant cut-off significance value, all p-values below are interpreted as statistically significant.

Results

The entries from 1629 persons were screened. Data from 1441 individuals were included into the final analysis. Exclusion reasons for the others were: duplicate entries (n = 48), critical data (identity for repeated measures) missing (n = 84), non-ACL-tear (n = 56). For those included, graft type distributions was as follows (always displayed as absolute numbers and percentage share): hamstrings (semitendinosus with or without gracilis tendon): n = 566 (39.3%); patellar tendon graft: n = 40 (2.8%); quadriceps tendon graft: n = 139 (9.6%); unknown: n = 696 (48.3%). Secondary injuries /issues distribution was: isolated ACL rupture: n = 938 (65.1%); lateral ligament (Ligamentum collaterale tibiale) involvement: n = 70 (4.9%); meniscal tear: n = 414 (28.7%); unhappy triad (Combined ACL, Meniscus medialis, and Ligamentum collaterale tibiale tea): n = 15 (1%); unknown: 4 (0.3%).

The samples’ mean age was 29.4 years (standard deviation 11.8 years), 592 females and 849 males with a mean body mass index of 24.8 kg/m2 (standard deviation 3.5 kg/m2) were analyzed. Mean time passed between the injury and the reconstruction was: median 50 days, mean 135 days (standard deviation 193 days).

Figure 1 displays the individual data for the kinesthesia/joint position sense and postural balance measurements (angle reproduction and y-balance test). The corresponding analysis’ outcomes are highlighted in Table 2. Mean pain intensity during measurement was 0.51 points (standard deviation 1.32 points) for the reconstructed side. Variance explained by interindividual differences was 7% (angle reproduction), and 29% (Y-Balance).

Fig. 1
figure 1

Scatterplots for kinesthesia (angle reproduction, above) and postural balance (y-balance test, below). Individual data (dots) for the days since reconstruction (x-axis) and the values of the outcomes (y-axis), separated for females and males, are displayed

Full size image
Table 2 Estimates, confidence intervals, and statistical outcomes for the mixed models for s kinesthesia and balance (A: kinesthesia, angle reproduction error, B: postural balance, Y-Balance composite score); both times separated by the reconstructed and unimpaired leg
Full size table

Figure 2 displays the individual data for the vertical jump and hop measurement (Drop jump and vertical hop). The corresponding analysis’ outcomes are highlighted in Table 3. Mean pain intensity during measurement was 0.36 points (standard deviation 1.05 points). Variance are explained by interindividual differences was 29% (Drop jumps), and 3% (Vertical hop).

Fig. 2
figure 2

scatterplot for the vertical jump and hop measurement drop jump (above) and vertical hop (below). Individual data (dots) for the days since reconstruction (x-axis) and the values of the outcomes (y-axis), and the corresponding regression curves are displayed

Full size image
Table 3 Estimates, confidence intervals, and statistical outcomes for the mixed models for the vertical jump and hop measurements: drop jump (above) and vertical hop (below); both times separated by the reconstructed and unimpaired leg
Full size table

Figure 3 displays the individual data for the horizontal jump and hop measurements (speedy jumps, side hops, and single leg hops for distance). The corresponding analysis’ outcomes are highlighted in Table 4. Mean pain intensity during measurement was 0.42 points (standard deviation 1.12 points) for the reconstructed leg only. Variance are explained by interindividual differences was 43% (speedy jumps), 14% (side hops), and 7% (single leg hops for distance).

Fig. 3
figure 3

Scatterplots for the horizontal jump and hop measurements speedy jumps, side hops, and single leg hops for distance. Individual data (dots) for the days since reconstruction (x-axis) and the values of the outcomes (y-axis), separated for females and males, and the corresponding regression curves are displayed

Full size image
Table 4 Estimates, confidence intervals, and statistical outcomes for the mixed models for horizontal jump and hop measurements speedy jumps (A), side hops (B), and single leg hops for distance (C); each time separated by the reconstructed and unimpaired
Full size table

Discussion

Function after anatomic ACL reconstruction is influenced by several interacting factors. Adopting a repeated measure cohort design, we evaluated the contribution of time between injury and surgery, time since reconstruction, age, gender, pain, graft type, and concomitant injuries as isolated and interactive contributors to inertial sensor-assessed motor function after ACL reconstructions in a multiple linear mixed model regression approach. We found that many contributors interactively impact on different functional abilities during the RTS-process after ACL reconstruction. When (all analysis performed in omnibus models) other relevant contributors are considered, the reconstructed leg function was associated with the time from injury to reconstruction (angle reproduction error, vertical hopping height, side hops) and positively with the time passed since reconstruction (vertical hopping height, side hops, single leg hop for distance). A further negative contributor to the reconstructed leg’s performance was pain intensity. Angle reproduction was worse when the lateral ligament was involved and better outcomes in the Y-Balance composite score and vertical hopping height were observed in patella grafts reconstructed knees. Most of the outcomes, independent of the leg, were also different between sexes/genders (males often showed larger function values), and affected by increasing age (negatively). The unimpaired leg was, further, mostly influenced by the time between injury and reconstruction (negatively, vertical hopping height, and side hops) and time since reconstruction (positively, Drop Jump knee displacement, vertical hopping height and side hops); but also, in parts, from the graft type of the reconstructed leg ( better outcomes in the single leg hop for distance after patella tendon graft).

Knowing these contributors, their contributive value (estimate), and their interaction is helpful for the function-based and deficit-oriented rating and management of rehabilitation and RTS strategies. Improving or even restoring functional abilities and thus decreasing the identified deficit may consequently lead to a decrease in the subsequent injury risk [7].

As the putatively unimpaired leg is affected by the injury, reconstruction, and all the measures after the injury, too, an individualized and side-dependent comparison of the function after ACL-reconstruction may be more accurate than, as it is usually done, a rating the functional status during RTS using limb symmetry indexes (LSI). The LSI approach may overestimate the knee function after ACL reconstruction [31]. This critique against the LSI is not new. Our findings, however, add a somewhat new aspect to this discussion. When the LSI and not each leg is considered, patient demographics or even intra-operative predictors do not predict the achievement of a symmetrical muscle function [32]. Our findings showed that the putatively unimpaired leg is affected by (in parts) different, but still injury-related variables, than the ACL-reconstructed leg such as graft type. This is another hint for the fact that an ACL injury is rather a global central than a local peripheral problem is given [33,34,35].

A function rating by estimated preinjury capacity level calculations be more constructive than LSIs-based ratings [31]. Supported by the present findings, an isolated view on the change over time of the functional abilities of the reconstructed side may be equally constructive. Despite the critique, LSIs are, nevertheless, predictive for a second ACL injury [4,5,6]. The comparable small effect sizes of the prediction of a second injury might, in parts, be explained by the limited value of LSIs in performance and biomechanical measures. Targeting the identified impairments in functional ability directly may reduce ACL injury risk in healthy limbs in male athletes playing level 1 sports [36].

Our findings of the relevance of the time span between injury and reconstruction is, generally in accordance with the most recent evidence. Here, the authors conclude, that “early ACLR was superior to elective delayed ACLR in terms of the Lysholm score at 2 years and the IKDC score” [37]. In contrast, it is also important to notice that most of the clinical outcomes were not different between early and late reconstruction [37]. As the measures between the ACL injury and reconstruction, like pre-operative rehabilitation affects function, they must be considered, likewise [38]. We found that the way how the time between injury and reconstruction potentially affects post-surgery functional outcomes is not equal between sides: of the unaffected leg, side hops were negatively, vertical hops positively associated with time between injury and reconstruction. A potential explanation may be found in the injury-related reduced afferent neural input from the injured side and the herewith associated globally reduced motor control output [34]. In a more complex motor control task like side hops, the limited output may be more relevant than in a somewhat simple strength/acceleration task like vertical jumps. Speculatively, this may leads to the decrease in side hops performance. Here, prehabilitation strategies may be more helpful to restore the functional ability than the potential impairment due to the lack of afferent input to the central nervous system. This assumption is highly speculative but leads to interesting experimental future study rationales. The time between injury and reconstruction is derived from many factors. Inter alia, non-coping athletes are often reconstructed before the restoration of their preinjury functional abilities [39].

Although the time passed since reconstruction was found to be a relevant predictor of function, performing one single assessment at the hypothetical end of the RTS process is not constructive as, of course and once more proved by our results, many other factors contribute to the final functional outcome [40]. Multiple repetitive measurements, aiming to monitor and verify the course of the RTS process, is more promising [40]. These repetitive measurement approach over time both considers time and (functional) status factors and was found to feasible in an athletic RTS-setting [41].

Numerous reconstruction-specific factors were also important. Previous findings of better functional outcomes after quadriceps tendon autografts, when compared with hamstrings autograft, could not be reproduced: both graft types showed comparable functional values (hamstrings even slightly better in Y-balance) [13]. Patella graft was found to be associated with better outcomes than the other graft types. When these findings are rated, one must consider that patella autografts may lead to severe reconstructed-site pain [13]. As pain was considered as independent variable in the present analyses, higher pain values during objectively equal functional tasks lead to an increase in the total model values. The impact of concomitant (meniscus and collateral ligament injuries, or even unhappy triad) injuries on the functional outcomes is, generally, in accordance with current literature [15, 16]. Reasons for this association can be found in a certain accumulation of injury- and surgery-derived functional deficits of the two (or more) injuries [42, 43].

Limitations

The association of function and age, sex/gender pain intensity/perception during performance is not surprising, but must also be considered when function should be rated more holistic and cumulated in one model [14]. Here, the interactive calculation of the various factors in total models where the interrelationship can be operationalized (and not only single contributors to a dependent variable), must be considered as a strength of this analysis. However, we only reported associations/observations and no experimentally derived effects. That must definitely be considered as a major limitation. It is, for example, not always known why a certain measurement was undertaken at a certain time point. Furthermore, the tests themselves are (mostly) valid and reliable, the objectivity (inter-rater-reliability) of performing them in a non-laboratory clinical setting is not. Numerous further factors are known to contribute to post-reconstruction function. Here, pre-injury functional status and level of physical activity as well as the amount and type of pre- and post-surgery rehabilitation, or further functional outcomes such as strength, and more chaotic hop tests are mentioned as the (potentially) most relevant.

Conclusion

Numerous factors such as time between injury and reconstruction, time since reconstruction, age, gender, pain, graft type, and concomitant injuries are predictive for the individual values and courses of functional abilities after ACL reconstruction. Some of the contributors to motor function such as rehabilitation measures and time until surgery can be modified. Other contributors, such as age, gender, and concomitant injuries cannot be impacted. They, however, must be considered when the post-reconstruction function is rated. It might not be enough to assess factors isolated; the knowledge on their interactive contribution to motor function is helpful for the management of the reconstruction, deficit-oriented function-based rehabilitation, and individualized return to sports strategies.