Functional rehabilitation strategies for enhancing neuromuscular control after ACL reconstruction: A systematic review

Abstract

Background:

Arthrogenic muscle inhibition (AMI) and altered neuromuscular control commonly arise after anterior cruciate ligament (ACL) injury and may persist both preoperatively and after anterior cruciate ligament reconstruction (ACLR), potentially impairing functional recovery.

Objective:

To synthesise randomised controlled trial (RCT) evidence on functional rehabilitation interventions targeting neuromuscular control after ACLR, using electromyography (EMG)-derived outcomes.

Methods:

PubMed, Cochrane Library, Embase, CINAHL, Web of Science, and SPORTDiscus were searched from January 1, 2015, to December 4, 2025; duplicates were removed ($n=63$). RCTs involving individuals undergoing ACLR (pre-operative) and/or following ACLR (post-operative) were included if they reported EMG-derived neuromuscular outcomes following functional rehabilitation. Due to heterogeneity in interventions, EMG normalisation, and outcome definitions, a structured narrative synthesis was performed.

Results:

Nine RCTs (total $n=263$) were included. One trial examined preoperative blood flow restriction training (BFRT) preconditioning (5 sessions over 8 days before surgery), whereas the remaining interventions were delivered postoperatively. Jump training reduced quadriceps–hamstring co-contraction by 50% ($p<0.001$). Preoperative BFRT preserved quadriceps endurance at 4 weeks post-ACLR compared with sham (sham reduction: $97\pm 85$ s) and increased root mean square (RMS) EMG amplitude by $54\pm 58\%$ at week 4. Aquatic proprioceptive training improved muscle activation similarity and magnitude ($p<0.01$). Sand-based training and cross-education interventions increased EMG activity during gait phases ($p<0.05$). Vibration therapy and postoperative BFRT produced mixed or non-significant effects. Evidence for several intervention types was limited to single trials and should be interpreted cautiously.

Conclusion:

Functional rehabilitation strategies show limited but promising potential to improve neuromuscular outcomes following ACLR. Phase-specific approaches may be appropriate: early-phase interventions (e.g., BFRT preconditioning, vibration therapy, and cross-education) may help mitigate AMI and preserve neuromuscular capacity, whereas later-phase strategies (e.g., plyometrics and aquatic proprioceptive training) may support movement quality and motor control. Future RCTs should prioritise standardised EMG methodology and clinically meaningful endpoints (e.g., return-to-sport and reinjury).

Introduction

Anterior cruciate ligament (ACL) injuries are common among physically active individuals and frequently occur during non-contact tasks such as jump-landing or rapid directional changes.^1,2 These injuries compromise knee joint stability and increase the risk of secondary meniscal or cartilage damage, ultimately contributing to long-term functional limitations.^3,4 ACL reconstruction (ACLR) is therefore widely performed to restore mechanical stability and facilitate return to sport; however, full functional recovery is often unsuccessful.¹ Nearly half of patients develop post-traumatic osteoarthritis within 15 years, and reinjury rates exceed 20%.^5,6

Quadriceps weakness and neuromuscular deficits are among the most persistent complications following ACLR.^7,8 Strength deficits greater than 20% frequently remain and contribute to gait asymmetry, abnormal joint loading, and compromised functional performance.⁹ These deficits are strongly linked to arthrogenic muscle inhibition (AMI), which arises from joint effusion, pain, and disrupted afferent signaling.¹⁰ Since AMI alters neuromuscular recruitment strategies, rehabilitation must address not only muscle strength but also movement quality and motor control.^11,12 Recent clinical practice guidelines, including those from the American Academy of Orthopedic Surgeons, emphasise restoring functional stability and protecting joint health throughout rehabilitation.¹³

More than 70% of ACL injuries occur without direct contact, highlighting the importance of modifying biomechanical risk factors such as dynamic knee valgus, quadriceps dominance, reduced hamstring contribution, and excessive co-contraction.¹⁴ As a result, recent rehabilitation approaches emphasise functional, task-oriented interventions that enhance neuromuscular control. These include core and hip strengthening, perturbation training, unstable-surface training (such as sand-based exercise),¹⁵ body-weight supported (BWS) plyometrics,¹⁶ aquatic proprioceptive training,¹⁷ vibration therapy,¹⁸ and blood flow restriction training (BFRT).¹⁹ Emerging approaches, such as cross-education and neurophysiologically targeted vibration modalities, have also demonstrated potential benefits.

Recent systematic reviews and meta-analyses have evaluated selected rehabilitation components after ACLR, for example, BFRT and strength/performance outcomes such as vertical jump metrics, but they rarely synthesise electromyography (EMG)-derived neuromuscular control outcomes across diverse functional interventions.^1,8 To date, however, existing systematic reviews have typically focused on a single intervention category or lacked simultaneous evaluation of neuromuscular activation, biomechanical performance, and functional movement quality.^20,21,22 These limitations hinder the development of comprehensive rehabilitation models. In contrast to prior reviews that have typically examined a single intervention category or emphasised strength and functional outcomes alone, this review advances the field by (i) restricting inclusion to randomised controlled trials (RCTs), (ii) prioritizing EMG-derived neuromuscular outcomes as mechanistic indicators of AMI-related activation failure, and (iii) providing a structured synthesis by intervention type and perioperative phase (pre-operative versus post-operative). This approach enables clearer phase-based interpretation and more practical clinical guidance for integrating functional rehabilitation strategies after ACLR.

Therefore, the purpose of this systematic review was to synthesise evidence from RCTs evaluating functional physical therapy interventions around ACLR (pre- and post-operative phases), with a focus on their effects on neuromuscular activation, biomechanical risk factors, and functional performance. The findings aim to provide practical guidance for designing phase-specific rehabilitation programs.

Methods

Study design

This systematic review followed the methodological recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement and the Cochrane Handbook for Systematic Reviews of Interventions (version 6.3). The review protocol was not prospectively registered (e.g., PROSPERO); however, eligibility criteria, outcomes, and the synthesis plan were defined a priori and applied consistently throughout screening and data extraction. To minimise selective reporting, we prespecified the eligibility criteria, primary outcome domains (EMG-derived neuromuscular measures), and subgrouping logic (perioperative phase and intervention type) before screening; standardised extraction and risk-of-bias forms were used, and any post hoc decisions were documented in the data-extraction log.

Search strategy

A structured search strategy combining controlled vocabulary (e.g., MeSH) and free-text keywords related to ACLR, AMI, EMG, neuromuscular activation, motor control, and functional exercise was developed. Six electronic databases (PubMed, Cochrane Library, Embase, CINAHL, Web of Science, and SPORTDiscus) were searched from January 2015 to December 4, 2025. The complete search strategies and run dates for each database are provided in Appendix A. A sample PubMed search syntax was: (“anterior cruciate ligament reconstruction”[MeSH] OR ACLR OR “ACL reconstruction”) AND (electromyography OR EMG OR neuromuscular OR “muscle activation” OR “motor control”) AND (rehabilitation OR exercise OR training OR proprioceptive OR plyometric). Reference lists of eligible studies were also screened to identify additional records. The complete search strategies for all databases and the run date are provided in Appendix A.

Eligibility criteria (PICOS)

Participants: Individuals undergoing ACLR (pre-operative) and/or following ACLR (post-operative).

Intervention: functional, task-oriented, or sensorimotor-focused programs.

Comparison: no intervention, conventional therapy, or alternative exercise.

Outcomes: EMG amplitude, co-contraction index, activation similarity, central activation ratio, and functional task performance.

Study design: RCTs.

Studies were excluded if they did not involve ACLR, did not apply functional interventions, did not report neuromuscular or EMG outcomes, or were non-peer-reviewed (e.g., theses, conference abstracts) (Supplementary Table S1).

Selection process

Three independent reviewers screened all titles and abstracts, followed by full-text assessment using predefined PICOS criteria. Disagreements were resolved by consensus discussion; if disagreements persisted, all three reviewers jointly made the final decision by rational consensus based on the predefined criteria. Full-text screening included verification of intervention classification, EMG outcome reporting, and perioperative status (pre-operative or post-operative). Reasons for exclusion were documented at each stage to ensure transparency. Inter-rater reliability for initial screening was substantial (Fleiss’ $\mathrm{\text{kappa}}=0.790$) (Fig. 1).

Fig. 1.

PRISMA flow chart of study selection.

Risk of bias assessment

Risk of bias was evaluated independently by two reviewers using the Cochrane Risk of Bias Tool (RoB 1.0), with each domain rated as low, unclear, or high risk. Disagreements were resolved through discussion until consensus was reached; if consensus could not be reached, a third reviewer adjudicated. Domains assessed included random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting.

Data synthesis

Given heterogeneity in EMG normalisation procedures, task conditions, intervention duration, and the timing of postoperative assessments, meta-analysis was not feasible. Therefore, we performed a structured narrative synthesis, grouping studies first by perioperative timing (pre-operative versus post-operative) and then by intervention type, with study counts reported for each subgroup. Findings were summarised within predefined neuromuscular outcome categories (e.g., activation amplitude, co-contraction index, frequency-domain characteristics, activation similarity).

Results

Study characteristics

Nine RCTs involving 263 participants were included. Trials were grouped by perioperative timing (pre-operative versus post-operative) and by intervention type. Mean age across studies ranged from 16.5 to 35.0 years, with most samples comprising young adults. Studies were published between 2016 and 2025 and were conducted in the United States, Iran, and other countries. All participants were individuals with ACL rupture undergoing ACLR, although graft types were not consistently reported (Table 1).

Table 1.

Study characteristics of included RCTs ($n=9$).

Sources	Participants (M/F)	Age (yrs)	Timing relative to ACLR	Graft type	Intervention	Duration
Pamukoff et al.9	20 (6/14)	$21.1\pm 1.2$	$50.7\pm 21.3$ mon post-op	16 PT, 3 HS, 1 AG	WBV/LMV (30 Hz)	Single
Elias et al.16	19 (5/14)	JTBW: $21.1\pm 3.4$; JTBWS: $24.9\pm 5.9$	JTBW: $17\pm 13.5$; JTBWS: $18\pm 12.5$ mon post-op	JTBW: 4 PT, 3 HS, 2 cadaveric; JTBWS: 7 PT, 3 HS	Jump training	8 wks
Žargi et al.23	20 (16/4)	BFR: $34\pm 6$; SHAM: $35\pm 5$	Pre-op (5 sessions over 8 days before ACLR); follow-up at 4 & 12 wks post-op	Semitendinosus–gracilis (hamstring) autograft	BFR (pre-op)	5 sessions/8 days
da Costa et al.18	44 (44/0)	$27.4\pm 6.2$	$17\pm 1.4$ mon post-op	Gracilis–semitendinosus	WBV (50 Hz)	Single
Curran et al.19	34 (15/19)	$16.5\pm 2.7$	Intervention started at 10 wks post-op	BPTB ($n=25$), STG ($n=6$), QT ($n=3$)	Con/Ecc $+$ BFRT	8 wks
Hajouj et al.24	40 (40/0)	18–35 (mean $=$ mid-20s)	Not explicitly stated; all were athletes after ACLR, in rehabilitation phase	Not specified	Aquatic PT	10 wks
Sheikhalizade et al.15	28 (28/0)	22–25	$>6$ mon post-op	Not specified	Sand exercises	8 wks
Sheikhalizade et al.25	28 (28/0)	22–25 (mean $=$ 24)	Not explicitly stated; all had prior ACLR and were in a chronic, stable phase	Not specified (ACLR type not reported)	Sand gait training	8 wks
Payandeh and Daneshmandi26	30 (30/0)	Exercise: $29.15\pm 7.89$; Control: $28.97\pm 7.18$	$>6$ mon post-op	Not specified	Cross-training	8 wks

Notes: ACLR, anterior cruciate ligament reconstruction; AG, allograft; BFR, blood flow restriction; BFRT, blood flow restriction training; BPTB, bone–patellar tendon–bone; Con, concentric; Ecc, eccentric; HS, hamstring; JTBW, jump training body weight only; JTBWS, jump training with body weight support; LMV, local muscle vibration; PT, patellar tendon; QT, quadriceps tendon; STG, semitendinosus–gracilis; WBV, whole-body vibration.

Risk of bias summary

These methodological limitations reduce confidence in intervention-specific effects, particularly where evidence is derived from single trials. Because participant and personnel blinding are often infeasible in exercise-based trials, performance bias may have influenced functional task performance and EMG outcomes. Accordingly, findings should be interpreted as preliminary and warrant confirmation in larger, better-reported RCTs.

Methodological quality ranged from low to moderate. The most frequent concerns were unclear reporting of random sequence generation and insufficient blinding of participants and personnel. Allocation concealment was adequately described in only a subset of studies. Blinding of outcome assessment was more consistently implemented, and attrition and selective reporting biases were generally low (Fig. 2).

Fig. 2.

Risk of bias graph and summary.

Intervention overview

Functional rehabilitation interventions were categorised as vibration therapy (two studies),^9,18 BFRT (two studies; one preoperative, one postoperative),^19,23 jump training (one study),¹⁶ aquatic proprioceptive training (one study),²⁴ sand-based training (two studies),^15,25 and cross-education training (one study).²⁶ Intervention durations ranged from single-session protocols to multiweek programs lasting up to 10 weeks, with training frequencies of two to three sessions per week in most long-term protocols.

Effects on neuromuscular control

Postoperative jump training consistently reduced quadriceps–hamstring co-contraction during landing tasks, with approximately a 50% decrease in co-contraction index reported following an 8-week program.¹⁶ These improvements were maintained at follow-up, indicating durable neuromuscular adaptations. BWS jump training further allowed high-volume plyometric exposure while managing joint load, which is advantageous in early rehabilitation stages.

Postoperative aquatic proprioceptive training, when combined with accelerated land-based rehabilitation, significantly improved activation similarity and sensorimotor output compared with land-based rehabilitation alone.^20,24 These findings suggest that the aquatic environment’s combination of reduced joint loading and enriched proprioceptive input may facilitate improvements in neuromuscular control strategies.

BFRT showed perioperative timing effects. In the preoperative preconditioning trial, the sham group demonstrated a $\sim 50\%$ reduction in quadriceps endurance at 4 weeks post-ACLR ($97\pm 85$ s), whereas the BFRT group maintained endurance and showed increased RMS EMG amplitude ($+54\pm 58\%$ at week 4).²³ In contrast, postoperative high-load BFRT did not demonstrate additional neuromuscular benefits beyond conventional high-load strengthening.¹⁹

Postoperative vibration therapy induced acute increases in EMG amplitude and central activation ratio, likely through modulation of afferent sensory pathways, although findings were protocol dependent and not uniformly positive across studies.^9,18

Postoperative cross-education training, in which only the non-surgical limb was strengthened, generated moderate-to-large improvements in EMG activation of the surgical limb, consistent with a possible cross-limb transfer effect.²⁶ The summary of neuromuscular outcomes is presented in Table 2.

Table 2.

Summary of EMG-derived neuromuscular outcomes and main findings ($n=9$).

Sources	Muscles	Variables	Main findings
Pamukoff et al.9	VM, VL, RF	RMS, CAR	$\uparrow$EMG & CAR*
Elias et al.16	VL, BF	CoI	$\downarrow$CoI 50%***
Žargi et al.23	VM (sEMG)	RMS (VM sEMG)	$\uparrow$RMS**; preserved endurance (sham endurance reduction: $97\pm 85$ s at 4 wks post-op)
da Costa et al.18	VM, VL	RMS	NS
Curran et al.19	VM, VL	RMS, CAR	NS
Hajouj et al.24	VM, VL, RF, BF, ST	MAG, SI	$\uparrow$MAG & SI**
Sheikhalizade et al.15	VM, VL, RF, BF, ST, TA, Gas, GM	Freq	$\uparrow$Frequency*
Sheikhalizade et al.25	VM, VL, RF, BF, ST, TA, Gas, GM	RMS	$\uparrow$Loading phase*
Payandeh and Daneshmandi26	VM, VL, RF, BF, TA, Gas	RMS	$\uparrow$Swing & stance*

Notes: BF, biceps femoris; CAR, central activation ratio; CoI, co-contraction index; Gas, gastrocnemius; GM, gluteus medius; MAG, magnitude; NS, not significant; RF, rectus femoris; RMS, root mean square; SI, similarity index; ST, semitendinosus; TA, tibialis anterior; VM, vastus medialis; VL, vastus lateralis. *$p<0.05$; **$p<0.01$; ***$p<0.001$. $\uparrow$ increased; $\downarrow$ decreased.

Discussion

This systematic review demonstrates that functional rehabilitation interventions can meaningfully improve neuromuscular control following ACLR, as shown across nine RCTs reporting improvements in EMG-based outcomes and movement quality.

However, the evidence base remains small (nine RCTs) and intervention- and EMG-method heterogeneity is substantial. Accordingly, intervention-specific conclusions should be interpreted as preliminary and context-dependent, particularly when supported by single trials.

Jump training yielded some of the most robust and durable effects, reducing quadriceps–hamstring co-contraction and improving landing mechanics.¹⁶ These changes likely reflect improved coordination between agonist and antagonist muscle groups and reduced quadriceps dominance, key factors in safe deceleration and landing strategies.³

Aquatic proprioceptive training significantly enhanced activation similarity and sensorimotor output.²⁴ The aquatic environment’s buoyancy, hydrostatic pressure, and multidirectional resistance may enhance afferent feedback and improve sensorimotor integration; however, direct central neural adaptations were not assessed, and mechanistic interpretations should be considered hypothesis-generating.^20,26,27

Sand-based training challenges postural control and gait under unstable conditions, increasing muscle activation and improving frequency-domain characteristics during walking.^15,25 These adaptations suggest that sand-based exercise may refine gait mechanics and dynamic stability in later rehabilitation phases.^28,29

Vibration therapy produced acute increases in EMG amplitude and central activation ratio, likely through enhanced afferent input; however, inconsistencies and parameter sensitivity limit definitive conclusions.^9,18

BFRT showed the greatest promise when applied preoperatively or when high-load training was not yet feasible, helping preserve neuromuscular capacity and early postoperative muscle endurance.²³ In contrast, postoperative high-load BFRT did not demonstrate advantages beyond those of conventional strengthening.^19,30,31 Once high-load resistance becomes tolerable, the benefits of BFRT appear to diminish.^32,33

Cross-education training demonstrated clinically meaningful cross-limb effects, with strengthening of the non-surgical limb enhancing activation in the reconstructed limb — valuable in early rehabilitation when surgical-side loading is limited.^26,34 Because the included trial(s) relied on surface EMG without neurophysiological measures (e.g., TMS, H-reflex), the underlying central mechanisms remain speculative.

Clinical relevance of EMG-derived outcomes warrants careful interpretation. EMG-based indices (e.g., activation amplitude, symmetry, and co-contraction) provide mechanistic insight into AMI-related activation failure and movement strategies, but they are surrogate markers. In observational work after ACLR, altered neuromuscular activation strategies during landing and gait have been linked to abnormal knee mechanics that may influence functional performance and potentially reinjury risk.^3,12 Accordingly, the intervention-specific EMG improvements summarised here should be viewed as plausibly supportive of safer movement patterns, while direct effects on return-to-sport, reinjury, and patient-reported outcomes remain insufficiently tested in RCTs. Additionally, contemporary orthopaedic clinical practice guidelines for osteoarthritis management continue to position structured exercise as a core treatment pillar, reinforcing the potential value of rehabilitation strategies that optimise neuromuscular control and joint loading profiles over the long term.¹³

Clinical implications: Based on the limited RCT evidence, a phase-based integration may be reasonable. In the perioperative/early phase when high-load training is not feasible, clinicians may consider strategies aimed at mitigating AMI and preserving neuromuscular capacity (e.g., BFRT preconditioning, vibration therapy, and cross-education), paired with progressive restoration of range of motion and gait. As loading tolerance increases, later-phase task-oriented approaches that target movement quality and motor control (e.g., graded plyometrics/jump training and aquatic proprioceptive training) may be introduced to refine landing and sensorimotor strategies. These recommendations should be individualised and applied cautiously because many intervention categories are supported by single trials and protocol details (dose, timing, and EMG tasks) vary substantially.

This review has several limitations. Heterogeneity in interventions, EMG acquisition/normalisation methods, and outcome definitions precluded meta-analysis and limited direct comparison across studies. Several trials had unclear reporting for randomisation and allocation concealment, and blinding was frequently not feasible, which may have introduced bias. Surface EMG was the primary neuromuscular measure in most studies, limiting inference about central neural mechanisms. In addition, the small number of trials per intervention category and the concentration of studies in a limited number of countries may limit generalizability and raise concerns about publication bias; because fewer than 10 studies were available, formal publication bias assessment (e.g., funnel plots) was not feasible.

Rehabilitation should be individualised based on patient factors, surgical details, postoperative stage, and return-to-activity goals. Functional interventions are most effective when integrated within a comprehensive, progressive rehabilitation framework rather than used in isolation.^1,4,10,11

Conclusions

Functional rehabilitation interventions demonstrate meaningful potential to improve neuromuscular control following ACLR; however, the overall certainty of evidence remains limited. The most consistent signals were observed for later-phase plyometric training and aquatic proprioceptive training, while early-phase approaches such as BFRT preconditioning, vibration therapy, and cross-education may help mitigate AMI and preserve neuromuscular capacity. Given that many intervention categories were supported by single RCTs and methodological quality was variable, these findings should be interpreted cautiously and validated in larger trials with standardised EMG methodology and clinically relevant outcomes.

Conflict of Interest

The authors have no conflicts of interest relevant to this paper.

Funding/Support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Author Contributions

Conceptualisation, H.P. and B.H; methodology, H.P. and M.K; literature search and data extraction, B.H. and M.K; risk of bias assessment, B.H. and M.K; data analysis, B.H and M.K; writing-original draft preparation, H.P. and M.K; writing-review and editing, H.P. and M.K. All authors have read and agreed to the published version of the paper.

Appendix A

The Supplementary Appendix are available at: https://www.worldscientific.com/doi/suppl/10.1142/S1013702526500083

ORCID

Minbong Kang https://orcid.org/0000-0001-7852-6887

Byoungha Hwang https://orcid.org/0000-0002-6293-9270

Hankyu Park https://orcid.org/0000-0001-6742-2409