Exercise Therapy in Nonspecific Low Back Pain among Individuals with Lower-Limb Amputation: A Systematic Review

Low back pain is very common condition that often becomes a long-lasting problem in prostheses users after lower limb amputation. The presented study aims to decide the potential benefits of exercise therapy on low back pain among lower limb amputees by using a systematic review. The PICO technique was used to answer the primary issue of this review: Does exercise treatment lessen the prevalence of low back pain in the population of lower limb amputees? Systematic review was conducted in the following databases: Medline-PubMed, EMBASE, Scopus, and Web of Science. Studies up to September 2010 published in English are included. Aim, target population, development and execution strategies, and treatment suggestions were among the data gathered. The primary outcomes of interest were exercise interventions as a therapy for low back pain but only two articles met including criteria. The search was broadened and 21 studies describing biomechanical changes in gait and pelvic-spine posture were analysed. This review indicates that movement therapy is a potential treatment strategy in low back pain among amputees. The major limitation of the study is the very heterogenous group of subjects in terms of amputation level, baseline activity level and comorbidities. We used a procedure that was registered in PROSPERO (CRD42022345556) to perform this systematic review of systematic reviews. There is a necessity of good quality research for concluding a consensus of exercise intervention.


Introduction
Amputation of the lower limb is a life-changing experience, in which there are alterations in physical and mental well-being [1]. An estimated 1.6 million people were living with the loss of limb in the year 2015; in the United States, 185,000 people undergo an amputation of a lower limb every year. Dysvascular amputations, described as secondary to complications of peripheral arterial disease or diabetes mellitus, are the most common [2]. The leading causes have been reported to vary depending on the region. In many lowand middle-income countries, trauma has been documented as the primary mechanism for limb loss [3]. The multitude of causes and complications of the amputation process make it very difficult to study. Traumatic amputees tend to be younger and better in shape than vascular disease or diabetic amputees. They are also able to make full recoveries from their injury to a point where they can autonomously walk and achieve everyday high-level functioning. Secondary musculoskeletal disability, such as low back pain (LBP), is among the most important sources of additive disability. In the amputee population, between 52-71% of amputees experience LBP [4,5]. The amount is much higher than in the non-amputee population (12-34%) [6]. There have been numerous systematic evaluations of exercises for low back pain but only in the non-amputee population. Search Strategy (("lower limb amputation" [MeSH Terms] OR "amputation" [All Fields] OR ("aboveknee amputation" [All Fields] OR "below-knee amputation" [All Fields] OR "limb loss" [All Fields]) AND ("low back pain" [MeSH Terms] OR ("LBP" [All Fields])) AND ("exercise" [MeSH Terms] OR "exercise" [All Fields] OR ("physical" [All Fields] AND "activity" [All Fields]) OR "physical therapy" OR ("exercise" [All Fields] AND "therapy" [All Fields]) OR "exercise therapy"))) AND (("2000/01/01" [PDAT]: "2022/12/31" [PDAT] AND English [lang]. The investigation includes English-language papers that were randomised and published between 1 January 2000 to 31 December 2021. The review was performed in agreement with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and PRISMA-P checklist is provided as an additional file. The reviewers were physiotherapist and academic researchers working with amputees' population daily.
Inclusion and exclusion criteria were defined for the examination of titles, abstracts, and extensive texts. The review examined works completed and published in English between the years 2000 and 2021. Nonspecific low back pain is defined as persistent low back pain that is not ascribed to an identifiable, recognised specific disease (e.g., infections, cancerous, osteoporosis, ankylosing spondylitis, fracture, inflammatory process, radicular syndrome or cauda equina syndrome) for the purposes of this review [19]. Exercise treatments were described as activities that were planned, systematic, and repeated that result in body movement and energy consumption by engaging skeletal muscles [20]. Posttreatment, short-term (closest to three months), intermediate-term (closest to 6 months), and long-term (closest to 12 months) follow-up.
Exclusion criteria included research published in a language other than English that were unrelated to low back pain in amputation patients. Studies were ruled out if the participants experienced acute or subacute low back pain, or if the circumstance was caused by certain disorders. Bachelor's, Master's, and Doctoral theses, as well as Letters to the Editor, Conference reports, and study protocols, were all rejected. Additional exclusion criteria included a lack of properly documented outcomes, availability to fulltext papers, and the absence of defined scales for low back pain. The participants in the trials included male and female patients between 18 and 65 years of age who had experienced amputation in the lower extremity. The electronic search was carried out by one reviewer. Two reviewers (AWS and AMB) independently evaluated the titles and abstracts. Two independent reviewers (AWS and AMB) examined these full text papers to determine eligibility. Disagreements were identified and handled between pairs of reviewers and where they required the engagement of a third reviewer. The detailed screening process will be shown in the following Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) flow diagram (PROSPERO nr CRD4202234555). In Excel, a data extraction form was specifically created. Information was gathered about general information (e.g., country, healthcare context, publication year, target population and presenting symptoms) (e.g., country, healthcare setting, publication year, target population and presenting symptoms), methods regarding assessed movement activity and implementation (e.g., [strength of] recommendations, any details regarding subgroups of amputees).
Using the Risk-of-Bias 2 tool, which is accessible on the Cochrane platform, the risk-ofbias analysis was carried out independently by two researchers. The analysis looked at the randomization procedure, variations from intended interventions, missing outcome data, measurement of outcome, and choice of the reported result as its five bias domains. Three to seven questions, with possible responses of Yes/Probably Yes/Probably No/No/No information, were included in each domain. The programme evaluated each domain's and the overall study's risk of bias based on the responses given.
To evaluate the possibility of bias in nonrandomized studies or therapies, the ROBINS-I analysis was conducted. Confounding participants, interventions, deviations from planned interventions, missing data, measurements, and reported outcomes were the six bias domains that were assessed in the analysis.

Results
Overall, a total of 866 references were found; of those, 675 articles were disregarded due to unrelated title, summaries, and/or language. Finally, 191 were entirely analysed and ultimately only 21 studies incorporated in the review (Figure 1). The ultimate number of publications included in the study was indicated by the risk-of-bias evaluation. The risk-ofbias analysis revealed that the included studies' quality is either moderate or questionable ( Figure 2 and Table 2). ulation and presenting symptoms), methods regarding assessed movement activity and implementation (e.g., [strength of] recommendations, any details regarding subgroups of amputees).
Using the Risk-of-Bias 2 tool, which is accessible on the Cochrane platform, the riskof-bias analysis was carried out independently by two researchers. The analysis looked at the randomization procedure, variations from intended interventions, missing outcome data, measurement of outcome, and choice of the reported result as its five bias domains. Three to seven questions, with possible responses of Yes/Probably Yes/Probably No/No/No information, were included in each domain. The programme evaluated each domain's and the overall study's risk of bias based on the responses given.
To evaluate the possibility of bias in nonrandomized studies or therapies, the ROB-INS-I analysis was conducted. Confounding participants, interventions, deviations from planned interventions, missing data, measurements, and reported outcomes were the six bias domains that were assessed in the analysis.

Results
Overall, a total of 866 references were found; of those, 675 articles were disregarded due to unrelated title, summaries, and/or language. Finally, 191 were entirely analysed and ultimately only 21 studies incorporated in the review (Figure 1). The ultimate number of publications included in the study was indicated by the risk-of-bias evaluation. The risk-of-bias analysis revealed that the included studies' quality is either moderate or questionable ( Figure 2 and Table 2).   Critical features of the included and disregarded articles are presented in graphs.

Exercise-Orientated Rehabilitation Programmes for Low Back Pain in Amputees
Despite many studies available on the low back pain topic and exercise in a nonamputee population, only two studies of this type relevant for amputees were found. The influence of the back school programme in lower limb amputees was examined by Anaforoglu et al. [21]. Twenty men, post-traumatic unilateral transfemoral amputees from the intervention group, performed 10 sessions in two weeks (5 days per week) with back health education and an exercise programme that included practical and theoretical information with individual exercises. Each session lasted about 1 h and was supervised by the physiotherapist. The control group had only a brochure on theoretical info on back health education and exercise pictures. The VAS scale, flexibility and the Oswestry disability index (ODI) were measured before interventions and performed again 1 month and 3 months after treatment. Results show that after 1 month, the pain perception and ODI score decreased significantly in the experimental group (respectively, VAS 34.1 SD 13 in group 1 and 52.8 SD 15.68 in group 2, p < 0.001; ODI score 9.55 SD 5.65 in group 1 and 14.85 SD 7.97 in group 2; p = 0.03). A similar result was obtained 3 months later (VAS 12.8 SD 8.31; in group 1 and 30.6 SD 10.93 in group 2, p < 0.001; ODI score 4.65 SD 3.61 in group 1 and 9.85 SD 5.39 in group 2; p = 0.001 [21]. Min Kyung Shin et al. [26] examined the influence of lumbar strengthening exercise training in lower limb amputees. A total of 19 subjects unilateral (11 patients) and bilateral (8 patients) posttraumatic amputees were enrolled in the study. There was no control group designed in the study. The exercise programme was conducted twice in a week for 8 weeks, for a sum of 16 sessions, each session was 30 min long and was made up of 14 exercises. Evaluations were performed twice; the first was 1 week before the programme started, while the second was following the 8 week programme. VAS, The Korean version of the Oswestry Disability Index (K-ODI), The Thomas test, and a trunk raising test were used to evaluate changes in participants. Before the project, the mean VAS score was 4.6 SD 2.2, the K-ODI score 12.4 SD 8.2. At the end of the programme, the mean VAS score was 2.6 SD 1.6 and the K-ODI score 11.4 SD 8.2 (p < 0.001). Abdominal muscle strength and back extensor strength considerably increased after the programme (Abdominal: 4.4 SD 0.7 before and 4.8 SD 0.6 after, p = 0.007; extensor strength before 2.6 SD 0.6 before and 3.5 SD 1.2 after, Life 2023, 13, 772 6 of 20 p = 0.007). There was no substantial difference among results in unilateral and bilateral amputees [26]. Study characteristics are showed in Table 3.

Spinopelvic Alignment and Relationship with Low Back Pain among Amputees
Facione et al. [27] examined the spinopelvic alignment in transfemoral amputees using radiologic imaging. 10 male and 2 female patients were classified into two groups: with and without low back pain (LBP). It should be mentioned that 11 subjects used a microprocessor-controlled knee and only one was a mechanical knee. There were 5 subjects in the low back pain group and 7 without it. To analyse postural alignment, biplanar low-dose x-rays of the full spine were made. The researchers found that four subjects with low back pain had an imbalanced sagittal posture (T9 tilt −13 0; SD: 4, LBP group and −9 0 SD:1 non-LBP group; p = 0.046). Furthermore, 8 subjects (6 LBP group and 2 non-LBP group) presented an abnormally low value of thoracic kyphosis (TK 26 0 SD:10, LBP group and 16 0 SD: 5 non-LBP group; p= 0.051). The mean angle TK in the non-LBP group was lower than in the LBP group (p = 0.0510).
A similar study was conducted by Matsumoto et al. [22] in which a relationship only between the lumbar lordosis angle and low back pain was examined using lateral radiological imaging. The authors decided to include 17 transfemoral amputee males, 9 were placed in the LBP group, and 8 in the non-LBP group. Pain levels were characterised using the Chronic Pain Grade questionnaire to rate current pain and intensity, psychological well-being was assessed using the SF-36 Mental Scale and the Roland-Morris Disability Questionnaire was used for 24-item classification of physical disability related to LBP. According to this study, there was no significant difference in the angle of lumbar lordosis (LLA) between two groups with and without LBP (LLA 46.1 0 SD: 12.4 0 LBP group and 51.0 0 SD: 12.6 0 non-LBP group, p = 0.43). Simultaneously there was also no significant difference in sacral inclination angle (SIA) between groups (SIA 38.3 0 SD:8.7 LBP group and 38.1 0 SD: 7.5 non-LBP group, p = 0.84). Study characteristics presented in Table 4.

Trunk Kinematics during Standing, Stepping or Sitting Activities
Alterations in trunk-pelvis and lumbar-spine kinematics were measured in three articles by Hendershot et al. [23][24][25]. The first article compares 8 males with unilateral lower leg amputation (transtibial and transfemoral) with 8 male non-amputees as a control group. There was no additional division into the LBP group and the non-LBP group, but those articles highlight changes that might be crucial in the potential development of LBP among amputees. All participants completed the Physical Activity Questionnaire (IPAQ) and their seated balance was assessed using an unstable chair that pivots on a low-friction ball-and-socket joint. Participants performed maximum voluntary contractions (MVC) in the flexion, extension and left/right lateral bending of the trunk. During MVCs, electromyographic (EMG) activities of the bilateral lumbar erector spine, rectus abdominis, and external oblique muscles were recorded [23]. All traditional measures of postural control were significantly higher among participants with lower leg amputation (95% ellipse area, RMS distance, and mean velocity). The RMS distances were higher in the A-P direction for both groups (RMS distance A-P 0.70 cm SD: 0.26 Transtibial; 0.82 cm SD: 0.26 transfemoral and 0.56 cm SD: 0.12 non-amputee group; p = 0.015). The mean normalised RMS muscle activity was higher among participants with lower leg amputation in the erector spinae (p < 0.0001), rectus abdominis (p < 0.0001), and external oblique (p = 0.0045).
In the second study, trunk kinematics and neuromuscular behaviours were compared between people with and without lower leg amputation who performed maximal voluntary standing contractions (MVC) in extension and left/right lateral bending [24]. The same study group that was described before (8 male amputees and 8 male non-amputees) was examined. During MVCs, electromyographic activity was in the same muscle group as before. Furthermore, participants were exposed to horizontal trunk perturbation. Postural displacements were measured with a laser displacement sensor. The main result is that during perturbations, the stiffness of the trunk (TS) and the maximum reflex force (MRF) were significantly lower (TS 13.2 N/mm SD: 2.5 in the non-amputee group; 10.0 N/mm SD: 2.1; p = 0.017 and MRF 71.7 N SD:12.3 in the non-amputee group and 55.3 N SD:11.8 in the amputee group; p = 0.017). Simultaneously, the effective trunk mass was comparable across groups and perturbation orientations. The third study had examined flexion-relaxation responses during asymmetric trunk flexion movements [25]. For the third time, the same group took part in the study (8 male amputees and 8 male non-amputees) but this time participants were standing in a fixed structure and movements of the pelvis and lower limbs were further minimised by a fixed pelvic confinement. Twenty-one different movements were performed with the report order of transverse rotation randomised. During each move, subjects flexed forward towards targets until reaching a relaxed, passive hanging position with minimal muscle activity and arms hanging relaxed and then returned to an upright standing position. The main results of these studies showed similar angles of peak lumbar flexion (p = 0.26) and peak nEMG values for both flexion (p = 0.10) and extension (p = 0.33) in sagittal-symmetric movements. In sagittal-asymmetric movements, the maximum lumbar flexion angles decreased significantly (p < 0.001) decreased with increasing transverse rotation among participants after lower limb amputation, bilateral similar (all p > 0.4). Additionally, during flexion, the peak nEMG values were similar between the groups (p = 0.16), transverse rotation angles (p = 0.72), and the direction (p = 0.24).
Actis et al. [28] examined subjects with transtibial amputation during the sit-to-stand task. They enrolled 8 participants with lower extremity amputation and 8 without amputation. The dominance of the limb was determined as the leg chosen to kick a ball and everyone completed the Oswestry Low Back Pain Questionnaire. The main task was five sit-to-stand trials with 42 kinematic markers to track feel, shanks, thighs, pelvis, and trunk. Muscle group activation was registered by EMG. The authors developed a musculoskeletal model. Only one individual had LBP that was more severe than minimal, according to the Oswestry questionnaire (30% score). Comparatively to non-amputees, subjects with lower limb amputations showed higher peaks and average L4-L5 compressive loads, peak sit-to-stand motion angles, trunk lateral bending, and axial rotation angular velocities. However, there were no variations in muscle activation that were statistically significant (0.05 < p < 0.1). Additionally, it mentions that the amputees generated higher vertical force in the intact limb than the control group due to their greater asymmetry.
Butowicz et al. [29,30] performed two particularly important kinematic investigations. The first analysis looked at how the existence of low back pain influences the joint coordination and balance of the trunk-lower limb during standing. A total of 40 participants were included in this study (23 with LBP and 17 non-LBP amputees). Eight Inertial Measuring Units (sternum, sacrum, bilateral foot, bilateral lower leg, bilateral upper leg) were used to measure the subjects' standing stillness for 30 s while their eyes were closed and opened. As a result, there was no trunk-hip coordination pattern that indicated the ability to maintain balance while keeping one's eyes open (Fuzzy Entropy (FE) 0.37 SD:0.08 for the non-LBP group and 0.4 SD:0.10 for the LBP group). With eyes closed, trunk-lower limb joint coordination patterns in the intact limb, such as extension/flexion patterns on the amputated side and flexion/extension patterns on the intact side, predicted FE in the LBP group [29].
The second article by Butowicz et al. [30] was carried out on 32 people with traumatic lower limb amputation and the participants were divided into two groups (LBP-19 and non-LBP-13). The participants were told to maintain their arms crossed and their seat level while sitting in an unsteady chair with their eyes open. Using an 18-camera motion capture system, the three-dimensional trunk-pelvis kinematics were examined with 12 retro-reflective markers. At the same time, EMG data were collected. Centre of Pressure (COP) measures, EMG, and trunk kinematics were compared between groups. As a result we find that there was no main effect of the group on the set of COP-based measures (Wilks' Λ = 0.84, F (10,18) = 0.74, p = 0.60, η 2 = 0.16) and there was a significant effect on trunk kinematic (Wilks' Λ = 0.46, F (6,19) = 3.52, p = 0.02, η 2 = 0.54) and muscle activity (Wilks' Λ = 0.69, F (4,15) = 3.46, p = 0.03, η 2 = 0.31) [30]. The trunk muscle forces and spinal loads during sit-to-stand and stand-to-sit activities were analysed by Shojaei et al. [31]. In the experimental group, there were 10 males with unilateral transfemoral amputation (TFA), and the control group was composed of 10 nonamputees. All participants were military personnel. Participants were asked to perform five consecutive sit-to-stand and reverse movements back to sitting position. There was a force platform under their feet and a 23-camera motion capture system analysing fullbody kinematics. The main effects were differences in peak compression (2556 N SD: 731 in TFA; 2208 N SD: 421 in the non-amputee group) and anteroposterior shear forces (373 N SD: 144 in TFA; 221 N SD: 118 in the non-amputee group). The results clearly show that there were larger spinal loads among the amputees during both sit-to-stand and reverse movement.
Another very important research focused on the kinetic effort of the trunk during ascent and descent of the steps was conducted by Gaffney et al. [32]. In this study, 7 men with unilateral transtibial amputation and 7 men who were able were enrolled. The task for participants was step ascent and descent from a 20 cm platform, while their movement was analysed by 8 near-infrared cameras (63 reflective markers were instrumented to obtain all body kinematics). Peak moments were compared between groups and between limbs during the loading phase. Peak posterior translational trunk moments during vertical thrust of ascent were higher in amputees with severed limbs than in intact limbs in the sagittal plane (p = 0.01, g = 1.52 (1.6 2.64)), which can place more strain on the lower back extensor muscles. In the transverse plane, the maximum axial translation moment toward the leading stance foot was higher among amputees when leading with prosthesis or intact limb compared to the healthy group during weight acceptance (p < 0.01, g = 2.36 (1.62 5.01) in transtibial amputees; p = 0.01, g = 1.47 (1.01 2.94) in the control group). During the descent movement in the sagittal plane, the maximal translational moments of the anterior and posterior trunks were greater among amputees when leading with the intact limb compared to control subjects (p < 0.01, g = 1.83 (1.48 3.7) experimental group and p = 0.01, g = 1.16 (0.4 3.16) in the control group). In the transverse plane, the peak moment of axial rotation of the trunk toward the leading stance foot was greater among amputees when they step onto the amputated or intact limb compared to the healthy control group (p = 0.01, g = 1.13 (0.01 3.78) experimental group p = 0.017, g = 1.45 (0.3 4.28)) [32].
Murray et al. [33] conducted similar research focussing on biomechanical compensations of the trunk and lower extremities during the step task. The participants in this study were divided into three groups. They included amputees with (n = 10) and without (n = 9) diabetes mellitus as well as an able-bodied control group (n = 11), in contrast to all previous research authors. The steps were 60 cm × 40 cm × 20 cm, put over an adjacent force plate, and the subjects stepped onto them while standing on an associated force plate. Participants in the two-step descent stepped onto the force plate after starting on the step. A total of 63 reflective markers were applied to the participants' bodies, and the trunk was modelled as a single stiff section. The findings showed that amputees had excessive and asymmetric trunk motion, as well as abnormal joint moments in the lower back and lower limbs when compared to persons with diabetes and healthy people. For all groups, the trunk flexed forward during the entire ascending step in the sagittal plane. Peak trunk flexion among amputees was bilaterally comparable (limb with amputation = 31.8 (7.2); limb with an intact limb = 31.4 (7.5); p = 0.8). Amputation patients, but not the diabetic group, showed more trunk flexion than the control group (21.6 SD: 8.60, p = 0.0017 for the amputated leg; p = 0.02 for the intact limb). In comparison to the diabetes group (p = 0.003) and the control group (p = 0.001), the transtibial amputee group produced more peak low back extension moments on the severed limb during ascent kinetics. Stepping onto the intact limb during step descent caused the experimental group to experience low back extension moments that were five times higher than those experienced while stepping onto the severed leg. Similar to step ascent, amputees without diabetes displayed a greater hip extension moment when stepping onto an artificial limb (control group p = 0.003, 22.6 SD: 5.50 TTA, 16.7 SD: 2.40 DM, 16.2 SD: 4.50). In Table 5, study characteristics are listed.

Trunk Kinematics during Walking Activities
Banks et al. [34] asked a crucially important question. During their study, researchers were analysing if lower back demands can be reduced by improving gait symmetry. The small experimental group consisted only of five amputees and five able-bodied participants. On a treadmill, the subjects were instructed to walk with varying degrees of asymmetry. Using a full-body OpenSim model that was assessed for gait, the L5/S1 vertebral joint forces were estimated for each level of asymmetry. An intriguing finding was that symmetrical gait did not significantly differ in joint forces.
Butowicz et al. [36] researched whether trunk muscle activation patterns were influenced by walking speed. They planned the experiment with eight unilateral amputees and 10 able-bodied control groups. The experimental task was to walk on a 15 m walkway at four different speeds (1, 1.3, 1.6 m/s and self-selected speed). Full-body kinematics were recorded by tracking the location of 51 surface markers, and the erector spine was monitored with electromyographic (EMG). Interesting results showed that there were no differences in the first onset of thoracis erector spinae (TES) during intact stance and any speed between amputees and the control group. However, during an intact stance, amputees activated TES for a higher percentage of the gait cycle and corresponded to increased lateral trunk flexion among amputees. On the contrary to the hypothesis of the researchers, the trunk ROM remained similar at all walking speeds. Fatone et al. [37] presented similar results during the study conducted in 23 amputees divided into two groups with low back pain (n = 12) and without low back pain (n = 11). Participants walked at a self-selected comfortable walking speed along the walkway while videotaping using an 8-camera digital motion analysis system. Again, the pelvic sagittal plane movement patterns were very similar between groups. Only a small increase in anterior pelvic tilt was observed in the LBP group. Based on the sagittal lumbar spine motion, there was no significant difference in the number of patients in the LBP group (W2(1, n = 21) = 0.43; p = 0.84; phi = 0.045). A total of 46% of the participants with LBP had the extension pattern, while 54% did not. The research of the kinematics of the lumbar spine during locomotion and low back pain among amputees was expanded by Morgenroth et al. [38] by including a new set of participants. They consisted of 6 healthy, able-bodied control groups and 17 amputees (split into the LBP group of 9 and the non-LBP group of 8). There were three dynamic walking tests run. Once more, there were no appreciable differences between the LBP and non-LBP groups in the sagittal or frontal plane of lumbar spine excursion during locomotion. Only the transverse plane rotation between the LBP and non-LBP groups showed statistical significance.
Golyski and Hendershot [39] questioned earlier studies and devised a plan for observing how the trunk and pelvis move during transient twists in amputees. Twenty spins comprising a 90-degree change in direction to the left and right were executed by eight participants with unilateral lower limb amputations and five able-bodied control groups. A 27-camera motion capture system was used to acquire and analyse full-body kinematic data. The primary findings were that, during spin turns, there were no significant differences in the frequency of any coordination mode during the stance or swing phase (p > 0.082). On the other hand, amputees were more likely to exhibit transverse plane pelvis phase coordination during step turns (p = 0.036).
In another study, researchers tried to analyse trunk-pelvic motion during walking with hip strength and knee joint moment among amputees. Butowicz et al. [40] took into account 24 male amputees and 8 able-bodied control groups. No one experienced LBP at the time of the tests. Isometric hip abductor strength was assessed using a hand-held dynamometer as participants walked at a speed of 1.3 metres per second on a 15 m boardwalk. Surprisingly, there was no connection between pelvic or trunk mobility and hip abductor strength. The scientists noticed that amputees had higher trunk lateral flexion and acceleration during stride than the healthy control group. Esposito and Wilken [41] decided to enrol a group of 16 transfemoral amputees and 12 able-bodied control groups and analyse the relationship between pelvic-trunk coordination in LBP.
Amputees were divided into two groups: 9 with LBP (TFA-LBP) and 7 without LBP (TFA-NP). Participants completed a walking activity on a 20 m walkway and their movement was captured with the motion capture system. The results showed that amputees with and without LBP exhibited coordination of transverse plane movement similar to that of able-bodied (p = 0.966). Study characteristics are presented in Table 6.

Discussion
The main goal of this review was to analyse differences in physical activity as an approach to treatment of LBP among amputees. The task was impossible to accomplish due to the extremely low number of articles corresponding to this topic. The impact of exercise treatment on LBP in the able-bodied population is widely researched. According to the most recent studies [42][43][44], exercise lowers pain when compared to no therapy, standard care, or a placebo. We can also discover evidence that exercise enhances capacity when compared to therapies such as electrotherapy or education, and it may be more beneficial than hands-on therapist treatment [45]. We also know that the most successful programmes included at least one or two sessions of Pilates or strength exercises each week. Sessions of shorter than 60 min of core-based, strength, or mind-body exercises are also beneficial, as are training programmes of 3 to 9 weeks of Pilates and core-based exercises [16]. When it comes to the amputee population, we can find two research articles that include some types of exercise in the treatment of LBP. Both with satisfactory results for participants who introduced trunk muscle exercise and a back school programme. We decided to include and analyse in this review articles that examine the biomechanics of the trunk, pelvis, and lower extremities that can lead to LBP during various tasks and activities. Although the included studies investigated the origins of LBP in amputees, many of them were of low quality and had significant limitations. The very limited number of trials in physical investigations is a critical limitation, implying that the results may not be typical of the overall population. Furthermore, comparing the amputee population with the able-bodied control group is debatable and makes it difficult to draw convincing conclusions. Another significant restriction is that previous research concentrated on traumatic amputees, although we know that peripheral arterial disease and diabetes are the leading causes of lower extremity amputations [46,47]. Traumatic amputation occurs more frequently in younger, active people who may not yet experience LBP. The authors also paid much less attention to the risk of factors for LBP. Only in four studies was LBP questioned with standardised tools, and in the general population we have a strong correlation between age, gender, abdominal obesity, smoking and the risk of LBP [48]. Furthermore, any current study that takes into account the psychological stress that can lead to LBP among amputees, but we know that limb amputation is an irreversible act that is sudden and emotionally devastating to all patients [49]. This review has identified topics that should be investigated more thoroughly in future studies. Several trunk and pelvic kinematic patterns occur in amputees that contribute to an additional load on the spine during different activities. Hendershot and Actis [22][23][24]28] examined activities such as sitting and stepping, proving that for people with LLA those tasks are more demanding for muscle and spine joints. Butowicz [30] points out that people with LLA and LBP demonstrated impaired postural control of the trunk. The persistence of LBP among amputees may be the result of neuromuscular adaptation in the proximal structures, but further research is needed to determine the reason behind these increases. Many of the older patients after LLA spend most of their days sitting in a sitting position. Taking into account the results of the presented studies, it seems very reasonable to address stability exercise and strengthening trunk movements as a golden standard of rehabilitation. A similar approach was taken on the able-bodied population resulting in very satisfying outcomes [50,51]. To maintain mobility after LLA, patients must adopt movement compensations to account for loss of knee or ankle function in the amputated limb. Reduced trunk muscle activation and increased intramuscular fat may be potential intervention targets after LLA [52]. Additionally, the intermuscular fat content of residual limb endings rises over time [53]. Murray [33] was the only author to take into account amputees with diabetes and performed stepping tasks among the experimental and control groups. Half of people with diabetes have nerve damage that influences their control of the neuromotor muscles. Murray's study shows that people with LLA and diabetes exhibited excessive and asymmetric trunk motion, which was assisted by an asymmetric load on the low back. There are many good quality articles that show a significant improvement in pain and quality of life after exercise therapy among patients with diabetes [54,55].
Walking is a highly repetitive task that exposes people with LLA to large alternations of spinal loads. To reduce the stress on the weaker hip abductor muscles on the side of the residual limb, people with LLA prefer to engage in considerably increased lateral bending towards the prosthetic limb during single-limb stance and double-limb support. Greater value in the intact limb imply that it is important in maintaining stability and optimising body progression throughout various tasks [56]. There is moderate to strong evidence that chronic LBP patients have different walking gaits than healthy controls in the able-bodied population [57]. Fatone and Morgenroth [37,38] tried to assess lumbar and pelvic kinematics during gait in transfemoral amputees. Both studies concluded that differences in lumbar and thoracic motion do not appear to be independently related to LBP. Furthermore, some studies that investigated the mechanics of altered amputees did not report the prevalence of LBP, so it is challenging to draw mechanical conclusions that relate altered gait to LBP. Esposito [41] concluded that individuals with LLA with and without LBP had similar transverse plane movement coordination to able-bodied people and that only amputees could have found the increases in speed problematic enough to achieve a change in transverse plane coordination. Naturally, it could lead to the statement that we should exercise gait patterns and improve gait symmetry in the LLA population, but Banks [34] tried to answer this theory in his research. The study revealed that training amputees to walk more symmetrically may not decrease low back demands since they already locomote at a desired degree of asymmetry, which clearly reduces L5/S1 joint loads. Exercise therapy using, for example, the Pilates method can improve weight discharge in gait and reduce LBP [58].
Gait and posture bring us to the use of a prosthetic leg during walking and other activities of daily living. The purpose of prosthetic design is to replicate the anatomy and function of the missing limb. [59]. The studies presented in this review did not take this into account because the designed groups were mainly composed of high-functioning people with LLA who often used electronic knee prostheses. Many prosthetic parameters, including limb weight, number of artificial joints, prosthetic length, and prosthetic attachment, may contribute to the beginning of LBP in the general population of patients with LLA [60]. Additionally, the weight of the prosthesis device itself demands more energy of the low back, hip muscles, and core musculature. This brings us to the conclusion that core muscle strength and back muscle endurance are needed to walk with a prosthesis efficiently. The benefit would be to focus on stabilisation exercise therapy. Resistance training (especially in the core and lumbar extensor muscles) may enhance gait metrics, according to emerging research [61,62].
This systematic review's limitations may have introduced some possible biases. The low confidence of evidence is a result of some of the included reviews' poor methodological quality and the RCTs that underlie them. The overall number of participants was low for the majority of outcomes and time periods. High levels of overlap were produced by the inclusion of the same RCTs in several evaluations, and the inconsistent findings of the review authors' methodological quality assessments of some RCTs are further a cause for concern.
LBP is one of the most frequent conditions for which people seek out primary care services from a physiotherapist or general practitioner. For this condition, it's crucial to offer precise, efficient management. To support the professions of physical medicine and rehabilitation, physical therapy, and prosthetics in achieving the objective of standardisation in the care of patients with LLA, evidence-based practise must go beyond just using outcome measures as evaluation tools. Understanding the patient's response to therapy (improvement, no change, decrease) enables doctors to provide care more quickly, lowering overall rehabilitation time and increasing patient outcomes [63].

Potential Benefits of Exercise-Based Rehabilitation in People with LLA
Current research in various therapeutic groups supports the advantages of strengthening, balance, and stability training on back pain. The goal of the exercise programme should be to improve muscle tissue quality, strength, endurance, and balance, as well as to reduce movement asymmetries during locomotion and weight-bearing duties. Amputees suffering from muscular atrophy and fibre type shifting may benefit directly from improvements in muscle growth, strength, and endurance. Resistance exercise has been shown to dramatically decrease discomfort [64]. Resistance training causes morphological modifications in skeletal muscle such as muscle fibre hypertrophy and cross-sectional area, as well as an increase in the amount of connective tissue surrounding muscle fibres. [65].
Improved strength, endurance, and stability of knee extensors and flexors in transtibial LLA may allow for improved stability of the intact limb during stance and propulsive forces provided by the limb during walking. Other advantages include better walking endurance, confidence in gait speed, and greater independence in everyday tasks [66]. Lunges, plank variants, and multidirectional reaching increase control and strength of the core, hip flexors, and adductors, which help to stabilise the spine. Core and balance training is critical for improving distal mobility and the generation of strong limb motions [67]. Following long-term general prescription recommendations, regular resistance exercise participation is suggested to address physiological and mechanical difficulties that support low back discomfort.
Exercise therapy may or may not be the best solution depending on a number of factors, including the patient's preferences and values, the physiotherapist's clinical skills, and the results of the study. This is in accordance with the principles of evidence-based treatments.

Conclusions
Our findings show that intervention groups employing an exercise intervention had considerably lower chronic low back pain than other therapies, but there are not enough trials to produce evidence-based practice. Exercise therapies, such as core strengthening and stability exercises, have the potential to significantly reduce low back discomfort in amputees. For future studies, better methodological research that compares the different amputee groups (TFA/TTA/able body/LBP/non-LBP/unilateral/bilateral) systemically in terms of personal factors influencing LBP, their anatomical structures, and their biomechanical movements under various conditions is needed for a better understanding of the mechanisms of LBP among people with LLA. • The results of this systematic review may have been biased because the only study language used in the included RCTs was English and the study sample sizes were small.

•
The exercise therapy was inconsistent throughout the included RCTs, and the controls' treatments varied as well.

•
High levels of heterogeneity within relatively small segments of the literature may have had an impact on the validity of the review.

•
The risk of bias of presented studies was moderate or debatable. Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.