1 Introduction

Amputation, which is defined as the total or partial removal of a limb, is a measure used to restore the health of a patient for whom there are no other viable options for recovering an injured limb (Souza et al. 2019). In 2017, it was estimated that 57.7 million people worldwide were living with traumatic limb amputation (McDonald et al. 2021). In the United States of America (USA), it is estimated that almost two million people have undergone limb amputation (Ambron et al. 2021), and in Brazil, only in 2021, 66,818 new limb amputations were performed, with more than half of this number, 37,124, referring to lower limb amputation (DATASUS, 2022).

The high prevalence of amputations is a significant public health problem, responsible for high morbidity and mortality rates, in addition to negatively impacting the socioeconomic and health aspects of the amputee (Souza et al. 2019). One of the problems that most affects amputees is the sensation of a phantom limb, a condition that affects up to 90% of people with amputation, which refers to the persistent, and generally painful, sensation of the removed limb (Ambron et al. 2021; Erlenwein et al. 2021). It is known that phantom limb pain (PLP) is associated with various factors, including changes in body schema (Erlenwein et al. 2021), and about 25% of people with amputation develop a phenomenon called phantom limb telescopy (PLT) (Stankevicius et al. 2021), in which the sensation is that the limb is shortened (Ambron et al. 2021). In addition, there is another type of pain that can affect amputees, residual limb pain (RLP), also known as stump pain, which is a painful sensation in the part of the limb that remained, usually in the stump, and is related to possible complications such as vascular abnormalities, skin diseases, soft tissue and bone disorders, wound healing difficulties, among others (Zaheer et al. 2021). About 47% of people who undergo amputation are affected by RLP (Ghoseiri et al. 2018).

Other challenges during the rehabilitation process of individuals with amputation refer to the physical conditions associated with walking, such as atypical gait patterns, asymmetry in the weight load between the residual limb and the prosthesis, decreased balance, increased fear of falling, and low perception of the prosthesis as part of the body itself (Bourque et al. 2019; Charkhkar et al. 2020; Escamilla-Nunez et al. 2020; Raspopovic 2020). Having a lower limb amputated increases the risk of falling compared to individuals without amputation (Neptune 2022), and 40.4% of falls suffered by amputees cause harm and hospital readmission (Safee and Osman 2021).

Virtual Reality (VR) is an emerging option for rehabilitation in the health sector (Stanica et al. 2020). In particular, the use of Immersive VR (IVR) rehabilitation programs - systems that enable complete immersion in a virtual environment using a head-mounted display (HMD) - suggests an increased capacity for brain plasticity toward self-recognition in a virtual body (Lee 2019). Even if individuals remain motionless, the Action Observation (AO) has the potential to produce the same neural activity as the performance of the task itself (Yoshimura et al. 2020). Specific areas within the premotor, motor, and parietal cortices are activated during the planning, execution, and observation of cognitive motor control tasks, and this association between areas is defined as the mirror-neuron system (MNS) (Yoshimura et al. 2020). Regarding PLP, the literature points to promising potential in rehabilitation with VR through its use to early restore the body schema and reduce sensorimotor inconsistencies (Erlenwein et al. 2021). Furthermore, there are several studies supporting the use of VR for physical rehabilitation and gait improvement in different health conditions, including training after amputation (Horsak et al. 2021).

Although there are successful studies on the use of this technique in clinical practice and research, the literature still lacks conclusive evidence regarding the intervention with IVR in amputee patients, as well as the protocols used (Herrador-Colmenero et al. 2018). In recently published systematic literature reviews on virtual reality as an intervention for Lower Limb Amputation (LLA), one focused on motor components and the other on pain management, concluded that the results were inconclusive due to a limited number of studies employing proper randomized controlled trials (RCT). The first review found only 4 RCTs, and the second review found only one (Hao et al. 2023; Cheung et al. 2023). It is evident that there is a need to understand why conducting RCTs with this specific population and intervention presents a challenge. Therefore, this research aims to make a valuable contribution to the scientific literature by conducting a small RCT to discuss the feasibility of trial studies concerning rehabilitation with IVR for patients with LLA, focusing on outcomes such as PLP, PLT, RLP, and balance. The results and discussion intend to assist investigators in better preparing their intervention designs for future research.

2 Materials and methods

2.1 Participants

Forty-one participants with LLA, who were patients of a Specialized Rehabilitation Center (SRC), a referral rehabilitation service for 37 municipalities in the state of Rio Grande do Sul (RS), Brazil, took part in this research. The sample consisted of adult males and females, with active medical records at the time of the intervention, who agreed to participate in the study and signed the Free and Informed Consent Form. The research was evaluated and approved by the Research Ethics Committee (CEP) of the Federal University of Health Sciences of Porto Alegre (UFCSPA) under the number 44582421.0.0000.5345.

To be eligible for the study, the participant had to be over 18 years old and have undergone a major lower limb amputation. The exclusion criteria were sensory deficits or vestibular system dysfunctions that made it impossible to use the IVR equipment, cognitive deficits that prevented the understanding of the protocols involving the IVR sessions, diagnosis of severe depression or severe psychiatric disorder, and presence of seizures, nausea, and/or vomiting in their clinical history. All criteria were verified in the patient’s records, according to the initial assessment of the SRC multidisciplinary team, and 103 individuals with LLA were able to participate in the study. All were invited to participate in the SRC facilities through individual scheduling and 41 accepted to take part in the study. These participants were randomly assigned in a 1:1 ratio, using a randomizer software to allocate participants into groups. Twenty individuals were assigned to the Control Group (CG), and 21 were assigned to the Intervention Group (IG).

During the experiment, 41 participants had signed the informed consent, but only 21 participants, 12 in the CG and 9 in the IG, completed the research. Of the 20 individuals who were lost, 10 dropped out before the initial assessment due to concerns that the meetings would disrupt their personal and work routines, with 15 remaining in the IG and 16 in the CG. During the implementation of the protocol, there were 7 more losses, all from the IG: 4 because they considered that the meetings disrupted their personal and work routines, 2 who did not adapt to the device and showed strong symptoms of malaise, and 1 who died. As there were only withdrawals in the IG, 4 participants from the CG were relocated to the IG, resulting in 12 participants in each group. However, 3 of the 4 relocated participants were lost: 1 had moved to another distant city, 1 considered that the meetings would disrupt their personal and work routine, and 1 did not adapt to the device and showed strong symptoms of malaise. Figure. 1 summarizes these movements of participants throughout the study. Considering that most participants were not residents of the city where the SRC was located, it may contribute to a high rate of withdrawal from the study.

Fig. 1
figure 1

Number of participants and losses during each step of the research

Full size image

2.2 Study design

All participants underwent an initial assessment lasting between one and two hours, during which an assessment form was filled out with anthropometric, demographic, and clinical data. The Activities-Specific Balance Confidence (ABC) Scale was applied, and information regarding the size of the phantom limb was collected to attest to PLT. The McGill Pain Questionnaire (MPQ) was also applied to gather information about RLP and PLP. For those who did not feel the phantom limb, phantom limb size information was not collected, nor was the MPQ applied for PLP. Subsequently, all participants in the CG continued with the conventional care they were already receiving at the SRC, while the IG participants, in addition to conventional care, also underwent the IVR protocol.

At each intervention, the participants were tested for pain at the beginning and end using the Numerical Analog Scale (NAS). After two consecutive sessions, they were evaluated for the feeling of presence in the virtual environment using the NAS, and about cybersickness using the Sickness Simulator Questionnaire (SSQ). At the conclusion of the protocol sessions, all initial assessments were repeated once. Pre and post-evaluations were performed using the double-blind method.

2.3 Evaluation procedures

The evaluation form was developed based on the questions recommended by the International Society of Prosthetics and Orthotics (ISPO) for participants in prosthetic research (Lemaire and Wong 2013). To determine the size of the phantom limb and the presence or absence of PLT, participants were asked to point to different parts of the phantom limb with their eyes closed, which was a technique previously used by Mayer et al. (2008). MPQ was used to characterize PLP and RLP. This questionnaire is widely used in pain research, including studies involving phantom pain (Perry et al. 2018), and it categorizes pain characteristics into four subgroups: sensory, affective, evaluative, and miscellaneous (Lima et al. 2021). Mobility and balance were assessed using the ABC Scale, which measures balance during the performance of 16 activities of daily living (Branco 2013). Both the ABC Scale and the MPQ were translated and adapted to Brazilian Portuguese (Marques et al. 2013; Pimenta and Teixeira, 1996).

2.4 RVI protocol

All procedures were carried out in a clinical room at the SRC, with the appropriate authorization from the management team. Patients were offered free transportation to the SRC by their city of residence, although some chose to travel by private vehicle. The scientific literature does not provide a clear consensus on the number of intervention sessions required to observe significant effects, ranging from 6 to 25 total sessions (Cassani et al. 2020), then, in this study, it was decided to conduct 16 sessions, striking a balance within this range. These sessions lasted 15 min each, as this duration allowed us to complete all sets of the four VR exercises and minimized discomfort associated with longer HMD sessions. To make it convenient for participants, especially those traveling from other cities for the research, and to intensify the protocol, we scheduled two sessions per week on the same day. Consequently, once a week, two sessions were performed in sequence, separated by a 10-minute rest interval, for a total of 8 weeks of intervention. In contrast, traditional rehabilitation sessions at the SRC lasted approximately 40 min, encompassing not only limb exercises but also therapist-assisted limb manipulation, along with longer rest periods between exercises, aspects not present in the VR sessions.

2.5 IVR equipment and virtual environment

During each session, participants remained seated with their torso supported and used an HMD equipped with a smartphone that projected the images directly onto the display. As the use of the IVR application consumes a significant amount of smartphone energy, we used two smartphones for this research. When one phone’s battery was depleted, it was placed to charge, and the sessions continued with the other phone, creating a smartphone relay. The Gear VR, manufactured by Samsung, was used as the HMD, and the smartphones used were the Galaxy S6 and Galaxy S7 models, also from Samsung.

The virtual environment (VE) in which the patients were immersed was developed by researchers from the Neuroscience and Virtual Reality Laboratory at the Federal University of Health Sciences of Porto Alegre (UFCSPA). It represents a room with physical exercise equipment, and the avatar, a virtual body that replaces the real body (Kokkinara et al. 2016), is viewed from a first-person perspective. During the session, the avatar performs between two or four sets of four lower limb exercises, including lunge with a barbell and weight plates on the shoulders, step up, squat with a barbell and weight plate on the shoulders, and knee extension with weight pads (Fig. 2). Participants were instructed to remain seated and observe the exercises without moving their bodies (Fig. 3), making it an intervention of AO of the exercise in the first person, but without any real movement of the body. Therefore, this protocol offered no risk of falling to the participants. In cases where participants reported discomfort or malaise, the IVR session was interrupted. All photographs of the participants were authorized according to the SRC protocol.

Fig. 2
figure 2

Avatar from a first-person perspective performing knee extension with weight pads

Full size image
Fig. 3
figure 3

Participants during AO IVR Session, seated with their torso supported and using an HMD equipped with a smartphone

Full size image

2.6 Data analysis

The results of the qualitative variables were presented as frequency and percentage, while the quantitative variables were presented as the mean and standard deviation for symmetrical data and median and interquartile range for asymmetrical data. Normality was checked using the Shapiro-Wilk test and by inspecting histograms. Groups were compared using Student’s t-test, Mann-Whitney test, Chi-square test, and Fisher’s exact test. Pre- and post-intervention comparisons were made within each group using Student’s t-tests for paired data, Wilcoxon test, and McNemar test. Bonferroni correction was performed for the applicable comparisons. Cohen’s d size estimates for parametric tests and r for non-parametric tests were also calculated. The cutoff point for effect size using Cohen’s (1988) criteria for parametric tests is 0.3 as a small effect, 0.5 as a medium effect, and 0.8 as a large effect. For non-parametric tests, 0.1 is considered for a small effect, 0.3 for a medium effect, and 0.5 for a large effect.

For each participant who completed the entire treatment, the average pain response in the stump and phantom limb was calculated at the beginning and end of the sessions. Initial and final measurements of the group were compared using the Wilcoxon test. The average sense of presence and symptoms were only calculated at the end of two consecutive sessions. Pain variation (stump and phantom) in the sessions was calculated for each subject by subtracting the average pain at the end from the average pain at the beginning of the sessions. Correlations between amputation time, age, immersion, symptoms, and pain variation in the stump and phantom limb were verified using Spearman’s correlation coefficient.

The Simulator Sickness Questionnaire (SSQ) score was calculated using the formula TSC = ([I] + [2] + [3]) x 3.74 (Kennedy et al. 1993). The variation of SSQ score was calculated as the final mean minus the mean at the beginning of the sessions.

The analyses were performed using SPSS statistical software (IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp.). Results with a p-value < 0.05 were considered significant.

3 Results

3.1 Participants

Tables 1, 2 and 3 present the sociodemographic, amputation, and prosthesis data for each group, respectively. There were no statistically significant differences between the groups at baseline, except for a significant correlation between the prosthesis material and the groups, with more participants in the IG using polypropylene plastic prosthesis, while more participants in the CG used acrylic resin. For the prosthesis data, only participants who had already taken their prosthesis home and were using it on a daily basis (n = 16) were considered.

3.2 Pre-intervention versus post-intervention results

The results of the ABC scale for balance, MPQ for RLP, PLP, absence, and presence of PLT are presented in Tables 4 and 5, and 6, respectively. No statistically significant differences were found between the CG and IG after the IVR protocol.

Table 1 Participants socio-demographic data per group
Full size table

3.3 Intervention group: pain, cybersickness and immersion

The results of the PLP and RLP assessments by NAS for the IG participants were compared to scores before the first session and after the last session and are presented in Table 7. There were no statistically significant differences between pre- and post-intervention scores. The mean and standard deviation results for RLP variation, PLP variation (variations calculated as the difference in pain measurement after the beginning of the sessions), sense of presence in the VE, and cybersickness symptoms in the group that underwent the intervention are shown in Table 8. The SSQ scores are shown in Table 9. The results for correlations of amputation time, RLP variation, PLP variation, VE immersion, and cybersickness symptoms with age and time since of amputation are presented in Tables 10, 11, and 12.

There was a significant correlation between RLP and amputation time (r= -0.742, p = 0.022). Removing the outlier participant (amputation time of 288 months), the significant correlation remained (r= -0.794, p = 0.019), as shown in Fig. 4. There were also statistically significant correlations for PLP and headache (r = 0.751, p = 0.020), PLP and variation in disorientation and oculomotor cybersickness symptoms, as shown in Figs. 5 and 6, (respectively r = 0.761, p = 0.017; r = 0.778, p = 0.014); RLP and increased salivation (r = 0.679, p = 0.044); age and variation in cybersickness symptoms of nausea, as shown in Fig. 7 (r= -0.760, p = 0.018); and Burping and Immersion (r= -0.853, p = 0.003).

Table 2 Participants amputation data per group
Full size table
Table 3 Participants prosthesis data per group
Full size table
Table 4 Activities-specific balance confidence scale results per group
Full size table
Table 5 McGill pain questionnaire results per group
Full size table
Table 6 – Phantom limb telecopy results per group
Full size table
Table 7 NAS results for PLP and RLP of intervention group participants
Full size table

4 Discussion

4.1 Balance

In a previous study that analyzed the neural activity of participants undergoing AO of upper limb movements in IVR and non-immersive VR, it was shown that those undergoing IVR exhibited improved rhythmic patterns with better discrimination of spatial characteristics in the brain (Choi et al. 2020). This finding supports the clinical results of another study that analyzed the performance of using an upper limb prosthesis and found that participants who underwent AO of a bilateral activity with the upper limbs in IVR performed significantly better in real life with the prosthesis than those who did not undergo AO in IVR or those who had undergone AO in non-immersive VR (tablet) (Yoshimura et al. 2020). However, in this study where participants were subjected to AO of lower limb exercises, the same effects were not observed, as there was no significant difference between the groups or between pre- and post-intervention for balance during specific activities.

Although no significant differences were observed in the ABC Scale scores after the intervention, the mean confidence of the IG increased to 63.9%, demonstrating higher balance scores (closer to 100% confidence). In comparison, that of the CG decreased to 47.4%. These scores suggest a promising improvement, even if not yet sufficient, and corroborates the practical significance observed in the Cohen’s d size effect between groups (0,56), as a medium effect. One hypothesis is that, unlike the upper limbs, which we typically observe when performing an activity, we do not habitually observe the lower limbs. Developing movement skills with the lower limbs primarily occurs through other means of sensory feedback, such as proprioceptive and tactile systems (Petry et al. 2018; Stroppa-Marques et al. 2019). To achieve better results, the VE of AO for lower limb rehabilitation purposes should likely incorporate balance exercises that commonly require the association of observation with the performance of using the lower limb, such as circuits with obstacles or moments that require greater observation of the lower limbs, such as performing exercises in front of a mirror.

Table 8 Mean and Standard Deviation of RLP variation, PLP variation, presence in VE and cybersickness symptoms in participants of Intervention group
Full size table
Table 9 Cybersickness symptoms comparison between first and last session into three categories and total
Full size table
Table 10 Correlation results of pain variations, time since amputation and presence in VE with age, time since amputation and pain variation in participants of the Intervention Group
Full size table

4.2 PLP, PLT and correlations

Recent studies have indicated significant changes in phantom limb pain (PLP) in individuals undergoing immersive virtual reality (IVR) treatment using pre- and post-intervention protocols for both upper and lower limbs (Ambron et al. 2018; Chau et al. 2017; Osumi et al. 2019; Rutledge et al. 2019). However, it should be noted that these studies used games that reproduced the person’s real movement in the virtual environment, which is different from what was carried out in this study, where a virtual environment of AO was prioritized.

Individuals with phantom limb sensations typically experience the feeling and movement of the missing limb, but they have no control over these sensations and actions (Demidoff et al. 2007). The fact that the protocol used in this study was an AO, which was also not controlled by the participants, may have contributed to the lack of change in pain patterns. Nevertheless, it is worth noting that there was no worsening of PLP in the intervention group, indicating that the treatment did not harm the patients.

On the other hand, the findings related to PLT are noteworthy. In the post-intervention period, patients in the control group (CG) who had previously experienced PLT and continued to experience phantom limb sensations did not report any changes in the size of the phantom limb. In contrast, two out of three patients in the intervention group (IG) who had not shown improvement in phantom limb sensations reported a reduction in PLT, and their phantom limb sensations became more aligned with their actual limb size. It is possible that seeing the avatar’s limb in its normal size in the immersive virtual reality (IVR) environment was beneficial for these patients, aiding in the reorganization of the phantom limb’s spatial representation in the patient’s body schema at the neurological level. This is particularly relevant since phantom limb sensations are associated with the reorganization of somatosensory and motor cortices (Costa et al. 2021). However, it should be noted that the sample size was not large enough to produce significant results in this regard.

Table 11 Correlation results of cybersickness symptoms with age, time since amputation, pain variation, presence in VE in participants of the Intervention Group
Full size table
Table 12 Correlation results of cybersickness categories and total variation with age, time since amputation, pain variation, presence in VE in participants of the Intervention Group
Full size table
Fig. 4
figure 4

RLP Variation for amputation time in months ( SPSS was used to create the artwork)

Full size image
Fig. 5
figure 5

Disorientation category variation for PLP variation (SPSS was used to create the artwork)

Full size image
Fig. 6
figure 6

Oculomotor category variation for PLP variation (SPSS was used to create the artwork)

Full size image
Fig. 7
figure 7

Nausea category variation for age variation (SPSS was used to create the artwork)

Full size image

Furthermore, in the intervention group, patients who reported more PLP descriptors were also more likely to experience the symptom of headache during the IVR sessions, and this correlation was found to be statistically significant. This data may be related to neurocognitive and neuromotor compensation resulting from the adaptation caused by the absence of the limb in the patient’s actual body (Sava et al. 2018).

4.3 RLP and correlations

Previous literature reviews (Ghoseiri et al. 2018; Xavier et al. 2021) have identified various pharmacological and non-pharmacological treatments for RLP and PLP, however virtual reality was tested as a treatment only for PLP. Although the changes found in this study regarding RLP were not significant, a very interesting correlation was found among the data of the patients who received the intervention. Patients reported an increase or decrease in RLP immediately after the IVR session, and this increase or decrease was correlated with the time since amputation. A statistically significant negative correlation was found between the NAS pain score and the time since amputation; in other words, the longer the time since amputation, the lower the RLP after the IVR session. In general terms, the IVR session demonstrated positive effects in reducing RLP as an immediate effect in patients with a longer time since amputation.

This finding is likely related to the potential of IVR as a pain distractor, as already found in other studies that evaluated IVR as an analgesic procedure in different populations, such as changing bandages in patients with burns and cystoscopy procedures (Georgen and Freitas, 2022; Marques et al. 2021). The fact that it is related to patients with a longer amputation time indicates that IVR seems to be more effective in treating chronic pain in the residual limb than acute and recent pain for this specific population. Further studies with a larger number of participants are necessary to investigate these effects and possibilities of use more comprehensively.

In addition, the patients who underwent the intervention and reported more PLP descriptors were also the ones who frequently experienced the symptom of increased salivation during the IVR sessions, and this correlation was statistically significant. This is an interesting finding, as there are reports in the scientific literature that suggest the region of the mouth and the face close to the mouth receives migrating innervation from the sensory areas of the phantom limb. Although this phenomenon is better known concerning the upper phantom limb (Lemons 2021).

4.4 Cybersickness and VE immersion

The only symptom that showed a statistically significant negative correlation with the sense of presence in the virtual environment was the feeling of burping. The greater the sense of immersion in the VE, the less frequently the symptom of burping appeared. Burping is classified as a subtype of cybersickness, specifically within the nausea subgroup. In this case, it is supragastric burping, which means that air does not enter the stomach, and the symptom is considered a voluntary response of patients to the unpleasant sensation in the upper abdomen or chest (Zad and Bredenoord 2020). The literature indicates that there are warning signs that predict burpings, such as abnormal tension or an uncomfortable feeling of pressure in the retrosternal region (Zad and Bredenoord 2020). Therefore, the correlation of this symptom with immersion may be linked to these factors. Two hypotheses can be considered. First, the participants did not feel present in the virtual environment during the IVR exercises but insisted on completing the entire session, which may have made them feel more tense than normal. The abnormal tension culminated in burping. Second, the patients did not feel present in the virtual environment due to sensory inconsistencies between movement and vision, and these inconsistencies generated nausea and abdominal discomfort. The participants chose to continue the protocol, and the digestive symptoms culminated in a feeling of pressure in the retrosternal region and, finally, burping.

4.5 Feasibility and cost-benefit analysis for study replication

The results suggest that implementing physical exercises using IVR may effectively improve balance and PLT. However, the study had limitations in terms of sample size, resulting in a lack of statistical validity. The potential benefits related to enhancing patient balance through a protocol that eliminates the risk of falling is a noteworthy aspect that deserves researchers’ attention. Similarly, the prospect of treating PLT is highly intriguing, particularly given the absence of effective treatments for this issue. Therefore, it is recommended to conduct new studies with larger sample sizes focusing on these two outcomes. However, there is a need to pay special attention to strategies for obtaining a larger sample size and engaging participants. This research was carried out in a facility serving amputee patients from 37 municipalities, covering a vast region of the state. Although the plan was to conduct a small RCT, all patients were invited to participate. Despite this extensive reach, the sample size was ultimately limited. Within this demographic, there is a high prevalence but reduced interest in research participation.

One strategy for addressing this situation is to consider multicenter partnerships to gather more comprehensive data. Currently, there are 305 accredited and operational SRCs throughout Brazilian territory, many of which are located in cities with academic research development centers (Ministério da Saúde do Brasil 2024). Researchers should be encouraged to seek collaborations among universities to conduct more robust work. Regarding costs, the program used for this research was developed by NERV and will be provided free of charge to researchers interested in using it for research purposes, with proper references to the developers and ensuring non-commercial use only - it can also be used outside of Brazil, since there is no speaking scenes during the video. The HMD and smartphones used in this research are low-cost products compared to other devices used in the IVR research context and can be easily acquired without the need for substantial funding. Ultimately, the most significant cost would be the time dedicated to research work. The protocol used in this research typically takes around 4 months between evaluations, intervention protocols, and reassessments, with 3 out of 5 working days of the week fully booked for intervention sessions with research subjects. However, despite the high time cost, it is understood that researchers already allocate dedicated time in their schedules for research activities. Additionally, the protocol’s application is straightforward and can even be conducted by undergraduate students when properly trained by their supervisors.

Furthermore regarding patient participation, the present research experienced a high rate of participant withdrawal. It is known that individuals who have undergone LLA tend to exhibit social phobia, depression, and lower quality of life (Tutak et al. 2020). These emotional aspects may have impacted treatment adherence; therefore, in future studies, it would be interesting to evaluate participants’ emotional aspects and investigate correlations between these variables and dropouts. Additionally, a strategy to reduce dropouts could be to offer home-based interventions, as this could avoid exposures that trigger social phobia. However, it should be considered that there would be additional costs in terms of travel time and financial expenses related to transportation.

At last, another outcome that should be further explored is the use of IVR as a non-pharmacological resource for analgesia in patients with LLA. The results of this research indicated significant findings related to this aspect in individuals with longer amputation times, but it was only tested with participants from the IG. It is recommended to expand the dataset with a larger sample size, and if possible, to conduct a methodological study that can compare more than one type of virtual environment, one with the avatar and another without it. This would aim to understand whether the perception of a complete avatar has implications for pain sensation, or if merely using a distracting device is enough to reduce pain.

4.6 Conclusion

This research represents a pivotal feasibility study conducted through a small RCT, providing crucial insights into the challenges and potential solutions for individuals with LLA. Through a comprehensive analysis of balance, phantom limb pain (PLP), residual limb pain (RLP), and associated factors, this study sheds light on the intricate dynamics of rehabilitation in this population. Despite inherent limitations in sample size and participant retention, the findings underscore the promising role of IVR as a rehabilitation tool. Specifically, IVR shows potential in reducing PLT and potentially enhancing balance, although further research is warranted to validate these outcomes. Moreover, the exploration of pain correlations, such as those with amputation time and IVR immersion, suggests new avenues for investigation. Strategies such as fostering multicenter collaborations and exploring home-based interventions emerge as promising approaches to address challenges related to sample size and participant retention. Overall, this feasibility study, conducted within the framework of a small RCT, lays a solid foundation for future research endeavors aimed at optimizing IVR interventions for improved outcomes in individuals with LLA.