The use of adapted boxing as a rehabilitation strategy in people with diverse health conditions: a systematic review

Boxing exercise programs without direct contact with an opponent, is an adapted physical activity that integrate a series of tasks that may have positive effects on people with diverse health conditions. This study aimed to analyze the scientific evidence available on the effects of using adapted boxing as a rehabilitation strategy on people with diverse health conditions. A systematic review of the scientific literature was performed on studies which dealt with the effects of boxing as a rehabilitation strategy in people with diverse health conditions. Two modified scales were applied to assess the methodological quality of the findings. A total of 10 studies were included in the analysis, 4 of them were experimental and 6 observational. None of the studies reported injuries or other health complications derived from the exercise programs. The interventions led to improvements in cardiovascular fitness, upper body strength, body composition, coordination, gait independence and gait speed. The performance of adapted boxing is beneficial for people with diverse health conditions. Future randomized trials with larger samples would be needed to determine the most appropriate therapeutic dosing for each population.


Introduction
People with diverse health conditions experience changes in body functions and structures arising from different disorders that can result in impairments, activity limitations or participation restrictions (Üstün, Chatterrji, Kostansjek, & Bickenbach, 2003). If difficulties exist in any of these three areas, it is considered disability by the International Classification of Functioning, Disability and Health (World Health Organization, 2001). The performance of physical exercise is considered a safe, low-cost, non-pharmacological alternative which may prove useful in the rehabilitation processes of people with diverse health conditions (Pedersen & Saltin, 2015).
Traditionally, rehabilitation strategies have been based on physical exercise and were focused on aerobic and resistance exercise (Johannsen et al., 2016;Penedo & Dahn, 2005). However, over the last few years, numerous alternative modalities have emerged and its effects have undergone research, as was the case with Pilates (Byrnes, Wu, & Whillier, 2018), some varieties of martial arts (Jahnke, Larkey, Rogers, Etnier, & Lin, 2010), dancing (Strassel, Cherkin, Steuten, Sherman, & Vrijhoef, 2011) or yoga (Ross & Thomas, 2010), among others. All these modalities have been the object of scrutiny and revision, which has led to productive scientific discussions to consider the implementation of these modalities in the exercise programs for different populations.
One of the most recent rehabilitation strategy proposals which could lead to beneficial effects on people with diverse health conditions consists in boxing without direct contact with an opponent exercise programs, adapted to the particularities of each health condition. This modality integrates a eujapa.upol.cz condition (Chaabène et al., 2015). It should be noted that boxing could also be performed virtually, using acti also boosts adherence (Naugle, Naugle, & Wikstrom, 2014;Sanders et al., 2015). Adherence is a crucial aspect for the success of the physical rehabilitation program (Rhodes & Fiala, 2009). Furthermore, previous studies have shown the feasibility of these adapted interventions (Lee, Psotta, & Vagaja, 2016).
Due to the interesting characteristics of boxing without contact or performed virtually in these populations, it is important that health and rehabilitation professionals can easily access all existing evidence regarding its effects (Dijkers, Murphy & Krellman, 2012). This is particularly important in order to have the knowledge on how to prescribe and adapt the boxing programs to meet the requirements of each health condition, as well as the personal needs and abilities of specific individuals. These objectives can be achieved by conducting systematic reviews that synthesize the scientific knowledge available on the subject (Higgins & Green, 2008).
Despite the apparent benefits of therapies based on adapted boxing , no systematic review has analyzed the efficacy and usefulness of real or virtual adapted boxing as a rehabilitation strategy in people with diverse health conditions. In addition, the absence of scientific evidence on its feasibility and safety at the time of implementing practical therapeutic interventions may contribute to a false perception of danger a rejection (Hermand, Mullet, & Rompteaux, 1999). For this reason, the present systematic review aimed to analyze the research focused on the impact of boxing in people with diverse health conditions. The objective is to provide updated information, which might be of interest to health professionals and practitioners responsible for prescribing physical exercise as part of a rehabilitation process.

Search strategy
A systematic review of studies focused on boxing as a rehabilitation strategy in people with different health conditions was carried out by searching PubMed, Cochrane Database of Systematic Reviews, Scopus, SportDiscus and Dialnet up to July 2018. The search was based on the Population, Intervention, Comparison and Outcome (PICO) strategy. Following the recommendations from regarding the population and intervention were used, in a combination of standardized MeSH and free-text terms. Therefore, the following combination of keywords and Boolean operators was of relevant references within included articles or different systematic reviews were performed manually.

Study selection and inclusion criteria
Following the PICO strategy, the inclusion criteria were: a) to perform physical exercise programs based on boxing as a rehabilitation strategy in people with diverse health conditions related to impairments, following the International Classification of Functioning, Disability and Health (World Health Organization, 2001); b) to report the specific effects of such practices; and, c) to be published in English, Spanish or Portuguese. There were no restrictions regarding age or the types of health conditions. Those interventions in which boxing was part of a combined therapy whose effects were not reported separately were excluded.
Two reviewers (MSL and MS) examined the titles and abstracts of each item independently and created a draft list of eligible studies. After an independent examination of the full text, both reviewers compared their results to produce a final selection. In case of disagreement, the opinion of a third reviewer was requested (CA). eujapa.upol.cz

Data analysis
Reviewers gathered their own data independently once they finished reading the full text of each potentially eligible study. Attention was paid to the general objectives, variables under analysis, instruments, number of participants, age, types of interventions included in the study, number of dropouts, adherence and results obtained after the intervention.
The methodological quality of each study was assessed by two reviewers (MSL and CA). The opinion of a third reviewer (JC) was requested in case of disagreement. For studies which tested the effect of boxing as a rehabilitation strategy, a scoring scale comprised of 10 items with a maximum rating of 15 points was employed (Appendix A). This scale has been applied in similar systematic reviews following previous recommendations (Ayán Pérez & Cancela, 2014). For studies which aimed at identifying the impact of boxing on the physiological parameters of people with diverse health conditions, a 11-item checklist was created (Yes, one point; No, zero points; max. score 11 points) using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (National Heart, Lung and Blood Institute, 2018) as a model. Items 5, 7, 9, 10 and 11 were included in order to adapt the scale to the designs of the included studies. The quality rating of this scale was interpreted as: < 4, poor; 5-6, fair; 7-9, good; 10-11, excellent.

Results
A total of 499 studies were pre-selected. After screening titles and abstracts, 34 full-text papers were reviewed. Only 10 studies were selected. Of them, 2 experimental and 2 quasi-experimental were focused on the effects of boxing as a rehabilitation strategy, and 6 observational in nature, analyzed the impact of boxing on the physiological variables of people with diverse health conditions ( Figure  1). No meta-analyses or prior review work could be found on this topic.

Objective and target population
Out of the four purely experimental studies, two analyzed the effects of interventions based on non-contact boxing training, where opponents had either abdominal obesity (Cheema, Davies, analyzed the feasibility of Wii TM Boxing in pediatric populations (Abdulsatar, Walker, Timmons, & Choong, 2013), whereas the fourth paid attention to the effects of boxing performance from a seated position in adults who had suffered a stroke (Park, Gong, & Yim, 2017).

Characteristics of the interventions
The duration of the programs described in the purely experimental studies varied from 1 to 12 weeks, with between one and four sessions per week lasting from 10 to 90 minutes. One study proposed a daily double session (Abdulsatar et al., 2013) and only one of the four studies included a follow-up phase (Combs et al., 2011). Boxing practice from a seated position was directed towards the improvement of upper body functions, gait and quality of life in general in patients who had suffered a stroke (Park et al., 2017), as well as to evaluate the feasibility and effectiveness of a 12-week boxing exercise program in contrast with walking at a comfortable pace in comparable physical exercise sessions. Combs et al. (2011) also analyzed the effects on balance, mobility and quality of life in patients with Parkinson, while Abdulsatar et al. (2013) focused on assessing the safety and feasibility of using AVGs as a rehabilitation strategy in children staying in a pediatric critical care unit setting.
In the case of those studies of a predominantly observational character, four studies proposed one session (Gaffurini et al., 2013;Howcroft et al., 2012;Hurkmans et al., 2011Hurkmans et al., , 2010, one suggested three interventions preceded by a familiarization meeting with the sample population (Kafri et al., 2014) andMat Rosly et al., (2017) proposed two sessions, but allowed participants to take part in as many as 14 if they so desired. Sessions lasted between 8 and 45 minutes. Five studies analyzed the impact of AVG boxing practice and contrasted it with other AVGs (Gaffurini et al., 2013;Howcroft et al., 2012;Hurkmans et al., 2011Hurkmans et al., , 2010Kafri et al., 2014). Kafri et al.(2014) carried out an additional comparison between sports AVGs and walking at a comfortable pace. Finally, one study compared virtual boxing with real heavy bag training (Mat Rosly et al., 2017). eujapa.upol.cz 3 Gaffurini et al., 2013;Hurkmans et al., 2011Hurkmans et al., , 2010Mat Rosly et al., 2017)

8
The main characteristics of each program are shown in Table 2.

26
Two studies (Abdulsatar et al., 2013;Combs et al., 2011) were categorized as quasi-experimental 27 and the other two (Cheema et al., 2015;Park et al., 2017) as experimental, since a randomization 28 process was performed, without specifying the method. No study monitored participants after the 29 intervention. One study did not provide details about the blinding (Park et al., 2017) and the other 30 three were single blinded (Abdulsatar et al., 2013;Cheema et al., 2015;Combs et al., 2011).

31
Information on the causes of dropout was included in all cases (Abdulsatar et al., 2013;Cheema et al., 32 2015;Combs et al., 2011;Park et al., 2017), but only two referred the reliability of the instruments 33 employed (Combs et al., 2011;Park et al., 2017). No study clarified whether there had been a control 34 of co-interventions in which patients may have taken part. Two of the studies did not report if the 35 groups were homogeneous (Cheema et al., 2015;Park et al., 2017) and the other two did not feature 36 groups to perform such comparison (Abdulsatar et al., 2013;Combs et al., 2011). The methodological 37 quality score varied from 7 (Abdulsatar et al., 2013) to 9 (Combs et al., 2011;Park et al., 2017), with a

39
Observational studies 40 Table 4 shows the results of the methodological quality assessment for the observational 41 studies (Gaffurini et al., 2013;Howcroft et al., 2012;Hurkmans et al., 2011Hurkmans et al., , 2010Kafri et al.,   Fat loss E: interval-based exercises preceded by a 5-min warm-up of continuous skipping at a self-selected intensity. Intervals were prescribed at 2:1 (i.e., 2 min of high intensity activity followed by 1 min of rest (standing or pacing) between intervals and exercises). Three intervals of each of the following five exercises were performed for a total of 30 min of high-intensity effort: (1) heavy bag, (2)  Warm-up of breathing and stretching exercises for major muscle groups in the trunk and extremities. 45-to 60-minute circuit training regimen of function, endurance, and punching activities that alternated between 3-minute training bouts and 1minute rest breaks. The session ended with a 15-to 20-minute cool-down that emphasized core stretching and strengthening and breathing exercises.
Duration (Wk/F/min): 12 and optional continuation/NR/90 Intensity: NR Progression: The activities were advanced by encouraging patients to train as intensely as they could tolerate by striving to complete more repetitions during each training bout. Gaffurini et al. (2013); Observational To increase EE, HR, VO2 and pulmonary ventilation After 10 minutes of rest, the subjects started playing each video game (WiiTM Sports bowling, tennis or boxing) in a randomized order, for 10 minutes with 5 minutes of rest in between for cardiorespiratory recovery. The program started with a warm-up session involving breathing and stretching of the trunk and limbs for 5 minutes, followed by mitt and sandbag hitting for 10 minutes, with a 2minute rest period. Thereafter, stretching of the trunk and limbs was performed for 5 minutes, similarly to the warm-up. C: resistance exercises and gait training. During the 30-minute conventional physical therapy session, neurodevelopmental treatment (NDT) and proprioceptive neuromuscular facilitation (PNF) were performed for 15 minutes each.
Duration (Wk/F/min): 6/3/30 Both E and C Intensity: NR Progression: Before the experiment, patients were trained to wear gloves in the sitting position, and hit mitts and a sandbag in various directions by applying different gestures. The training was conducted in the sitting position in the first and second weeks. In the third and fourth weeks, training was conducted below the hips. During the fifth and sixth weeks, the subjects were asked to hit the target while sitting and standing. EE: Energy Expenditure; F: Frequency; HR: Heart rate; QoL: Quality of Life; MET: Metabolic equivalent; NA: Not Applicable NR: Not reported; VO2: Oxygen consumption; VO2max: Maximum eujapa.upol.cz All but one (Gaffurini et al., 2013) specified and defined clearly the sample populations. In all of the 47 studies, the research question or objective was clearly stated (item nº 1), the participation rate of eligible 48 people was at least 50% (item nº 3). All subjects were selected or recruited from the same or similar 49 populations and during the same time period (item nº 4). The numbers of individuals at each stage of 50 the study were reported (item nº 5), and all the statistical methods used in the study were described (item 51 nº 11). Only one study (Kafri et al., 2014) justified sample size estimation. No study reported on the 52 validity and reliability of its evaluation instruments. Two studies analyzed the variables at baseline 53 (Hurkmans et al., 2011(Hurkmans et al., , 2010. Four (Gaffurini et al., 2013;Howcroft et al., 2012;Kafri et al., 2014;Mat 54 Rosly et al., 2017) analyzed exposure to the intervention on more than one occasion. Only one study 55 carried out a comparison between groups (Kafri et al., 2014). Overall, the methodological quality of the 56 d 9.

58
This systematic review provided information from experimental studies where the effects of adapted 59 boxing therapies on certain variables were investigated (Abdulsatar et al., 2013;Cheema et al., 2015; 60 Combs et al., 2011;Park et al., 2017). In addition, observational studies were found, in which the main 61 purpose was to determine the physiological effects of such therapies (Gaffurini et al., 2013;Howcroft et 62 al., 2012;Hurkmans et al., 2011Hurkmans et al., , 2010Kafri et al., 2014;Mat Rosly et al., 2017), particularly regarding 63 energy expenditure. Sample populations in the experimental studies selected showed great variability, to 64 the point of finding different health conditions in each analysis. Observational studies provided data on 65 three distinct population groups: patients who had suffered a stroke (Hurkmans et al., 2011;Kafri et al., 66 2014), patients with cerebral palsy (Howcroft et al., 2012;Hurkmans et al., 2010) and patients with spinal 67 cord injuries (Gaffurini et al., 2013;Mat Rosly et al., 2017). Despite the heterogeneity in terms of 68 intervention designs and sample populations, the assessment of methodological quality indicated that 69 70 71 Kafri et al., 2014;Mat Rosly et al., 2017).

72
This collection of studies did not report adverse effects on their respective patients and pointed to 73 the existence of beneficial effects derived from boxing practice as an intervention strategy in obese 74 (Cheema et al., 2015), Parkinson disease (Combs et al., 2011) and even post-stroke patients (Park et al.,

79
The results derived from observational studies indicated that virtual boxing practice required EEs 80 between 3.36 and 6.8 METs, which would imply that, either moderate or vigorous exercises can be 81 tolerated by people with diverse health conditions. In this regard, some typical activities of boxing 82 practice, like heavy bag hitting, represented an EE of 4.4 METs; an expenditure comparable to cycling or 83 Pilates (Ainsworth et al., 2011). These results are particularly interesting in the case of health conditions 84 with mobility limitations, such as cerebral palsy, that hinder the possibility to achieve adequate levels of 85 EE in these populations. Children with cerebral palsy usually present slow gait velocity due to restrictions 86 in stride length and cadence (Honkavaara & Rintala, 2010). While therapies like hippotherapy 87 (Honkavaara & Rintala, 2010) or Petra-running (Donnell, Verellen, Van de Vliet, & Vanlandewijck, 88 2010) can improve these parameters; adapted boxing focused on the upper body, could be an alternative 89 and complementing strategy in order to increase EE in this population.

91
eujapa.upol.cz Table 3. Summary description of the aims, outcomes and findings of the included studies.

First author and year
Aim Variables (measurement tools) Findings Adherence; dropouts Abdulsatar et al. (2013) To evaluate the safety and feasibility of virtual reality exercise as a novel acute rehabilitation intervention in a Pediatric Critical Care Unit setting. There were no adverse events attributable to the intervention. Upper limb activity during WiiTM was significantly greater than the average daily activity (p = 0.049). There were no statistical differences in grip strength (p = 0.20) or in the physiological measurements post WiiTM play as compared to baseline. The caregivers 0.4 for their impression of whether the intervention was safe and 5.3 ± 1.8 for potential benefit provided to their child. A strong correlation was found between the total play-time and caregiver perception of child enjoyment (p = 0.02).
NR; 1 Cheema et al. (2015) To assess the feasibility and effectiveness of a 12-week boxing training intervention compared with an equivalent dose of brisk walking in obese adults for fat loss. E: 79 ± 15%; 0 C: 55 ± 43%; 2 Combs et al.
To describe the effects of disease severity and duration of boxing training (short term and long term) on changes in balance, mobility, and quality of life for patients with mild or moderate to severe PD.
-General Balance (FRT) -Changes in functional standing balance over time (BBS) -Balance confidence (ABS -Mobility (TUG) -Walking endurance (6MWT) -Gait speed, cadence, stride length and step width (GaitRite Walkway System) -Activities of daily living and motor examination (subscales from UPDRS and PDQL) Improvements were found on at least 5 of the 12 outcome measures over the baseline at the 12-week test and showed continued improvements at the 24-and 36-week tests. Patients with mild PD typically showed improvements earlier than those with moderate to severe PD. NR; 1 Gaffurini et al. (2013) To provide evidence on the efficacy of a newer intervention for general health promotion in spinal cord injury population.
-VO2, pulmonary ventilation, heart rate and energy expenditure (metabolimeter) All parameters significantly increased from rest to bowling, tennis and boxing. Also, all of them showed a significant positive linear trend (p < 0.0001) while no higher order polynomials were significant. None of these parameters were significantly affected by the sequence of gaming.

First author and year
Aim Variables (measurement tools) Findings Adherence; dropouts Howcroft et al. (2012) To evaluate the potential of active video game (AVG) play for physical activity promotion and rehabilitation therapies in children with cerebral palsy (CP) through a quantitative exploration of energy expenditure, muscle activation, and quality of movement.
-HR (HR monitor) -VO2 and EE (cardiopulmonary testing unit) -Upper limb muscle activations (Electromiography System) -Upper limb kinematics (motion capture system) -Self-reported enjoyment (PACES) -Self-perceived exertion (OMNI perceived exertion scale) Moderate levels of physical activity were achieved during the dance (MET = 3.20 ± 1.04) and boxing (MET = 3.36 ± 1.50) games. Muscle activations did not exceed maximum voluntary exertions and were greatest for the boxing AVG and for the wrist extensor bundle. Angular velocities and accelerations were significantly larger in the dominant arm than in the hemiplegic arm during bilateral play. A high level of enjoyment was reported on the PACES (4.5 ± 0.3 out of 5).
NR; 0 Hurkmans et al. (2011) To determine the energy expenditure of chronic stroke patients while playing WiiTM Sports.
NR; 0 Hurkmans et al. (210) To determine energy expenditure of adults with bilateral spastic cerebral palsy while playing WiiTM Sports tennis and boxing.
-BMI (calibrated electronic scale) -Height (wall mounted metal anthropometer) -BFP (caliper and skinfold thickness) -EE during game play, sitting and standing, -Ventilation and RER (portable, indirect calorimeter) -HR (HR monitor) -Self-perceived exercise intensity (modified 0 to 10 Borg scale) Both WiiTM Sports tennis and boxing increased the VO2 above sitting (340% and 393%, respectively; p = 0.012), and above standing (218% and 256%, respectively; p = 0.012). Energy expenditure during Wii boxing (5.0 METs) was higher compared with energy expenditure during tennis (4.5 METs); however, this difference was not significant (p = 0.069). For all participants, the energy expenditure was greater than 3 METs during Wii tennis (range, 3.0 6.9 METs) and boxing (range, 3.6 6.8 METs) game play. The energy expenditure was greater than 6 METs for 1 participant during tennis and boxing game play and for 1 participant during boxing game play only. Perceived exertion was rated higher for WiiTM Sports boxing than for tennis; however, this difference was not significant (p = 0.076).
NR; 0 Kafri et al. (2014) To characterize the energy expenditure (EE), exercise intensity, and energy metabolism of individuals poststroke, while playing AVGs in different positions, from different consoles, and to compare the performance with comfortable walking and with able-bodied individuals.
-VO2, METs and RER (portable telemetric gas analysis system) -HR (HR monitor) -RPE (Borg Scale 6 to 20) the games (a question adapted from a questionnaire) 1.3 METs and was lower compared with able-bodied participants for XboxTM-Boxing (p = 0.001), WiiTM-Boxing in Standing (p = 0.01), Run (p < 0.001), and Penguin (p = ± 9.3 % predicted maximal HR and was lower compared with able-bodied participants for XboxTM-Boxing (p = 0.007) and Run (p = 0.005). For participants poststroke, EE of walking at a comfortable pace did not differ from boxing games in Standing or Run. For able-bodied participants only, the EE for XboxTM-Boxing was higher than WiiTM Boxing (6.5 ± 2.6 vs 4.4 ± 1.1; p = 0.02). EE was higher in standing versus sitting for poststroke (p = 0.04) and NR; 0 eujapa.upol.cz

First author and year
Aim Variables (measurement tools) Findings Adherence; dropouts able-bodied participants (p = 0.03). There were no significant group differences for RPEs. RER of playing in sitting approached anaerobic metabolism. XboxTM Boxing had the highest score of enjoyment in poststroke participants (4.9 ± 0.1) as well as both XboxTM and WiiTM Boxing AVGs in able-bodied participants (4.2 ± 1.2). Mat Rosly et al. (2017) To compare the physiological responses and user preferences between conventional heavy bag boxing against a novel form of video game boxing, known as exergaming boxing, in individuals with spinal cord injury.
-HR (HR monitor) -VO2, EE, total EE, minute ventilation and MET (indirect calorimeter) -RPE (Borg 0 to 10 scale) --constructed survey) Both exergaming and heavy bag boxing achieved moderate intensities of exercise with 4.3 ± 1.0 METs and 4.4 ± 1.0 METs, respectively, in a sitting position. Paired t-test revealed no significant differences (p < d = 0.02-0.49) in the physiological or perceived exertional responses between the two modalities of boxing. A significant preference (p < 0.05) was found for the exergaming boxing. There is evidence to suggest that boxing may be perceived as a high-risk activity (Hermand et al., 1999). In contrast, none of the selected studies reported adverse effects after the intervention and the number of dropouts was quite low, which provides evidence that reduces the expected risk perception. Another related aspect would be that of the participants´ motivation and appeal of the therapeutic program, which are two fundamental characteristics to favor adherence and predisposition to complete the intervention, and even more so in the case of people with diverse health conditions (Lui & Hui, 2009). In this line, Kafri et al. (2014) measured the level of enjoyment during the sessions and showed that, in both groups, the highest scores were registered in the case of boxing AVGs. Similarly, the caregivers in one of the studies (Abdulsatar et al., 2013) reported high perceptions of enjoyment during AVG-based interventions in the context of a pediatric intensive care unit. It should be noted that Mat Rosly et al.
methodological quality of the selection, these are not exempt from limitations which may prevent direct transference to physical exercise program design in populations with diverse health conditions. Firstly, regarding the included studies, the samples were small both in experimental and observational studies (Gaffurini et al., 2013;Howcroft et al., 2012;Hurkmans et al., 2011Hurkmans et al., , 2010Kafri et al., 2014;Mat Rosly et al., 2017), only two were randomized and just one of them featured a control group (Cheema et al., 2015;Park et al., 2017). Also, the interventions were short in general; therefore, arguments about potential improvements or design adequacy might have been influenced by this characteristic. In addition, the absence of a follow-up phase in experimental studies did not allow for a thorough evaluation of the variables over time. Furthermore, the lack of information regarding the control of co-interventions in which patients might have taken part for the duration of the physical exercise program could have interfered with adapted boxing proposals. Secondly, the methodological quality scales implemented did not assess the adequacy between the designs of the studies and the statistical analysis carried by the authors. Also, in these scales, the weight of the items in the total score or the established cut-off points could be debated. Thirdly, the existence of certain methodological limitations inherent to the review design e.g. language restrictions, gray literature not reviewed or publication bias should be considered as well.

Conclusions
The practice of adapted boxing in people with diverse health conditions may be applied as a feasible rehabilitation strategy with positive effects on cardiovascular fitness, upper body strength, body composition, coordination and parameters such as gait independence and gait speed. AVG virtual programs of adapted boxing stand as an alternative which is motivating and attractive for participants, entary lifestyles and problems derived from obesity and functional independence deterioration. Adapted boxing for rehabilitation purposes may contribute to comply with the international recommendations about physical activity and health (WHO, 2015).

Perspectives
Adapted boxing is a feasible physical therapy and can lead to improvements in different people´s health-related parameters with diverse health conditions. This information may be of interest for health professionals and practitioners responsible for the prescription of physical exercise as part of a rehabilitation process. In addition, the improvements in balance and mobility detected for patients with disease (Combs et al., 2011) could be similarly applicable to people with multiple sclerosis (Giannì, Prosperini, Jonsdottir, & Cattaneo, 2014) or fragile older adults (Ambrose, Paul, & Hausdorff, 2013). Nevertheless, future randomized studies with larger population samples, longer interventions, adequate follow-up periods and better control for confounding factors such as exercise contamination are necessary to determine the adequate therapeutic dosage for each health condition, and the extent of the potentially positive effects of adapted boxing.