Exercise Interventions for Improving Flexibility in People with Multiple Sclerosis: A Systematic Review and Meta-Analysis

Background and objectives: People with multiple sclerosis (MS) often experience limitations in joint range of motion, which is linked to spasticity and continued inactivity. Low flexibility levels in this population have been linked to postural problems and muscular pain. Therefore, the purpose of this study was to conduct a systematic review and a meta-analysis aimed at identifying the characteristics and methodological quality of investigations studying the effects of exercise interventions on the flexibility levels of people with MS. Materials and Methods: Three electronic databases (MEDLINE/PubMed, SPORTDiscus and Scopus) were systematically searched up to May 2019 for intervention studies focused on the effects of exercise on the flexibility levels of people with MS. A meta-analysis, including randomized controlled trials (RCT), which reported information regarding the effects of exercise on flexibility, was also conducted. The methodological quality of included studies was assessed using the Physiotherapy Evidence Database, and the Quality Assessment Tool for Before–After Studies, with no control group. The quality of the information reported, regarding the programs conducted, was assessed by means of the Consensus on Exercise Reporting Template (CERT) scale. Results: Seven studies, four RCTs and three uncontrolled investigations were finally selected. The methodological quality of the RCTs was considered “poor” in one study, and “good” and “excellent” in two studies and one investigation, respectively. The three uncontrolled studies showed a methodological quality between “fair” and “poor”. Following the CERT scale, four studies were graded as “high” and three as “low”. Findings from the meta-analysis indicated no significant effects on hamstring flexibility, or the range of motion in the hips, knees or ankles. Conclusions: There is preliminary evidence from individual studies which indicates that people with MS can improve their lower limb flexibility following participation in physical exercise programs, but the meta-analysis did not confirm these findings.


Introduction
Multiple sclerosis (MS) is a complex condition, characterized by the inflammation of the central nervous system, and it causes axonal or neuronal loss, demyelination and astrocytic gliosis [1]. It is one of the most common neurological disorders worldwide; an estimated 2.3 million people live with MS [2]. In many countries, it is the main cause of non-traumatic neurological disability in young adults [2]. Its incidence is higher among women, and in the Northern Hemisphere [3]. The course of the disability cannot be predicted, and its clinical presentation may vary greatly from patient to patient. There is no known cure for MS [4]. Current management guidelines are focused on reducing its exacerbation, through therapies aimed at reducing symptoms [4]. In the last 25 years, these therapies have shown to have a positive effect on these parameters [5], but the individual with MS still experiences an enormous impact on their level of function, which has a considerable negative effect on overall quality of life.
Physical exercise has been postulated as one of the non-pharmacological strategies of interest, due to its low cost and positive effects on the physical and mental health of the MS population [6]. In general, the majority of research on the effects different exercise types have on this group focuses on training programs, which aim to improve cardiovascular and muscular fitness, in view of the effects that these generally have on two symptoms of great prevalence, namely, fatigue and muscular weakness [7]. Most of the previous studies on this subject have focused on analyzing the effects of aerobic and resistance training, hoping to identify basic guidelines for the prescription of physical exercise for people with MS.
In this regard, it should be mentioned that a training pyramid has been proposed, with the purpose of prescribing physical exercise for persons with MS. The basis of this pyramid is formed of the passive range of motion (ROM) exercises, and then it progresses towards active flexibility exercises [8]. This is because persons with MS often experience limitations in articular movement, which are linked to spasticity and continued inactivity [9]. Reduced levels of muscular flexibility have also been linked to postural problems and muscular pain in this population [10]. Nevertheless, even though flexibility is considered one of the basic components of a healthy physical condition, no systematic reviews contributing to the information on the effects of exercise programs on the flexibility of persons with MS have been published so far, or basic guidelines for prescribing these programs. Therefore, there is a need to conduct a systematic review to provide scientific evidence regarding the type of programs specialists in neurorehabilitation should prescribe to persons with MS, as well as to describe the expected benefits of these programs. Under these circumstances, the objective of this study was to conduct a systematic review and a meta-analysis, aimed at identifying the characteristics and methodological quality of investigations that have studied the effects of exercise intervention on the flexibility levels of people with MS.

Search Process
The search strategy was selected and designed in order to find research studies which gave information on the effects of physical exercise on the flexibility of persons with MS. Three electronic databases were used for the search (Medline/PubMed, SPORTDiscus and Scopus), up to May 2019, using the key words and Boolean operators "Multiple Sclerosis" AND "Range of Motion" OR "Flexibility" OR "Stretching".

Selection Procedure and Eligibility Criteria
Eligibility was assessed by one author and supervised by a second author. Those studies which, having proposed a physical exercise program for persons with MS, included variables relating to flexibility amongst their outcomes, we re considered eligible. Investigations were excluded if: (a) the sample included participants with MS and other conditions, and data for each population was not reported separately; (b) exercise was included as an additional treatment arm, or it was performed as part of a combined therapy program and its effects could not be isolated; (c) the intervention was based on the performance of a single exercise training session; (d) the research was not published in a peer-reviewed journal written in English, French, Portuguese or Spanish.
The studies were examined independently, by reviewing the information which appeared in the titles and abstracts and classifying them as "selected" or "eliminated". The studies which did not provide sufficient information were registered as "doubtful", pending a subsequent reading of the full text.
Once this first selection was complete, the information was contrasted, and a second review was carried out with the assistance of the other co-authors. The studies which were then rejected were those which included: (a) combined exercise interventions which did not include a measure of flexibility; (b) interventions which included pharmacological treatment together with exercise. Finally, seven articles were chosen which fulfilled the requirements and which were related to the object of study of this review.

Data Extraction
The information in each study regarding design type, characteristics of the sample, the exercise program to be carried out, flexibility assessment tools, and effects of the program on flexibility, was extracted by a researcher onto a data log grid. The information was subsequently independently revised by a second researcher.

Assessment of Methodological Quality
The methodological quality of the studies that were considered randomized controlled trials (RCTs) was extracted directly from Physiotherapy Evidence Database (PEDro) [11]. If a study was not registered in this database, the PEDro scale was applied by two authors independently. The suggested cut-off points to categorize studies by quality were as follows: excellent (9-10), good (6)(7)(8), fair (4)(5), and poor (<3) [12].
For the studies with no control groups, the "Quality Assessment Tool for Before-After Studies with No Control Group" [13] tool was applied by two authors independently. This tool includes 12 questions, and authors must define the quality of each study ("poor", "fair" or "good") according to how much risk of bias they consider exists. In either procedure, in cases of disagreement, advice was sought from a third author.
The quality of the information reported with regard to the characteristics of the programs conducted was assessed by means of the Consensus on Exercise Reporting Template (CERT) scale, applied by a single author [14]. The scale contains 16 questions with scores between 1 and 19. A score of <9 is considered "low" methodological quality, and a score of ≥9 is considered "high".

Data Analysis
A meta-analysis, including RCTs which reported information regarding the effects of exercise on flexibility, both before and after the intervention, was performed when the same outcomes had been assessed in at least two studies in a comparable manner [15]. Pre-and post-intervention data was presented for the control and intervention groups as mean ± standard deviation (SD). In order to do this, standardized mean differences (SMD) and their 95% confidence intervals (CIs) were calculated, to assess the change in each selected variable. The SMD is the mean divided by the SD, and its calculation incorporated control and intervention sample sizes, pre-and post-intervention means, and SDs, for each of the selected outcome measures. To obtain the pooled effects, a fixed-effects model was used. In the event of a heterogeneity level (I 2 ) over 30%, a random-effects model was also applied [16]. This same procedure was conducted in order to analyze the pooled effects of the programs in the intervention groups, taking into account only the data from these groups. Forest plots displaying SMD and 95% CIs were used to compare these effects in the pre-and post-intervention measurements in the intervention groups. SMDs were significant when their 95% CIs excluded zero, while pooled SMD values of less than ±0.2, ranging from ±0.2 to ±0.8, and greater than ±0.8 indicated the existence of small, medium, or large effects, respectively. In all the studies, the relative effect size (ES), by means of Cohen's d, was calculated for all flexibility outcomes, by analyzing the intra-group pre-and post-intervention measurements. In addition, the absolute ES, comparing the groups in the study, was also calculated. These calculations incorporated the post-intervention sample sizes, as well as pre-and post-intervention means and SDs for each of the selected outcome measures. Following Cohen's classification [17], ESs were divided into trivial (d ≤ 0.2), small (d > 0.2), moderate (d > 0.5), large (d > 0.8), and very large (d > 1. 3). If the studies intended to be included did not report the required data, the authors were contacted and it was requested. All statistical analyses were performed using Stata 13.

Designs and Samples
Of the 834 references located, 62 were initially selected. Following a reading of the full text, a total of seven studies (four RCTs [18][19][20][21] and three uncontrolled investigations [22][23][24]) were included in the final analysis ( Figure 1). In accordance with the studies, which included information regarding the characteristics of the sample, a total of 163 participants (64.4% women), with an average age of 49.5 ± 5.5 years participated in the proposed interventions. The general characteristics of the studies included are shown in Table 1. ±0.8 indicated the existence of small, medium, or large effects, respectively. In all the studies, the relative effect size (ES), by means of Cohen's d, was calculated for all flexibility outcomes, by analyzing the intra-group pre-and post-intervention measurements. In addition, the absolute ES, comparing the groups in the study, was also calculated. These calculations incorporated the postintervention sample sizes, as well as pre-and post-intervention means and SDs for each of the selected outcome measures. Following Cohen's classification [17], ESs were divided into trivial (d ≤ 0.2), small (d > 0.2), moderate (d > 0.5), large (d > 0.8), and very large (d > 1. 3). If the studies intended to be included did not report the required data, the authors were contacted and it was requested. All statistical analyses were performed using Stata 13.

Designs and Samples
Of the 834 references located, 62 were initially selected. Following a reading of the full text, a total of seven studies (four RCTs [18][19][20][21] and three uncontrolled investigations [22][23][24]) were included in the final analysis ( Figure 1). In accordance with the studies, which included information regarding the characteristics of the sample, a total of 163 participants (64.4% women), with an average age of 49.5 ± 5.5 years participated in the proposed interventions. The general characteristics of the studies included are shown in Table 1.

Quality Assessment
The methodological quality of the RCTs was considered "excellent" [18] and "good" [19,20] in three of the studies analyzed, and "poor" in one [21] ( Table 2).

Quality Assessment
The methodological quality of the RCTs was considered "excellent" [18] and "good" [19,20] in three of the studies analyzed, and "poor" in one [21] ( Table 2).

Interventions
The selected studies carried out different interventions, based on the performance of aerobic exercise [23], strength exercises [21], or a combination of multiple forms of exercise (flexibility, balance and strength [18,22] or aerobic and strength [20]). Other research proposed interventions based on Pilates [19,21] or Tai chi [24]. Only one study included a specific flexibility program [21] (Table 1). The average duration of the training programs was 23.1 ± 7.1 weeks, with an average of 2 ± 0.8 sessions per week, with the exception of one study, which does not give the session times, but the times of the sets carried out; the average duration of a set was 67.5 ± 27.2 min.

Effects of the Programs on Flexibility
Of a total of seven studies, three provided information on the effects of the program on an intergroup and intragroup level [18][19][20]. The remaining four only provided information on an intragroup level [21][22][23][24]. Significant changes in different variables were observed after the completion of the proposed programs in five of the seven studies analyzed ( Table 1). The ES of the different studies is presented in Table 5. The studies from Ponzano et al., (2017) [21], Pereira et al., (2012) [22] and Husted et al., (1999) [24] did not report sufficient data to calculate the ES.

Lower Limb Range of Motion.
A total of three investigations [20,22,23] provided information on the effects of the exercise interventions on the lower limb ROM of the participants. All of these reported significant positive effects after the exercise program, which manifested in a static ROM improvement in the ankle plantar-flexion (right), knee flexion (right-left), hip flexion (right-left) and hip abduction (left) [22], hip flexion with knee extended (moderate to large ES), hip abduction (moderate to large ES), hip adduction (large to very large ES) and hip external rotation (very large ES) [23]. Improvements were also found in the ROM during the gait in the hip flexion-extension (small to moderate ES), knee flexion-extension (small to moderate ES), ankle dorsi-plantar-flexion (small to moderate ES) [20], ankle dorsi-flexion (large to very large ES), ankle plantar-flexion (large to very large ES), ankle angle at contact (large to very large ES), ankle angle at toe-off (moderate to large ES), knee flexo-extension (trivial ES), hip extension (trivial ES), hip flexo-extension (trivial ES), hip adduction (large to very large ES), hip abduction (very large ES), hip angle at contact adduction-abduction (very large ES) variables. When the meta-analysis was performed for the ROM, no significant effects were found comparing the baseline and post-intervention results in the intervention groups ( Figure 2).

Flexibility of the Posterior Kinetic Chain
Four of the articles reviewed measured the posterior kinetic chain [18,19,21,24]. There appear to be no significant differences in two of these [18,19]. However, in the study by Ponzano et al. [21], the posterior kinetic chain was assessed by means of the Sit and Reach test, and significant differences were found in two of the groups (G 1 y G 3 ; small to moderate ES). Significant differences were also found in G 1 using the spinal mouse test, which assesses the morphology of the rachis on the sagittal plane (moderate to large ES). Finally, the study by Husted et al. [24] mentions an improvement of 28% between pre-and post-test results, using the Hamstring Flexibility Test to measure the flexibility of the hamstrings through the spinal column. When the meta-analysis was performed for the Sit and Reach test, no significant effects were found, either when comparing the post-intervention effects in the intervention groups (Figure 3a, n = 39), or when comparing the intervention to the control groups (Figure 3b, n = 78).

Flexibility of the Posterior Kinetic Chain
Four of the articles reviewed measured the posterior kinetic chain [18,19,21,24]. There appear to be no significant differences in two of these [18,19]. However, in the study by Ponzano et al. [21], the posterior kinetic chain was assessed by means of the Sit and Reach test, and significant differences were found in two of the groups (G1 y G3; small to moderate ES). Significant differences were also found in G1 using the spinal mouse test, which assesses the morphology of the rachis on the sagittal plane (moderate to large ES). Finally, the study by Husted et al. [24] mentions an improvement of 28% between pre-and post-test results, using the Hamstring Flexibility Test to measure the flexibility of the hamstrings through the spinal column. When the meta-analysis was performed for the Sit and Reach test, no significant effects were found, either when comparing the post-intervention effects in the intervention groups (Figure 3a, n = 39), or when comparing the intervention to the control groups (Figure 3b, n = 78).

Upper Limb Range of Motion
The upper limb ROM (shoulder, elbow and wrist), differentiating the right side from the left side, was assessed in just one of the articles reviewed [22]. The authors reported that the intervention had significant effects on the extension (right and left) and abduction (left) of the shoulder, on the flexion (right and left) of the elbow and on the extension (right) of the wrist.

Upper Limb Flexibility
Measurements of upper limb flexibility were taken only by means of the Back Scratch test in one of the studies [18], in which no significant differences were found between the pre-test and post-test results, when adjusted by age. However, when adjusted by gender and by pre-intervention values, improvements were found in both groups.

Upper Limb Range of Motion
The upper limb ROM (shoulder, elbow and wrist), differentiating the right side from the left side, was assessed in just one of the articles reviewed [22]. The authors reported that the intervention had significant effects on the extension (right and left) and abduction (left) of the shoulder, on the flexion (right and left) of the elbow and on the extension (right) of the wrist.

Upper Limb Flexibility
Measurements of upper limb flexibility were taken only by means of the Back Scratch test in one of the studies [18], in which no significant differences were found between the pre-test and post-test results, when adjusted by age. However, when adjusted by gender and by pre-intervention values, improvements were found in both groups.

Discussion
In this research, scientific evidence of the effectiveness of physical exercise, for the improvement of aspects relating to flexibility in persons with MS, has been examined. In order to achieve this objective with maximum precision, the search was programmed to find the highest number of studies which focused only on exercise interventions, although it was not limited to the study of RCTs, for various methodological reasons. Firstly, when a limited number of RCTs are found, it is difficult to draw solid conclusions. Therefore, the inclusion of non-RCT studies may be useful to obtain a better view of the current interventions, with a view to informing future research [25]. Secondly, non-RCT studies may provide relevant information on feasibility, adverse effects, or response rate, in said interventions [26]. Finally, non-RCT studies may include important detailed information on the characteristics of the interventions carried out, such as the number and duration of sessions, type of exercises, rests, intensity, volume, etc. Thus, in this review, the spectrum of results is wider, providing greater clarity as to the state of the problem, having extracted the data and conclusions.
An important finding in this research is the fact that only three of the seven studies selected demonstrated good methodological quality. This would suggest that current scientific evidence is limited and, therefore, future RTCs are necessary in this line of investigation. However, valid information can be extracted from all the studies analyzed, since they provided interesting data regarding the characteristics of the interventions performed, which could be useful for rehabilitation professionals who deal with this population.
The results obtained in the studies reviewed suggest that, when the intention is to increase ROM in persons with MS, aerobic and strength exercises appear to be the most effective. On the other hand, when the intention is to increase muscular extensibility, alternative physical therapies, such as Pilates and Tai chi, would appear to be the most effective strategies. There is a remarkable lack of studies which base their research on exercises that seek only to improve flexibility, or which propose a single stretching program to see the resulting effects. This may be because exercise-based rehabilitation therapies focus more on strategies designed to reduce the impact of principal symptoms, such as fatigue (through aerobic training) [27] or muscular weakness [28]. Stretching programs have generally aimed to treat of spasticity [29], and so the effect of this therapy on the ROM or muscular extensibility of persons with MS who are not affected by this symptom remains unknown.
Despite the limited number of studies found, it was possible to conduct a meta-analysis in which the data could be synthesized. Meta-analysis allows us to evaluate whether the size of the effect is consistent and if the effect may be considered strong, and, therefore, the size of the effect may be estimated with greater precision than with a single study. Moreover, if the size of the effect varies, that variation may be described and, potentially, explained [30].
The results of the conducted meta-analysis indicated that there were no significant improvements in any of the variables analyzed. In the case of the Sit and Reach test, although included studies showed a great degree of homogeneity, these results could be influenced by the type of test chosen to assess flexibility. A tendency towards improvement in favor of the intervention group could be observed only when the evolution of the control group is not considered. With regard to ROM assessment of hip, knee and ankle joints, significant effects, in favor of the intervention groups, we re not shown. The only joint in which a tendency towards improvement was observed in was the ankle. It would be worth considering whether the effects of exercise tend to occur sooner in the ankle due to its relative anatomic simplicity, compared to the knee or the hip, which involve greater muscle groups, and where factors such as propioception and balance play a fundamental role [31]. The limited sample size, and the heterogeneity of the studies included, limits the possibility of explaining the results, and, therefore, studies with greater homogeneity in their measurement protocols are necessary, in order for more decisive conclusions to be reached.
Systematic reviews serve to provide more certainty about the scientific evidence presented in a research paper. The results of these reviews form the basis for the development of new research on the same topic. In this regard, the present review brings to light that the majority of studies related to the effects of exercise on the flexibility of persons with MS focused on the lower limbs, perhaps due to these areas being more affected than the upper limbs, in persons with MS [32]. However, mobility in the upper body is fundamental, as three out of four patients with MS face a functional decrease in the upper extremity [33], with a negative impact on daily activities, such as grooming, showering, eating and writing, which reduces quality of life and results in greater dependency [34]. Therefore, future lines of research should analyze the effects of exercise programs on upper body ROM.
As far as the authors are aware, this is the first review study which focuses on the effects that exercise programs have on flexibility in persons with MS. However, despite its originality, this study presents a series of limitations which should be recognized. First, a small number of studies were found, and their methodological quality left room for improvement. Second, the samples were composed of persons affected on a low level (EDSS 1-4), and the majority of the studies were focused on the lower limbs. Third, the low number of RCTs required that non-RCT studies were also included in the review. Fourth, the heterogeneity of exercise interventions and the outcomes measured should also be considered. Therefore, scientific evidence regarding the effect of exercise programs on MS can be rated as weak and could be improved upon. Additionally, there are inherent limitations in the design of this study, mainly due to language restrictions and the fact that grey literature was not sourced, factors which would affect the number of studies ultimately located.

Conclusions
There is preliminary scientific evidence from individual studies, which would indicate that persons affected with MS can improve their lower limb flexibility following participation in physical exercise programs, but the meta-analysis did not confirm these findings. Future, quality studies are necessary to verify the results, and these should take into account the effect that this type of program has on the upper limbs, in order to be able to form solid prescriptions, intended to improve flexibility among persons with MS.