**Integrative Neuromuscular Training in Young Athletes, Injury Prevention, and Performance Optimization: A Systematic Review**

#### **Borja Sañudo 1,\*, Juan Sánchez-Hernández 1, Mario Bernardo-Filho 2, Ellie Abdi 3, Redha Taiar <sup>4</sup> and Javier Núñez <sup>5</sup>**


Received: 6 August 2019; Accepted: 10 September 2019; Published: 12 September 2019

**Abstract:** The aim of this systematic review was to evaluate the current evidence by assessing the effectiveness of integrative neuromuscular training programs in injury prevention and sports performance in young athletes. Different data sources were analyzed up to January 2018. Eligible studies contained information on population (young athletes), intervention (neuromuscular training), comparator (control group or another exercise intervention), outcomes (injury prevention or sport performance), and study design (randomized trials or prospective studies). The trials were restricted based on the language (English) and for publication date (after 1 January 2007). Fourteen randomized controlled trials were included: Seven included dynamic stability-related outcomes. Three assessed the coordination performing fundamental movements and sport-specific skills, while other five studies analyzed muscle strength and two assessed plyometric tests. Agility was evaluated in three studies and speed tests were also considered by four studies. Finally, fatigue resistance in three studies and injury risk in four were assessed. This review provides evidence that integrative neuromuscular training programs can enhance performance and injury prevention in young athletes, taken into account that adherence to the training program is adequate. Collectively, well-designed, randomized studies are necessary to collaborate with the present findings.

**Keywords:** neuromuscular training; strength; injury prevention; young athletes

#### **1. Introduction**

Integrative neuromuscular training (INT) is defined as a training program aimed to enhance physical fitness and prevent the aggregation of the neuromuscular deficits, along with the improvements of the motor competence, especially in youth with a lower level of motor skills [1]. This type of training in sync with the improved sports-related movement skills could also have an impact on injury prevention [2]. Further, recent literature has displayed that the absence of this training's type before or during the adolescence can lead to imbalances and incorrect movement patterns. This fact might be associated with a greater injury risk [3–5], especially in young female athletes [6] and in those sports involving a high number of landings or change of direction maneuvers, such as football or

basketball [7]. Although young athletes present a lower injury ratio in comparison with adults, these injuries entail a longer recovery period. The injuries could even lead to a cessation of the practice due to fear of recurrence [8], with consequent effect on their careers as athletes [9,10].

Recently, Fort-Vanmeerhaeghe et al. [11] developed a classification of the INT components that included two blocks: (a) The first block focused on the development of fundamental movement skills, which includes coordination, strength, plyometric, agility, dynamic stability, speed, and fatigue resistance training, and (b) in the second block sport-specific movement skills were included [11]. It is suggested that all these contents must be exercised in order to improve performance and reduce injury risks. By coordination, within the fundamental movement skills, authors usually refer to locomotor, manipulation, and stability skills (variety of movements and multitasking, including unanticipated reactions). It is believed that these abilities allow control and optimize different sport-specific movements, thus decreasing the injury risks in youth athletes [11]. Additional aspects, such as strength, have been extensively analyzed in the literature. Resistance training might include core and lower limb positions and stability but also the application of upper-body and lower-body exercises and pushing/pulling strength exercises [11]. Strength deficits have been associated with less neuromuscular control and increasing the appearance of injuries in the young population [12].

Moreover, plyometric [13], considered as the development of stretch shortening cycle ability and agility training (development of skills at maximum speed integrating changes of direction actions) [11] also have been suggested in order to improve the performance and decrease injury risks in this population group. Literature also highlighted dynamic stability training (balance training that includes dynamic actions) as a means to manage neuromuscular control because of the improvements in the sensorimotor system, which enhances joint dynamic stability and may also reduce the risk of injury in youth athletes [14]. Furthermore, it seems that speed is the main determinant in the performance of many sports. The inability to produce speed in sprint, optimally during childhood, may decrease the possibility of achieving high competitive levels [15]. This term is closely related with agility and would include the development of skills at maximum speed [11]. On the other hand, fatigue resistance training could be defined as the development of skills under fatigue conditions [11]. It is known that neuromuscular fatigue is an important risk factor in numerous sports injuries [16].

Despite the importance of this type of training, especially in young athletes, the number of studies that include all INT contents in their intervention programs is scarce. The knowledge on its effect of injury prevention and performance has to be determined in this population. Therefore, this systematic review focuses on the analysis of the current evidence assessing the effectiveness of INT programs on injury prevention and sports performance in young athletes.

#### **2. Methods**

#### *2.1. Eligibility Criteria*

For this systematic review, the Preferred Reporting Items for Systematic Reviews, Meta-Analyses (PRISMA) statement, and checklist were used. The inclusion criteria were studies performed in the young population that assessed the efficacy of INT training compared to a control group, with other types of training or with no training. The population considered was both young and athletes. Children up to 11 years old in girls and 13 years old in boys in addition to adolescents between ages of 12 and 18 years old for girls and 14–18 years old for boys, as previously defined by Lloyd et al. [17] were examined. Exclusively, the studies including two or more components of INT (i.e., fundamental and specific movement skills, strength, plyometrics, speed, agility, coordination, dynamic stability, and fatigue resistance) in the training program, which were compared to a control group (i.e., randomized controlled trial or prospective study) were considered. Further, only studies that clearly detailed parameters of exercise containing a description of at least training intensity or volume were included. Finally, only studies published in English after 1 January 2007 were assessed.

#### *2.2. Search Strategy*

A literature search was performed in the following electronic databases: PubMed, Cochrane Central Register of Controlled Trials, Web of Science, CINAHL, MEDLINE, and SPORTDiscus. The last search was conducted in January 2018. The search strategies varied according to the different databases and used the following systematic search terms: Neuromuscular control or neuromuscular training or integrative neuromuscular training and strength training, or plyometric training or speed training or agility training or fundamental movement skills or specific movements skills or coordination training or dynamic stability or fatigue resistance and youth athletes or young or adolescents and injury prevention or sport performance. Language was limited to English and participants were all human.

#### *2.3. Data Collection Process and Quality Analysis*

Two authors (J.S. and B.S.) independently screened titles and abstracts to choose potential eligible studies. Full-text articles were obtained and independently evaluated for inclusion in the present review based on the inclusion criteria. Disagreements were resolved by consensus and, if necessary, the study's authors were contacted for clarification guidance. The flow chart can be observed in Figure 1. Each study that met the inclusion criteria was abstracted for information regarding: Number of trained and untrained athletes, age and sex, sport, main and secondary outcomes, main results, INT contents included, frequency, duration, and intensity of the training program.

**Figure 1.** Flow diagram of the search process of integrative neuromuscular training effects in injury prevention and performance on youth athletes.

The methodological quality was individually assessed by using an adaptation of the Cochrane methods [18], which included the generation of randomization sequence, the allocation concealment, blinding of participants or assessors and description of withdrawals and dropouts (see Table 1).


**Table 1.** Quality assessment of studies.

#### **3. Results**

The literature search identified 471 potentially relevant studies (Figure 1). Following the screening of the titles, 305 studies were excluded in addition to 97 duplicates. Concerning the eligibility criteria, the remaining 69 studies were analyzed. Ultimately, 55 studies were excluded, 27 were due to the mean age which was higher than 18 years old, 13 included only one component of INT in their training protocols, seven were performed before 2007 and eight studies did not include a control group. This resulted in 14 included studies to be analyzed in the present review.

#### *3.1. Characteristics of the Included Studies*

Table 2 shows that most studies were executed exclusively on young female population [20,25–28, 30–32], four studies analyzed both male and female athletes [19,22,23,29], while two studies [21,24] included only male athletes.

Seven studies analyzed soccer players [22,25–27,30–32] and other studies were performed in handball [19], tennis [24] and basketball players [28]. Moreover, we found one study in players of different sports (i.e., field hockey, volleyball, soccer, and basketball) [20], and three remaining studies were developed in unspecified sports [21,23,29]. Regarding the age of the athletes and the peak of height velocity (PHV), which normally occurs around the age 12 in females and 14 in males [33], in twelve studies [19–22,24–28,30–32] mean age was post-PHV (>14 years old), while only in two studies [23,29] the mean age of the athletes was pre-PHV (<14 years old).


**Table2.**Characteristicsoftheincludedstudies.

#### *Appl. Sci.* **2019**, *9*, 3839


**Table 2.** *Cont.*

**Table 2.** *Cont.* NMT: Neuromuscular training group; PLYO: Plyometric group; CORE: Core group; CTG: Comprehensive group; RG: Regular group; CG: Control group; DJ: Drop jump; SEBT:excursion balance test; CMJ: Countermovement jump test; RSA: Repeated sprint ability; BESS: Balance error scoring system; IRR: Injury rate ratio.

players (IRR = 0.28, 95% CI 0.10 to 0.79).

 Star

#### *3.2. Outcomes Measures*

Dynamic stability related outcomes were assessed in seven studies [19,21,23,25,27,28,32]. As defined by Fort-Vanmeerhaeghe et al. [11], the term "dynamic stability" was considered as the training of lower limb dynamic stabilization (with three categories of progression: Static balance, dynamic balance, and dynamic stabilization) and core dynamic stability. Further studies assessed coordination performing fundamental movements [19,20,26] and sport-specific skills [31], while other studies analyzed muscle strength [21,23,24,27,31] and plyometric [27,32] tests. Agility [21,23,27] and speed [21,24,27,31] tests were also considered by different studies. Finally, fatigue resistance [23,24,29] and injury risk [22,29,30,32] were also examined.

Among the selected studies, five [19,21,25,28,32] found significant improvements in dynamic stability. Four of them reported significant increments in Star excursion balance test (SEBT) in the intervention group compared with control participants [21,25,28,32]. Furthermore, positive changes in single leg stability (one-leg hop test) were found [19]. Nevertheless, two studies [23,27] did not report changes between-groups in this outcome.

Significant changes compared to the control group were also observed in coordination performing fundamental movement skills (i.e., jumping and landing technique) in two studies [19,20]. Barendrecth et al. [19] showed that the in-season training group had the greatest benefits on knee kinematics and single leg stability. Although, Brown et al. [20] showed a greater peak knee flexion of landings in the neuromuscular training group in both bilateral (*p* = 0.027) and unilateral landings (*p* = 0.076). By contrast, non-significant between-group differences were observed by Klugman et al. [26]. With respect to the sport-specific skills, the only study that assessed this outcome [31] observed no differences between group performances of soccer players after a period of neuromuscular training.

Among the five studies that assessed muscle strength, three reported significant improvements [21, 23,24]. Fernandez et al. [24] showed a 2% increment in countermovement jump (CMJ) in the experimental group. Moreover, Chaouachi et al. [21] found improvements in 1RM leg press test, CMJ and maximal hopping test in the neuromuscular training group compared with the control group. This agrees with Faigenbaum et al. [23] who reported significant changes in the training group (girls in physical education classes) in the curl up, long jump and single leg hop tests. Notwithstanding, two studies [27,31] did not find any improvement in strength after the training protocol. Neither of the two studies which analyzed plyometric performance reported positive changes between training and control groups [27,32].

Once the agility tests were analyzed, again contradictory results were found. Uniquely, the study performed by Chaouachi et al. [21] reported increments in shuttle run performance (*d* = 0.52). However, non-significant improvements were seen on agility tests in two additional studies [23,27]. This controversy persists in the speed tests results, while two studies found significant differences in 10-m sprint [21,24] and one in 30-m sprint [21], the remaining two studies did not show changes in this outcome [27,31].

With respect to the outcome of fatigue resistance, one study [23] conducted in girls, found positive changes. Moreover, a different study [24] observed significant changes in repeated sprint ability (RSA), with performance increments of ≈1.5%, but did not find changes in the 30-15 intermittent fitness test. Contrarily, Richmond et al. [29] showed that the experimental group, submitted to strength and dynamic stability training, had significant improvements in oxygen uptake (VO2max) in the Leger 20-meter shuttle run test.

With respect to the injury risk, three studies [22,29,32] found significant improvements in the training group compared to the control group. In their study, Emery et al. [22] reported lower injury rates in the training group (2.08 injuries/1000 h) when compared to the control group (3.35 injuries/1000 h). Moreover, Richmond et al. [29] presented an injury rate of 7.1 per 100 students in the experimental group and 14.5 per 100 students in the control group. In a study, Steffen et al. [32] included an experimental group that performed a protocol combining strength, plyometrics, dynamic stability, and

agility during four months with two, three sessions per week. It was reported that the injury risk is significantly lower in the high adherence group in lower extremities (IRR = 0.28, 95% CI 0.10 to 0.79) when compared with the control group. Uniquely, one of the analyzed studies [30] described no significant changes between groups in injury risk after a period of strength, plyometric, and dynamic stability training (training group = 3.6 injuries/1000 h, control group = 3.7/1000 h).

#### *3.3. Characteristics of the Intervention Programs*

The strength and dynamic stability showed in Table 3 were the contents of INT and were included in the analyzed studies [19–23,25–32]. Plyometric training was used in ten studies [19–22,24,26,28,30–32]. Three studies [19,20,27] included coordination performance fundamental movement skills training in their intervention programs, while four [22,24,28,32] used agility training and two-speed training [20,24].

The length of the exercise programs varied in the included studies from six weeks to eight months. Study duration was six weeks in two studies [20,28], eight weeks in four [21,23–25], and 10 weeks in other three studies [19,26,31]. The remaining studies had a duration of: 11 weeks [27], 12 weeks [29], 16 weeks [32], 20 weeks [22], and eight months [30].

The frequency of the sessions for most studies was two times per week [19,23–25,27,28]. The exceptions were four studies in which participants had three training sessions per week [20,21,26,31] and two studies that had a variable duration between two and three sessions per week [29,32]. Conjointly, in one study [30], the intervention group was trained 15 consecutive sessions and thereafter one session per week. Information regarding the number of training sessions per week was not provided in one study [22].

The duration of the sessions was between 15 and 20 min in most studies [19,22,23,27,29–32], throughout the time, in other three studies, the sessions lasted 60 min or more (20,25,28). Three studies [21,24,26] did not provide any information related to the duration of the sessions. In these studies, participants performed five to six exercises of strength, plyometric and dynamic stability training [21,26], or plyometric, agility and speed training after eight minutes of dynamic warm-up [24].


**Table**




NMT: Training group; CG; control group; PLYO: Plyometric training group; Bal-PLYO: Balance and plyometric training group; CORE: Core training group; CTG: ComprehensiveRG: Regular training group; INT: Integrative neuromuscular training; PE: Physical education; wk: weeks.

 group;

#### **4. Discussion**

To the best of our knowledge, this is the first systematic review that analyzes the effects of INT in young population considering the contents used in the intervention programs. The primary purpose of this research was to determine which contents of INT provide better results in terms of injury prevention and performance in young athletes. However, there was a desire to advance one step further in the description of the different parameters that determine the application of the training type (i.e., which frequency, intensity, and volume can optimize the benefits of this type of training).

#### *4.1. Characteristics of the Included Studies*

Most of the studies in this review included only female athletes [20,25–28,30–32], in addition, there is a lack of studies assessing exclusively male athletes [21,24]. This fact makes it difficult to extrapolate the results to other populations. Furthermore, there is a clear tendency to analyze only soccer players [22,25–27,30–32], suggesting the need to extend the results in other sports. Once the age of the participants was analyzed, we found again that most of the studies focused on post-PHV athletes [19–22,24–28,30–32] and only two studies [23,29] analyzed the effects of INT on pre-PHV athletes. This is an important factor, since we cannot generalize the results obtained from athletes at different maturation stages [34]. In their study, Granacher et al. [34] suggested training different contents depending on the athletes' maturation stage. The study considered that focus on coordination, agility, dynamic stability, and strength is more significant in early stages, which aim to improve the movement techniques at a controlled speed. Whilst in a more advanced stage, youths should be trained more frequently on plyometrics, core, and strength exercises with a greater sport specificity and execution speed. This is a tendency generally followed by the reviewed articles. We observed that in the particular studies that trained pre-PHV athletes only strength and dynamic stability training were utilized [23,29], while the studies performed on post-PHV athletes included alternative contents, such as plyometric training [19–22,24,26,28,30–32].

#### *4.2. Outcomes Measured*

Concerning the effects of INT in youth injury prevention and performance, it was suggested that dynamic stability can be improved [19,21,25,28,32] with an INT program. Notwithstanding, most studies reported improvements in this outcome. While two studies [23,27] did not find changes, possibly by the cause of the training, it did not have the magnitude or stimulus required to enhance this outcome [23,27], also possibly due to a low player attendance in the training sessions [27].

When the coordination of performing fundamental movement skills was assessed, we found some contradictory results. While significant changes after an INT protocol were reported in two studies [19,20], one study [26] observed non-significant between-group changes. Again, an insufficient application of exercise could give an explanation to these results as the authors themselves indicated. There might be a dose response relationship for improving the fundamental movement skills with INT [26]. By taking a closer look at this study it was found that the training protocol consisted of only six exercises focused on hamstring strength, plyometrics, and dynamic stability. Nevertheless, there were no specific coordination exercises including fundamental movement skills. Furthermore, the same tendency can be observed with the sport-specific skills. One study [31] did not report between-group differences. Subsequently, an INT program of 15 min per session composed of 10 exercises focusing on strength, plyometric, and dynamic stability training was accomplished. It also concluded that the training volume and intensity for each of the exercises were too low to result in performance improvements. It is necessary that at least 20 min of INT replace the ordinary warm-up exercises used by the team [31].

As previously reported, there is also a controversy in strength and plyometrics after a neuromuscular training protocol. Further, three of five studies found positive changes in strength [21,23,24], while the remaining of two [27,31] did not report any differences between groups. One possible explanation to this lack of effect could be attributable, at least in one of the studies [27], to a low training adherence (59.6 ± 14.3%). It is assumed that a higher adherence to the training program would be necessary to improve this outcome [27,31]. Along the same line, non-significant changes between groups were found in plyometric tests after a period of INT in two studies [27,32]. This was probably due to a combination of the lack of stimulus and low adherence to the training program [27,32].

Moreover, contradictory results were found once more with respect to agility and speed tests. Most of the studies did not report significant changes between-groups in agility [23,27] or speed [27,29] performance. As it was described above, it appears that the adherence to the training programs and training stimulus should have been greater in order to enhance agility or speed performance [23,27,29]. Notwithstanding, another possible explanation of the discrepancies in this outcome is the number of contents performed. It has been suggested that performing a 15 min INT program, two times per week, training on only strength and dynamic stability contents [23,27,29], are not enough to improve agility and speed performance.

Fatigue resistance was the only INT content that improved in the three studies, which measured this capacity [23,24,29]. Therefore, it can be suggested that an INT program can improve this outcome in young athletes. This can happen after a period of training of 8–12 weeks with two/three sessions per week focused on around 15 min training of dynamic stability and strength contents [23,29] or plyometrics, agility, and speed contents [24].

Finally, regarding the injury surveillance, most of the analyzed studies observed a decrease in the rate of injuries [22,29,30,32]. Exclusively, one study [30] did not report significant results, probably considering the low compliance of the training program. The intervention teams included the INT program (composed of 20 min sessions where 10 exercises were performed focusing on strength, dynamic stability, and plyometrics) in only 60% of their training sessions during the first half of the season [30]. Due to these findings, the injury risk in athletes is likely to be reduced after an INT program with adequate adherence.

#### *4.3. Characteristics of the Intervention Programs*

In the literature, strength training is considered as one of the best paradigms to enhance physical performance in youth athletes [35,36]. Together with dynamic stability training, the INT content was used more in the training programs analyzed in the current study [19–23,25–32]. Furthermore, strength deficits have been associated with less neuromuscular control, which increases the occurrence of injuries mainly in the lower limbs [12]. In this systematic review, those studies which included strength training in their INT programs reduced the rates of injury [22,29,32]. However, these changes were also accompanied by improvements in dynamic stability [19,21,25,28,32], functional movement skills [19,20], strength [21,23], agility [21], speed [21], or fatigue resistance [23,29]. Nonetheless, four [26,27,30,31] of the 13 articles [19–23,25–32] that used strength training did not find positive changes in any of the outcomes measured. This generates a controversy due to the program characteristics in these articles, which were very similar to those studies that did find significant improvements.

Numerous explanations might be suggested for these discrepancies: (a) Application of exercise, (b) age of participants, and (c) contents included in the program. It could be speculated that it is the combination of training contents, rather than the sum of them individually, that confers the benefits on these young people. In fact, the four studies that did not find positive changes employed only three contents: Strength, plyometrics, and dynamic stability in three studies [26,30,31] and strength, dynamic stability and coordination in another one [27]. The age should be another key factor when programing this content. While pre-PHV athletes will have a neuromuscular type of adaptation, exercises must focus on techniques, postural control, and circum-PHV. Post-PHV athletes will not have just neuromuscular but also structural adaptations. The proposed task should additionally focus on improving technique in more advanced and sport-specific exercises and increasing execution speed [11]. Among the aforementioned studies with limited results, all were post-PHV (>14 years old) and one [31] presented the highest mean age of all analyzed studies. This suggests the need for an adjustment of adequate training loads and progression in each content.

Dynamic stability was the other main content used in the intervention programs [19–23,25–32]. Evidence supports the need to train this content due to improvements in the sensorimotor system, which can improve neuromuscular control. This leads to a better joint dynamic stability and, therefore, a decrement in the injury risk [37]. In fact, the youth athletes' inability to maintain postural balance in static and dynamic actions has been associated with an increased likelihood of injury [38]. In the reviewed articles, those which included dynamic stability training in their INT programs also included strength training [19–23,25–32]. Thus, similar improvements were found for both contents, which allowed us to suggest that dynamic stability along with strength training might be a good strategy to improve performance and prevent injuries in young athletes. The literature shows that dynamic stability should be trained through exercises focused on the lower limbs and core [39], which challenges the feed-forward mechanism. This is described as anticipated actions that occur before the sensorimotor system detects changes in the environment [40]. This mechanism is the most important factor of training in order to maintain balance in landing, deceleration and cut off maneuvers, which helps to decrease injury risk [41]. Nonetheless, the lack of these types of exercises can be clearly detected in the training programs described in the reviewed articles [19–32].

Plyometric training was the other content of INT widely included in the intervention programs [19–22,24,26,28,30–32], which has been shown to improve performance and to decrease injury risk in young athletes [13]. In the current review, studies which included plyometric training in their INT programs improved dynamic stability [19,21,28,32], functional movement skills [19,20], strength [21,24], agility [21], speed [21,24], fatigue resistance [24] tests, and reduced injury risk [22,32]. Only three [26,30,31] of the 10 studies [19–22,24,26,28,30–32] that used plyometric training did not report significant changes between-groups. One of the particularities of this INT content is the need for an adequate progression. This progression should start with low-intensity exercises performed at a slow velocity with a proper technique and develop at later stages to higher velocity and intensity drills [13].

In various studies [19,20,27], coordination of fundamental movement skills' training was included in their intervention programs. From these studies, we found two reported significant improvements in fundamental movement skills [19,20] and dynamic stability tests [19]. One study [27] could not find significant changes probably due to the low compliance with the training program (mean player attendance at the training sessions was 59.6 ± 14.3%). These skills are particularly important in pre PHV athletes during sport-specific skills development in circum-PHV and post-PHV athletes [42]. Coordination should be trained at early stages due to the greater neural plasticity that children have at this age [43]. Moreover, children are encouraged to be involved in a variety of sports at this age in order to develop neural adaptations for many skills before sport specialization [11]. Likewise, a proper movement competency not only enhances physical ability but, in addition, is suggested to decrease injury risk [44]. However, none of the studies that included coordination in the program subsequently analyzed the injury rates. Furthermore, it should also be highlighted that the protocol of sport-specific skills was not clearly defined in the analyzed studies. Throughout the same line, four articles [22,24,28,32] that included agility training in their protocols were found. Literature has defined agility as the combination of decision-making process and change of direction [45]. Decision-making will improve overall during adolescence based on the experience acquired by the athletes. Nevertheless, the change of direction technique can be enhanced in pre-PHV [15], probably as the result of the great neural plasticity of these youth athletes [43]. From the reviewed articles that used agility training in their protocols, we found improvements in dynamic stability [28,32], strength [24], speed [24], fatigue resistance [24] tests, and injury prevention [22,32]. Hence, it would be interesting to include these outcomes in the youth training programs to enhance performance and to reduce injury risk.

Regarding speed training, evidence suggests developing this INT content across the different stages of growth [15]. From the reviewed literature, we found two studies [20,24] that included speed training in their programs. These studies reported significant improvements in strength [24], speed [24], fatigue resistance [24], and fundamental movement skills [20]. Athletes should start training with the high-velocity capacity and the proper running biomechanics, with plyometric, coordination, and sprint technique exercises. Prior to puberty and before maturation, these parameters are easier to change [46]. However, post-PHV athletes may improve more with a combination of neural basis training mainly focused on strength training [15]. Contrarily, it is remarkable that none of the reviewed studies included detailed fatigue resistance training in their protocols. Since we know that neuromuscular fatigue is an important risk factor in numerous sports injuries [16], central fatigue can decrease the ability to perform complex tasks in sports, which will reduce performance and increase injury risk [47]. Again, at early stages, this content should be focused on the correct technical execution of basic movements (pre PHV) to gradually progress towards small-sided games with higher intensity. Once an athlete has enough competencies, a combination of small-sided games and high-intensity interval training should be included in the training program [11,48].

If the parameters of the exercise are analyzed, the length of the exercise programs varied between six weeks to eight months, with most programs lasting eight weeks [21,23–25]. As it was reported in previous literature, at least four weeks are necessary to generate adaptations [15]. Nonetheless, the goal of INT will be to achieve long-term athlete development, with a multi-year training approach following all stages of growth in youth athletes [15]. Moreover, training frequency varies between one and three sessions per week, with two days per week at the most used frequency [23–25,27,28,31]. According to this finding, manifestly practicing high-intensity INT with two/three days per week on non-consecutive days may be sufficient to produce adaptations in youth athletes [49]. Moreover, the duration of the sessions (which was between 15 min and 60 min or more) was 15 min on average in most of the selected studies [22,23,27,29,31]. With respect to this outcome, the literature suggests sessions of 60 min [50] or between 30 and 90 min for young athletes [51]. However, for physical education classes, it is also scientifically proven that 15 min sessions are effective in reducing injury risk [51]. Further analysis of this parameter in future studies is warranted.

It is interesting to note that while most studies disclosed significant improvements with these characteristics [19–25,28,29,32], some others [26,27,30,31] did not find improvements with similar protocols in terms of length of the program, frequency, duration, or INT contents trained. Therefore, it was hypothesized that the main reason for this controversy is the grade of adherence to the training program. Due to the fact that those studies which could not find significant improvements in any outcome measure, reported low adherence to the protocols as the main cause of their results [26,27,30,31]. Consequently, this makes the training stimulus lower than previously planned.

#### *4.4. Limitations and Strengths of This Review*

The main limitation of our systematic review was the lack of a detailed description of the training protocols employed by the included studies. However, in order to overcome this limitation, the authors were contacted to collect the missing data (although the lack of some data persisted in some cases). Moreover, the methodological quality of the included studies is moderate as a result of notable bias in at least six of 14 studies reported. Further, the heterogeneity of the studies (e.g., sample size) limits the discussion of the results. Moreover, only studies published in English language were considered in the current review.

The main strength of this study is the description of the most used INT contents in the literature and their effects on injury prevention and sports performance in the youth population. Additionally, this review analyzed the program characteristics, reporting the most common length, frequency, and duration used by the studies, which accomplished our inclusion criteria.

#### **5. Conclusions**

In conclusion, this review presents that INT programs can enhance performance and injury prevention in young athletes, as long as the adherence to the training program is adequate. The

most INT contents used in the reviewed studies were strength, dynamic stability, and plyometrics. These contents were followed by coordination of fundamental movement skills, speed, and agility training. However, none of the studies included all the discussed contents in the same protocol. Finally, lack of protocols that included fatigue resistance and specific movement skills training was reported. This is contrary to the recommendations of the literature highlights, with respect to the benefits of discussed INT contents in young athletes.

Future studies are recommended to analyze the effects of a protocol that include all INT contents in young athletes at different growth stages. Researchers need to pay specific attention to pre-PHV athletes and in boys, where a lack of evidence in the literature was found. Needless to say, new lines of research should include INT programs with more drills that challenge the feed-forward mechanisms. This has been described in the literature as one of the most important factors in order to decrease injury risk.

**Author Contributions:** Conceptualization, B.S. and J.S.-H.; methodology, M.B.-F.; formal analysis, R.T.; investigation, B.S.; writing—original draft preparation, B.S. and J.N.; writing—review & editing, B.S. and J.N.; supervision, M.B.; review & editing, E.A.

**Funding:** This research received no external funding.

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

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **3D Device for Forces in Swimming Starts and Turns**

**Karla de Jesus 1,2,3,4,\*, Luis Mourão 1,2,5, Hélio Roesler 6, Nuno Viriato 2,7, Kelly de Jesus 1,2,3,4, Mário Vaz 2,7, Ricardo Fernandes 1,2 and João Paulo Vilas-Boas 1,2**


Received: 29 June 2019; Accepted: 16 August 2019; Published: 30 August 2019

**Abstract:** Biomechanical tools capable of detecting external forces in swimming starts and turns have been developed since 1970. This study described the development and validation of a three-dimensional (six-degrees of freedom) instrumented block for swimming starts and turns. Seven force plates, a starting block, an underwater structure, one pair of handgrips and feet supports for starts were firstly designed, numerically simulated, manufactured and validated according to the Fédération Internationale de Natation rules. Static and dynamic force plate simulations revealed deformations between 290 to 376 με and 279 to 545 με in the anterior-posterior and vertical axis and 182 to 328.6 Hz resonance frequencies. Force plates were instrumented with 24 strain gauges each connected to full Wheatstone bridge circuits. Static and dynamic calibration revealed linearity (*R*<sup>2</sup> between 0.97 and 0.99) and non-meaningful cross-talk between orthogonal (1%) axes. Laboratory and ecological validation revealed the similarity between force curve profiles. The need for discriminating each upper and lower limb force responses has implied a final nine-force plates solution with seven above and two underwater platforms. The instrumented block has given an unprecedented contribution to accurate external force measurements in swimming starts and turns.

**Keywords:** sports engineering; biomechanics; ground reaction forces; swimming; performance

#### **1. Introduction**

The analysis of swimmers' performance has traditionally examined spatiotemporal variables representing the start, turn and clean swimming distance [1]. The start and turn are fundamentally different skills than free swimming, which does not necessarily indicate a similar level of start or turning performance [1]. Generally, at the elite level, it is not only swimming speed that wins the races but rather the start and turn where most expert swimmers are travelling at their fastest velocity [2,3]. For instance, the start phase of the 50 m men's freestyle at the 2019 FINA Champions Swim Series at

Budapest has showed that 15 m after the start take-off phase, the second-placed swimmer was 0.08 s slower and the final race time difference was 0.15 s. Moreover, over a 200 m event, the turn contributes 21% to total race performance and progressively more as race distance increases [4].

External forces generated during the ventral starts using instrumented starting blocks have been measured since 1971 with a uniaxial force platform placed at the edge of the pool, which had assessed horizontal forces applied by the swimmers' feet during the conventional circular arm swing technique [5]. Cavanagh et al. [6] had optimised Elliot and Sinclair's solution including eight strain gauges in a horizontal bar to measure the horizontal and vertical forces applied by the hands and feet during the grab ventral start technique. Three-dimensional forces were assessed in starts since 2003 by Naemi et al. [7] revealing that the grab start performed with hands in between feet was less stable in preventing platform twist. Horizontal forces applied by feet and hands in the backstroke start were firstly measured in 2011 (de Jesus) using a uniaxial platform and a load cell, being fixed on the starting wall and on the handgrip, respectively.

The changes in the starting block design in 2008 implied adjustments in the previously developed instrumented platforms for start (e.g., [2]) force analyses. Mason and co-authors [2] have presented an instrumented block comprised of four triaxial force sensors placed in a main force plate fixed over the starting block, inclined rear plate, handles for ventral and backstroke start and underwater platform with holes over the front surface. Recently, researchers have shown a new device to measure independently three-axial forces exerted on hands and feet in the ventral kick start technique [8], but limitations have been identified on hand placement dependence. Despite the relevant shortcomings for start analysis that the previous systems have provided, none of them are able to narrow in a unique solution for all ventral, backstroke and relay start force analysis possibilities, assuming laterality effects and independency in hands and feet force assessment. The horizontal and vertical force components determine steering start strategies, being the most assessed. However, lateral responses are essentially a controlling movement in starts [7,9,10] and it can be better understood when assessing each swimmer's limb force contribution.

The description of forces generated during swimming turns also started in the 1970s using uniaxial force plates to assess the horizontal component in tumble and open turns (e.g., [11]). Swimming turns external forces analysis using three-dimensional force plates is still scarce, with the flip technique and its variants being the most commonly investigated (e.g., [12,13]). A double underwater tri-axial force plate solution developed for independent assessment of forces applied by each foot in a backstroke start can also provide a direct measurement of the combined forces exerted during turns and the determination of foot position. Furthermore, in some turn techniques, coaches can assess laterality data if the foot location is compatible with each force platform (e.g., [13]). Mason and co-authors' underwater force platform design has a multitude of holes system to reduce wave effects [2]. However, this configuration limits the possibility to use the same force plate for water wave effect analyses [14].

Coaching and commercial ongoing-instrumented starting blocks are still lacking some final integrated solution for start and turn force analyses that could inspire new biomechanical research directions. In a competitive, rapid uptake market such as sports equipment, it is important to keep searching for new and improved designs and materials at affordable prices [15]. Commercial force plate prices are ~\$20,000, which can usually be very expensive for coaches and biomechanists [16]. Beside the high prices, technical assistance issues and spare parts' availability are sources of many difficulties for labs and researchers [17]. Thus, due to the high cost of existing force plate systems, the development of simpler low cost, adjustable and accurate models for biomechanical analysis is desirable for force measurements in swimming start and turn techniques (cf. [18]).

A dynamometric unit composed by independent 3D force plates, a starting block, an underwater structure, ventral and dorsal start handgrips and feet supports is original. It presents crucial potentialities, particularly regarding forces and momenta measurement in individual ventral and backstroke starts, relays and turning techniques. As a versatile and low cost system, the force plates can be used uncoupled from the dynamometric unit to measure active (e.g., [19]) and passive swimmers' drag (e.g., [14]), and underwater gait ground reaction forces (e.g., [20]). The current study aimed to design, construct and validate an instrumented swimming start block, emphasizing geometry description, numerical simulation, sensors bonding, calibration, experimental and ecological validation procedures.

#### **2. Materials and Methods**

#### *2.1. 3D Geometric Computer Aided Design*

Force plates, the starting block, underwater structure, handgrips and feet supports were 3D designed using a solid modelling computer aided design software (SolidWorks 2012, Dassault Systèmes, SOLIDWORKS Corporation, Waltham, MA, USA). Each force plate and handgrip was framed to achieve proper sensibility with a high rigidity and reduced mass. The start block project prioritized low deformation to support seven to nine force plates, handgrips and an underwater structure. In addition, force plates, the start block, underwater structure, handgrips and feet support dimensions (Figure 1) were conditioned to comply with the Fédération Internationale de Natation rules (FINA; FR 2.7 and FR 2.10).

**Figure 1.** The seven force plates solution distributed in the start block for individual limb force-moment assessments: Each force plate (FP 1 to 7) and feet supports (back plate and wedge) are also shown.

#### *2.2. Force Plates Spatial Layout*

Contrarily to arbitrarily multiplying force platforms on the block, it is worth it to consider a minimum number of them. Such a number considers the limbs involved in the starting technique force development and the set of limb positioning on the starting block/wall. The rationale relies on the following: If a sole limb force is measured then the centre of pressure ( <sup>→</sup> *COP*) lies on the limb contact area. On the other hand, if a sole platform measures simultaneously two-limb forces then the centre of pressure lies between the limbs and perhaps outside each limb contact area. Such measurement can mask postural constraints necessary for maximum performance data interpretation. The five above water force plates configuration was firstly developed for independent swimmers' upper and lower limbs kinetic analysis in individual ventral start techniques (e.g., grab and track start).

Subsets of analogous platforms have been preferable, aiming for the comparison of force plates behaviour, which implied that two pairs of platforms were similar (300 mm × 250 mm above water and 600 mm × 300 mm underwater). This resemblance provides an interchangeable ability, which might be of great importance if mass production or maintenance were considered. The fifth above water force plate is 500 mm × 440 mm, enabling support for the rear foot as used in the track start technique.

#### *2.3. Force Plates Geometry*

The main requirements of the load sensor followed Lywood et al.'s geometry [21], which consisted of instrumented rectangular section bars oriented to be most sensitive to *x*, *y* and *z* force and torque components (cf. [22]), obviating most cross-talk. A waterproof sensor choice enables force plates immersion into the bottom of the swimming pool (~2 m). Sensor locus, distant from the centre anchorage points, allows a better interpolation approach. As the Roesler [20] platform design complied with these previous requirements, it was selected to serve as the testing tool.

It has already been mentioned that Roesler's [20] force plate topology allows a more accurate → *COP* determination, as well as direct and independent 3D forces and moment measurements without interference among them [18]. Each force plate core was designed to be manufactured in galvanized steel and is essentially composed by two vertical and two horizontal beams plus two lateral boxes. The beams, contrarily to the ring or pylon, can acquire the applied load with better accuracy and precision, also allowing better minute change capture in the strain throughout the top plate and not just at the corners [23]. Above and underwater top (Figure 2a,b) and bottom (Figure 2c,d) force plates were built in duralumin to minimise force plate mass. The mounting apparatus plays a crucial role in providing accurate and reliable measurements, with the force plate top and core unattached with commercial bushings.

**Figure 2.** Above and underwater top (**a**,**b** panels) and core (**c**,**d** panels) 300 mm × 250 mm and 600 mm × 300 mm force plates top.

#### *2.4. Start Block, Underwater Structure, Handgrips and Feet Support*

Two start block projects were previously designed: (i) a bulky and solid structure and (ii) a lattice block with a declination support, which was replaced by a lattice galvanized steel structure with zero inclination due to excessive mass and reduced stability showed by the previous structures. The starting block was projected to be fixed over an underwater structure attached vertically to the swimming pool wall by front and rear edges, being similar to a previous structure used by Pereira et al. [13,14] and de Jesus et al. [24] for flip turn and backstroke start kinetics assessment. The first underwater structure included a two independent force plates support. The underwater structure evolved from a heavier to a lattice form, with its version being slighter, including holes with 100 mm distance between them placed at different heights on both force plates. It presented a flat rectangular surface for swimming turn analysis (replying the touch pad; FINA, FR 2.13), with a hollow area for underwater force plates embedding, as previously used in turn analysis [13,14].

Handgrips were projected to be independent and framed in galvanized steel, being the first design very versatile to be easily used in ventral and dorsal start technique analysis. However, the first design had shown a handgrip positioning dependency on measured strain signals that did not allow real training and competitive swimmers' movement. In fact, strain gauges would be bonded to the handgrip pipes, obliging swimmers to position their hands on a fixed place to allow comparisons (this was eliminated by the handgrips fixation on each lateral force plate top). With two force plates fixed, each one on the start block lateral, forces and moments could be measured and handgrips positioning could be located. The second handgrip project was based on a simple and outdated handgrip version (e.g., [24]), being updated for two horizontal (i.e., the highest and the lowest, 0.43 and 0.56 m above water surface, respectively) and vertical bars, following the OSB11 starting block configurations. The final handgrip prototype received fine arrangements due to the existent pipe profile. The adjustable feet support for ventral and backstroke start was framed in galvanised steel and nylon (FINA, FR 2.7 and 2.10, respectively), allowing the five rear foot authorised positions.

#### *2.5. Finite Element Analysis*

We conducted static structural simulations using modelling software for finite element analysis (Ansys v.12.1, ANSYS Inc., Canonsburg, PA, USA), thus enabling predictions on how the dynamometric unit would strain under isolated and integrated conditions. Dynamic simulations were applied to verify resonance frequency, equivalent stress, equivalent strain and deformations. Based on Roesler et al. [20] geometry and sensor location definition, a 8000 N load was vertically and antero-posteriorly applied to confirm the previously determined sensor location (Figure 3a). The 8000 N load was simulated to allow force plate use in other data collection purposes that had depicted increased ground reaction forces (e.g., long distance jump, 15.2 times body weight; [25]).

Static and dynamic force plate simulations were performed with a core, top and mounting apparatus (i.e., polyethylene bushing and screw). Simulations with the starting block and handgrips were conducted with a 2500 N (centrally located; Figure 3b) and 2000 N load (vertical and antero-posterior; Figure 3c) in the commonly used handgrip positioning for ventral and backstroke start [10]. The underwater structure with force plates vertically mounted was simulated with the gravity at sea level (i.e., 9.81 m/s2) and values of total deformation, equivalent stress and equivalent elastic strain have also been obtained. The most refined mesh for force plates, the starting block, underwater structures and handgrip simulations was composed of pyramids and cobbled with a 1 mm length and 39,197 nodes and 12,431 elements, 76,344 nodes and 15,724 elements, 273,837 nodes and 107,653 elements, and 46,372 nodes and 17,459 elements, respectively.

**Figure 3.** Strain numerical simulation in the assembled force plate (**a** panel), starting block (**b** panel), and handgrips (**c** panel).

#### *2.6. Electrical Circuit*

Following structures manufacturing, strain gauges were bonded to the platforms as sensing elements due to the previous research group background and short budget available. Each force plate was instrumented with 24 waterproof strain gauges (Kyowa, Electronic Instruments, KFW-5-120-C1-5M2B, Tokyo, Japan), arranged in six independent full Wheatstone bridges, minimising temperature effects. Those strain gauges were internally bonded to each force plate core and positioned as depicted in Figure 4a–d.

**Figure 4.** Strain gauge zones: Vertical force (y), antero-posterior (x) and medium-lateral (z) moment (top and bottom view, **a**,**b** panel), antero-posterior (x) force (**c** panel) and medium-lateral (z) force and vertical moment (**d** panel).

After bonding, each strain gauge received an additional protection of a two-part polybutadiene resin designed for re-enterable splice protection (Scotchcast TM re-enterable electrical insulation resin 2123, 3M TM, St. Paul, MN, USA), minimising chlorine wear. Strain gauge wires were brazed in a full Wheatstone bridge configuration (six for each force plate), silicone protected. Each full Wheatstone bridge was connected to a shielded and unfilled cable and provided data from each variable of interest (i.e., horizontal, vertical and lateral forces and momenta).

The six-shielded unfilled cables were connected to an analogue-to-digital converter module to transmit full Wheatstone bridge signals (NI9237 50 kS/s/ch 24 bit-4Channels, National Instruments Corporation, NI™, Austin, TX, USA) and to its scanner chassis (NI CompactDAQ 9172 and 9188 with 8 slots, National Instruments Corporation, NI™, Austin, TX, USA) through RJ50 connectivity, which interfaced with the computer. Strain gauges were identified by digit codes and cables wiring had also been standardised with colour code jacket terminals nearby the RJ50 plug. The dynamometric system specs included 42 strain plus two trigger channels acquiring at a 2000 Hz sampling rate.

Custom-designed data processing software was created in LabView 2013 (SP1, NI™, Austin, TX, USA) to acquire, plot and save the strain readings of each three axial force and moment of force component from each force plate. The executable file was programmed to record data in a total of 8 s (4 s before and after the trigger signal), which was a strategy implemented for the full relay start force acquisitions, anticipatory individual start (e.g., pre-activation) and turn actions (e.g., wave drag). Each force and moment of force curve profile was observed in real data time-acquisition and data acquisition files were later operated to transform strain signals into force-moment (using MatLab R2014a, The MathWorks Incorporated, Natick, MA, USA) routines for the required matricial conversion operations.

#### *2.7. Static Calibration*

Static calibration was performed on dry land before force plate proper use and applied both to a load and to an unload sequence with 10 kg individual masses (up to 50 kg in each positioning; cf. [8,18]) allowing the correspondence between strain and applied load. The vertical force component was calibrated at five positions with the use of a tension, compression machine (Instron 8804 Servohydraulic Fatigue Testing System, Instron®, Ilinois Tool Works Inc., Norwood, MA, USA) that laid the right load on each force plate top centre. On antero-posterior and lateral axis force calibration, platforms were vertically fixed on the wall and the load was applied on each centre of interest through a stainless-steel cable connection. A stainless-steel cable was fixed through holes made on the lateral of each force plate top (three per edge). Forces and moment were calibrated using the central and lateral holes (respectively).

#### *2.8. Laboratory Experimental Validation*

An experimental validation was completed comparing the results of a rigid body theoretical free rotation around a pivot point <sup>→</sup> *COP* fall force pattern [26] and the associated strain signal pattern generated. Rigid body inertia moment of the inverted pendulum had to be assessed previously to know mass distribution to allocate its centre of mass and to calculate the centre of mass to the centre of pressure distance. This assessment implies the mass and geometrical dependency of moment of inertia and dynamic behaviour observed in force patterns generated, obviating the use of any force platform except gravity acceleration knowledge.

#### *2.9. Ecological Experimental Validation*

The instrumented start block was tested in a 25 m long and 1.90 m deep indoor swimming pool for real data acquisitions, which were then qualitatively compared with starting and turning data previously presented in the literature. All experimental procedures conformed to the requirements stipulated in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) and were approved accordingly by the local research ethics committee. Swimmers and parents and/or guardians (when participants were under 18 years old) provided informed written consent before data collection.

Nine well-trained, healthy and able-bodied male swimmers (mean and standard deviations: 21.09 ± 5.64 years, stature 1.74 ± 0.05 m, and body mass 71.32 ± 11.03 kg) and eight age-group male swimmers engaged on a regular basis in regional- and national-level competitions (mean ± SD: 12.4 ± 0.6 years old, 155.1 ± 13.6 cm of height, 44.6 ± 10.9 kg of body mass, 14.1 ± 5.3% of body fat, with 3.5 ± 1.4 years of experience in competitive swimming and 3.3 ± 0.7 (2–4) of Tanner maturation scale by self-evaluation) volunteered to participate in the start and turn validation study, respectively.

Swimmers performed a familiarisation period with each start and turn technique studied (cf. [27,28]). For the start protocol, each swimmer randomly performed three maximal 15 m trials of backstroke with vertical handgrips, track and one step relay starts. Swimmers performing the turn protocol were also tested in a set of three trials for the open backstroke to the breaststroke medley turning technique. Trials started and finished from the mid-pool (at 12.5 m from the turning wall) and swimmers were instructed to swim in and out at maximum speed until the 12.5 reference [13,28]. Rest periods of 2 min were observed between each repetition in both protocols.

A start signal complying with the FINA SW 4.2 rule was produced through an official device (StartTime IV acoustic start, Swiss Timing Ltd., Corgémento, Switzerland) and instrumented to simultaneously generate an auditory signal and export a pulse to the force plates with convenient signal conditioning. In the one step relay start protocol and open backstroke to breaststroke turn the starter device (without sound) was triggered 7.5 m before the incoming swimmer had touched the wall.

#### **3. Results and Discussion**

#### *3.1. Simulations*

Table 1 presents strain in antero-posterior and vertical axes (8000 N centred) and resonance frequency of the first, third and final prototypes of above water 300 mm × 250 mm, 500 mm × 440 mm and underwater 600 mm × 300 mm platforms, corroborating Roesler's [20] suggestions for strain values between 100 and 500 με. The force plates design optimisation was a compromise among maximum rigidity, minimum mass and high frequency, which allows small deformations, uncoupling, good linearity and low hysteresis (c.f. [17]). The specific force plate application determined proper resonance frequencies and the maximal vertical load was supported. The waterproof force plate used by Roesler [20], with 500 mm × 500 mm framed in galvanized steel, obtained a 35 Hz resonance frequency, which was considered sufficient for underwater applications due to over damping effects. However, in the current project, force plate versatility has been prioritized and values higher than ~140 Hz were required (cf. [20]) as they can be both used independently of the starting block and out of the water. Commercial force platforms with a natural frequency ≥150 Hz have been well accepted for triple jump ground reaction forces assessment (e.g., [25]).


**Table 1.** Antero-posterior and vertical strain and their resonance frequency above 300 mm × 250 mm, 600 mm × 300 mm and 500 mm × 440 mm underwater force plates of first, third and final prototypes.

Table 2 shows results from one example of static simulation from each force plate considering all 24 strain gauge responses, which allow noticing maximal cross-talk less than 5% (cf. [20,21]). In fact, Roesler's [19] findings have shown ~3% of maximal interference among loads when 800 N had been applied. In addition, Lywood et al. [21] reported a cross-talk between orthogonal axes less than 5% for 40 N and 10 N of vertical and horizontal forces applied. The 300 mm × 250 mm force plate laterally positioned on the starting block with fixed handgrip was simulated with a 2000 N load applied vertically on the handgrips, indicating an expectable relevant lateral force (strain gauges 17, 18, 23, 24; cf. Figure 4d). When 8000 N was centrally and vertically applied on a 600 mm × 300 mm force plate, only strain gauges responsible for this measurement responded (i.e., 1 to 12, vertical force, antero-posterior and lateral moment; cf. Figure 4a,b).


**Table 2.** The 24 strain gauge responses when a 2000 N and 8000 N vertical load was applied on 300 mm × 250 mm and 600 mm × 300 mm force plates.

The starting block total deformation under a 2500 N centre vertical load was 0.00030553 m. The gravity at sea level (9.81 m/s2) tested over the underwater structure and two force plates vertically fixed on it revealed a maximal deformation of 0.00012322 m. Moreover, using the same standard gravity at sea level, the underwater structure with force plates has showed a maximum of 0.00000813 Pa and 0.0000409 m/m, considering equivalent von-Mises stress and equivalent von-Mises elastic strain, respectively, indicating short stress gradients in the underwater structure regions. An antero-posterior 2000 N load applied both on the lowest and on the highest horizontal and vertical handgrips revealed 200, 165 and 115 maximal με. A vertical 2000 N load applied on the lowest and on the highest horizontal and vertical handgrips indicated 591, 585 and 205 maximal με.

#### *3.2. Calibrations*

The calibration regression equation for the antero-posterior and medium-lateral axis of 300 mm × 250 mm and 600 × 300 mm force plates is depicted in Figure 5a–d. Results evidenced the previously noticed linearity (*R*<sup>2</sup> ranging between 0.97 and 0.99) and non-meaningful cross-talk between orthogonal axes (small and negligible; <5%) when quantifying any couple of force plate output signals (cf. [20,21]). Calibration results of 300 mm × 250 mm force plates are depicted considering the forces applied on handgrips positioning previously simulated.

**Figure 5.** *Cont.*

**Figure 5.** Force plates calibration results: 300 mm × 250 mm force plate antero-posterior force (**a** panel), medium-lateral force (**b** panel), 600 mm × 300 mm force plate antero-posterior force (**c** panel), and medium-lateral force (**d** panel).

Figure 6 presents vertical force, antero-posterior and lateral moment calibration (in normalised strain measure) for the underwater 600 mm × 300 mm force plate, evidencing linearization and cross-talk ~5% between orthogonal sensors (cf. [18,20]). The force plate design used has revealed greater sensibility in responses to the vertical loads, corroborating Lywood [21].

**Figure 6.** Force plates calibration results: 600 mm × 300 mm force plate vertical force.

Repeatability for loads above 300 N and below 40 N has revealed within 5% and 10–15% with a 95% confidence level. The <sup>→</sup> *COP* locus was more uncertain farther from the platform centre with a reasonable radial around centre dependency.

#### *3.3. Laboratory Experimental Validation*

Due to in situ installation procedures, usage and aging, force plates accuracy may decrease, which can be propagated to the already calculated kinetic quantities. Based on these limitations, some research groups have developed systems to assess force plate accuracy using ad hoc designed in situ devices [8,26]. In the current study, static calibrations were followed by dynamical calibrations performed with a rigid body falling procedure [26], revealing homogeneity of static calibration results, particularly in the time = 0 force time curve of the falling body. The simultaneous correlation coefficient between the posterior strain filtered signal (moving average of 32 samples) and the theory generated by a previous application in MatLab R2014 (MathWorks Inc., Natick, MA,USA) has shown a value of 0.95.

#### *3.4. Ecological Experimental Validation*

The horizontal force–time curve exerted by hands and produced during a backstroke start data acquisition is similar to the one peak force profile previously presented (~0.5 BW; [24]). Most of the force–time curves displayed in backstroke start studies have analysed the horizontal component exerted on feet (e.g., [24,29]), which is similar to the dual peak force profile found in the current study (Figure 7a), with swimmers performing ~1.6 BW in the both peak forces. Vertical backstroke start forces exerted by feet have shown similar values to those found in de Jesus et al. (i.e., ~1.0 BW; [10,29]). Lateral force exerted by feet has depicted a peak force instant before the hands-off. Horizontal and vertical force–time curves exerted on feet and hands obtained in the new track start technique performed with the back plate (Figure 8) have registered similar rear (~800 N and ~1000 N) and front foot (~800 N and 600 N; cf. [1,8]) as well as hand force profiles (~150 N and ~800 N; cf. [8]). Previous studies have evidenced that the test–retest reliability of the kinetic gait parameters in the aquatic environment presented poor (medial lateral force component) to excellent reliability (vertical and antero-posterior force component; [30]).

(**a**)

**Figure 7.** *Cont.*

(**c**)

**Figure 7.** Backstroke start horizontal (**a** panel), vertical (**b** panel) and lateral (**c** panel) forces from seven trials of one swimmer (**c** panel).

**Figure 8.** Individual antero-posterior (x) and vertical (y) force—time curves representing forces exerted on left and right hands (LH and RH), as well as rear and front foot (ReF and RF) in the new track start technique synchronised with the trigger (trg).

In relay start techniques, the horizontal force–time curve exerted on the feet and obtained during one step start technique (Figure 9) revealed a similar finding as obtained by Takeda et al. [31], although these authors had not measured right and left forces exerted on each foot.

**Figure 9.** Horizontal force–time curve profiles from eight swimmers at the one step relay start technique: Rear foot (red line), front left and right foot (green and blue line, respectively).

Tri-axial forces measured in one underwater force plate during the open backstroke to breaststroke changeover have evidenced two distinct turn phases, the hand contact (~4.60 to 5.50 s) followed by swimmers' rotation (~4.80 s to 7.30 s) and the push-off (~7.30 to 7.80 s; Figure 10). Previous studies have already mentioned that the push-off phase depicted explosive lower limb movements since the first feet contact (~7.20 s) to the end of push-off (~7.80 s), which has been mainly observed in the horizontal force component [13,28]. Purdy and co-authors [31] have found ~193 N for the peak horizontal ground reaction force in female swimmers performing the open turn technique.

**Figure 10.** Individual backstroke to breaststroke open medley turning technique force–time curve with horizontal, vertical and medium-lateral components (purple, yellow and red line, respectively).

The current and the previously developed instrumented starting block configurations have shown restrictions regarding independent hand force and momentum measures with natural swimmers' hands and feet placement during all ventral start techniques. (e.g., [2,6,8]). The solution found in the current project was the replication of four more tri-axial force plates of 300 mm × 125 mm designed and numerically simulated (Figure 11a–d) to be placed in the frontal start block edge. An 800 N vertical, and 200 N medium-lateral and horizontal load were applied and registered 280.11, 109.89 and 117.34 με maximal deformation and a maximal 286.45 Hz frequency. The four-force plates solution

enables swimmers to grasp the top and bottom of the starting block surface, being versatile and valid to independently measure each swimmer's hand and foot forces and momenta generated regardless of the ventral starting technique used.

**Figure 11.** The four 300 mm × 125 mm force plates in static and dynamic simulation conditions for vertical, antero-posterior and medium lateral deformation (**a**–**c** panels), as well as the frequency analysis (**d** panel).

Notwithstanding the relevance of the dynamometric unit solution developed for start and turn analysis, it is recognised that future work is needed to improve main limitations: (i) finite element analysis revealed important data for project manufacturing. However, irregularities in the starting pool wall cannot be controlled and could affect measurements; (ii) despite static and dynamic calibration procedures, force plate linear responses were revealed, suggesting a less time-consuming in situ calibration procedure should be developed to avoid the detachment of each force plate from the dynamometric central; and (iii) increasing underwater force plate placement possibilities beyond side-by-side but also in an up and down location would cover independent feet force measurements from other turn techniques.

#### **4. Conclusions**

The dynamometric central for individual and relay swimming start and turn techniques framed according to FINA facility rules has evidenced reliable and accurate external force data. This device is ecologically valid and versatile, being able to be used in integrated or in future independent force plate analysis, as bow wave measurements in turns, passive drag and jumping techniques.

**Author Contributions:** Conceptualization, K.d.J. (Karla de Jesus), L.M., H.R., M.V. and J.P.V.-B.; data curation, K.d.J. (Karla de Jesus), L.M., M.V., R.F. and J.P.V.-B.; formal analysis, K.d.J. (Karla de Jesus), L.M., H.R., N.V., M.V. and J.P.V.-B.; funding acquisition, K.d.J. (Karla de Jesus), M.V., R.F. and J.P.V.-B.; investigation, K.d.J. (Karla de Jesus), K.d.J. (Kelly de Jesus), M.V., R.F., J.P.V.-B.; methodology, K.d.J. (Karla de Jesus), LM., H.R., N.V., K.d.J. (Kelly de Jesus), M.V., R.F., J.P.V.-B.; software, L.M., H.R., N.V., M.V., project administration; K.d.J. (Karla de Jesus), R.F., J.P.V.-B.; resources, K.d.J. (Karla de Jesus), R.F., J.P.V.-B.; supervision; K.d.J. (Karla de Jesus), L.M., M.V., R.F., J.P.V.-B.; validation; K.d.J. (Karla de Jesus), L.M., K.d.J. (Kelly de Jesus), J.P.V.-B.; visualization, K.d.J. (Karla de Jesus), R.F., J.P.V.-B.; writing—original draft, K.d.J., (Karla de Jesus), L.M., H.R., N.V., K.d.J. (Kelly de Jesus), R.F., J.P.V.-B.; writing—review and editing, K.d.J. (Karla de Jesus), L.M., M.V., R.F., J.P.V.-B.

**Funding:** This research was funded by the [Coordination for the Improvement of Higher Education Personnel—CAPES] grant number [BEX 0761/12-5/2012-2015], [Foundation for Science and Technology—FCT] grant number [EXPL/DTP-DES/2481/2013-FCOMP-01-0124-FEDER-041981], [CAPES-FCT] grant number [99999.008578/2014-01], SANTADER grant number [PP/IJUP2011/123] and [Amazonas State Research Support Foundation - FAPEAM] grant number [POSGRAD 2017 FAPEAM, 002/2016]. Funders had no role in study design, data collection and analysis, decision to publish, or manuscript preparation.

**Acknowledgments:** Leandro dos Santos Coelho, Alexandre Igor Araripe Medeiros and Phornpot Chainok provided valuable feedback on manuscript drafts.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Isokinetic Strength in Peritoneal Dialysis Patients: A Reliability Study**

#### **Daniel Collado-Mateo 1,\*, Francisco Javier Dominguez-Muñoz 1, Zelinda Charrua 2, José Carmelo Adsuar 1, Eugenio Merellano-Navarro <sup>4</sup> and Armando Manuel Raimundo 2,3**


Received: 5 July 2019; Accepted: 26 August 2019; Published: 29 August 2019

**Abstract:** Although there are studies assessing the effects of interventions on the knee strength of patients undergoing dialysis, there are no previous studies investigating the test–retest reliability of isokinetic measures in people undergoing peritoneal dialysis. The objective of this study was to determine the relative and absolute reliability of peak torque and work measurements for isokinetic concentric knee and elbow extension and flexion in peritoneal dialysis patients. Thirty-one patients undergoing peritoneal dialysis (19 males) participated in the current study. All isokinetic tests were performed using a Biodex System 3. Participants performed three concentric repetitions of each test (flexion or extension) with the dominant limb (knee and elbow) at 60◦/s. Peak torque (Nm) and work (J) were extracted. The intraclass correlation coefficient (ICC), standard error of measurement (SEM), and smallest real difference (SRD) were calculated. The results showed that all knee peak torque and work measures had an ICC of >0.90. On the other hand, the ICC for peak torque and work in the elbow concentric extension was <0.90, while the remaining elbow-related variables achieved an excellent reliability. Therefore, isokinetic dynamometry is a reliable technique to evaluate peak torque and work for concentric flexion and extension in both the knee and elbow joints in patients undergoing peritoneal dialysis.

**Keywords:** kidney; torque; exercise; physical fitness; peritoneal dialysis

#### **1. Introduction**

Peritoneal dialysis (PD) patients present a condition resulting from multiple physiological and behavioral changes which contribute to muscle wasting [1]. Recent studies showed that PD patients often present decreased levels of physical activity (PA) matching a sedentary lifestyle [2]. These circumstances significantly affect the health condition of those patients and are associated with significant morbimortality [3]. Moreover, the nutritional status and sedentary behavior directly affect muscle function, exercise performance, physical function, strength, and health-related quality of life [4–6]. Although the proportion of patients active during leisure time is low [4], they need adequate levels of strength to accomplish daily life activities such as the need to displace, walking up and down stairs, maintaining a standing position, etc.

Exercise programs are recommended to increase lean body mass, strength, and physical functioning in frail elderly persons and those with chronic diseases, including PD patients [5]. Precise and sensitive strength tests are required to appropriately extract conclusions in studies focused on the evaluation of strength. Among these tests, the hand grip dynamometer and the isokinetic dynamometer are two of the most reliable and widely used devices. In this regard, the handgrip dynamometer has been previously used to evaluate muscle strength of the upper limb in patients undergoing dialysis [4,7]. Isokinetic dynamometry has been previously used to evaluate the effects of different interventions and therapies on knee flexion and extension strength in dialysis patients [8,9], whereas, to our knowledge, there is no study evaluating upper limb strength using an isokinetic dynamometer in this population.

Although there are studies assessing the effects of interventions on the knee strength of patients undergoing dialysis, to our knowledge, no previous study has investigated the reliability of knee isokinetic procedures adapted for use in PD patients with poor muscle strength. In healthy individuals, the use of isokinetic dynamometry to evaluate the knee strength when subjects receive adequate instructions and are familiar with the equipment is sufficiently reliable [10,11], but the reliability in PD patients still remains unknown, which impairs and limits the appropriate interpretation of results. In this regard, reliability is considered an important prerequisite for the correct interpretation of isokinetic dynamometry data, which allows the clinician to identify whether or not a genuine change has occurred [12]. Therefore, the aim of the current study was to determine the relative and absolute intra-session reliability of peak torque and work measurements for isokinetic concentric knee and elbow extension and flexion in PD patients.

#### **2. Materials and Methods**

#### *2.1. Participants*

Patients were recruited from the Nephrology Unit from the Espírito Santo Hospital of Évora, Portugal. The inclusion criteria to participate in the study were as follows: (1) be a peritoneal dialysis patient for at least 6 months; (2) do not have any impediment to perform the strength tests according to the physician's criteria; and (3) give their informed consent to participate in the study. A total of 49 patients fulfilled the first two inclusion criteria and were invited to participate. Of them, 31 (12 women and 19 men) agreed to participate in the study and signed the informed consent. The University of Évora ethics committee approved the protocol of this study, which was conducted in accordance with the updated World Medical Association's Declaration of Helsinki for human studies [13]. Of the 31 participants, 18 were on continuous ambulatory peritoneal dialysis while 14 were on automated peritoneal dialysis. Patients were having PD for a mean of 943 ± 552 days.

#### *2.2. Instrumentation*

All isokinetic tests were performed using a Biodex System 3 quick-set isokinetic dynamometer (Biodex Corp., Shirley, NY, USA) and System 3 software (version 3.40). Body fat and lean percentage were assessed by dual-energy X-ray absorptiometry (DXA—Hologic QDR, Hologic, Inc., Bedford, MA, USA). Finally, physical activity level was assessed using the ActiGraph accelerometer device with dimensions 3.8 × 3.7 × 1.8 cm (27 g). All participants were asked to use an accelerometer on the right hip, near the iliac crest, during seven consecutive days.

#### *2.3. Procedures*

The procedure is depicted in Figure 1. The intra-session reliability of the measurements was evaluated. All tests were conducted by the same researcher. Only the dominant arm and knee were tested. The dominant arm was defined as the arm used to write and the dominant knee was defined as that of the preferred kicking leg. Knee and elbow protocols followed the Biodex Isokinetic System 3 quick-set application/operation manual instructions [11]. Prior to the implementation of the protocols, all the subjects performed 15 min of warm-up with joint mobilization and stretching.

1. Knee protocol: Participants were seated in a seatback tilt at 85◦. The dynamometer orientation and dynamometer tilt were 45◦ and 0◦, respectively. The participant's axis of rotation of the knee was aligned with the dynamometer shaft. All patients were informed about the tasks they were going to perform and performed two familiarization and warm-up repetitions. The weight of the leg was recorded using the dynamometer software, and gravity adjustments were made. All participants were asked to perform three concentric movements of the knee involving alternative extension and flexion at 60◦/s. Reliability was calculated between the second and the third repetition. The rest interval was 2 min long. This protocol has been used previously in the scientific literature [14,15]. The participants were verbally encouraged during the tests.

2. Elbow protocol: Participants were seated in a seatback tilt at 85◦. The seat orientation was 15◦. The dynamometer orientation and dynamometer tilt were 15◦ and 0◦, respectively. Participants were stabilized with shoulder, waist, and thigh straps. They were informed about the tasks and performed two repetitions, aimed to warm-up and also to get used to the position, the angular speed, and the proposed task. The weight of the arm was recorded using the dynamometer software, and gravity adjustments were made. All participants were asked to perform three concentric movements of the elbow involving alternative extension and flexion at 60◦/s. Reliability was calculated between the second and the third repetition. The rest interval was 2 min long. The participants were verbally encouraged during the tests.

**Figure 1.** Study procedure. DXA: dual-energy X-ray absorptiometry.

#### *2.4. Measures*

Peak torque (Nm) and work (J) were extracted from the System 3 software for knee and elbow flexion/extension. Peak torque is defined as "the single highest torque output recorded throughout the range of motion of each repetition". Work is defined as "the output of mechanical energy" and is represented by the area under the curve of torque versus angular displacement [16].

#### *2.5. Statistical Analysis*

The statistical analysis was performed following the criteria used in previous studies [12,14,15]. Absolute values for peak torque and work were obtained as means and standard deviations. Differences in the descriptive characteristics between men and women were evaluated using independent samples *t*-test.

Relative reliability was estimated using the ICC3,1 (intraclass correlation coefficient, two-way mixed single measures) with 95% confidence intervals across the two test repetitions [17]. An ICC higher than 90 was interpreted according to Munro et al. as excellent [18]. A paired sample *t*-test was

performed to analyze differences in the mean values of the isokinetic variables between Repetitions 2 and 3.

Absolute reliability was determined by calculating the standard error of measurement (SEM; SEM = SD, where SD is the mean SD of Repetition 2 and Repetition 3) and the smallest real difference (SRD; SRD = 1.96) [19]. Additionally, the SEM and SRD were converted to percentages in order to facilitate the comparability of errors of measurement with those in other studies. These percentages were calculated as follows: SEM% = (SEM/mean peak torque or work of the two repetitions)·100 and SRD% = SRD/(mean peak torque or work of the two repetitions)·100.

Isokinetic variables were also correlated with anthropometric and body composition variables using bivariate Pearson's correlation.

#### **3. Results**

The characteristics of patients undergoing PD are reported in Table 1. The mean age was 48.45 (13.39) and the mean body mass index (BMI) was 24.35 (3.67), which is close to the overweight threshold. Differences between men and women were observed in age, body composition, height, and weight. In general terms, there was sedentary behavior, with more than 18 h/day of sedentary time and about 5000 daily steps.


**Table 1.** Characteristics of patients undergoing peritoneal dialysis (*N* = 31).

Data reported as mean ± SD. \* Significantly higher compared to the other group based on results from independent samples *t*-test. BMI: body mass index.

Table 2 shows the peak torque and work at 60◦/s in each of the two repetitions for all variables. There were no statistically significant differences in any test. In general, the mean peak torque and work of male patients were higher than values from female participants.

Table 3 summarizes the ICC values and the 95% confidence intervals, as well as the SEM and SRD in absolute values and percentages. All knee peak torque and work measures had an ICC greater than 0.90, which is excellent according to the classification by Munro, Visintainer, and Page [18]. On the other hand, peak torque in the elbow concentric extension in males and work in the elbow concentric extension in males and the general population showed an ICC lower than the threshold for excellent reliability. The remaining elbow-related variables achieved an excellent reliability. Regarding peak torque, the SRD% in the general population ranged between 18.41% in the elbow concentric flexion and 29.44% in the knee concentric extension. Regarding work, the SRD% in the general population ranged between 26.15% in the elbow concentric extension and 33.63% in the knee concentric flexion.


**Table 2.** Summary of isokinetic peak torque and work at 60◦/s in two repetitions (*N* = 31).

Note: Values are mean ± standard deviation; \* Paired sample *t*-test was performed to analyze differences in the mean values of the isokinetic variables between Repetitions 2 and 3.



Abbreviations: CI, confidence interval; ICC, intraclass correlation coefficient; SEM, standard error of measurement; SEM%, standard error of measurement as a percentage; SRD, smallest real difference; SRD%, smallest real difference as a percentage.

Table 4 shows the correlations between isokinetic strength and different variables such as age, height, weight, BMI, fat mass, and lean mass. The variable "height" was significantly correlated with all isokinetic strength outcomes, while BMI was not correlated with any variable.


**Table 4.** Correlations between isokinetic strength and anthropometric and body composition variables.

\* *p* < 0.05; \*\* *p* < 0.01; BMI, Body mass index.

#### **4. Discussion**

The main finding of this study was that the test–retest reliability of elbow and knee concentric flexion is good or excellent in PD patients. These results were similar to those observed in sit-to-stand-to-sit, six-minute walk, one-leg heel-rise, and handgrip strength tests in people undergoing hemodialysis [20] or in the incremental shuttle walk test, the estimated maximum repetition for quadricep strength, and VO2peak by cardiopulmonary exercise testing in non-dialysis chronic kidney disease [21]. However, little was known about the reliability of physical function tests in PD patients and, to our knowledge, this is the first study aimed to evaluate the test–retest reliability of isokinetic measures in this population.

The evaluation of physical function is relevant since it is strongly related to quality of life, independence, and the ability to perform activities of daily living. Patients suffering from chronic kidney disease may have a higher risk of having low strength levels and low muscular mass, which is commonly associated with a higher risk of mortality [22]. However, most of the studies aimed to evaluate physical function in patients undergoing dialysis have been conducted with hemodialysis patients [23], while further research is needed in PD patients.

According to Zuo et al. [24], exercise capacity may be reduced in about 96% of PD patients. This reduction in exercise tolerance could be partially determined by age, sex, and body composition and is strongly associated with health-related quality of life in this population [25]. In this regard, the reduced quality of life and physical function may be similar in patients undergoing hemodialysis or PD [26]. However, Kang et al. [27] observed more favorable mental and physical components in hemodialysis patients compared with PD patients. Although the differences between the two dialysis modalities in terms of physical function and quality of life still remain unclear, patient satisfaction is commonly higher in patients undergoing PD [26,27]. Therefore, physical function is a relevant measure that should be included in comprehensive health assessments in dialysis patients. The current study provides test–retest reliability parameters that should be used to interpret health-related physical function evaluations in this population.

Isokinetic dynamometry is considered the gold standard for dynamic muscle performance testing [28,29]; thus, clinicians and researchers should be encouraged to use this device to conduct their physical function evaluations. Previous studies have reported the test–retest reliability of isokinetic measures in several populations different from patients undergoing dialysis. In this regard, reliability results obtained in the current study are similar to those obtained in postmenopausal women with osteopenia [30], women with fibromyalgia [15], patients with knee osteoarthritis [31], or persons with chronic stroke [32]. Furthermore, the current study not only reports the reliability parameters of measures of the lower limb but also assesses the reliability of the elbow flexion and extension strength tests. Therefore, future research may use both the absolute and the relative reliability parameters reported here to evaluate whether an observed change in lower or upper limb strength represents a true change or not.

Almost every measure achieved an excellent reliability according to the classification by Munro, Visintainer, and Page [18]. The ICC in elbow flexion in men was lower than 0.90; thus, it was classified as "good" but not "excellent". On the other hand, the reliability of this measure was excellent among

women. Regarding the SRD, which is the parameter that defines the threshold to consider an observed change as "true" or "real", the values for the knee peak torque ranged between 20.70% in the knee concentric flexion of women and 31.29% in the knee concentric extension of men. Regarding the elbow's peak torque, it ranged between 11.84% in the concentric flexion of women and 24.56% in the concentric extension of men. Overall, higher reliability was observed in women compared to men, and the SRD was lower in the concentric flexion compared to the concentric extension. This was also observed when assessing the work. Although hypothetical, these sex differences might be related to the significant differences between men and women observed at baseline, i.e., age, height, weight, fat mass, and lean mass. However, given the relatively low sample size, this finding must be taken with caution.

The current study has some limitations. First, the participants did not undergo a familiarization session because of time and financial constraints. Second, although the sample size was similar to or even higher than those from previous studies [15,30,31], results from the stratification by men and women must be taken with caution. Third, the mean age of males was 52.42, which means that they were more than 10 years older than the women; thus, comparison between males and females could be influenced by that age difference. Despite these three limitations, the current study provides useful information about the reliability of isokinetic measures in people undergoing PD.

#### **5. Conclusions**

Isokinetic dynamometry is a reliable technique to evaluate peak torque and work for concentric flexion and extension in both the knee and the elbow joints in patients undergoing PD. Although results from division by sex must be taken with caution, higher reliability was observed in women compared to men and in the concentric flexion compared to the concentric extension. In addition to the reported good or excellent reliability based on ICC values, the present study provided novel SRD data which are associated to measurement error and individual variability and will assist healthcare professionals in interpreting treatment effects on isokinetic strength in this population.

**Author Contributions:** Conceptualization, N.B. and A.M.R.; Data curation, Z.C., N.B. and A.M.R.; Formal analysis, D.C.-M., F.J.D.-M. and J.C.A.; Investigation, D.C.-M., Z.C., J.C.A. and E.M.-N.; Methodology, N.B.; Supervision, J.C.A., N.B., E.M.-N. and A.M.R.; Writing—original draft, D.C.-M.; Writing—review and editing, D.C.-M., F.J.D.-M., Z.C., J.C.A., N.B., E.M.-N. and A.M.R

**Funding:** This research received no external funding.

**Acknowledgments:** The authors acknowledge the support provided by Manuel Amoedo in the evaluations of the patients.

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

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Relationship between Kinematic Variables of Jump Throwing and Ball Velocity in Elite Handball Players**

**Abdel-Rahman Akl 1, Ibrahim Hassan 2, Amr Hassan 3,\* and Phillip Bishop <sup>4</sup>**


Received: 14 July 2019; Accepted: 16 August 2019; Published: 20 August 2019

**Abstract:** The purpose of this pilot study was to evaluate the relationship between the kinematic variables of the right hand and left leg with ball velocity during jump-throwing phases in handball for better-informed training. We investigated ball velocity and the key kinematic variables of jump throwing during different throwing phases in three strides. Ten right-handed male handball professional players who had competed in the Egyptian Handball Super League participated in this study. Jump throwing performance was divided into three phases (cocking, acceleration and follow-through), which included eight events during the throwing. Five trials were captured for each player, and a 3D analysis was performed on the best trial. Results indicated that the velocity of the throwing hand was the most important variable during jump throwing, which was correlated with ball velocity during the three phases of performance in four events: Initial contact (IC) (r = 0.66\*), initial flight (IF) (r = 63\*), maximum height of the throwing hand (Max-HH) (r = 0.78\*) and ground contact (GC) (r = 0.83\*). In addition, the initial flight was the most important event in which players need to be using the best angles during performance, particularly the shoulder angle.

**Keywords:** jump throwing ability; 3D motion analysis; acceleration

#### **1. Introduction**

Team handball is an Olympic sport played worldwide at a professional level in several countries. Recently, handball has received increased attention in research studies, especially in biomechanics [1–8].

Jump throwing is an essential task in handball and is used frequently from different positions when a player shoots at the goal. As has been mentioned in earlier studies, handball players perform about 48,000 throws during the season, with the mean throwing speed of 130 km·h−<sup>1</sup> [9], they and commonly use (73–75% of the time) jump throwing throws during the competition [5,7]. Hence, jump throwing and players' ability to accelerate the ball with the arm throw are vital demands during a handball game.

During jump throwing in handball, ball velocity and jump height range are key factors for better throwing performance [10]. Thus, most studies have investigated ball velocity [5,11,12].

Kinematic variables contribute to the velocity of the ball in order to find the fundamental procedures for improving handball players. As such, previous studies have documented the kinematic results of jump throwing performance in handball. Wagner, Buchecker, von Duvillard and Muller [7] found a significant difference of ball velocity, body height and weight between elite and lower performance levels. As well as, Wagner, Pfusterschmied, von Duvillard and Müller [5] compared the ball velocity and throwing accuracy between jump and standing throws. In addition, van den Tillaar and Ettema [8] investigated the contribution of upper-extremity, trunk, and lower-extremity movements, and van den

Tillaar [2] compared the range of motion with throwing kinematic variables. Plummer and Oliver [1] investigated the effects of fatigue on kinematic and kinetic changes of upper-extremity jump throwing.

Many kinematic experiments have studied ball throwing for team handball players and investigated the movement phases during the ball throwing [7,8,13,14]. Few studies have examined the kinematic variables effects and established relationships with ball velocity using 3D biomechanical tools. The handball coaches tend to focus on improving the throwing velocity for players [15].

Thus, through the investigation in the biomechanics of throwing velocity in elite handball players for the purpose of providing guidance to coaches and players desiring to improve this key aspect of performance, previous studies have shown a distinctive contribution of a velocity–accuracy trade-off and have concluded that kinematic variable as displacement along the velocity accuracy is achieved as an important contributed factor of decision-making. Therefore, ball velocity is considered the main factor for high quality jump throwing towards the goal [2,16,17].

In this regard, Stirn, et al. [18] opined that there are many factors contributing to the final velocity of the ball at release. To evaluate these factors, different demanding and time-consuming acquisition and analysis methods are required, including kinematic and electromyography assessments. Wagner, Pfusterschmied, von Duvillard and Müller [5] argued that the lower extremity in this type of skill plays an important role to drive the upper-extremity during performance. Therefore, throwing performance is considered to be the final outcome of an efficient kinetic chain.

The drive leg step before the take-off during the throwing jump skill could be an essential part of throwing to provide a support for the transfer of momentum through the pelvis and trunk to the dominant throwing arm. The kinematic investigation of the current study should be helpful for identifying the optimum jumping throwing mechanics. Consequently, we evaluated the relationship between ball velocity and some kinematic variables to identify key variables which should contribute to improving jump throwing performance.

Few studies have examined kinematic variable effects and established relationships with ball velocity using 3D biomechanical tools. Our goal for this study was to evaluate the correlation between key variables and throwing velocity, with the goal of providing information useful to coaches and players for improving jump-throwing performance. Therefore, we hypothesized that there would be positive and strong relationships between the kinematic variables of the right hand and left leg with ball velocity, and we determined key kinematic variables of jump throwing.

#### **2. Materials and Methods**

#### *2.1. Subjects*

Ten right-handed male handball players participated in current study (age: 20.8 ± 1.21 years; body mass: 82.8 ± 8.57 kg; height: 189.6 ± 8.65 cm, training experience: 9 years). They were part of a professional team which competed in the Egyptian Handball Super League. The study was approved by the institutional ethics committee of studies and research, and each player's consent was obtained.

#### *2.2. Procedures*

After a 15-min warm-up including general and shoulder-specific mobility exercises, as well as stretching and familiarization with the protocol, participants performed jump-throws after three running steps while positioned in front of the goal. Players were instructed to throw the ball as fast as possible. A total of five successful attempts were recorded for each player, with a one-minute rest between attempts. The best attempt, according to velocity of the ball, was selected for 3D analysis. The jump throwing skill was broken into three phases: The cocking phase, the acceleration phase, and the follow-through phase. Eight events were identified, beginning with the touch down (this event was determined when the jump leg touched the force platform starting the take-off).

The second event was the maximum ground reaction force (this event was designated as when the peak ground reaction force was achieved). Next was take-off (this event was designated as when the jump leg left the force platform). Maximum arm cocking was designated as when the arm reached the maximum back swing. Maximum height of center of mass was defined as when the center of mass achieved maximum height.

The maximum height of throwing arm was defined as when the throwing hand was at the maximum height. Ball release was defined as when the ball was released from the throwing arm, and the last event was landing, which was designated as when the player touched down from flight during performance (Figure 1).

**Figure 1.** Jump throwing phases (cocking phase, acceleration phase, follow-through phase), and events (touch down, maximum ground reaction force (GRF), take-off, maximum arm cocking, maximum height of center of mass (COM), maximum height of throwing arm, ball release, and landing) during performance.

The kinematic variables of the right-hand (throwing arm) were measured using a 3-D motion capture system (Simi Reality motion analysis V. 9.0.6, eight synchronized Basler scA640-120gc-High-Speed Cameras were used at a 100 Hz frequency) that tracked the position of the reflective markers on anatomical landmarks according to the Hanavan model [19,20], (Figure 2). The 3D coordinates were the X (medio–lateral), Y (anterior–posterior) and Z (vertical) directions. Throwing performance was evaluated by the ball velocity [21], which was calculated by creating a rigid body of the ball using software and tracked. The angular kinematics of the joints were derived from relative angles between the two relevant segments (sagittal plane). A strain gage force platform (MP4060®, Bertec Corporation, Columbus, OH, USA) was used to determine the touch down and maximum ground reaction force events. The standard setting of software for filtering were used, and differentiation was used to calculate the kinematics variables [22].

#### *2.3. Statistical Analysis*

The Pearson coefficient of correlation was used to determine the strength of the relationships between each kinematic variable and ball velocity during each of the jump throwing phases (Table 1). Asterisk signs above the number represent significant differences between kinematic variables and ball velocity, and (\*) indicates *p* ≤ 0.05 for statistical procedures (SPSS 21, V. 21 Statistics for Windows. IBM Corp, Armonk, NY, USA).

**Figure 2.** Full body marker set in (**A**) anterior (**B**) posterior view. The markers were placed at tempus head (TEMP), acromion left/right (LAC, RAC), epicondylus lateralis (elbow lateral) left/right (LLE, RLE), trochanter major left/right (LMT, RMT), Condylus lateralis left/right (LCL, RCL), malleolus lateralis left/right (LML, RML), foot tip left/right (LFT, RFT), vertebra C7 (VC7), vertebra Th8 (TH8), mid spina Iliaca posterior superior (SAC), middle finger base joint left/right (LMF, RMF), and heel left/right (LH, RH).

#### **3. Results**

The relationships between kinematic variables and ball velocity are shown in Tables 1 and 2 during each jump throwing phase (see Figure 1).

Table 1 provides the Pearson correlation coefficients between ball velocity and kinematic variables (displacement, velocity, and acceleration) at three investigated phases for eight events. No significant correlation was seen between ball velocity and right-hand kinematic variables during the cocking phase (displacement) and the acceleration phase (acceleration).

During the cocking phase, the ball velocity showed moderate correlations with the right arm velocity at initial contact (IC) (Y and R coordinates) (r = 0.66, r = 0.63) and initial flight (IF) (Y) (r = 0.63). Furthermore, there was a relatively strong correlation (r = 0.72) between the right arm acceleration at maximum ground reaction force (Max-GRF) (Y) and ball velocity. Through the acceleration phase, right arm displacement at maximum height of body (GC—ground contact) during flight (Max-GCH) (Y) had the highest correlation coefficients during this phase (r = 0.83) with ball velocity, and at release ball (RB) (Y) a strong correlation (r = 0.77) was observed. Likewise, ball velocity had moderate and strong correlations with the right arm displacement at maximum height of the throwing hand (Max-HH) (R) and RB (R), (r = 0.67 and r = 0.79, respectively). In addition, strong and moderate correlations were seen during the acceleration phase between velocity of the right hand and the ball velocity at Max-HH (Y) and Max-HH (R), (r = 0.78 and r = 0.70, respectively). As shown in Table 1 during the follow-through

phase, the kinematic variables of the right hand such as the displacement at GC (Y, R) and velocity at GC (Y) had the highest correlation coefficients (r = 0.84, r = 0.88, and r = 0.83, respectively). A moderate correlation was observed between ball velocity and the right-hand acceleration (r = 0.68) at GC (Y) during this phase.


**Table 1.** Correlation between right-hand center of mass kinematic variables and ball velocity during jump throwing phases in handball. N = 10.

Note: Coordinates: X (medio–lateral), Y (anterior–posterior), Z (vertical), and R (resultant); IC: Initial contact (end of last step); Max-GRF: Maximum Ground Reaction force; IF: Initial flight; Max-BC: Maximum back cooking; Max-GCH: Maximum height of body GC during flight; Max-HH: Maximum height of the throwing hand GC during flight; RB: Release ball; GC: Ground contact; \* Correlation were considered significant at *p* < 0.05.

**Table 2.** Correlation between right hand and left leg angular variables and ball velocity during jump throwing phases in handball.


Note: IC: Initial contact (end of last step); Max-GRF: Maximum ground reaction force; IF: Initial flight; Max-BC: Maximum back cooking; Max-CGH: Maximum height of body GC during flight; Max-HH: Maximum height of the throwing hand GC during flight; RB: Release ball; GC: Ground contact; \* Correlation were considered significant at *p* < 0.05.

The Pearson correlation coefficients between ball velocity with the right hand and left leg angular variables (angle, velocity, acceleration) during the three phases can be seen in Table 2. There were no significant correlation coefficients observed between the ball velocity and the kinematic variables of right elbow during all phases except a moderate negative performance score (r = −0.64) of angle at Max-HH during the acceleration phase. Likewise, during the follow-through phase, the velocity of the left knee had a moderate negative correlation (r = −0.67) at GC. Right shoulder velocity and acceleration had a moderate correlation with ball velocity at IF during the cocking phase (r = 0.69, r = 0.67, respectively). At RB during the acceleration phase, the right shoulder velocity had a moderate negative correlation (r = −0.67). Left hip acceleration during the cocking phase had a moderate negative correlation with ball velocity at Max-GRF and IF (r = −0.65, r = −0.67, respectively), and the angle of the left hip had a strong correlation (r = 0.73) at GC during the follow-through phase. The angle of left ankle during the cocking phase had a strong negative correlation with performance at IC and Max-GRF (r = −0.71, r = −0.80, respectively).

#### **4. Discussion**

The current study examined the relationships between ball velocity and kinematic variables during the jump throwing of elite handball players. The main aim was to determine the relationships between ball velocity and right hand (throwing hand) linear kinematics, as well as right arm and left leg (take-off leg) angular kinematics during three investigated phases over eight events.

The results in Table 1 show that the kinematic variables at maximum ground reaction force, maximum height of body center of gravity during flight, release ball and ground contact events were correlated strongly with ball velocity. These findings support a previous study which indicated important kinematic variables of the standing throw in handball [23].

The analysis of the cocking phase in our study indicated a moderate correlation between ball velocity and the velocity of the right hand at IC (Y, R) and IF (Y). This finding indicates that the primary velocity of elbow and the initial flight with the ball increased the ball velocity. During the same phase, the right-hand acceleration had a strong correlation with ball velocity at Max-GRF (Y). This strong correlation indicates that the Max-GRF during the jump throwing improved the horizontal velocity of the right hand (elbow and wrist joint) before ball release, which led to improving the arm swing during throwing.

The displacement of the right hand during the acceleration phase was strongly correlated with ball velocity at Max-CGH (Y), RB (Y) and RB (R). In addition, the Max-HH (R) during this phase correlated moderately with the ball velocity. This finding implies that maximum height of body GC and the dominant throwing hand increase the kinematic requirements of the hip and shoulder to maintain a high velocity of ball and performance; van den Tillaar and Ettema [24] showed some kinematic changes during overarm throwing related to the elbow extension and internal rotation.

Furthermore, the velocity of the right hand correlated strongly with Max-HH (Y) and moderately with Max-HH (R). The positive correlation with ball velocity and maximum heights of the throwing hand (Y, R) coordinates during the acceleration phase implies that arm motion continues until ball release time, which generates a ball rotation around the central axis, increases and generates the required release ball velocity. This supports the study of Werner, et al. [25], who suggested that the arm acceleration phase is the dynamic phase between the maximum external shoulder rotation and the instant of ball release.

During the follow-through phase, the results indicated that a higher correlation between ball velocity and the displacement of the right hand at GC (Y and R), and the velocity of the right hand had a high correlation at GC (Y). However, a moderate correlation was seen during this phase with the acceleration of the right hand at GC (Y). A strong correlation of displacement and velocity were found during this phase, so we suggest that when players perform the jump throw, they must increase their movement horizontally to enable forward movement at the landing. Hirashima, et al. [26] indicated that the ability of skilled throwers to optimize the throwing arm event of inertia during the arm cocking and arm acceleration phases highlights the importance of the arm movement to reach a high ball velocity.

Our findings in Table 2 indicate that the moderate negative correlation between ball velocity and the right elbow angle during the Max-HH event. This finding indicates that the smaller angle at the elbow joint assists in the improvement of a higher ball velocity at release for longer throwing. This finding supports previous studies [8,15,23], which indicated that the elbow extension and internal rotation velocity were factors for fast shoulder throwing, wherein about 73% of the contribution of ball velocity was based on the maximal internal rotation velocity of the shoulder and maximal elbow extension during the throw.

A moderate correlation was seen between the ball velocity and the right shoulder velocity and acceleration variables at IF during the cocking phase. In addition, a moderate correlation was observed between the right shoulder velocity at RB and ball velocity during the acceleration phase. This finding indicates that player who moves forward at take-off also maintains a higher projection angle with the wrist hyperextension at the ball release, which leads to a higher flight time style. This result is seen in the correlation between ball velocity and the releasing of ball during the acceleration phase. During the acceleration phase, the maximal angular velocities of the elbow extension, wrist flexion, and the internal rotation of the shoulder joint make a substantial contribution to overarm throwing in team handball [23].

The left hip and right hand had moderate correlations with ball velocity at Max-GRF and with IF events during the cocking phase. During the follow-through phase at GC, a strong correlation was observed between the angle of the left hip and ball velocity. These negative correlations with ball velocity during the cocking phase indicated that handball players were using a greater angle of projection to reach for optimum position before releasing the ball and generate a required power with the throwing arm. Thus, higher velocities of the arm joints could be produced [23], which could result in higher external rotation while throwing. This finding confirmed that the movement of the large body segments such as legs and trunk assist with the reduction of energy in the throwing arm in order to reduce the loads, especially on the shoulder and elbow [25,27,28].

The ball velocity had a moderate correlation with the velocity of the left knee at GC during the follow-through phase and the left ankle angle at the IC and Max-GRF events during the cocking phase (Table 2). The negative correlation between left knee velocity with ball velocity implies that players had a greater knee flexion in the follow-through phase for preparation for landing. Olsen, et al. [29] and Koga, et al. [30] reported that landing after a jump throwing with the knee near full extension was important to prevent ACL injuries in team handball players.

It should be noted that these findings were derived from a small sample of homogenous elite handball players. We selected elite players because we wanted data from high-level performance. A small sample was used in this pilot study in order to evaluate the utility of this approach. Finding strong and moderate correlations in a small, homogenous sample suggests that these findings are very conservative. We hope that future investigations will expand on these preliminary findings.

A limitation was that the sample size in the present study was small (ten right-handed male handball players), and we analyzed the best attempt of the five successful performed attempts. All our participants were elite, meaning that our sample was homogenous, making correlations more challenging to identify. We were willing to accept these limitations in this pilot study because our goal was to identify the correlations for top performers, since it was assumed that these participants represent the best performances. However, future investigations should recruit a larger number of participants and/or increase the number of attempts.

#### **5. Conclusions**

Based on the main findings from this study, the velocity of the right hand at IC and IF events (anterior–posterior direction) correlated with ball velocity during the cocking phase as a preparation phase that accelerates the arm throwing. The increases in right-hand acceleration at Max-GRF, Max-HH events (anterior-posterior direction) led to the improvement of the horizontal velocity of the right hand (elbow and wrist joint) before the RB event achieved the highest ball velocity. Furthermore, the results indicated that the change in angular variables of shoulder joint are greater than those of the elbow. Thus, we recommend that, in training programs, coaches focus on increasing their player's movement horizontally to enable forward movement at the landing when evaluating the performance level of handball players.

Also, we recommend that coaches to use special power and reaction time exercises to reduce the dominant foot contact time at the jump moment and before throwing the ball. This would allow the player to jump higher before the ball release event and keep the maximum velocity of the ball while throwing performance.

**Author Contributions:** Conceptualization, A.-R.A. and I.H.; methodology, A.-R.A., I.H. and A.H.; software, A.-R.A.; validation, A.-R.A., I.H., P.B. and A.H.; formal analysis, A.-R.A.; investigation, A.-R.A. and I.H.; resources, A.-R.A.; data curation, A.-R.A. writing—original draft preparation, A.-R.A., I.H. and A.H.; writing—review and editing, P.B.; visualization, A.H.; supervision, P.B.; project administration, A.-R.A.

**Funding:** This research received no external funding.

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

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Morphology-Related Foot Function Analysis: Implications for Jumping and Running**

**Peimin Yu 1, Liangliang Xiang 1, Minjun Liang 1, Qichang Mei 2, Julien S. Baker <sup>3</sup> and Yaodong Gu 1,2,\***


Received: 14 July 2019; Accepted: 5 August 2019; Published: 8 August 2019

**Abstract:** Barefoot and shod running has received increased attention in recent years, however, the influence of morphology-related foot function has not been explored. This study aimed to investigate morphology-related jumping and running biomechanical functions in habitually barefoot and shod males. A total of 90 barefoot males (Indians) and 130 shod males (Chinese), with significant forefoot and toe morphology differences, participated in a vertical jump and running test to enable the collection of kinematic and kinetic data. The difference of pressure distribution in the hallux and forefoot was shown while jumping and running. The unrestricted forefoot and toes of the barefoot group presented flexible movement and leverage functions to expand the forefoot loading area during performance of the two tasks. Findings related to morphology functions, especially in the forefoot and toe may provide useful information for footwear design.

**Keywords:** foot morphology; toes function; biomechanics; barefoot; jumping; running

#### **1. Introduction**

Human feet are the basic terminal structures that support human walking, running, jumping, and other locomotion. The foot is a complex structure that controls balance and movement [1,2]. Foot morphology has been studied since the early 20th century [3]. Previous studies have demonstrated that the foot differs significantly between habitually barefoot and shod people [4–6], and differences in the kinetics of walking, running, and jumping have been observed [7–9].

Different foot morphology may also be a contributory factor for injury during motion [10], and may also influence physical activity performance [11,12]. There are many reasons for morphological differences in humans, which include disease, foot malfunctions, genetics, and deformity [13]. Research findings have indicated that external factors, such as footwear, may deform foot structure, and result in conditions such as hallux valgus (HV) [3,14]. HV could induce foot dysfunction [15], influence foot morphology [16], and may impair quality of life [17], which may result in depression and pain [18].

In addition, when compared to habitually shod populations, habitually barefoot populations demonstrate more toe separation [3,4,14]. Studies on foot morphology have focused on the width and length of the foot [6], and several studies have investigated the morphological differences between the hallux and other toes [4,14]. However, whether these differences influence the motions needed for physical activity is unclear.

Jumping is a typical movement in many sports, and has attracted much attention from the research community [19,20]. Jumping performance has been evaluated using a one-foot and a two-feet jump [19], and toe flexor function has also been examined [20]. Furthermore, the countermovement jump has been important to support clinicians in the medical diagnosis of muscle power during prolonged recovery periods following ankle injuries [21]. The contribution of the forefoot and toes has been evaluated while performing the vertical jump, and kinematics, kinetics, and spatiotemporal parameters have been recorded and analyzed [22].

Lieberman et al. [23] indicated that habitually barefoot populations and shod populations present different foot strike patterns. Habitually barefoot populations land on the forefoot, then bring down the heel, and have been observed landing with a flat foot, but seldom on the heel. Habitually shod populations mostly land with a rearfoot strike. The elevated and cushioned heel of the modern running shoe may be a contributory factor that has facilitated the differences in the strike patterns observed. However, strike patterns have been observed to be variant, even between shod or barefoot populations, in recent studies [2,7,8,24]. In spite of the conflicting opinions about barefoot locomotion, it has gained in popularity in recent years, and is now included in athletic training [25], recreational running [26], and rehabilitation [27]. A previous study has revealed the foot shape and function differences in native barefoot walkers [5] and runners [24]. The morphological differences between habitually barefoot and shod runners were found to exist in the forefoot and toe regions [4]. However, morphology based on the functions of the forefoot and toes while performing vertical jumping and running has not been investigated.

Therefore, the purpose of this study was to examine morphology-related performance differences while conducting vertical jumping and running tasks between habitually barefoot males and shod males. A further aim was to explore any functional differences in the forefoot and toes, based on foot morphological characteristics. It was hypothesized that the lower extremity kinematics and plantar forefoot loading distribution would be different due to the morphological differences in the forefoot and toes region.

#### **2. Materials and Methods**

#### *2.1. Participants*

The sample size was calculated prior to this study using the power package in R-3.6.1 (effect size = 0.5, α level = 0.05, power value = 0.9, type: two-sample, alternative: two sided). A total of 90 barefoot males (Indians) and 130 shod males (Chinese), who presented significant forefoot and toe morphology differences in a previously published study [4], volunteered to participate in the vertical jumping and running test to enable collection of kinematic and kinetic data. All participants were students in the University and had a history of running or other physical activities. Participants of Indian ethnicity originated from South India (Kerala state), were running or taking part in physical activities barefoot since birth, and wore slippers during daily life. Participants of Chinese ethnicity were shod runners since birth and wore different kinds of shoes in daily life. Participants with hallux valgus, high-arched foot, flat foot, diabetic foot, or any other foot deformities were excluded via foot scan prior the test. None of the participants had sustained injuries or surgeries to their lower limbs in the previous half year.

Data for 62 barefoot males (age: 22 ± 1.9 years; weight: 65 ± 8.6 kg; height: 1.69 ± 0.16 m), presenting with a forefoot strike during running, and 112 shod males (age: 23 ± 2.8 years; weight: 66 ± 7.8 kg; height: 1.71 ± 0.11 m), presenting with a rearfoot strike during running, were included for analysis via post data procession. This study, with detailed guidelines for participants' safety and experimental protocols, was approved by the Human Ethics Committee at the Research Institute of Ningbo University ARGH20160819. The study was conducted in accordance with the declaration of Helsinki. Prior to the test, all subjects gave informed consent, with full knowledge of test procedures and requirements.

#### *2.2. Experiment Protocol*

The test protocol was consistent with a previously reported experiment [23], which was published from our laboratory recently [1,24]. After completion of foot scanning, participants revisited the motion capture lab for experimental vertical jumping and running tests. Participants were instructed to warm up and to familiarize themselves with the lab environment for 5 min prior to data collection. Before data collection, three familiarization trials were performed for each task.

While performing the vertical jump, participants stood on the ground in an akimbo position (right foot on the force platform) to reduce the interference from the upper body during performance of a maximal vertical jump. Each participant completed six trials with the right foot on the force platform (Model 9281B, Winterthur, Switzerland).

Running tests were conducted on a runway in the lab. Subjects performed barefoot running with the right foot striking the force platform, which was located in the middle of the runway and was used for kinetic data collection. The force platform and pressure data were used to assist in the definition of striking patterns following a previously established protocol [28,29]. Each participant performed six trials of running using a self-selected running speed, to present natural strike patterns during running and a collection of biomechanical characteristics. For both jumping and running sessions, there were 30 s rest intervals between each trial to minimize the effect of fatigue.

The pressure platform (Novel EMED System, Munich, Germany) was reported to have high reliability correlations (>0.7) [30], and the insole (Novel Pedar System, Munich, Germany) plantar pressure distribution system also displayed excellent reliability correlations (>0.9) [31]. The pressure plate was used to record barefoot jumping and running plantar pressure distribution data with a frequency of 100 Hz. The in-shoe plantar pressure measurement system was placed in the shoes for collection of the shod jumping and running plantar pressure distribution data among habitually shod males, with a frequency of 100 Hz. The habitually shod males (shod) performed shod running wearing shoes that were the same brand and model, for consistency.

#### *2.3. Data Acquisition*

Previous studies have outlined data collected from insole pressure sensors and pressure plates and show high reliability [32]. All the anatomical region division analysis was performed in the Novel Database in the data post-procession based on an auto-masking algorithm [33]. For trials of barefoot and shod vertical jumping, only the data in the forefoot and toes were included. The collected plantar pressure data while performing vertical jumping were separated into the push-off and landing phases for analysis. Thus, the plantar surface was divided into five anatomical regions: medial forefoot (MF), central forefoot (CF), lateral forefoot (LF), hallux (H), and other toes (OT), as this study mainly focused on the instant push-off and landing phase of the vertical jump. For trials using barefoot and shod running, the plantar surface was divided into eight anatomical regions, including medial rearfoot (MR), lateral rearfoot (LR), medial midfoot (MM), lateral midfoot (LM), medial forefoot (MF), lateral forefoot (LF), hallux (H), and other toes (OT). The variables for jumping and running included peak pressure, contact area, and the pressure–time integral of each anatomical region.

The kinematic test used the eight-camera Vicon motion analysis system (Oxford Metric Ltd., Oxford, UK) to collect the lower extremity kinematic data, with a frequency of 200 Hz. Sixteen reflective points (diameter: 14 mm) were attached with adhesive tape on the lower limbs of subjects, following a previously published protocol [34]. The anatomical landmarks included the anterior–superior iliac spine, posterior–superior iliac spine, lateral mid-thigh, lateral knee, lateral mid-shank, lateral malleolus, second metatarsal head, and calcaneus. A Kistler Force Platform (Model 9281B, Winterthur, Switzerland) was used to record ground reaction forces (GRFs), with a frequency set at 1000 Hz, to define the running foot striking patterns and contact time. The force platform was zero-levelled prior to testing each participant. The on and off force platform was defined from the value of vertical GRF as 20 N. Participants were required to strike the force platform with the right foot while performing the running and jumping tests on the force platform. The variables of running included spatiotemporal parameters, such as stride length, stride time and contact time, peak angles during stance, and joints range of motion (ROM) in a gait cycle. The spatiotemporal parameters were generated from the Workstation in the Vicon Nexus software (v1.8.5), including hip, knee, and ankle angles in the

sagittal plane, coronal plane, and horizonal plane, computed from the Vicon Plug-in-Gait Model using established protocols [20,30]. Vertical jump height was calculated by Equation (1) [35]:

$$\text{Jumpleigh}(\text{m}) = \frac{9.80 \text{m} \cdot \text{s}^{-2} \times \text{flighttime} (\text{s})^2}{8}. \tag{1}$$

#### *2.4. Statistical Analysis*

Normal distribution was checked for all variables, including jump height, peak pressure, pressure time integral, and contact area of vertical jumping, and running spatiotemporal parameters, such as stride length, stride time and contact time, running peak angles during stance, and joints range of motion in a gait cycle. Independent-sample *T* tests were used to analyze the significance of kinematic, plantar loading, and spatiotemporal variables between the barefoot and shod group. SPSS 18.0 (SPSS Inc., Chicago, IL, USA) software was used for the analysis, with statistical significance set at *p* < 0.05.

#### **3. Results**

After calculation and comparison of jump height, there were no significant differences between the height of the barefoot jump (386.4 ± 13.6 mm) and shod jump (408.2 ± 12.9 mm), with *p* > 0.05.

As shown in Figure 1, during the take-off phase (left), significant differences (*p* < 0.05) were found between barefoot and shod jumping in H (*p* = 0.02 and 0.01), MF (*p* = 0.018 and 0.029), and CF (*p* = 0.026 and 0.03) for peak pressure and the pressure–time integral. Significance for the contact area was also found in H (*p* = 0.032).

**Figure 1.** The peak pressure, pressure time integral, and contact area in the anatomical regions during the push-off (**a**) and landing (**b**) phases of the vertical jump. lateral forefoot (LF), central forefoot (CF), medial forefoot (MF), hallux (H), and other toes (OT). \* indicates significance between variables, *p* < 0.05.

During the landing phase (right), significant differences were found between barefoot and shod jumping in H (*p* = 0.016 and 0.021), MF (*p* = 0.026 and 0.031), and CF (*p* = 0.04 and 0.033) for peak pressure and the pressure time integral. For contact area, significant differences were found in H (*p* = 0.034) and CF (*p* = 0.02).

As measured from the running test, participants' running speeds were self-selected as comfortable from the generated spatiotemporal parameter. The comparison of collected spatiotemporal parameters, including stride length, stride time, and contact time, in one gait cycle between barefoot and shod running, are presented in Table 1.

**Table 1.** The spatiotemporal parameters between barefoot and shod running.


Note: \* Significance between barefoot and shod runners, *p* < 0.05.

As shown in Figure 2, the stances of barefoot and shod running were highlighted with solid (33.2 ± 0.7%) and dashed (37.5 ± 0.8%) vertical lines, which were calculated from the percentage of contact time in stride time. The peak angles during the stance were thus obtained between barefoot and shod running, and statistical significance was highlighted with a red dotted line with an asterisk (**\***), with Table 2 presenting detailed values.

**Figure 2.** Joint angles curves of the ankle, knee, and hip in sagittal, frontal, and horizontal planes during one gait cycle. Red dotted lines with \* indicate a significant difference, *p* < 0.05.

There was a significant difference between the foot strike angle of the ankle between shod and barefoot running, with the foot strike angle of shod running at 17.1 ± 4.3◦, and barefoot running at −7.2 ± 3.9◦ (minus indicates plantarflexion), *p* = 0.00. Internal and external ankle rotation also showed a significant difference, *p* < 0.05. The maximal rotation angle during the push-off phase of the stance was 3.24 ± 2.26◦ (shod running) and −3.76 ± 1.5◦ (barefoot running). Barefoot running showed significantly larger ankle ROM than shod running, *p* = 0.00 (Table 3).

The knee joint contact angles while foot landing were 12.33 ± 8.45◦ (shod) and 0.1 ± 2.3◦ (barefoot), showing significance (*p* = 0.012) (highlighted in Figure 2). Smaller knee joint ROM in the sagittal plane was also observed, with *p* = 0.021 (Table 3).

For hip movement, shod running (38.79 ± 7.81◦) presented a larger flexion angle than barefoot running (30.12 ± 5.66◦) while landing (*p* = 0.03) (Table 2). A greater internal rotation angle was observed for barefoot running (27.21 ± 3.66◦) than shod running (14.21 ± 2.66◦), as the foot was landing (*p* = 0.32) (Figure 2). Shod running presented significantly larger ROM than barefoot running in the gait cycle (Table 3).

**Table 2.** Peak joints' angles between barefoot and shod running during stance.


\* Significance between barefoot and shod runners.

**Table 3.** Lower extremity joints' range of motion (ROM) between barefoot and shod running in gait cycle.


\* Significance between barefoot and shod runners.

Peak pressure, contact area, and the pressure–time integral are shown in Figure 3. For peak pressure, MR, LR, LM, LF, and H showed significant differences between shod and barefoot running. Specifically, barefoot running demonstrated less peak pressure in MR (*p* = 0.00) and LR (*p* = 0.00) than shod running. In contrast, barefoot running showed larger peak pressure in LM (*p* = 0.028), LF (*p* = 0.019), and H (*p* = 0.005) than shod running. For the pressure–time integral, shod running showed a larger pressure–time integral in MR (*p* = 0.00), LR (*p* = 0.00), and MF (*p* = 0.02) than barefoot running. In contrast, barefoot running indicated a larger pressure–time integral in LM (*p* = 0.03), LF (*p* = 0.009), and H (*p* = 0.028) than shod running. For the contact area, shod running presented a larger area in MR (*p* = 0.00), LR (*p* = 0.00), and MM (*p* = 0.00) than barefoot running.

**Figure 3.** Foot pressure of barefoot and shod running. Medial rearfoot (MR), lateral rearfoot (LR), medial midfoot (MM), lateral midfoot (LM), medial forefoot (MF), lateral forefoot (LF), hallux (H), and other toes (OT). \* indicates significance, *p* < 0.05.

#### **4. Discussion**

This study aimed to analyze foot morphology-related jumping and running biomechanics and evaluate any potential functional differences. Participants in this study were from different parts of Asia, with a barefoot group from Indian ethnicity and a shod group from Chinese ethnicity. The main findings were that, (i) during the push-off and landing phases of the vertical jump, the separate hallux of barefoot individuals shared loading from the metatarsals, and thus expanded the loading concentrated region; (ii) during the push-off phase of running, there were plantar pressure differences in the hallux and forefoot of barefoot individuals compared with shod individuals; (iii) barefoot individuals with separate toes presented a flexible range of motion, particularly in the coronal plane of the ankle, sagittal plane of the knee, and horizontal plane of the hip.

Hallux angle has been reported to be different among populations of different ethnicities [11]. However, few studies have focused on the minimal distance between the hallux and the interphalangeal joint of the second toe. Compared with results of our previous study [4], barefoot groups in this study had a larger distance and smaller hallux angle, while the shod group had a larger hallux angle and smaller distance. It may be concluded that the barefoot group had more flexible hallux than the

shod group [1,3,36]. Lambrinudi et al. [36] reported that if the separate hallux has ambulatory and prehensile functions, it could work fundamentally the same way as the fingers. However, wearing shoes may block these prehensile and separate functions of the toes, due to the sharp-headed or ill-fitted space restrictions [3,6,14]. The hallux angle and minimal distance are the basis of the morphological differences for the vertical jumping and running test in this study.

Results from the vertical jump test indicated that the hallux presented larger plantar loading in the barefoot group compared with the shod group (Figure 1). During the push-off phase, the plantar loading of the barefoot group was larger under the hallux, while the pressure of the shod group was larger under the medial forefoot and central forefoot. The same pressure time integrals were presented in these anatomical regions, which may imply that the hallux of the barefoot group was used predominantly, while the forefoot of shod group was used primarily. Moreover, the peak pressure of the barefoot reduced in the forefoot regions while landing. This suggests that the gripping function of the hallux could firm and expand the supporting base during the push-off and landing phases by the separate toes [3,36]. In addition, large loading under the hallux could reduce the impact force to the forefoot [24]. Previous research has reported that excessive loading under the metatarsal head area (forefoot) would lead to forefoot injuries [7]. These findings imply that the foot morphology related to toe gripping functions may have a possible link with forefoot metatarsal stress injuries. However, this study did not investigate the injury risks between the two population groups. The jumping height showed no significance, which implies that the morphological differences may not be linked with jumping performance, or there may be a limitation in the akimbo position. Further research is needed and should focus on jumping performance via comprehensive kinematic analysis.

Research pertaining to habitually barefoot and shod people has received increased attention in recent years. Different ethnicities [4,5,24], pathological factors [37], and different forms of sport participation [10] could influence foot morphological differences. Among all the barefoot and shod participants, biomechanical data for the forefoot strike barefoot running, and rearfoot strike shod running were included in this study. Barefoot running was reported to be different to minimalist, racing, or regular shoe conditions in a previous biomechanical study of experienced runners [38]. The extrinsic muscles of the foot presented reduced muscular activity during barefoot running, for instance, peroneus longus [39,40]. Other controversial opinions were proposed between the minimalist and barefoot running when compared to traditional shoe running [41,42]. This study focused on the biomechanics from the forefoot and toe morphology, thus, shod running from the barefoot group and barefoot running from the shod group were not performed during the test, which was aimed to reduce the acute response of altering to shod (for the barefoot group) or barefoot (for the shod group) running.

During the running test, each participant performed running at a comfortable speed so as to present natural running biomechanics [7,43]. The results indicated significant differences in strike length, strike time, and contact time between the shod and barefoot groups, which are consistent with previous studies [2,7,8]. In previous barefoot running studies, the barefoot group was observed to reduce these spatiotemporal parameters [8,44,45]. The stride time of the barefoot group was significantly less than the shod [46]. The running performance of the barefoot group was characterized by landing on the forefoot, and the ankle changed from plantarflexion to dorsiflexion in the sagittal plane, which contributed to the greater dorsiflexion angle during the stance. The shod running resulted in landing with the rearfoot, and the ankle in the dorsiflexion position, which could explain the ankle angle difference as the foot strikes. Different foot strike patterns could be a reason for the strike time differences observed [8].

The observed knee contact flexion angle and peak flexion angle difference of shod running in this study may be a compensatory movement (with larger sagittal knee ROM), resulting from the previously established greater knee impact while rearfoot shod running [9,45,47]. As shown by the hip flexion angle, the contact flexion angle was larger than that of barefoot running, and this could explain the increased stride length of shod running and, although not significant, an about hip flexion-extension ROM was observed.

In terms of the plantar pressure distribution, the barefoot group showed smaller peak pressure and pressure–time integral than the shod group in the rearfoot. This may have been caused by the rear foot landing during shod running, which results in a larger contact area in MR and LR. The difference in contact area in MM may be related to the uppers and soles of the footwear while shod running. Owing to the forefoot strike of barefoot running, the larger peak pressure and pressure time integral to the LM and LF may result from the landing impact. This finding could explain the previously reported forefoot metatarsus fatigue injury due to the repetitive impact and lack of cushioning protection from footwear [1,24]. The hallux showed an increased peak pressure and pressure–time integral during barefoot running, and, while not significant, this was also observed for the contact area. This may result from the active gripping motion of the separate hallux "leverage function" expanding the push-off supporting area (fulcrum) [1,36]. Thus, less loading was found in the MF compared with shod running, which presented greater MF loading and smaller H loading. The greater ankle ROM in the coronal plane and peak angle while pushing off may be kinematic evidence for the active toes function related to the morphological differences in this study. As reported, the function of the remaining toes may also be used for balance and stability control under static and dynamic conditions [48], and it may also be useful during running and jumping performances. Further benefits from the toes in relation to balance and coordination, especially contributions to long distance endurance racing and related events [49], needs further investigation.

Several limitations should be considered in this study. Firstly, participants were physically active males in their early twenties, which may be a limiting factor for generalizing findings from this study to different age groups and genders. Secondly, this study lacked information on the vertical jump biomechanics between the two-population groups, which should be a future research project to investigate potential differences in jumping performance. Thirdly, the entire test was conducted in a lab-based environment, and it is possible the jumping and running biomechanical performances may be different in an outdoor environment.

Previous research revealed that running injuries are a multifactorial issue, including systemic factors, training (experience), health factors, and lifestyle factors [50]. The foot type, structure, or morphology were considered for musculoskeletal injuries in several studies [10,51–53]; however, morphology-related foot functions have been rarely investigated. This needs investigation as a potential contributory factor for injury research.

#### **5. Conclusions**

This study analyzed the morphology-related jumping and running biomechanical functions of habitually barefoot and shod males. The unrestricted forefoot and toes of the barefoot group presented flexible movements and a leverage function to expand the forefoot loading area during jumping and running. Findings from the study in relation to morphology-related functions, especially the contribution of the forefoot and toes, may provide useful information for footwear design and injury prevention.

**Author Contributions:** Conceptualization, P.Y., M.L., Q.M. and Y.G.; methodology, P.Y., L.X., Q.M., J.S.B. and Y.G..; validation, L.X., M.L. and Q.M.; investigation, P.Y. and L.X.; writing—original draft preparation, P.Y., L.X. and Q.M.; writing—review and editing, J.B. and Y.G.

**Funding:** This research was funded National Natural Science Foundation of China (No.81772423), National Key R&D Program of China (2018YFF0300903) and K.C. Wong Magna Fund in Ningbo University.

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

#### **References**

1. Mei, Q.; Fernandez, J.; Hume, P.; Gu, Y. Investigating biomechanical function of toes through external manipulation integrating analysis. *Acta Bioeng. Biomech.* **2016**, *18*, 87–92. [CrossRef]


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Applied Sciences* Editorial Office E-mail: applsci@mdpi.com www.mdpi.com/journal/applsci

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18