Balance Training Programs in Athletes - a Systematic Review.

It has become almost routine practice to incorporate balance exercises into training programs for athletes from different sports. However, the type of training that is most efficient remains unclear, as well as the frequency, intensity and duration of the exercise that would be most beneficial have not yet been determined. The following review is based on papers that were found through computerized searches of PubMed and SportDiscus from 2000 to 2016. Articles related to balance training, testing, and injury prevention in young healthy athletes were considered. Based on a Boolean search strategy the independent researchers performed a literature review. A total of 2395 articles were evaluated, yet only 50 studies met the inclusion criteria. In most of the reviewed articles, balance training has proven to be an effective tool for the improvement of postural control. It is difficult to establish one model of training that would be appropriate for each sport discipline, including its characteristics and demands. The main aim of this review was to identify a training protocol based on most commonly used interventions that led to improvements in balance. Our choice was specifically established on the assessment of the effects of balance training on postural control and injury prevention as well as balance training methods. The analyses including papers in which training protocols demonstrated positive effects on balance performance suggest that an efficient training protocol should last for 8 weeks, with a frequency of two training sessions per week, and a single training session of 45 min. This standard was established based on 36 reviewed studies.

Introduction

It has become almost routine practice to incorporate balance exercises into training programs for athletes from different sports, fall prevention programs for the elderly and rehabilitation programs. The objectives and benefits seem obvious, e.g., performance improvement and injury prevention as commonly cited goals (Hrysomallis 2011; Kümmel et al., 2016; Lesinski et al., 2015). However, the type of training that is most efficient still remains unclear, and the frequency, intensity and duration of exercise that would be most beneficial have not yet been determined. The main goal of this review was to establish whether a gold standard of balance training exists in this field.

Posture and balance control are fundamental in daily life to safely accomplish any type of movement and motor task that involves displacement of body segments or the entire body. Balance is the process of maintaining the body’s center of gravity (CoG) vertically over the base of the support, and it relies on rapid and continuous feedback from visual, vestibular and somatosensory structures for the subsequent execution of smooth and coordinated neuromuscular actions (Winter, 1995; Zatsiorsky and Duarte, 1999). Efficient postural balance not only reduces the risk of body imbalance, fall, or subsequent injuries, but also contributes to the optimization of motor performance in a number of athletic disciplines (Hrysomallis, 2007; McGuine et al., 2000; Watson, 1999).

Each sport involves specific motor skills that require the completion of particular postures and movements (Hrysomallis et al., 2006; Maurer et al., 2006; Paillard, 2017). Balance is an important factor in many athletic skills, but the relationship between sports competition results and balance is not yet fully understood (Adlerton et al., 2003; Hrysomallis, 2011). A lower level of balance is associated with injuries, such as sprains and, muscle, tendon and ligament strains among others (McGuine et al., 2000; Emery and Meeuwisse, 2010; Eils et al., 2010). Maintaining a standing posture on a stable surface is a major determinant of balance. A sway analysis in a simple task, such as quiet standing, is used as a variable of its description (Visser et al., 2008; Duarte and Freitas, 2010). However, controversy exists in the literature regarding the influence of balance training on athletes’ performance and balance improvement, as well as injury prevention.

Literature search

The following review is based on papers that were found through computerized searches of PubMed and SportDiscus from 2000 to 2016. There is no general consensus in the literature regarding what to call training programs and exercise, therefore, we searched for various terms of training programmes. Based on a Boolean search strategy, consistent with previous meta-analyses on the effects of balance training (Kümmel et al., 2016; Lesinski et al., 2015; Zemková 2014), the following search terms were (individually or in various combinations) used: “balance training” OR “proprioceptive training” OR “core stability training” OR “injury prevention”, OR “postural control” AND “injury prevention” AND “sport” OR “athletes” OR “basketball” OR “baseball” OR “volleyball” OR “football” OR “soccer” OR “handball” OR “tennis” OR “ski” OR “runners” OR “judo” OR “taekwondo” OR “capoeira” OR “figure skating” OR “bicycling”. The search was limited to English language and full-text original articles.

Study selection

Only the studies that met the following criteria were included: (1) the participants of an intervention and a control group had to be healthy at the time of the study, (2) the study subjects were in an age range of 7-30 years old, (3) balance tests were performed before and after the intervention programs. Studies were excluded if (1) they did not meet the criteria for CTs (Control Trials), (2) the PEDro scale was lower than four (Table 3), or (3) balance training was not described in detail. We made an exception for the four papers (PEDro 3) because of the better quality description of training protocols. The reviewers conducted the literature review independently, based on inclusion and exclusion criteria. In total, 50 studies met the inclusion criteria for review (Figure 1).

Table 3

Physiotherapy evidence database (PEDro) scores of the reviewed studies-

References	Eligibility criteria specified	Subjects randomly allocated to groups	Allocation concealed	Groups similar at baseline	Blinding of all subjects	Blinding of all therapists	Blinding assessors	Dropout < 15%	Intention-to-treat method	Statistical comparison between	measures and measures	Score
Benis et al.. (2016)	+	+	-	+	+	-	-	+	-	+	+	7
Gioftsidou et al.. (2012)	+	+	-	-	-	-	-	+	+	+	+	6
Malliou et al.. (2004)	+	+	-	-	-	-	-	+	+	+	+	6
Mahieu et al.. (2006)	+	+	-	+	-	-	-	+	+	+	+	7
Zech et al.. (2014)	+	+	+	+	-	-	-	+	+	+	+	8
Pau et al. (2012)	+	-	-	+	-	-	-	-	+	+	+	5
Daneshjoo et al.. (2012)	+	+	-	+	-	-	-	+	+	+	+	7
Heleno et al.. (2016)	+	+	-	+	-	-	+	+	+	+	+	8
Saunders et al. (2013)	+	+	-	-	-	-	-	+	+	+	+	6
Gioftsidou et al.. (2006)	+	+	-	+	-	-	-	-	+	+	+	6
Sato and Mokha (2009)	+	+	-	+	-	-	-	-	-	+	-	4
Matin et al.. (2014)	-	+	-	+	-	-	-	+	+	+	-	5
Romero-franco et al.. (2012)	-	+	-	+	-	-	-	+	+	+	+	6
Myer et al.. (2006)	+	+	-	-	-	-	-	-	+	+	+	5
Lust et al.. (2009)	+	+	-	-	-	-	-	+	+	+	+	6
Dobrijevic et al.. (2016)	+	+	-	-	-	-	-	+	+	+	+	6
Pfile et al.. (2016)	+	-	-	-	-	-	-	+	-	+	+	4
Hammami et al. (2016)	+	-	-	-	+	-	-	+	+	-	+	5
Emery and Meeuwisse (2012)	+	+	-	+	-	-	-	+	+	+	+	7
Verhagen et al.. (2004)	+	+	+	+	-	-	+	+	+	+	+	9
Petersen et al.. (2005)	-	-	-	-	-	-	-	+	+	+	-	3
Imai et al.. (2014)	+	+	-	+	-	-	-	-	+	+	+	6
O’Malley et al.. (2016)	+	+	+	+	-	-	-	+	+	+	+	8
Trecroci et al.. (2015)	+	+	+	+	-	-	-	+	-	+	+	7
Steib et al.. (2014)	+	+	-	+	-	-	-	+	-	+	+	6
Valovich et al.. (2009)	+	-	-	+	-	-	+	-	-	+	+	5
Eisen et al.. (2010)	+	+	-	+	-	-	-	-	-	+	+	5
Holm et al.. (2004)	+	-	-	-	-	-	-	+	-	+	+	4
Asadi et al.. (2015)	+	+	-	+	-	-	-	+	+	+	+	7
Verhagen et al.. (2005)	-	+	-	-	-	-	-	+	-	+	+	4
Kachanathu et al.. (2014)	+	+	-	+	-	-	-	-	-	+	+	5
Cankaya et al.. (2015)	-	+	-	-	-	-	-	-	-	+	+	3
Karadenizli (2016)	+	-	-	-	-	-	-	+	-	-	+	3
Ahmadabadi et al.. (2015)	+	+	-	+	-	-	-	+	-	+	-	5
Iacono et al.. (2014)	+	+	-	+	-	-	-	+	-	+	+	6
Karadenzili (2016)	+	+	-	+	-	-	-	+	-	+	+	6
Eils et al.. (2010)	+	+	-	+	-	-	-	+	+	+	+	7
Soligard et al..(2008)	+	+	-	+	-	-	-	+	+	+	+	7
Kraemer and Knobloch (2009)	-	-	-	-	-	-	-	+	-	+	+	3
Owen et al.. (2013)	-	+	-	+	-	-	-	+	+	+	-	5
McGuine and Keene (2006)	+	+	-	+	-	-	-	+	+	+	+	7
McHugh et al.. (2007)	+	-	-	-	-	-	-	+	-	+	+	4
Steffen et al.. (2008)	+	+	+	+	-	-	-	+	+	+	+	8
Cumps et al.. (2007)	-	-	-	+	-	-	-	+	-	+	+	4
Mandelbaum et al.. (2005)	-	-	-	-	-	-	-	+	+	+	+	4
Soderman et al.. (2000)	+	+	-	+	-	-	-	-	-	+	+	5

“+” indicates a “YES” score; “-” indicates a “NO” score

Figure 1

A flowchart illustrating the different phases of the search and study selection

Balance training

Training methods

No general agreement may be found in the literature regarding which terms should be used to summarize training programs that aim at the improvement of postural stability (Kümmel, Kramer, Giboin, Gruber, et al., 2016). Some authors (Verhagen et al., 2005; Cumps et al., 2007; Kachanathu et al., 2014; Hammami et al., 2016) described balance or core stability exercises in their training programs. Others (Benis et al., 2016; Hammami et al.; Malliou et al., 2004; 2016; Pau et al., 2012; Verhagen et al., 2002; Zech et al., 2010) described neuromuscular or proprioceptive training and included multi-intervention programs with a combination of balance, strength, plyometric, and sport-specific exercises. Some authors describe the implemented exercises as balance training (Verhagen et al., 2005; Gioftsidou et al., 2006), and others call it sensorimotor training (Heleno et al., 2016; Pauet al., 2011), neuromuscular training (Zech et al., 2014; Benis et al., 2016) or proprioceptive training (Eils et al., 2010; Malliou et al., 2004; Mandelbaum et al., 2005). However, the most common term used seems to be balance training. Therefore, as in other systematic reviews (Kümmel et al., 2016), we use the term “balance training” to describe any training program primarily directed at the improvement of postural stability, regardless of the term used in the studies. Each of the training program described above presents a large variety of exercises. The balance training interventions consisted of balance exercises on both a stable and unstable surface, with or without recurrent destabilization during performance (Cumps et al., 2007; Hübscher et al., 2010; McHugh et al., 2007; Soderman et al., 2000; Verhagen et al., 2002, 2005; Zech et al., 2010). In some studies, training programs also included exercises with visual feedback (Malliou et al., 2004).

Frequently, studies that examined neuromuscular or proprioceptive training interventions similar to balance training included balance exercises on stable and unstable platforms with or without perturbations of postural control (Hübscher et al., 2010; Zech et al., 2010). Some authors also described neuromuscular training as multi-intervention programs with a combination of balance, weight, plyometric and sport-specific agility drills to address all aspects of neuromuscular control (Holm et al., 2004; Hübscher et al., 2010; Mandelbaum et al., 2005; Myer et al., 2009). In some papers, the authors implemented plyometric training alone to improve balance or combined it with balance exercises (Asadi et al., 2015; Manolopoulos et al., 2015; Myer et al., 2006; Pfile et al., 2013, 2016).

Balance assessment

To assess static and dynamic balance, some researchers used clinical and laboratory tests. Balance tests were performed before and after an intervention program. In some reports, also strength, aerobic endurance and specific performance were assessed (Hammami et al., 2016; Imai et al., 2014; Kang et al., 2013; Manolopoulos et al., 2015; Myer et al., 2006).

Static balance was evaluated using simple tests such us the stork test (Daneshjoo et al., 2012; Hammami et al., 2016) or the single leg stance (SLS) test (Dobrijević et al., 2016; Kang et al., 2013; Karami et al., 2014). These tests require the participants to keep their hands on the hips and maintain the foot of their non-tested leg at the knee level with their eyes open or closed. The participants attempted the task a few times, and the best scores were recorded for further analysis. More sophisticated procedures were performed on a force plate (FP) which can monitor the movement of the center of pressure (COP). Different variables derived from the path of the COP during the single leg stance test (Ahmadabadi et al., 2015; Malliou et al., 2004; Saunders et al., 2013), the quiet standing (QS) test (Cankaya et al., 2015; Pau et al., 2011; Steib et al., 2016) or the limit of stability (LOS) test (Mahieu et al., 2006; Romero-Franco et al., 2012) have been used as measures of balance. A balance assessment can also be conducted on an unstable surface. One example is a kinesthetic ability trainer (KAT) (Holm et al., 2004). The KAT consists of an electronic moveable platform that is supported by a small pivot at its central point. The stability of the platform is controlled by pressure that varies in a circular pneumatic bladder between the platform and the base of the unit. High pressure indicates an inflated platform (stable), while low pressure a deflated platform (unstable). An unstable surface makes the balance test more dynamic and possibly more applicable in a sports context.

Dynamic balance was assessed by the Balance Error Scoring System (BESS) (Imai et al., 2014; Mcleod et al., 2009), the Star Excursion Balance Test (SEBT) (Eisen et al., 2010; Filipa et al., 2012; Sato and Mokha, 2009) and the Y-Balance Test (YBT) (Trecroci et al., 2015; Benis et al., 2016; Hammami et al., 2016). The BESS consists of 6 separate 20 s balance tests that the subjects perform in different stances and on different surfaces. The test comprises 3 stance conditions (double-leg, single-leg, and tandem stance) and 2 surfaces (firm and foam). All trials are performed with the eyes closed (Finnoff et al., 2009). Errors are recorded as the quantitative measurement of postural stability under different testing conditions. Another test, originally described by Gray (Gray, 1995) as a rehabilitation tool, the SEBT, is a series of single-limb squats using the non-stance limb to perform maximal reach in order to touch a point along 1 of 8 designated lines on the ground. The lines are arranged in a grid that extends from a center point and are 45° from one another. The reach distances are normalized to leg length. The YBT, was inspired by clinical applications of the SEBT (Coughlan et al., 2012). The participants push the reach-indicator block with one foot in the anterior, posteromedial and posterolateral directions while standing on the other foot on a central footplate. Some researchers used the Modified Star Excursion Balance Test (MSEBT), where the subjects performed movements in the same directions as in the YBT (Zech et al., 2014; Heleno et al., 2016). Dynamic balance was also evaluated by a jumping test. For example, Heleno et al. (2016) conducted the Side Hop Test (SHT) with lateral jumps, and the Figure Eight Test (F8) using forward jumps with rotation. O’Malley et al. (2016) used the Landing Error Scoring System (LESS). The LESS identifies poor jump-landing techniques, such as decreased knee and hip flexion motion, knee valgus, and hip internal rotation, which can cause greater joint loading. Zech et al. (Zech et al., 2014) assessed the time to stabilization (TTS) following single-leg jump landing.

To assess proprioception and the stability of the upper and lower limbs, the isokinetic dynamometer (ID) was used to evaluate joint position sense (Daneshjoo et al., 2012).

Equipment and exercises

We found that there were numerous balance exercises that effectively improve static and dynamic balance. Training methods included exercises on stable and unstable surfaces in anterior/posterior and mediolateral directions, with or without recurrent destabilization (e.g., ball throwing or catching, strengthening exercises, or external perturbation applied by a partner) (Cumps et al., 2007; DiStefano et al., 2009; Hübscher et al., 2010; McHugh et al., 2007; Paillard, 2017; Soderman et al., 2000; Verhagen et al., 2002, 2005; Zech et al., 2010). The balance training programs typically included progression of the exercises. In some studies, balance exercises were performed first with the eyes open and then with the eyes closed in order to increase the difficulty (Hammami et al., 2016; Heleno et al., 2016; McGuine and Keene, 2006; Verhagen et al., 2005). Additionally, the balance training programs included transitions from a double-leg stance to a single-leg stance (Gioftsidou et al., 2006; O’Malley et al., 2016; Pau et al., 2011) on a stable or unstable surface (Eisen et al., 2010; Manolopoulos et al., 2015; Steib et al., 2016).

Occasionally, the authors also used exercises with visual feedback, such as moving a cursor to a target by shifting the weight (Malliou et al., 2004) or maintaining a single-leg stance on a board (Gioftsidou et al., 2006). For these types of exercises, the Biodex Stability System was used. In the studies, wobble boards that allow for multiplanar movement ( Eisen et al., 2010; Holm et al., 2004; Hrysomallis, 2007; Soderman et al., 2000), tilt boards permitting uniplanar movement (Dobrijević et al., 2016; Hrysomallis, 2007), BOSUs (Myer et al., 2006; Romero-Franco et al., 2012), foam mats (McHugh et al., 2007), inflated rubber discs (Saunders et al., 2013) and Swiss balls ( Kang et al., 2013; Sato and Mokha, 2009) were frequently used. These devices were used for different movements such as tilting, rotating, squatting, hopping, jumping, throwing and catching a ball (Eisen et al., 2010; Daneshjoo et al., 2012; Myer et al., 2006; Soligard et al., 2008). These activities were also combined with resistance exercises while balancing (Filipa et al., 2012; Petersen et al., 2005; Romero-Franco et al., 2012). In some papers, the authors implemented plyometric training alone to improve balance or combined plyometric training with balance exercises. These exercises emphasized jumping movements with feedback regarding technical performance and proper limb alignment (Asadi et al., 2015; Manolopoulos et al., 2015; Myer et al., 2006; Pfile et al., 2013, 2016). The plyometric exercises consisted of athletic positions, squat jumps, line jumps, bounding in place, and box drops, among others (Asadi et al., 2015; Myer et al., 2006). Core stability training was also used to improve balance. Some authors understood core exercises as bracing the abdominal muscles with low intensity limb movements (Kachanathu et al., 2014); however, most authors applied global training of larger superficial muscles around the abdominal and lumbar regions (Filipa et al., 2012; Iacono et al., 2014; Lust et al., 2009; Myer et al., 2006; Sato and Mokha, 2009). Core stability training included front planks, side planks, back bridges, quadruped exercises and exercises on a Swiss ball.

The influence of balance training on balance in various sport disciplines

The most widely studied disciplines were soccer (Cankaya et al., 2015; Daneshjoo et al., 2012; Gioftsidou et al., 2006; Imai et al., 2014), basketball (Asadi et al., 2015; Benis et al., 2016; Mcleod et al., 2009; Pfile et al., 2016) and handball (Holm et al., 2004; Karadenizli, 2016a, 2016b; Steib et al., 2016). The majority of the studies revealed significant differences between the groups after the interventions (Asadi et al., 2015; Daneshjoo et al., 2012; Kachanathu et al., 2014; Mcleod et al., 2009; O’Malley et al., 2016; Pfile et al., 2016; Steib et al., 2016). However, a few publications were found that did not show any significant influence of balance training on balance in various sport disciplines (Benis et al., 2016; Eisen et al., 2010; Sato and Mokha, 2009; Zech et al., 2014).

The most widely studied disciplines were soccer (Cankaya et al., 2015; Daneshjoo et al., 2012; Gioftsidou et al., 2006; Imai et al., 2014), basketball (Asadi et al., 2015; Benis et al., 2016; Mcleod et al., 2009; Pfile et al., 2016) and handball (Holm et al., 2004; Karadenizli, 2016a, 2016b; Steib et al., 2016). The majority of the studies revealed significant differences between the groups after the interventions (Asadi et al., 2015; Daneshjoo et al., 2012; Kachanathu et al., 2014; Mcleod et al., 2009; O’Malley et al., 2016; Pfile et al., 2016; Steib et al., 2016). However, a few publications were found that did not show any significant influence of balance training on balance in various sport disciplines (Benis et al., 2016; Eisen et al., 2010; Sato and Mokha, 2009; Zech et al., 2014).

The majority of the study interventions used full training units (Dobrijević et al., 2016; Filipa et al., 2012; Hewett et al., 1999; Kachanathu et al., 2014; Myer et al., 2006), but several authors applied balance training only as a warm-up (Ahmadabadi et al., 2015; Holm et al., 2004; O’Malley et al., 2016; Trecroci et al., 2015). Among the various exercise types, balance training and neuromuscular training (Eisen et al., 2010; Kang et al., 2013; Verhagen et al., 2005; Matin et al., 2014) were most commonly incorporated, followed by plyometric exercises and core stability training (Asadi et al., 2015; Karadenizli, 2016a, 2016b; Lust et al., 2009; Sato and Mokha, 2009).

To assess static balance, the authors mostly applied the SLS test (n = 18) with 13 studies that used measurements conducted on a force plate. The most common procedure was the QS test (n = 16), and the LOS test was used twice. In the analyzed studies, the researchers mainly relied on the SEBT (n = 13), YBT (n = 4) and BESS (n = 2) to assess dynamic balance. Although the SEBT was the most popular in these studies, at least two difficulties accompanied this procedure. In many cases, the SEBT was assessed only in three directions that corresponded to the YBT (Filipa et al., 2012; Imai et al., 2014; Heleno et al., 2016). In addition, in several studies, the results were presented as composite reach scores (Daneshjoo et al., 2012; Eisen et al., 2010; Pfile et al., 2016). Therefore, even if the differences reached significance, it was not possible to ascertain the direction in which the improvement occurred. The detailed characteristics of the training protocols and tested tasks are shown in Table 1.

Table 1

Influence of balance training on balance in various sports disciplines

	Subjects				Training Modality
Reference	N/Sex	Age (years)	Status Training	Discipline	D (min)	F (n/week)	T (week)	Training Type	Device + Procedure	Conclusions
Lust et al. (2009)	IG:open	20.00 ± 1.54	NR	baseball	30-45	3	6	CST	no device,	The OKC/CKC/CS
	kinetic								a single test	group and the
	chain/closed								consisted of a	OKC/CKC group
	kinetic chain								continuous	demonstrated
	(OKC/CKC):								alternating	significantly
	12M, open								procedure to	greater scores than
	kinetic chain/								lift one hand	the control group
	closed kinetic								to touch the	after training.
	chain/core								line then lift
	stability								the other
	(OKC/CKC/CS								hand to touch
	): 13M								the line for 15
	CG: 15M								s
Asadi et al. (2015)	IG (PLT): 8 M	IG (PLT): 20.1 ±	amateur	basketball	30	2	6	PLT	SEBT	After a 6-week
	CG	0.8								training period,
	(Basketball): 8	CG : 20.5 ± 0.3								the PLT + BT
	M									group showed
										significant
										improvements in
										all directions,
										whereas the
										basketball group
										did not show any
										significant
										changes.
Benis et al. (2016)	IG: 14 F	IG: 20 ± 2	national	basketball	30	2	8	NMT	YBT	Improvement over
	CG: 14 F	CG: 20 ± 1	league							baseline scores
			players							was noted in the
			practicing 4							posteromedial and
			times a week							posterolateral
			for 2 hours							reach directions
										and in the
										composite YBT
										scores of the
										experimental
										group. No
										differences in
										anterior reach
										were detected in
										either group.
										Differences were
										noted in
										postintervention
										scores for
										posteromedial
										reach and
										composite scores
										between the
										experimental and
										control groups.
McLeod et al. (2009)	IG:37 F	IG: 15.6 ± 1.1	competitive	basketball	90	2	6	NMT	BESS	Trained subjects
	CG: 25 F	CG: 16.0 ± 1.3						(functional	SEBT	scored
								strengthen		significantly fewer
								ing, PLT,		BESS errors on the
								agility BT)		single-foam and
										tandem-foam
										conditions at the
										posttest than the
										control group and
										demonstrated
										improvements on
										the single-foam
										compared with
										their pretest, the
										authors found a
										significant
										decrease in total
										BESS errors in the
										IG at the posttest
										compared with
										their pretest and
										the CG.
Imai et al. (2014)	IG: 10 M	IG: 16,5 ± 0,5	high	soccer	NR	3	12	CST	FP: SLS (EO,	Significant
	CG: 9 M	CG: 16,1± 0,6	school soccer						EC) 20s/ 2x	differences in the
			club, practice						SEBT	posterolat. and
			six times per							posteromed.directi
			week							ons between the
										pre and post test.
										Significantly lower
										values of length of
										COP between the
										pre and post test.
Pfile et al. (2016)	IG: 11 F	IG: 19.40 ± 1.35	11 Division I	basketball	about	2	6	NMT+PLT	SEBT-3	The mean
			women’s		30				directions x3	composite reach
			basketball						LESS	significantly
									(Landing	improved over
									error scoring	time. LESS scores
									system)	significantly
										improved over
										time
Saunders et al. (2013)	IG:14 F	14.7 ± 4.5	1 year of	figure	20	3	6	NMT	FP: SLS 15	No statistically
	CG:12 F		competition	skating					s/3x, SLL 15	significant
			experience						s/3x	differences
			2h of on-ice							between the
			practice per w.							groups
Ahmadabadi et al. (2015)	IG: 8 F	9.62 ± 1.45	more than	gymnastics	30	3	4	BT	FP: QS (EO,	A significant
	CG: 8 F		three years of						EC)	increase in balance
			athletic						SLS – 30 s	performance, a
			experience							significant increase
										in dynamic and
										static balance with
										double feet
Dobrijević et al. (2016)	IG: 33 F	7-8	recreational	gymnastics	60	2	12	PT	no device	After
	CG: 27F								SLS (EO, EC)	proprioceptive
									time to losing	training, the
									balance	experimental
										group significantly
										improved
										performance in all
										the tests for
										maintaining a
										balance position.
Holm et al. (2004)	IG: 35 F	23 (± 2.5)	elite division	handball	about	min. 3	NR	NMT	Balance KAT	There was a
			14.9 (± 3.2)		15	during			2000: SLS	significant
			years, 4.7 (±			5-			(right, left leg)	improvement in
			2.8) years at			7weeks			x3, 2-leg	dynamic balance
			the top level			1 during			dynamic test	between test 1 and
			experience			the			x3	test 2. The effect
			10 to 11			season			custom made	on dynamic
			h/week - total						device:	balance was
			number of						assessment of	maintained 1 year
			hours						knee	after training. For
									kinesthesia	static balance, no
										significant changes
										were found. For
										the other variables
										measured, there
										were no statistical
										differences during
										the study period.
Karadenizli (2016)	IG: 16 F	14.57 ± 0.92	3.66 ± 0.63	handball	NR	2	10	PLT	FP: QS (EO,	Significant
			years sport						EC), SLS – 30s	differences were
			experience							observed between
									Dynamic	the pre- and post-
									Balance -	test of plyometric
									Slalom Test –	education training
									60 s	of flexibility,
										standing long
										jump, left leg
										ellipse
										area at unipedal
										static balance.
Verhagen et al. (2004)	29 (F/M)	IG: 22.5 ± 2.4	second and	volleyball	NR	2	5.5	BT	FP: SLS, QS	Balance training
	IG: 10	IG (volleyball):	third							did not lead to a
	IG (volleyball):	23.6 ± 3.2	volleyball							reduction in the
	8	CG: 25.5 ±7.8	players							centre of pressure
	CG: 11									excursion in a
										general population
										consisting of non-
										injured and
										previously injured
										subjects.
Malliou et al. (2004)	30 (IG: 15 M/F,	19.3 ± 9	no	skiing	20	4	NR	PT	BBS: SLS 20	No statistically
	CG: 15 M/F)		experience						s/3x (right,	significant
									left leg)	differences
										between the
										groups were
										found.
Karadenizli et al. (2016)	IG: 14 F	IG:15.64 ± 0.82	3.5 years of	handball	NR	2	10	PLT	FP: SLS (right,	The IG made
	CG: 12 F	CG: 15.38 ± 0.92	sport						left leg) – 30 s	significantly
			experience							greater
										improvements
										than the CG in the
										SLS (left).
Matin et al. (2014)	IG: 12 M	11.34 ± 3.68	the	fitness	60	3	4	NMT	SLS 3x	Neuromuscular
	CG: 12 M		representative						Dynamic test:	training can
			physical						(jumping) five	enhance important
			fitness team						scores were	factors of static
			of						dedicated for	and dynamic
			the						covering the	balance and the
			elementary						mark	results showed a
			schools						and five	significant increase
									scores for	in performance of
									holding the	the individuals
									balance	participating in
									stance as	neuromuscular
									static for 5 s	training.
Eisen et al. (2010)	36 F/M	18-22	NR	soccer/basket	NR	3	4	BT	SEBT	There was no
	IG (dynadisc):			ball						difference for each
	12									group
	IG (rocker									individually, and
	board): 12 CG:									no difference
	12									between trained
										and untrained legs
										within a subject
Zech et al. (2014)	IG: 15	IG: 15.7 ± 3.9	first regional	hockey	20	2	10	NMT	FP: 3x jump-	All balance
	CG: 15	CG: 14.1 ± 1.4	youth						landing time	measures except
			divisions						to	the medial-lateral
									stabilization	TTS improved
									(TTS),	significantly over
									SLS 30 s/3x	time in both
									(preferred	groups. Significant
									leg)	group by time
									MSEBT	interactions were
									BESS	found for the BESS
										score. The IG
										showed greater
										improvements
										after 10 weeks of
										training in
										comparison to the
										CG.
Myer et al. (2006)	IG (PLT): 8 F	IG 15.9+/-0.8	not less than	voleyball	90	3	7	IG: PLT	FP: a single-	The percentage
	IG2 (CST+BT):	IG2 15.6+/-1.2	4 years of					IG2:	leg hop and	change from the
	11 F		experience					CST+BT	BT x3	pretest to posttest
									(randomized	in vertical ground
									trials on each	reaction force was
									side)	significantly
										different between
										the PLT and
										CST+BT groups
										considering the
										dominant side.
Romero-franco et al. (2012)	IG: 16 M	IG: 22.5 ± 5.12	NR	running	30	3	6	PT	FP: QS (EO,	Significant
	CG: 17 M	CG: 21.18 ± 4.47							EC) 2x52s	differences were
									BBS: EO 3x20	found in stability
									s, LOS in 8	in the medial-
									different	lateral plane with
									directions	EO, gravity center
										control in the right
										direction and
										gravity center
										control in the back
										direction after the
										exercise
										intervention in the
										IG.
Sato and Mokha (2009)	IG:12 F/M	IG:37.75 ± 10.63	recreational	running	NR	4	6	CST	SEBT	CST had no
	CG: 8 F/M	CG: 39.25 ±	and							significant
		10.81	competitive							influence on scores
										measured by the
										SEBT or any GRF
										variables.
Mahieu et al. (2006)	IG: 6 F, 11 M	9-15	competitive	skiing	30	3	6	VT	SMART	No significant
	CG: 8 F, 8 M								Balance	differences except
									Master: LOS 8	for directional
									s/8x, rhythmic	control during the
									weight shift	LOS and the left-
									left /right,	right excursion of
									forward/back	the rhythmic
									ward	weight shift test
										were found.
Cankaya et al. (2015)	IG: athletes 25	11	NR	soccer	40	3	8	BT	FP: QS (EO,	Balance
	M, sedentary:								EC),	performance of the
	25 M								SLS – 30 s	athletes and
	CG: 25 M								clockwise	sedentary group
									rounds 5 x 60s	improved
										compared to the
										CG.
Daneshjoo et al. (2012)	CG: 12 M; IG	CG: 19.7 ± 1.6	professional	soccer	20-25	3	6	FIFA 11:	ID: JPS	Both warm up
	(FIFA 11): 12	IG FIFA 11:	(five year					BT + ST +	SEBT	programs
	M, IG	19.2 ± 0.9	experience of					PT	Stork Stand	improved
	(HarmoKnee):	IG Ham o	playing					Harm o	Balance Test	proprioception in
	12 M	Knee: 17.7 ± 0.4	soccer at					Knee: BT +		the dominant leg
			professional					ST + CST		at 45° and 60° knee
			level)							flexion. Dynamic
										balance assessed
										by the SEBT also
										showed
										improvement in
										both groups, with
										the HarmoKnee
										group showing
										significant
										difference when
										compared to the
										CG.
Iacono et al. (2014)	IG: 10 M	IG: 18.7 ± 0.67	competitive	soccer	NR	4	5	CST	FP: SLS (EO,	CST significantly
	CG: 10 M	CG: 19 ± 0.63	players						EC) – 3x20 s	improved static
									SEBT	and dynamic
										balance
Gioftsidou et al. (2006)	39 (CG:13, IG -	16 ± 1	The young	soccer	20	3	12	BT	BBS: SLS 20	Significant
	before		championship						s/3x (right,	differences in the
	appropriate		of the first						left leg)	IG after training.
	training: 13 M,		Greek
	IG – after		division
	appropriate
	training: 13)
Heleno et al. (2016)	IG: 12 M	14-16	players with	soccer	NR	3	5	SMT	SLS,	After a five-week
	CG: 10 M		a minimum						Side Hop Test	training program,
			of 3 years of						(SHT),	the intervention
			training						Figure of	group obtained
			experience;						Eight Test	significant results
			participation						(F8)	in the
			in state and						MSEBT	F8, SHT and SEBT,
			national							as well as in the
			competitions;							following
			training 5							variables: area of
			times a week							pressure of sway
										center (COP),
										mean velocity and mean frequency of
										COP
Trecroci et al. (2015)	IG: 12 M	11.3 ± 0.70	sub-elite	soccer	15	2	8	BT	YBT	Significantly
	CG: 12 M		players							greater
										improvements in
										the YBT
Manolopoulos et al. (2015)	IG: 20 (ST: 10	ST: 21.3 ± 1.3	amateur	soccer	NR	2	8	ST	FP: stork	COP (cm) in
	M	SMT: 22 ± 1.7						ST + SMT	stance, raise	anteriorposterior
	SMT: 10 M)								the heel off	and mediolateral
									the ground –	axes decreased
									5 s	significantly after
										training
Kachanathu et al. (2014)	IG: 23 M	18 ± 2	NR	soccer	60	Phase-I:	4	CST	Double	Significant
	CG: 23 M					6			Straight Limb	differences
						Phase-II:			Lowering test:	of dynamic
						6			x3 SEBT: 8	balance and core
						Phase-			directions x3	stability in the IG
						III: 3				compared to the
										CG
Granacher et al. (2016)	IG: 12 M	12-13	first division	soccer	NR	2	8	BT	Standing	Results indicated
	CG: 12 M		Tunisian					PLT	Stork Test,	that BT provided
									YBT	significantly
										greater
										improvements in
										the YBT
Gioftsidou et al. (2012)	IG1: 13	22.7 ± 3.5	first Greek	soccer	20	IG1: 6	IG1: 3	BT	BBS: SLS 20 s	Both training
	IG2: 13		division			IG2: 3	IG2: 6		x3 (each leg)	groups
	CG: 12								Balance	demonstrated
									board: SLS	significant
									time to lose	improvements on
									balance	Biodex stability
										tests. Similarly for
										the balance board,
										the results
										revealed
										significant
										improvements for
										both IGs.
Alyson et al. (2012)	IG: 13 F	IG: 15.4 ± 1.5	competitive	soccer	50	2	8	NMT	SEBT	After NMT,
	CG: 7 F	CG:14.7 ± 0.8								subjects
										demonstrated a
										significant
										improvement in
										the SEBT score on
										the right and left
										limb.
O’Malley et al. (2016)	IG: 41 M	IG: 18.6 (18.4-	teams were	2 teams:	15	2	8	GAA 15	YBT	There was a
	CG: 37 M	18.8)	required to	1 football				(Gaelic	LESS	greater reduction
		CG: 18.3 (18.1-	train at	1 hurling				Athletic	(Landing	in mean LESS
		18.5)	least twice					Associatio	Error Scoring	score in favour of
			per week.					n) training	System)	the IG post
								program		exercise training.
										Clinically and
										statistically
										significant
										improvements in
										dynamic balance
										and jump-landing
										technique
										occurred in
										collegiate level
										Gaelic football and
										hurling players.
Pau et al. (2012)	IG: 13 F	IG:13.2 ± 0.2	0-3 years of	voleyball	20-30	2-3	6-9	NMT	FP: QS	The IG exhibited
	CG: 13 F	CG: 13.0 ± 0.1	experience						(EO,EC) – 20 s	smaller sway areas
									SLS 10 s	in EC conditions in
										the bipedal stance,
										while the other
										variables were
										unaffected. BT also
										reduced sway area
										and A-P COP
										displacements of
										the nondominant
										limb for SLS.
Kang et al. (2013)	36 M	NR	middle	weightlifters	NR	NR	8	BT	SLS (EC)	Significant
	IG (high school):		school: exp.							changes
	8		of 25.44							were found in one-
	IG (middle		months; high							leg standing time
	school: 8		school: exp.							with eyes closed in
	CG (high		of 55.44							the IG.
	school): 8		months
	CG (middle
	school): 8

NR = non reported; IG = intervention group; CG = control group; F = females; M = males; n = group size; PT = prioproceptive training; BT = balance training; CST = core stability training; PLT = plyometric training; ST = strength training; SLS = single leg stance; NMT = neuromuscular training; D = training duration (min); F = frequency (n/week); T = duration of the intervention (week); FP = force plate; BBS = biodex balance system; SEBT = star excursion balance test; ID = isokinetic dynamometry; EO = eyes open; EC = eyes close; QS = quiet standing; BESS = balance error scoring system; YBT = Y balance test; SLL = single leg landing; SMT = sensory motor training

The influence of balance training on injury prevention

Balance control is a crucial factor in sports and an important component of common motor skills. Disturbances in balance control can increase the risk of injuries during high intensity activities (Burke-Doe et al., 2008; McGuine et al., 2000). The importance of balance control in the prevention of damage and musculoskeletal injuries during sports performance has been emphasized (Carolyn A Emery, 2005) and investigated in many studied cases (see review by Hrysomallis, 2007). Although the cause of injury is not always known, several risk factors for impairment in balance during training have been indicated (McKay et al., 2001). These factors include an insufficient warm-up (Woods et al., 2007), poor flexibility (Hartig and Henderson, 1999; Zakas et al., 2005), muscle imbalances (Parry and Drust, 2006; Croisier et al., 2008), muscle weakness (Croisier, 2004; Junge and Dvorak, 2004), neural tension (Turl and George, 1998), fatigue (Worrell, 1994), and previous injuries (Ekstrand et al., 2011; Parry and Drust, 2006).

The most common sports injuries (60%) are sprains, dislocations, and ligament ruptures that occur at the knees and ankles and at the hands, elbows, and shoulders (Conn et al., 2003; Hawkins and Fuller, 1999; Powell and Barber-Foss, 1999; Schneider et al., 2006). Improving balance in athletes by appropriate training has proven to engender positive effects on the reduction of the discussed injuries (Hrysomallis, 2007). Exercises may be included into a training program as part of an injury prevention strategy or with the primary goal of improving an athlete’s performance (Sannicandro et al., 2014). According to Hrysomalis (2007), Hűbscher (2010) and other authors, we are able to distinguish different design concepts and components of preventive exercises for balance including plyometrics, strengthening, balancing, endurance and stability, with a different approach to preventive programs (Heidt et al., 2000; Myklebust et al., 2003; Soderman et al., 2000). The results of our analysis of the relationship between different balance prevention training protocols and injuries are shown in Table 2. It indicates the effectiveness of balance training in reducing the incidence of sports injuries among athletes. The analysis of the prevention programs contains the results for team sports (such as basketball, soccer, handball, and volleyball), mainly because of their specificity (high-risk of injuries), which may consequently cause long-term disabilities for the injured player (Lohmander et al., 2004; Myklebust and Bahr, 2005; von Porat et al., 2004).

Table 2

Relationship between different balance prevention training and injuries

	Subjects				Training Modality
References	N/Sex	Age (years)	Status Training	Discipline	D (min)	F (n/week)	T (week)	Training Type	Conclusions
Eils et al. (2010)	n = 198	IG1	performance	basketball	20	1	NR	PT	The risk of sustaining an
	IG1 = 81 49M : 32F	22.6 ± 6.3	level of the				(all		ankle injury was
	CG1 = 91 54M : 37F	CG1	players				season)		significantly reduced in
	IG2 = 8 4M : 4F	25.5 ± 7.2	varied						the IG by approximately
	CG2 = 8 4M : 4F	IG2	between the						35%. The IG showed a
		24.3 ± 2.9	seventh						significantly more stable
		CG2	highest						SLS concerning the
		25.9 ± 8.2	(Kreisliga)						mediolateral direction.
			and the						The degree of error for
			highest						10-dorsiflexion and 15-
			league						plantarflexion and the
			(Bundesliga)						mean error were
			in Germany						significantly reduced in
									the posttest in the IG, but
									not in the CG.
Soligard et al. (2008)	n =1892 F	13-17	at least two	soccer	20	NR	NR	Running	There was a
	IG = 1055		training				(8 months-	exercises	significantly lower risk
	CG = 837		sessions a				all season)	BT	of injuries overall,
			week in					CST	overuse injuries, and
			addition to					PLT	severe injuries in the IG.
			match play
Kraemer and Knobloch (2009)	IG = 24 F	21 ± 4	German	soccer	3000	NR	NR	PT	One year after training
			premier				(3 years)	BT	implementation,
			league					PLT	noncontact injuries
									decreased significantly
									by 65% (p = .021).
									Overall, the mean injury
									rate of all noncontact
									injuries during all
									intervention periods
									significantly decreased
									by 42% (p = .045) versus
									the control period.
Owen et.al. (2013)	n = 67 M	IG = 28.6 ± 3.75	competitive	soccer	NR	2	NR	BT	During the intervention
	IG = 44	CG=27.4 ± 4.85	players				(2 seasons:	ST	season, the number of
	CG = 23						2008-2010)	CST	muscle strain/tears was
								FT	less (25% of total
									injuries) than the control
									season (52% of total
									injuries).
Timothy et al. (2006)	n = 765 F/M	IG = 16.4 ± 1.2	high school	basketball	10	3	NR	BT	A reduced risk of an
	IG =373	CG = 16.6 ± 1.1	students	soccer			(all		ankle sprain was
	CG = 392		trained by				season)		observed after
			certified						intervention.
			coaches
Eils et al. (2010)	n = 198	IG1	performance	basketball	20	1	NR	PT	The risk of sustaining an
	IG1 = 81 49M : 32F	22.6 ± 6.3	level of the				(all		ankle injury was
	CG1 = 91 54M : 37F	CG1	players				season)		significantly reduced in
	IG2 = 8 4M : 4F	25.5 ± 7.2	varied						the IG by approximately
	CG2 = 8 4M : 4F	IG2	between the						35%. The IG showed a
		24.3 ± 2.9	seventh						significantly more stable
		CG2	highest						SLS concerning the
		25.9 ± 8.2	(Kreisliga)						mediolateral direction.
			and the						The degree of error for
			highest						10-dorsiflexion and 15-
			league						plantarflexion and the
			(Bundesliga)						mean error were
			in Germany						significantly reduced in
									the posttest in the IG, but
									not in the CG.
Soligard et al. (2008)	n = 1892 F	13-17	at least two	soccer	20	NR	NR	Running	There was a
	IG = 1055		training				(8 months-	exercises	significantly lower risk
	CG = 837		sessions a				all season)	BT	of injuries overall,
			week in					CST	overuse injuries, and
			addition to					PLT	severe injuries in the IG.
			match play
Kraemer and Knobloch (2009)	IG = 24 F	21 ± 4	German	soccer	3000	NR	NR	PT	One year after training
			premier				(3 years)	BT	implementation,
			league					PLT	noncontact injuries
									decreased significantly
									by 65% (p = .021).
									Overall, the mean injury
									rate of all noncontact
									injuries during all
									intervention periods
									significantly decreased
									by 42% (p = .045) versus
									the control period.
Owen et.al. Knobloch (2013)	n = 67 M	IG = 28.6 ± 3.75	competitive	soccer	NR	2	NR	BT	During the intervention
	IG = 44	CG=27.4 ± 4.85	players				(2 seasons:	ST	season, the number of
	CG = 23						2008-2010)	CST	muscle strain/tears was
								FT	less (25% of total
									injuries) than the control
									season (52% of total
									injuries).
Timothy et al. Knobloch (2006)	n = 765 F/M	IG = 16.4 ± 1.2	high school	basketball	10	3	NR	BT	A reduced risk of an
	IG =373	CG = 16.6 ± 1.1	students	soccer			(all		ankle sprain was
	CG = 392		trained by				season)		observed after
			certified						intervention.
			coaches
Malachy et al. (2007)	n = 175	15-18	high school	football	10	2	13	SLS	The injury incidence for
	IG = 175		students					BT	the players after the
									intervention was
									significantly lower than
									the combined injury
									incidence before the
									intervention (p < .01).
Cumps et al.(2007)	n = 50 M/F	IG= 17.7 ± 3.9	elite youth	basketball	10	3	22	BT	Relative risks showed a
	IG = 26	CG= 18.0 ± 2.7	and young					SLS	significantly lower
	CG = 24		senior					PLT	incidence of lateral ankle
			basketball					Dynamic	sprains in the IG
			players					exercises	compared to the CG.
Mandelbaum et al. (2005)	IG1: 1041 F	14-18	competitive	soccer	20	NR	NR	stretching	During the first period
	CG1: 1905 F		female youth				(2 season)	ST	(IG; CG1), there was an
			soccer					PLT	88% decrease in ACL
	IG2: 844 F		players in a					Agility	injury in the IG subjects
	CG2: 1913 F		southern					NMT	compared to the control
			California						group. In the second
			soccer league						period (IG2; CG2) there
									was a 74% reduction in
									ACL tears in the IG
									compared to the age-
									and skill-matched
									controls.
Verhagen et al. (2005)	IG = 641	IG= 24.4 ± 2.8	the second	voleyball	5	NR	NR	BT	Significantly fewer ankle
	CG = 486	CG= 24.2 ± 2.5	and third				(one	SLS	sprains in the IG were
			Dutch				season		found compared to the
			volleyball				2001/2002)		CG. A significant
			divisions;						reduction in the ankle
			experience in						sprain risk was found
			years 13.3 ±						only for players with a
			2.3						history of ankle sprains.
Soderman et al. (2000)	n = 140 F IG = 62 CG = 78	IG= 20.4 ± 4.6 CG= 20.5 ± 5.4	players of the second and third Swedish divisions	soccer	15	NR	NR (12 weeks)	BT	The results showed no significant differences between the groups with respect either to the number, incidence, or type of traumatic injuries of the lower extremities.
Emery et al. (2012)	n = 744 M/F	IG: U13-15=46.6%	first and	soccer	30	NR	20	NMT	There was a 38%
	IG = 380	U16-18=53.4%	second					BT	reduction in all injury in
	CG = 364	CG: U13–	Calgary					ST	the IG compared with
		15=48.9% U16–	youth					Agility	the CG and a 43%
		18=51.1%	division of					Stretching	reduction in acute-onset
			indoor						injury.
			football
Hewett et al. (1999)	n = 1263 F/M	high school	high school	soccer,	60-90	3	6	NMT	The untrained group
	IG = 366 FCG = 463 F	students	students,	basketball,				PLT	demonstrated an injury
	CGPopulation = 434		females were	volleyball					rate 3.6 times higher
	M		players,						than the trained group
			males were						and 4.8 times higher
			not						than the male control
									group. The trained
									group had a significantly
									lower rate of noncontact
									injuries than the
									untrained group (p =
									0.01).
Petersen et al. (2005)	N = 276 F	NR	2 of the	handball	10	3	8	PT	Ankle sprain was the
	IG = 134		teams were					PLT	most frequent diagnosis
	CG = 142		from the						in both groups with 11
			third highest						ankle sprains in the CG
			league;						and 7 ankle sprains in
			4 teams were						the
			of a superior						IG. The knee was the
			amateur						second frequent injury
			level;						site. In the CG, 5 of all
			4 teams were						knee injuries
			at a lower						were anterior cruciate
			amateur level						ligament (ACL)
									ruptures, while in the IG
									only one.
			competitive						There was no difference
			players with						between the IG and CG
			13.3 ± 2.1						in performance from the
			hours of						pre to post-test for any of
			football						the tests used.
			activities per					CST
	n = 36 F		week and					BT
Steffen et al. (2008)	IG = 18	16-18 ( 17.1 ± 0,8)	that had been	football	15	NR	10	PLT
	CG = 16		involved in					ST
			organized
			football for
			10 ± 1.5 years

NR = non reported; IG = intervention group; CG = control group; F = females; M = males; n = group size; PT = prioproceptive training; BT = balance training; CST = core stability training; PLT = plyometric training; ST = strength training; SLS = single leg stance; NMT = neuromuscular training; D = training duration (min); F = frequency (n/week); T = training duration (week)

Conclusions

In most of the reviewed articles, balance training has proven to be an effective tool for the improvement of postural control. However, a few articles stated that such effect did not occur (Eisen et al., 2010; Mahieu et al., 2006; Malliou et al., 2004; Sato and Mokha, 2009; Saunders et al., 2013; Verhagen et al., 2005), and a few studies, in which the effect was not reflected in all balance measures, suggested that balance training did not influence all of the dimensions of postural control (Benis et al., 2016; Holm et al., 2004; Pau et al., 2011; Zech et al., 2014). In some cases where the authors carried out both static and dynamic tests, significant results occurred in only one test type. Therefore, we would recommend the execution of both types of tests to decrease the risk of making inappropriate or global conclusions that training is ineffective in general.

Another issue is that the duration of training was heterogeneous. In most cases, it was approximately 40-50 min and was implemented as a full training session; however, in some articles, this time was rather short, i.e., only 10-20 min. In several studies, duration was not reported (Table 1). No gold standard is apparent in this field; therefore, it is difficult to make a global conclusion about the effectiveness of various types of balance training. Moreover, we are aware that it may be very difficult to establish one model of training that would be appropriate for each sport discipline, including its characteristics and demands. The main aim of this review was to identify a training protocol that is most commonly used and may lead to improvements in balance. Therefore, on the basis of analyses including papers in which training protocols resulted in positive effects on balance performance, it may be stated that an efficient training protocol should last for 8 weeks, with a frequency of two training sessions per week, and a single training session of 45 min.