Gait Retraining for Performance and Injury Risk

Izzy 5. Moore, Tom Goom and Kelly J. Ashford

Highlights

• • There is no such thing as the ‘correct way to run’ or ‘perfect running form’; a universal style of running does not appear to exist, particularly regarding economical running.
• • Gait retraining may need to be undertaken only by some runners, and prescription should be based upon an individual assessment that includes other clinical considerations.
• • Male and female runners should be considered separately due to important discrepancies in running biomechanics between the sexes, and the potential influence of breast motion on running economy in females.
• • Gait retraining has been used successfully to lower the risk of developing certain injuries and to reduce the pain associated with specific overuse injuries; however, modification of running style redistributes loads, which may increase injur)' risk in other parts of the body.
• • Internal focus of attention verbal cues seems more effective at producing desired technique changes, potentially by invoking constructive conscious control of ones gait to mediate improvements of a habitual movement pattern.
• • The overall injury rate is similar between barefoot runners and those who wear trainers; however, small amounts of barefoot running may be useful as a gait retraining, conditioning and rehabilitation approach in some runners.

Introduction

The biomechanics of running has been well researched, with early work dating back to the 1930s. Running, like other movements, is a skill. Yet, our innate ability to run without instruction from an early age means many individuals never receive training specifically relating to this skill. Fundamentally, this means that how most individuals run is a consequence of the body self-organising. Many constraints are placed upon the body as it self-organises, including anthropometries (height, leg length), injury history, experience and exposure, and although rather a simplistic view, metabolic cost has often been cited as the primary driver of gait selection. Other factors that appear to contribute to movement selection include minimising muscle excitation, effort requirements and reducing the ‘jerkiness’ of motion (Hreljac, 2000). Theoretically, pain may also contribute to gait selection due to avoidance behaviours. The gait an individual uses is therefore a complex interplay of contributing constraints and drivers.

Two juxtaposing views can be presented by practitioners and researchers that either running gait should be left untouched, or that it must be changed to prevent injury and improve performance. To this end, this chapter will discuss how and when to modify gait and seek answers to the following questions: Is there an economically optimal running technique? Can and should we modify uninjured gait patterns to prevent injury? How can gait retraining be used during lower limb injury rehabilitation? And, is there a role for footwear and barefoot running in modifying gait?

Gait Retraining and Performance

The pursuit of the ‘perfect running form’ is likely to be a fruitless task, particularly when often sex-aggregated data are used to assess associations between running economy and gait. Females produce greater relative stride lengths (stride length divided by leg length), use higher stride frequencies at the same velocity, opt to use a lower duty factor by spending less time in contact with the ground (Nelson et al., 1977) and have greater pelvic anterior tilt than their male counterparts (Schache et al., 2003). They are, therefore, not using a gait that is a scaled- down version of what males use due to typically being shorter and weighing less.

Adopting a one-size-fits-all approach to running gait has led to different forms of running being advocated and specific values being recommended for certain gait characteristics. For example, there appears to be a gravitation towards wanting runners to produce a cadence of 180 steps per minute, but there is currently no evidence to support this approach. Whilst there is no basis for this, several modifiable intrinsic and extrinsic running biomechanical factors and their effect on running economy have been identified (see Table 13.1), with the propulsive phase being highlighted as having the strongest direct relationships with running economy (Moore, 2016). It is important to note that unless otherwise stated in this section, the term ‘runners’ is used to represent ‘male’ runners due to paucity of female-specific studies.

Economical Spatiotemporal Characteristics

From an economical perspective, large changes in stride length and frequency are to be avoided in trained runners. Numerous studies have shown that increasing or decreasing stride lengths and frequencies by >6% increases the volume of oxygen consumed, meaning economy is worsened. In fact, trained runners exhibit incredibly economical stride lengths and frequencies, which tend to be within 3% of their optimal stride length/frequency (Cavanagh and Williams, 1982; de Ruiter et al., 2013). The mathematically optimal stride length/ frequency is quantified as the stride length/frequency that would produce the lowest metabolic cost of running. In addition, small changes (<3%) to stride length do not appear to affect running economy (Moore, 2016; Craighead et al., 2014). Therefore, stride length appears to be able to operate within an optimum window, which is 97-100% of the self-selected stride length (Moore, 2016).

In untrained runners, there is likely to be greater opportunity to modify stride length/ frequency to reap economical benefits. De Ruiter and colleagues (2013) reported untrained

TABLE 13.1 Modifiable intrinsic and extrinsic running biomechanics and their effect on running economy (RE)

Note: GRF = ground reaction force. CoM = centre of mass. Source: Adapted from Moore (2016).

runners to be 8% away from their optimal stride frequency, suggesting such runners may benefit from a gait retraining intervention to increase their stride frequency towards one that minimises metabolic cost. It is interesting to note that in both untrained and trained runners, the preference is to overstride, i.e. produce strides that are longer than optimal, and therefore produce stride frequencies that are lower than optimal. Anthropometries such as leg length and height may place constraints on which combination of stride length and frequency can be achieved by an individual; however they are not the primary determinants (Cavanagh and Kram, 1989). It is likely to be a complex interaction between the physiological constraints, energy' demand and metabolic cost that lead to an individual self-selecting a given stride length.

Ground contact time is the metabolically expensive phase of the gait cycle (Arellano and Kram, 2014), meaning swing time is the less metabolically expensive phase. Intuitively, this could be interpreted to mean shorter ground contact times and longer swing times are desirable. Associations between the time spent on the ground and running economy have, however, been conflicting (Moore, 2016). Reasons for such conflicting evidence could be the underlying muscle-tendon properties of each runner. Short ground contact times require high magnitudes of force to be produced rapidly. Therefore, this strategy may be suited to individuals with a higher proportion of fast twitch fibres and favour the use of elastic energy' storage and return (Lussiana et al., 2019). Conversely, long ground contact times mean force is produced slowly, suggesting a greater reliance on slow twitch muscle fibres, but enables greater forward translation of the centre of mass (Lussiana et al., 2019). Ground contact time has since been shown to be an optimised gait characteristic in male and female runners (although only one female was part of the cohort) (Moore et al., 2019a). This means, like stride length and frequency, male and female runners choose ground contact times that minimise the metabolic cost of running (enhance running economy).

Interestingly, 60% of the male and female runners in a study by Moore and colleagues (2019a) had ground contact times that were slightly shorter than their predicted optimal ground contact time, whilst 40% had slighter longer than optimal times (Figure 13.1). The collective understanding about ground contact time highlights two different strategies: (1) short ground contact times leading to low duty factors; and (2) long ground contact times leading to high duty factors. Duty factor is the proportion of the gait cycle one foot spends in contact with the ground. Trained runners, through a combination of self-optimisation and training, are likely to possess the strategy that works best for their musculoskeletal system but may need slight modifications to fine-tune their gait. An individual assessment of economical running would help inform training requirements for both trained and untrained runners to identify if small or large refinements to technique are warranted.

The amount a runner’s centre of mass rises and falls during each gait cycle in the vertical direction is known as vertical oscillation. Increasing vertical oscillation leads to a worsening of running economy (Tseh et al., 2008), possibly because the work performed against gravity is increased (Moore, 2016). However, no study has tried to minimise vertical oscillation and measure the effect on running economy. Minimising vertical oscillation will, however, have repercussions for other gait characteristics given the interaction with spatiotemporal characteristics. In particular, it will reduce stride length (Adams et al., 2018), increase stride frequency (Adams et al., 2018) and lengthen ground contact time (Adams et al., 2018). Further, if stride length, stride frequency and ground contact time have been shown to be optimised gait characteristics in runners, vertical oscillation could also be considered optimised. In support of this, Lussiana and colleagues (2019) identified different magnitudes of vertical oscillation depending upon the duty factor of a runner. Runners with a low duty factor had shorter ground contact times relative to stride time, larger vertical oscillations and lower horizontal centre of mass displacements than runners with a high duty factor. Yet, there was no difference in running economy between high and low duty factor runners. Interestingly, there was an equal proportion of females (40%) in the low and high duty factor groups, showing that both types of strategies can be adopted by females and males. Practitioners should be aware of the potential change in vertical oscillation that accompanies changes in other spatiotemporal characteristics but may not need to target it specifically as an uneconomical gait characteristic.

Economical Kinetics

Early work demonstrated that the vertical force required to support an individual’s body mass under gravity, known as body weight, is one of the primary determinants of running economy, along with the time spent applying force (ground contact time) (Kram and Taylor, 1990). However, whilst running at a constant velocity, the average vertical force applied to each stride will always be equal to the body weight of a runner and, thus, cannot be modified during a run to the same extent that ground contact time can be (see the ‘Economical spatiotemporal characteristics’ section).

FIGURE 13.1 Relationship between the deviations from self-selected running gait characteristics (%) and from the metabolic cost during self-selected gait (%). (A) Example of self-selected ground contact time longer than optimal. (B) Example of self-selected ground contact time within 1% of optimal. (C) Example of self-selected ground contact time shorter than optimal. Solid black lines represent ground contact time. Solid grey lines represent leg stiffness. Optimal gait characteristics that minimise metabolic cost are identified by circles (black (•) = ground contact time; grey (•) = leg stiffness). Dashed lines highlight the corresponding X and Y values for optimal gait characteristics

Source. Adapted from Moore et al. (2019b).

Running at a constant velocity requires alternating periods of deceleration and acceleration. Reducing the amount of velocity lost during the braking (deceleration) phase of stance would reduce the amount of propulsion (acceleration) required to generate momentum. This would beneficially impact economy, as the propulsive impulse along with the vertical force to support a runner’s body weight incur the greatest metabolic cost (Arellano and Kram, 2014).

Understanding the interaction between braking and propulsion has been relatively unexplored in runners, but the seminal work by Chang and Kram (1999), which applied external horizontal forces to the body, identified that braking incurs a lower metabolic cost than propulsion in males and females. Yet, to achieve a constant velocity, braking and propulsive impulses must be equal. It therefore follows that reducing the change in centre of mass velocity during braking would be a beneficial economical strategy' to adopt. This strategy appears to be the case only when sex-aggregated data are used (Folland et al., 2017). When females are considered separately, greater velocity changes are associated with better running economy (Williams et al., 1987). Further, through two studies on female beginner runners, Moore and colleagues (2012b, 2016) observed reduced knee extensor moments during braking, greater alignment of their legs with the ground reaction force (GRF) during propulsion, greater propulsive forces and stiffer calf musculo-tendon units as the runners became more economical. However, only the change in GRF alignment was associated with the change in economy, meaning females appear to be able to reduce muscular force requirements during braking and enable a more economical push-off to be produced, potentially through a more efficient stretch-shortening cycle in the calf musculo-tendon unit. The direct relationships observed between propulsive kinetics and economy may in fact be mediated by braking kinetics. It is not known why female and male runners appear to exhibit different braking and propulsion strategies. Studies need to be designed to capture the potential female-specific factors that contribute to this (e.g. breast biomechanics) rather than the continued focus on only the lower limb to understand economical running gait.

Arguably, leg stiffness receives a lot of attention in relation to economical movement. As a quantity derived from a theoretical model of running gait (the spring-mass model), the application is perhaps muddied by a lack of awareness around the assumptions of the model and eagerness to consider this quantity as an actual, tangible gait characteristic. Nevertheless, the storage and reutilisation of elastic energy' is seen as a fundamental component of running gait. Therefore, being able to store and/or reutilise more elastic energy by having a stiffer musculoskeletal system intuitively should minimise the volume of oxygen required to produce the level of energy' needed to run a given distance.

It was previously thought that greater leg stiffness would lead to superior running economy (Moore, 2016), but this has been shown only in correlational studies (Dalleau et al., 1998) or inferred from knee and ankle flexion angles during stance (Folland et al., 2017; Moore et al., 2014b). Notably, adopting a compliant, crouched running gait (‘Groucho running’) with exaggerated knee flexion worsens running economy. Yet, like spatiotemporal gait characteristics (e.g. stride frequency, stride length and ground contact time), leg stiffness appears to be a relatively optimised gait characteristic in males and females (Moore et al., 2019a). Furthermore, trained runners appear to have a homogenous leg stiffness modulation strategy involving force production, which is not evident in untrained runners (Bitchell et al., 2019). Both male and female runners predominately favour stiffer limbs than is optimal for running economy, which allows them to produce short ground contact times (Moore et al., 2019a). If this strategy is combined with longer strides than optimal as reported by Cavanagh and Williams (1982), it potentially exposes runners to large braking forces and vertical loading rates. As a result, we recommend individually assessing runners. It is possible that optimising stride length, frequency, ground contact time and leg stiffness using an individual running assessment to improve running economy may also be beneficial for injury risk, if braking forces and impact loading rates are reduced (Hreljac et al., 2000; Napier et al., 2018).

Economical Kinematics

A large number of lower limb kinematics have been associated with running economy in males and females (Moore, 2016), with the act of swinging the leg contributing to 7% of the metabolic cost of running (Arellano and Kram, 2014). Many of the lower limb kinematics have been identified only within one study and often conflicting relationships are presented for the same kinematic variable. For example, high magnitudes of peak knee flexion during stance may be good for economy but could also be bad for economy (Williams and Cavanagh, 1987; Folland et ah, 2017). Greater peak knee flexion may indicate faster knee flexion velocities, which are associated with smaller changes in centre of mass velocity during braking and propulsion (Williams et ah, 1987), whilst lower peak knee flexion may indicate a stiffer lower limb during stance (see the ‘Economical kinetics’ section for further detail). Nevertheless, one consistent finding, reported for both male and female runners, is that a better running economy is associated with less leg extension at toe-off (Moore et ah, 2012b; Williams and Cavanagh, 1987; Moore et ah, 2014b; Cavanagh et ah, 1977). This can be achieved through less knee extension, less ankle plantarflexion or a combination of both. It is hypothesised that having a degree of flexion at this point of the gait cycle may allow the muscles to operate at favourable lengths to produce force or that this places the leg in a favourable position going into swing, reducing the degree of movement required and reducing the moment of inertia of the leg.

When considering the initial ground contact from an economical perspective, an individual’s habitual foot strike does not appear to be detrimental to their running economy at any speed, so targeting this gait characteristic for endurance runners to aid performance improvements is therefore not supported by evidence. Indeed, for male and female rearfoot strikers, imposing a forefoot strike on them appears to worsen running economy at slow and medium speeds (Gruber et al., 2013b).

Much attention has been directed towards understanding the lower limb contributions to running and running economy, with less known about the torso, head and upper extremities. Hinrichs (1987) was the first to quantify a rotational coupling in angular momentum between the upper and lower extremities. Therefore, swinging the arms enables runners to counter the vertical momentum of the lower limbs, as well as facilitating vertical oscillation of the centre of mass and minimising head, shoulder and torso rotation. As a result, restricting arm movement, by holding them behind your back or on top of your head, increases the metabolic cost of running and thus, worsens running economy (Arellano and Kram, 2011). Collectively, owing to the upper and lower limb coupling, it is recommended that runners are encouraged to maintain their natural arm swing.

As discussed earlier, female runners should be considered separately to male runners. In addition to differences in common gait characteristics, females must consider the interaction of these factors with their breast biomechanics. Unfortunately, this is an under-researched field, but some meaningful advancements have been made in recent years (e.g., Milligan et al., 2015; Milligan, 2013). Breasts move in a three-dimensional figure of eight during each gait cycle. A breast impact shock occurs following each foot strike and can lead to many females suffering breast pain whilst running. Given the rotational coupling of the upper and lower extremities, greater breast motion may lead to greater counteracting motion of the lower limbs. There is growing evidence to support this theory, with breast kinematics being shown to affect knee angle (Milligan, 2013), step length (Milligan, 2013), trunk lean (Milligan et al., 2015) and medio-lateral forces (White et al., 2009). Ensuring female runners have appropriate breast support to minimise breast motion serves two potential functions: (1) eliminating or reducing pain associated with running; and (2) reducing unwanted rotational motion at the lower limb and medial-lateral forces that may contribute to increased metabolic cost (worsening of economy).

Individual Assessment for Economical Running

Optimising running form for economy improvements based on current evidence traditionally requires laboratory-based testing (e.g. motion capture and gas analysis systems). For many this may not be available. However, surrogate measures, such as heart rate and video/app analysis, can be used in combination with freely available software (Moore, 2019) or an automated spreadsheet (Moore, 2020) that can predict the gait with the lowest heart rate response. This software and spreadsheet have been developed to support practitioners for performing high- level testing of individuals with minimal equipment.

Using a combination of a metronome and verbal instructions, such as “run to the beat”, practitioners can either manipulate step frequency or ground contact time. To manipulate step frequency, one must first determine the individual’s self-selected cadence using a video app or visual inspection. Then set the metronome to a beat that represents their self-selected cadence ±5% and self-selected cadence ±10%. Example formulas to help calculate these are shown in Table 13.2. To manipulate ground contact time, repeat the first step to determine self-selected cadence. Then set the metronome to the self-selected cadence and instruct individuals using the verbal cues provided in Table 13.3. In our laboratory, we have found that providing appropriate time to familiarise individuals to the protocol is crucial before testing. Manipulating ground contact in this way will enable leg stiffness optimisation to also be determined.

In summary, there is a need to consider female and male runners separately due to the conflicting findings and potential effects of breast motion on economical gait characteristics. We argue that there is no such thing as a perfect running form, particularly regarding economical running, and whilst associations previously identified are informative, they should not be used as a blueprint of how everyone should run. Instead, we strongly recommend undertaking an individual economical running gait assessment by systematically examining the effects of modifying a specific gait characteristic for each runner. This can be done for cadence, stride

TABLE 13.2 Example cadence manipulation equations

(Continued)

(Continued)

Note: Italics denotes author suggestions. Other information is taken from literature.

Camera feedback can be provided by connecting a camera to a screen in front of the individual running. The camera can be placed at die side or behind the individual to provide them with sagittal and frontal (back) views.

length, ground contact time and leg stiffness using video analysis, heart rate and open-access software to predict the mathematically optimal value that minimises heart rate (a surrogate measure for economy).

Gait Retraining and Injury

The most common running-related injuries that occur are in the lower limb and are chronic in nature, in part due to the repetitive loading that accompanies the act of running. Gait retraining has been proposed as both an injury prevention and injury rehabilitation strategy for lower limb running-related injuries. Whilst gait retraining for performance focuses on improving running economy, when used for injury purposes it focuses on redistributing the load a runner experiences. The external forces that act upon the lower limb cannot be eliminated altogether, but in some instances, they can be attenuated by modifying technique and being transferred to other joints and/or structures. Table 13.3 outlines gait characteristics that can be modified through gait retraining and the specific injuries they are designed to address. Female and male runners have been included in gait retraining studies, with no sex-specific differences in outcomes noted to date, meaning in this section ‘runner’ refers to both males and females. This section will cover the most common gait characteristics that have been the focus of research.

Is Injury Prevention Possible by Changing Running Gait?

It is widely accepted that injury is multifactorial. Using gait retraining as a preventative strategy assumes a mechanical basis to the development of the injury, where symptoms result from the changes to a tissue, which are associated with overload to that tissue. Based on this assumption, it follows that if we can change the way the musculoskeletal system is loaded (e.g. frequency, direction, magnitude, rate, variability, point of application), the load experienced by the tissue would not exceed its capacity and symptoms would not occur. In support of this approach, Chan and colleagues (2018) demonstrated that a two-week gait retraining intervention to reduce vertical loading rates in untrained runners, prior to a 12-month period of running, led to a 62% lower rate of injury than if no gait retraining was undertaken. The gait retraining approach applied in this intervention encouraged individuals to eliminate the vertical impact peak force, suggesting a forefoot strike pattern was adopted. Lower loading rates mean vertical stiffness of the body is reduced during impact. If a rate-dependent relationship between loading and bone injury is present in humans, like it is for other animals (Burr et al., 1985), then lower loading rates could be associated with lower risk ofbony injuries. It is notable that the study by Chan and colleagues (2018) decreased the occurrence of patellofemoral pain and plantar fasciitis but increased the occurrence of Achilles tendinopathy and calf strain possibly due to the higher ankle plantarflexor moments (Nunns et al., 2013) required during forefoot striking. Therefore, individuals with a previous history of calf or Achilles injuries are advised to avoid this type of gait retraining approach. Conversely, individuals with a history of knee injuries may benefit from this gait retraining strategy.

Not all running-related injuries share a clear relationship with load and many injuries may not present with any structural changes to the tissue. The association between tissue changes, such as pathology, and pain is also unclear, with pathology frequently present in asymptomatic individuals (Horga et al., 2019). This calls into question a purely mechanical approach to injury occurrence. Consequently, injury prevention based on the underlying assumption that injury is simply a mechanical issue cannot be universally applied.

In addition, using a gait retraining approach to prevent injuries will likely mean idealising and pursuing a certain running gait. Like performance, and as can be seen in the study by Chan and colleagues (2018), one running gait is unlikely to suit everyone. Whilst these critiques are also true for injury rehabilitation, there is far more supporting evidence that gait retraining reduces pain in symptomatic runners; it may also provide a greater opportunity for a more individually tailored running form to be encouraged.

Rehabilitating Lower Limb Injuries Using Gait Retraining

Pain reduction is undoubtedly one of the most important outcomes for an individual undergoing injury rehabilitation. Gait retraining can be an effective rehabilitation strategy' but can be performed only on individuals who are able to run whilst injured. Modifying running gait has successfully reduced pain associated with patellofemoral pain syndrome (anterior knee pain) (Bramah et al., 2019; Esculier et al., 2017; Noehren et al., 2011; Roper et al., 2016; Willy et al., 2012b), iliotibial band syndrome (Hunter et al., 2014) and chronic exertional compartment syndrome (Breen et al., 2015; Diebal et al., 2012). Understanding which gait characteristics to target requires knowledge of anatomy and how it functions, as well as knowledge of the movement that causes pain. There is also the need to consider the anatomy and how it functions with structures proximal and distal to the site of pain. For example, the hip and/ or foot should be considered when assessing patellofemoral pain. Table 13.3 provides an overview of the different gait characteristics that have been targeted, and we will focus on the three most common ones: cadence, tibial acceleration and hip kinematics.

One of the simplest technique changes that has been employed is an increase in cadence, which is usually undertaken with an individual running on a treadmill contacting the treadmill belt in time to the beat of a metronome. Example calculations to manipulate cadence are provided in the ‘Individual Assessment for Economical Running’ section (p. 192). Typically, between 7.5% and 10% increases in cadence have been used (Bramah et al., 2019; Willy et al.,

• 2016) and are associated with a number of kinetic and kinematic changes (Heiderscheit et al., 2011). For example, the foot is horizontally placed closer to the centre of mass at initial contact, peak horizontal braking and vertical impact forces are reduced, estimated knee reaction forces and work at the knee are lowered, work at the ankle increases and the tibia is more vertically aligned (Heiderscheit et al., 2011; Lieberman et al., 2015). Faded feedback approaches and one-off visits can be used to change cadence. However, it is important to note that large changes to cadence will worsen running economy (Moore, 2016). Encouragingly, even just a 3% increase in cadence can reduce knee pain and, vertical and braking forces during running (Diss et al., 2018), whilst also not being detrimental to running economy (Bruton et al.,
• 2017) . Making incremental changes to cadence until perceived pain has reduced would allow a practitioner to potentially mitigate the magnitude of change in running economy by settling on the minimum cadence increase required to reduce pain.

Peak tibial acceleration, sometimes referred to as tibial shock, appears to be higher in female runners who have a history of stress fractures (Milner et al., 2006). It has since become a focus of gait retraining research. Previous work has shown that reducing tibial acceleration also enables male and female runners to reduce instantaneous and average loading rates (Crowell et al., 2010; Crowell and Davis, 2011), which may be beneficial for reducing patellofemoral pain or general lower limb injury risk. Lowering peak tibial acceleration can be achieved with some simple verbal instructions as well as using real-time biofeedback (see Table 13.3). If tibial acceleration can be relayed in real-time, researchers have used a 50% reduction in peak tibial acceleration as a target value (Crowell et al., 2010; Crowell and Davis, 2011). This can be achieved by some runners, but inter-individual variation exists regarding the level that can be achieved. Such variation is also present when changing cadence and foot angles at initial contact. If tibial acceleration cannot be relayed in real time, the verbal instructions used by others means runners are likely to be exhibiting a flatter foot (midfoot) or forefoot strike pattern to produce lower tibial accelerations (Clansey et al., 2014). Therefore, instructing runners to alter their foot strike pattern will likely achieve the same outcome. Shortening stride lengths has also been shown as an effective way to reduce tibial acceleration in male runners (Mercer et al., 2002).

During the stance phase, the lower limb must attenuate the external force it is exposed to. Several kinematic strategies exist to achieve this. Whilst the largest motion of the lower limb is observed in the sagittal plane, a greater number of frontal plane movements have reported associations with injury (retrospectively and prospectively), particularly at the pelvis, hip and knee. The collapsing of the lower limb during ground contact time can be visually seen and quantified as hip and thigh adduction, knee abduction and contralateral pelvic drop. Some of these kinematics are exhibited in runners who have a history of, or develop, iliotibial band syndrome (van der Worp et al., 2012) and who have patellofemoral pain syndrome (Neal et al., 2016). Changing such a movement pattern can be achieved using gait retraining within a relatively short space of time (2 weeks) (Willy et al., 2012b), and may in fact be superior to trying to do so through a strength-based intervention which typically takes 6 weeks (Willy and Davis, 2011). Along with the time requirements, there are two other reasons to favour adopting a gait retraining approach. Firstly, lower limb muscle strength consistently displays very poor-to-no relationship with running kinematics, meaning higher strength is not associated with less or more movement (Baggaley et al., 2015; Schmitz et al., 2014; Fukuchi et al., 2014). Secondly, increasing lower limb muscle strength does not lead to changes in running kinematics (Ferber et al., 2011; Willy and Davis, 2011). If a practitioner feels there is a need to change frontal plane pelvis, hip and knee kinematics, then verbal instructions and biofeedback, rather than strength interventions, will enable a runner to do so.

In summary, gait retraining has been successfully used to reduce the incidence of certain injuries, yet increase the incidence of others, in addition to reducing the pain associated with several overuse lower limb injuries. Modifying a runner’s cadence and/or tibial acceleration produces several kinematic and kinetic changes that may contribute to redistributing the load applied to the painful body region. Specific frontal plane lower limb kinematics can also be targeted and altered within a relatively short time.

How to Perform Gait Retraining

Gait retraining means providing instructions and/or feedback to an individual to modify specific gait characteristics. It can be undertaken with either a performance or injury-related goal in mind. Depending upon the goal in mind, different gait characteristics may be targeted. Firstly, we must consider how to perform gait retraining, which applies to both performance and injury goals. Gait retraining starts with a systematic assessment of running gait alongside a clinical examination to determine a runner’s goals and needs. A runner’s spatiotemporal characteristics, kinematics and kinetics will be examined alongside their symptoms during gait, which may include pain and when it occurs in the gait cycle, as well as how it changes with gait retraining or changing constraints, such as speed and gradient.

Once an assessment is complete, then goals of gait retraining can be created, and outcome measures can be used to determine effectiveness. In the treatment of injury this may involve reducing symptoms during running, or alteration of provocative movement patterns while measures of economy or performance can be utilised if they are the main goal. An integral part to the effectiveness of any proposed gait retraining intervention is the application of appropriate motor learning principles. While it is beyond the scope of this chapter to cover in-depth coverage of these principles, we will present some key considerations for practitioners.

Developing Verbal Instructions

Developing verbal instructions to facilitate bringing about a modified running gait is a challenge. But, it is important to note, all studies employ a verbal instruction during gait retraining, sometimes as the only way to modify gait and other times in combination with biofeedback and/or self-modelling (see Table 13.3). First, a practitioner must decide on which gait characteristic they wish to modify in an individual. It is preferable to select one characteristic at a time, rather than providing an individual with multiple characteristics to focus on. Then the language used to instruct an individual must be considered. Three main types of instructions are typically used: (1) analog)'; (2) external focus of attention and (3) internal focus of attention. There are both clinical (Franklyn-Miller, 2016) and coaching (Pedley et al., 2017) examples where analogies are used during gait retraining. However, analogies suffer from a similar issue as external focus of attention instructions, in that they may be ambiguous. They may also lead to unwanted gait changes (Phillips et al., 2017) and produce inconsistent gait responses between individuals (Kleynen et al., 2019). The most common types of verbal instructions used are external and internal focus of attention and the underlying theory is discussed next.

Focus of Attention and Constructive Conscious Control

Wulfand Lewthwaite’s (2016) Optimising Performance Through Intrinsic Motivation And Learning (OPTIMAL) theory provides an excellent overview of performance-related motor learning. The authors advocate for the use of an external focus of attention, where one’s attention is directed towards the movement effect (task outcome) during training to promote skill learning and development, rather than the use of an internal focus of attention where attention is directed towards one’s body and its movement (task execution). There is a substantial body of evidence to support this approach e.g., greater throwing accuracy (Lohse et al., 2010), jumping higher (Wulfand Dufek, 2009) or further (Wu et al., 2012).

To date, gait retraining studies have reported employing an external focus of attention verbal cue to manipulate gait (e.g. Chow et al., 2014). However, inspection of the methods (the inclusion of visual feedback alongside the cue) and verbal cues (which include an internally focused section preceding the externally focused section: “we’re aiming to change foot strike, so run quietly” (Moore et al., 2019b) or “run with your knees pointing towards the wall”) suggests the terminology used has muddied the waters between an external and internal focus of attention. This hybrid model of verbal cueing is often used by clinicians, so whilst the internal aspect may be beneficial, the external portion of the cues have the potential to induce unwanted changes to running technique that consequently may increase loading of the painful body region. Indeed, Moore and colleagues (2019b) have shown that by using this type of cue, a hybrid technique response is produced. Meaning individuals produced technique responses somewhere in between the responses of individuals exposed to either an internal (most effective at producing the desired change) or external cue (least effective at producing the desired change).

This hybrid approach supports a body of work that advocates the importance of utilising an internal focus as a mediator for continuous improvement (Toner and Moran, 2015). Specifically, rather than trusting movements will change based on unconscious trial and error, it has been suggested that increasing somatic reflection (bodily awareness) promotes the refinement of, and/or changes in, habitual movement patterns. Therefore, it appears that the theoretical concept of‘constructive conscious control’ is relevant in this context. For example, take a runner who is experiencing a low-level knee pain. Providing a verbal instruction to run without the knee pain, without providing clear, internally focused instructions about what kinematic strategies to adopt (based on clinical assessments) to facilitate this is likely to be unhelpful. Thus, for runners who want to modify gait because of injury, or to improve performance, a clear, unambiguous kinaesthetic reference point (an internal focus) allowing them to refine their running gait in a systematic way may be advantageous. It is important for practitioners to be mindful of the current debate and consider the potential importance and effectiveness of utilising internal focus of attention cues for their clients during gait retraining.

Considerations for the Provision of Gait Training

Literature often uses the same verbal instruction for each runner to allow a hypothesis to be tested. However, within a coaching and clinical setting, individually tailored verbal instructions are encouraged. It is possible that this may mitigate the ambiguity issue associated with externally focused instructions or analogies. Additionally, asking for the individual’s interpretation of the verbal instruction is recommended. This may mean using an iterative approach, where trial and error is used before settling on a given instruction. Providing these instructions every 20-30 seconds during the initial stages of gait retraining balances the need to remind individuals, without becoming too annoying for both runner and practitioner (Moore et al., 2019b).

If possible, a period of learning the modified gait should occur over several sessions. A typical approach is to adopt a faded-feedback paradigm, whereby instructions and/or feedback are gradually withdrawn until none are provided and the individual must run with the modified gait unaided. Faded-feedback over eight sessions has been used to good effect to treat running-related patellofemoral pain (Noehren et al., 2011; Willy et al., 2012b). Single-session gait retraining can achieve desired kinematic changes (Moore et al., 2019b; Bramah et al., 2019), but retention of the modified running gait may be more difficult to acquire.

Biofeedback and Self-Modelling

Gait retraining can be performed using a variety of biofeedback methods, whereby gait characteristics are relayed back to an individual as they run. Table 13.3 provides an overview of the different variables that have been simultaneously calculated and visually displayed to individuals whilst running on a treadmill. Many of the biofeedback techniques employed in research require sophisticated, expensive research equipment, such as three-dimensional motion analysis or instrumented treadmills. Encouragingly, there are some simpler, cheaper alternatives. Frontal plane gait characteristics (e.g. hip adduction, step width) can be viewed using a mirror, whilst sagittal and frontal plane (back view) gait characteristics can be viewed by syncing a video camera with a display screen positioned in front of the runner. A practitioner can then move the video camera to the side or back of the individual to highlight target gait characteristics. Another option is the use of wearable technology. Willy and colleagues (2016) were able to use a global positioning system (GPS) watch and foot pod to determine cadence whilst individuals ran outside. The GPS watch was then programmed to display cadence using a faded-feedback paradigm. Wearable technology can also be used to help quantify workload variables and monitor running rather than relying on self-report measures (Stiles et al., 2018). But, the accuracy and reliability of the technology' used must also be considered, given the scarcity of validated wearables on the market (Peake et al., 2018).

Self-modelling refers to the observation of yourself successfully performing the desired movement. In the context of gait retraining, this can be achieved by videoing the individual producing the specified technique change and asking them to review this video before they go running. This is a form of‘self-modelling’known as positive self-review (Dowrick, 1999). It has been used to reduce anterior knee pain during a cadence-focused gait retraining intervention and increase confidence in ability to perform the modified gait (Diss et al., 2018). However, the authors recommend the need to use criteria-based progression rather than time-based progression, meaning each individual may need a different number of gait retraining sessions to achieve the desired technique changes.

To potentially facilitate the transference of a newly acquired running technique to other constraints such as different surfaces, footwear or gradient, feedforward self-modelling may be useful. This would require an individual to be filmed performing successfully under these different constraints and providing an individual with such videos. This has been shown to be effective in learning more complex, higher-level moves during trampolining (Ste-Marie et al., 2011), but has yet to be undertaken during running gait retraining. Encouraging results from Zhang and colleagues (2019) have shown that gait retraining on a level treadmill with the aim of reducing tibial acceleration did transfer to overground running and uphill running, but not to downhill running. Runners may need to be reminded to pay specific attention to producing the modified gait during the initial stages of gait retraining to help facilitate future retention.

Gait Retraining in Practice

Gait retraining can be conducted both on a treadmill and overground, yet most of the evidence is derived from treadmill running. Using a treadmill has several advantages: (1) runners can modify gait without worrying about maintaining a specific speed; (2) runners can practice the modified gait at a variety of speeds and/or inclines; (3) biofeedback can be displayed in front of a runner and (4) practitioners can easily view and video multiple steps, providing an opportunity to see gait strategies each runner is employing. Facilitating the transference of the modified gait to overground could be undertaken through the use of feedforward selfmodelling. Verbal instructions can also serve as useful reminders that a runner can use when practicing the modified gait during overground running.

Gait Retraining for Performance

Using a biomechanically derived training intervention to enable a runner to change gait characteristics, e.g. ground contact time and cadence, seems to be achievable. Indeed, we advocate this approach above an exercise/conditioning training intervention in the first instance, as there is very limited evidence showing that exercise- or conditioning-based interventions change desired gait characteristics of a runner (Giovanelli et al., 2017; Gomez-Molina et al.,

2018) . Conversely, short biomechanical interventions (<15 days) have shown some promising results; for instance, male and female runners may benefit from 15 to 30 minutes of running at higher cadences for 10 to 15 days (Quinn et al., 2019; Morgan et al., 1994). Morgan and colleagues (1994) reported that runners were able to successfully modify cadence towards their optimal cadence using this type of training approach and improve their running economy. Whilst, these biomechanical changes take about half the training time to produce as physiological adaptations do, we recommend caution with using consistent gait modifications for such long periods of time as it may impart a high cardiovascular and musculoskeletal demand, which may not be suitable for all runners.

Producing greater knee and/or ankle flexion at toe-off could be targeted for potential improvements in running economy. A verbal instruction such as ‘keep your knee slightly flexed at push-off’ (internal cue) or ‘drive forwards’ (external cue) may lead a runner to modify their toe-off leg orientation. This will need to be monitored visually by recording the runner in the sagittal plane and ensuring the desired technique change is being produced.

Gait Retraining for Injury Rehabilitation

Here we will outline two scenarios for using gait retraining in practice for lower limb injury rehabilitation. Scenario A is a runner coming into the clinic for a one-off visit. Scenario В is a runner engaged in a treatment programme that includes multiple visits to the clinic over a 1-month period.

Scenario A: A male runner presents with patellofemoral pain during running, and assessment reveals several gait characteristics which may be increasing load on this body region: the runner has minimal knee flexion at initial contact and their foot lands a large distance in front of their hip, they exhibit high peak knee flexion at mid-stance and a kinetic assessment reveals a high peak braking force. His self-selected cadence running at 12 knvh^-1 was 158 steps per minute.

Increasing step rate by 10% has been found to reduce overstriding and peak knee flexion (Heiderscheit et al., 2011; Lenhart et al., 2014) as well as being associated with decreased peak braking force (Napier et al., 2019). Recent research has found that just one session of implementing this strategy' can be effective in runners with patellofemoral pain (Bramah et al.,

2019) . In this runner’s case we may experiment with 3%, 5%, 7.5% and 10% increases in cadence within the session to determine a strategy that achieves our goals without increasing effort excessively (see Table 13.2 for example calculations). A metronome could be used to create this change, and a GPS watch could provide additional feedback to determine whether the increase was achieved.

Videos would be taken, and kinetic data collected while running with these cues to examine changes in kinematics and peak braking force. Further feedback is then provided to optimise the changes. If kinetic data are not available, videos would suffice for this intervention.

In the case of a single session of this nature, providing a runner with ongoing instruction and feedback is important. The runner could continue to use the metronome while running and monitor his cadence using a GPS watch.

Scenario В: A female runner presents with lateral hip pain, and gait analysis reveals two kinematic findings which may increase load on sensitive tissue: increased peak hip adduction and contralateral pelvic drop during the stance phase on the symptomatic leg.

These findings may be associated with a narrow stride width (Brindle et al., 2014); however increasing stride width can be problematic as a narrow stride is thought to minimise energetic cost (Arellano and Kram, 2011). In a clinical setting, this means that we should avoid increasing stride width excessively, and a runner may need more feedback and multiple sessions to optimise this new running technique.

If the knees brush together as a result of increased hip adduction, then cueing, ‘don’t let the knees touch’ can be a simple and effective strategy' to reduce hip adduction. Instructing the runner to ‘run wider’ or to run on either side of a line on a track can also be effective. In addition, increasing step rate has also been found to reduce peak hip adduction, so that can be considered too (Bramah et al., 2019). In our experience, verbal feedback when applying these cues, alongside visual feedback from a mirror or video, is especially important when making these changes, and several sessions may be required with gradually reducing feedback.

In summary, we argue that there is a need to develop constructive conscious control during gait retraining to enable an individual to acquire the desired technique changes to modify their gait. Using an internal focus of attention may facilitate this to a larger extent than an external focus of attention or analogy will. Devising individually tailored verbal instructions is recommended, along with criteria-based progression, whereby individuals are only progressed to another instruction and/or phase of gait retraining if they have achieved the desired technique change. Biofeedback has shown promising results, particularly when included as part of a faded-feedback paradigm, but traditionally biofeedback has required sophisticated computer algorithms and equipment to create. Wearable technology is allowing biofeedback to be undertaken ‘in the field’, yet currently only selected characteristics can be quantified precisely (e.g. cadence). Self-modelling is encouraged to enable individuals to review success- fill gait retraining performances to facilitate learning.

Gait Retraining Using Barefoot or Minimalist Footwear

Adding mass to the foot during running will worsen running economy. For every 100 g added to each foot, running economy will worsen by 1% as the metabolic cost of running increases (Frederick, 1984; Fuller et al., 2014). When mass is added to the lower limb, it increases its moment of inertia (resistance to move). The further away from the hip (point of rotation) that the mass is added, the greater the effect on moment of inertia, and thus a more pronounced effect on running economy is observed (Myers and Steudel, 1985; Martin, 1985). However, a review by Fuller and colleagues (2014) reported that running in a shoe weighing less than 220 g (or 440 g per pair) would not lead to a significant worsening of running economy compared to running barefoot. This makes the choice of footwear an important consideration for a runner from an economical perspective, with lighter shoes or no shoes at all (barefoot) being superior.

It is important to note that barefoot running is seen as a specific running form, and minimalist footwear is often recommended as a way of achieving this running form whilst providing protection to the sole of the foot. This means the possible benefits that may be associated with running barefoot or in minimalist footwear are argued by some to not just be a function of removed mass but are due to the ‘barefoot running form’ being adopted. This is typically described as: reducing your stride length and thus, increasing stride frequency, producing greater knee flexion upon initial contact with the ground, and striking the ground with the ball of your foot (forefoot striking or most likely toe-striking). The reduction in stride length appears to come from the foot being brought closer to the centre of mass at initial contact as the knee flexes more and the tibia rotates forwards, placing the tibia in a near vertical position. Additionally, a forefoot or toe-striking pattern occurs, and the ankle is plantarflexed at initial contact. The combination of the kinematic changes means the effective mass being accelerated upon ground contact is reduced, and there is a distinct lack of vertical impact force in the force-time curve (Lieberman et al., 2010).

Unfortunately, when barefoot running is often discussed, the notion of it being about a running form is overlooked or misunderstood. Simply taking your shoes off will not mean a runner will adopt the running form outlined here; there are of course many factors at play. Factors that influence responses to being barefoot are: level of somatosensory feedback (Moore et al., 2014b), pain tolerance, surface stiffness (Allison et al., 2013), and the speed and time a runner is exposed to being barefoot (Moore and Dixon, 2014). The first two factors, somatosensory feedback and pain tolerance, are why exposing individuals to barefoot running may be a useful therapeutic tool during gait retraining and injury rehabilitation. This is because individuals may subconsciously adapt their running kinematics to reduce pain and attenuate forces being applied to the musculoskeletal system. Holistically, this could be a mechanism to explore how an individual ‘self-protects’ and tries to redistribute lower limb load.

It would appear intuitive that increasing the variability of how individuals modulate load distribution through technique alterations may increase the tissue capacity of the musculoskeletal system, in so much as parts of the musculoskeletal system are now being exposed to higher levels of load than perhaps they would during a runner’s normal gait. Exposure to variable loading patterns may explain why runners who use two or more pairs of different trainers during their running programme have a lower rate of injury than those who use just one pair of trainers (Malisoux et al., 2015). Variability can be induced in several other ways (e.g., changing terrain and gradient); however, if practitioners are turning to barefoot running to prevent injury or improve running economy, verbal instructions may be needed to facilitate learning the barefoot running technique.

Regarding injury risk, barefoot runners have similar injury rates to runners who wear trainers (shod runners). However, the most commonly injured body regions differ. Barefoot runners tend to injure the plantar surface of the foot, the calf musculotendon unit and the ankle, whilst shod runners tend to injure their hip, knee and plantar fascia (Altman and Davis, 2016). Load distribution during barefoot and shod running supports these different trends. Barefoot running produces greater ankle plantarflexion moments and ankle joint work compared to shod running, which produces greater knee flexion moments and knee joint work. If runners are interested in transitioning footwear i.e. from shod to minimalist or barefoot, a phased transition should be undertaken over several weeks. This should include gradual exposure to the footwear or removal of it, in addition to exercises to condition the musculotendon units that will be exposed to greater loads. However, it remains debatable whether such a change is likely to lead to improvements in running economy or performance and if these outweigh the notable risk of injury during transition.

Summary

• • We recommend undertaking an individual economical running gait assessment rather than drawing upon cross-sectional, correlational studies that appear to present an ‘economical running technique’. Pursuing such a running form assumes there is a perfect running form that should be targeted for all individuals, which is not the case.
• • Strengthening exercises and the retraining of other movements (e.g. single leg squats) do not translate to gait modifications. Therefore, if a practitioner wishes to target a specific running gait characteristic, gait retraining is recommended. Gait retraining can improve running economy and reduce pain within 15 days, whilst physiological adaptations are shown to take much longer.
• • Common gait modifications for injury rehabilitation that can be targeted outside of a research laboratory are: an increase in cadence, a reduction in tibial acceleration, increasing hip adduction and reducing contralateral pelvic drop
• • Internal focus of attention verbal cues are more effective at producing desired technique changes, potentially by invoking constructive conscious control of one’s gait to mediate improvements of a habitual movement pattern.
• • Lightweight footwear (<220 g per shoe) or barefoot running will incur a lower metabolic cost than running in footwear >220 g per shoe. The overall rate or injury is similar between barefoot and shod running, but specific injuries may become more or less common. Practitioners should also be mindful that not all individuals will adopt a ‘barefoot running form’ by simply taking their shoes off, and therefore the desired gait modifications may not be achieved.

14