The ankle-foot complex plays a crucial role in human walking due to its contributions to both balance and forward propulsion [1]. Plantarflexor muscle weakness, due to neuromuscular disorders such as stroke, can disrupt the biomechanics of gait, leading to slower walking speeds and a less efficient gait pattern [2,3]. To facilitate mobility and an efficient walking pattern, ankle-foot orthoses (AFOs) may be prescribed.

There exists a wide range of AFOs in a variety of shapes, materials, and stiffnesses, with varying effects on patient gait biomechanics [4,5]. These design parameters are often used to optimize biomechanical outcomes for individual patients. For example, increasing the bending stiffness of an AFO can reduce knee flexion during the second rocker in patients with crouch gait [6]. Similarly, increasing footplate stiffness can maximize energy storage and return, reducing the energetic cost of walking [7].

Adapted footwear such as rocker shoes have been investigated independent of AFO use and shown to affect gait kinematics and kinetics [8,9]. As AFOs are typically worn with footwear, it is essential to consider how footwear design interacts with AFOs to influence gait. When combined, footwear and AFO form an integrated orthotic system where the shape and bending stiffness of both devices play a combined role in influencing how forces are transmitted through the lower limb during gait. Consequently, footwear adaptations can alter the biomechanical effects of AFOs on gait. Despite these interactions, the specific impact of different footwear parameters on AFO performance remains unclear [10]. The footwear design can either complement or counteract the intended mechanical function of the AFO. For instance, when the ankle is immobilized with a rigid AFO, increasing the heel-sole differential (HSD) can increase knee flexion during midstance [11]. Conversely, a rocker shoe with a proximal apex position may shorten stance phase duration [12] and reduce the energy storage and return of a carbon fiber AFO, potentially negating its benefits.

Although the importance of the interaction between AFOs and footwear is recognized, most existing research has focused on sagittal plane alignment, particularly the shank-to-vertical angle (SVA), and often within specific patient populations [4,10,11]. While systematic reviews have examined the biomechanical effects of footwear or AFOs independently [13,14], no systematic review to date has examined the biomechanical effects resulting from the interactions between ankle-foot orthosis and footwear combinations (AFO-FCs) in adults. The aim of this review is to provide a systematic overview of the current evidence on AFO-FCs. Specifically, it evaluates how footwear modifications, when paired with AFOs, influence gait biomechanics in both healthy adults, who provide insight into gait adaptations as controlled models, and patients with gait pathology. These findings are intended to inform clinical recommendations and identify directions for future research. A background on ankle-foot biomechanics and the theoretical effects of AFO-FCs in cases of dorsiflexor and plantarflexor weakness is first introduced to provide context for the findings of this review. Existing literature on AFO-footwear interactions in both patient populations and healthy adults is then synthesized to establish an evidence-based overview of the biomechanical outcomes resulting from AFO-footwear interactions.

Human locomotion can be broken down into 2 main phases, stance and swing. The stance phase is typically defined as the moment the heel of the foot strikes the ground until the foot leaves the ground again [1]. During stance, the ankle-foot complex can be likened to the rolling of a wheel as the center of pressure (COP) moves from heel to toe [15]. Within stance, 3 subphases, known as rockers, can be distinguished as the COP progresses from heel to ankle to forefoot in a curved path relative to a fixed lower-leg coordinate system [16]. The period between initial contact and the forefoot touching the ground makes up the first rocker, also known as the heel rocker. Following this is the second, or ankle, rocker, where dorsiflexion takes place, and the foot remains flat on the ground. The third, forefoot rocker, begins as the heel leaves the ground and concludes with toe-off. The 3 rockers are illustrated in Figure 1.

Together, these 3 rockers make up part of a phenomenon known as the “roll-over shape” (ROS). This term refers to the ellipsoidal trajectory of the COP relative to a fixed frame of reference at the ankle [17]. Figure 2 illustrates the transformation from the laboratory reference frame (A) to the ankle-foot reference frame (B). The ROS provides a useful framework for evaluating gait adaptations to AFOs, as it can demonstrate changes in timing and kinematics, providing insight into gait adaptations with a single measure. Figure 3 provides examples of the ROS for healthy individuals (A), two theoretical ROS for patients with calf muscle weakness (B), and visualizations of the ROS when walking with a rigid AFO, both without (C) and with an increased HSD (D). In healthy walking, the ROS is an arc that maintains a relatively constant radius even when walking speed or body weight is altered, due to physiological adaptations of the ankle-foot complex [17,18]. This consistent shape provides a predictable rollover, smoothing step-to-step transitions, allowing humans to walk with minimal energy expenditure [15,19]. Individuals with calf muscle weakness may have a shortened or flattened ROS of the ankle-foot complex compared to the typical arc [20]. Dorsiflexor weakness primarily affects the first rocker, while plantarflexor weakness impairs forward progression of the COP during the second rocker and reduces push-off power during the third rocker [1].

AFOs are commonly prescribed to support patients with calf muscle weakness during gait [3,10,22]. While both passive and active AFOs (exoskeletons) exist, this review focuses on passive AFOs, which are more typically used by patients in daily life. Passive AFOs range from solid polypropylene designs to energy-returning carbon fiber leaf springs and can be tailored to control ankle joint motion to restore a more optimal gait pattern. However, while AFOs can improve stability and help re-establish a smooth ROS [20], their increased stiffness and limited range of plantarflexion motion may reduce push-off power and limit gains in energy efficiency [23].

Footwear plays an important role in gait biomechanics, with design features such as the HSD, rocker soles, and heel hardness influencing how the COP progresses through each rocker relative to the ankle coordinate system [16,24,25]. When combined with AFOs, footwear characteristics can influence orthotic effects. Together, AFO-footwear adaptations offer a powerful means to restore the ROS and improve gait biomechanics in patients with plantar- or dorsiflexor- muscle weakness. However, their effects may differ across the 3 rockers, leading to complex interaction effects between AFO-FCs and gait biomechanics. The following sections summarize the biomechanical effects of the most commonly used AFO-footwear adaptations in the sagittal plane.

During the first rocker, the dorsiflexors generate a resistive moment that controls plantarflexion, gradually lowering the forefoot to the ground [1]. This negative work from the dorsiflexors leads to smooth knee and hip flexion and supports the controlled transition of the center of mass (COM) from one stance limb to the next [19].

When plantarflexion-restrictive AFOs are used to compensate for dorsiflexor weakness, this gradual transition of the COM is disrupted [26]. The reduced dorsiflexor contribution to the energy transfer between limbs and absence of controlled plantarflexion motion increases angular momentum at the knee and hip. This can result in rapid COP progression toward the midfoot, accompanied by increased knee and hip flexion during early stance [27]. Certain footwear adaptations may mitigate the adverse biomechanical effects of absent plantarflexion during the first rocker. For instance, a shoe with a rocker distal to the heel positions the COP more anteriorly at initial contact [28]. This reduces the ground reaction force (GRF) moment arm at the ankle and knee [28], reducing angular momentum and potentially promoting a smoother COM transition in patients using rigid AFOs. In contrast, a rocker proximal to the heel with a small radius combined with a rigid AFO could theoretically increase the GRF moment arm, thereby increasing angular momentum at the knee and hip.

In healthy gait, once the foot is fully in contact with the ground, the gradually increasing plantarflexor activity generates a plantarflexor moment about the ankle joint, decelerating shank inclination and knee flexion [29]. During this phase, near-isometric activation of the plantarflexors combined with ankle dorsiflexion stretches the Achilles tendon, allowing elastic potential energy to be stored and later released as push-off power in terminal stance [29].

Throughout the second rocker, gradually increasing plantarflexor activity shifts the COP toward the forefoot. This progression lengthens the GRF moment arm relative to the ankle joint, contributing to knee and hip extension during terminal stance.

In the case of plantarflexor muscle weakness, insufficient force generation at the ankle may lead to rapid and excessive shank inclination, ankle dorsiflexion, and knee flexion, as well as increased knee and hip extensor moments [1,3]. Excessive knee flexion impairs elastic energy storage in the Achilles tendon and reduces the moment-producing capacities of the ankle plantarflexors, resulting in a flatter ROS and reduced push-off power [20]. A dorsiflexion-resistant AFO can help modulate the rate of forward shank inclination by providing an external plantarflexion moment. This reduces excessive knee and hip flexion, leading to a smoother ROS [3]. Positioning the forefoot apex more distally can prolong the second rocker and extend COP progression [30], potentially increasing external plantarflexion moments and hip and knee extension in AFO users. Conversely, a more proximal apex shortens the second rocker, leading to restricted COP progression and reduced knee extension and plantarflexor moments [12,31].

During the third rocker, the heel lifts off the ground and COP progression and ankle plantarflexor moments approach their peak. At this stage, body weight rapidly transitions from single-limb to double-limb support and elastic energy stored in the Achilles tendon is rapidly released, generating positive ankle power and facilitating a smooth redirection of the COM into the subsequent stance phase, as well as rapid swing initiation of the trailing limb [1]. Adequate timing and sufficient magnitude of peak ankle power ensure rapid rotation of the shank-foot system over the forefoot [19].

The increase in plantarflexion moment during the second rocker contributes to forward progression of the COP and a gradual increase in ankle angular velocities [1]. As the COP progresses anteriorly toward the forefoot, it approaches the shoe’s forefoot apex, which effectively acts as a rotational axis for the foot-shoe system. When the COP reaches this apex position, the external dorsiflexion moment generated by the GRF can no longer balance the internally generated plantarflexion moment, resulting in rotation of the foot about the forefoot apex and heel lift-off, marking the onset of the third rocker.

AFOs with energy-storing elements (such as springs or carbon fiber elements) can store mechanical energy during the second rocker and release it at push-off (third rocker) to aid propulsion. A more distal apex position could delay the onset of the third rocker, increasing COP progression, peak GRF moment arms, and ankle dorsiflexion, leading to greater energy storage in spring-like AFOs and enhanced push-off power when the energy is released in the third rocker. Additionally, changes in rocker radius can modulate timing and duration of foot-to-ground contact and shank kinematics during the third rocker, where a smaller radius accelerates shank inclination and energy release (smaller ROS arc), while a larger radius decelerates it, prolonging the third rocker phase (larger ROS arc). Table 1 summarizes the theoretical biomechanical effects across all 3 rockers for 3 types of footwear adaptations when used with an AFO that restricts both plantar- and dorsiflexion.

This systematic review was registered with PROSPERO (International Prospective Register of Systematic Reviews; CRD42024625819) and is reported in line with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline [32].

A systematic database search was performed by the primary reviewer using PubMed, Embase, Scopus, and Web of Knowledge. The search was performed using the following phrases: ankle foot ortho*, AFO, AFO-FC, orthotic*, orthosis, orthoses, splint, combined with biomech*, gait, locomotion, walk*, kinemat*, and adaptation, shoe, footwear, rocker, roll-over, insert, wedge, and tuning. Papers were retrieved from the beginning of the respective database history until December 2024. A follow-up search was conducted in May 2026. The complete search string for Scopus is:

TITLE-ABS-KEY((“ankle foot ortho*” OR “orthotic device” OR AFO OR “AFO-FC” OR (ankle AND (orthotic* OR orthosis OR orthoses OR splint))) AND

(biomech* OR gait OR “gait analys*” OR locomotion OR walk* OR kinemat* OR tuning OR manipulation) AND (shoe* OR footwear OR rocker* OR “roll-over” OR insert* OR wedge* OR tuning))

The remaining search strings, including database-specific keywords, can be found in Multimedia Appendix 1.

Titles were excluded if they mentioned nonadult participants, animal participants, or cadavers. Review papers, protocols, or papers that investigated only movements other than walking, surgical techniques or surgery-related outcomes, or devices that were not AFOs, lower leg casts, or footwear were also excluded in the title exclusion process. Rigid lower leg (plaster) casts were considered to behave in a similar manner to rigid AFOs and were eligible for inclusion.

Abstracts were excluded if the abstract did not mention both footwear and AFOs and did not describe effects of footwear adaptation on gait while using an AFO. The abstract must have described walking with an AFO in at least two footwear conditions. Abstracts were excluded when footwear was manipulated in only the leg contralateral to the AFO.

Papers written in English, French, German, or Dutch were eligible for full-text selection. Inclusion criteria required that the study described or visualized the AFOs, footwear, and footwear modifications. To qualify, the footwear adaptation must have been applied externally to the AFO, either between the AFO and the shoe, the shoe itself, or between the shoe and the ground. Additionally, at least one of the following outcome measures must have been examined: sagittal plane lower-limb joint kinetics and kinematics, sagittal plane whole-body COM kinetics and kinematics, spatiotemporal walking parameters, muscle-tendon dynamics, lower-limb EMG (electromyography), COP, GRF, plantar pressure, or metabolic cost.

Screening was performed with the use of Catchii, a systematic review management tool [33]. Two reviewers independently screened papers at each stage of the inclusion process. Reviewers AM and CG performed the title and full-text screening, and the abstract inclusion process was performed by AM and JMH. Disagreements were resolved through discussion between each pair of reviewers. For each stage of the selection process, papers were excluded based on the first exclusion criterion that was met, and only that criterion was recorded.

The quality of the included research was assessed using a modified Downs and Black Checklist [34]. This version of the checklist scored question 27 binarily, rather than out of the original 5 points, and is more suited to observational studies. Questions regarding follow-up were omitted from the checklist as gait studies are typically within-participant designs without follow-up. The final checklist was scored out of a potential 25 points. Adjusting for the omitted questions, a study could be considered “excellent” quality if it scored 23-25 points, “good” between 17 and 22 points, “fair” between 12 and 16 points, and “poor” if marked less than 12 points [35]. Only studies scoring “fair” or higher would be considered to have sufficient quality for data extraction.

A modified version of the AFO reporting checklist developed by Ridgewell et al [36] was used to evaluate the quality of AFO reporting. The checklist evaluates 3 categories: participant details, AFO details, and the experimental protocol. This checklist was developed for studies on children with cerebral palsy, and as such the section on participant details was excluded for the purpose of this review. Scoring followed the guideline provided by Ridgewell et al [36]. Reviewers AM and CG independently scored both checklists. Any disagreements were resolved through discussion.

Data were extracted by AM. A custom checklist was used to extract footwear design parameters. This checklist includes design descriptors such as outsole hardness and apex position but does not include manufacturing items such as last type or sole material. Only outcome measures listed in the inclusion criteria were included in this review. Data were systematically synthesized using a structured biomechanically informed framework. Data were synthesized first by footwear adaptation (eg, HSD), then by the phase of gait in which outcomes were measured, and finally by the specific outcome measure. For clarity purposes, all moments are reported as internal moments, with any values reported as external moments transformed by AM.

The systematic database search, including the updated search, resulted in 3064 papers after duplicates were removed. A total of 118 full texts were evaluated for inclusion in the review. Of these, 9 papers were written in German, and 1 paper in French, all of which were excluded due to not being primary research. A total of 14 studies were included in the review, following the inclusion procedure outlined in Figure 4.

An overview of the participants, interventions, and study designs can be found in Table 2. The majority (10/14) of studies were within-subjects designs evaluating hemiparetic patients after stroke using solid dorsal AFOs.

A total of 204 participants (108 healthy, 84 patients after stroke, and 12 with unilateral ankle weakness or pain) were included across all 14 studies. Seven papers examined AFO-footwear combinations in patients with hemiparesis secondary to stroke [37,38,40,43,44,47,48]. Hemiparetic participants ranged across studies from 11 months to over 10 years post stroke onset. Six studies recruited only healthy participants [28,39,41,42,45,49], and one study enrolled both healthy controls and participants with unilateral foot or ankle weakness or pain [46]. Of the 8 studies that included patients, 5 studies recruited participants who used AFOs in their daily lives [37,38,43,46,47], 1 study included only participants who had not used AFOs in their daily lives [48], and 2 studies did not specify whether AFOs were used in daily life [40,44]. Nine studies included gait speed as an outcome measure [28,37,38,40,42-44,47,48]. Three studies let participants walk at a self-selected walking speed and did not report it as an outcome. Of these 3 studies, one study controlled the speed between the trials [45], and 2 studies did not specify whether speed was controlled between conditions [41,49]. One study set a target gait speed based on the participants' leg length and a Froude number of 0.16 and used this for all conditions [46]. One study did not report gait speed as an outcome or specify if gait speed was controlled [39].

A summary of the AFOs included in this review and their properties can be found in Table 3. A total of 19 different AFOs were used across the 14 included studies. Eight of the AFOs used a solid dorsal shell design constructed out of polypropylene [37,38,40,42-44,47]. One study used 2 AFOs with a “plastic shell,” which are assumed to be polypropylene dorsal shell AFOs, one with plantarflexor resistance and one with a plantarflexion stop [48]. Four AFOs were carbon-fiber, of which 2 AFOs had ventral shells [41], 1 AFO a proximal cuff [46], and 1 AFO a dorsal shell [45]. While a walking boot was also investigated by Kearney et al [41], results from the cast were not included, as it did not meet the inclusion criteria of having the footwear adaptation be between the AFO and the shoe or ground. One AFO had a hinged ankle joint with adjustable medial and lateral struts which attached directly to an experimental shoe [28]. One study used 2 AFOs with the same design with either a medial or lateral strut [49]. These 2 AFOs were counted as 1 AFO in Table 3 as they were otherwise identical. The AFO had free dorsiflexion and plantarflexion but did not permit frontal plane motion. No information about any parameters other than strut position was given or possible to infer for this AFO. This study also examined a stirrup brace which restricted frontal plane movement and was considered an AFO. A plaster cast [39] was reported and categorized as an AFO for the purpose of this review. Including the cast, 11 AFOs had full-length footplates [37,39-42,44,46,47]. The length of the footplate was not stated or possible to infer from images for 7 AFOs [38,43,45,47-49]. Eight of the footplates were rigid [37-39,42,44,45,47], one was flexible throughout the footplate [45], and one was flexible at the metatarsophalangeal joint [43]. Two AFOs had indeterminable footplate stiffnesses, with the only description given as one AFO being “more flexible,” and the other AFO being “less flexible” [41]. The footplate stiffness was not given or impossible to infer for 7 AFOs [28,40,47-49].

The footwear adaptations used for each study can be found in Table 4. Footwear adaptations are reported in the same manner as the respective study. Seven papers investigated adaptations at the heel of the shoe. The majority of these focused on increasing the HSD, either with insole wedges [41,45,46] or through altering the heel height of the shoe itself [47]. An overview of these adaptations can be examined in Table 5. One study used insoles to manipulate the SVA, but did not specify the shape, height, or location of these insoles [38]. One paper examined the effects of changing the shape and hardness of the heel of an experimental shoe [28]. One paper examined the effects of increasing the HSD using wedges, which were then translated to the design of a custom shoe and combined with a stiff rocker outsole [43]. Design parameters for that rocker outsole were not given; however, from the image provided, it can be inferred that an apex position proximal to the metatarsophalangeal joint was likely used.

Three studies compared standard shoes to rocker shoes [40,44,48]. All 3 studies used the same forefoot rocker design of an apex position at approximately 65% of the length of the shoe from the heel, and a rocker angle of 15° with a curved forefoot and stiff outsole. Farmani et al (2016) [37] studied the effects of adding a rocker sole directly to the footplate of a rigid polypropylene AFO compared to the same AFO with only a layer of nonslip rubber added to the footplate [37]. This rocker sole also had an apex position at approximately 65% of the shoe length with a beveled forefoot with a rocker angle of 15°, and as such is considered functionally a rocker shoe for the purpose of this review.

Dhalla et al [39] examined the effects of different types of cast shoes when used with a rigid walking cast. Schmalz et al [49] examined the effects of mediolateral wedging, in this instance applying the wedges externally to the shoe. It is unclear whether the footplate of the AFO and wedge were the same length. A study from Arvin et al [42] compared a conventional solid dorsal AFO with a shoe to an AFO of the same manufacturer but with a shank inclined in dorsiflexion without a shoe. Both the traditional AFO with a shoe and the modified AFO worn without a shoe had the same SVA of 5°-7° inclination [42].

This review used a biomechanical framework to synthesize evidence on the interacting effects of footwear parameters and AFOs on gait outcomes in patients and healthy young adults. While it is evident that footwear adaptations can optimize gait biomechanics beyond what can be achieved by an AFO and standard shoe alone, large knowledge gaps remain, preventing empirical validation of many theoretical and biomechanical assumptions and adoption in clinical practice. The effects of plantar pressure [39,41] will be omitted from the discussion to focus on kinetics and kinematics.

Participants both with and without pathologies were included in this review, as healthy walkers can provide a valuable model for the biomechanical effects of AFO-FCs. Indeed, in the few studies that examined the same AFO type, footwear adaptation, and outcome measure, it was found that both patients after stroke and healthy walkers have increased knee flexion during the second rocker when the HSD is increased [38,43,45]. However, this heterogeneity in participants does require discretion when interpreting the results, as adaptations may not be transferable to all patient populations.

A wide range of AFO designs were included in this study; however, the majority (11/19) could be generalized as rigid AFOs which restrict both plantarflexion and dorsiflexion and can be assumed to have negligible differences in their mechanical properties when interpreting sagittal plane gait variables.

A very specific set of exclusion criteria was used for this review. While the number of included studies reflects similar reviews and the existing knowledge gap, a broader inclusion protocol could have been beneficial. For instance, the inclusion of papers that did not use an AFO for all conditions but had multiple footwear adaptations could still have provided insight into the effects different footwear adaptations may have on biomechanics. Furthermore, data were extracted by a single author, and while the included results were examined by the other 2 authors of this review, it is possible that some data were overlooked or inconsistently collected during the inclusion process, introducing a risk of subjective bias and potentially reducing the reproducibility of this review’s findings. As all studies included at least one significant finding, there may be a risk of publication bias. However, due to the design of the included studies, a formal assessment of publication bias was not performed.

While the papers included in this study were mostly of “good” quality, none could be considered “excellent.” This is largely due to no studies being blinded; however, due to the physical nature of the interventions, it would be difficult to truly blind participants and researchers. It is important to note that 4 of the 5 studies using forefoot rockers belong to the same institution. While each paper from that group reported different outcome measures, the same rocker parameters were used for all 4 papers.

A major limitation of this review is the lack of empirical descriptors given for interventions. Without empirical descriptors, it is difficult to draw scientifically sound conclusions from the presented outcomes. Many AFO details were inferred, or the details provided were too vague to derive conclusions from. For instance, 2 AFO designs were described as “more stiff” or “less stiff” than the other [41]. As AFO stiffness has a significant impact on knee and ankle kinematics, whether that difference in stiffness is 1 or 5 Nm/deg can have a substantial influence on gait outcomes [53]. Compounding this issue, very few studies gave sufficient detail about their adaptations, and there was often minimal description of the footwear being used. This lack of transparency makes it difficult to replicate findings or understand the specific mechanical contributions of the footwear-AFO combinations being tested.

The results demonstrate that footwear adaptations can be used to optimize AFO effects for individual patients. However, while the biomechanical effects of increasing the HSD with a rigid AFO are well explored, there exists limited information regarding the effects of other outsole design parameters and compliant AFOs. Further understanding of the interactions between AFOs and rocker outsole parameters is required to maximize AFO effectiveness across different patient populations. Future research should investigate the effects of varying forefoot parameters, including rocker radius, apex position, and footplate stiffness, in combination with a wider range of AFO designs including energy-storing AFOs. Explicit and consistent reporting of AFO and footwear design properties is necessary to make meaningful conclusions on biomechanical outcomes.