Heel-to-toe drop of running shoes: a systematic review of its biomechanical effects

Abstract Heel-to-toe drop (HTD) values may modify the running biomechanics. However, more than twenty additional footwear characteristics may contribute to this too. The aim of this review was to identify and systematise the specific effects caused by different HTDs on running biomechanics. A systematic search was carried out in seven online databases and Footwear Science according to PRISMA protocol for studies that included the practice of endurance running in running shoes with different HTD types. A modified Downs and Black checklist was used to assess the risk of bias. Characteristics of the studies and footwear, the equipment used, and biomechanical outcomes were extracted for qualitative synthesis. Twelve studies were included. Only one had a randomised control trial design, which was classified as ‘good’ (score = 24) quality. The studies reported thirty-nine kinematic and sixteen kinetic variables. HTD ranged between −8 to 16 mm. HTDs did not produce modifications in contact and flight time, stride frequency, and stride length. Some controversial evidence supports that the foot strike pattern changes towards forefoot strike only with HTD 0 compared to HTD 8–10. All evidence indicates that HTD values modified neither ankle, knee, or hip kinematics. Greater evidence supports that the modification of HTD in running shoes does not modify the values of GRF. Despite the controversial nature of the data, the trend is that a lower HTD shows a higher vertical loading rate. The different HTDs modified neither of the joint moments. The negative HTD could be an interesting variable to consider for future research. In addition, it is necessary to study the relationship between the HTD values and different combinations of heel height and forefoot height.


Introduction
The history of athletic footwear design has traced a cyclical path. Until the mid-1960s, the vast majority of running footwear had features that could be classified today as 'minimalist footwear' (Coetzee et al., 2018), (i.e. highly flexible with low cushioning and absence of control features) (Davis, 2014). Subsequently, and just before the beginning of the 'running fever' of the '70s, New BalanceV R introduced the Trackster model, which included a more robust midsole in the heel area. There, it started a technological revolution that resulted in a trend in the design and construction of footwear oriented towards more cushioning and support elements for the foot (Werd et al., 2010). This trend reached its peak during 2013 and 2014, an instance that coincides with the publication of the bestseller novel Born to Run (McDougall, 2009) and of critical studies focussed on the advantages of barefoot running or running with minimalist shoes (Bramble & Lieberman, 2004;Lieberman et al., 2010). These publications have promoted the production and sale of running shoes with a minimalist design; however, there is no conclusive evidence about their benefits (Haman & Dowling, 2015).
The reasons the design of running shoes has changed throughout history have been: 1) to improve performance, 2) to protect against injury, and 3) to provide comfort (Abi an et al., 2012;Lafortune, 2008). Unfortunately, although footwear has been evolving, it has not been possible to prevent running from being an activity with a high incidence of overuse injuries, a value that can reach 29% of runners per year (Jungmalm et al., 2020), or even 42.7% with these injuries being located primarily in the knee (28%), ankle-foot (26%) and leg segment (16%) joints (Francis et al., 2019). The barefoot running trend has been presented as a solution to the high incidence of overuse injuries due to the support kinetics changes that have been observed (Hall et al., 2013;Perkins et al., 2014), such as the disappearance of the transient impact peak (characteristic of running in shoes) (Lieberman et al., 2010;Sun et al., 2018), variables that could be considered as predisposing factors for overuse injuries (Van Der Worp et al., 2015).
Minimalist footwear has a highly flexible sole, a weight of fewer than 200 grams, a heel height equal to or less than 20 mm, and a Heel-to-Toe Drop (HTD) of 7 mm or less (Coetzee et al., 2018). Whose aim is to emulate the acute effects that have been described during barefoot running. However, it would appear that these changes would not occur primarily due to footwear conditions but instead due to a shift in foot strike pattern. Forefoot strike pattern (FFS) (where the ball of the foot strikes the ground) may be facilitated by running barefoot or/in minimalist footwear (Hall et al., 2013). Thus, it has been shown that shifting towards an FFS produces virtually the same effects as barefoot running (Futrell et al., 2020;Xu et al., 2021). Likewise, the evidence indicates that runners with a FFS are less injured than those who run with a rearfoot strike pattern (RFS) (where the heel strikes the ground first) (Daoud et al., 2012).
According to the above, the relationship between foot strike pattern, minimalist footwear, and barefoot running could be summarised as follows: To decrease the incidence of injuries, modifying the GRF values during plantar support could be convenient. This effect can be achieved by running with FFS. This type of foot strike pattern is favoured for barefoot running (Lai et al., 2020). The use of minimalist footwear is the transition between barefoot running and wearing running shoes. However, there is evidence to refute these assertions. Thus, running in minimalist shoes or barefoot does not always produce changes in foot strike pattern (Hatala et al., 2013). Likewise, these biomechanical changes can be reached through other methods (Barcellona et al., 2017;Dolenec et al., 2015). The individual response is another factor to take into account (Tam et al., 2016).
Due to the relationship between foot strike pattern and footwear condition (Lai et al., 2020), it would seem logical to propose that running footwear design should be oriented towards a progressive use of FFS. HTD is defined as the sole height difference between the rearfoot and forefoot (Esculier et al., 2015) (Figure 1). Some evidence indicates that lower HTD values modify the foot strike pattern (Chambon et al., 2015;Horvais & Samozino, 2013;Malisoux et al., 2017;TenBroek et al., 2013TenBroek et al., , 2014 and decrease injury risk in occasional runners (Malisoux et al., 2016). However, more than 20 additional footwear characteristics may contribute to changing the biomechanics and modifying injury incidence of runners but are not adequately measured and/or reported in the current literature (Hoitz et al., 2020;Ramsey et al., 2019).
It seems logical to raise the need to study the specific effects caused by modifications in the HTD of running shoes. However, in most studies where HTD has been modified, others such as weight, stack height, stability, motion control technologies, and flexibility have also been changed, variables proper to the definition of minimalist footwear (Esculier et al., 2015). It is for this reason that it is necessary to identify those specific biomechanical modifications that cause different HTD values in running shoes, which allows us to propose as the objective of this review to identify and systematise the specific effects caused by different Heel-to-Toe Drop values of running shoes on biomechanical variables in endurance runners.

Protocol and registration
This systematic review was carried out in accordance with the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) (Liberati et al., 2009). The review protocol was registered on the International Prospective Register of Systematic Review (PROSPERO) website on April 21, 2021 and updated on June 1, 2021 (Registration number CRD42021246208).

Eligibility criteria
The articles considered for the review were chosen according to the PICOS methodology, whose inclusion criteria were: (1) Population: Regular runners over 18 years of age, of both sexes; (2) Type of Intervention: Practice endurance running in sports shoes with different HTD values; (3) Type of Comparison (Control): Practice of endurance running performed with shoes with a different heel-to-toe drop and barefoot running are compared. In addition, it included studies whose primary independent variable was HTD and was stated within the study objective; (4) Type of Outcomes: Biomechanical (kinematic, kinetic or EMG) modifications of endurance running technique; (5) Type of Studies: Randomised control trial (RCT), prospective, retrospective, or clinical case studies, performed in laboratory or field. The exclusion criteria were: (1) Reviews (systematic or narrative) and meta-analyses; (2) Studies that did not include biomechanical variables; (3) Studies that did not state the variable HTD within the study objective; and (4) Studies that used non-athletic footwear (Mcpoil, 1988).

Electronic Search
A systematic search was carried out in the following databases: MEDLINE (through PubMed), Web of Science, Scopus, SPORTDiscus, Taylor & Francis, ProQuest, Springer, and Footwear Science Journal between 05/13/2021 to 05/20/2021. A final search was carried out between 01/10/2023 to 01/18/2023. The search was conducted in English, considering articles published to the date of the review.

Study selection
After removing duplicate studies, using a standard form, three independent reviewers (CSR, VPO, and EHH) proceeded to read titles and abstracts, recording potentially relevant studies. The PDF files of these studies were then retrieved and fulltext reviewed to select those that met the eligibility criteria. A fourth reviewer (CAR) was considered to define the inclusion of a study when there was no initial consensus.

Data collection process
The same three reviewers (CSR, VPO, and EHH) independently used a standard form to record study characteristics, footwear characteristics and biomechanical outcomes from each study.

Search Results
Twelve studies were included in this review, of which only one corresponds to a randomised control trial (Malisoux et al., 2017) (Figure 2).

Study characteristics
The characteristics of the studies analysed are summarised in Table 1. Only one of the studies presented a randomised control trial (RCT) design, whose duration was 6 months, or ended when the subjects completed 500 km of training volume (Malisoux et al., 2017). The other studies corresponded to a cross-sectional (CS) design.
The most used warm-up was running at low or selfselected speed for a period between 5 and 10 minutes. Only two studies did not specify the warm-up performed (Chambon et al., 2015;Mo et al., 2020).
Ten studies have given participants time to familiarise themselves with the experimental footwear, ranging from 20 seconds to 7 minutes. Two studies did not inform if any familiarisation activity was performed .

Quality assessment
Regarding the evaluation of the methodological quality of the included studies, the agreement between the two evaluators was good (Kappa 0.711). Only one study obtained the score to be classified as 'Good' methodological quality (Malisoux et al., 2017), while eight were evaluated as 'Fair' (Besson et al., 2019;Fu et al., 2022; Gij on-Noguer on et al., 2019; Mo et al., 2020;Moody et al., 2018;Yu et al., 2021;Zhang et al., , 2022, and three 'Poor' (Chambon et al., 2015;Horvais & Samozino, 2013;Richert et al., 2019). The main aspects of risk of bias presented by the studies were those related to external validation and internal validation, specifically identifying confounders within their studies (Table 2).

Variables studied and equipment used
The studies measured 39 kinematic and 16 kinetic variables (Table 3). All studies analysed kinematic variables. Two of them did not include kinetic variables in the design or in the results (Gij on-Noguer on et al., 2019; Malisoux et al., 2017).
The most frequently measured kinematic variables were contact time (n ¼ 9), ankle dorsiflexion at initial contact (n ¼ 7), stride frequency (n ¼ 7), and foot/ground angle at initial contact (n ¼ 6). The equipment most used to measure was the motion capture system with optoelectronic cameras (n ¼ 10), most set at a capture frequency of 200 Hz. Of these, the most employed system was the Vicon NexusV R system (n ¼ 6).
The most frequently measured kinetic variables were vertical loading rate (n ¼ 7) and transient peak (n ¼ 5) through force platforms synchronised with the motion capture system (n ¼ 9); of these, two studies used treadmills instrumented with force platforms. Only one study performed calculations that allowed estimating leg stiffness and vertical stiffness values without employing a specific measurement instrument (Horvais & Samozino, 2013).

Footwear characteristics
Only three studies reported that the experimental shoes were equal in all their characteristics, except HTD (Mo et al., 2020;Zhang et al., , 2022. Subsequently, we found three studies that modified HTD in conjunction with heel height. Of these, two studies also reported slight differences in weight (Chambon et al., 2015;Malisoux et al., 2017), unlike a third study that did not provide this information (Horvais & Samozino, 2013).
The HTD of the shoes analysed varied between À8 mm and 16 mm. The most frequent values were 0 (n ¼ 8), 4 (n ¼ 5), 10 (n ¼ 5), 12 (n ¼ 4), and 8 mm (n ¼ 4). One study considered the biomechanical effects that negative HTD produced . In three studies, the barefoot running condition was also used (Chambon et al., 2015;Moody et al., 2018;Richert et al., 2019), which was used as Records excluded because they do not report results based on heel-to-toe drop (n = 5) Excluded because it analyzes the effects of insoles inserted inside the footwear (n = 1)  a control in one of the studies (Moody et al., 2018). In four studies, the authors specified the brand and model of footwear used. In one of them, four different brands and models of footwear were used (Moody et al., 2018), while the three studies used shoes of the same brand but of different models (Gij on-Noguer on et al., 2019; Horvais & Samozino, 2013;Richert et al., 2019). Eight studies specified forefoot, and heel thickness, five specified shoe weight, and four studies reported midsole material hardness (Table 4).

Biomechanical outcomes
The results of the studies included in this review are summarised in Table 5. There, a comparison is made of the measurements achieved between the different types of HTD, which were grouped into five categories: negative HTD; HTD 0 mm; HTD 4 to 7 mm; HTD 8 to 10 mm, and HTD 12 to 16 mm. This way, the effect of the different HTD on the studied variables can be appreciated. To simplify the presentation of the results, this section only mentions the changes in which HTD showed significant effects on the biomechanical variable. When more than two types of HTD were compared and/or another factor was present, attention was paid to the result of the post hoc analyses. It is worth mentioning that the table does not include the results reported by the study of Horvais and Samozino (2013) because their results are expressed in terms of correlation, unlike the other studies. Therefore, their results are mentioned in the text.

Spatiotemporal variables
The spatiotemporal variables were most frequently reported in the included studies. However, neither contact time, flight time, nor stride length showed significant differences between the different types of HTD. One of the five studies that measured stride length reported significative differences, finding greater stride length in HTD 8 than HTD 12 mm (Mo et al., 2020). In addition, Besson et al. (2019) studied the duration of the braking and push-off phases, reporting that subjects running with HTD 0 mm spent less time in the braking phase and more time in the push-off phase than HTD 6 and HTD 10 mm.
The ankle angle at initial contact was the only kinematic joint variable that showed differences between HTD. Only two studies evidenced changes in this variable, where the foot is supported with more plantarflexion when running barefoot than when using shoes and when comparing HTD 0 with HTD 6 and HTD 10 (Besson et al., 2019). When Table 2. Quality assessment of the studies (Downs & Black, 1998).    subjects ran with HTD 6.5, the ankle was more plantarflexion than HTD 16 too (Fu et al., 2022). Most studies did not report differences between the HTD (Chambon et al., 2015;Malisoux et al., 2017;Mo et al., 2020;Moody et al., 2018;Richert et al., 2019;.
In the frontal plane, HTD À8 produced more ankle movement amplitude than HTD 8 . Neither of the other ankle variables was modified for HTD.
In another joint, knee angle at initial contact was considered in five studies, but none of them reported differences between different types of HTD. Knee ROM during the stance phase showed a lesser value in HTD À8 than HTD 8 in the frontal and sagittal plane . Neither other knee variables showed differences between the different HTDs, except knee peak flexion angle, where HTD 0 showed a lower angle than HTD 5, HTD 10, and HTD 15 . However, this variable did not show differences in another study (Moody et al., 2018).
No variables related to the hip joint showed differences between the different types of HTD.
Two studies included the metatarsophalangeal joint (MTPJ) in the measurements. Range of motion (ROM) in the sagittal plane showed a lower value in HTD À8 than in HTD 8 , and a lower value in HTD 16 than in HTD 6.5 (Fu et al., 2022). Yu et al. also studied the ROM in the frontal plane, and they did not find differences between HTD types.
The study of Horvais & Samozino (Horvais & Samozino, 2013) expressed their results in terms of correlations, finding a positive correlation between HTD and foot strike angle. Thus, as the HTD value decreased, the foot strike angle decreased while the subjects ran to 3.9 m/s (r ¼ 0.62; p ¼ 0.013), and 4.7 m/s (r ¼ 0.66; p ¼ 0.008). The study also reports that HTD was positively correlated with contact time (r ¼ 0.85; p < 0.001 to 3.9 m/s, and r ¼ 0.83; p < 0.001 to 4.7 m/s) and duty factor, corresponding to contact time expressed relative to stride time, (r ¼ 0.87; p < 0.001 to 3.9 m/s, and r ¼ 0.87; p < 0.001 to 4.7 m/s), and negatively correlated with flight time (r ¼ À0.88; p < 0.001 to 3.9 m/s, and r ¼ 0.83; p < 0.001 to 4.7 m/s).

Kinetic variables
The vertical loading rate expressed in terms relative to the body weight of the subjects was the most studied variable (Table 4), with seven studies. Most studies did not show differences between different HTDs. However, in the studies that found significant differences, the vertical loading rate tended to be higher when HTD was lower. There, HTD À8 showed a higher value than HTD 8 . The same in HTD 0 with HTD 6 (Besson et al., 2019), HTD 0 with HTD 10 (Besson et al., 2019;, HTD 0 with HTD 15 , HTD 4 with HTD 8 (Richert et al., 2019), HTD 4 with HTD 12 (Richert et al., 2019), and HTD 6.5 with HTD 16 (Fu et al., 2022).
Four studies considered joint moments as interest variables. The peak knee moment was significantly smaller in HTD 12 (3.8 ± 0.5 Nm/kg), and HTD 8 (3.7 ± 0.4 Nm/kg) than in HTD 4 (3.6 ± 0.5 Nm/kg) (Richert et al., 2019). However, neither Fu et al. (2022) nor  found differences between the HTD used in the experimental design. No significative differences were found in the ankle or hip moments.
Leg stiffness was included in one study. In there, it correlated positively and statistically significantly with HTD (r ¼ 0.82, p < 0.001) when subjects ran at 3.9 m/s, and negatively and statistically significantly when they ran at 4.7 m/s (r ¼ À0.52, p < 0.043) (Horvais & Samozino, 2013). Zhang et al. (2022) included variables related to the patellofemoral joint (PFJ) stress. They found lower values in Peak PFJ stress, knee extension moment at the time of peak PFJ stress, PFJ force at the time of peak PFJ stress, and quadriceps muscle force in HTD 0 than in HTD 10 and HTD 15.

Discussion
The aim of this review was to identify and systematise the specific effects caused by different Heel-to-Toe Drop values of running shoes on biomechanical variables in endurance runners.
Knee ROM Horizontal Plane during stance phase: ROM in HTD À8 decreased by 0.9 compared with HTD 8 (p ¼ 0.018) .
Peak knee extension moment: No significative differences .
Ankle peak eversion angle: No significative differences .
Knee peak abduction angle: No significative differences .
Peak knee abduction moment: No significative differences .
Peak ankle dorsiflexion moment: No significative differences .
Peak ankle eversion moment: No significative differences .
Knee extension moment at the time of peak PFJ stress: No significant differences (Zhang et al., 2022). PFJ force at the time of peak PFJ stress: No significant differences (Zhang et al., 2022). Quadriceps muscle force: No significant differences (Zhang et al., 2022).
Ankle peak eversion angle: No significative differences .
Knee peak abduction angle: No significative differences .
Peak knee flexion moment: No significative differences .
Peak ankle dorsiflexion moment: No significative differences .
Quadriceps muscle force: HTD showed lower value (67.6 ± 18.8) than HTD 10 (79.8 ± 15.2) (p < 0.017) (Zhang et al., 2022). (continued) possible to conclude that HTD modifications do not produce a modification in contact time, flight time, stride frequency, and stride length. The data from the studies are virtually unanimous. However, it should be mentioned that the main comparisons are made between HTD 0 and other HTDs. However, comparisons between HTD of higher value always tend to show no effect. The inclusion of footwear with negative HTD in the studies was interesting, but the only study with these characteristics did not analysed spatiotemporal variables. To resume the effects that different HTDs produce over kinematic variables, Table 7 is exposed. Apparently, there is evidence that supports the fact that subjects modify the foot strike pattern towards forefoot strike only when they use HTD 0 in comparison with HTD 8-10. In this case, this hypothesis is supported by four studies (Besson et al., 2019;Chambon et al., 2015;Mo et al., 2020;, and one of them did not find differences (Malisoux et al., 2017), but that is the one that has an RCT design and consequently, the best score in Downs & Black quality assessment tool. This fact may partly counteract the evidence presented in favour of modifying the foot strike pattern.
A study used the foot strike angle as a method (Mo et al., 2020). In despite of the significative differences reported, this finding should not be considered because the values were not enough to change a rearfoot strike to a midfoot or forefoot strike since the mean values were in all cases > 8 (Altman & Davis, 2012). On the other hand, the strike index was modified by negative HTD from rearfoot strike to midfoot strike . This was the only study that included this aspect of the foot strike pattern. All other comparisons between HTDs showed no effect on the foot strike pattern.
According to the above, in general, we can state that there is not enough evidence to support the idea that the modification of the HTD contributes to modifying the plantar support pattern.
Considering the results of the reviewed studies, it seems that the foot strike pattern is an independent variable with respect to BFR or running shoes with different HTD measurements (Nunns et al., 2013), which is coherent with two studies that report the same effects caused by the use of minimalistic running shoes can be reached through precise instructions, downplaying the role of shoe characteristics or running barefoot (Barcellona et al., 2017;Dolenec et al., 2015). In the same way, there is a study that reports that the response to the use of different types of running practice (with or without running shoes) is individual to each subject (Tam et al., 2016). This allows us to question the role that the HTD variable has in the twenty-eight characteristics that the running shoes can present (Ramsey et al., 2019), which suggests that it is necessary to relativise the importance of HTD 0 over HTD > 4 mm in kinematic, kinetic and spatiotemporal changes.
With respect to joint kinematics, all available evidence indicates that different HTD values modified neither ankle, knee, or hip kinematics. Again, the results of Yu et al. (2021) are interesting since using a negative HTD shoe showed a greater ankle range of movement in a frontal plane. This increases the instability of the joint. However, only one study has added this type of running shoe.
To resume the effects that different HTDs produce over kinetic variables, Table 8 is exposed. We could observe a greater amount of evidence that supports the point that the modification of HTD in running shoes does not modify the values of GRF. The vertical loading rate was the most studied variable. Most studies stated that this variable did not change when the subjects used different types of HTD.

HTD 8-10
The number of studies reporting (yes) and those reporting no effect (no) is indicated in parentheses.
design with poor to fair methodology quality. It is interesting that the only HTD range studied was between À8 and 16 mm, when currently, there are running shoes with over 20 mm, and even some prototypes with negative values of HTD (G amez-Pay a et al., 2018). In our opinion, the negative HTD could be an interesting variable to consider for subsequent research. In addition, it is necessary to study the relationship between the HTD values and different combinations of heel height and forefoot height, such as the study carried out by Richert et al. (2019).

Disclosure statement
No potential conflict of interest was reported by the author(s).