Lower limb biomechanics during running in individuals with achilles tendinopathy: a systematic review

Shannon E Munteanu; Christian J Barton

doi:10.1186/1757-1146-4-15

Review
Open Access
Published: 30 May 2011

Lower limb biomechanics during running in individuals with achilles tendinopathy: a systematic review

Shannon E Munteanu^1,2 &
Christian J Barton^1,3

Journal of Foot and Ankle Research volume 4, Article number: 15 (2011) Cite this article

28k Accesses
50 Citations
11 Altmetric
Metrics details

Abstract

Background

Abnormal lower limb biomechanics is speculated to be a risk factor for Achilles tendinopathy. This study systematically reviewed the existing literature to identify, critique and summarise lower limb biomechanical factors associated with Achilles tendinopathy.

Methods

We searched electronic bibliographic databases (Medline, EMBASE, Current contents, CINAHL and SPORTDiscus) in November 2010. All prospective cohort and case-control studies that evaluated biomechanical factors (temporospatial parameters, lower limb kinematics, dynamic plantar pressures, kinetics [ground reaction forces and joint moments] and muscle activity) associated with mid-portion Achilles tendinopathy were included. Quality of included studies was evaluated using the Quality Index. The magnitude of differences (effect sizes) between cases and controls was calculated using Cohen's d (with 95% CIs).

Results

Nine studies were identified; two were prospective and the remaining seven case-control study designs. The quality of 9 identified studies was varied, with Quality Index scores ranging from 4 to 15 out of 17. All studies analysed running biomechanics. Cases displayed increased eversion range of motion of the rearfoot (d = 0.92 and 0.67 in two studies), reduced maximum lower leg abduction (d = -1.16), reduced ankle joint dorsiflexion velocity (d = -0.62) and reduced knee flexion during gait (d = -0.90). Cases also demonstrated a number of differences in dynamic plantar pressures (primarily the distribution of the centre of force), ground reaction forces (large effects for timing variables) and also showed reduced peak tibial external rotation moment (d = -1.29). Cases also displayed differences in the timing and amplitude of a number of lower limb muscles but many differences were equivocal.

Conclusions

There are differences in lower limb biomechanics between those with and without Achilles tendinopathy that may have implications for the prevention and management of the condition. However, the findings need to be interpreted with caution due to the limited quality of a number of the included studies. Future well-designed prospective studies are required to confirm these findings.

Peer Review reports

Background

Achilles tendinopathy is a common musculoskeletal disorder that can impair physical function in daily living, occupation and sporting environments. The prevalence of Achilles tendinopathy has been reported to be greater in males [1]. The condition accounts for between 8 and 15% of all injuries in recreational runners [2–4] and has a cumulative lifetime incidence of approximately 24% in athletes [5]. Although Achilles tendinopathy is common in athletes, one-third of patients with chronic Achilles tendinopathy are not physically active [6]. In some settings, approximately 30% of patients who present with this condition undergo surgical treatment [6, 7].

Achilles tendinopathy is considered a multifactorial condition, with both extrinsic and intrinsic factors thought to contribute to its development [8–10]. Proposed extrinsic risk factors include altered weightbearing surfaces (excessively hard, slippery or uneven) [8, 10], inappropriate footwear [8, 10, 11], training errors [10], use of specific medications such as fluoroquinolones [12] and the type of exercise activity (e.g., sports involving the stretch-shorten cycle such as running or jumping) [5]. Proposed intrinsic risk factors include previous injury [8], increased age [13], presence of specific genetic variations such as polymorphisms occurring within the COL5A1 and tenascin-C genes [14], male gender [15], increased adiposity and/or metabolic disorders [16, 17], pre-existing tendon abnormalities [18], triceps surae inflexibility [10, 19], hormonal status [20–22] and abnormal lower limb biomechanics [8, 10, 15, 23].

Alterations in lower limb biomechanical characteristics including temporospatial parameters, lower limb kinematics, dynamic plantar pressures, kinetics (ground reaction forces and joint moments) and muscle activity are frequently associated with Achilles tendinopathy [8, 15, 23]. One biomechanical factor commonly considered to be associated with Achilles tendinopathy is the presence of excessive foot pronation [8]. Clement et al. [10] originally proposed that excessive pronation of the foot may lead to Achilles tendinopathy through two mechanisms. First, excessive pronation of the foot is speculated to create greater hindfoot eversion motion, resulting in excessive forces on the medial aspect of the tendon and subsequent microtears. Second, abnormal pronation of the foot is thought to lead to asynchronous movement between the foot and ankle during the stance phase of gait, resulting in a subsequent 'wringing' effect within the Achilles tendon. This 'wringing' effect is theorised to cause vascular impairment within the tendon and peritendon [10] and elevated tensile stress [24] leading to subsequent degenerative changes in the Achilles tendon. In addition to kinematic theories, altered lower limb muscle function (timing, amplitude or co-ordination of contractions of the triceps surae) [23–26] and altered lower limb kinetics [11, 24, 25, 27] have also been speculated to be risk factors for Achilles tendinopathy by increasing tendon loading.

Several studies have been performed to investigate the association between abnormal lower limb biomechanics and Achilles tendinopathy. Critiquing and summarising results from these studies is now required to assist in the development of; (i) preventative strategies, and; (ii) specific and effective management strategies for the condition. However, at present, the aetiology of Achilles tendinopathy is not clearly understood [8]. Therefore, the aim of the present study was to perform a systematic review of the existing literature (prospective cohort and retrospective case-control studies) to identify, critique and summarise lower limb biomechanical factors associated with Achilles tendinopathy.

Methods

Inclusion and exclusion criteria

Prospective cohort and case-control studies evaluating biomechanical factors associated with mid-portion Achilles tendinopathy (i.e., 2-6 cm proximal to its insertion) were considered for inclusion. The inclusion criteria required participants to be described as having: midsubstance tendinopathy of the Achilles, Achilles tendinitis, tenosynovitis or tendinosis [28]. Additional terms such as Achilles tendinopathy, tenopathy, tendinosis, partial rupture, paratenonitis, tendovaginitis, peritendinitis and achillodynia have also been used to describe the problems of non-insertional pain associated with the Achilles tendon so were also used [29]. Measures of interest were gait characteristics including temporospatial parameters, lower limb kinematics, dynamic plantar pressures, kinetics (ground reaction forces and joint moments) and muscle activity.

Unpublished studies, case-series studies, non-peer-reviewed publications, intervention studies, studies not involving humans, reviews, letters, opinion articles, non-English articles and abstracts were excluded. Studies which included participants with concomitant injury or pain from structures other than the mid-portion of the Achilles tendon (e.g., insertional Achilles tendon pathology) or that failed to localise the pathology in the tendon were excluded.

Search Strategy

MEDLINE (OVID) (1950-), EMBASE (1988-), CINAHL (1981-), SPORTDiscus and Current Contents (1993 week 27-) electronic databases were searched in November 2010 (week 3). A generic search strategy was formulated [28, 30] and the results are reported in Additional Data File 1.

Review process

All titles and abstracts found were downloaded into Endnote version XI (Thomson Reuters, Philadelphia, PA) giving a set of 2701 citations. The set was cross-referenced and any duplicates were deleted, leaving a total of 1575 citations. Each title and abstract was evaluated for potential inclusion by two independent reviewers (SEM and CJB) using a checklist developed from the inclusion/exclusion criteria outlined above (see Additional File 2). If insufficient information was contained in the title and abstract to make a decision on a study, it was retained until the full text could be obtained for evaluation. Any disagreements regarding studies were resolved by a consensus meeting between the two reviewers.

Methodological quality assessment

The methodological quality of each included study was assessed using 16 items (maximum score of 17) of the 'Quality Index' considered relevant for assessing prospective cohort and case-control study designs (Table 1) [31]. The original Quality Index scale consisting of 26 items was shown to have high internal consistency (KR-20 = 0.89), test-retest (r = 0.88) and inter-rater (r = 0.75) reliability and high criterion validity (r ≥ 0.85) [31]. Two reviewers (SEM and CJB) applied the quality index to each included study independently, and any scoring discrepancies were resolved through a consensus meeting.

Table 1 Modified Downs and Black Quality Index results, and inter-rater reliability for each item and total score

Full size table

Statistical analysis

Inter-rater reliability of each item of the Quality Index was evaluated using unweighted kappa and percentage agreement statistics, and the overall score was evaluated using the intra-class correlation coefficient (ICC_3,1) with corresponding 95% confidence intervals (CIs).

Means and standard deviations for all continuous data were extracted and effect sizes (Cohen's d) (with 95% CIs) calculated to allow comparison between each study's results. To allow visual comparison, effect sizes were entered into forest plots. Categorical data (e.g. frequency of foot type) was compared between groups using odds ratios (with 95% CIs) transformed to effect sizes (with 95% CIs) as described by Chinn et al. [32] Calculated effect sizes were considered statistically significant if their 95% CI did not cross zero. If inadequate data were available from original studies to complete effect size calculations, attempts were made via email to contact the study's corresponding author for additional data.

Sample sizes (limbs analysed), the presence or absence of symptoms, participant demographics (gender, age, BMI, mass, height, duration of symptoms and sporting experience) and biomechanical analysis details were also extracted to assist in interpretation of findings.

Results

Following the search, nine studies were deemed appropriate for inclusion [2, 11, 19, 24, 25, 27, 33–35]. This included two prospective cohort [2, 19] and seven case-control study designs [11, 24, 25, 27, 33–35]. There were no disagreements amongst reviewers. One study [33] did not contain appropriate data to complete effect size calculations, meaning data extraction (effect size calculations) was performed on a total of eight studies [2, 11, 19, 24, 25, 27, 34, 35].

Quality assessment of included studies

All individual items from the Quality Index scale demonstrated high inter-rater reliability (kappas ≥ 0.57) with percentage agreement ≥ 77.8% (Table 1). The total score obtained from the Quality Index scale demonstrated high inter-rater reliability (ICC_3,1 = 0.98).

Additional data

Additional data required to complete effect size calculations was provided by Baur et al. [11]. Additionally, Van Ginckel et al. [2] provided revised data for some reported variables which were reported erroneously in their manuscript.

Methodological data to assist interpretation of results

Table 2 shows the samples sizes and population characteristics. Table 3 shows the biomechanical analysis details of each of the included studies.

Table 2 Sample sizes and population characteristics from each included study

Full size table

Table 3 Lower limb biomechanical analyses, gait characteristics and footwear conditions of included studies

Full size table

Differences in lower limb biomechanics between those with and without Achilles tendinopathy

Temporospatial gait characteristics

Four [11, 24, 33, 34] studies controlled gait velocity. Of the remaining five studies [2, 19, 25, 27, 35], only one [27] reported temporospatial data, with effect size calculations indicating no differences in velocity, stride length, stride time or stride frequency between cases and controls. Additionally, another study [35] reported that no significant differences in gait velocity were evident between groups but did not present supporting data.

Lower limb kinematics

Three studies investigated frontal plane rearfoot kinematics (Figure 1) [25, 34, 35]. Those with Achilles tendinopathy displayed greater rearfoot eversion range of motion when shod (d = 0.92) but not unshod [34] and greater eversion range of motion of the ankle/rearfoot (d = 0.67) [35]. Effect size calculations for all other frontal plane rearfoot kinematics comparisons were not statistically significant.

Four studies investigated tibial segment and ankle joint kinematics (Figure 2) [24, 27, 34, 35]. Donoghue et al. [34] showed reduced maximum lower leg abduction (barefoot) in cases (d = -1.16). Ryan et al. [35] showed reduced maximum ankle dorsiflexion velocity in cases (d = -0.62). All other tibial segment and ankle kinematic comparisons were not significantly different between groups [24, 27, 34, 35].

Three studies performed analyses for knee and hip kinematics (Figure 3) [24, 27, 34]. Azevedo et al. [27] reported that the magnitude of knee flexion between heel strike and midstance was significantly reduced in cases (d = -0.90). Effect size calculations for all other knee joint kinematics comparisons were not significantly different between groups [24, 27, 34]. There were no statistically significant effects for comparisons in sagittal plane hip kinematics [27].

Plantar pressure parameters

A large number of plantar pressure parameters were analysed across three studies [2, 11, 19] (Figures 4A-D and 5). A prospective study by Van Ginckel et al. [2] showed that those who developed Achilles tendinopathy demonstrated significantly reduced displacement of the posterior-anterior component of the centre of force at last foot contact (d = -0.95), posterior-anterior displacement of the centre of force during forefoot push-off phase (d = -0.75), total posterior-anterior displacement of the centre of force (d = -0.95) and medio-lateral force distribution under the metatarsal heads at forefoot flat (d = -0.93) (Figure 4A). Further those who developed Achilles tendinopathy displayed reduced timing of initial contact at the second metatarsal head region (d = -1.00) (Figure 4B), relative peak force at the medial heel (d = -0.73), time to peak force at the lateral heel (d = -1.08) and at the medial heel (d = -0.72) regions (Figure 4C). Additionally, increases were found for peak force at the fifth metatarsal head region (d = 0.84) (Figure 4C) and force-time integral at the fifth metatarsal head region (d = 0.81) (Figure 4D) in those who developed Achilles tendinopathy [2].

Figure 5 shows that lateral deviation of the centre of pressure in the rear-and mid-foot (Alat [barefoot]) was significantly reduced in cases (d = -0.98) [11]. The frequency of dynamic pes planus or pes cavus (assessed using dynamic arch index in both barefoot and shod conditions) was not significantly different between those who did and did not develop Achilles tendinopathy [19].

Lower limb external kinetics

One study analysed lower limb joint moments (Figure 6). Peak tibial external rotation moment was significantly reduced in cases (d = -1.29) [24].

Three studies analysed ground reaction forces [11, 25, 27] (Figure 7A-C). The normalised time to first vertical peak (d = 19.54) [25] and normalised time to minimum vertical peak (d = 22.69) [25] were significantly increased (delayed) in cases (Figure 7A). The normalised time to second vertical force (d = -19.50) [25] was significantly reduced (earlier) in cases (Figure 7A). The second normalised vertical peak force (d = 0.52) [25] and the vertical impulse (barefoot) were significantly increased in cases (d = 0.70) (Figure 7A) [11].

The normalised time to maximum braking force (d = -56.1) [25] and normalised time (% stance) to maximum propulsive force (d = -26.5) [25] were significantly reduced (earlier) in cases (Figure 7B). The normalised maximum braking force (d = 0.46) [25], normalised average braking force (d = 0.52) [25] and pushing impulse (shod) (d = 0.74) [11] were significantly increased in cases (Figure 7B).

The normalised time to maximum lateral force was significantly reduced (earlier) (d = -12.05) [25] and normalised time to maximum medial force was significantly increased (delayed) (d = 13.25) [25] in cases (Figure 7C). The normalised maximum lateral force was significantly increased (d = 0.57) [25] in cases (Figure 7C).

Lower limb muscle function

Two studies performed comparisons of lower limb muscle function (amplitude and/or timing) [11, 27] (Figures 8 and 9A-D). Azevedo et al. [27] reported no significant effects for the amplitude of lateral gastrocnemius at pre-and post-heel strike between cases and controls. Baur et al. [11] showed that the amplitude of lateral gastrocnemius to be significantly reduced during weight acceptance (shod and barefoot) (d = -1.50 and-2.46 respectively) but significantly increased during push-off (shod and barefoot) (d = 0.69 and 1.26 respectively) in cases. Further, the total time of activation of lateral gastrocnemius (shod and barefoot) (d = 0.80 and 1.21 respectively) [11] was significantly increased in cases. Baur et al. [11] investigated medial gastrocnemius function and showed that cases displayed significantly increased amplitude during push-off (shod) (d = 0.86). There were no other significant effects for the amplitude or timing of onset of this muscle (Figure 8).

Azevedo et al. [27] showed that the amplitude of tibialis anterior was significantly reduced at pre-heel strike (100 ms before heel strike) in cases (d = -1.00). Baur et al. [11] showed the amplitude of tibialis anterior during weight acceptance (shod) (d = 1.06) and push-off (barefoot) (d = 1.93) to be significantly increased in cases. Further, the onset of activation of tibialis anterior (shod and barefoot) (d = 0.65 and 0.67 respectively) [11] was significantly increased (delayed) in cases (Figure 9A).

Baur et al. [11] showed the amplitude of peroneus longus during pre-activation (shod) (d = 0.76) and during push-off (barefoot) (d = 0.83) to be significantly increased in cases. Azevedo et al. [27] reported the amplitude of peroneus longus at post-heel strike (100 ms post-heel strike) to be significantly reduced (d = -0.67) in cases (Figure 9B).

Baur et al. [11] investigated soleus muscle function and showed that those with Achilles tendinopathy displayed significantly reduced amplitude during pre-activation (shod) (d = -1.49) and weight acceptance (barefoot) (d = -1.48) but increased during push-off (shod and barefoot) (d = 0.72 and 1.95 respectively). Further, the total time of activation (shod and barefoot) was significantly increased (d = 0.96 and 0.68 respectively) in cases [11] (Figure 9C).

At the hip and knee joints, the amplitude of rectus femoris and gluteus medius post-heel strike (100 ms post-heel strike) were significantly reduced (d = -1.4 and-1.1 respectively) in cases [27] (Figure 9D).

Discussion

The aim of the present systematic review was to identify, critique and summarise lower limb biomechanical factors associated with Achilles tendinopathy. This review is timely to enhance the development of effective intervention and prevention strategies for the condition. Nine studies [2, 11, 19, 24, 25, 27, 33–35] evaluating lower limb biomechanics in those with Achilles tendinopathy were identified, with eight [2, 11, 19, 24, 25, 27, 34, 35] containing sufficient data to complete effect size calculations.

Quality

In agreement with other studies [30, 36, 37] that have used Quality Index [31], high inter-rater reliability for the selected items used in this study was found. Methodological quality was varied, with scores ranging between 4 and 15 out of 17. Several studies did not clearly describe participant characteristics (Item 3) [11, 25, 33, 34] or discuss whether participants invited (Item 11) [11, 24, 25, 27, 33–35] or recruited were representative of entire population (Item 12) [11, 27, 33–35]. This limits the ability of any findings to be applied to a broader population. None of the case-control studies [11, 24, 25, 27, 33–35] blinded their outcome assessors (Item 15) making it possible that some of the associated results may have been biased. Several included studies did not clearly describe confounding variables (Item 5) [11, 19, 25, 33–35] or adjust for these in their analyses (Item 25) [11, 19, 33, 34]. Additionally, the validity and reliability of outcome measurements used was not reported by any of the studies (Item 20) [2, 11, 19, 24, 25, 27, 33–35]. One study [11] analysed both limbs of each participant, and pooled data for both limbs within the case group, despite participants in the case group having unilateral symptoms. Two case-control studies [33, 34] excluded participants that displayed a rigid foot type in the Achilles tendinopathy but not in the control group. This introduces significant recruitment bias into their studies.