A three-year prospective comparative gait study between patients with ankle arthrodesis and arthroplasty

Clinical Biomechanics 54 (2018) 42–53

Contents lists available at ScienceDirect

Clinical Biomechanics journal homepage: www.elsevier.com/locate/clinbiomech

A three-year prospective comparative gait study between patients with ankle arthrodesis and arthroplasty

T

Ava D. Segala, Krista M. Cyra, Christina J. Stendera, Eric C. Whittakera, Michael E. Hahna,b,1, ⁎ Michael S. Orendurffa,2, William R. Ledouxa,b,c, , Bruce J. Sangeorzana,c a

Center for Limb Loss and MoBility, Rehabilitation Research and Development, Department of Veterans Affairs, Seattle, WA, USA Department of Mechanical Engineering, University of Washington, Seattle, WA, USA c Department of Orthopaedics & Sports Medicine, University of Washington, Seattle, WA, USA b

A R T I C LE I N FO

A B S T R A C T

Keywords: End-stage ankle arthritis Kinematics Kinetics Biomechanics Push-off power Heel strike transient

Background: End-stage ankle arthritis is a debilitating condition that often requires surgical intervention after failed conservative treatments. Ankle arthrodesis is a common surgical option, especially for younger and highly active patients; however, ankle arthroplasty has become increasingly popular as advancements in implant design improve device longevity. The longitudinal differences in biomechanical outcomes between these surgical treatments remain indistinct, likely due to the challenges associated with objective study of a heterogeneous population. Methods: Patients scheduled for arthroplasty (n = 27) and arthrodesis (n = 20) were recruited to participate in this three-year prospective study. Postoperative functional outcomes were compared at distinct annual time increments using measures of gait analysis, average daily step count and survey score. Findings: Both surgical groups presented reduced pain, improved survey scores, and increased walking speed at the first-year postoperative session, which were generally consistent across the three-year follow-up. Arthrodesis patients walked with decreased sagittal ankle RoM, increased sagittal hip RoM, increased step length, and increased transient force at heel strike, postoperatively. Arthroplasty patients increased ankle RoM and cadence, with no changes in hip RoM, step length or heel strike transient force. Interpretation: Most postoperative changes were detected at the first-year follow-up session and maintained across the three-year time period. Despite generally favorable outcomes associated with both surgeries, several underlying postoperative biomechanical differences were detected, which may have long-term functional consequences. Furthermore, neither technique was able to completely restore gait biomechanics to the levels of the contralateral unaffected limb, leaving potential for the development of improved surgical and rehabilitative treatments.

1. Introduction Patients who suffer from ankle arthritis often experience severe and debilitating pain, leading to impaired function and reduced quality of life. Functional deficits typically manifest as shorter stride length,

slower walking speed, decreased ankle motion, altered joint loading, and self-reported reduced function (Agel et al., 2005; Brodsky et al., 2011; Glazebrook et al., 2008; Khazzam et al., 2006; Segal et al., 2012). The primary cause of end-stage ankle arthritis is previous trauma (Agel et al., 2005; Brockett and Chapman, 2016; Saltzman et al., 2005;

Abbreviations: A1m, peak ankle plantar flexion moment (terminal stance); A1a, peak ankle power absorption; A2a, peak ankle power generation; Abd, abduction; Add, adduction; Abs, absorption; deg, Degrees; Dorsi, dorsiflexion; Ext, extension; Flex, flexion; Gen, generation; GRFZ, vertical ground reaction force; HST, heel strike transient; K1a, peak knee power absorption (early stance); K2a, peak knee power generation (mid-stance); K3a, peak knee power absorption (early swing); K4a, peak knee power absorption (terminal swing); H1a, peak hip power generation (early stance); H2a, peak hip power absorption (terminal stance); H3a, peak hip power generation (early swing); K1m, peak knee extension moment (early stance); K2m, peak knee flexion moment (mid-stance); H1m, peak hip extension moment (early stance); H2m, peak hip flexion moment (mid-stance); H3m, peak hip extension moment (terminal swing); P0, baseline (preoperative) session; P1, first year follow-up (postoperative) session; P2, second year follow-up (postoperative) session; P3, third year follow-up (postoperative) session; Plantar, plantar flexion; m/s, meters/second; min, minute; MFA, musculoskeletal function assessment survey; RoM, range of motion; s, seconds; SE, standard error; SF-36, Short Form (36 questions) survey ⁎ Corresponding author at: VA Puget Sound, MS 151, 1660 S. Columbian Way, Seattle, WA 98108, USA. E-mail address: [email protected] (W.R. Ledoux). 1 Director, Bowerman Sports Science Clinic, Department of Human Physiology, University of Oregon, Eugene, OR. 2 Current Affiliation: Director, Motion & Sports Performance Laboratory, Stanford Children's Health, Palo Alto, CA, USA. https://doi.org/10.1016/j.clinbiomech.2018.02.018 Received 18 August 2017; Accepted 26 February 2018 0268-0033/ Published by Elsevier Ltd.

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

2. Methods

Valderrabano et al., 2004), which lends itself to a younger and more diverse population compared to degenerative hip and knee arthritis (Brown et al., 2006; Michael et al., 2008; Saltzman et al., 2005). This further complicates diagnosis, analysis, and generalized treatment for ankle arthritis. Understanding the changes in biomechanics associated with ankle surgery can help improve treatment and quality of life for these patients. The standard of care for patients with end-stage ankle arthritis has historically been a fusion of the tibiotalar ankle joint, also referred to as ankle arthrodesis (Espinosa and Klammer, 2010; Pedowitz et al., 2016; Takakura et al., 1999). However, total ankle replacements (arthroplasty) have become more accepted and practiced due to substantial improvements in implant design and durability. In a recent retrospective study (n = 78), arthroplasty implant survival rate was reported as 97.5% at a mean follow-up time of 5.2 years (Hofmann et al., 2016). However, other studies reported a higher revision rate still exists for arthroplasty compared to arthrodesis (Daniels et al., 2014; SooHoo et al., 2007). Daniels et al. reported the revision rate at a mean follow-up of 5.5 years was 17% for arthroplasty (n = 232) compared to 7% for arthrodesis (n = 89) (Daniels et al., 2014). Prospective gait studies of patients who have undergone either arthrodesis or arthroplasty reported improved spatiotemporal and joint biomechanical parameters compared to preoperative measures for both surgery types (Brodsky et al., 2011; Brodsky et al., 2016; Choi et al., 2013; Queen et al., 2012; Valderrabano et al., 2007). Brodsky et al. reported increased postoperative (mean 15 months) step length, walking speed, hip joint range of motion (RoM), and ankle moment for the affected limb of arthrodesis patients compared to preoperative values (Brodsky et al., 2016). Choi et al. showed similar results to Brodsky et al. for arthroplasty patients, including increased sagittal plane ankle RoM at a mean follow-up time of 37.2 months (Choi et al., 2013). However, neither surgery has been shown to completely restore gait to a functional level equivalent to control populations, and the altered ankle biomechanics of these patients can potentially lead to compensatory strategies in both the ipsilateral and contralateral lower limb joints with an increased risk of comorbidity. Few prospective gait studies have directly compared the pre- to postoperative changes in lower limb biomechanics between both surgical procedures. One study by Flavin et al. examined three-dimensional biomechanics using digital-motion capture of arthrodesis, arthroplasty, and control volunteers (n = 14 per group) preoperatively and at one-year postoperatively (Flavin et al., 2013). Their results showed that arthroplasty patients increased stride length, cadence, and dorsiflexion postoperatively, while arthrodesis subjects increased plantar flexion despite similar total ankle RoM postoperatively. These results provide an initial comparison of the postoperative function between these surgical procedures; however, their study lacks consistent multi-year follow-ups to confirm the effect of surgery over time, and does not report on the biomechanics of the lower limb joints proximal to the affected ankle. Our prospective study directly compares the lower limb (ankle, knee and hip) biomechanics between arthrodesis and arthroplasty patients at multiple, consistent time periods. The goal of this study is to compare the changes in lower limb biomechanics between arthrodesis and arthroplasty patients through spatiotemporal, kinematic, and kinetic measures collected preoperatively and at one-year postoperative intervals for three consecutive years. Based on our initial findings of the functional limitations of end-stage ankle arthritis (Segal et al., 2012) and from a smaller subset of patients (n = 9 each) after one-year follow-up (Hahn et al., 2012), we hypothesized that both surgical groups would maintain a reduction in pain and improved walking speed after three-years follow-up; however, we also predicted that divergence in ankle RoM between arthroplasty and arthrodesis patients would lead to additional kinematic and kinetic compensations compared to baseline.

2.1. Recruitment Qualifying patients scheduled for either ankle arthrodesis or arthroplasty to treat end-stage ankle arthritis provided informed consent to participate in this Institutional Review Board-approved study. Inclusion criteria were the diagnosis of end-stage ankle arthritis as defined by the presence of pain and failed conservative care (i.e., bracing, life-style modifications, physical therapy), age of 18 years or older, and ambulatory without an assistive device with the primary impediment of ankle arthritis. Patients were excluded if they had received any recent (< 1 year) surgical lower-extremity interventions, presented with neurological, metabolic or orthopedic impairment that might affect walking ability or the presence of rheumatoid arthritis. 2.2. Protocol This prospective, non-randomized study involved an initial evaluation at baseline prior to ankle surgery (P0), followed by three annual post-surgical follow-up sessions (P1, P2, P3). Prior to the baseline session, two expert orthopedic surgeons explained the risks and benefits of each procedure and then allowed patients to choose their surgical preference. Two tibiotalar arthroplasty devices each with 2-component, fixed bearing designs were used for all ankle replacements (Salto Talaris® Ankle, Integra LifeSciences, Plainsboro, NJ; Agility™ Ankle System, DePuy Synthes, Johnson and Johnson, Warsaw, IN, USA). Ankle arthrodesis was a fusion of the tibiotalar joint using internal screw fixation. All surgeries involved standard open techniques without arthroscopy. Pre- and postoperative example radiographs are presented in Hahn et al. (Hahn et al., 2012). All patients followed a standard postsurgical rehabilitation protocol consisting of an initial six-week nonweightbearing period, followed by a progressive weight-bearing period and gradual return to normal daily activities. 2.2.1. Gait analysis During each laboratory visit, patient height, weight, and standard anthropometric measurements were taken according to Vicon's requirements for static and dynamic modeling (Vicon, Centennial, CO, USA). Thirty-five, 14 mm reflective markers were placed on each patient's upper and lower limbs, torso, pelvis, and head at locations consistent with Vicon's Plug-In Gait full body marker set. Patients were then asked to walk barefoot at their self-selected walking speed across a 10-m walkway with four embedded force platforms (2 AMTI BP400600, Watertown, MA, USA; 2 Bertec FP4060-NC, Columbus, OH, USA). Several practice walking trials were completed to identify a starting position that allowed for patients to naturally strike each force platform with a single limb. Patients completed five repeated trials while marker trajectories were collected with a 12-camera Vicon MX system at 120 Hz and later filtered with the Woltring quantic spline algorithm (Vicon) with a mean-square-error value of 20. Ground reaction force (GRF) data were simultaneously collected at 1200 Hz with the force platforms. Spatiotemporal variables included walking speed, step length, step width, stance duration, step duration, and cadence. Step length was measured as the distance along the direction of progression between the heel marker positions from heel strike to opposite limb heel strike, where affected or unaffected step length was defined by the first heel strike. Step width was defined in a similar manner, except was measured as the distance in the mediolateral direction between heel marker positions. Stance duration was the time from heel strike to toe off for each limb. Step duration was the time from heel strike to opposite limb heel strike, where affected or unaffected limb was defined by the first heel strike. Cadence was defined as the average number of steps per minute of the affected and unaffected limbs. Lower extremity joint angles (kinematics) were calculated using the 43

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

separated by surgery type, as well as the total change between P0 and each follow-up session across surgery type for the affected limb. All pair-wise differences were adjusted for multiple comparisons within the model using simultaneous inference (Hothorn et al., 2008). Mean differences by follow-up session and surgery type for the affected and unaffected limbs at P0 are reported in Tables 2b–5 and Supplemental Tables S1–S3 (Joint kinetics (moments and powers), survey scores, and step count data demonstrated few differences between surgery types; therefore, the key results are summarized in the results text with the associated tables included as supplemental material for completeness). In the text, the overall mean affected limb changes in outcomes from P0 are summarized by averaging the changes across follow-up sessions (mean (SE), [95% Confidence Interval], P-value). Additionally, we tested the surgery type by session interaction to assess whether an overall difference in the pattern of change across all follow-up sessions by surgery type was present. A second set of regression models were also performed to adjust for speed by including speed as an independent covariate. For brevity, speed adjusted differences were only included when significance was detected because of the speed effect. Secondary analyses to assess limb asymmetry were tested by including both sets of limbs in the regression models with limb as an independent covariate. Finally, t-tests (chi-square test for difference in surgery type by sex) were used to test for differences in outcome between surgery types at baseline. Significance was set at P < 0.05. Analyses were performed using R 3.3.2 (Team, 2016) with the lme4 package for linear mixed effects regression and multcomp package (Hothorn et al., 2008) for multiple comparisons.

Plug-In Gait dynamic model (Vicon), which transforms the recorded marker trajectories according to the static pose to define relative joint motion between adjacent segments. Sagittal ankle, knee, and hip RoM were measured using discrete peaks defined in early to mid-stance (RoM1) and in mid-stance to terminal stance (ankle) or mid-stance to mid-swing (knee, hip) (RoM2). Lower extremity net internal joint moments and powers (kinetics) were calculated using standard inverse dynamic techniques (Winter, 1991) with Nexus software (Vicon) and normalized to patient body weight. The vertical ground reaction force (GRFZ) discrete peaks included: (1) heel strike transient (HST) peak, extracted when a negative slope was detected between heel strike and the first maximum before midstance, (2) HST prevalence, calculated as the total peak occurrence from the binary output of HST peak (1 = peak occurred, 0 = peak did not occur) divided by the total number of trials, (3) GRFZ1max, measured as the peak GRFZ from early to mid-stance, (4) GRFZ2max measured as the peak GRFZ from mid-stance to terminal stance and (5) GRFZmin, measured as the minimum GRFZ between GRFZ1max and GRFZ2max. Discrete peaks for the affected and unaffected limbs were extracted using MATLAB software (MathWorks, Inc., Natick, MA, USA). 2.2.2. Step count Each patient wore a StepWatch™ 3 Activity Monitor (Modus Health, Edmunds, WA, USA; 70 × 50 × 20 mm, 38 g; daily step count accuracy > 98% (Foster et al., 2005; Resnick et al., 2001; Shepherd et al., 1999)) on their ankle for 1–2 weeks within a month of each testing session (baseline: 11.4 (3.0) days (mean (SD)); follow-up sessions: 13.5 (1.2) days) to objectively measure daily activity. Fewer average days were collected prior to the baseline session (range: 5–15 days) because some patients were recruited at their preoperative appointment, which was < 2 weeks before surgery. Follow-up sessions (range: 10–15 days) were combined because the average number of days collected varied by less than half of a day. The analyzed step metrics were average number of steps taken per day (step total) and the average number of steps taken at low (< 15 strides/min), medium (15–40 strides/min), and high (> 40 strides/min) step intensity, similar to previous research (Brandes et al., 2008).

3. Results 3.1. Patient withdrawal and demographics Baseline sessions were collected for patients scheduled for either ankle arthrodesis (n = 20) or ankle arthroplasty (n = 27) between the years 2007 and 2011. Explanations for missed follow-up sessions were similar between surgical groups (Table 1); the most common being the participant was doing well and did not return for out-of-town follow-up exams (arthrodesis, n = 5; arthroplasty, n = 5). Few patients in each group withdrew due to continued pain after surgery (arthrodesis, n = 2; arthroplasty, n = 1). Patients with at least one follow-up session were included in the analysis (arthrodesis, n = 13; arthroplasty, n = 20). Three patients in each group only completed the first-year follow-up session. Also, three arthroplasty patients were having revisions from previous ankle replacements 5–9 years prior to baseline. While these patients may have worse outcomes, we considered them to be representative of this heterogeneous population and, therefore, chose to include them in the analysis. The measured monthly time increments for each follow-up session were the same between surgery types (P = 0.52) and therefore, combined across surgeries (P1 = 12.7 (0.3) months, [12.2, 13.2] (mean (SE), [95% Confidence Interval]); P2 = 24.6 (0.3) months, [24.0, 25.2]; P3 = 37.0 (0.3) months, [36.4, 37.6]). The patient group demographics at baseline were similar between surgery types, except the arthrodesis patients were slightly (+Δ6 cm) taller (Table 2a). The changes from baseline to three-year follow-up session between surgery types (Arthroplasty-Arthrodesis Difference) in body weight (+Δ5.5 kg) and BMI (+Δ2.0) were significantly different, with the changes primarily associated with decreased body weight (−Δ4.7 kg) and, thus, BMI (−Δ2.0) for arthrodesis patients (Table 2b).

2.2.3. Surveys Patients also completed two validated questionnaires at baseline and each follow-up session. The Musculoskeletal Function Assessment (MFA) (Zhang and Jordan, 2010) is a general arthritis functional measure consisting of 100 yes or no questions that reveals how a patient's musculoskeletal disorder affects function during activities of daily living (e.g., household chores, walking in the community). Higher scores indicate reduced function. The SF-36 (Medical Outcomes and Trust, Boston, MA, USA) is a short-form health survey consisting of 36 questions broken into eight categories (McHorney et al., 1994; Ware Jr. and Sherbourne, 1992). Our study only included the two most relevant categories: physical function (PF) and bodily pain (BP), with lower scores indicating reduced function or more pain, respectively. Patients were also asked to rate their pain on average over the past month from 0 (no pain) to 10 (pain as bad as it could be) at each visit. 2.3. Statistical methods Linear mixed effects regression was used to test for changes in outcomes from baseline (P0) across all follow-up sessions, and for differences in these changes by surgery type. Outcome metrics were the dependent variables, study session (modeled as a three-category variable with P0 as the reference group), surgery type, and session-by-surgery type interaction were the independent variables. Random effects were modeled to account for the variability in mean outcomes across study participant. For the biomechanical outcomes, an additional set of random effects were included to address repeated measures per session. To test our primary hypotheses, pair-wise differences were assessed for changes in outcome between P0 and each follow-up session

3.2. Spatiotemporal variables All patients significantly increased walking speed across sessions (+Δ0.14 m/s, [0.06, 0.22], P < 0.001), with similar improvements between surgery types (surgery by session interaction, P= 0.64). 44

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

Table 1 Explanations for missed follow-up sessions by surgery type for all participants who completed a baseline session.

Doing well and did not return for out-of-town follow-up visits Lack of interest Pain Comorbidity (total hip/knee replacement, talonavicular fusion, midfoot fusion, foot drop) Cancer diagnosis Deceased Insurance denial Unknown TOTAL

Arthrodesis

Arthroplasty

5 0 2 3 0 1 1 1 13

5 2 1 5 1 0 0 1 15

[6.7, 13.6], and for arthroplasty: −Δ6.4 deg. [3.7, 9.2], both P < 0.001). Affected limb sagittal RoM1 (early to mid-stance) remained consistent at the knee after either surgery (< Δ2 deg), but increased at the hip only for arthrodesis patients across sessions (RoM1: +Δ3.8 deg. [0.5, 6.2], P = 0.020). Affected limb sagittal hip RoM2 (mid-stance to mid-swing, +Δ3.9 deg. [0.4, 7.3] P = 0.025) also increased only for arthrodesis patients. At the knee, the change in RoM2 was similar in magnitude for arthrodesis patients (+Δ4.3 deg. [−0.2, 8.8], P = 0.13), but this difference did not reach statistical significance. The changes at the hip were associated with increased walking speed (i.e., the increases were non-significant after adjustment for speed) and when separated by session, were significant in the first- and second-year follow-up sessions (Table 4).

However, when separated by session, the change in speed reached statistical significance in only two of the six pairwise comparisons (Table 3). Arthrodesis patients had a significant increase in step length for both limbs across session (+Δ0.08 m [0.02, 0.13], P = 0.002). In contrast, arthroplasty patients only increased unaffected limb step length (+0.05 m, [0.01, 0.10], P = 0.009), but also increased cadence (+Δ7 steps/min [0, 13], P = 0.033) and decreased affected limb step duration across session (−Δ0.04 s [−0.07, 0.00], P = 0.021). These changes in step length, cadence and step duration were generally consistent for each pairwise follow-up comparison (Table 3). 3.3. Joint kinematics (Fig. 1) A significant increase in early to mid-stance affected limb sagittal ankle RoM1 across follow-up sessions (+Δ4.1 deg. [0.7, 7.5], P = 0.015) was found for arthroplasty compared to arthrodesis and was consistent across each pairwise follow-up comparison (ArthroplastyArthrodesis Difference, Table 4). Arthrodesis patients also presented decreased mid- to terminal stance affected limb sagittal ankle RoM2 across sessions (speed adjusted: −Δ3.9 deg. [0.3, 7.6], P = 0.029), which was significant in the first- and second-year follow-up sessions after adjusting for walking speed (Table 4); however, both surgery group's affected ankle RoM2 remained significantly reduced compared to their unaffected limb across sessions (affected-unaffected limb for arthrodesis: −Δ10.2 deg.

3.4. Joint kinetics (Figs. 2, 3) No differences in peak sagittal plane joint moments and powers were detected between surgery types. Additionally, all small postoperative increases in mid- to late stance ankle and hip moments (Table S1) and changes in knee and hip powers (Table S2) for both surgeries were associated with increased walking speed. 3.5. Vertical ground reaction force (Fig. 4) There was a borderline significant difference across sessions in HST

Table 2a Baseline session (P0) mean (standard error, SE) and range for patient demographics, survey scores, step count and walking speed separated by surgery type (arthrodesis, arthroplasty) for the participants included in the analysis, with associated t-test P-values. Bold values indicate a significant difference (P < 0.05). Mean (SE)

Demographics Weight (kg) Height (m) BMI Age (years) Sex: % females (No. of females/total) Survey scores MFAa SF-36 physical functionb SF-36 body painb Pain scorec Step count (steps/day) Total Low frequency Medium frequency High frequency Speed (m/s) a b c

P-value

Arthrodesis

Arthroplasty

(n = 13) 90.6 (5.0) 1.75 (0.08) 29.4 (1.2) 53.4 (9.8) 23% (3/13)

(n = 20) 86.7 (4.0) 1.69 (0.12) 30.9 (1.0) 59.9 (8.7) 60% (12/20)

39.1 (4.5) 46.2 (5.9) 34.4 (6.0) 5.8 (0.6) 7662 (1094) 2101 (264) 4153 (685) 1400 (409) 1.02 (0.06)

Range Arthrodesis

Arthoplasty

0.53 0.03 0.35 0.09 0.10

66.8–115.5 1.61–1.88 24.2–36.5 37–71

50.7–113.6 1.51–1.89 21.7–39.0 46–81

35.1 (3.6) 44.8 (5.0) 40.1 (5.1) 6.6 (0.5)

0.59 0.86 0.38 0.42

18–59 5–80 0–58 3–10

16–73 10–80 10–68 2–9

8505 (890) 2585 (214) 4635 (556) 1300 (337) 1.00 (0.05)

0.56 0.14 0.64 0.86 0.97

2628–20,254 856–3666 1316–13,692 79–6174 0.55–1.35

3960–15,392 796–4288 1641–10,620 184–2762 0.53–1.33

Musculoskeletal Function Assessment (MFA): lower score indicates improved function (range: 0–100). Short Form 36 (SF-36) physical function and body pain: higher score indicates improved function or reduced pain (range: 0–100). Pain score: lower score indicates less pain (range: 0–10).

45

Arthroplasty

Arthroplasty

−3.3 [−7.3, 0.7], 0.16 −0.1 [−3.2, 2.9], 1.0 3.2 [−1.8, 8.2], 0.46 −1.3 [−2.7, 0.2], 0.13 −0.2 [−1.3, 0.9], 1.0 1.1 [−0.8, 2.9], 0.56

Arthrodesis

P2–P0

Arthroplasty

−4.7 [−8.7, −0.8], 0.010 0.7 [−2.4, 3.9], 0.99 5.5 [0.4, 10.6], 0.027 −1.6 [−3.1, −0.2], 0.017 0.4 [−0.8, 1.5], 0.97 2.0 [0.1, 3.9], 0.029

Arthrodesis

P3–P0

46

b

a

P1-P0

−0.06 [−0.13, 0.01], 0.15 −0.05 [−0.10, 0.01], 0.13 −0.06 [−0.15, 0.02], 0.24 −0.06 [−0.13, 0.01], 0.091

−0.02 [−0.05, 0.01], 0.45 −0.04 [−0.06, −0.01], 0.003 −0.02 [−0.06, 0.02], 0.65 −0.04 [−0.07, 0.00], 0.030 −0.02 [−0.08, 0.03], 0.68 −0.04 [−0.08, 0.01], 0.15 −0.03 [−0.08, 0.02], 0.52 −0.04 [−0.08, 0.00], 0.043

0.72 (0.03) 0.71 (0.02) 0.76 (0.03) 0.75 (0.03)

0.56 (0.02) 0.56 (0.02) 0.58 (0.02) 0.58 (0.02)

Arthroplasty

0.09 [0.02, 0.15], 0.003 0.07 [0.01, 0.12], 0.013

0.04 [−0.01, 0.09], 0.23 0.05 [0.00, 0.10], 0.022

0.15 [−0.02, 0.33], 0.10 0.15 [0.01, 0.29], 0.036 3 [−7, 12], 0.93 7 [−1, 14], 0.10 −0.01 [−0.04, 0.02], 0.83 0.00 [−0.02, 0.02], 1.0

Arthrodesis

P3-P0

−0.02 [−0.07, 0.03], 0.81 −0.04 [−0.07, 0.00], 0.087 −0.03 [−0.08, 0.03], 0.72 −0.04 [−0.09, 0.00], 0.077

−0.06 [−0.13, 0.01], 0.16 −0.05 [−0.11, 0.00], 0.089 −0.06 [−0.13, 0.02], 0.22 −0.06 [−0.12, 0.00], 0.079 −0.06 [−0.14, 0.02], 0.19 −0.07 [−0.14, −0.01], 0.023 −0.06 [−0.15, 0.02], 0.25 −0.07 [−0.14, −0.01], 0.028

0.03 [−0.03, 0.08], 0.62 0.05 [0.00, 0.09], 0.037

0.08 [0.02, 0.15], 0.008 0.06 [0.00, 0.11], 0.055

0.02 [−0.03, 0.08], 0.74 0.06 [0.02, 0.11], 0.002

0.09 [0.02, 0.15], 0.007 0.08 [0.02, 0.14], 0.002

0.53 (0.03) 0.53 (0.02) 0.59 (0.03) 0.55 (0.02)

0.13 [0.00, 0.26], 0.063 7 [0, 13], 0.034 0.00 [−0.03, 0.02], 1.0

Arthroplasty

0.14 [−0.02, 0.30], 0.12 4 [−4, 12], 0.66 0.00 [−0.03, 0.03], 1.0

Arthrodesis

0.18 [0.02, 0.34], 0.020 0.12 [−0.01, 0.25], 0.078 4 [−4, 11], 0.65 6 [0, 12], 0.031 −0.01 [−0.04, 0.02], 0.89 −0.01 [−0.03, 0.02], 0.98

Arthroplasty

P2-P0

1.02 (0.06) 1.00 (0.05) 107 (4) 108 (3) 0.12 (0.01) 0.09 (0.01)

Arthrodesis Arthroplasty Arthrodesis

P0

MEAN DIFFERENCES by session [95% Confidence Intervals], P-values

Stance duration = time period from heel strike to toe off. Step duration = time period from heel strike to contralateral limb heel strike.

Speed (m/s) Cadence (steps/min) Step width (m) Step length (m) Affected limb Unaffected limb Stance duration (s)a Affected limb Unaffected limb Step duration (s)b Affected limb Unaffected limb

Session

MEAN (SE)

Table 3 Mean (SE) spatiotemporal variables for the baseline session (P0) and mean differences from P0 by session (P1, P2, P3) and [95% confidence intervals] separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected) where applicable, with associated P-values. Bold text indicates a significant difference (P < 0.05).

−1.0 [−4.4, 2.5], 0.98 0.9 [−2.0, 3.7], 0.97 1.8 [−2.6, 6.3], 0.87 −0.4 [−1.6, 0.9], 0.98 0.2 [−0.9, 1.2], 1.0 0.5 [−1.1, 2.1], 0.97

Arthrodesis

Arthrodesis

Arthroplasty

P1–P0

Mean differences by session [95% confidence intervals], P-value

P0

Weight (kg) 90.6 (5.0) 86.7 (4.0) Arthroplasty-Arthrodesis Difference BMI 29.4 (1.2) 30.9 (1.0) Arthroplasty-Arthrodesis Difference

Session

Mean (SE)

Table 2b Mean (SE) patient demographics at baseline session (P0) and mean differences from P0 by session (P1–P3) with [95% Confidence Intervals] separated by surgery type (arthrodesis, arthroplasty) with associated P-values. Bold values indicate a significant difference (P < 0.05).

A.D. Segal et al.

Clinical Biomechanics 54 (2018) 42–53

Ankle RoM1 (early to mid-stance) Affected limb Arthroplasty-Arthrodesis Difference Unaffected limb Ankle RoM2 (mid- to terminal stance) Affected limb Adjusted for speed Unaffected limb Knee RoM1 (early to mid-stance) Affected limb Unaffected limb Knee RoM2 (mid-stance to mid-swing) Affected limb Unaffected limb Hip RoM1 (early to mid-stance) Affected limb Unaffected limb Hip RoM2 (mid-stance to mid-swing) Affected limb Unaffected limb

Session

47 17.5 (1.1) 17.4 (1.1) 25.0 (1.3) 14.3 (1.1) 9.9 (1.3) 53.7 (1.9) 51.6 (1.7) 38.7 (1.3) 38.5 (1.2) 39.4 (1.3) 39.3 (1.1)

17.4 (1.4) 15.9 (1.6)

49.7 (2.3) 50.3 (2.1)

39.2 (1.6) 40.4 (1.4)

38.8 (1.6) 40.0 (1.3)

19.6 (1.0)

18.1 (1.2)

18.9 (1.3) 18.8 (1.3) 25.0 (1.6)

13.5 (0.7)

16.2 (0.9)

Arthroplasty

0.4 [−2.4, 3.3], 1.0

1.1 [−1.9, 4.0], 0.87

4.0 [0.6, 7.5], 0.012*

4.2 [0.7, 7.8], 0.011*

−0.2 [−3.5, 3.0], 1.0

−0.3 [−2.2, 1.6], 1.0

4.2 [0.4, 7.9], 0.020*

4.2 [0.5, 7.8], 0.019*

3.8 [−1.0, 8.6], 0.19

−2.6 [−6.7, 1.5], 0.37 −4.3 [−8.2, −0.4], 0.023

0.9 [−2.7, 4.6], 0.97

Arthroplasty

1.1[−2.0, 4.1], 0.88

3.2 [−0.6, 7.1], 0.15

3.3 [−0.3, 7.0], 0.086

4.8 [−0.6, 10.3], 0.10

−0.3 [−4.1, 3.6], 1.0

−1.6, [−5.3, 2.2], 0.75 −3.1 [−6.5, 0.4], 0.11

−0.4 [−3.2, 2.4], 1.0

0.5 [−2.5, 3.4], 1.0

Arthroplasty

1.0 [−2.2, 4.1], 0.92

0.6 [−2.4, 3.5], 1.0

−0.3 [−4.7, 4.1], 1.0

−1.7 [−3.9, 0.5], 0.21

0.9 [−2.2, 3.9], 0.92 −0.1 [−2.9, 2.7], 1.0

−2.1 [−5.0, 0.8], 0.27 2.2 [−0.2, 4.5], 0.080 4.3 [0.5, 8.0], 0.016

Arthrodesis

P3–P0

−1.1 [−3.7, 1.4], 0.81

−0.1 [−3.4, 3.2], 1.0 −1.1 [−4.2, 2.1], 0.86

−2.4 [−5.2, 0.5], 0.15 1.6 [−0.6, 3.9], 0.28 4.0 [0.4, 7.6], 0.023

Arthrodesis

P2–P0

−0.6 [−4.1, 2.8], 0.99 −1.6 [−4.7, 1.6], 0.60

4.2 [−0.2, 8.6], 0.073

0.3 [−2.0, 2.7], 1.0

−2.7 [−6.9, 1.5], 0.36 −4.4 [−8.3, −0.5], 0.019

−1.9 [−4.4, 0.9], 0.33 2.3 [0.0, 4.5], 0.045 4.1 [0.6, 7.7], 0.013

Arthrodesis

Arthrodesis

Arthroplasty

P1–P0

Mean differences by session [95% confidence intervals], P-values

P0

Mean (SE)

Table 4 Baseline session (P0) mean (SE) sagittal joint ankle range of motion (deg) in early to mid-stance (RoM1) and mid-stance to terminal stance/mid-swing (RoM2) separated by surgery type (arthrodesis, arthroplasty) and by limb (affected, unaffected), mean differences from P0 by session (P1, P2, P3) and [95% confidence intervals] separated by surgery type for the affected limb with associated P-values, and mean differences by surgery (Arthroplasty-Arthrodesis Difference) separated by session when a significant difference was detected. Bold text indicates a significant difference (P < 0.05). An asterisk (*) identifies significant comparisons that become non-significant after adjusting for walking speed.

A.D. Segal et al.

Clinical Biomechanics 54 (2018) 42–53

Heel strike transient (HST) peak Affected limb Unaffected limb HST prevalence (%) Affected limb Unaffected limb GRFZ1max(early to mid-stance) Affected limb Unaffected limb GRFZ2max (mid- to terminal stance) Affected limb Unaffected limb GRFZ min (mid-stance) Affected limb Unaffected limb

Session

P1–P0

48 10.0 (0.1) 10.4 (0.2) 9.8 (0.1) 10.2 (0.1) 8.3 (0.1) 8.2 (0.1)

9.5 (0.2) 10.2 (0.2)

8.3 (0.2) 8.1 (0.2)

75 (7) 19 (8)

82 (8) 40 (10)

9.8 (0.1) 10.0 (0.2)

4.1 (0.3) 5.1 (0.5)

4.2 (0.3) 6.3 (0.4)

0.3 [−0.1, 0.8], 0.26

−0.1 [−0.4, 0.3], 1.0

−0.3 [−0.6, 0.0], 0.11 −0.4 [−0.8, 0.1], 0.21

0.3 [−0.1, 0.8], 0.24

−0.6 [−1.0, −0.2], < 0.001*

6 [−24, 35], 1.0

−8 [−30, 14], 0.92

0.2 [−0.3, 0.7], 0.87

1.4 [0.5, 2.4], < 0.001

0.4 [−0.4, 1.3], 0.66

Arthrodesis

0.1 [−0.2, 0.4], 0.93

0.3 [−0.1,0.7], 0.20

4 [−22, 30], 1.0

1.3 [0.3, 2.3], 0.004*

Arthroplasty

P2–P0

Mean differences by session [95% confidence intervals], P-values

Arthrodesis Arthroplasty Arthrodesis

P0

Mean (SE)

0.6 [0.1, 1.2], 0.005*

0.3 [0.0, 0.7], 0.12

4 [−25, 34], 1.0

1.4 [0.4, 2.5], 0.002

Arthrodesis

0.3 [−0.1, 0.7], 0.27

0.1 [−0.2, 0.4], 0.90

−29 [−53, −6], 0.007*

0.6 [−0.2, 1.5], 0.27

Arthroplasty

−0.2 [−0.6, 0.1], 0.46 −0.6 [−1.1, −0.1], 0.007* −0.2 [−0.6, 0.2], 0.78

0.1 [−0.3, 0.5], 0.96

0.0 [−0.4, 0.4], 1.0

−9 [−32, 14], 0.89

0.3 [−0.5, 1.1], 0.87

Arthroplasty

P3–P0

Table 5 Mean (SE) vertical ground reaction force (GRFZ, N/kg) at four discrete peaks for the baseline session (P0) separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected) and mean differences from P0 by session (P1, P2, P3) and [95% confidence intervals] separated by surgery type for the affected limb with associated P-values. Bold text indicates a significant difference (P < 0.05). An asterisk (*) identifies significant comparisons that become non-significant after adjusting for walking speed.

A.D. Segal et al.

Clinical Biomechanics 54 (2018) 42–53

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

Fig. 1. Average sagittal plane joint angles (deg) across the gait cycle (%) for baseline (P0) and three annual follow-up sessions (P1, P2, P3) separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected). Discrete peaks to measure the range of motion (RoM) for each joint in early to mid-stance (RoM1: A1-A2, K1-K2, H1-H2) and mid-stance to terminal stance (ankle) or mid-swing (knee, hip; RoM2: A2-A3, K2-K3, H2-H3) are depicted for each joint.

Fig. 2. Average sagittal plane joint moments (Nm/kg) across the gait cycle (%) for baseline (P0) and three annual follow-up sessions (P1, P2, P3) separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected). Discrete peaks (A1m, K1m, K2m, H1m − H3m) are depicted for each joint.

49

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

Fig. 3. Average sagittal plane joint powers (W/kg) across the gait cycle (%) for baseline (P0) and three annual follow-up sessions (P1, P2, P3) separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected). Discrete peaks (A1p, A2p, K1p − K4p, H1p − H3p) are depicted for each joint.

Fig. 4. Average vertical ground reaction force (GRF, N/kg) across the gait cycle (%) for baseline (P0) and three annual follow-up sessions (P1, P2, P3) separated by surgery type (arthrodesis, arthroplasty) and limb (affected, unaffected). Discrete peaks (HST, GRFZ1max, GRFZ2max, GRFZmin) are depicted in the first plot.

3.6. Surveys and step count

peak between surgery types and a corresponding borderline difference in HST prevalence (surgery by session interaction, HST peak: P = 0.056, HST prevalence: P = 0.098), both of which became significant after adjusting for walking speed (HST prevalence used an average walking speed across trial; surgery by session interaction, HST peak: P = 0.0007, HST prevalence: P = 0.020). For arthrodesis patients, the HST peak increased (+Δ1.4 N/kg [0.4, 2.3], P < 0.001) and the HST prevalence remained consistent across sessions (+Δ5% [−18%, 27%], P = 1.0). In contrast, for arthroplasty patients, HST peak was more consistent (+Δ0.4 N/kg [−0.3, 1.2], P = 0.39) and tended to be less prevalent across sessions (−Δ15% [−34, 0], P = 0.17), with the most pronounced decrease (−Δ29%) occurring at the third follow-up session (Table 5). Finally, a steeper valley at midstance (GRFZmin) was found for arthrodesis patients across sessions (−Δ0.5 [0.1, 0.9], P = 0.006), which was primarily associated with increased walking speed.

At baseline, survey scores and step counts were similar for both surgical groups (Table 2a). Postoperatively, both patient groups reported similar improvements in function and reductions in pain across session (surgery by session interaction, P > 0.35) with decreased MFA scores (−Δ16.2 [11.6, 20.7], P < 0.001), increased SF-36 physical function (+Δ28.0 [19.2, 36.7], P < 0.001) and body pain scores (+Δ36.5 [26.8, 46.1]), and decreased pain scores (−Δ4.2 [3.1, 5.2], P < 0.001). Improvements were also consistent by follow-up session (Table S3). Finally, no changes in total steps per day or steps at low-, medium- or high-intensity per day were detected for either surgery type. 4. Discussion The purpose of this prospective study was to compare the postoperative changes in lower limb biomechanics of patients after ankle 50

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

with few detectable differences related to either surgery. Additionally, postoperative ankle push-off power remained approximately 30% reduced for both surgeries compared to their respective unaffected limbs, which was similar to a prior study of arthroplasty patients and their matched controls (Valderrabano et al., 2007). Brodsky et al. reported an even larger average difference (~50%) in peak push-off power between limbs in a group of arthrodesis patients (Brodsky et al., 2016). However, the reduced affected limb postoperative push-off power did not lead to appreciable increases in the unaffected limb vertical ground reaction force, as predicted by dynamic walking models that suggest increased collision forces result from reduced push-off power on the contralateral limb (Adamczyk and Kuo, 2009). Reduced push-off power may also lead to increased peak knee external adduction moments and an increased risk of knee osteoarthritis (e.g., Foroughi et al., 2009), which was approximately four times as prevalent in patients 22 years (12–44 range) after ankle arthrodesis (~40%; Coester et al., 2001) versus the general population over the age of 60 years (Zhang and Jordan, 2010). Therefore, we performed a secondary analysis to examine the unaffected limb peak knee external adduction moment. There was a trend of increased peak knee moment for arthrodesis patients (Table S1, Fig. S1), but we were unable to detect a difference possibly due to the relatively small and variable population. Despite relatively small magnitudes (< 5 deg), the ankle RoM results supported our hypotheses of increased RoM for arthroplasty and decreased RoM for arthrodesis. The simplistic single segment foot model may have underestimated differences between surgeries since it does not account for the foot's complex structure and therefore, cannot distinguish distal foot joint contributions. However, this work reinforces that both surgical procedures were not able to fully restore ankle RoM during gait as compared to the unaffected limb, and that the predominant deficiency persisted in late stance plantar flexion, similar to Brodsky et al. (Brodsky et al., 2016). A similar difference in ankle RoM has also been reported compared to a control group (Singer et al., 2013). The arthroplasty patients' restricted ankle RoM post-surgery may be attributed to (1) the 2-component, fixed bearing implant design, (2) inherent limited use of the ankle after years of pain management, and (3) insufficient rehabilitation leading to loss of proprioception, plantar flexor weakness, and/or the formation of scar tissue (Valderrabano et al., 2007). In contrast, the reduced ankle RoM is an expected outcome for arthrodesis patients, yet is contrary to Flavin et al. (Flavin et al., 2013). To clarify the kinematic foot joint contributions, a followup analysis is underway using a multi-segment foot model to further study ankle arthritic patients. Several limitations should be considered when interpreting the findings of this study. A small sample size of a heterogeneous population can contribute to increased variability and the inability to detect small differences. We chose the linear mixed effects model to account for missing data, which maximizes the number of subjects and trials included in the analysis. The prospective study design and comparison to the unaffected limb further improves the study's power to detect differences. However, non-significant findings should be interpreted cautiously. One of the pitfalls of a longitudinal study is that at the time of publication, the procedures may be outdated due to the constant development of new techniques and devices. However, with increased device longevity, many patients will continue to use older technology and may benefit from a better understanding of their gait biomechanics to adapt daily behavior and improve outcomes.

arthrodesis or arthroplasty in one-year time intervals for three consecutive years. We found postoperative changes predominantly occurred at the first-year follow-up session, and these outcomes were generally maintained across the 3-year period. Exceptions included weight loss for the arthrodesis patients and decreased impact loading for arthroplasty patients at the third-year follow-up compared to baseline. Increased hip RoM for arthrodesis patients and increased cadence for arthroplasty patients were also more pronounced in the first two years post-surgery; however, the increased trends remained in the third follow-up session. Finally, while both patient groups comparably increased walking speed across follow-up sessions, they used different strategies to increase speed. Both surgical groups walked at a faster pace postoperatively (12% average increase), resulting in an average speed that was similar to a prior study of arthroplasty patients and their matched controls (Valderrabano et al., 2007). This finding is in contrast to several other studies that also reported increased walking speed after ankle arthroplasty (Dyrby et al., 2004; Flavin et al., 2013; Ingrosso et al., 2009) and arthrodesis (Brodsky et al., 2016; Flavin et al., 2013), but the speeds in these studies remained approximately 20% reduced compared to matched controls. Despite achieving equivalent speeds in this study, each surgical group implemented a different strategy to increase speed. Arthrodesis patients increased their affected step length while arthroplasty patients increased cadence and reduced step duration. The increased step length approach used by arthrodesis patients was accompanied by increased hip RoM, attributed mostly to increased terminal stance hip extension, and similar to prior study (Brodsky et al., 2016). The increased hip RoM may be compensatory movement for reduced ankle RoM. In contrast, the arthroplasty patients adopted a faster cadence while maintaining consistent step length and hip RoM postoperatively. Therefore, arthrodesis patients took fewer and longer steps with greater hip RoM to cover the same distance compared to arthroplasty patients, possibly at the expense of altered limb loading. Despite demonstrating the ability to walk at a near normal speed postoperatively and reduced pain, daily activity outside the laboratory did not increase for either surgery as measured through a lack of change in daily step count. However, small trends of increased activity may exist that we were unable to detect due to our small and variable sample (Table S3). Altered step length and joint motion may influence limb positioning at initial foot contact, increasing impact loading and transient stresses acting on the lower extremities. Despite the possible correlation between elevated impulsive forces at heel strike and pathological conditions including degenerative joint disease (Collins and Whittle, 1989; Radin, 1987; Radin et al., 1991; Radin and Paul, 1971; Whittle, 1999) and the reduced ability for those with knee pathology to attenuate high frequency impact loads (Chu et al., 1986), the effect of arthrodesis or arthroplasty surgery on the HST peak remains unclear. At the three-year follow-up, arthrodesis patients presented a consistent 25% greater affected limb HST peak that occurred more than twice as frequently (~86%) compared to their unaffected limb (~40%) and the general population (33%) (Radin et al., 1986). In contrast, the affected limb HST peak for arthroplasty patients was similar to their preoperative levels and less prevalent (~46%) at the third follow-up session compared to baseline (75%). Two prior studies also depicted the presence of a HST peak after arthrodesis, which was absent after arthroplasty, but neither study included this metric in their analysis (Flavin et al., 2013; Piriou et al., 2008). Patients may be able to mitigate these impact loads by wearing shoes that absorb transient forces and facilitate proper limb positioning at initial contact, with certain viscoelastic materials able to provide cushioning and reduce forces by as much as 46% (Lafortune and Hennig, 1992). Therefore, arthrodesis patients who avoid barefoot walking may minimize impulsive loading (Johnson, 1988; Voloshin and Wosk, 1981; Whittle, 1999) and improve long-term outcomes. The increased impact loading for arthrodesis patients did not translate into postoperative changes in sagittal moments and powers,

5. Conclusions Biomechanical studies of ankle arthrodesis or arthroplasty patients can assess functional improvements at one-year post-surgery, which are maintained after three years. However, longer-term outcomes (> 5 years) require further exploration. Both surgeries improved pain and function with small biomechanical changes between surgeries that may affect longer-term outcomes. 51

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

Supplementary data to this article can be found online at https:// doi.org/10.1016/j.clinbiomech.2018.02.018.

adduction moment with biomechanical variables in osteoarthritis: a systematic review. Knee 16 (5), 303–309. Foster, R.C., Lanningham-Foster, L.M., Manohar, C., McCrady, S.K., Nysse, L.J., Kaufman, K.R., Padgett, D.J., Levine, J.A., 2005. Precision and accuracy of an ankle-worn accelerometer-based pedometer in step counting and energy expenditure. Prev. Med. 41 (3–4), 778–783. Glazebrook, M., Daniels, T., Younger, A., Foote, C.J., Penner, M., Wing, K., Lau, J., Leighton, R., Dunbar, M., 2008. Comparison of health-related quality of life between patients with end-stage ankle and hip arthrosis. J. Bone Joint Surg. Am. 90A (3), 499–505. Hahn, M.E., Wright, E.S., Segal, A.D., Orendurff, M.S., Ledoux, W.R., Sangeorzan, B.J., 2012. Comparative gait analysis of ankle arthrodesis and arthroplasty: initial findings of a prospective study. Foot Ankle Int. 33 (4), 282–289. Hofmann, K.J., Shabin, Z.M., Ferkel, E., Jockel, J., Slovenkai, M.P., 2016. Salto Talaris total ankle arthroplasty: clinical results at a mean of 5.2 years in 78 patients treated by a single surgeon. J. Bone Joint Surg. Am. 98 (24), 2036–2046. Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in general parametric models. Biom. J. 50 (3), 346–363. Ingrosso, S., Benedetti, M.G., Leardini, A., Casanelli, S., Sforza, T., Giannini, S., 2009. GAIT analysis in patients operated with a novel total ankle prosthesis. Gait Posture 30 (2), 132–137. Johnson, G.R., 1988. The effectiveness of shock-absorbing insoles during normal walking. Prosthetics Orthot. Int. 12 (2), 91–95. Khazzam, M., Long, J.T., Marks, R.M., Harris, G.F., 2006. Preoperative gait characterization of patients with ankle arthrosis. Gait Posture 24 (1), 85–93. Lafortune, M.A., Hennig, E.M., 1992. Cushioning properties of footwear during walking: accelerometer and force platform measurements. Clin. Biomech. 7 (3), 181–184. McHorney, C.A., Ware Jr., J.E., Lu, J.F., Sherbourne, C.D., 1994. The MOS 36-item ShortForm Health Survey (SF-36): III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med. Care 32 (1), 40–66. Michael, J.M., Golshani, A., Gargac, S., Goswami, T., 2008. Biomechanics of the ankle joint and clinical outcomes of total ankle replacement. J. Mech. Behav. Biomed. Mater. 1 (4), 276–294. Pedowitz, D.I., Kane, J.M., Smith, G.M., Saffel, H.L., Comer, C., Raikin, S.M., 2016. Total ankle arthroplasty versus ankle arthrodesis: a comparative analysis of arc of movement and functional outcomes. Bone Joint J. 98-B (5), 634–640. Piriou, P., Culpan, P., Mullins, M., Cardon, J.N., Pozzi, D., Judet, T., 2008. Ankle replacement versus arthrodesis: a comparative gait analysis study. Foot Ankle Int. 29 (1), 3–9. Queen, R.M., De Biassio, J.C., Butler, R.J., DeOrio, J.K., Easley, M.E., Nunley, J.A., 2012. Changes in pain, function, and gait mechanics two years following total ankle arthroplasty performed with two modern fixed-bearing prostheses. Foot Ankle Int. 33 (7), 535–542. Radin, E.L., 1987. Osteoarthrosis. What is known about prevention. Clin. Orthop. Relat. Res. 222, 60–65. Radin, E.L., Paul, I.L., 1971. Response of joints to impact loading. I. In vitro wear. Arthritis Rheum. 14 (3), 356–362. Radin, E.L., Whittle, M.W., Yang, K.H., Jeffereson, R.J., Rodgers, M.M., Kish, V.L., O'Connor, J.J., 1986. The heelstrike transient, its relationship with the angular velocity of the shank and the effects of quadriceps paralysis. In: Lantz, S.H. (Ed.), Advances in Bioengineering, Ed. K.A. American Society of Mechanical Engineers, New York. Radin, E.L., Burr, D.B., Caterson, B., Fyhrie, D., Brown, T.D., Boyd, R.D., 1991. Mechanical determinants of osteoarthrosis. Semin. Arthritis Rheum. 21 (3 Suppl. 2), 12–21. Resnick, B., Nahm, E.S., Orwig, D., Zimmerman, S.S., Magaziner, J., 2001. Measurement of activity in older adults: reliability and validity of the Step Activity Monitor. J. Nurs. Meas. 9 (3), 275–290. Saltzman, C.L., Salamon, M.L., Blanchard, G.M., Huff, T., Hayes, A., Buckwalter, J.A., Amendola, A., 2005. Epidemiology of ankle arthritis: report of a consecutive series of 639 patients from a tertiary orthopaedic center. Iowa Orthop. J. 25, 44–46. Segal, A.D., Shofer, J., Hahn, M.E., Orendurff, M.S., Ledoux, W.R., Sangeorzan, B.J., 2012. Functional limitations associated with end-stage ankle arthritis. J. Bone Joint Surg. Am. 94 (9), 777–783. Shepherd, E.F., Toloza, E., McClung, C.D., Schmalzried, T.P., 1999. Step activity monitor: increased accuracy in quantifying ambulatory activity. J. Orthop. Res. 17 (5), 703–708. Singer, S., Klejman, S., Pinsker, E., Houck, J., Daniels, T., 2013. Ankle arthroplasty and ankle arthrodesis: gait analysis compared with normal controls. J. Bone Joint Surg. Am. 95 (24), e191(1-10). SooHoo, N.F., Zingmond, D.S., Ko, C.Y., 2007. Comparison of reoperation rates following ankle arthrodesis and total ankle arthroplasty. J. Bone Joint Surg. Am. 89 (10), 2143–2149. Takakura, Y., Tanaka, Y., Sugimoto, K., Akiyama, K., Tamai, S., 1999. Long-term results of arthrodesis for osteoarthritis of the ankle. Clin. Orthop. Relat. Res. 361, 178–185. Team, R.C., 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing Available from. https://www.R-project.org/. Valderrabano, V., Hintermann, B., Dick, W., 2004. Scandinavian total ankle replacement: a 3.7-year average followup of 65 patients. Clin. Orthop. Relat. Res. 424, 47–56. Valderrabano, V., Nigg, B.M., von Tschamer, V., Stefanyshyn, D.J., Goepfert, B., Hintermann, B., 2007. Gait analysis in ankle osteoarthritis and total ankle replacement. Clin. Biomech. 22 (8), 894–904. Voloshin, A., Wosk, J., 1981. Influence of artificial shock absorbers on human gait. Clin. Orthop. Relat. Res.(160), 52-6.

Conflict of interest All products identified in this article were independently purchased through commercial channels. No commercial party having a direct financial interest in the results of the research reported in this article has or will confer a benefit on the authors or on any organization with which the authors are associated. Funding This work was funded by the Department of Veterans Affairs Rehabilitation Research and Development Service grants A4513R and A4843C. The contents do not represent the views of the U.S. Department of Veterans Affairs or the United States Government. Acknowledgements The authors would like to thank Jane Shofer, MS, for conducting the statistical analyses and Eric Rohr, Elise Wright, Hannah Sutton, Marisa Benich and Jacynda Wheeler who contributed to subject recruitment, data collection or data processing. Registration of clinical trials This non-randomized, prospective study entitled “Treatment Outcomes for Ankle Arthritis” was registered in clinicaltrials.gov on October 20, 2006 (Identifier: NCT00391365, Study ID# F4513-R) and was last updated on March 15, 2017. https://clinicaltrials.gov/ct2/ show/NCT00391365?term=sangeorzan&rank=2. References Adamczyk, P.G., Kuo, A.D., 2009. Redirection of center-of-mass velocity during the stepto-step transition of human walking. J. Exp. Biol. 212 (Pt 16), 2668–2678. Agel, J., Coetzee, J.C., Sangeorzan, B.J., Roberts, M.M., Hansen Jr., S.T., 2005. Functional limitations of patients with end-stage ankle arthrosis. Foot Ankle Int. 26 (7), 537–539. Brandes, M., Schomaker, R., Mollenhoff, G., Rosenbaum, D., 2008. Quantity versus quality of gait and quality of life in patients with osteoarthritis. Gait Posture 28 (1), 74–79. Brockett, C.L., Chapman, G.J., 2016. Biomechanics of the ankle. Orthop. Traumatol. 30 (3), 232–238. Brodsky, J.W., Polo, F.E., Coleman, S.C., Bruck, N., 2011. Changes in gait following the Scandinavian Total Ankle Replacement. J. Bone Joint Surg. Am. 93 (20), 1890–1896. Brodsky, J.W., Kane, J.M., Coleman, S., Bariteau, J., Tenenbaum, S., 2016. Abnormalities of gait caused by ankle arthritis are improved by ankle arthrodesis. Bone Joint J. 98B (10), 1369–1375. Brown, T.D., Johnston, R.C., Saltzman, C.L., Marsh, J.L., Buckwalter, J.A., 2006. Posttraumatic osteoarthritis: a first estimate of incidence, prevalence, and burden of disease. J. Orthop. Trauma 20 (10), 739–744. Choi, J.H., Coleman, S.C., Tenenbaum, S., Polo, F.E., Brodsky, J.W., 2013. Prospective study of the effect on gait of a two-component total ankle replacement. Foot Ankle Int. 34 (11), 1472–1478. Chu, M.L., Yazdani-Ardakani, S., Gradisar, I.A., Askew, M.J., 1986. An in vitro simulation study of impulsive force transmission along the lower skeletal extremity. J. Biomech. 19 (12), 979–987. Coester, L.M., Saltzman, C.L., Leupold, J., Pontarelli, W., 2001. Long-term results following ankle arthrodesis for post-traumatic arthritis. J. Bone Joint Surg. Am. 83-A (2), 219–228. Collins, J.J., Whittle, M.W., 1989. Impulsive forces during walking and their clinical implications. Clin. Biomech. 4 (3), 179–187. Daniels, T.R., Younger, A.S., Penner, M., Wing, K., Dryden, P.J., Wong, H., Glazebrook, M., 2014. Intermediate-term results of total ankle replacement and ankle arthrodesis: a COFAS multicenter study. J. Bone Joint Surg. Am. 96 (2), 135–142. Dyrby, C., Chou, L.B., Andriacchi, T.P., Mann, R.A., 2004. Functional evaluation of the Scandinavian Total Ankle Replacement. Foot Ankle Int. 25 (6), 377–381. Espinosa, N., Klammer, G., 2010. Treatment of ankle osteoarthritis: arthrodesis versus total ankle replacement. Eur. J. Trauma Emerg. Surg. 36 (6), 525–535. Flavin, R., Coleman, S.C., Tenenbaum, S., Brodsky, J.W., 2013. Comparison of gait after total ankle arthroplasty and ankle arthrodesis. Foot Ankle Int. 34 (10), 1340–1348. Foroughi, N., Smith, R., Vanwanseele, B., 2009. The association of external knee

52

Clinical Biomechanics 54 (2018) 42–53

A.D. Segal et al.

Winter, D.A., 1991. The Biomechanics and Motor Control of Human Gait: Normal, Elderly and Pathological. University of Waterloo Press, Waterloo. Zhang, Y., Jordan, J.M., 2010. Epidemiology of osteoarthritis. Clin. Geriatr. Med. 26 (3), 355–369.

Ware Jr., J.E., Sherbourne, C.D., 1992. The MOS 36-item short-form health survey (SF36). I. Conceptual framework and item selection. Med. Care 30 (6), 473–483. Whittle, M.W., 1999. Generation and attenuation of transient impulsive forces beneath the foot: a review. Gait Posture 10 (3), 264–275.

53