Gait & Posture 70 (2019) 24–29

Contents lists available at ScienceDirect

Gait & Posture

journal homepage: www.elsevier.com/locate/gaitpost

Physical activity and age-related biomechanical risk factors for knee T

osteoarthritis
Jocelyn F. Hafera,b, , Jane A. Kentb, Katherine A. Boyerb,c
⁎

a
School of Kinesiology, University of Michigan, Ann Arbor, MI, United States
b
Department of Kinesiology, University of Massachusetts Amherst, Amherst, MA, United States
c
Department of Orthopedics and Physical Rehabilitation, University of Massachusetts Medical School, Worcester, MA, United States

ARTICLE INFO ABSTRACT

Keywords: Background: Knee osteoarthritis (OA) is a highly prevalent disease leading to mobility disability in the aged that
Muscle strength could, in part, be initiated by age-related alterations in knee mechanics. However, if and how knee mechanics
Muscle power change with age remains unclear.
Older gait Research question: What are the impacts of age and physical activity (PA) on biomechanical characteristics that
Co-activation
can affect the loading environment in the knee during gait?
Knee mechanics
Methods: Three groups (n = 20 each, 10 male and 10 female) of healthy adults were recruited: young (Y, 21–35
years), mid-life highly active (MHi, 55–70 years, runners), and mid-life less active (MLo, 55–70 years, low PA).
Outcome measures included knee kinematics and kinetics and co-activation during gait, and knee extensor
muscle torque and power collected at baseline and after a 30-minute treadmill trial to determine the impact of
prolonged walking on knee function.
Results: At baseline, high-velocity concentric knee extensor power was lower for MLo and MHi compared with Y,
and MLo displayed greater early (6.0 ± 5.8 mm) and peak during stance (11.3 ± 7.8 mm) femoral anterior
displacement relative to the tibia compared with Y (0.2 ± 5.6 and 4.4 ± 6.8 mm). Also at baseline, MLo
showed equal quadriceps:hamstrings activation, while Y showed greater relative hamstrings activation during
midstance. The walking bout induced substantial knee extensor fatigue (decrease in maximal torque and power)
in Y and MLo, while MHi were fatigue-resistant.
Significance: These results indicate that maintenance of PA in mid-life may impart small but measurable effects
on knee function and biomechanics that may translate to a more stable loading environment in the knee through
mid-life and thus could reduce knee OA risk long-term.

1. Introduction anterior displacement relative to the tibia [9] compared with young
adults. These differences are also characteristic of symptomatic knee
Knee osteoarthritis (OA) is a mobility-limiting, age-related disease OA [9,10]. The co-occurrence of the altered knee mechanics with age
for which there is a 44% lifetime risk in American adults [1]. Main- and knee OA suggests these changes may precede the onset of symp-
tenance of physical activity (PA) throughout the lifespan may reduce toms thereby supporting a mechanical pathway for idiopathic knee OA
OA risk [2], possibly by mitigating changes in several knee OA bio- initiation [11]. However, as highlighted in a recent meta-analysis, few
mechanical risk factors that are also associated with aging (e.g., de- studies have quantified the age-related changes in knee mechanics and
creased quadriceps strength, increased muscle co-activation, altered if, how, and in whom these changes occur remains unclear [7].
knee mechanics during gait) [3–5]. As rates of knee OA initiation in- Abnormal knee mechanics could partially result from decreased
crease rapidly from age 50–70 years [6], quantifying age-related muscle strength. Around mid-life, decreases in knee extensor muscle
changes in biomechanical OA risk factors in mid-life, and the potential torque and power [12] and increased knee extensor fatigue following
for PA to mitigate them, is critical. dynamic contractions begin to occur [13]. Decreased knee extensor
There is initial evidence that healthy adults at mid-life or older walk function is associated with knee OA initiation [14] and altered knee
with less knee flexion at heel strike and a smaller range of motion at the flexion angles during gait in individuals at risk of knee OA post-ACL
knee [7], greater knee flexion in midstance [8], and greater femoral rupture [15] or with current knee OA [16]. Because low knee extensor

⁎
Corresponding author at: School of Kinesiology, University of Michigan, 401 Washtenaw Ave., Ann Arbor, MI, 48109, United States.
E-mail address: johafer@umich.edu (J.F. Hafer).

https://doi.org/10.1016/j.gaitpost.2019.02.008
Received 15 November 2018; Received in revised form 15 January 2019; Accepted 12 February 2019
0966-6362/ © 2019 Elsevier B.V. All rights reserved.
J.F. Hafer, et al. Gait & Posture 70 (2019) 24–29

strength is itself a risk factor for knee OA, further loss of strength with 2.2.1. Physical activity monitoring
fatigue could amplify knee OA risk. Changes in knee mechanics may All participants wore triaxial accelerometers (GT3X, Actigraph,
also result from altered muscle activation. Initial evidence suggests Pensacola, FL) at the hip for 7 days. PA data included ≥4 days of ≥10 h
mid-life adults with and without knee OA have greater muscle activa- wear, including ≥1 weekend day. Accelerometer data were used to
tion across the knee compared to young adults [17], which could alter calculate average weekly activity counts and moderate-to-vigorous PA
knee joint loading [18]. (MVPA) minutes [22].
If changes in strength or muscle activation drive age-related changes
in knee mechanics, we would expect greater deviations in knee me- 2.2.2. Gait analysis
chanics in adults with poorer knee extensor function. Higher PA is as- Overground gait was captured before and after the 30MTW.
sociated with greater knee extensor muscle torque and power in mid- Kinematics and kinetics of participants’ right leg were captured using
life and older adults [19], and has been shown to not increase the risk an 11-camera motion analysis system (Oqus, Qualisys, Göteborg,
[20] and possibly protect against [2] and slow the progression of knee Sweden) with 2 force plates (AMTI, Watertown, MA). Marker and force
OA [21]. However, there is a lack of information on the role of PA in data were collected at 200 and 2000 Hz, and low-pass filtered at 8 and
biomechanical risk factors for knee OA, specifically, knee extensor 15 Hz, respectively. Five acceptable trials were captured at each of 2
muscle strength and fatigability and muscle co-activation and knee speeds: preferred and fixed (1.4 m·s−1). Acceptable trials involved the
mechanics during gait. participant cleanly hitting a force plate with their right foot at a speed
This study’s primary aim was to determine whether knee extensor within 5% of the other trials for that condition (monitored via photo-
muscle function (here, maximal torque and power), co-activation across gates).
the knee during gait, and knee mechanics differ between young (Y) and Thigh and shank segments were modeled using the Point Cluster
mid-life adults, and between mid-life adults with high (MHi) or low Technique (PCT). PCT is a previously-validated [23,24] marker con-
(MLo) PA levels. We hypothesized that MLo would be weaker, have figuration and algorithm optimized for calculation of the 3 rotations
altered muscle co-activation, and display different knee mechanics and 3 translations at the knee joint using clusters of markers on the
compared to Y and MHi. Further, even if individuals are similar at thigh (10 markers) and shank (7 markers). Pelvis, thigh, shank and foot
baseline, the potential for daily bouts of activity to induce muscle fa- coordinate systems were established during a static trial from anatomic
tigue may predispose certain populations to fatigue-related changes in markers (anterior and posterior iliac spine, iliac crest, greater tro-
knee mechanics or muscle activation and, potentially, increased knee chanter, medial and lateral femoral epicondyles, medial and lateral ti-
joint loads. Therefore, our secondary aim was to test the impact of an bial plateau, medial and lateral malleoli, calcaneus and 5th metatarsal).
acute exercise bout on knee extensor muscle function and knee me- The foot and pelvis were tracked by their anatomic markers. Externally-
chanics during gait. We hypothesized MLo would display greater knee referenced joint moments were calculated using inverse dynamics.
extensor fatigue in response to a bout of walking, and have corre- Knee kinematic outcomes included flexion angle at heel strike,
spondingly greater changes in knee mechanics and co-activation com- midstance peak and range of motion during stance; peak adduction
pared to both Y and MHi. angle during loading response; and femoral anterior displacement re-
lative to the tibia at heel strike, at the first peak of the vertical ground
reaction force, average over stance, and peak during stance. Knee ki-
2. Methods netics included the first peak extension and adduction moments, and
the peak flexion moment. For descriptive purposes, sagittal hip and
2.1. Participant selection ankle ranges of motion during stance and peak flexion and/or extension
moments were also reported.
Three groups were recruited: highly active mid-life adults (MHi;
55–70 years, running ≥15 miles/wk), less active mid-life adults (MLo; 2.2.3. Knee extensor muscle function testing
55–70 years, ≤3 30-minute moderate exercise bouts/wk), and young Maximal isometric torque (Nm·kg−1) as well as peak concentric and
adults (Y; 21–35 years; recreationally active). Groups included equal eccentric isokinetic knee extensor power (W·kg−1) at 90 and 270°·s−1
male and female numbers. MHi were runners to ensure a vigorously were collected before and after the 30MTW using an isokinetic dy-
active group that could be quantified using accelerometry. Y were re- namometer (HUMAC NORM, CSMi, Stoughton, MA). Concentric and
creationally active (but not regular runners) as this activity profile eccentric power were collected in a single motion. At baseline, two sets
matched that of highly active mid-life adults in preliminary data col- of three repetitions were performed for each test (isometric, con/ec-
lection. All participants completed Knee Osteoarthritis Outcome Score centric at 90°·s−1, con/eccentric at 270°·s−1) with 30 s rest between sets
questionnaires to verify absence of knee symptoms. Scores were similar and 2 min rest between tests. Isometric repetitions included 5 s con-
between groups (Supplementary Table S-1). All participants had tractions followed by 5 s rest. After the 30MTW one set of each test was
BMI < 30 kg·m−2, were free of significant musculoskeletal injury his- collected with 15 s rest between tests. For isometric torque, the knee
tory, cardiovascular or neurological pathology, and chronic pain. Power was flexed 60° relative to full extension and isokinetic power was col-
calculations indicated 12–19 participants per group were needed to lected across 70° of knee motion.
detect meaningful differences (Supplementary Table S-2). Participants
completed IRB-approved informed consent prior to data collection. 2.2.4. 30-minute treadmill walk (30MTW)
After baseline gait and strength testing, participants performed the
30MTW. Treadmill speed was set to the pace of the 400 m walk from
2.2. Study protocol visit one. If this pace was not comfortable, treadmill speed was adjusted
in increments of 0.1 mph until the speed felt “normal.” Treadmill in-
Participants completed two study visits ≥7 days apart. Visit one cline was increased to 3% at minutes 7, 17, and 27 for one minute and
included a timed 400 m walk at self-selected pace to determine tread- then returned to level. This protocol was designed to mimic 30 min of
mill speed for visit two (see 30-minute treadmill walk). Participants also exercise an individual might complete during a typical day, and has
practiced strength testing and were given a PA monitor. Visit two in- been shown to cause knee extensor fatigue in older women [25].
cluded overground walking gait analysis; knee extensor muscle testing;
and a 30-minute treadmill walk (30MTW) with electromyography 2.2.5. Knee muscle co-activation
(EMG) collection. During this visit, participants wore standard neutral Co-activation was calculated using surface EMG collected at
shoes (RC550, New Balance, Boston, MA). 2000 Hz. Electrodes (Trigno, Delsys, Natick, MA) were placed on the

25
J.F. Hafer, et al. Gait & Posture 70 (2019) 24–29

rectus femoris, vastus lateralis, vastus medialis, biceps femoris, and (p = 0.006 for Y vs. MHi and MLo; Fig. 1A). During the second minute
semitendinosus. Ten consecutive strides were extracted from the second of the 30MTW groups differed in co-activation at midstance (p = 0.03),
and final minutes of the 30MTW. Gait events were identified using an where MLo had greater quadriceps:hamstrings co-activation compared
accelerometer on the lower leg. Each signal had the mean offset re- to Y (post-hoc p = 0.04; MLo DCCR=-0.01, Y DCCR=-0.22, Fig. 2A).
moved and was band-pass filtered at 20–500 Hz, rectified, and lowpass Baseline knee mechanics were similar between groups at both
filtered at 20 Hz. Each muscles’ resulting signal was then normalized to walking speeds. The fixed speed results are presented here (Table 2)
its average stance activation over the 10 strides during the second with preferred speed results in supplements (Tables S-3, S-4). MLo had
minute of the 30MTW. greater femoral anterior displacement than Y at the time of the first
Directed co-contraction ratios (DCCRs) were calculated between the peak of the vertical ground reaction force, and greater peak femoral
quadriceps (average of rectus femoris and vasti) and hamstrings anterior displacement. Knee joint angles did not differ between groups
(average of biceps femoris and semitendinosus) [10]. DCCRs were (Fig. 3, Table 2). Y had greater knee extension moments in early stance
calculated at each gait cycle point t for each stride s: compared to both MHi and MLo, and there was a trend towards MHi
If quadriceps activation > hamstrings activation: having larger knee flexion and first peak adduction moments than Y
and MLo (Table 2).
(average of hamstrings linear envelopes )t , s
DCCRt , s = 1
(average of quadriceps linear envelopes )t , s
3.2. Response to 30MTW
else,
(average of quadriceps linear envelopes )t , s The 30MTW elicited knee extensor fatigue (decrease in torque or
DCCRt , s = 1 power) across most contraction velocities in MLo and Y but not in MHi
(average of hamstrings linear envelopes )t , s
(Fig. 1B). MLo fatigued more than MHi in concentric contractions at
DCCRs range between 1 and -1 where 1 indicates exclusive quad- 270°·s−1 (post-hoc p = 0.008, Fig. 1B). During terminal swing, MHi had
riceps and -1 exclusive hamstrings activation. Values near 0 indicate a decrease vs. MLo’s increase in quadriceps:hamstrings co-activation,
relatively equal activation of the two muscle groups. DCCRs were (post-hoc p = 0.05, Fig. 2B).
averaged across the 10 strides from the second and final minute of the Changes in knee kinematics after the 30MTW were small (< 1.5°
30MTW, and then over specific phases of the gait cycle: terminal swing or < 1 mm) and not different between groups (Table 3). Knee flexion
(final 15% of swing); and early, mid, and late thirds of stance. moments changed differently between groups (Table 3), with MHi
displaying a small decrease and Y a small increase in response to the
2.2.6. Statistics 30MTW (post-hoc p = 0.03).
Data were normally distributed, thus primary outcome variables
were compared between groups using one-way MANOVAs with sig- 4. Discussion
nificance set at p ≤ 0.05. For the primary aim, outcome variables were
compared between groups at baseline (baseline overground gait and The overall study aim was to determine the impact of age and PA on
muscle strength, co-activation from 30MTW minute 2). For the sec- measures that may affect the loading environment in the knee, and thus
ondary aim, changes in outcome variables pre- to post-30MTW were knee OA risk, during gait. We hypothesized that MLo compared to both
compared between groups (post-pre overground gait and muscle MHi and Y would have greater knee flexion angles and femoral anterior
strength, minute 30-minute 2 co-activation data). Where significant displacement during stance, altered knee moments, lower knee extensor
main effects were found, Tukey’s post-hoc tests were performed. muscle torque and power, and greater muscle co-activation. In partial
Baseline and post-30MTW knee extensor torque and power were com- support of this hypothesis, MLo were weaker than Y in concentric
pared using paired t-tests to test for knee extensor muscle fatigue. contractions, and had greater femoral anterior displacement and mid-
stance co-activation patterns compared to MHi and Y. Further, we hy-
3. Results pothesized that these group differences would be amplified by a
30MTW. This hypothesis was largely not supported, but MLo showed
Group characteristics are shown in Table 1. Due to technical issues, greater decreases in high-velocity knee extensor power than MHi. Our
muscle co-activation data from the second minute of the 30MTW in- results suggest that high levels of PA in mid-life may mitigate me-
clude 58 participants (n = 18 MLo) and for the final minute include 54 chanical risk factors for knee OA, particularly concentric knee extensor
participants (n = 18, 19, and 17 Y, MHi, and MLo, respectively). MLo muscle power and femoral anterior displacement.
had fewer MVPA minutes compared to Y and MHi, and all groups dif- Similar to previous research [13], baseline knee extensor power and
fered in PA counts (Table 1). torque differed by age during concentric contractions. As expected, only
MLo were weaker than Y at the moderate contraction velocity of
3.1. Baseline comparison 90°·s−1, while both mid-life groups were weaker than Y at 270°·s−1. The
lack of a group difference in eccentric power aligns with studies com-
At baseline, groups differed in knee extensor power at 90°·s−1 paring young and older (> 65 years) adults that have found smaller
(p = 0.01) and 270°·s-1 (p = 0.002). Knee extensor power was lower in [26] and later [27] declines in eccentric relative to concentric knee
MLo compared to Y during concentric contractions at 90°·s-1 (post-hoc extensor power with age. MLo’s activity level of nearly 150 min per
p = 0.01) and in both MHi and MLo compared to Y at 270°·s−1 week of MVPA appeared to not be sufficient to preserve low-velocity

Table 1
Group characteristics reported as Mean (SD). MVPA: moderate to vigorous physical activity. Y: young group. MHi: highly active mid-life group. MLo: less active mid-
life group. * different from Y; + different from MHi.
Group n (#male) Age (years) Height (m) Mass (kg) Preferred walking speed Treadmill walking speed Weekly MVPA Weekly counts
(m·s−1) (m·s−1) minutes (x10−3)

Y 20 (10) 27.8 (3.5) 1.72 (0.09) 69.8 (11.8) 1.40 (0.15) 1.39 (0.14) 393.5 (162.0) 2509 (783)
MHi 20 (10) 61.9 (4.0) 1.68 (0.11) 64.4 (12.9) 1.35 (0.12) 1.35 (0.12) 473.5 (216.5) 3340 (1152) *
MLo 20 (10) 62.9 (3.9) 1.71 (0.11) 69.9 (11.7) 1.35 (0.12) 1.35 (0.12) 147.7 (110.1) *+ 1504 (633) *+

26
J.F. Hafer, et al. Gait & Posture 70 (2019) 24–29

Fig. 1. Knee extensor torque and power at baseline (A) and

change after the 30 min treadmill walk (B). Mean ± SE. Y:
young group. MHi: highly active older group. MLo: less active
older group. A: ^ indicates MLo different from Y; * indicates
MHi and MLo different from Y. B: # indicates significant de-
crease from baseline. + indicates greater change in MHi than
MLo.

risk of knee OA than MHi in terms of knee extensor power.

At baseline, only MLo had greater quadriceps:hamstrings co-acti-
vation at midstance compared to Y. This could indicate that the quad-
riceps remain activated longer after early stance in MLo relative to Y
and MHi. This finding is similar to the increased co-activation pre-
viously reported in older compared to young adults [28] and may
suggest an earlier onset of age-related gait changes in MLo. While ex-
ternal joint moments and muscle activation do not directly assess loads
in the knee, the combination of trends toward age-related increases in
peak knee moments and greater co-activation would be consistent with
higher contact forces in the knee in MLo [10,18]. MLo also had greater
femoral anterior displacement compared to Y. Femoral anterior dis-
placement appears to increase from young asymptomatic adults to mid-
life and older asymptomatic adults and adults with symptomatic knee
OA [9]. A combination of greater femoral anterior displacement with
greater co-activation and knee moments could transfer higher loads to
unconditioned cartilage [11], thereby increasing risk for knee OA in-
itiation. These baseline results suggest that less active mid-life adults
may be progressing along a trajectory towards knee OA initiation, and
that, in agreement with recent literature [21], high levels of PA may
protect against OA via an impact on biomechanical risk factors for knee
OA.
The 30MTW induced knee extensor fatigue in MLo and, somewhat
surprisingly, in Y, with variation in the magnitude of this effect across
contraction velocities. The lack of concurrent changes in co-activation
and gait mechanics suggests a relative insensitivity of gait mechanics to
moderate fatigue in healthy young and middle aged adults. Based on
limited literature, older mobility-intact adults may use only ≈25% of
Fig. 2. DCCR for quadriceps vs. hamstrings at baseline (A) and change after the
their knee extensor power during gait [29]. With a conservative as-
30 min treadmill walk (B). Mean ± SE. Positive values indicate greater quad-
sumption that MLo had the same relative effort as older adults in the
riceps activation relative to hamstring activation. Values near 0 indicate rela-
tively equal activation between the two muscle groups. A: * indicates MLo literature, their ≈20% decrease in power would still leave a ≈55%
different from Y. B: + indicates MHi different from MLo. reserve of knee extensor strength for maintaining gait.
The current study has some limitations. While MLo were less active
than MHi and Y, their MVPA of ≈148 min·wk−1 approached minimum
concentric muscle strength. As decreased knee extensor strength at
exercise guidelines and exceeded reported averages for adults age 50+
moderate contraction velocities (60-120°·s−1) has been associated with
[30]. A less active group may have resulted in larger differences both at
knee OA incidence [5], our results suggest that MLo may be at greater

27
J.F. Hafer, et al. Gait & Posture 70 (2019) 24–29

Table 2
Baseline knee kinematics and kinetics. KF: knee flexion; KA: knee adduction; FAD: femoral anterior displacement; KE: knee extension; HF: hip flexion; HE: hip
extension; ADF: ankle dorsiflexion; Y: young group; MHi: mid-life highly active group; MLo: mid-life less active group. Where significant main effects were found,
data are bolded and post-hoc p-values are reported. * indicates Y different from MLo, ^ indicates Y different from MHi.
Y MHi MLo p-value post-hoc

Mean SD Mean SD Mean SD

Stride length (m) 1.49 0.06 1.45 0.10 1.47 0.08 0.22 na
KF Heel strike (°) 5.3 4.6 4.9 4.5 6.1 5.6 0.77 na
KF Midstance (°) 20.1 5.2 22.8 6.0 22.8 5.2 0.20 na
KF Stance ROM (°) 38.9 3.5 38.6 4.1 40.4 3.8 0.28 na
KA Midstance (°) 2.0 2.7 1.3 3.4 1.2 3.4 0.74 na
FAD Heel strike (mm) −9.0 7.1 −6.0 7.5 −5.2 7.0 0.23 na
FAD at first VGRF peak (mm) 0.2 5.6 3.9 4.9 6.0 5.8 < 0.01 * < 0.01
FAD Stance average (mm) 2.7 4.0 5.8 4.5 6.1 5.3 0.04 na
FAD Max stance (mm) 4.4 6.8 8.4 7.2 11.3 7.8 0.02 *0.01
KE Moment (%BW·Ht) −1.9 0.4 −1.5 0.6 −1.5 0.5 0.02 *0.03, ^0.04
KF Moment (%BW·Ht) 2.6 1.3 3.5 1.4 2.8 1.0 0.08 na
KA Moment (%BW·Ht) −2.9 0.7 −3.4 0.6 −3.1 0.8 0.07 na
Hip ROM (°) 42.9 5.1 44.5 4.2 45.4 5.5
Ankle ROM (°) 26.1 4.0 26.1 3.9 28.3 4.7
HF Moment (%BW·Ht) −3.8 0.7 −4.3 0.9 −4.0 1.3
HE Moment (%BW·Ht) 4.5 1.0 4.5 1.1 4.0 1.3
ADF Moment (%BW·Ht) −9.3 0.7 −9.1 0.7 −8.9 1.3

Fig. 3. Group mean knee kinematic and

kinetic data through the gait cycle. Y:
young group; MHi: mid-life highly ac-
tive group; MLo: mid-life less active
group. Lettered arrows indicate discrete
variables of interest, reported in
Table 2: a: knee flexion at heel strike; b:
knee flexion at midstance; c: femoral
anterior displacement (FAD) at heel
strike; d: FAD at first vertical ground
reaction force peak; e: FAD maximum
during stance; f: first peak knee exten-
sion moment; g: peak knee flexion mo-
ment; h: first peak knee adduction mo-
ment. P-values noted where p ≤ 0.05
for main comparisons of group.

baseline and in response to the 30MTW. However, these mid-life groups relative to the tibia, midstance muscle co-activation, and knee exten-
had similar weekly minutes in light to moderate PA (2252 ± 566 for sion moment in early stance) differ by age and habitual PA. Our results
MHi and 1893 ± 555 for MLo), suggesting that differences observed suggest high PA mitigates some age-related biomechanical risk factors
are a result of the vigorous activity performed by MHi. Our ability to for knee OA in healthy mid-life adults. The well-controlled cohorts in
test that MLo would display greater fatigue and larger changes in gait this study allowed for discrimination of factors that could alter the
mechanics than Y and MHi may have been limited by the moderate loading environment for knee joint cartilage based on age or decreased
fatigue we observed in MLo. Finally, as with many studies of asymp- PA alone, independent of the many comorbidities that additionally alter
tomatic adults, we did not have radiographs and cannot exclude the cartilage health. Our results support a role of PA (independent of its
possibility of asymptomatic knee OA. tendency to reduce body weight and associated health problems) in
Readers should be aware of the limitations of methodologies for wellness interventions or rehabilitation programs designed to reduce
calculating joint kinematics and the implications of a methodology risk factors of knee OA.
relative to the hypotheses being tested. We selected PCT to measure
tibiofemoral motion as our study questions necessitated that partici-
Conflict of interest statement
pants walk naturally at their preferred gait speed overground. In a PCT
validation study using a tibial Ilizarov frame [23], average bone loca-
The authors declare no conflicts of interest.
tion error was 0.08 mm. If similar error were assumed for femur loca-
tion, maximum error expected in tibia vs. femur location would be
0.16 mm, well below the group differences reported in our study. Acknowledgements
We have shown that variables that may reflect the loading en-
vironment in the knee joint (i.e., anterior displacement of the femur The authors thank Marquis Hawkins, PhD for statistical guidance.
This project was supported by an American College of Sports Medicine

28
J.F. Hafer, et al. Gait & Posture 70 (2019) 24–29

Table 3
Changes in knee kinematics and kinetics in response to the 30MTW. KF: knee flexion; KA: knee adduction; FAD: femoral anterior displacement; KE: knee extension;
HF: hip flexion; HE: hip extension; ADF: ankle dorsiflexion; Y: young group; MHi: mid-life highly active group; MLo: mid-life less active group. Where significant
main effects were found, data are bolded and post-hoc p-values are reported. * indicates Y different from MLo, ^ indicates Y different from MHi.
Y MHi MLo p-value post-hoc

Mean SD Mean SD Mean SD

Stride length (m) 0.00 0.04 0.00 0.02 0.01 0.04 0.34 na
KF Heel strike (°) 0.2 2.5 1.1 1.9 1.1 1.7 0.31 na
KF Midstance (°) 0.0 1.9 0.0 2.0 0.5 1.7 0.61 na
KF Stance ROM (°) −0.4 1.5 −1.4 2.7 −1.1 1.8 0.29 na
KA Midstance (°) −0.3 1.0 −0.4 1.3 −0.5 1.3 0.90 na
FAD Heel strike (mm) −0.8 3.8 0.4 4.0 −0.3 5.0 0.69 na
FAD at first VGRF peak (mm) −0.6 3.7 0.0 4.2 −0.3 4.7 0.88 na
FAD Stance average (mm) −0.7 3.7 −0.4 3.8 0.1 3.9 0.83 na
FAD Max stance (mm) −0.1 3.9 0.2 4.4 0.3 3.8 0.64 na
KE Moment (%BW·Ht) −0.2 0.3 0.0 0.4 −0.1 0.2 0.39 na
KF Moment (%BW·Ht) 0.2 0.4 −0.1 0.3 0.0 0.5 0.03 ^0.03
KA Moment (%BW·Ht) −0.1 0.2 −0.1 0.3 −0.1 0.3 0.64 na
Hip ROM (°) 1.0 1.9 0.7 1.2 0.9 1.6
Ankle ROM (°) 0.5 1.7 0.1 1.7 0.3 1.7
HF Moment (%BW·Ht) −0.1 0.5 0.0 0.6 −0.1 0.5
HE Moment (%BW·Ht) 0.1 0.4 −0.1 0.6 −0.2 0.2
ADF Moment (%BW·Ht) 0.0 0.3 0.2 0.3 0.2 0.4

Foundation Doctoral Research Grant (JFH). D. Byrd, Reduced quadriceps strength relative to body weight: a risk factor for knee
osteoarthritis in women? Arthritis Rheum. 4111 (1998) 1951–1959.
[15] M. Lewek, K. Rudolph, M. Axe, L. Snyder-Mackler, The effect of insufficient
Appendix A. Supplementary data quadriceps strength on gait after anterior cruciate ligament reconstruction, Clin.
Biomech. Bristol Avon (Bristol, Avon) 171 (2002) 56–63.
Supplementary data associated with this article can be found, in the [16] A.M. Murray, A.C. Thomas, C.W. Armstrong, B.G. Pietrosimone, M.A. Tevald, The
associations between quadriceps muscle strength, power, and knee joint mechanics
online version, at https://doi.org/10.1016/j.gaitpost.2019.02.008. in knee osteoarthritis: a cross-sectional study, Clin. Biomech. (Bristol, Avon) 3010
(2015) 1140–1145.
References [17] D. Rutherford, M. Baker, I. Wong, W. Stanish, The effect of age and knee osteoar-
thritis on muscle activation patterns and knee joint biomechanics during dual belt
treadmill gait, J. Electromyogr. Kinesiol. 34 (2017) 58–64.
[1] L. Murphy, T.A. Schwartz, C.G. Helmick, J.B. Renner, G. Tudor, G. Koch, [18] L.C. Tsai, S. McLean, P.M. Colletti, C.M. Powers, Greater muscle co-contraction
A. Dragomir, W.D. Kalsbeek, G. Luta, J.M. Jordan, Lifetime risk of symptomatic results in increased tibiofemoral compressive forces in females who have undergone
knee osteoarthritis, Arthritis Rheum. 599 (2008) 1207–1213. anterior cruciate ligament reconstruction, J. Orthop. Res. 3012 (2012) 2007–2014.
[2] I.J. Wallace, S. Worthington, D.T. Felson, R.D. Jurmain, K.T. Wren, H. Maijanen, [19] K.M. Tarpenning, M. Hamilton-Wessler, R.A. Wiswell, S.A. Hawkins, Endurance
R.J. Woods, D.E. Lieberman, Knee osteoarthritis has doubled in prevalence since the training delays age of decline in leg strength and muscle morphology, Med. Sci.
mid-20th century, Proc. Natl. Acad. Sci. U. S. A. 11435 (2017) 9332–9336. Sports Exerc. 361 (2004) 74–78.
[3] D.D. Dunlop, J. Song, P.A. Semanik, L. Sharma, J.M. Bathon, C.B. Eaton, [20] J. Qin, K.E. Barbour, M.C. Nevitt, C.G. Helmick, J.M. Hootman, L.B. Murphy,
M.C. Hochberg, R.D. Jackson, C.K. Kwoh, W.J. Mysiw, M.C. Nevitt, R.W. Chang, J.A. Cauley, D.D. Dunlop, Objectively measured physical activity and risk of knee
Relation of physical activity time to incident disability in community dwelling osteoarthritis, Med. Sci. Sports Exerc. 502 (2018) 277–283.
adults with or at risk of knee arthritis: prospective cohort study, BMJ 348 (2014) [21] A. Bricca, W. Wirth, C.B. Juhl, J. Kemnitz, D.J. Hunter, C.K. Kwoh, F. Eckstein,
g2472. A.G. Culvenor, Moderate physical activity may prevent cartilage loss in women with
[4] K.A. Boyer, T.P. Andriacchi, G.S. Beaupre, The role of physical activity in changes in knee osteoarthritis: data from the Osteoarthritis Initiative, Arthritis Care Res.
walking mechanics with age, Gait Posture 361 (2012) 149–153. (Hoboken) (2018).
[5] N.A. Segal, J.C. Torner, D. Felson, J. Niu, L. Sharma, C.E. Lewis, M. Nevitt, Effect of [22] P.S. Freedson, E. Melanson, J. Sirard, Calibration of the computer science and ap-
thigh strength on incident radiographic and symptomatic knee osteoarthritis in a plications, Inc. Accelerometer. Med. Sci. Sports Exerc. 305 (1998) 777–781.
longitudinal cohort, Arthritis Rheum. 619 (2009) 1210–1217. [23] E.J. Alexander, T.P. Andriacchi, Correcting for deformation in skin-based marker
[6] S.A. Oliveria, D.T. Felson, J.I. Reed, P.A. Cirillo, A.M. Walker, Incidence of symp- systems, J. Biomech. 343 (2001) 355–361.
tomatic hand, hip, and knee osteoarthritis among patients in a health maintenance [24] M. Kozanek, A. Hosseini, F. Liu, S.K. Van de Velde, T.J. Gill, H.E. Rubash, G. Li,
organization, Arthritis Rheum. 388 (1995) 1134–1141. Tibiofemoral kinematics and condylar motion during the stance phase of gait, J.
[7] K.A. Boyer, R.T. Johnson, J.J. Banks, C. Jewell, J.F. Hafer, Systematic review and Biomech. 4212 (2009) 1877–1884.
meta-analysis of gait mechanics in young and older adults, Exp. Gerontol. 95 (2017) [25] S.A. Foulis, S.L. Jones, R.E. van Emmerik, J.A. Kent, Post-fatigue recovery of power,
63–70. postural control and physical function in older women, PLoS One 129 (2017)
[8] E.F. Chehab, T.P. Andriacchi, J. Favre, Speed, age, sex, and body mass index pro- e0183483.
vide a rigorous basis for comparing the kinematic and kinetic profiles of the lower [26] T. Hortobagyi, D. Zheng, M. Weidner, N.J. Lambert, S. Westbrook, J.A. Houmard,
extremity during walking, J. Biomech. 58 (2017) 11–20. The influence of aging on muscle strength and muscle fiber characteristics with
[9] J. Favre, J.C. Erhart-Hledik, T.P. Andriacchi, Age-related differences in sagittal- special reference to eccentric strength, J. Gerontol. A Biol. Sci. Med. Sci. 506 (1995)
plane knee function at heel-strike of walking are increased in osteoarthritic patients, B399–406.
Osteoarthr. Cartil. 223 (2014) 464–471. [27] R.S. Lindle, E.J. Metter, N.A. Lynch, J.L. Fleg, J.L. Fozard, J. Tobin, T.A. Roy,
[10] T.L. Heiden, D.G. Lloyd, T.R. Ackland, Knee joint kinematics, kinetics and muscle B.F. Hurley, Age and gender comparisons of muscle strength in 654 women and
co-contraction in knee osteoarthritis patient gait, Clin. Biomech. (Bristol, Avon) men aged 20-93 yr, J. Appl. Physiol. 1997 (835) (1985) 1581–1587.
2410 (2009) 833–841. [28] T. Hortobagyi, S. Solnik, A. Gruber, P. Rider, K. Steinweg, J. Helseth, P. DeVita,
[11] T.P. Andriacchi, S. Koo, S.F. Scanlan, Gait mechanics influence healthy cartilage Interaction between age and gait velocity in the amplitude and timing of antagonist
morphology and osteoarthritis of the knee, J. Bone Joint Surg. Am. 91 (Suppl 1) muscle coactivation, Gait Posture 294 (2009) 558–564.
(2009) 95–101. [29] C.M. Beijersbergen, U. Granacher, A.A. Vandervoort, P. DeVita, T. Hortobagyi, The
[12] M.P. Murray, G.M. Gardner, L.A. Mollinger, S.B. Sepic, Strength of isometric and biomechanical mechanism of how strength and power training improves walking
isokinetic contractions: knee muscles of men aged 20 to 86, Phys. Ther. 604 (1980) speed in old adults remains unknown, Ageing Res. Rev. 122 (2013) 618–627.
412–419. [30] R.P. Troiano, D. Berrigan, K.W. Dodd, L.C. Masse, T. Tilert, M. McDowell, Physical
[13] J.K. Petrella, J.S. Kim, S.C. Tuggle, S.R. Hall, M.M. Bamman, Age differences in activity in the United States measured by accelerometer, Med. Sci. Sports Exerc. 401
knee extension power, contractile velocity, and fatigability, J. Appl. Physiol. 2005 (2008) 181–188.
(981) (1985) 211–220.
[14] C. Slemenda, D.K. Heilman, K.D. Brandt, B.P. Katz, S.A. Mazzuca, E.M. Braunstein,

29