Abstract
BACKGROUND:
Deep knee bending has been reported as an occupational hazard to workers who have to adopt such postures. High knee joint moments have been associated with knee osteoarthritis initiation and progression.
OBJECTIVE:
This study aimed to compare four high knee flexion postures (dorsiflexed and plantarflexed kneeling, and flat-foot and heels-up squatting) to determine which one results in lower knee joint flexion and ab/adduction moments.
METHODS:
Forty-three participants performed five trials of each posture. Peak (for descent/ascent) and mean (for the static hold) external knee flexion and ab/adduction moments were analyzed for each posture using 2-way ANOVAs and post-hoc pairwise comparisons.
RESULTS:
It was observed that the flat-foot squat resulted in significantly lower knee flexion moment compared to the other three postures (4.63±0.99 % BW·H during the static phase, and 5.83±1.24 % BW·H and 5.94±1.24 % BW·H during descent and ascent phases, respectively). During ascent phase, significant differences was indicated in peak adduction moments for the flat-foot squat in comparison to both styles of kneeling.
CONCLUSIONS:
When high knee flexion is required but posture is not dictated, flat-foot squat will reduce exposures to high knee moments.
Introduction
High knee flexion postures such as kneeling and squatting (knee flexion angle exceeding ∼120°) are commonly used in industries such as construction, mining or childcare. Epidemiological studies report that the risk of developing knee osteoarthritis (OA) is elevated (odds ratio 1.9–2.46, 95% confidence interval) in work conditions that require kneeling or squatting postures [1–3]. First, the altered joint congruency as the knee moves through extreme flexion angles can result in the loading of atypical structures within the knee that may not be conditioned to withstand this load [4]. Second, high moments at the knee during such postures can contribute to the high risk of cartilage degradation [5, 6]. A greater knee adduction moment has been found to contribute to both the initiation and progression of knee OA [7–12]. In a study of participants with medial knee OA, Erhart-Hledik et al. [11] reported that the peak knee flexion moment was negatively associated with posterior tibial cartilage thickness for those with less severe OA. They also found that the peak knee adduction moment was negatively associated with cartilage thickness on the medial side in those with more severe OA. It is expected that high abduction moments can cause a more valgus alignment, which along with the flat/concave shape of the tibial plateau on the lateral side, can increase the stress on the lateral meniscus during high flexion postures. Radiographic, MRI, and biomechanical studies prove the strong association between valgus alignment and lateral knee OA development and progression [13–17]. According to Teichtahl et al. [15], the risk of lateral joint space narrowing can rise by 62% due to valgus alignment. Therefore, it is expected that the higher flexion and ab/adduction moments during deep knee bending could represent a higher risk of cartilage degradation.
Previous studies have investigated knee motion and the tibiofemoral joint forces and moments for specific high flexion postures in exercise and activities of daily living [6, 18–28]. List et al. [22] and Gullett et al. [25] determined the flexion moments in the knee joint for the descent and ascent phases of barbell squatting. They described an increase in the difference between the two phases with an external load, and a greater knee extensor moment when the load is on the back than in the front [22, 25]. Hemmerich et al. [19] investigated only the kinematics of the knee joints during squat, kneel, and sitting cross-legged postures. They reported more than 140° of flexion and up to 10° of abduction for such high flexion activities [19]. Higher knee flexion moments have been associated with higher flexion angles during squatting and kneeling [20, 26]. Also, measurement of knee joint moments and forces using instrumented knee implants shows an increase in knee abduction moment with increasing knee flexion angle [27, 28]. Nagura et al. [20] studied the kinematics and net flexion moment at the knee joint during double-leg and single-leg kneeling. They showed that the maximum flexion moment occurred at high flexion angles, and that the double-leg rise (ascent) and descent resulted in the highest flexion moment at the knee [20]. Pollard et al. [6] quantified the net moments and forces at the knee joint when the participants were in squatting and one-knee kneeling postures. They reported that the highest flexion moment occurred in the one-knee kneel and the highest adduction moment occurred in the heels-up squat [6]. However, that study did not investigate variations in squatting and kneeling postures (e.g. heels-up squat (HS), flat-foot squat (FS), dorsiflexed kneel (DK), or plantarflexed kneel (PK)). Also, limited quantitative data is available on the joint moments during the transition (descent/ascent) phases of variations of high flexion activities.
Due to distinctive anatomic characteristics, women might exhibit different motion patterns and consequently different kinetics from men while performing a similar activity. Women reportedly demonstrate a more valgus position than men during the single-leg squat due to a wider pelvis, which changes the angle of the femur relative to the tibia [29]. Zeller et al. [29] reported larger knee flexion angles in women performing the single-leg squat; however, the difference was not statistically significant. There is evidence that knee joint laxity is higher for women in comparison to men [30]. This increased laxity can result in a higher range of joint motion, and consequently higher flexion and abduction moments at the knee joint [20, 26–28].
This study aimed to compare peak (during descent and ascent phases) and mean (during the static hold of deep flexion) moments at the knee joint across four high flexion postures (HS, FS, DK, and PK). The objective was to determine which posture(s) minimized the external flexion and ab/adduction moments. Internal-external rotation moments were not considered within the scope of this study due to the greater errors in the transverse plane data caused by the soft tissue artifact [31, 32].
It was hypothesized that that PK would experience the highest knee joint flexion and abduction moments in the static phase because a higher peak flexion angle would be reached during the PK compared to the other three postures [21, 33]. The logic that higher moments would be associated with higher flexion angles was based on previous studies [20, 26–28]. Also in the static phase, female subjects were expected to experience higher flexion and abduction moments since they were expected to exhibit larger flexion angles [29, 30]. During the dynamic transitional phases, it was hypothesized that both kneeling styles would result in higher peak flexion and ab/adduction moments than the squatting styles because, when kneeling is performed as an asymmetric task, one knee makes contact with the ground before the other. Thus, kneeling requires full, or near full, body weight transfer to a single leg during the task.
Methods
Forty-three participants were recruited (ages19–32): 20 males (height = 1.7±0.63 m, mass = 72.6±12.7 kg) and 23 females (height = 1.6±0.69 m, mass = 60.0±6.9 kg). Physical activity levels were not officially surveyed, however, most participants indicated that they were recreationally active. All participants confirmed that they had no previous knee injuries. Participants gave informed consent to participate in this study, which was approved by the University of Waterloo Ethics Board.
Participants completed five trials of the four postures (Fig. 1). These 20 trials were fully randomized by a random number generator and were completed in a single session for each participant. A trial consisted of stepping forward onto two force plates, descending to the required squat or kneel position, an 8-second static hold in the fully flexed position, and ascending to standing. Kneeling transitions to or from standing were done in an asymmetric style, where the participant moved through a deep lunge position and one knee made (or ended) contact with the floor before the other. The supporting (front) leg during the lunge was the lead leg; the leg with the knee that made ground contact first was the trail leg. Participants were permitted to choose their left or right leg as the lead leg, but were required to maintain the same lead leg for all trials. Participants were given specific consistent instructions for each style of squatting and kneeling, and were required to practice each of the four postures at least twice until they confirmed that they were comfortable with the posture.

Four high flexion poses used in this experiment: Flat-foot squatting (top left), heels-up squatting (top right), dorsiflexed kneeling (bottom left), and plantarflexed kneeling (bottom right).
Ground reaction forces (GRF) were collected using two (for squatting) or four (for kneeling once the knees made contact with the ground) force plates (OR6-7; Advanced Mechanical Technology, Inc., Watertown, MA, USA). Kinematic data were collected using an 18-camera Optotrak motion capture system (Northern Digital Inc., Waterloo, ON, Canada) and rigid bodies on the feet, shanks, thighs, and the skin over the sacrum. Kinematic and kinetic data were processed using Matlab (Mathworks, Inc., Natick, MA) and Visual 3D (C-Motion Inc., Germantown, MD). Kinematic and kinetic raw data was filtered with a dual pass 2nd order Butterworth filter with a 6 Hz cut off frequency [34]. Local coordinate systems were created as described in Table 1. The external knee joint moments were expressed in the knee joint coordinate system (JCS), in which the femoral medial-lateral axis was the “flexion” axis, the tibial distal-proximal axis was the internal rotation axis, and ab/adduction was interpreted about the mutually perpendicular “floating” axis [35]. The JCS is considered a standard convention by the International Society of Biomechanics for interpretation of the kinematics of the lower limb joints [36]. Because the rationale for the hypotheses was based on established relationships between knee joint moments and angles, the JCS was used to express both the kinematics and kinetics in this study [37, 38].
Local coordinate system definitions for the foot, tibia and thigh segments
The descent phase was defined to begin when the participant made the first contact with a force plate (exceeding a 10 N threshold). The participant was required to be in the static pose 5 seconds later. The static pose was held for 8 seconds. The ascent was unconstrained with respect to time. However, the end of the ascent phase was defined as the last instant of contact on the force plate (no longer exerting a minimum of a 10 N force).
External flexion and ab/adduction moments were expressed in units of percent body weight times height (% BW·H). During transitions, the peaks of these normalized moments were determined from each trial. The peak moments from individual trials were averaged to calculate a set of mean peak moments for each posture per participant during both the ascent and descent phases. During the static phase, the means of these moments were calculated for each static phase and then averaged across trials for each participant. Participant means were then averaged to determine grand means for each posture during each phase. When a participant’s frontal plane moment curve demonstrated no abduction or no adduction moments, that participant was excluded when calculating the grand mean abduction or adduction moment, respectively.
The lead leg was analyzed during the kneeling postures because the lead leg experiences greater moments of force in the sagittal and frontal planes than the trail leg [39]. It was assumed that the two squatting postures were performed symmetrically. Therefore, only the right leg was analyzed. Statistics were run in Matlab (Mathworks, Inc., Natick, MA) using two-way repeated measures (pose (4)×gender (2)) ANOVAs on the three phases for the peak (in descent/ascent phases) and mean (in static phase) flexion and ab/adduction moments (a total of 9 ANOVAs). In cases where a significant pose main effect was detected, a post-hoc Tukey test was used to determine which pairs of postures were significantly different. The post-hoc comparisons of the two styles within a given task (kneel or squat) were of particular interest to see if the style of kneeling or squatting had a significant effect on the knee joint moments. A critical value of 0.05 was used to determine statistical significance. Since the rationale for the hypotheses involves the knee flexion angle in the static phase, a two-way, repeated measures (pose (4)×gender (2)) ANOVA was run on the mean of the flexion angle for the static phase only.
The peak (for descent and ascent phases) and mean (for static phase) flexion and ab/adduction moments across the four postures are summarized in Table 2. A comparison between FS (with lowest average peak flexion angle) and PK (with highest average peak flexion angle) showed a 39.7% lower flexion moment and a 14% lower abduction moment during the static hold of the flat-foot squat. FS was also the posture with the lowest peak flexion and ab/adduction moments in ascent phase (Table 2). During descent phase, FS resulted in the smallest peak flexion and abduction moments, while HS was the posture with the smallest peak adduction moment.
Comparison of the flexion and ab/adduction moments (% BW·H) across four high flexion postures (Means with different letter superscripts were significantly different)
Comparison of the flexion and ab/adduction moments (% BW·H) across four high flexion postures (Means with different letter superscripts were significantly different)
*indicates a pose main effect (p < 0.05). ▴indicates a gender main effect (p < 0.05). 135 participants. 240 participants. 339 participants. 45 participants. 58 participants. 63 participants. 74 participants.
There were no significant pose-gender interactions for any of the knee moments or the flexion angles in any phase. A pose main effect was found for the flexion moment during all phases (p < 0.001). Post-hoc analysis showed a significant difference between FS and all other postures. In all cases, even when there were style differences within kneeling or within squatting, the post-hoc analysis indicated no statistically significant difference in flexion moments between DK and HS. No pose main effect was found on the ab/adduction moments during the static or descent phases. However, during ascent, a pose main effect was found for the peak adduction moments (p = 0.002). Post-hoc analyses showed a significant difference for FS in comparison to both styles of kneeling. There was a main effect of gender on the mean abduction moment during the static phase and peak abduction moment during the transition phases (p < 0.005), as well as the peak flexion moment in the ascent phase(p = 0.05).
There was a pose main effect on the mean knee flexion angle (Table 3) during the static phases (p < 0.001). Post-hoc analyses indicated that FS resulted in significantly lower peak flexion angle (138.11°±15.2°) in comparison to all other postures. The peak knee abduction moment during transition phases occurred at angles within the range of the deepest flexion angles for each posture (average of 135.84°±24.58° for the descent phase and 136.8°±24° for the ascent phase). However, the peak adduction moment during transitions was reached at an average of 46.67°±38.91° flexion for the descent phase and 46.78°±37.57° for ascent phase (Table 3).
Comparison of the mean flexion angle (degrees) of the knee joint during the static phase and the mean flexion angles corresponding to peak moments in the descent/ascent phases (Means with different letter superscripts were significantly different based on post-hoc analyses of a pose main effect.)
This study investigated the effect of posture and gender on flexion and ab/adduction moments at the knee joint by comparing four high flexion activities: HS, FS, PK, and DK. It was expected that PK would experience the highest knee joint flexion and abduction moments in the static phase. This hypothesis is accepted only for flexion moments. PK produced peak flexion moments significantly greater than all other postures. This hypothesis was based on the logic that PK would have the highest static flexion angles. That postulation was proved incorrect; the mean static phase flexion angles during PK were not significantly different from those during DK or HS. In the static phase, female subjects were expected to experience higher flexion and abduction moments. This hypothesis is confirmed only for the abduction moments. This hypothesis was based on the logic that females would exhibit higher flexion angles in the static phase than males. This logic is rejected; there was no main effect of gender and no pose-gender interaction on the static flexion angles. During the dynamic transitional phases, it was expected that both kneeling styles would result in higher peak moments than the squatting styles. This hypothesis is rejected. In all cases, DK peak moments were not significantly different than the moments during HS.
During PK, a flexion angle more than 150° was experienced, while the subject was in a sitting back posture (in contrast to the more upright position of the thigh and torso in squatting or DK) with a greater proportion of the body mass of the participants distributed towards their feet instead of their knees. Such an increase in the ground reaction force at the feet during PK resulted in a larger ankle moment on the shank. This increased ankle moment would drive the overall increase in flexion moment observed at the knee during static PK.
A goal of this study was to compare posture styles within posture types (e.g. FS vs. HS or PK vs. DK) to see if one style should be chosen over another when the task dictates a particular posture type. Significant differences were found between styles of squatting; the FS resulted in lower peak and mean flexion moments in the descent/ascent and static phases respectively compared to the HS. The peak of the flexion moment during the descent and ascent phases both corresponded on average with the greatest knee flexion angle attained (Table 3). The FS had a lower flexion moment potentially because the center of pressure (COP) at the foot was more posterior (nearer the center of the foot) than in HS, where the COP was under the metatarsals. The posterior translation of the COP from HS to FS would result in a shorter moment arm from the COP to the ankle, and thus a lower flexion moment at the ankle (assuming the vertical GRF acting at the COP in both postures is equal). This lower ankle moment, used in the calculation of the knee moment, would result in a lower knee flexion moment during FS. The effect of gravity on the tibial segment could also account for a portion of the difference between FS and HS. During the FS, the shank was oriented relatively vertically in comparison to HS, where the shank was oriented more horizontally (Fig. 1). The more horizontal orientation would increase the moment arm length from the center of mass of the tibia (where the force of gravity was applied in the model) to the proximal end of the tibia, where the knee joint moment was calculated. This increased moment arm would produce a larger moment about the knee joint created by the gravitational force, resulting in a greater knee flexion moment. In HS, the anterior positioning of the COP, longer moment arm from the COP to the ankle, and the more horizontally oriented shank make the loading condition in the sagittal plane similar to DK, which may explain why there was no significant difference in knee flexion moments between those two postures.
Kneeling types were not significantly different with respect to the ab/adduction moment during any phase; however, the peak flexion moment during the ascending transition was significantly higher for PK compared to DK. During static kneeling, PK likely had a higher mean knee flexion moment because a greater proportion of the participant’s weight was over their feet. As described previously, a greater portion of the total GRF applied at the feet instead of the knees would mean a larger force applied at a greater distance from the knee in the sagittal plane.
The descent/ascent phases had higher moments of force for all four high flexion postures compared to their corresponding static phase. An average increase of 19% in flexion moments, 46.5% in adduction moments, and 14% in abduction moments were demonstrated for the peak knee moment in comparison to the mean values of the static hold. It could mean that the dynamic phases present a higher risk for acute injury, which could lead to further knee joint complications such as knee OA [9, 40]. Given that people who use high flexion for occupational or religious purposes may kneel or squat up to 15 times a day, the repeated exposure to high moments could be problematic [41]. Depending on the purpose of the high knee flexion activity, the posture may be held for a prolonged period of time. During the prolonged hold, a chronic exposure mechanism of knee OA initiation and progression may be more relevant. During non-pathological gait, the peak knee flexion moment is typically reported to be approximately 3% BW·H, which occurs for a very short period, albeit very frequently [42]. During kneeling, the peak flexion moment was as high as 9.13±1.45% BW·H for PK in the ascent phase, which is over twice as high as is reported during gait. The lowest mean flexion moment in a static hold was 4.63±0.99% BW·H for FS, which is still higher than is typically experienced throughout gait. Along with a high flexion moment, the static postures require 138°–153° of knee flexion, which may indicate that atypical structures within the knee are being loaded at magnitudes greater than are typically encountered, resulting in an increased risk of OA due to chronic exposure.
The current study shows that during all phases, the female subjects experienced significantly lower knee abduction moments (mean and peak) and not statistically significant greater peak knee adduction moment for all postures in comparison to the moments in male subjects. In this study, the moments were normalized by body weight times height. This normalization method is reported to be effective in reducing moment variability due to sex [42]. Several studies investigated the differences in kinematics and kinetics of the lower extremities between males and females during different athletic activities such as running, side- cutting, drop-landing, and squatting [18, 43–47]. These studies mostly reported a smaller range of knee flexion angle, greater peak knee valgus position, higher quadriceps muscle and lower hamstring muscle activation, and lower external abduction moments in females. Zeller et al. [29] reported larger flexion angles in women performing single-leg squats; however, the reported data were not statistically significant between genders. The findings of the current study for the knee flexion and ab/adduction moments are consistent with those of Kernozek et al. [45], which showed no significant difference in the peak flexion moment and significantly lower peak external knee abduction moment in females performing drop landings. Such differences can be associated with anatomical characteristics between male and female subjects. The lower extremity alignment and a larger Q-angle (the angle between femoral and the tibial axes in the frontal plane) in females could be associated with an increased knee valgus position and differences in muscle activity [45, 46]. Also, a less flexed posture of the torso and thigh in females during the descent phase of the squat may contribute to the differences in the knee joint position and moments [18, 45].
Preliminary statistical analysis (two-way repeated measures (pose (2) ×leg (2, i.e. right vs. left)) ANOVAs on the mean flexion and adduction angle in static phase) showed no main effect of the leg (p > 0.34) and no posture x leg interaction (p > 0.65). A similar statistical analysis on the moments (peak for the transitions and mean during static phase) showed no significant main effect of leg on flexion, adduction and abduction moments (p > 0.15) except for the peak adduction moment in descent phase (p = 0.003). An average of 0.78±0.69 and 0.79±0.71 (% BW.H) difference was observed between the peak adduction moments on the left and right knees in flat-foot and heels-up squats, respectively. Therefore, the assumption of symmetric squat is a limitation in the analysis of the adduction moment during the descent phase of the squat postures.
Another limitation of the current study was that torso angle was not measured and as a result, the effect of upper body flexion on the position of the knee joint in deep knee bending and the moments at the joint could not be analyzed.
The majority of previous studies on kneeling and squatting did not account for thigh-calf contact [4, 21]. However, neglecting thigh-calf contact could cause overestimation of the knee joint moment and the tibiofemoral joint contact forces in high flexion positions. In an initial assessment of deep squatting (beyond 150° of knee bending), Zelle et al. [48] found that incorporating thigh-calf contact decreased the compressive knee force by approximately 60%. Thigh-calf contact forces were not measured in this study, thus they were not included in the rigid link segment model.
Conclusion
The differences that exist between kneeling and squatting postures, as well as between styles of kneeling and squatting, underline the importance of selecting a joint saving posture when engaging in high flexion activities. In situations with no prescribed requirements for the posture, FS should be the posture of choice when aiming to minimize external knee joint flexion moments. The combination of peak flexion moment and peak adduction moment is a contributing parameter in prediction of the peak of medial contact force at the knee joint, which is a suspected contributor to medial tibiofemoral OA [9, 49]. FS can be challenging for those unfamiliar with the posture, however, based on its popularity as a resting posture and a toileting posture in some cultures [50], it is expected that it can be learned with practice. If FS is unobtainable, HS or DK should be employed. While HS produced lower external knee joint flexion moments than DK, they were not significantly lower. Also, kneeling is a more stable posture due to the larger base of support and thus might be more sustainable or safer than HS in applications where prolonged high flexion is required and the FS cannot be accomplished. Especially in cases where squatting is not possible or practical, DK should be used over PK.
Conflict of interest
None to report.
Footnotes
Acknowledgments
Funding for this study was provided by the Natural Sciences and Engineering Research Council (NSERC) (grant number 418647).
