Kinematic differences between OA males and OA females
The first purpose of this study was to examine gender-differences in gait kinematics between individuals with and without knee OA. The current study significantly builds upon previous research wherein the focus has been limited to only a single joint or plane of motion [5–8]. Moreover, the current investigation involved ankle, knee and hip joints, as well as foot and pelvis kinematics, for all three planes of motion, in an attempt to better understand the etiology of knee OA which is more prevalent in females as compared to males.
The results of the current study show that of the 112 discrete variables of interest, only kinematic variables for frontal plane knee and hip joint motion were significantly different between OA men and women during treadmill walking (Table 2). Specifically, OA females demonstrated greater knee abduction at touchdown and during swing as compared to OA males. These results are in contrast to Astephen Wilson et al. [5] who, using a similar PCA approach, reported that the only knee joint kinematic differences between OA males and OA females were in the sagittal plane knee angle range during stance. However, these authors [5] investigated severe knee OA patients prior to, and following total knee arthroplasty (TKA), and they only examined the differences in waveform shapes, not discrete variables, so comparisons with the current results are difficult. In the present study, OA females also exhibited significantly greater hip adduction angle at the maximum peak during stance in comparison to their male counterparts. Therefore, a novel finding of this study is that frontal plane hip and knee kinematics appear to be different between males and females, and the differences at the hip and the knee persist in both healthy and OA-symptomatic individuals.
In the sagittal plane, while McKean et al. [4] and Astephen Wilson et al. [5] both reported that females with moderate-to-severe knee OA exhibited reduced knee ROM angles across the gait cycle, similar results were not evident in the present investigation for subjects with mild-to-moderate knee OA. However, these findings are similar to Sims et al. [8] who also reported that knee ROM in OA females was no different compared to OA males with K-L grades of 1–4. It should be noted that the average walking speed observed by Sims et al., [8] (1.1 m/s) was more similar to that of the current study (1.1 m/s), contrary to McKean et al. [4] (1.3 m/s) and Astephen Wilson et al. [5] (0.9 m/s). The current results also suggest that the maximum peak knee flexion angle during the swing phase did not differ between OA males and OA females. This result is in contrast to Kaufman et al. [7], who observed this difference in 9 males and 11 females with OA. Therefore, based on the disparate findings amongst the current study and previous studies, further research may be necessary to better understand sagittal plane knee kinematics between OA males and OA females.
Kinematic differences between healthy males and healthy females
The current study found overall agreement as compared to previous investigations involving gender differences in gait kinematics for older adults [18, 19, 31]. Specifically, in the current study healthy females exhibited significantly greater maximum peak hip adduction during the stance phase of gait as compared to healthy males (Table 2). These results are similar to most previous studies that have also reported differences in frontal plane hip joint angles between young, middle-aged and older healthy males and females during both walking and running [14, 15, 18, 19, 25, 31, 32]. It has been suggested that increased frontal plane hip motion, together with hip abductor muscle weakness, may be a factor related to why healthy females are more likely to experience a musculoskeletal injury such as patellofemoral pain [33] or iliotibial band syndrome [34], as compared with their male counterparts. In addition, these results are also in support to previous studies [15, 16, 31] wherein healthy females exhibited significantly greater knee abduction at touchdown in comparison to their male counterparts (Table 2).
In contrast to investigations involving younger and middle-aged healthy adults [14–16, 30], the current study found no significant differences in knee external rotation angles nor were there differences in hip internal rotation angles between healthy older males and females. A possible reason for these contradictory findings may be the subtle changes in gait associated with biological aging [30, 35]. The mean age of the subjects in the current study was 53.26 years, while the subjects in aforementioned studies were in their twenties or forties. Therefore, it appears that gender-specific gait kinematic differences for healthy older adults are dissimilar to those previously found for healthy younger adults.
Kinematic differences between healthy subjects and OA subjects
The second purpose of this study was to assess the differences in gait kinematics between healthy gender-matched subjects as compared with their knee OA counterparts. There were no significant differences between healthy males and OA males or differences between healthy females and OA females in the original discrete variables. These results are partially agreement with the results of Ko et al. [11] and Weidow et al. [13] who reported no significant differences between healthy and OA subjects in knee kinematics for both gender-specific groups. On the other hand, these results are in contrast to a study by McKean et al. [4] who reported OA females exhibited less sagittal plane knee and ankle kinematics based on the PCA features of the gait waveforms as well as a study by Manetta et al. [10] who reported that OA males exhibited less knee flexion ROM during stance based on the discrete variables as compared to their healthy counterparts.
It is interesting to note that the standard deviations of the discrete variables, as well as the variability in waveform data, were both larger for OA affected males and females. This finding suggests an overall pattern of increasing variance, and possibly, individualized responses to disease progression, making characterization of the group as a whole, more challenging, especially when sample sizes are limited as in many of the aforementioned studies. Further research utilizing large sample sizes and sub-typing of OA individuals may provide valuable insight into characterizing gait changes in response to OA [36].
Multivariate analysis and classification model
When the number of biomechanical variables is high and the between-group differences are relatively small, multivariate analysis and machine learning methods can provide insight into group biomechanical characteristics. This study clearly shows that a PCA and SVM approach can provide insight into complex relationships of biomechanical gait variables, as compared to multiple univariate analysis methods. However, this approach does have a trade-off in the interpretability of the result, as the feature vectors used to separate genders or disease states often include data from many different joints and planes. It is therefore advisable to combine both approaches for a more comprehensive understanding of biomechanical differences between groups.
To our knowledge, no previous investigations have utilized the PCA and SVM approach to discriminate between male and female subjects with and without knee OA during walking. Deluzio and Astephen [37] used a PCA and a linear discriminant analysis (LDA) approach to discriminate between healthy and knee OA mixed-gender groups and reported a classification accuracy of 92 %. The results of the current study show that classification accuracies of 98–100 % are possible for discrimination between males and females for both healthy and OA subject groups as well as between healthy and OA subjects for male and female subject groups using the PC scores as the input features for the SVM classifier. Although not as effective as the PCA approach, the original discrete variables still produced classification accuracies of 71–86 % when used as input features for the SVM classifier. Thus, careful consideration of the final interpretation of the data, as well as the desire for high classification accuracy, are both needed when deciding on a statistical approach.
Limitations
Limitations to the current research study are acknowledged. We did not collect ground reaction force, or electromyography data and thus neither body kinetics, nor muscle activation patterns, were included in the analysis. However, Boyer et al. [18] reported no differences in the normalized ground reaction force between healthy and knee OA groups. Moreover, we chose to use joint kinematic angles to simplify the clinical interpretation of the results and shed some light on the greater prevalence of this disease in the female population as compared with males. Other clinical measures could also be incorporated to better understand the underlying mechanisms of knee OA. For example, future studies should include joint kinetics and ground reaction force data along with other clinical variables such as muscle strength, passive range of motion, muscle activation or knee stability to gain a greater understanding of sex-related differences in walking gait biomechanics, in an OA-affected population. In addition, self-reported pain and function scores, along with KL grade were only used for inclusion into the current study. Future studies should include these measures as previous studies have been shown to provide a better understanding gait kinematic patterns within distinct sub-groups of patients [38].
Since the subjects involved in the current study were all experiencing knee OA at the time of testing, and had been experiencing pain on most days of the previous week, cause and effect relationships cannot be established between the etiology of knee OA and walking biomechanics. However, the cross-sectional information gleaned from the current study has the potential to inform gender-specific rehabilitation and treatment approaches. Future prospective studies involving subjects, grouped by age and gender, will be an invaluable addition to the literature.
Confounding factors may exist between the groups studied, including pain (as previously mentioned), gait speed and BMI. We acknowledge that walking speed in individuals with knee OA may influence a number of gait biomechanical variables [39–41], however, the walking speeds of OA males (a mean of 1.134 m/s and a range of 1.01–1.23) and OA females (a mean of 1.146 m/s and a range of 1.06–1.21) in the present study were similar across groups and comparable to the self-selected normal walking speeds of a mixed-gender group of moderate knee OA patients in previous studies (e.g., a mean of 1.13 m/s and a range of 0.9–1.4 m/s [40]). BMI is also known to be an influential factor in the study of gait biomechanics [42, 43], and both hip and knee frontal plane kinematics in particular can be affected. There is, however, no consensus on the exact nature of these effects, and further research is needed to separate contributions of gender, BMI and joint disease to changes in overall gait.