The aims of this study were first to assess the differences in gait STPs and gait variability measured with IMUs during a 10-meter walk test between participants with PsA with foot pain, and age and sex-matched healthy participants and second, to investigate the relationships between gait STPs and variability and clinical outcomes of foot pain and disability.
Our findings showed significant differences in all the STPs between participants with PsA and matched controls. These differences included lower cadence, gait speed, stride length, swing time, and foot strike angle and, higher gait cycle duration and double support time in the PsA group compared to the healthy controls. However, only cadence, gait speed, and gait cycle duration remained significantly different after adjusting the differences for BMI. Nearly 50% of our PsA sample had a BMI above 30 kg/m2 which is not surprising because obesity is an important comorbidity of PsA . Moreover, obesity is known to alter STPs which has been suggested to be a strategy to lower joint loadings . Therefore, it is logical that BMI affected the differences in STPs between participants with PsA and controls in our study.
A few previous studies showed some alterations in STPs in people with PsA but not all of them demonstrated significant differences between PsA participants and healthy controls. For instance, Woodburn et al. demonstrated a significant decrease in gait speed in people with PsA with enthesitis (p= 0.014) . Hyslop et al. showed that except for stride length, there were no significant differences in cadence, gait speed, and double support time, between participants with PsA with enthesitis and healthy controls . Similarly, Wilkins et al. reported no significant differences in cadence, gait speed, and double support time in PsA with and without dactylitis compared to healthy participants . It is important to mention that all these studies included participants with a younger mean age compared to that reported in our study. In a recent systematic review, age has been shown to have significant effects on STPs in healthy adults . Thus, the more significant between-group differences demonstrated in the present study could be attributed to a combined effect of age and disease. Moreover, in the study by Hyslop et al. the participants were matched for BMI which was normal in PsA and control participants. This could explain the non-significant differences reported in their findings. In addition, in this latter study, even though patients with confirmed enthesitis were included, low to moderate levels of foot pain were reported by the authors which can also help explain their findings. In our study, although nearly 90% of the PsA participants were managed on DMARDs/biologicals and most of them had normal CRP levels, a high prevalence of simultaneous forefoot and rearfoot pain and moderate to severe levels of self-reported foot pain and disability were demonstrated. Clinically important differences in STPs between PsA and healthy participants and strong correlations between foot pain, foot function, and STPs, especially gait speed, were also demonstrated. Interestingly, these correlations were not affected by the CRP levels, disease duration, and lower limb pain since none of these clinical parameters was significantly correlated to STP. Although direct comparison between pain levels reported in Hyslop et al. and those reported in the present study cannot be done due to the different measurement tools used, our findings suggest that foot pain may play a major role in gait alterations in people with PsA.
Based on gait speed values, we were able to discriminate between three PsA subgroups. PsA participants who had gait speed values below 1.0 m/s, had higher FFI scores than those for whom gait speed was comprised between 1.0 m/s and 1.2 m/s and those with gait speed above 1.2 m/s. We didn’t have enough power to statically test the differences in the FFI scores between these three subgroups. However, knowing that the MCID for the FFI total score is 7 points, we were able to show that differences between these gait speed-based subgroups could be clinically significant. This suggests that gait speed may be a relevant metric not only to assess gait alteration in people with PsA but also to have more objective insight into the impact of the disease on self-reported foot pain and disability.
Results from studies addressing gait STPs in patients with RA are coherent with our study. For example, a previous systematic review on gait analysis of the lower limb in patients with RA showed they tend to walk slower, with a longer gait cycle, a shorter step length, a longer double support time, and a lower cadence compared to healthy subjects . These findings were confirmed in a recent metanalysis that reported a significant decrease in gait speed, stride length, and cadence and a significant increase in double support in patients with RA compared to healthy participants. Similarly to the present study, this metanalysis also reported large effect sizes for the differences between RA and healthy participants (Effect size (95% CI) were 1.55 (0.83 to 2.27); 1.66 (1.49 to 1.84); 0.97 (0.45 to 1.49)) and 1.01 (0.66 to 1.36) for gait speed, stride length, cadence and double support time, respectively .
It appears that, walking slower with shorter steps is a common compensatory strategy that people with arthritic foot disease use to reduce loads and pain in the affected joints and to increase stability [53, 54]. It has been reported that reducing gait speed leads to lower joint flexion and extension moments in hip, knee, and ankle joints  and that reducing step length allows for a decrease in the vertical ground reaction forces [56–58]. Moreover, double limb support in contrast to single limb support and swing (% GCT), is the most stable phase during gait and all these parameters represent the ability of the patient to transfer their body weight on the affected limb . Our findings, similarly to previous studies in RA patients showed a significant increase of double support and a reduction in the swing phase . This suggests that spending more time on both feet could be an adaptive approach to increase stability and reduce pain during gait.
Analysis of gait variability is a clinically relevant parameter in the evaluation of gait and responses to interventions and is a viable option for the quantitative evaluation of gait stability . To our knowledge, gait variability has never been investigated in people with PsA or other populations with foot involvement associated to arthritic joints disease. In our study, the mean stride time variability was higher in the PsA goup (4.49 ± 3.56%) compared to the control group (2.32 ± 0.72%), and above the normative values reported for stride time variability (1.1–2.6%)  indicating an increased gait instability. This is consistent with novel findings from a recent study that reported an increased risk of falling in people with PsA . Increased gait variability and instability could be ascribed to pain, muscles weakness, restricted range of motion, and a decrease of proprioception caused by inflammation in the foot joints and the surrounding structures . However, we did not find significant correlations between foot pain and stride time variability. Findings from a recent study reporting a significant alteration of static and dynamic balance in people with PsA, also showed that there were no correlations between balance parameters, foot pain and foot function . This suggests that pain may not be a determinant of gait variability and that this metric could be accepted as an independent gait parameter that should be assessed systematically in people with PsA. However, this needs to be confirmed in larger and longitudinal studies. Further studies are also needed to investigate the involvement of muscle weakness, reduced range of motion, and alterations of the proprioceptive system in gait variability.
Our study showed that none disease-related parameters (disease duration and CRP levels) were not correlated with self-reported foot pain and function which is consistent with results from a previous study conducted in people with spondyloarthritis . Gait spatiotemporal parameters and especially gait speed, however, were strongly correlated to these clinical outcomes. It would be interesting to investigate these associations in larger and longitudinal studies. It would be also relevant to assess the associations between gait parameters and important clinical domains such as disease activity, global function, and fatigue.
Body-worn IMUs for gait analysis are more than ever used in clinical assessment and clinical studies in several neurological diseases such as Parkinson’s disease, stroke, multiple sclerosis, and other conditions that increase the risk of falling. These systems are easy to use, time and cost-effective, do not require special equipment or expertise, could be used in any setting and recent evidence suggests that they could accurate and reliable to measure STPs in people with axial spondyloarthritis and PsA [21, 22]. This study suggests that body-worn IMUs could be useful to obtain an objective measure of functional mobility in people with PsA.
There are some limitations to this study. First, given the small sample size and the uneven distribution of males and females in our study sample, we cannot generalize the findings to the population. Second, we included patients based on their subjective perception of foot involvement. Although from a clinical perspective, the patients’ perception of pain and disability is a vital criterion, adding ultrasonography/MRI data to confirm the presence of enthesopathy, tendinopathy, synovitis, and /or bone erosions would have given more insight into the severity and progression of foot involvement. Third, CRP levels were documented from the participant’s clinical records which led to missing data and a delay (up to 3 months in a few participants) between CRP levels assessment and data collection. Moreover, important clinical domains including disease activity, skin disease activity and fatigue were not assessed which could significantly limit the proper description of the study cohort. Also, it is important to mention that gait variability was assessed over a 10-meter distance. Ideally, future studies should include longer distances while assessing this metric. Finally, the presence or absence of foot deformity were recorded in a qualitative manner (presence/absence). Using standardized tools such as the structural index could have been more relevant to ensure comparability between studies.