Fall frequency among individuals with post-polio syndrome (PPS) is estimated at approximately 70%, and one-third of those who fall sustain fragility fractures in the polio-involved limb [1
]. Fall management programs for the PPS population mostly aim to manage asymmetric gait and reduce energy requirements [3
]. Gait is mainly improved by assistive devices [4
] and orthoses, balance training, and knee extensor strengthening exercises.
Non-fatiguing strengthening and exercises involving isokinetic, isometric, and endurance muscular training, have been demonstrated to improve symptoms of muscular fatigue and pain in patients with PPS [5
]. Reduced energy requirements are advocated to diminish the symptoms of fatigue. This is obtained by pacing strategies, including scheduling rest periods during the day, as well as during activities, and general lifestyle modifications, including weight control and modification of daily activities [8
One of the factors that contributes to the occurrence of falls in PPS patients is extensive muscle weakness, particularly of the knee extensors of the affected leg. This weakness exists either as a remnant of the primary infection or appears and progresses later in life as a part of PPS. Another risk factor for falls is the fear of falling. Previous studies reported fear of falling in 60–95% of PPS patients [1
]. Additional risk factors for falls in PPS patients include fatigue, muscle and joint pain, reduced sensation in the legs, depression, reduced gait performance, and impaired dynamic balance [1
Imbalance in PPS patients can be explained by asymmetric gait, caused by asymmetrical involvement of the muscles, chronic compensatory, musculoskeletal deformities or contractures, and excessive surgical or orthotic alternations. Balance confidence in this population might also be impaired by reduced proprioceptive input during muscle contraction, caused by altered sensibility of muscle spindles at different contraction levels [14
Post-polio syndrome patients may develop chronic compensatory musculoskeletal deformities or contractures, which allows them to increase ambulation speed and balance. However, excessive alternation of these compensatory mechanisms by an orthosis may worsen the mobility of the patient and induce falls [15
]. A consequence of the deformities can be quantified by leg-length discrepancy, which was found to be another predictor of falls in PPS patients [2
Presently, the acquirement of spatio-temporal gait data—e.g., step length and stance duration—has become frequent, due to the availability of simple wearable hardware or pressure-sensing mats (see Table 1
summarizing previous studies of PPS gait). These data hold within them variables of symmetry and variation, which provide important characterization of the PPS gait. However, while these data have been thoroughly quantified for PPS patients [16
], their ability to predict falls, has yet to be explored. As summarized in Table 1
, only four of the reviewed studies that performed instrumented gait analysis with PPS patients, collected data relating to fall or fear of fall. Two of which used low-technology (stop watch and pedometer) so that calculation of gait symmetry or variability could not be achieved. The remaining two researchers did not use the data to find which gait parameters best predict falls in this population. We therefore aimed (i) to analyze the correlation between spatio-temporal gait data and fall measures (fear and frequency of falls); and (ii) to test whether the gait parameters are predictors of the aforementioned fall measures in PPS patients.
The novelty of the current study is in its main finding, showing that the risk of falls in PPS patients can be predicted by simple parameters of the gait pattern: CV of the stance duration of the weak, CV of double support of the contralateral limb, and double support time (percent of the gait cycle) of the weak limb. These measures can be obtained in a clinical setting, but also using wearable sensors outside of the clinic. We believe that monitoring the gait pattern during daily activities will increase our ability to predict falls. Importantly, we believe our results may be proven a crucial factor in devising a real-time feedback system for the patient, alerting him or her regarding the need for rest or use of an assistive device, in order to prevent an imminent fall and possible related injury.
The gait characteristics of our study population are consistent with the results of our previous exploration of PPS gait, where we reported no association between the gait pattern of the PPS patients and sex, age, body mass index (BMI), education, and marital status [16
]. Therefore, the gait pattern may essentially result from the physical status and compensation mechanisms. We also showed that PPS patients who use orthoses and/or walking aids ambulated with a smaller base width and better stance and swing durations symmetry compared to PPS patients who require these aids [16
]. In the previous study, we did not calculate the CV of the spatio-temporal gait parameters. In the current study, we found that the gait variability differs between these two groups. Mainly, the CV of the base width, calculated separately for each leg, was smaller for the PPS group who use orthoses and/or walking aids compared to the PPS group who do not use them. This finding can be explained by the additional measurements described in Table 2
. Specifically, the PPS group who require walking aids showed weakness of the quadricep muscle, which explains their lower score of balance confidence, i.e., lower ABC scores. Both muscle weakness and lower scores of balance confidence are most likely the cause for their wider base width, slower gait velocity and reduced cadence, also demonstrated by lower TUG scores. We surmise that this gait strategy of slower gait and wider base of support allows these patients to better control their gait variability, expressed by lower CV of the base width and double support duration.
The ABC scale is a valid and reliable tool for prediction of falls. It was shown to be a strong predictor falls for elderly people [32
], individuals with Parkinson’s disease [34
], post stroke patients [35
], and individuals with multiple sclerosis [36
]. In PPS patients, it was reported that participants (n = 415) who had higher ABC scores were less likely to have risk of falls (p
= 0.028) [37
]. We therefore chose it as a secondary outcome measure for risk of falls. In our study, the ABC scores showed strong negative correlation with the TUG score and were moderately associated with the cadence and velocity of the gait. Additionally, the gait velocity was the best predictor of the ABC scale. It is not surprising that the gait velocity and cadence are related to the balance confidence. The patient will reduce gait velocity and cadence when feeling unsure of his or her stability while ambulating on different terrains. However, we expected that the time to complete the TUG, which includes both walking and turning, will also be compromised by low security in one’s balance. Although negative correlation between the ABC scale and the TUG score was reported in individuals with Parkinson’s disease (r = −372, p
< 0.01) [38
] and in older women (r = −0.39, p
< 0.001) [39
], proposed explanations for this unexpected finding were not presented by the authors. Since both the ABC scale and the TUG score were shown to be predictors of falls in various populations, we assume that there are different factors not shared between the two tests which serve as predictors of falls. These factors should be identified in future studies in order to better understand the motor and psychological mechanisms that contribute to the risk of falls.
The gait cadence and velocity were also negatively associated with the number of falls in the last year. PPS patients who walked slower and reported lower confidence in their balance also reported frequent falls. Here, we found positive correlation with the TUG score so that patients who took longer time to complete the TUG, also reported more falls. So much so that it seems that the motor components that play a role in the TUG are related to the actual risk of falls in PPS patients. Two more gait parameters positively associated with the frequency of falls were the symmetry index of the swing duration and the CV of the step length of the weak limb. That between-limb asymmetry of the swing duration and high variability of the step length in the weak limb were associated with a higher number of falls. These two parameters are related to one another, since asymmetry in the swing durations between the limbs means that the patient may drop down his or her leg sooner (or later) during the gait cycle, thereby decreasing (or increasing) the step length. This might cause an inadequate response to the natural movement of the body’s center of gravity during locomotion, resulting in loss of balance and inevitable fall. However, these two gait parameters were not the best predictors for the reported occurrence of falls. The three predictors of falls were the CV of the stance duration of the weak limb, CV of double support of the contralateral limb, and double support time (percent of the gait cycle) of the weak limb. Double support duration is an important gait measure found to increase with age as an attempt to increase stability [40
]. Additionally, the double support duration was found to be approximately 2% (of the gait cycle) higher in fallers compared to non-fallers [41
]. However, to the best of our knowledge, this is the first study to use the CV of the double support duration, along with other spatio-temporal gait parameters, in order to predict frequency of falls. Indeed, the CV of the double support duration was another predictor of falls in PPS patients, along with the CV of the stance duration. These findings might prove to be important when devising a predictive algorithm to monitor and provide real-time notifications to patients at risk of falls using the state-of-the-art technology.
Our results are supportive of new algorithms and applications that allow continuous measurement of gait velocity and cadence via the smartphone, placed on the body, in a bag, or on a belt [42
]. Additional technologies that acquire spatio-temporal gait parameters include E-Textile-Based wearable socks [44
], Inertial Measurement Units (IMU) [45
], Hall-effect sensors [46
], and smartwatches [47
]. A thorough list of wearable sensors used for gait analysis is detailed in [46
]. These technologies allow remote patient monitoring. Importantly, we believe that our results support the integration of real-time feedback and alerting algorithms that can protect the patient in case of dangerous gait patterns that predict an imminent fall.
The limitations of this study include the subjective report of the subjects regarding their history of falls in the last year and balance confidence. Although these reports might be sensitive to the subject’s interpretation, they are frequently used for the study of risk of falls [1
]. Another limitation is the single-day measurements in lab settings that might not be demonstrative of the daily gait pattern of the patient.