Cerebral palsy (CP) is a group of permanent diseases caused by a nonprogressive lesion in the developing brain, where posture and movement disorders lead to activity limitations [1]. Balance and gait disorders, which are common in CP, cause loss of independence in activities of daily living and psychosocial negative effects by restricting the functionality of the child [2-4].

Increasing functionality is the ultimate goal in CP rehabilitation, thus improving motor capacity is of paramount importance [4,5]. Motor capacity includes walking speed, walking endurance, mobility, and balance factors. Clinical rehabilitation programs using forward, backward, sideways, and cross walking training on various surfaces are known to improve balance and kinematic gait parameters [6-10]. In neuro-rehabilitation, backward walking training (BWT) is an alternative method which has been recently applied for improving motor capacity. Backward walking, although using the same rhythm cycle, is not thought of as a simple reversal of forward walking since it requires much higher physiological, sensory and perceptual effort [11-13]. BWT has been reported to ameliorate balance, postural control, walking parameters, and motor function in studies on the healthy population [9,14], stroke patients [10] and the elderly [5] in the literature. Moreover, it may have biomechanical and physiological advantages over forward walking, further enhancing lower extremity muscle strength [15] and cardiopulmonary response [7].

While forward walking training (FWT) on treadmill for children with CP has been investigated for more than three decades with moderate improvements in gait speed and endurance [16], recent investigations have shown that BWT improves gross motor skills, balance, and walking parameters in children with CP [17-24]. These studies focused mainly on BWT effects on balance and gait parameters using specially designed non-standard (proprietary) test and training equipment, while its impact on mobility and walking endurance was not examined. However, current knowledge in neurorehabilitation confirmed by our experience shows that improvements in motor skills in neurological patients do not reflect on walking endurance at the same rate. Therefore, the main purpose of our study was to determine whether BWT improves walking endurance as well as balance and mobility in children with CP using only nonproprietary and easily accessible clinical assessment and evaluation tools which may be used in simple clinical settings.

Demographic and clinical information of the participants are given in Table 1. There were no significant differences in weight, height, and body mass index (BMI) between the control and training groups. Almost twothirds of the children were at the spastic hemiparetic type CP level. No sex-related statistically significant differences were detected in their baseline assessment data.

During BW T sessions, very few (only 4) children achieved the maximal target speed (1.6 m/s) and this for only a short time (around one half minute). One child could not adapt to BWT and dropped from the study.

Statistics for all outcome measurement results are shown in Table 2. There were no significant differences between groups in any of the outcome measures evaluated before starting the training sessions. No sex-related statistically significant differences of outcome variations were found. All evaluations were found normally (Gaussian) distributed according to Shapiro–Wilk statistics.

Highly significant improvements in all outcome measures evaluated for the BWT group have been observed, while control group member assessments remained almost stationary during the training period. Training group 2MWT distance and PBS scores improved by 3.5% and their mean TUG Test time was 5.1% shorter after BWT. Moreover, their 10MWT duration was reduced by 6.1%, corresponding to an average relative short-distance walking speed improvement of 7.4%. The corresponding effect sizes, as evaluated with Cohen’s-d measure, have been computed as d=0.61 (PBS), d=0.26 (TUG), d=0.23 (10MWT) and d=0.18 (2MWT).

The eventuality of the interdependence of the outcome assessments of the study and their posttraining variations was investigated using Pearson’s linear correlation analysis. Weak correlations were detected between TUG improvements with those assessed for PBS (r=-0.39, p<0.05) and measured for 10MWT (r=0.41, p<0.05). In addition, no dependence on BMI, MMAS and GMFCS levels were identified. Moreover, all posttraining outcome variations were found independent of their baseline assessed values.

This study was planned to examine the effects of treadmill BWT on walking speed, walking endurance, mobility, and balance in children with CP as evaluated with the outcome measures we selected to assess these factors affecting motor capacity. Our 8-week-long supplemental BWT program improved significantly all measured outcome assessments. These improvements were independent from the participants clinical characteristics and baseline performance levels.

The results suggest that BWT, as an unusual and different practice, is effective in improving walking endurance, balance, and walking speed, as it works the musculoskeletal system differently, requiring more balance, coordination skills, and attention.

Improvements in mobility (TUG) correlated with those in balance (PBS) and short-time walking speed (10MWT). Since mobility involves both balance and ambulatory speed factors, this observed interdependence should be expected. On the other hand, the independence of balance (PBS) improvements from those for walking speed (10MWT) and walking endurance (2MWT) may point to a dual but independent effect of BWT on balance and ambulatory performance.

The independence of observed training-induced outcome variations from their baseline assessment scores as well as from MMAS and GMFCS levels suggests that the progress achieved in the training group with BWT intervention is uniform, i.e., does not depend on the child’s neurological impairment severity and therefore is not limited to children at predetermined impairment levels.

The effect of FWT on the treadmill on children with CP have been investigated for more than three decades; a comprehensive review of pertinent quantitative research. In 2019, Witherspoon et al. [16] reported significant moderate improvements in gait speed and endurance. Although BWT benefit in neurorehabilitation was established knowledge, it has only been recently used as an alternative to FWT in children with CP. The first identified BWT pilot study (without a control group) by Kim et al. [23] in 2013 with 12 participants aged 5–15 used video-based real-time gait analysis in which moderate improvements on gait speed and balance (in GMFM-88 dimensions D and E) were claimed. Later, in a series of studies by a separate research group, the effect of BWT was investigated for children with CP aged 5–7 at GMFCS levels II–III [18,19]; 5–9 at I–III [23]; 7–11 at II–III [20,21]; and 10–14 at I–II [17,22] concentrating on gait parameters including speed as well as on static and dynamic balance outcomes. This latter series of studies had 12 to 30 participants each with half of them assigned to control groups and used an additional body weight support system when appropriate. BWT was applied 3 times per week for 12 weeks, except only for 6 weeks in [18,19]. A metareview of their findings can be found in [5]; they observed dramatic (>50%) improvement in gait speed which they attributed mostly to improved strength and ~7% average amelioration of balance parameters. Only one study [17] compared backward walking (on the ground without using a treadmill) with FWT with no significant advantage of one over the other method. In all these past studies, specially designed proprietary systems were utilized to analyze gait speed and structural balance, none made use of commonly accepted clinical measures such as 10MWT, 2MWT, or TUG.

Although we found the progress seen after BWT statistically highly significant, the relative ameliorations achieved were small, all around 5% corresponding to small Cohen’s-d effect sizes (around d=0.2) except for PBS with a medium size effect variation (d=0.61). The effect reported in previous work were larger: the gait speed improvement reported in previous studies for 10MWT exceeded 50% in [17,20,24]; although with very high dispersion and was reported as 7.6% by Kim et al. [23], 2013 similar to our measured value of 6.1%. However, in these studies the participants walked at a much slower rate — by four times in [17,20,24] and twice slower in [23] — and their speed measurement method was different (selfselected comfortable rate vs. an encouraged competitive one in this study 10MWT). Likewise, only two studies [17,23] reported clinically reproducible (GMFM-88) balance improvements of 6%–8% and 8%–15% in dimensions D and E respectively. We observed only a ~3.5% score increase using PBS, which is over 90% correlated, thus comparable, with GMFM-88. One must note that our BWT program regimen was lighter than that applied in previous studies: 2 sessions per week vs. 3 sessions per week; 15 minutes per session vs. 15–25 minutes per session; 8 weeks duration vs. 12 weeks duration. This considerable difference in training time may partially explain the relatively smaller effect we achieved with BWT compared to previously reported values. Although the effect size, as evaluated with Cohen’s-d measure, may be judged medium for balance improvement (d>0.6) and weak for those in mobility, walking speed and walking endurance (d≈0.2), we must conclude the progress we observed, although significant, is below clinically significant thresholds [33,34]. Longer period or more intensive application of BWT may improve the impact on clinical outcomes.

This study contribution may be summarized as follows. First, although our exercise training methods were mostly similar to those applied in previous studies, unlike them, we used only nonproprietary and easily reproducible clinical scales for evaluating eventual progress; this may be used in further studies in a clinical setting to easily track improvements in motor capacity. Second, this study is the first to examine the effect of BWT on walking endurance. Finally, BWT effect on mobility was also investigated and it is the first report on this motor capacity factor.