Reliability of instrumented movement analysis as outcome measure in Charcot-Marie-Tooth disease: results from a multitask locomotor protocol.

Some neurodegenerative diseases at early stage may not drastically affect basic gait ability, whereas more demanding locomotor tasks are more prone to disease-induced abnormalities. In this study, we evaluated the interday test-retest reliability, 4-6 weeks apart, of instrumented movement analysis on a group of 20 subjects with Charcot-Marie-Tooth (CMT) disease considering a set of kinematic and kinetic curves and related parameters obtained during natural walking (NW) and faster walking, heel and toe-walking, step ascending and descending. Results showed that the reliability was good for NW, with the exception of trunk curves, pelvic tilt and EMG profiles (moderate reliability), and trunk ROM in sagittal/transverse plane (poor reliability). Comparing our results with literature, CMT patients did not present a greater variability during NW than healthy subjects or patients with diseases of CNS. Additional locomotor tasks showed a slight reduction of reliability, although the moderate-to-good level shown in NW was almost never reduced to poor. Most of SEM values (absolute measurement errors) were smaller than 5°, a clinically acceptable threshold. In particular THS, an ankle joint related parameter computed across heel and toe-walking tasks, showed an optimal reliability (ICC=0.95, SEM=2.7°) and correlation with CMT clinical scores. Toe and heel-walking and step ascending tasks maximised the number of parameters with a moderate-to-good correlation with patients' clinical status. We concluded that, in addition to natural walking, more challenging locomotor tasks are good candidates to provide reliable and sensitive outcome measures for CMT patients.

Introduction

Charcot–Marie–Tooth disease (CMT) is the most frequent inherited neuromuscular disorder (prevalence 17–40/100000) [1] and presents different genetic forms with similar clinical symptoms, i.e. distal limb muscle wasting and weakness, skeletal deformities, distal sensory loss and reduced deep tendon reflexes [2]. The underlying pathogenic mechanism let the identification of a primary demyelinating variety (CMT1, CMT4) and a primary axonopathy (CMT2), although intermediate forms are recognised (CMTX1). The most predominant genetic forms are CMT1A (40–50%), associated with a chromosome duplication involving the Peripheral Myelin Protein-22 gene, and CMTX1 (10%). Symptoms often start in childhood and the disease then follows a slowly progressive disabling course, although there is a large variability in severity, which may occur also within the same CMT type and even the same family. This aspect highlights the crucial role of functional assessment for a correct individual planning of CMT patients’ clinical management.

There is still no pharmacological treatment for CMT and clinical approaches rely on physiotherapy, orthotics and surgical treatment of skeletal deformities. Ascorbic acid has been shown effective in an animal model of CMT1A and randomized controlled trials have been performed in humans, but thus far the results have been dismaying [3-5]. One major problem in conducting clinical trials is the slow disease course, making a major challenge the identification of sensitive-to-change outcome measures. Beside clinical and neurophysiological assessments, a quantitative evaluation of locomotor functions would provide important outcome measures with a higher sensitivity to progression-related or therapy-induced changes.

The progression of motor and sensory dysfunction usually follows a disto-proximal gradient, starting from the intrinsic foot muscles and then spreading to the leg muscles, thus affecting locomotor functions. The gait pattern is characterised by foot-drop (a reduced capacity to lift the forefoot during the swing phase of walking due to dorsiflexor muscles weakness) and a compensatory increase of hip and knee flexion [6,7]. Moreover, patients who have also a plantar flexor deficit, may present a clumsy gait pattern, with reduced cadence and step-length, and a broader support area [7].

Instrumented gait analysis, which allows for a precise analysis of human gait through optoelectronic systems, is a technique increasingly adopted in several locomotor disorders (for example cerebral palsy [8], adult hemiplegics [9] and focal dystonia [10]) for the planning of therapeutic interventions and the outcome assessment. Recently, some authors documented the abnormalities of gait biomechanics of CMT [6] and found subgroups with specific functional deficits [7]. However, the reliability of GA in CMT disease must be assessed, as already done in other diseases [11-13].

As at an early stage CMT may not drastically affect the basic gait ability, a multitask protocol including more demanding locomotor functions, like toe- and heel-walking, step ascending and descending, should be considered to increase sensitivity [14], as shown for patients with facioscapulohumeral dystrophy [15]. Accordingly, the aim of the present study was to assess the inter-session reliability of a multitask locomotion protocol of movement analysis applied on a group of CMT patients.

Methods

Subjects

Inclusion criteria were a diagnosis of CMT based on clinical, electrodiagnostic and genetic criteria. Exclusion criteria were presence of other neurological diseases or unrelated clinical conditions affecting locomotor functions; inability to walk barefoot without assistance and/or orthoses; previous double or triple arthrodesis at the ankle limiting joint ROM.

Twenty CMT subjects have been included, 13 males and 7 females, 14 with CMT1A, 4 with CMTX1 and 2 with CMT2. Mean age was of 24.6 years (SD: 17.1; range: 8–68), mean height 160.7 cm (SD: 14.9; range: 138–187), mean weight 53.4 kg (SD:13.9; range: 36–81) and mean Charcot–Marie–Tooth Neuropathy Examination Score (CMTES)1 of 6.3 (SD: 2.9; range: 2–11).

All subjects gave informed consent and the protocol was approved by the local ethical committee.

Study design

A single evaluator, test–retest design on 2 days, 4–6 weeks apart, was adopted to assess the reliability of output parameters.

Instrumentation and procedure

Kinematic data were collected with a 9-camera SMART-E motion capture system (BTS, Milano, Italy) sampling at 60 Hz and using the total-body LAMB marker set [16]. Two force plates (Kistler, Winterthur, Switzerland) were used to measure ground reaction forces (GRF) at 960 Hz sampling frequency. Surface electromyography signals (EMG) were recorded with a 8-channel and 1 kHz sampling frequency wireless ZeroWire electromyograph (Aurion, Milano, Italy), using 10 mm diameter electrodes at 20 mm inter-electrodic distance and 10–200 Hz bandpass filtering. Tested muscles included Tibialis Anterior, Soleus, Gastrocnemius Medialis, Peroneus Longus, Rectus Femoris, Vastus Medialis, Biceps Femoris and Gluteus Maximus of the most affected side or, in case of symmetry, of the right side.

The experimental protocol included 6 locomotor tasks: walking at natural (NW) and fast speed (FW), toe-walking (TW) and heel-walking (HW), step ascending (SA) and descending (SD). Each task was practiced and then repeated until five successful trials were recorded. The subject was asked to walk along a 12 m long walkway at self-selected speed (NW) and then at fastest speed, without running (FW). TW and HW tasks were performed at self-selected speed, asking the subject to maximise the lift of heel and toe from ground during walking. SA task consisted in climbing a two-step stair (step height: 18 cm, step depth: 30 cm), starting from ground level and stopping on the second step. The contrary was performed in SD task. Only the first step (a rigid wooden structure positioned on one force plate) provided GRF data. All tasks were performed barefoot and no instruction was given to the subjects about targeting the platform.

Data analysis

Markers’ coordinates were low-pass filtered (6 Hz cut-off frequency). Individual anthropometric parameters were computed from a static calibration trial, and used for estimation of internal joint centers. These, in turn, enabled calculation of trunk [17], pelvis and lower limb kinematics [16] during the locomotor tasks. Inverse dynamics let to compute moments and powers at the ankle, knee and hip joints. Dynamic data were computed in absolute terms as suggested by Fortin et al. [13] for test–retest reliability studies. Spatio-temporal gait parameters were normalized to subject's body height (BH). All kinematic and kinetic variables were time-normalized as a percentage of the whole stride duration (0–100%) and ensemble averages were then calculated. EMG signals underwent rectification, low-pass filtering (Butterworth 5th order, 3 Hz cutoff frequency) and dynamic maximum normalization (to the maximum value across all the trials of a task) for each muscle independently.

Specific values (maximum, minimum, values at specific time points within the gait cycle) were selected on each variable, in order to evidence specific clinical signs (e.g. foot drop, plantar flexor failure, joint ROM and their changes during heel- and toe-walking, compensatory movements at hip and knee during swing). Each parameter was averaged across the five trials. A detailed nomenclature of these parameters is presented in Table 1 and most of them are shown in Fig. 1.

Table 1

Mean and standard deviation values of parameters for each locomotor task for the considered population (N = 20 subjects with CMT). Codes for tasks: NW = natural speed walking; FW = maximum speed walking; TW = toe-walking; HW = heel walking; SA = step ascending; SD = step descending.

			Mean ± St Dev
Parameter	Code	Unit	NW	FW	TW	HW	SA	SD
Spatio-temporal
Gait speed normalized to body height	SN	[%BH/s]	70.4 ± 11.7	100.5 ± 18.4	55.4 ± 17.6	36.6 ± 14.0	29.2 ± 3.5	28.8 ± 3.7
Cadence	Cadence	[steps/min]	112.2 ± 9.6	140.7 ± 16.0	110.1 ± 19.5	103.3 ± 22.8	73.5 ± 6.5	91.1 ± 9.6
Stride length normalized to body height	SLN	[%]	74.9 ± 7.0	86.2 ± 8.9	59.4 ± 13.9	42.8 ± 13.4	29.0 ± 2.3	33.5 ± 6.1
Stance duration as percent of gait cycle	Stance%	[%]	61.3 ± 2.8	60.2 ± 4.3	64.7 ± 5.3	68.5 ± 6.1	68.2 ± 3.0	70.0 ± 4.0
First double support duration as % of gait cycle	Double%	[%]	11.1 ± 3.3	9.4 ± 2.8	15.5 ± 6.5	18.6 ± 6.5	17.6 ± 3.4	16.1 ± 5.5
Kinematics
Trunk ROM in sagittal plane (flexion	TSROM	[°]	4.6 ± 2.3	6.2 ± 5.2	5.9 ± 2.5	7.1 ± 2.6	14.8 ± 6.3	11.5 ± 4.4
Trunk ROM in frontal plane (bending	TFROM	[°]	5.6 ± 2.4	6.0 ± 3.1	7.0 ± 2.9	8.5 ± 3.1	14.8 ± 6.1	14.6 ± 7.5
Trunk ROM in transverse plane (rotation	TTROM	[°	8.3 ± 4.0	8.5 ± 6.1	12.8 ± 8.9	13.1 ± 8.1	14.1 ± 5.4	17.6 ± 7.0
Hip angle at foot strike	HSFS	[°]	31.6 ± 8.5	36.5 ± 9.6	30.7 ± 7.6	28.3 ± 10.0	51.1 ± 72.3	20.6 ± 9.2
Hip angle at foot off	HSFO	[°]	−0.9 ± 10.9	−2.1 ± 12.0	4.4 ± 11.9	16.6 ± 13.3	11.7 ± 10.7	32.7 ± 14.4
Hip extension peak	HSPK	[°]	32.3 ± 8.7	37.1 ± 9.4	31.1 ± 7.6	29.7 ± 10.7	68.9 ± 9.3	33.6 ± 13.7
Hip flex/extension ROM	HSROM	[°]	48.2 ± 5.4	55.1 ± 6.4	41.1 ± 6.1	31.5 ± 8.3	76.6 ± 64.8	39.0 ± 7.9
Hip ab/adduction ROM	HFROM	[°]	12.2 ± 3.0	14.7 ± 4.7	11.1 ± 4.4	13.1 ± 4.1	20.1 ± 7.3	19.9 ± 6.2
Knee angle at foot strike	KSFS	[°]	12.3 ± 6.4	17.1 ± 6.9	18.3 ± 8.5	5.8 ± 5.3	66.0 ± 8.7	14.4 ± 4.5
Knee angle at foot off	KSFO	[°]	42.1 ± 5.8	43.7 ± 8.5	37.9 ± 10.0	37.0 ± 14.1	18.9 ± 6.5	93.9 ± 5.5
Knee flex/extension ROM	KSROM	[°]	62.0 ± 6.0	67.4 ± 6.1	49.2 ± 8.6	45.3 ± 18.5	66.2 ± 9.8	88.6 ± 8.8
Knee flexion peak in stance	KSPK1	[°]	42.1 ± 5.8	43.8 ± 8.5	38.3 ± 9.9	37.1 ± 13.9	70.1 ± 7.6	93.9 ± 5.6
Knee flexion peak in swing	KSPK2	[°]	68.4 ± 6.7	71.2 ± 7.5	58.4 ± 10.0	46.8 ± 18.1	77.8 ± 8.1	101.9 ± 7.8
Ankle angle at foot strike	ASFS	[°]	−29.2 ± 7.0	−29.1 ± 8.1	−42.1 ± 8.6	−27.2 ± 8.6	−25.0 ± 9.7	−43.4 ± 7.6
Ankle angle at foot off	ASFO	[°]	−33.2 ± 8.2	−35.8 ± 7.2	−48.1 ± 10.1	−26.2 ± 10.1	−42.8 ± 8.3	−35.0 ± 11.5
Ankle dorsi/plantaflexion ROM	ASROM	[°]	30.4 ± 9.4	31.5 ± 12.3	28.8 ± 8.1	21.4 ± 10.8	42.6 ± 7.9	47.7 ± 8.4
Ankle dorsi/plataflexion ROM in swing	ASROMsw	[°]	15.2 ± 4.5	15.7 ± 6.7	16.7 ± 6.1	8.8 ± 5.2	24.0 ± 6.5	16.0 ± 7.8
Foot angle at foot strike	FoSFS	[°]	−11.9 ± 8.2	−13.1 ± 8.7	−34.7 ± 7.8	−10.4 ± 7.7	−31.0 ± 7.7	−37.7 ± 9.3
Mean ankle angle TW– mean ankle angle HW	THS	[°]	−13.3 ± 8.9
Kinetics
Hip positive mechanical work	HW+	[J]	13.4 ± 7.7	23.9 ± 10.4	11.7 ± 7.3	12.0 ± 8.0	27.5 ± 15.2	9.9 ± 7.6
Hip negative mechanical work	HW−	[J]	−5.3 ± 2.7	−9.0 ± 4.5	−5.4 ± 4.2	−2.7 ± 3.1	−18.5 ± 72.2	−4.6 ± 5.2
Knee positive mechanical work	KW+	[J]	5.1 ± 5.1	7.0 ± 6.3	4.8 ± 6.0	3.4 ± 4.4	37.4 ± 19.4	12.4 ± 22.9
Knee negative mechanical work	KW-	[J]	−11.3 ± 4.7	−17.2 ± 7.8	−7.8 ± 4.2	−6.7 ± 5.0	−6.4 ± 4.2	−52.9 ± 29.6
Ankle power positive peak	APwrPK	[W]	129.5 ± 47.5	152.4 ± 70.4	72.8 ± 30.6	38.7 ± 29.1	136.6 ± 51.8	65.0 ± 65.0
Ankle positive mechanical work	AW+	[J]	13.1 ± 4.2	15.2 ± 6.0	13.1 ± 6.0	4.3 ± 3.3	21.1 ± 8.3	13.3 ± 8.6
Ankle negative mechanical work	AW−	[J]	−7.9 ± 5.4	−7.7 ± 6.6	−11.9 ± 6.7	−5.3 ± 3.3	−5.9 ± 3.7	−25.3 ± 13.2
First peak of vertical GRF	GRFvMax1	[N]	587.2 ± 156.1	690.2 ± 207.0	618.0 ± 163.9	550.6 ± 169.1	558.9 ± 158.5	833.6 ± 281.3
Second peak of vertical GRF	GRFvMax2	[N]	600.6 ± 147.3	662.1 ± 162.4	567.7 ± 138.7	558.4 ± 155.2	613.6 ± 157.4	505.7 ± 164.6
Minimum between vertical GRF peaks	GRFvMin	[N]	401.2 ± 115.3	260.1 ± 86.5	399.1 ± 137.1	453.1 ± 116.8	377.8 ± 86.3	390.8 ± 114.2

Fig. 1

Time course of kinematic and kinetic variables selected for CMC analysis for the group of 20 CMT patients (average, solid black lines; ±SD, shaded area). Arrows indicate the parameters selected for ICC and SEM analysis. Vertical lines represent, respectively, contralateral foot off, contralateral foot strike and ipsilateral foot off (average, solid lines; ±SD, dotted lines). Only task NW is represented.

Intersession reliability was quantified by the coefficient of multiple correlation (CMC), the intraclass correlation coefficient (ICC) and the standard error of measurement (SEM).

The CMC is an index of consistency of waveforms between two sessions and has a range between 0 (no consistency) and 1 (perfect reliability). Kinematic curves were depurated of their mean value before calculating CMCs as suggested by Kadaba et al. [18]. Correlation between CMCs and age was assessed by means of Pearson's Rho coefficient.

Intersession reliability on selected parameters was assessed by the ICC. Model 2,k (with k = 5 repetitions) was adopted between the 6 possible versions [19], since the subjects were assessed by the same rater and the parameters were the average over the five repetitions for each task. As suggested by Portney and Watkins [20], CMC and ICC values above 0.75 indicated good reliability, those between 0.5 and 0.75 moderate reliability and those under 0.5 poor reliability.

McGinley et al. [21] recommended that studies reporting reliability of gait analysis data include absolute measures of measurement error such as the SEM:where SD is the standard deviation of the parameter values from all subjects [19].

Finally, the correlation of movement analysis’ parameters with clinical scales (CMTES: Charcot–Marie–Tooth Examination Scores [22], ONLS: Overall Neuropathy Limitations Scale [23], Walk-12 [24]) was assessed by Spearman's rank correlation coefficient (ρ). ρ values greater than 0.75 indicated good correlation and those between 0.5 and 0.75 moderate correlation. The level of statistical significance was set at 0.05.

Results

Fig. 1 shows, for the whole group of CMT patients, the mean (±SD) profiles of the main kinematic and kinetic variables for the NW task. In Table 1 mean and SD values of all parameters are reported. Since this study focused on the reliability analysis of curves/parameters and not on their pattern/values, the discussion on the latter will be addressed in a forthcoming study.

No subject reported pain during any test that could have interfered with the locomotor tasks and affected the results. Table 2 reports CMC values of all variables for all tasks. As regards NW, all kinematic curves showed good reliability, excluding pelvic tilt and trunk angles which had moderate reliability. Joint moments, powers and GRFs presented good reliability. All EMG curves showed moderate reliability, excluding peroneus longus and rectus femoris in FW which showed, respectively, good and poor reliability.

Table 2

Test–retest reliability for each locomotor task: CMC values (average over all subjects) of the present study, compared to results from Kadaba et al. [18] (healthy adults) and McGinley et al. [21] (average values of several studies including CP and stroke subjects). CMCs from Kadaba are the mean of right and left limb CMCs, which the author reported distinct. Codes for tasks as in Table 1 Code for reliability level: bold type = good (CMC ≥ 0.75); plain type = moderate (0.5 ≤ CMC < 0.75); shaded type = poor (CMC < 0.5).

Curve	CMC
	NW	FW	TW	HW	SA	SD	Kadaba et al. 1989 NW	McGinley et al. 2009 NW
Angular displacement
Trunk flexion	0.65	0.61	0.68	0.59	0.71	0.64	–	–
Trunk bending	0.71	0.55	0.61	0.63	0.59	0.55	–	–
Trunk rotation	0.64	0.63	0.62	0.64	0.51	0.60	–	–
Pelvic tilt	0.72	0.58	0.64	0.55	0.80	0.72	0.65	0.56
Pelvic obliquity	0.93	0.85	0.89	0.82	0.94	0.90	0.94	0.85
Pelvic rotation	0.94	0.90	0.93	0.84	0.78	0.75	0.83	0.72
Hip flexion	0.99	0.97	0.97	0.94	0.98	0.94	0.99	0.96
Hip adduction	0.93	0.87	0.81	0.80	0.86	0.81	0.95	0.89
Hip rotation	0.85	0.76	0.77	0.73	0.78	0.78	0.83	0.62
Knee flexion	0.99	0.94	0.94	0.91	0.95	0.97	0.99	0.96
Ankle flexion	0.96	0.91	0.91	0.86	0.92	0.92	0.97	0.93
Foot flexion	0.99	0.95	0.96	0.93	0.95	0.96	–	–
Foot progression	0.81	0.76	0.76	0.75	0.66	0.75	0.85	0.55
Hallux extension	0.93	0.88	0.86	0.74	0.71	0.86	–	–
Joint moment
Hip	0.96	0.92	0.91	0.89	0.92	0.85	0.97	–
Knee	0.89	0.82	0.74	0.70	0.91	0.84	0.94	–
Ankle	0.97	0.92	0.94	0.87	0.90	0.89	0.98	–
Joint power
Hip	0.93	0.77	0.83	0.74	0.90	0.65	–	–
Knee	0.88	0.72	0.76	0.61	0.87	0.84	–	–
Ankle	0.91	0.77	0.83	0.69	0.72	0.84	–	–
GRF
Antero-posterior	0.95	0.97	0.79	0.71	0.80	0.89	0.99	–
Medio-lateral	0.93	0.97	0.94	0.97	0.97	0.91	0.94	–
Vertical	0.86	0.87	0.63	0.78	0.56	0.75	0.99	–
EMG
Tibialis Anterior	0.71	0.72	0.61	0.65	0.62	0.57	0.84	–
Gastroc. Medialis	0.70	0.66	0.66	0.58	0.66	0.56	0.89	–
Soleus	0.74	0.68	0.67	0.57	0.64	0.54	–	–
Peroneus Longus	0.70	0.80	0.74	0.55	0.62	0.56	–	–
Rectus Femoris	0.64	0.45	0.57	0.53	0.63	0.58	0.84	–
Vastus Medialis	0.67	0.61	0.60	0.54	0.72	0.62	0.86	–
Biceps Femoris	0.64	0.65	0.67	0.60	0.61	0.55	0.82	–
Gluteus Maximus	0.68	0.61	0.63	0.60	0.72	0.52	0.84	–

Concerning the other locomotor tasks, CMCs were in general smaller than in NW, with the exception of trunk flexion and pelvic tilt for SA. However, most curves showed the same level of reliability (moderate or good) presented in NW. In particular, focusing on the sagittal plane, only hallux extension in HW and SA, knee moment in TW and HW, and all joint power in HW showed moderate reliability instead of the good reliability presented in NW. The majority of curves (86%) did not show significant correlation with age, while a moderate level of correlation (Pearson's Rho > 0.5; maximal value = 0.68) has been found only for 23 out of 186 curves, across all tasks. To determine the effective relevance of this finding, we split the whole group into two subgroups of 10 young (<18 years) and 10 adult (≥18 years) patients for comparison: only 6 curves showed significantly different CMCs between groups (p < 0.05, U-test). In particular, adult subgroup presented better reliability than young for ankle and hip power (HW), pelvis rotation (SA) and pelvic tilt (NW) – good vs. moderate CMCs – and for trunk rotation (SA and SD) – moderate vs. poor CMCs.

Table 3 reports ICC and SEM values of the parameters for all tasks. The global framework showed good or moderate reliability for the largest part of the selected parameters and tasks. On the other hand, trunk kinematics, especially in sagittal and transverse planes, were critical for all tasks except SD. Few kinetic parameters showed poor reliability and only for specific tasks (KW+ for HW and SD; HW− for SA; APwrPK and AW+ for SD). THS, which was calculated across the additional tasks TW and HW, had very high reliability (ICC = 0.95). Comparing ICCs and SEMs of additional tasks to NW, reliability appeared unchanged or lower, with a few exceptions, for all tasks similarly.

Table 3

Test–retest reliability for each locomotor task: ICC (model 2,k) and SEM values (in absolute units, as in Table 1) of the present study. Codes for tasks as in Table 1 Code for reliability level: bold type = good (ICC ≥ 0.75); plain type = moderate (0.5 ≤ ICC < 0.75); shaded type = poor (ICC < 0.5). Units of SEM values are the same as in Table 1.

Parameter	Code	ICC-SEM
		NW	FW	TW	HW	SA	SD
Spatio-temporal
Gait speed normalized to body height	SN	0.95–3.67	0.78–10.12	0.89–7.98	0.85–6.96	0.74–2.69	0.82–2.39
Cadence	Cadence	0.91–1.99	0.71–5.77	0.95–3.02	0.91–4.79	0.60–3.21	0.75–3.70
Stride length normalized to body height	SLN	0.96–2.16	0.78–6.09	0.87–7.15	0.87–6.33	0.90–1.13	0.68–4.42
Stance duration as percent of gait cycle	Stance%	0.92–1.00	0.59–3.34	0.84–2.65	0.86–3.21	0.63–2.37	0.61–3.08
First double support duration as % of gait cycle	Double%	0.93–1.11	0.72–2.19	0.97–1.64	0.84–3.62	0.73–2.43	0.82–2.92
Kinematics
Trunk ROM in sagittal plane (flexion)	TSROM	0.22–2.41	0.27–5.02	0.20–2.71	0.56–2.18	0.43–5.70	0.71–4.31
Trunk ROM in frontal plane (bending)	TFROM	0.77–1.35	−0.04–4.07	0.53–2.69	0.37–3.43	0.63–4.84	0.83–5.70
Trunk ROM in transverse plane (rotation)	TTROM	0.34–3.96	0.49–5.11	0.17–9.36	0.17–8.40	0.45–5.72	0.95–2.35
Hip angle at foot strike	HSHS	0.82–4.26	0.82–4.79	0.80–4.62	0.75–6.29	0.87–3.91	0.85–4.38
Hip angle at foot off	HSTO	0.92–4.43	0.91–4.99	0.88–5.81	0.82–7.22	0.84–5.34	0.89–5.90
Hip extension peak	HSPK	0.84–4.27	0.82–4.74	0.81–4.62	0.80–5.92	0.90–3.67	0.91–5.24
Hip flex/extension ROM	HSROM	0.97–1.55	0.84–4.56	0.83–3.63	0.91–3.07	0.93–2.44	0.96–2.55
Hip ab/adduction ROM	HFROM	0.92–1.46	0.83–2.54	0.83–2.31	0.81–2.33	0.80–3.56	0.90–2.33
Knee angle at foot strike	KSHS	0.87–3.71	0.89–3.42	0.93–3.49	0.89–3.14	0.90–3.32	0.87–2.85
Knee angle at foot off	KSTO	0.93–3.03	0.90–4.03	0.95–3.44	0.96–4.40	0.90–2.74	0.66–4.50
Knee flex/extension ROM	KSROM	0.95–2.09	0.63–5.23	0.84–4.90	0.97–4.51	0.87–4.13	0.93–3.19
Knee flexion peak in stance	KSPK1	0.93–3.03	0.90–4.03	0.95–3.52	0.96–4.40	0.93–2.71	0.66–4.50
Knee flexion peak in swing	KSPK2	0.92–2.75	0.83–4.44	0.89–5.40	0.96–5.22	0.86–3.12	0.90–3.39
Ankle angle at foot strike	ASHS	0.81–4.33	0.85–4.55	0.79–5.15	0.92–3.72	0.85–5.02	0.74–5.04
Ankle angle at foot off	ASTO	0.78–4.88	0.73–6.03	0.90–4.07	0.84–6.15	0.73–5.74	0.85–5.23
Ankle dorsi/plantaflexion ROM	ASROM	0.98–2.17	0.98–2.51	0.92–3.87	0.98–2.49	0.91–3.83	0.88–4.21
Ankle dorsi/plataflexion ROM in swing	ASROMsw	0.87–2.74	0.80–3.91	0.93–2.54	0.92–2.63	0.88–3.23	0.86–4.09
Foot angle at foot strike	FoSHS	0.87–4.32	0.90–4.15	0.73–4.44	0.85–4.46	0.56–6.76	0.79–4.79
Mean ankle angle TW − mean ankle angle HW	THS	0.95–2.72
Kinetics
Hip positive mechanical work	HW+	0.93–2.29	0.74–8.07	0.94–2.34	0.92–2.00	0.98–2.74	0.96–1.45
Hip negative mechanical work	HW−	0.92–1.08	0.89–1.58	0.93–1.44	0.70–1.56	0.34–10.68	0.93–2.19
Knee positive mechanical work	KW+	0.97–1.11	0.83–2.94	0.73–3.37	0.37–4.85	0.93–4.99	0.06–36.36
Knee negative mechanical work	KW−	0.86–2.43	0.84–3.15	0.94–1.12	0.85–2.15	0.94–1.40	0.77–14.03
Ankle power positive peak	APwrPK	0.92–15.75	0.92–21.25	0.67–26.55	0.70–25.44	0.87–21.25	0.41–137.59
Ankle positive mechanical work	AW+	0.93–1.40	0.95–1.62	0.74–4.25	0.79–2.29	0.94–2.47	0.39–25.17
Ankle negative mechanical work	AW−	0.95–1.61	0.96–1.74	0.69–5.67	0.82–2.09	0.92–1.62	0.83–7.59
First peak of vertical GRF	FvMax1	0.98–24.05	0.97–47.05	0.88–73.06	0.95–42.65	0.95–44.10	0.97–81.09
Second peak of vertical GRF	FvMax2	0.96–40.78	0.94–53.47	0.85–75.07	0.95–43.53	0.89–76.21	0.99–24.40
Minimum between vertical GRF peaks	FvMin	0.98–22.36	0.81–61.35	0.94–41.05	0.96–33.70	0.97–21.62	0.99–20.13

The results of correlation analysis are summarised in Table 4 (see caption for details). The tasks which presented the larger number of parameters with a correlation, at least moderate, with clinical scores were HW (number of parameters: 34), TW (25) and SA (26).

Table 4

Correlation between parameters and clinical scores (Spearman's rank coefficient—ρ). Only significant correlations (p < 0.05) with a correlation coefficient > 0.5 are reported. Codes for tasks as in Table 1 Codes for scores: C = CMTES; W = Walk12; O = ONLS. A ρ value between 0.5 and 0.75 is marked “+” or “−” according to the sign; a ρ value between 0.75 and 1 is marked “++” or “−−”. Last column and row report the number of significant correlations observed, respectively, in each row and column (ρ values higher than 0.75 are counted twice).

Parameter	Code	NW	FW	TW	HW	SA	SD	Σ by row
Spatio-temporal
Gait speed normalized to body height	SN	W+						1
Cadence	Cadence						W+	1
Stride length normalized to body height	SLN	O−						1
Stance duration as percent of gait cycle	Stance%	C+ O+		C+	W+			4
First double support duration as % of gait cycle	Double%				C+ W+ O+			3
Kinematics
Trunk ROM in sagittal plane (flexion)	TSROM					W + +		2
Trunk ROM in frontal plane (bending)	TFROM					W+	W+	2
Trunk ROM in transverse plane (rotation)	TTROM			W+			W+	2
Hip angle at foot strike	HSFS					W+	C+	2
Hip angle at foot off	HSFO			C+ O+	C+ O+	W+		5
Hip extension peak	HSPK					W+		1
Hip flex/extension ROM	HSROM	C+ O+	C+		C+			4
Hip ab/adduction ROM	HFROM				C+ O+	C+ O+	C+ W+	6
Knee angle at foot strike	KSFS			W+				1
Knee angle at foot off	KSFO			C+ O+	C+			3
Knee flex/extension ROM	KSROM				C+			1
Knee flexion peak in stance	KSPK1			C+ W+ O+	C+			4
Knee flexion peak in swing	KSPK2				C+			1
Ankle angle at foot strike	ASFS	W−	O−		C− W− O−−	C− O− W−		9
Ankle angle at foot off	ASFO				O−		W−	2
Ankle dorsi/plantaflexion ROM	ASROM		C+		C+ W+			3
Ankle dorsi/plataflexion ROM in swing	ASROMsw			C− O− W−		C− O− W−		6
Foot angle at foot strike	FoSFS	O−			C− O− W−			4
Mean ankle angle TW − mean ankle angle HW	THS			C+ O+ W+			3
Kinetics
Hip positive mechanical work	HW+	C+ O+	O+	C+ O+	C+ O+	O+ W+		9
Hip negative mechanical work	HW−						W−	1
Knee positive mechanical work	KW+			W+	C+		W+	3
Knee negative mechanical work	KW−			C− O−	C−	W−	O−	5
Ankle power positive peak	APwrPK					C− O−		2
Ankle positive mechanical work	AW+					C− O− W−		3
Ankle negative mechanical work	AW−		O− W−		C− O−	O−	O+ W+	7
First peak of vertical GRF	GRFvMax1		O+	O+	O+		C+ W+ O+	6
Second peak of vertical GRF	GRFvMax2	O+	O+	W+ O+	C+ O+	C+ W+ O+		9
Minimum between vertical GRF peaks	GRFvMin			C+ O+	O+			3
Σ by column		11	8	25	34	26	15	119

Discussion

Reliable and valid measures are particularly important in the assessment of patients with CMT, since such disease has slow progression and variable clinical expression. Nevertheless, only few impairment and disability outcome measures, including CMTNS/CMTES, ONLS and 10-m timed walking (T10MW), have been tested and shown to present substantial to excellent reliability [22,25].

Gait analysis is a powerful tool to quantitatively characterise the locomotor functions of patients with gait disturbances, including those presented by CMT patients [7]. However, the reliability of such data should be known to distinguish real changes in patient's condition from non-significant variability.

In the present study we assessed, in 20 CMT subjects, the test–retest reliability of variables and parameters extracted from a multitask gait analysis protocol. To our knowledge, this is the first study on reliability of gait analysis in CMT patients.

As a general picture, test–retest reliability of all curves (see Table 2) and parameters (see ICC values in Table 3) was good for natural walking (NW) with the exception of trunk curves, pelvic tilt and EMG profiles, which showed moderate reliability and trunk ROM in sagittal and transverse plane, which presented a poor reliability. Although moderate correlation between CMC and age was found for a few curves, the global picture does not substantiate relevant influence of age on curve reliability.

Comparing NW CMCs to those reported from Kadaba et al. [18] (relative to normal adults, natural speed), little difference was observed: arbitrarily considering noticeable a difference in CMC greater than 0.05, in present study pelvic tilt and rotation had higher CMCs, while vertical GRF lower CMC; all EMG curves presented lower CMCs. Comparing NW to McGinley et al. [21] (median of CMCs from 9 published studies on populations heterogeneous in age and pathology, including CP children and adults with stroke), present study CMCs were higher for pelvic obliquity, tilt and rotation, hip rotation and foot progression, not significantly different for the other variables. It can be concluded that CMT patients do not present a greater variability of kinematic and kinetic walking pattern at natural velocity than healthy subjects or patients with diseases of CNS. Concerning EMG profiles, the lower reliability found in our CMT patients when compared to literature data on healthy subjects, might be explained by the presence of muscle denervation and/or the adoption of variable muscular activation patterns to compensate for distal muscular weakness.

Additional locomotor tasks showed, in general, a slight reduction of reliability of curves and parameters. This can be explained considering the increase of intra-subject variability associated to a more challenging task than natural walking. However, in most cases, the level of reliability (moderate-to-good) shown in NW was not worsened, with the exception of knee positive work in HW and SD, hip negative work in SA and ankle positive work and power peak in SD, which are therefore not suitable for longitudinal studies. The analysis of SEM values (Table 3), indicating the absolute measure of measurement error in a test–retest paradigm, confirms this overall picture, with a general slight increase of SEM values associated to additional locomotor tasks compared to NW. As expected, parameters with smaller ICC showed higher SEM values, confirming their poor level of reliability. In particular, as regard angular parameters, most of them presented a SEM smaller than 5°, the value indicated by McGinley et al. [21] as error's threshold for clinical misinterpretation.

CMT patients typically present a distal-to-proximal progression of motor and sensory dysfunction. Therefore, depending on the stage of the pathology, a given locomotor test might be the most indicated in terms of sensitivity: heel and toe walking at initial stages, when only distal muscles are involved, step ascending and descending, at a more advanced stage, when also proximal muscles might be affected. In particular, THS, i.e. the difference between the mean ankle flexion/extension angle measured during toe walking and that measured during heel walking, showed an optimal reliability (ICC = 0.95, SEM = 2.7°), a significant correlation with CMT clinical scores, and, most likely, high sensitivity to distal motor deficiency, since it depends on ankle muscles recruitment capability. We therefore propose THS as an outcome measure for evaluation of early stage CMT patients.

Positive mechanical work at all lower limb joints in SA was greater than in all other tasks, while negative mechanical work at knee and ankle was greatest in SD. This reflects the biomechanics of step ascending and descending, which is characterised by a production and absorption of mechanical work needed, respectively, to raise and to lower the body mass [26]. Interestingly, the reliability associated to those parameters (HW+, KW+, AW+ in SA and KW−, AW− in SD) was always good, indicating that they can be adopted as outcome measures of concentric and eccentric muscular contraction capability of lower limb muscles in CMT patients.

Toe walking, heel walking and step ascending tasks maximised the number of parameters showing a moderate-to-good correlation with the clinical status of CMT patients (as measured by CMTES, ONLS and Walk-12). This suggests that in CMT, the different stages of the disease has a greater influence on those challenging tasks than on natural walking.

To conclude, instrumented analysis of gait and additional challenging locomotor tasks is a good candidate to provide reliable and sensitive outcome measures for CMT patients. An in-depth analysis of the tables provided in the present work, would help the reader to choose the task and the parameters of interest based on their reliability, associated measurement error and level of correlation with clinical scores.

Conflict of interest statement

Authors declare there are no commercial relationships which may lead to a conflict of interests.