Original Research
Response and Prediction of Improvement in Gait Speed From Functional Electrical Stimulation in Persons With Poststroke Drop Foot Michael W. O’Dell, MD, Kari Dunning, PhD, Patricia Kluding, PhD, Samuel S. Wu, PhD, Jody Feld, DPT, Jivan Ginosian, MS, Keith McBride, DPT Objective: To describe changes in and predictors of comfortable gait speed (GS-C) after using a foot-drop stimulator (FDS; Bioness L300; Bioness Inc, Valencia, CA) for 42 weeks in persons who had sustained a stroke. Design: Secondary analysis of prospective assessments. Setting: Multicenter clinical trial. Participants: A total of 99 subjects who had sustained a stroke 3 months earlier and who had GS-C 0.8 m/s and drop foot with a mean age of 60.7 years and a poststroke time of 4.8 years. Methods: GS-C was assessed at baseline and at 30 weeks with and without use of an FDS (therapeutic effect) and at 6, 12, 30, 36, and 42 weeks with use of an FDS (total effect). After subjects participated in 8 physical therapy sessions, an FDS was used for ambulation over the course of 42 weeks. Main Outcome Measurements: Changes in mean GS-C over time, FDS “responder” status defined as either 0.1 m/s gain in GS-C (the minimal clinically important difference [MCID]) or advancing by one Perry Ambulation Category (PAC), and the incidence and nature of adverse events (AEs). Results: A total of 74 (75%) and 69 (70%) of 99 subjects completed assessments at 30 weeks and 42 weeks, respectively. Baseline GS-C was 0.42 m/s without use of an FDS and 0.49 m/s with use of an FDS. GS-C improved to 0.54 m/s at 30 weeks without use of an FDS (effect size ¼ 0.75) and 0.54, 0.55, 0.58, 0.60, and 0.61 m/s at 6, 12, 30, 36, and 42 weeks with use of an FDS, respectively (effect size 0.84 at 42 weeks). Half of the subjects achieved a maximum GS-C by 12 weeks. Approximately 18% were PAC responders and 29% were MCID responders for 30-week therapeutic effect, and 55% were PAC responders and 67% were MCID responders for 42-week total effect. After logistic regression, the following factors emerged as the strongest predictors of FDS responders: younger age, faster baseline GS-C and Timed Up and Go, and balance. At 42 weeks, 60% reported a device-related AE; 92% were mild and 96% were anticipated. Conclusions: When an FDS was used, GS-C improved progressively over 42 weeks, with 50% of patients achieving a clinically meaningful 42-week total effect and 50% achieving a maximum GS-C by 12 weeks. Younger patients with greater mobility levels may benefit most from use of an FDS. AEs were frequent, mild, and reversible. PM R 2014;-:1-15
INTRODUCTION Seven million Americans are living with the effects of a stroke, and nearly 800,000 new and recurrent cases of stroke are reported annually [1]. Impairments resulting from stroke include language and cognitive deficits, visual changes, and contralateral hemiplegia, among many others [2,3]. Sustained distal weakness in the hemiparetic leg is manifested by an inability to adequately dorsiflex the foot during the swing phase of gait and can impede mobility profoundly [4]. This condition, known as “drop foot,” occurs in up to 20% of persons who have had a stroke [5]. Drop foot is seen with either weakness of the anterior PM&R 1934-1482/14/$36.00 Printed in U.S.A.
M.W.O. Department of Rehabilitation Medicine, New YorkePresbyterian Hospital/Weill Cornell Medical Center, Box 142, Room Baker F1600, 525 East 68th ST, New York, NY 10065. Address correspondence to: M.W.O.; e-mail: [email protected] Disclosure: nothing to disclose K.D. Department of Rehabilitation Sciences, College of Allied Health Sciences, Division of Biostatistics and Epidemiology, College of Medicine, University of Cincinnati, Cincinnati, OH Disclosure: nothing to disclose P.K. Department of Physical Therapy and Rehabilitation Sciences, University of Kansas Medical Center, Kansas City, KS Disclosure: nothing to disclose S.S.W. Department of Biostatistics, University of Florida, Gainesville, FL Disclosure: paid consultancy, Bioness Inc J.F. Department of Community and Family Medicine, Duke University School of Medicine, Durham, NC Disclosure: employment, Bioness Inc J.G. Department of Clinical Research, Bioness Inc, Valencia, CA Disclosure: employment, Bioness Inc K.M. Department of Marketing, Bioness Inc, Valencia, CA Disclosure: employment, Bioness Inc Research support: This clinical trial was funded by Bioness Inc, Valencia, CA. This study is registered under clinical trials registration number NCT01138995. Submitted for publication June 19, 2013; accepted January 1, 2014.
ª 2014 by the American Academy of Physical Medicine and Rehabilitation Vol. -, 1-15, --- 2014 http://dx.doi.org/10.1016/j.pmrj.2014.01.001
1
2
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 1. Published studies of surface foot-drop stimulation and gait speed in persons with chronic stroke (n > 10) Study, Year (Reference) Current analyses
Total N (Stroke, n)
F (%)
Mean Age, y
Mean Time Poststroke, y
Mean Baseline GS (no FDS), m/s
F/U Time, wk
F/U, %
IE*
99
48
60.7
4.8
0.42
30
75%
42
70%
0.07 17% 0.07 17%
Everaert et al, 2013 (23)
38
33
57.1
0.53
0.52
6
88%
0.07 15% 0.08 16% NR
Taylor et al, 2013 (22)
62
47
59.6
4.8
0.49
>14.3y
90%
van Swigchem et al, 2010 (24)
26
19
52.8
3.2
1.02 (with AFO)
8
96%
Stein et al, 2010 (25)
73 (26)
46
58.8
7.5
0.64
44
32%
0.05 6.5%
Sabut et al, 2010 (28)
16
20
49.5
1.5
0.37
12
100%
NR
Hausdorff and Ring, 2008 (14)
24 (21)
17
54.0
5.8
0.53
8
100%
Laufer et al, 2009 (26)
16 (13)
7
55.0
5.3
0.62
52
67%
0.09 17% NR
Stein et al, 2006 (16)
26 (12)
NR
57
NR
0.69
12
100%
Taylor et al, 1999 (27)
140 (111)
NR
55.4
5.4
0.57
18
93%
Burridge et al, 1997 (15)
16
38
52.3
3.6
0.64
12
100%
Granat et al, 1996 (29)
19
16
56
7
0.87
11
84%
0.02x 3% 0.07 12% 0.04 6% NR
F ¼ female; GS ¼ gait speed; FDS ¼ foot-drop stimulator; F/U ¼ follow-up; IE ¼ immediate effect; TrE ¼ training effect; ToE ¼ total effect; ThE ¼ therapeutic effect; AE ¼ adverse event; NR ¼ not reported; TUG ¼ Timed Up and Go; BBS ¼ Berg Balance Scale; GS-C ¼ comfortable gait speed; 10-MWT ¼ 10-meter walk test; ODFS ¼ Odstock Dropped Foot Stimulator; PAC ¼ Perry Ambulation Category; MCID ¼ minimal clinically important difference; AFO ¼ ankle foot orthosis; TBI ¼ traumatic brain injury; 6-MWT ¼ 6-meter walk test; 6/10-MWT ¼ 6- to 10-meter walk test; PCI ¼ physiologic cost index. * Numbers are gains in gait speed in m/s; % is percent gain in gait speed from baseline as defined in text. y Average of all assessments taken after 100 days. z Assuming no FDS at baseline (not stated in study). x Estimated or calculated value for this table.
muscles, spasticity of the posterior muscles of the leg, or both. Subsequent development of plantar-flexion contracture at the ankle further complicates limb clearance. Drop foot contributes to undesirable compensatory movement patterns, slowed gait speed, limited functional mobility, and an increased risk of falls [6-8]. A number of treatment strategies for drop foot may be used depending on the underlying cause. Contracture at the ankle joint in a person with good ambulation potential may be managed surgically [9]. Posterior compartment spasticity is treated with exercise and a combination of oral, injection, or intrathecal medications [10]. The most common treatment is an ankle-foot orthosis (AFO), which is a brace that holds the ankle in a neutral position and improves limb clearance during the swing phase of gait. AFOs have significant drawbacks, however, including limitation of ankle mobility, which contributes to contracture development [11], possible undermining of motor recovery [12], difficulty
in standing from a seated position [6], and poor acceptance by the wearer as a result of discomfort and the perception of undesirable aesthetics [13]. Another strategy to combat drop foot is the use of functional electrical stimulation (FES). FES is used to stimulate branches of the common peroneal nerve just below the knee, thus activating dorsiflexion and eversion of the foot. A footdrop stimulator (FDS) is a commercially available FES system designed specifically for use during ambulation. These devices synchronize surface stimulation of the nerve with the swing phase of the gait through the use of either pressure sensors or heel switches placed in the shoe [14,15] or a tilt sensor [16]. The potential effects of FDS devices have been described previously by Dunning et al [17]. Gait speed has been studied most often, but any relevant parameter could be used. An immediate device effect refers to changes that occur at once (ie, within minutes) when “FDS off” is compared to “FDS on.” A training effect above and beyond the
PM&R
Vol.
-,
Iss.
-,
2014
3
Table 1. Continued Device Type, Skin-Related AE, GS Determination Method for Reported Data, Other
TrE*
ToE*
ThE*
Predictors of GS
0.09 18% 0.12 25%
0.16 38% 0.19 45%
0.12 29% NR
0.10 19% NR
0.17 37% 0.18 45% NR
0.12 26% 0.11 24% NR
ThE at 30 wk related to age, TUG, BBS, time poststroke ToE at 42 wk related to age, GS-C, BBS, hemorrhagic stroke, partial sensory loss NR
0.25 32%
0.19 25%
NR
0.10z 27% 0.18 34% 0.29 47% 0.08 12% 0.15 26% 0.13 20% 0.06 7%
NR
NR
WalkAide, AE not reported, 10-MWT, demographics for stroke only, FDS effects includes 15 nonstroke subjects, F/U % ¼ 13/41 nonprogressive subjects EMS (CyberMedic Corp), AE not reported, 10-MWT
NR
NR
L300, no AE observed, 6-MWT results included 3 TBI subjects
0.17 27% 0.03x 4% 0.08 14% 0.01 2% 0.07 8%
NR
L300, AE not reported, 6-MWT, extension of Hausdorff and Ring’s study WalkAide, skin irritation in “a few” subjects, 10-MWT, some data include nonstrokes ODFS, AE not reported, 10-MWT, data are stroke-specific, long-term F/U reported previously ODFS, AE not reported, 10-MWT
0.05 5% 0.20 24% NR 0.09 15% NR 0.06x 9% 0.08 13% 0.09 13% NR
NR NR
ToE not related to age, duration of drop foot ToE not related to time poststroke, ThE related to right hemiparesis ToE related to greater baseline PCI but not to initial gait speed NR
immediate device effect may occur as a person uses the FDS over weeks to months. The total effect is the sum of immediate and training effects. The immediate, training, and total effects reflect an orthotic benefit of FDS. Finally, a therapeutic effect represents a benefit that persists even without FDS and is sometimes termed a “carry-over effect” [15]. The underlying mechanism for the therapeutic and training effects probably results from neuroplastic changes, improved peripheral strength, and cardiopulmonary conditioning, as well as changes in other systems [17,18]. A few meta-analyses [19-21] and several FDS studies [14-16,22-30] have demonstrated improved gait parameters with use of an FDS in persons who have had a stroke. Table 1 compares a number of parameters and gait speed among published surface FDS studies, including 10 subjects who had sustained a stroke. A recent study that examined therapeutic FDS effect was not included because the authors used a different methodology to determine gait
L300, 48.5% of subjects with skin irritation, 10-MWT
WalkAide, 10% skin irritation (for all FDS users), 10-MWT ODFS, 22% mild skin irritation, long-term F/U of Taylor et al 1999 (27), 52% PAC and MCID responders at 16.5 m, 10-MWT L300, 15% skin irritation, 10-MWT, no measurements without FDS
FES Medical Electronics, AE not reported, 6/10-MWT
speed [31]. Therapeutic and total effects are frequently demonstrated with a wide array of research designs. It is important to consider that, beyond their use as an adjunct treatment during physical therapy (PT) sessions, modern FDS systems function like durable medical equipment meant for long-term use at home and in the community. Despite this long-term use, inspection of Table 1 reveals few studies in which investigators comment on the trajectory of or factors predicting clinical response over time; this information could help clinicians identify appropriate candidates for an FDS device. The Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy (FASTEST) study is a large, multicenter, randomized, controlled, single-blinded clinical trial conducted from 2010 to 2013 in which investigators compared the impact of an FDS versus an AFO on gait speed and other parameters in persons who had sustained a stroke and had drop foot. Details of the FASTEST methodology
4
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
[17] and 30 weeks’ comparison between devices [30] have been previously published. Of the 197 subjects entered in the study, 99 were randomized to the FDS arm and underwent extensive clinical assessment at baseline and 30 weeks with and without the device to assess therapeutic effect and 5 times with the device (at 6, 12, 30, 36, and 42 weeks) to assess the immediate device, training, and total effects. Because the original FASTEST analysis did not address the trajectory or prediction of response to the FDS, isolated consideration of these 99 subjects provides an opportunity to examine the clinical response to an FDS in persons who have sustained a stroke. Therefore the objectives of this secondary analysis of FASTEST data are to (1) characterize changes in comfortable gait speed (GS-C) over time with use of an FDS; (2) calculate the immediate device, training, total, and therapeutic effects of an FDS; (3) determine variables that predict FDS “responders” versus “nonresponders” for a 30-week therapeutic effect and 42-week total effect; and (4) assess adverse events (AEs) over 42 weeks of use.
METHODS
persons identified as having subacute (3-6 months) and chronic (>6 months) stroke who demonstrated drop foot as a significant limitation in gait. The drop foot had to be severe enough to require an AFO and, if necessary, a new or modified AFO was provided by the sponsor to ensure safety during the first weeks of FDS training. Subjects walked slowly (GS-C 0.8 m/s) but required the assistance of no more than one person to walk 10 m with or without an assistive device. In addition to requiring an AFO, all subjects were required to demonstrate adequate and tolerable peroneal nerve stimulation with the FDS unit before randomization. After their eligibility was confirmed, subjects were randomized to the AFO or FDS groups by the sponsor with the use of covariate adaptive randomization to ensure target allocations for age and time poststroke in 4 subgroups as follows: 3-6 months poststroke and age 65 years or Medicare beneficiary (target N ¼ 7); >6 months poststroke and age 6 months poststroke and age 65 years or Medicare beneficiary (target N ¼ 150).
Subjects and Randomization
Interventions and Assessments
A detailed description of the FASTEST trial design and methods has been previously published [17]. Participants were recruited at 11 sites throughout the United States, with each site obtaining institutional review board approval and informed consent before performing any assessment. Inclusion and exclusion criteria are outlined in Table 2. Salient to the current analyses, the FASTEST trial included
Hereafter, we describe interventions related only to the 99 subjects randomized to the FDS arm. During the first 6 weeks of the study, subjects participated in 8 sessions of physical therapy (a total time of 225-285 minutes). Two sessions per week were provided during the first 2 weeks and one session per week was provided during the next 4 weeks. The first visit focused on device fitting and settings;
Table 2. Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy study inclusion and exclusion criteria Inclusion Criteria
Exclusion Criteria
Ankle dorsiflexion neutral when stimulated with FDS in sitting and standing positions Adequate ankle and knee stability during gait 3 mo from at least 1 stroke of any etiology Foot drop caused by stroke sufficient to require use of an AFO (as assessed by the site’s physical therapist and orthotist) Adequate cognition and communication abilities (24/30 on the Mini Mental State Examination) or having a competent caregiver 18 years Able to safely walk at least 10 m with a maximum assist of 1 person Comfortable 10-m gait speed 0.80 m/s Medically stable
Fixed plantar flexion contracture 5 Adequate response to FDS Pain 4/10 in the affected leg Any other interventional research studies Demand-type cardiac pacemaker, defibrillator, electrical, or metallic implants Significant swelling/edema in the leg extending up to the knee Chronic skin problems or lesions near FDS stimulation site Pregnant/planning to be pregnant 6 weeks from botulinum toxin to weak leg or arm, or planned during the course of the study Expectation of change in medications for spasticity Unstable seizure disorder Pre-existing orthopedic conditions affecting ambulation Complete vs hemisensory loss in leg FDS or FES use >3 hours in 6 mo before enrollment Major depression (PHQ-9 >10) not managed by a health care provider Participating or planning physical or occupational therapy
Outlined are the inclusion and exclusion criteria of the Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy trial. All 99 of the subjects in the current study met these criteria. FDS ¼ foot-drop stimulator; AFO ¼ ankle foot orthosis; FES ¼ functional electrical stimulation; PHQ-9 ¼ Patient Health Questionnaire, 9-item version.
PM&R
during visits 2-4, device education, initial gait training, and a home exercise program were provided; and the remaining visits focused on gait training with the FDS. The specific FDS device used in this study was the Bioness L300 Foot Drop System (Bioness Inc, Valencia, CA), which comprises a stimulation cuff with electrodes affixed on the anterolateral aspect positioned below the knee, a control unit, and an inshoe pressure sensor under the foot. The stimulation unit initially is configured by a clinician, who uses a handheld, wireless computer interface. Gait and stimulation parameters (amplitude 0-100 mA, duration 100-300 microseconds, frequency 20-45 pps, and gait parameters, including ramp up/down 0-2 seconds) are individualized to achieve a functional, comfortable, and timely muscle contraction. The pressure sensor detects “heel off” at the start of swing phase and wirelessly initiates peroneal nerve stimulation, facilitating dorsiflexion and eversion of the ankle. When the foot makes “initial contact” at the start of stance phase, the stimulation is paused [17]. Standardized protocols developed by the sponsor, who has more than 5 years of market experience, were used by all sites for initial fitting of the FDS, gait training, the wearing schedule, the conditioning program, and device education. Home-exercise programs were individualized and focused on gait training with the FDS device. During the first 3 weeks of use, subjects gradually increased the time spent walking with the FDS from 15 minutes per day to all day. In addition, the cyclic training function (5 seconds on and 8 seconds off) of the device was used 15 minutes twice a day for the first week and 20 minutes twice a day for an additional 2 weeks to help condition the muscles stimulated during ambulation. An AFO initially was used as needed to ensure safe ambulation when the FDS was not being used. After the initial conditioning phase, subjects were to use the FDS exclusively for ambulation. At the time of the trial, the manufacturer provided hydrogel electrodes as the primary medium for transcutaneous conduction. Written skin care guidelines were reviewed and issued during the initial fitting and reviewed throughout the training period. Device compliance (number of steps per day) was estimated with use of a step activity monitor (StepWatch; Orthocare Innovations, Oklahoma City, OK) in weeks 6 and 24.
Outcomes Outcomes were obtained without the FDS at baseline and at 30 weeks to examine therapeutic effect. Outcome measures were also obtained with the FDS at baseline and at 6, 12, 30, 36, and 42 weeks to examine immediate device, training, and total effects. Routine, scheduled follow-ups were completed at weeks 16, 20, and 24. Outcome testing was performed by a physical therapist blinded to group assignment at baseline, 6, 12, and 30 weeks and not blinded at 36 and 42 weeks. To maintain blinding, a shroud of vinyl fabric was secured over the lower leg and shoe to conceal the
Vol.
-,
Iss.
-,
2014
5
FDS device. Therapists received training from the sponsor and passed an on-site competency test for consistency in assessing outcomes. The primary outcome in the FASTEST trial and the dependent variable for this secondary analysis is GS-C as assessed with the 10-m walk test [32-35]. Independent predictor variables included baseline GS-C, demographic and stroke characteristics, and clinical assessments encompassing primarily impairment and activity limitation as defined by the International Classification of Function [36]. Clinical assessments included the lower extremity FuglMeyer Assessment (LEFMA), Timed Up and Go (TUG), endurance via the 6-minute walk test, gait speed “as fast as possible,” and the Berg Balance Scale. All outcome measures are valid and reliable in persons who have sustained a stroke and have previously been described in detail [17]. The occurrence and types of AEs were assessed at all study visits. When an AE occurred, study personnel assigned a severity and determined whether it was caused by the device. The AE was also designated as an “anticipated AE” if it was included in the initial Bioness L300 filing with the Food and Drug Administration device approval process. The incidence of falls and injury were self-reported by participants and/or their caregivers retrospectively for the 6 months before baseline and prospectively at each study visit.
Statistical Analysis The FASTEST trial established a recruitment target of 206 subjects that was based on an 80% power to detect a minimal clinically important difference (MCID) of 0.1 m/s between the FDS and AFO groups [37] with the use of a 2-sample t-test with a 2-sided 0.05 alpha level. Other considerations determining the sample size have been discussed previously [30]. To assess potential bias from missing data, paired t-test, Mann-Whitney U test, or c2 tests were used to explore baseline variables associated with subjects who were “completers” and “noncompleters” at the 30- and 42week assessments. For objectives 1 and 2, descriptive statistics (ie, mean, median, range, standard deviation, and 95% confidence intervals [CIs]) were used to characterize changes in GS-C over 42 weeks of FDS use. The proportion of subjects achieving their maximum GS-C with an FDS at each assessment time also was noted. Paired t-tests were used for pairwise comparisons of GS-C between baseline and selected follow-up assessments to assess FDS orthotic and therapeutic effects. Cohen’s effect size [38] also was calculated for selected comparisons. A linear mixed model [39] was used to assess statistically significant changes in GS-C with use of an FDS among consecutive assessments, baseline through 42 weeks. For objectives 1 and 2, only data for completers were analyzed. Descriptive statistics only were used to address AEs in objective 4. To address objective 3, dichotomous categories of FDS “responder” and “nonresponder” were defined by either (1)
6
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 3. Demographic and baseline characteristics for 99 subjects randomized to changes in a foot-drop stimulator Variable
Overall (n [ 99)
Completers at 42 Weeks (n [ 69) (%)
Noncompleters at 42 Weeks (n [ 30) (%)
P Value*
Age, y Comfortable gait speed, m/s (m SD) Fast gait speed, m/s (m SD) 6-minute walk test, feet (m SD) Lower extremity Fugl-Meyer assessment (m SD) Timed Up and Go, s (m SD) Berg Balance Scale (m SD) Gender Female Male Prestudy AFO New prescription Prestudy prescription Ethnicity White Black Asian Hispanic Other Marital status Divorced Married Other Single Widowed Handed Left Right Education Did not complete high school High school graduate College/trade school Postgraduate school Graduate school Time poststroke, mo 3-6 6 Stroke side Left Right Vascular distribution MCA ACA PCA Basal ganglia Cerebellum Brainstem Unknown Other Stroke location N/A Cortical Subcortical Stroke type N/A Hemorrhagic Infarction
60.71 12.24 0.42 0.21
60.00 12.29 0.44 0.22
62.33 12.17 0.38 0.17
.386 .204
0.56 0.32 152.07 90.04 20.14 5.18
0.59 0.34 159.36 96.02 20.23 5.16
0.50 0.25 135.30 73.24 19.93 5.32
.223 .224 .794
34.34 27.73 40.31 9.00
33.53 28.05 40.55 9.11
36.21 27.35 39.77 8.86
.661 .692
48 51
32 (66.7) 37 (72.5)
16 (33.3) 14 (27.5)
.525
43 56
24 (55.8) 45 (80.4)
19 (44.2) 11 (19.6)
.008
54 26 5 10 4
41 (75.9) 14 (53.8)
13 (24.1) 12 (46.2) 5 (100.0)
.0002
10 (100.0) 4 (100.0)
17 56 2 17 7
11 37 2 13 6
6 (35.3) 19 (33.9)
.620
9 90
7 (77.8) 62 (68.9)
2 (22.2) 28 (31.1)
.580
7 23 48 6 15
4 20 31 6 8
3 (42.9) 3 (13.0) 17 (35.4)
.064
9 90
8 (88.9) 61 (67.8)
1 (11.1) 29 (32.2)
.189
46 53
36 (78.3) 33 (62.3)
10 (21.7) 20 (37.7)
.084
45 4 3 12 1 3 18 13
30 3 2 11
15 1 1 1 1
.405
3 (100.0) 11 (61.1) 9 (69.2)
7 (38.9) 4 (30.8)
12 48 39
8 (66.7) 33 (68.8) 28 (71.8)
4 (33.3) 15 (31.3) 11 (28.2)
.926
4 15 80
3 (75.0) 13 (86.7) 53 (66.3)
1 (25.0) 2 (13.3) 27 (33.8)
.280
(64.7) (66.1) (100.0) (76.5) (85.7)
(57.1) (87.0) (64.6) (100.0) (53.3)
(66.7) (75.0) (66.7) (91.7)
4 (23.5) 1 (14.3)
7 (46.7)
(33.3) (25.0) (33.3) (8.3) (100.0)
PM&R
Vol.
-,
Iss.
-,
7
2014
Table 3. Continued Variable
Overall (n [ 99)
Completers at 42 Weeks (n [ 69) (%)
Noncompleters at 42 Weeks (n [ 30) (%)
46 53
32 (69.6) 37 (69.8)
14 (30.4) 16 (30.2)
Hemisensory Intact Partial
P Value* .979
Outlined are baseline characteristics for 99 subjects before the initiation of a 42-week trial of foot-drop stimulator use and those subjects who did (n ¼ 69) and did not (n ¼ 30) complete the final 42-week evaluation. At 30 weeks, 74 (75%) of subjects completed the assessment (data not shown). For both time periods, subjects who used their prestudy AFO and were white or Hispanic were more likely to complete the assessment (in all cases P < .02). AFO ¼ ankle foot orthosis; MCA ¼ middle cerebral artery; ACA ¼ anterior cerebral artery; PCA ¼ posterior cerebral artery; N/A ¼ data not available. *P values are from c2 or 2-sample t-test for categorical or continuous variables, respectively.
achieving the MCID of 0.1 m/s [37] or (2) advancing by at least 1 Perry Ambulation Category (PAC; 0.8 m/s) at any time before the designated assessment [40,41]. Baseline GS-C without a device was used as the comparison. The proportion of PAC and MCID responders was calculated for 30-week therapeutic effect and 42-week total effect. For the responder analyses, noncompleters were classified as nonresponders. Univariate correlations between FDS responder status and baseline demographic and clinical variables used the Spearman rank correlation coefficient. Significant variables were subsequently entered into a multivariate logistic regression (after backward selection) to predict FDS “responder” by definition (PAC and MCID) and effect type (30 weeks therapeutic and 42 weeks total). For all analyses, P < .05 was considered significant, except for choosing variables for the logistic regression, where P < .10 was used.
Data Management and Quality A secure Web-based electronic data capture system (Medidata Rave, New York, NY) was used for clinical data collection and management. Third-party monitors performed regular visits at each site to review and verify all study data in source documents.
RESULTS Sample Characteristics More than 1200 potential subjects were screened; 389 provided informed consent and completed further in-person evaluation. A total of 197 subjects were enrolled; 99 were randomized to the FDS arm and 98 to the AFO arm. The most common reasons for exclusion after consent were inadequate ankle motion (n ¼ 83), GS >0.8 m/s (n ¼ 23), and inability to confirm drop foot (n ¼ 19). Compared with the AFO group, the FDS group had more female subjects (c2, P ¼ .016) and patients who had sustained an ischemic stroke (c2, P ¼ .021). One hundred eighteen of 197 subjects (60%) required a new or modified AFO at study entry. Adaptive covariate randomization for the 99 FDS subjects
yielded the following distribution: 3-6 months poststroke and age 65 years or Medicare beneficiary, total n ¼ 4, subjects >65 years ¼ 2; >6 months poststroke and age 6 months poststroke and age 65 years or Medicare beneficiary, total n ¼ 77, subjects >65 years ¼ 44. Table 3 displays baseline demographic and clinical characteristics for these 99 subjects, comparing completers and noncompleters at 42 weeks. In general, our sample was relatively young, welleducated, racially diverse, and substantially impaired (mean GS-C 0.42 0.21 m/s [without an FDS] and LEFMA of 20.1 5.18 points) and were in the chronic phase of recovery, with a mean time poststroke of 4.8 5.3 years. Seventy-four of 99 subjects (75%) completed the 30-week evaluation and 69 of 99 (70%) completed the 42-week evaluation. For both long-term assessments, completers were more likely to be using their prestudy AFO and to be white or Hispanic (both cases P < .02, data not shown for 30 weeks). Subjects completing the 6-, 12-, and 36-week assessments numbered 94 (95%), 85 (86%), and 70 (71%), respectively. By 30 weeks, 8 of the 25 subjects (32%) who withdrew did so for reasons related to the FDS device, with no additional withdrawals related to the FDS device by 42 weeks (8/30, 27%). The group took a mean of 2092 and 2369 steps per day in weeks 6 and 24, respectively, reflecting an activity level within a broad range reported in persons who have had a stroke, but still well below the activity level of even sedentary, healthy persons [42].
Characterization of GS-C Change and Effect (Objectives 1 and 2) Figure 1 depicts the mean GS-C for assessments at baseline and 30 weeks without use of an FDS (in black) and assessments at baseline and at 6, 12, 30, 36, and 42 weeks with use of an FDS (in gray). An absolute increase in mean GS-C with use of an FDS is observed at each consecutive time point, with the linear mixed model analysis indicating a significant time effect (P < .0001 based on an F test; df ¼ 6; denominator (den) df ¼ 485; F value ¼ 5.45.) Significance levels for GS-C pairwise comparisons corresponding to
8
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Figure 1. Mean walking speed with and without use of a foot-drop stimulator (FDS) over time. The mean comfortable gait speed over multiple assessments is outlined. Gait speed at baseline and 30 weeks was measured both with an FDS (gray bar) and without an FDS (black bar). The arrows indicate the statistical significance of pairwise comparisons for baseline gait speed with and without use of an FDS to speeds after 6, 12, 30, 36, and 42 weeks (W) of device use. Values at the top of the figure indicate training effect and values at the bottom represent total (6W, 12W, 30W, 36W, and 42W) and therapeutic (30W) effects. Note that the comparison of gait speed with and without use of an FDS at baseline was not significant (NS).
Figure 1 are as follows: immediate device effect (P ¼ .062); training effect at 6 and 12 weeks (P ¼ not significant [NS]) and at 30, 36, and 42 weeks (all P < .03); total effect at 6, 12, 30, 36, and 42 weeks (all P < .005); and therapeutic effect at 30 weeks (P < .0001). A 28.6% increase in GS-C (Cohen’s effect size ¼ 0.75) was seen for 30-week therapeutic effect and a 45% improvement (Cohen’s effect size ¼ 0.84) was seen for 42 weeks’ total effect. For 30 weeks’ therapeutic effect, 18 of 99 (18%) were PAC responders and 81 were not (56/99 [57%] nonresponders and 25/99 [25%] noncompleters), whereas 29 of 99 (29%) were classified as MCID responders and 70 of 90 were not (45/99 [45%] nonresponders and 25/99 [25%] noncompleters). Regarding 42-week total effect, 54 of 99 (55%) were PAC responders and 45 of 99 were not (15/99 [15%] nonresponders and 30/99 [30%] noncompleters), whereas 67 of 99 (68%) were classified as MCID responders and 32 of 99 were not (2/99 [2%] nonresponders and 30/99 [30%] noncompleters). Figure 2 depicts the cumulative proportion of responders (total effect only) under both responder definitions. It also tracks the proportion of subjects achieving their maximum GS-C with an FDS at each assessment point. Approximately half of the sample achieved their maximum GS-C by 12
weeks, with 20% achieving that maximum on their very first use of the FDS (ie, baseline with the FDS). Forty percent of the group achieved their maximum after the 30-week assessment.
Univariate and Multivariate Outcome Prediction (Objective 3) Table 4 outlines univariate (variables having at least one correlation at P < .10) and multivariate correlations between baseline parameters and PAC and MCID responders for the 30-week therapeutic effect and 42-week total effect. For the 30-week therapeutic effect, the patterns of univariate correlates were quite similar between responder definitions. Clinical variables, with the notable exception of LEFMA (data not shown, P > .3 for both FDS responder definitions), held the strongest associations. A shorter time poststroke and better education were significant for MCID only. After logistic regression, younger age (odds ratio [OR] 0.94, 95% CI 0.89-0.99; P ¼ .0098) and faster baseline TUG (OR 0.94, 95% CI 0.89-0.99; P ¼ .0263) predicted PAC responders, whereas the Berg Balance Scale (OR 1.14, 95% CI 1.06-1.22; P ¼ .0007) and shorter time poststroke
PM&R
Vol.
-,
Iss.
-,
2014
9
Figure 2. Maximum gait speed (Max Speed) and proportion of minimal clinically important difference (MCID) and Perry Ambulation Category (PAC) responders over time: total effect only. The cumulative proportion of subjects who were classified as PAC and MCID responders for total effect at each assessment point is shown. Also tracked is the proportion of subjects who achieved their maximum gait speed at each time point. Missing data for maximum gait speed were handled by carrying the last-available value forward. Missing data for PAC and MCID were considered nonresponders (see text). w ¼ week; FDS ¼ foot-drop stimulator.
(OR 16.64, 95% CI 2.78-99.34; P ¼ .0021) predicted MCID responders. For 42 weeks’ total effect, univariate patterns differed somewhat between FDS responder definitions. Clinical measures, once again not including LEFMA (data not shown, P > .25 for both FDS responder definitions), held strong associations with both responder definitions. PAC responders were associated with female gender and receiving a new AFO prescription at randomization. MCID responders were associated with having a hemorrhagic stroke, a nonintact sensory examination, a cortical stroke, and a better education. After regression, younger age (OR 0.94, 95% CI 0.90-0.98; P ¼ .002) and faster baseline GS-C (OR 1.40, 95% CI 1.10-1.77; P ¼ .006) remained significant predictors of PAC responders, whereas better balance (OR 1.17, 95% CI 1.09-1.26; P < .0001), hemorrhagic stroke (OR 14.87, 95% CI 1.63-135.79; P ¼ .0167), and a nonintact sensory examination (OR 0.28, 95% CI 0.093-0.84; P ¼ .0224) predicted MCID responders.
AEs (Objective 4) Over 42 weeks, 18 serious AEs were reported among the FDS group, none of which was related to the device. A total of 160 AEs were related to the FDS in 59 of 99 subjects (59.5%). Of the 160 AEs, 92% were classified as mild, 8% as
moderate, and none as severe, and 154 of 160 (96%) were classified as “anticipated.” As expected, 50% of the AEs attributed to the FDS were related to reversible skin issues due to surface stimulation (eg, blister, rash, contact dermatitis, skin irritation, and excoriation). Over 42 weeks, 43 falls occurred in 34 subjects, with 24 falls (56%) related to the device (22 of the falls were classified as mild AEs and 2 were classified as moderate AEs).
DISCUSSION With an estimated incidence of 20% among survivors, as many as 1.4 million Americans who have sustained a stroke may be living with drop foot, and many have decreased gait speed as a result [5]. Strategies to increase the speed of ambulation, including the use of an FDS, may hold the potential to improve community integration and functional status and even decrease the risk of death among persons who have had a stroke [33-35]. The FASTEST trial is the largest study to date to compare changes in gait between users of an FDS and an AFO in persons who have sustained a stroke and have drop foot. Although both groups had significant gains at 30 weeks, no difference was found between groups [30]. Subjects who used the FDS had more AEs at 30 weeks, but the AEs were not serious, and the vast majority were both mild and anticipated. User satisfaction,
10
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 4. Univariate and multivariate analyses predicting foot drop stimulator responders (n ¼ 99) Success Defined by PAC Univariate Model Variable* Therapeutic effect at 30 wk Age GS-C GS-F 6-MWT LEFMA TUG BBS Education, HS graduate Time poststroke, 3-6 mo Total effect at 42 wk Age GS-C GS-F 6-MWT TUG BBS Gender, female Prestudy AFO, new Stroke location, cortical Stroke type, hemorrhagic Sensation loss, partial
Multivariate Model
Coeff Est
P Value
Coeff Est
P Value
OR (CI)
0.077 0.0426 0.297 0.012 0.030 0.074 0.120 0.539 1.584
.0013 .0046 .0013 .0004 .5636 .0099 .0032 .5709 .037
0.064
.0098
0.94 (0.89-0.99)
0.065
.0263
0.94 (0.89, 0.99)
1.073 1.393 1.284 1.009 1.040 1.103 1.901 1.758 1.594 0.633 0.243
.0004 .0004 .0002 .0008 .0027 .0003 .0411 .0904 .1727 .3045 .5729
0.068 0.335
.002 .006
0.94 (0.90-0.98) 1.40 (1.10-1.77)
Listed are variables having at least one univariate correlation with PAC or MCID responders status at P < .10 for total and therapeutic effects. Noncompleters are classified as nonresponders. Variables emerging as significant after multivariate logistic regression (with backward selection) have coefficient estimate, P value, and OR with 95% CIs listed to the right. PAC ¼ Perry Ambulation Category; MCID ¼ minimal clinically important difference; Coeff Est ¼ coefficient estimate; OR ¼ odds ratio; CI ¼ upper and lower bounds 95% confidence intervals; GS-C ¼ comfortable gait speed; GS-F ¼ fast gait speed; 6-MWT ¼ 6-minute walk test; LEFMA ¼ lower extremity Fugl-Meyer assessment; TUG ¼ Timed Up and Go; BBS ¼ Berg Balance Scale; HS ¼ high school; AFO ¼ ankle-foot orthosis. * Only variables with P < .10 are included in this table. Variables tested but not listed include ethnicity, marital status, handedness, side of stroke, and vascular distribution of stroke.
however, was greater statistically in the FDS group at 30 weeks compared with the AFO group. This current secondary analysis of the FDS arm provides a platform to explore the response to FDS devices during an extended period in persons who have sustained a stroke. Although the overall clinical benefit of an FDS may be similar to that of an AFO, candidate selection and the mechanism and timing of improvement may differ. With the use of a number of approaches, we confirmed that the benefits of an FDS on GS-C in persons who have had a stroke are clinically and statistically significant from baseline status. When using an FDS, a majority of subjects achieved an important clinical end point by 42 weeks. We confirmed substantial therapeutic and total effects for FDS use, with Cohen’s effect sizes of 0.75 and 0.84, respectively, which are well above the recently reported pooled effect size of 0.29 for physiotherapy mobility interventions in persons in the chronic phase of a stroke [43]. With a mean time poststroke of 4.8 years, our findings fuel continuing doubt regarding the concept of a functional “plateau” in persons in the chronic phase of a stroke [44].
Characteristics of FDS Response The characterization of how GS-C responds to use of an FDS over time in persons who have had a stroke fulfills our first and second objectives. Our subjects used an FDS as durable medical equipment on a daily basis, as opposed to an adjunct to exercise during PT sessions [14]. We first discuss the 3 orthotic effects. GS-C with use of FDS increased progressively by a total of 45% (0.42-0.61 m/s) over 42 weeks compared with baseline without use of the device (total effect, see Figure 1). This gain is somewhat better than other published data [15,16,23,25,27-29], given a wide variation in methodologies (see Table 1). Two other factors might have accounted for at least some of this long-term improvement. First, many persons who have had a stroke are deconditioned [44] and socially isolated [45]. Given the travel, staff interaction, and effort required for a clinical trial, simply participating might have enhanced their physical mobility and contributed to gains. Second, natural recovery might have contributed but should be minimal given a mean time poststroke of 4.8 years, with only 10% of the sample having a poststroke time of
Response and Prediction of Improvement in Gait Speed From Functional Electrical Stimulation in Persons With Poststroke Drop Foot Michael W. O’Dell, MD, Kari Dunning, PhD, Patricia Kluding, PhD, Samuel S. Wu, PhD, Jody Feld, DPT, Jivan Ginosian, MS, Keith McBride, DPT Objective: To describe changes in and predictors of comfortable gait speed (GS-C) after using a foot-drop stimulator (FDS; Bioness L300; Bioness Inc, Valencia, CA) for 42 weeks in persons who had sustained a stroke. Design: Secondary analysis of prospective assessments. Setting: Multicenter clinical trial. Participants: A total of 99 subjects who had sustained a stroke 3 months earlier and who had GS-C 0.8 m/s and drop foot with a mean age of 60.7 years and a poststroke time of 4.8 years. Methods: GS-C was assessed at baseline and at 30 weeks with and without use of an FDS (therapeutic effect) and at 6, 12, 30, 36, and 42 weeks with use of an FDS (total effect). After subjects participated in 8 physical therapy sessions, an FDS was used for ambulation over the course of 42 weeks. Main Outcome Measurements: Changes in mean GS-C over time, FDS “responder” status defined as either 0.1 m/s gain in GS-C (the minimal clinically important difference [MCID]) or advancing by one Perry Ambulation Category (PAC), and the incidence and nature of adverse events (AEs). Results: A total of 74 (75%) and 69 (70%) of 99 subjects completed assessments at 30 weeks and 42 weeks, respectively. Baseline GS-C was 0.42 m/s without use of an FDS and 0.49 m/s with use of an FDS. GS-C improved to 0.54 m/s at 30 weeks without use of an FDS (effect size ¼ 0.75) and 0.54, 0.55, 0.58, 0.60, and 0.61 m/s at 6, 12, 30, 36, and 42 weeks with use of an FDS, respectively (effect size 0.84 at 42 weeks). Half of the subjects achieved a maximum GS-C by 12 weeks. Approximately 18% were PAC responders and 29% were MCID responders for 30-week therapeutic effect, and 55% were PAC responders and 67% were MCID responders for 42-week total effect. After logistic regression, the following factors emerged as the strongest predictors of FDS responders: younger age, faster baseline GS-C and Timed Up and Go, and balance. At 42 weeks, 60% reported a device-related AE; 92% were mild and 96% were anticipated. Conclusions: When an FDS was used, GS-C improved progressively over 42 weeks, with 50% of patients achieving a clinically meaningful 42-week total effect and 50% achieving a maximum GS-C by 12 weeks. Younger patients with greater mobility levels may benefit most from use of an FDS. AEs were frequent, mild, and reversible. PM R 2014;-:1-15
INTRODUCTION Seven million Americans are living with the effects of a stroke, and nearly 800,000 new and recurrent cases of stroke are reported annually [1]. Impairments resulting from stroke include language and cognitive deficits, visual changes, and contralateral hemiplegia, among many others [2,3]. Sustained distal weakness in the hemiparetic leg is manifested by an inability to adequately dorsiflex the foot during the swing phase of gait and can impede mobility profoundly [4]. This condition, known as “drop foot,” occurs in up to 20% of persons who have had a stroke [5]. Drop foot is seen with either weakness of the anterior PM&R 1934-1482/14/$36.00 Printed in U.S.A.
M.W.O. Department of Rehabilitation Medicine, New YorkePresbyterian Hospital/Weill Cornell Medical Center, Box 142, Room Baker F1600, 525 East 68th ST, New York, NY 10065. Address correspondence to: M.W.O.; e-mail: [email protected] Disclosure: nothing to disclose K.D. Department of Rehabilitation Sciences, College of Allied Health Sciences, Division of Biostatistics and Epidemiology, College of Medicine, University of Cincinnati, Cincinnati, OH Disclosure: nothing to disclose P.K. Department of Physical Therapy and Rehabilitation Sciences, University of Kansas Medical Center, Kansas City, KS Disclosure: nothing to disclose S.S.W. Department of Biostatistics, University of Florida, Gainesville, FL Disclosure: paid consultancy, Bioness Inc J.F. Department of Community and Family Medicine, Duke University School of Medicine, Durham, NC Disclosure: employment, Bioness Inc J.G. Department of Clinical Research, Bioness Inc, Valencia, CA Disclosure: employment, Bioness Inc K.M. Department of Marketing, Bioness Inc, Valencia, CA Disclosure: employment, Bioness Inc Research support: This clinical trial was funded by Bioness Inc, Valencia, CA. This study is registered under clinical trials registration number NCT01138995. Submitted for publication June 19, 2013; accepted January 1, 2014.
ª 2014 by the American Academy of Physical Medicine and Rehabilitation Vol. -, 1-15, --- 2014 http://dx.doi.org/10.1016/j.pmrj.2014.01.001
1
2
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 1. Published studies of surface foot-drop stimulation and gait speed in persons with chronic stroke (n > 10) Study, Year (Reference) Current analyses
Total N (Stroke, n)
F (%)
Mean Age, y
Mean Time Poststroke, y
Mean Baseline GS (no FDS), m/s
F/U Time, wk
F/U, %
IE*
99
48
60.7
4.8
0.42
30
75%
42
70%
0.07 17% 0.07 17%
Everaert et al, 2013 (23)
38
33
57.1
0.53
0.52
6
88%
0.07 15% 0.08 16% NR
Taylor et al, 2013 (22)
62
47
59.6
4.8
0.49
>14.3y
90%
van Swigchem et al, 2010 (24)
26
19
52.8
3.2
1.02 (with AFO)
8
96%
Stein et al, 2010 (25)
73 (26)
46
58.8
7.5
0.64
44
32%
0.05 6.5%
Sabut et al, 2010 (28)
16
20
49.5
1.5
0.37
12
100%
NR
Hausdorff and Ring, 2008 (14)
24 (21)
17
54.0
5.8
0.53
8
100%
Laufer et al, 2009 (26)
16 (13)
7
55.0
5.3
0.62
52
67%
0.09 17% NR
Stein et al, 2006 (16)
26 (12)
NR
57
NR
0.69
12
100%
Taylor et al, 1999 (27)
140 (111)
NR
55.4
5.4
0.57
18
93%
Burridge et al, 1997 (15)
16
38
52.3
3.6
0.64
12
100%
Granat et al, 1996 (29)
19
16
56
7
0.87
11
84%
0.02x 3% 0.07 12% 0.04 6% NR
F ¼ female; GS ¼ gait speed; FDS ¼ foot-drop stimulator; F/U ¼ follow-up; IE ¼ immediate effect; TrE ¼ training effect; ToE ¼ total effect; ThE ¼ therapeutic effect; AE ¼ adverse event; NR ¼ not reported; TUG ¼ Timed Up and Go; BBS ¼ Berg Balance Scale; GS-C ¼ comfortable gait speed; 10-MWT ¼ 10-meter walk test; ODFS ¼ Odstock Dropped Foot Stimulator; PAC ¼ Perry Ambulation Category; MCID ¼ minimal clinically important difference; AFO ¼ ankle foot orthosis; TBI ¼ traumatic brain injury; 6-MWT ¼ 6-meter walk test; 6/10-MWT ¼ 6- to 10-meter walk test; PCI ¼ physiologic cost index. * Numbers are gains in gait speed in m/s; % is percent gain in gait speed from baseline as defined in text. y Average of all assessments taken after 100 days. z Assuming no FDS at baseline (not stated in study). x Estimated or calculated value for this table.
muscles, spasticity of the posterior muscles of the leg, or both. Subsequent development of plantar-flexion contracture at the ankle further complicates limb clearance. Drop foot contributes to undesirable compensatory movement patterns, slowed gait speed, limited functional mobility, and an increased risk of falls [6-8]. A number of treatment strategies for drop foot may be used depending on the underlying cause. Contracture at the ankle joint in a person with good ambulation potential may be managed surgically [9]. Posterior compartment spasticity is treated with exercise and a combination of oral, injection, or intrathecal medications [10]. The most common treatment is an ankle-foot orthosis (AFO), which is a brace that holds the ankle in a neutral position and improves limb clearance during the swing phase of gait. AFOs have significant drawbacks, however, including limitation of ankle mobility, which contributes to contracture development [11], possible undermining of motor recovery [12], difficulty
in standing from a seated position [6], and poor acceptance by the wearer as a result of discomfort and the perception of undesirable aesthetics [13]. Another strategy to combat drop foot is the use of functional electrical stimulation (FES). FES is used to stimulate branches of the common peroneal nerve just below the knee, thus activating dorsiflexion and eversion of the foot. A footdrop stimulator (FDS) is a commercially available FES system designed specifically for use during ambulation. These devices synchronize surface stimulation of the nerve with the swing phase of the gait through the use of either pressure sensors or heel switches placed in the shoe [14,15] or a tilt sensor [16]. The potential effects of FDS devices have been described previously by Dunning et al [17]. Gait speed has been studied most often, but any relevant parameter could be used. An immediate device effect refers to changes that occur at once (ie, within minutes) when “FDS off” is compared to “FDS on.” A training effect above and beyond the
PM&R
Vol.
-,
Iss.
-,
2014
3
Table 1. Continued Device Type, Skin-Related AE, GS Determination Method for Reported Data, Other
TrE*
ToE*
ThE*
Predictors of GS
0.09 18% 0.12 25%
0.16 38% 0.19 45%
0.12 29% NR
0.10 19% NR
0.17 37% 0.18 45% NR
0.12 26% 0.11 24% NR
ThE at 30 wk related to age, TUG, BBS, time poststroke ToE at 42 wk related to age, GS-C, BBS, hemorrhagic stroke, partial sensory loss NR
0.25 32%
0.19 25%
NR
0.10z 27% 0.18 34% 0.29 47% 0.08 12% 0.15 26% 0.13 20% 0.06 7%
NR
NR
WalkAide, AE not reported, 10-MWT, demographics for stroke only, FDS effects includes 15 nonstroke subjects, F/U % ¼ 13/41 nonprogressive subjects EMS (CyberMedic Corp), AE not reported, 10-MWT
NR
NR
L300, no AE observed, 6-MWT results included 3 TBI subjects
0.17 27% 0.03x 4% 0.08 14% 0.01 2% 0.07 8%
NR
L300, AE not reported, 6-MWT, extension of Hausdorff and Ring’s study WalkAide, skin irritation in “a few” subjects, 10-MWT, some data include nonstrokes ODFS, AE not reported, 10-MWT, data are stroke-specific, long-term F/U reported previously ODFS, AE not reported, 10-MWT
0.05 5% 0.20 24% NR 0.09 15% NR 0.06x 9% 0.08 13% 0.09 13% NR
NR NR
ToE not related to age, duration of drop foot ToE not related to time poststroke, ThE related to right hemiparesis ToE related to greater baseline PCI but not to initial gait speed NR
immediate device effect may occur as a person uses the FDS over weeks to months. The total effect is the sum of immediate and training effects. The immediate, training, and total effects reflect an orthotic benefit of FDS. Finally, a therapeutic effect represents a benefit that persists even without FDS and is sometimes termed a “carry-over effect” [15]. The underlying mechanism for the therapeutic and training effects probably results from neuroplastic changes, improved peripheral strength, and cardiopulmonary conditioning, as well as changes in other systems [17,18]. A few meta-analyses [19-21] and several FDS studies [14-16,22-30] have demonstrated improved gait parameters with use of an FDS in persons who have had a stroke. Table 1 compares a number of parameters and gait speed among published surface FDS studies, including 10 subjects who had sustained a stroke. A recent study that examined therapeutic FDS effect was not included because the authors used a different methodology to determine gait
L300, 48.5% of subjects with skin irritation, 10-MWT
WalkAide, 10% skin irritation (for all FDS users), 10-MWT ODFS, 22% mild skin irritation, long-term F/U of Taylor et al 1999 (27), 52% PAC and MCID responders at 16.5 m, 10-MWT L300, 15% skin irritation, 10-MWT, no measurements without FDS
FES Medical Electronics, AE not reported, 6/10-MWT
speed [31]. Therapeutic and total effects are frequently demonstrated with a wide array of research designs. It is important to consider that, beyond their use as an adjunct treatment during physical therapy (PT) sessions, modern FDS systems function like durable medical equipment meant for long-term use at home and in the community. Despite this long-term use, inspection of Table 1 reveals few studies in which investigators comment on the trajectory of or factors predicting clinical response over time; this information could help clinicians identify appropriate candidates for an FDS device. The Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy (FASTEST) study is a large, multicenter, randomized, controlled, single-blinded clinical trial conducted from 2010 to 2013 in which investigators compared the impact of an FDS versus an AFO on gait speed and other parameters in persons who had sustained a stroke and had drop foot. Details of the FASTEST methodology
4
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
[17] and 30 weeks’ comparison between devices [30] have been previously published. Of the 197 subjects entered in the study, 99 were randomized to the FDS arm and underwent extensive clinical assessment at baseline and 30 weeks with and without the device to assess therapeutic effect and 5 times with the device (at 6, 12, 30, 36, and 42 weeks) to assess the immediate device, training, and total effects. Because the original FASTEST analysis did not address the trajectory or prediction of response to the FDS, isolated consideration of these 99 subjects provides an opportunity to examine the clinical response to an FDS in persons who have sustained a stroke. Therefore the objectives of this secondary analysis of FASTEST data are to (1) characterize changes in comfortable gait speed (GS-C) over time with use of an FDS; (2) calculate the immediate device, training, total, and therapeutic effects of an FDS; (3) determine variables that predict FDS “responders” versus “nonresponders” for a 30-week therapeutic effect and 42-week total effect; and (4) assess adverse events (AEs) over 42 weeks of use.
METHODS
persons identified as having subacute (3-6 months) and chronic (>6 months) stroke who demonstrated drop foot as a significant limitation in gait. The drop foot had to be severe enough to require an AFO and, if necessary, a new or modified AFO was provided by the sponsor to ensure safety during the first weeks of FDS training. Subjects walked slowly (GS-C 0.8 m/s) but required the assistance of no more than one person to walk 10 m with or without an assistive device. In addition to requiring an AFO, all subjects were required to demonstrate adequate and tolerable peroneal nerve stimulation with the FDS unit before randomization. After their eligibility was confirmed, subjects were randomized to the AFO or FDS groups by the sponsor with the use of covariate adaptive randomization to ensure target allocations for age and time poststroke in 4 subgroups as follows: 3-6 months poststroke and age 65 years or Medicare beneficiary (target N ¼ 7); >6 months poststroke and age 6 months poststroke and age 65 years or Medicare beneficiary (target N ¼ 150).
Subjects and Randomization
Interventions and Assessments
A detailed description of the FASTEST trial design and methods has been previously published [17]. Participants were recruited at 11 sites throughout the United States, with each site obtaining institutional review board approval and informed consent before performing any assessment. Inclusion and exclusion criteria are outlined in Table 2. Salient to the current analyses, the FASTEST trial included
Hereafter, we describe interventions related only to the 99 subjects randomized to the FDS arm. During the first 6 weeks of the study, subjects participated in 8 sessions of physical therapy (a total time of 225-285 minutes). Two sessions per week were provided during the first 2 weeks and one session per week was provided during the next 4 weeks. The first visit focused on device fitting and settings;
Table 2. Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy study inclusion and exclusion criteria Inclusion Criteria
Exclusion Criteria
Ankle dorsiflexion neutral when stimulated with FDS in sitting and standing positions Adequate ankle and knee stability during gait 3 mo from at least 1 stroke of any etiology Foot drop caused by stroke sufficient to require use of an AFO (as assessed by the site’s physical therapist and orthotist) Adequate cognition and communication abilities (24/30 on the Mini Mental State Examination) or having a competent caregiver 18 years Able to safely walk at least 10 m with a maximum assist of 1 person Comfortable 10-m gait speed 0.80 m/s Medically stable
Fixed plantar flexion contracture 5 Adequate response to FDS Pain 4/10 in the affected leg Any other interventional research studies Demand-type cardiac pacemaker, defibrillator, electrical, or metallic implants Significant swelling/edema in the leg extending up to the knee Chronic skin problems or lesions near FDS stimulation site Pregnant/planning to be pregnant 6 weeks from botulinum toxin to weak leg or arm, or planned during the course of the study Expectation of change in medications for spasticity Unstable seizure disorder Pre-existing orthopedic conditions affecting ambulation Complete vs hemisensory loss in leg FDS or FES use >3 hours in 6 mo before enrollment Major depression (PHQ-9 >10) not managed by a health care provider Participating or planning physical or occupational therapy
Outlined are the inclusion and exclusion criteria of the Functional Ambulation: Standard Treatment Versus Electrical Stimulation Therapy trial. All 99 of the subjects in the current study met these criteria. FDS ¼ foot-drop stimulator; AFO ¼ ankle foot orthosis; FES ¼ functional electrical stimulation; PHQ-9 ¼ Patient Health Questionnaire, 9-item version.
PM&R
during visits 2-4, device education, initial gait training, and a home exercise program were provided; and the remaining visits focused on gait training with the FDS. The specific FDS device used in this study was the Bioness L300 Foot Drop System (Bioness Inc, Valencia, CA), which comprises a stimulation cuff with electrodes affixed on the anterolateral aspect positioned below the knee, a control unit, and an inshoe pressure sensor under the foot. The stimulation unit initially is configured by a clinician, who uses a handheld, wireless computer interface. Gait and stimulation parameters (amplitude 0-100 mA, duration 100-300 microseconds, frequency 20-45 pps, and gait parameters, including ramp up/down 0-2 seconds) are individualized to achieve a functional, comfortable, and timely muscle contraction. The pressure sensor detects “heel off” at the start of swing phase and wirelessly initiates peroneal nerve stimulation, facilitating dorsiflexion and eversion of the ankle. When the foot makes “initial contact” at the start of stance phase, the stimulation is paused [17]. Standardized protocols developed by the sponsor, who has more than 5 years of market experience, were used by all sites for initial fitting of the FDS, gait training, the wearing schedule, the conditioning program, and device education. Home-exercise programs were individualized and focused on gait training with the FDS device. During the first 3 weeks of use, subjects gradually increased the time spent walking with the FDS from 15 minutes per day to all day. In addition, the cyclic training function (5 seconds on and 8 seconds off) of the device was used 15 minutes twice a day for the first week and 20 minutes twice a day for an additional 2 weeks to help condition the muscles stimulated during ambulation. An AFO initially was used as needed to ensure safe ambulation when the FDS was not being used. After the initial conditioning phase, subjects were to use the FDS exclusively for ambulation. At the time of the trial, the manufacturer provided hydrogel electrodes as the primary medium for transcutaneous conduction. Written skin care guidelines were reviewed and issued during the initial fitting and reviewed throughout the training period. Device compliance (number of steps per day) was estimated with use of a step activity monitor (StepWatch; Orthocare Innovations, Oklahoma City, OK) in weeks 6 and 24.
Outcomes Outcomes were obtained without the FDS at baseline and at 30 weeks to examine therapeutic effect. Outcome measures were also obtained with the FDS at baseline and at 6, 12, 30, 36, and 42 weeks to examine immediate device, training, and total effects. Routine, scheduled follow-ups were completed at weeks 16, 20, and 24. Outcome testing was performed by a physical therapist blinded to group assignment at baseline, 6, 12, and 30 weeks and not blinded at 36 and 42 weeks. To maintain blinding, a shroud of vinyl fabric was secured over the lower leg and shoe to conceal the
Vol.
-,
Iss.
-,
2014
5
FDS device. Therapists received training from the sponsor and passed an on-site competency test for consistency in assessing outcomes. The primary outcome in the FASTEST trial and the dependent variable for this secondary analysis is GS-C as assessed with the 10-m walk test [32-35]. Independent predictor variables included baseline GS-C, demographic and stroke characteristics, and clinical assessments encompassing primarily impairment and activity limitation as defined by the International Classification of Function [36]. Clinical assessments included the lower extremity FuglMeyer Assessment (LEFMA), Timed Up and Go (TUG), endurance via the 6-minute walk test, gait speed “as fast as possible,” and the Berg Balance Scale. All outcome measures are valid and reliable in persons who have sustained a stroke and have previously been described in detail [17]. The occurrence and types of AEs were assessed at all study visits. When an AE occurred, study personnel assigned a severity and determined whether it was caused by the device. The AE was also designated as an “anticipated AE” if it was included in the initial Bioness L300 filing with the Food and Drug Administration device approval process. The incidence of falls and injury were self-reported by participants and/or their caregivers retrospectively for the 6 months before baseline and prospectively at each study visit.
Statistical Analysis The FASTEST trial established a recruitment target of 206 subjects that was based on an 80% power to detect a minimal clinically important difference (MCID) of 0.1 m/s between the FDS and AFO groups [37] with the use of a 2-sample t-test with a 2-sided 0.05 alpha level. Other considerations determining the sample size have been discussed previously [30]. To assess potential bias from missing data, paired t-test, Mann-Whitney U test, or c2 tests were used to explore baseline variables associated with subjects who were “completers” and “noncompleters” at the 30- and 42week assessments. For objectives 1 and 2, descriptive statistics (ie, mean, median, range, standard deviation, and 95% confidence intervals [CIs]) were used to characterize changes in GS-C over 42 weeks of FDS use. The proportion of subjects achieving their maximum GS-C with an FDS at each assessment time also was noted. Paired t-tests were used for pairwise comparisons of GS-C between baseline and selected follow-up assessments to assess FDS orthotic and therapeutic effects. Cohen’s effect size [38] also was calculated for selected comparisons. A linear mixed model [39] was used to assess statistically significant changes in GS-C with use of an FDS among consecutive assessments, baseline through 42 weeks. For objectives 1 and 2, only data for completers were analyzed. Descriptive statistics only were used to address AEs in objective 4. To address objective 3, dichotomous categories of FDS “responder” and “nonresponder” were defined by either (1)
6
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 3. Demographic and baseline characteristics for 99 subjects randomized to changes in a foot-drop stimulator Variable
Overall (n [ 99)
Completers at 42 Weeks (n [ 69) (%)
Noncompleters at 42 Weeks (n [ 30) (%)
P Value*
Age, y Comfortable gait speed, m/s (m SD) Fast gait speed, m/s (m SD) 6-minute walk test, feet (m SD) Lower extremity Fugl-Meyer assessment (m SD) Timed Up and Go, s (m SD) Berg Balance Scale (m SD) Gender Female Male Prestudy AFO New prescription Prestudy prescription Ethnicity White Black Asian Hispanic Other Marital status Divorced Married Other Single Widowed Handed Left Right Education Did not complete high school High school graduate College/trade school Postgraduate school Graduate school Time poststroke, mo 3-6 6 Stroke side Left Right Vascular distribution MCA ACA PCA Basal ganglia Cerebellum Brainstem Unknown Other Stroke location N/A Cortical Subcortical Stroke type N/A Hemorrhagic Infarction
60.71 12.24 0.42 0.21
60.00 12.29 0.44 0.22
62.33 12.17 0.38 0.17
.386 .204
0.56 0.32 152.07 90.04 20.14 5.18
0.59 0.34 159.36 96.02 20.23 5.16
0.50 0.25 135.30 73.24 19.93 5.32
.223 .224 .794
34.34 27.73 40.31 9.00
33.53 28.05 40.55 9.11
36.21 27.35 39.77 8.86
.661 .692
48 51
32 (66.7) 37 (72.5)
16 (33.3) 14 (27.5)
.525
43 56
24 (55.8) 45 (80.4)
19 (44.2) 11 (19.6)
.008
54 26 5 10 4
41 (75.9) 14 (53.8)
13 (24.1) 12 (46.2) 5 (100.0)
.0002
10 (100.0) 4 (100.0)
17 56 2 17 7
11 37 2 13 6
6 (35.3) 19 (33.9)
.620
9 90
7 (77.8) 62 (68.9)
2 (22.2) 28 (31.1)
.580
7 23 48 6 15
4 20 31 6 8
3 (42.9) 3 (13.0) 17 (35.4)
.064
9 90
8 (88.9) 61 (67.8)
1 (11.1) 29 (32.2)
.189
46 53
36 (78.3) 33 (62.3)
10 (21.7) 20 (37.7)
.084
45 4 3 12 1 3 18 13
30 3 2 11
15 1 1 1 1
.405
3 (100.0) 11 (61.1) 9 (69.2)
7 (38.9) 4 (30.8)
12 48 39
8 (66.7) 33 (68.8) 28 (71.8)
4 (33.3) 15 (31.3) 11 (28.2)
.926
4 15 80
3 (75.0) 13 (86.7) 53 (66.3)
1 (25.0) 2 (13.3) 27 (33.8)
.280
(64.7) (66.1) (100.0) (76.5) (85.7)
(57.1) (87.0) (64.6) (100.0) (53.3)
(66.7) (75.0) (66.7) (91.7)
4 (23.5) 1 (14.3)
7 (46.7)
(33.3) (25.0) (33.3) (8.3) (100.0)
PM&R
Vol.
-,
Iss.
-,
7
2014
Table 3. Continued Variable
Overall (n [ 99)
Completers at 42 Weeks (n [ 69) (%)
Noncompleters at 42 Weeks (n [ 30) (%)
46 53
32 (69.6) 37 (69.8)
14 (30.4) 16 (30.2)
Hemisensory Intact Partial
P Value* .979
Outlined are baseline characteristics for 99 subjects before the initiation of a 42-week trial of foot-drop stimulator use and those subjects who did (n ¼ 69) and did not (n ¼ 30) complete the final 42-week evaluation. At 30 weeks, 74 (75%) of subjects completed the assessment (data not shown). For both time periods, subjects who used their prestudy AFO and were white or Hispanic were more likely to complete the assessment (in all cases P < .02). AFO ¼ ankle foot orthosis; MCA ¼ middle cerebral artery; ACA ¼ anterior cerebral artery; PCA ¼ posterior cerebral artery; N/A ¼ data not available. *P values are from c2 or 2-sample t-test for categorical or continuous variables, respectively.
achieving the MCID of 0.1 m/s [37] or (2) advancing by at least 1 Perry Ambulation Category (PAC; 0.8 m/s) at any time before the designated assessment [40,41]. Baseline GS-C without a device was used as the comparison. The proportion of PAC and MCID responders was calculated for 30-week therapeutic effect and 42-week total effect. For the responder analyses, noncompleters were classified as nonresponders. Univariate correlations between FDS responder status and baseline demographic and clinical variables used the Spearman rank correlation coefficient. Significant variables were subsequently entered into a multivariate logistic regression (after backward selection) to predict FDS “responder” by definition (PAC and MCID) and effect type (30 weeks therapeutic and 42 weeks total). For all analyses, P < .05 was considered significant, except for choosing variables for the logistic regression, where P < .10 was used.
Data Management and Quality A secure Web-based electronic data capture system (Medidata Rave, New York, NY) was used for clinical data collection and management. Third-party monitors performed regular visits at each site to review and verify all study data in source documents.
RESULTS Sample Characteristics More than 1200 potential subjects were screened; 389 provided informed consent and completed further in-person evaluation. A total of 197 subjects were enrolled; 99 were randomized to the FDS arm and 98 to the AFO arm. The most common reasons for exclusion after consent were inadequate ankle motion (n ¼ 83), GS >0.8 m/s (n ¼ 23), and inability to confirm drop foot (n ¼ 19). Compared with the AFO group, the FDS group had more female subjects (c2, P ¼ .016) and patients who had sustained an ischemic stroke (c2, P ¼ .021). One hundred eighteen of 197 subjects (60%) required a new or modified AFO at study entry. Adaptive covariate randomization for the 99 FDS subjects
yielded the following distribution: 3-6 months poststroke and age 65 years or Medicare beneficiary, total n ¼ 4, subjects >65 years ¼ 2; >6 months poststroke and age 6 months poststroke and age 65 years or Medicare beneficiary, total n ¼ 77, subjects >65 years ¼ 44. Table 3 displays baseline demographic and clinical characteristics for these 99 subjects, comparing completers and noncompleters at 42 weeks. In general, our sample was relatively young, welleducated, racially diverse, and substantially impaired (mean GS-C 0.42 0.21 m/s [without an FDS] and LEFMA of 20.1 5.18 points) and were in the chronic phase of recovery, with a mean time poststroke of 4.8 5.3 years. Seventy-four of 99 subjects (75%) completed the 30-week evaluation and 69 of 99 (70%) completed the 42-week evaluation. For both long-term assessments, completers were more likely to be using their prestudy AFO and to be white or Hispanic (both cases P < .02, data not shown for 30 weeks). Subjects completing the 6-, 12-, and 36-week assessments numbered 94 (95%), 85 (86%), and 70 (71%), respectively. By 30 weeks, 8 of the 25 subjects (32%) who withdrew did so for reasons related to the FDS device, with no additional withdrawals related to the FDS device by 42 weeks (8/30, 27%). The group took a mean of 2092 and 2369 steps per day in weeks 6 and 24, respectively, reflecting an activity level within a broad range reported in persons who have had a stroke, but still well below the activity level of even sedentary, healthy persons [42].
Characterization of GS-C Change and Effect (Objectives 1 and 2) Figure 1 depicts the mean GS-C for assessments at baseline and 30 weeks without use of an FDS (in black) and assessments at baseline and at 6, 12, 30, 36, and 42 weeks with use of an FDS (in gray). An absolute increase in mean GS-C with use of an FDS is observed at each consecutive time point, with the linear mixed model analysis indicating a significant time effect (P < .0001 based on an F test; df ¼ 6; denominator (den) df ¼ 485; F value ¼ 5.45.) Significance levels for GS-C pairwise comparisons corresponding to
8
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Figure 1. Mean walking speed with and without use of a foot-drop stimulator (FDS) over time. The mean comfortable gait speed over multiple assessments is outlined. Gait speed at baseline and 30 weeks was measured both with an FDS (gray bar) and without an FDS (black bar). The arrows indicate the statistical significance of pairwise comparisons for baseline gait speed with and without use of an FDS to speeds after 6, 12, 30, 36, and 42 weeks (W) of device use. Values at the top of the figure indicate training effect and values at the bottom represent total (6W, 12W, 30W, 36W, and 42W) and therapeutic (30W) effects. Note that the comparison of gait speed with and without use of an FDS at baseline was not significant (NS).
Figure 1 are as follows: immediate device effect (P ¼ .062); training effect at 6 and 12 weeks (P ¼ not significant [NS]) and at 30, 36, and 42 weeks (all P < .03); total effect at 6, 12, 30, 36, and 42 weeks (all P < .005); and therapeutic effect at 30 weeks (P < .0001). A 28.6% increase in GS-C (Cohen’s effect size ¼ 0.75) was seen for 30-week therapeutic effect and a 45% improvement (Cohen’s effect size ¼ 0.84) was seen for 42 weeks’ total effect. For 30 weeks’ therapeutic effect, 18 of 99 (18%) were PAC responders and 81 were not (56/99 [57%] nonresponders and 25/99 [25%] noncompleters), whereas 29 of 99 (29%) were classified as MCID responders and 70 of 90 were not (45/99 [45%] nonresponders and 25/99 [25%] noncompleters). Regarding 42-week total effect, 54 of 99 (55%) were PAC responders and 45 of 99 were not (15/99 [15%] nonresponders and 30/99 [30%] noncompleters), whereas 67 of 99 (68%) were classified as MCID responders and 32 of 99 were not (2/99 [2%] nonresponders and 30/99 [30%] noncompleters). Figure 2 depicts the cumulative proportion of responders (total effect only) under both responder definitions. It also tracks the proportion of subjects achieving their maximum GS-C with an FDS at each assessment point. Approximately half of the sample achieved their maximum GS-C by 12
weeks, with 20% achieving that maximum on their very first use of the FDS (ie, baseline with the FDS). Forty percent of the group achieved their maximum after the 30-week assessment.
Univariate and Multivariate Outcome Prediction (Objective 3) Table 4 outlines univariate (variables having at least one correlation at P < .10) and multivariate correlations between baseline parameters and PAC and MCID responders for the 30-week therapeutic effect and 42-week total effect. For the 30-week therapeutic effect, the patterns of univariate correlates were quite similar between responder definitions. Clinical variables, with the notable exception of LEFMA (data not shown, P > .3 for both FDS responder definitions), held the strongest associations. A shorter time poststroke and better education were significant for MCID only. After logistic regression, younger age (odds ratio [OR] 0.94, 95% CI 0.89-0.99; P ¼ .0098) and faster baseline TUG (OR 0.94, 95% CI 0.89-0.99; P ¼ .0263) predicted PAC responders, whereas the Berg Balance Scale (OR 1.14, 95% CI 1.06-1.22; P ¼ .0007) and shorter time poststroke
PM&R
Vol.
-,
Iss.
-,
2014
9
Figure 2. Maximum gait speed (Max Speed) and proportion of minimal clinically important difference (MCID) and Perry Ambulation Category (PAC) responders over time: total effect only. The cumulative proportion of subjects who were classified as PAC and MCID responders for total effect at each assessment point is shown. Also tracked is the proportion of subjects who achieved their maximum gait speed at each time point. Missing data for maximum gait speed were handled by carrying the last-available value forward. Missing data for PAC and MCID were considered nonresponders (see text). w ¼ week; FDS ¼ foot-drop stimulator.
(OR 16.64, 95% CI 2.78-99.34; P ¼ .0021) predicted MCID responders. For 42 weeks’ total effect, univariate patterns differed somewhat between FDS responder definitions. Clinical measures, once again not including LEFMA (data not shown, P > .25 for both FDS responder definitions), held strong associations with both responder definitions. PAC responders were associated with female gender and receiving a new AFO prescription at randomization. MCID responders were associated with having a hemorrhagic stroke, a nonintact sensory examination, a cortical stroke, and a better education. After regression, younger age (OR 0.94, 95% CI 0.90-0.98; P ¼ .002) and faster baseline GS-C (OR 1.40, 95% CI 1.10-1.77; P ¼ .006) remained significant predictors of PAC responders, whereas better balance (OR 1.17, 95% CI 1.09-1.26; P < .0001), hemorrhagic stroke (OR 14.87, 95% CI 1.63-135.79; P ¼ .0167), and a nonintact sensory examination (OR 0.28, 95% CI 0.093-0.84; P ¼ .0224) predicted MCID responders.
AEs (Objective 4) Over 42 weeks, 18 serious AEs were reported among the FDS group, none of which was related to the device. A total of 160 AEs were related to the FDS in 59 of 99 subjects (59.5%). Of the 160 AEs, 92% were classified as mild, 8% as
moderate, and none as severe, and 154 of 160 (96%) were classified as “anticipated.” As expected, 50% of the AEs attributed to the FDS were related to reversible skin issues due to surface stimulation (eg, blister, rash, contact dermatitis, skin irritation, and excoriation). Over 42 weeks, 43 falls occurred in 34 subjects, with 24 falls (56%) related to the device (22 of the falls were classified as mild AEs and 2 were classified as moderate AEs).
DISCUSSION With an estimated incidence of 20% among survivors, as many as 1.4 million Americans who have sustained a stroke may be living with drop foot, and many have decreased gait speed as a result [5]. Strategies to increase the speed of ambulation, including the use of an FDS, may hold the potential to improve community integration and functional status and even decrease the risk of death among persons who have had a stroke [33-35]. The FASTEST trial is the largest study to date to compare changes in gait between users of an FDS and an AFO in persons who have sustained a stroke and have drop foot. Although both groups had significant gains at 30 weeks, no difference was found between groups [30]. Subjects who used the FDS had more AEs at 30 weeks, but the AEs were not serious, and the vast majority were both mild and anticipated. User satisfaction,
10
O’Dell et al
GAIT SPEED AND FES IN POSTSTROKE DROP FOOT
Table 4. Univariate and multivariate analyses predicting foot drop stimulator responders (n ¼ 99) Success Defined by PAC Univariate Model Variable* Therapeutic effect at 30 wk Age GS-C GS-F 6-MWT LEFMA TUG BBS Education, HS graduate Time poststroke, 3-6 mo Total effect at 42 wk Age GS-C GS-F 6-MWT TUG BBS Gender, female Prestudy AFO, new Stroke location, cortical Stroke type, hemorrhagic Sensation loss, partial
Multivariate Model
Coeff Est
P Value
Coeff Est
P Value
OR (CI)
0.077 0.0426 0.297 0.012 0.030 0.074 0.120 0.539 1.584
.0013 .0046 .0013 .0004 .5636 .0099 .0032 .5709 .037
0.064
.0098
0.94 (0.89-0.99)
0.065
.0263
0.94 (0.89, 0.99)
1.073 1.393 1.284 1.009 1.040 1.103 1.901 1.758 1.594 0.633 0.243
.0004 .0004 .0002 .0008 .0027 .0003 .0411 .0904 .1727 .3045 .5729
0.068 0.335
.002 .006
0.94 (0.90-0.98) 1.40 (1.10-1.77)
Listed are variables having at least one univariate correlation with PAC or MCID responders status at P < .10 for total and therapeutic effects. Noncompleters are classified as nonresponders. Variables emerging as significant after multivariate logistic regression (with backward selection) have coefficient estimate, P value, and OR with 95% CIs listed to the right. PAC ¼ Perry Ambulation Category; MCID ¼ minimal clinically important difference; Coeff Est ¼ coefficient estimate; OR ¼ odds ratio; CI ¼ upper and lower bounds 95% confidence intervals; GS-C ¼ comfortable gait speed; GS-F ¼ fast gait speed; 6-MWT ¼ 6-minute walk test; LEFMA ¼ lower extremity Fugl-Meyer assessment; TUG ¼ Timed Up and Go; BBS ¼ Berg Balance Scale; HS ¼ high school; AFO ¼ ankle-foot orthosis. * Only variables with P < .10 are included in this table. Variables tested but not listed include ethnicity, marital status, handedness, side of stroke, and vascular distribution of stroke.
however, was greater statistically in the FDS group at 30 weeks compared with the AFO group. This current secondary analysis of the FDS arm provides a platform to explore the response to FDS devices during an extended period in persons who have sustained a stroke. Although the overall clinical benefit of an FDS may be similar to that of an AFO, candidate selection and the mechanism and timing of improvement may differ. With the use of a number of approaches, we confirmed that the benefits of an FDS on GS-C in persons who have had a stroke are clinically and statistically significant from baseline status. When using an FDS, a majority of subjects achieved an important clinical end point by 42 weeks. We confirmed substantial therapeutic and total effects for FDS use, with Cohen’s effect sizes of 0.75 and 0.84, respectively, which are well above the recently reported pooled effect size of 0.29 for physiotherapy mobility interventions in persons in the chronic phase of a stroke [43]. With a mean time poststroke of 4.8 years, our findings fuel continuing doubt regarding the concept of a functional “plateau” in persons in the chronic phase of a stroke [44].
Characteristics of FDS Response The characterization of how GS-C responds to use of an FDS over time in persons who have had a stroke fulfills our first and second objectives. Our subjects used an FDS as durable medical equipment on a daily basis, as opposed to an adjunct to exercise during PT sessions [14]. We first discuss the 3 orthotic effects. GS-C with use of FDS increased progressively by a total of 45% (0.42-0.61 m/s) over 42 weeks compared with baseline without use of the device (total effect, see Figure 1). This gain is somewhat better than other published data [15,16,23,25,27-29], given a wide variation in methodologies (see Table 1). Two other factors might have accounted for at least some of this long-term improvement. First, many persons who have had a stroke are deconditioned [44] and socially isolated [45]. Given the travel, staff interaction, and effort required for a clinical trial, simply participating might have enhanced their physical mobility and contributed to gains. Second, natural recovery might have contributed but should be minimal given a mean time poststroke of 4.8 years, with only 10% of the sample having a poststroke time of
