Comparison of One-Year Postinjury Mobility Outcomes Between Locomotor Training and Usual Care After Motor Incomplete Spinal Cord Injury

Abstract

Objectives:

To compare 1-year mobility outcomes of individuals with traumatic motor incomplete spinal cord injury (miSCI) who participated in standardized locomotor training (LT) within the first year of injury to those who did not.

Methods:

This retrospective case-control analysis conducted with six US rehabilitation hospitals used SCI Model Systems (SCIMS) data comparing 1-year postinjury outcomes between individuals with miSCI who participated in standardized LT to those who received usual care (UC). Participants were matched on age, gender, injury year, mode of mobility, and rehabilitation center. The primary outcome is the FIM Total Motor score. Other outcomes include the FIM Transfer Index, FIM Stairs, and self-reported independence with household mobility, community mobility, and stairs.

Results:

LT participants reported significantly better FIM Total Motor (difference = 2.812, 95% confidence interval [CI] = 5.896, 17.282) and FIM Transfer Index scores (difference = 0.958, 95% CI = 0.993, 4.866). No significant between-group differences were found for FIM Stairs (difference = 0.713, 95% CI = −0.104, 1.530) or self-reported household mobility (odds ratio [OR] = 5.065, CI = 1.435, 17.884), community mobility (OR = 2.933, 95% CI = 0.868, 9.910), and stairs (OR = 5.817, 95% CI = 1.424, 23.756) after controlling for multiple comparisons.

Conclusion:

LT participants reported significantly greater improvements in primary and secondary measures of mobility and independence (FIM Total Motor score; FIM Transfer Index) compared to UC participants. Self-reported mobility outcomes were not significant between groups.

Return to walking function is a priority after injury.^1-4 Locomotor training (LT)—the repetition of stepping-like patterning on a treadmill and/or over ground with or without body-weight support—has been shown to improve walking speed, endurance, and balance after motor incomplete spinal cord injury (miSCI).^5-10 In the largest LT study published to date,¹¹ statistically significant improvements in balance and walking measures (Berg Balance Scale, 6-minute walk test, 10-meter walk test) were reported in 196 individuals with miSCI after completing a standardized LT protocol initiated after discharge from inpatient rehabilitation. However, these types of intensive LT protocols (90-minute sessions completed 5x/week) are not considered usual care (UC) for individuals with miSCI, but instead LT is delivered with numerous differences in dose, intensity, frequency, modality, and guiding principles.^11,12

UC to promote walking recovery after SCI is difficult to define and often encompasses a variety of interventions and strategies dictated by therapist experience and resource availability.^13,14 A previous publication reported that 88% of all individuals with SCI received some outpatient (OP) therapy over the first year of injury (n = 1032),¹⁵ while another reported that individuals with SCI receive a median of 75 hours of OP physical therapy (PT) during the first year of injury.¹⁶ Therefore, OP therapy offers an important opportunity to maximize walking recovery after SCI, especially as inpatient lengths of stay have been dramatically reduced since the 1970s.^17,18

Due to practical and resource limitations, no prospective, randomized controlled trial has been conducted to compare LT and UC outcomes under equivalently specified protocols. In fact, there is no previously published evidence directly comparing mobility outcomes between individuals with miSCI who received an OP LT program to individuals who received UC. The objective of this study is to compare mobility between individuals with miSCI, preserved motor function below the neurologic level of injury, who received at least 30 sessions of LT within the first year of injury and those who received UC to inform intervention delivery and mobility recovery within the first year after miSCI.

Methods

A retrospective (nested case-control) analysis was completed using data obtained from the SCI Model Systems (SCIMS) National Database. The SCIMS, funded by the National Institute on Disability, Independent Living and Rehabilitation Research (NIDILRR), was established in 1970 to study the course and recovery of outcomes following traumatic SCI. SCIMS centers obtain demographic and injury-related data at admission and discharge from inpatient rehabilitation using medical record abstraction and patient/family preinjury questionnaire interviews. Follow-up surveys are then given to individuals with SCI who have been discharged from these centers at 1 and 5 years post injury and every 5 years thereafter.

The NeuroRecovery Network (NRN) is a collaboration of centers established by the Christopher and Dana Reeve Foundation in 2004 focused on designing and implementing standardized intervention protocols to maximize recovery after SCI. The NRN LT protocol was implemented in a standardized manner at six OP therapy centers across the United States specializing in SCI rehabilitation. The protocol includes 55 minutes of manually facilitated repetitive standing and stepping on a treadmill with partial body-weight support (BWS) (Figure 1) followed immediately by 30 minutes of over ground mobility training (Figure 2) at a frequency of 5 days per week.¹⁹

Figure 1.

Locomotor training with body-weight support.

Figure 2.

Over ground walking training.

Participant data for this study were obtained from the six NRN LT centers that also contributed data to the SCIMS database from 2005 to 2018. LT participants were selected according to the following inclusion criteria: (a) diagnosed with traumatic miSCI, (b) completed inpatient rehabilitation at a SCIMS center, and (c) completed at least 30 sessions of the NRN LT protocol prior to their 1-year follow-up SCIMS interview. UC participants were defined as individuals with traumatic miSCI who completed inpatient rehabilitation at an SCIMS center that was also a NRN center but who did not participate in the NRN LT program. LT and UC participants were matched in a 1:1 manner based on age (±5 years), time since injury (±3 years), motor function designated by American Spinal Cord Injury Association Impairment Scale (AIS C or D), sex, mode of mobility at time of discharge from inpatient rehabilitation (walk or wheelchair), and SCI center.²⁰ More severe injuries²¹ (AIS C and wheelchair use) and older age²² may negatively impact recovery after SCI, while neurologic and functional recovery may differ with sex.^23,24 Time since injury and rehabilitation center were matched to control for longitudinal and center differences in rehabilitation philosophy. Matching was completed using a SAS macro (SAS 9.4).

Outcomes

The FIM Total Motor score at 1 year post injury was the primary outcome of this study.²⁵ The FIM contains 13 motor items that are rated on a seven-point ordinal scale from complete dependence (0) to complete independence (7). The motor subscale includes eating, grooming, bathing, upper body dressing, lower body dressing, toileting, bladder management, bowel management, bed to chair transfer, toilet transfers, shower transfer, locomotion, and stairs. Secondary FIM outcomes in this analysis include a FIM Transfer Index (combined score for wheelchair to bed, toilet, and shower transfers) and FIM Stairs. One-year mobility-related self-report questions also were included as secondary outcomes in this analysis: (1) Are you able to walk (with or without mobility aid) for 150 feet in your home? (2) Are you able to walk (with or without mobility aid) for one street block outside? (3) Are you able to walk (with or without mobility aid) up one flight of steps?

Data analysis

SAS version 9.4 was used for all statistical analyses. Demographic differences were analyzed between paired LT and UC participants using a McNemar test for categorical variables. Paired t tests were used to compare means of normally distributed continuous variables; nonparametric Wilcoxon signed-rank tests were used to compare medians for skewed data. Linear mixed-effect models were fit for normally distributed continuous outcomes and presented as mean differences with a 95% confidence interval (CI). Generalized linear mixed-effects models were fit for nominal outcomes assuming a binomial distribution and a logit link with outcomes presented as adjusted odds ratios (OR) with 95% CIs. Models were adjusted for the following relevant patient and clinical factors: marital status, education level, inpatient hospitalization days, inpatient discharge FIM Total Motor score, and International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI) Total Motor score. As there were multiple outcomes, each with unique missing data rates, all pairs that could contribute outcome data were included in each outcome model. Furthermore, there could be missing outcome data from one member of the pair; however, the mixed-effect modeling strategy does not require that the pair be excluded from analysis and is able to incorporate all available information from the pair. Effect sizes were calculated to reflect the overall intervention effect across continuous outcomes expressed in Cohen d. Effect sizes were interpreted as small (d = 0.2-0.5), medium (d = 0.5-0.8), or large (d ≥ 0.8). Bonferroni corrections were used to adjust for comparisons of groups across six outcomes (α = 0.05/6 = 0.0083).

Results

Seventy-two matched pairs were obtained from the SCIMS national database. The LT group's average age was 38.8 (SD = 16.1) whereas UC was 38.7 (SD = 15.9) (p = .782) (see Figure 3). Significant differences (p < .05) were noted between those who received LT and UC at discharge from inpatient rehabilitation, with the LT group having more married individuals with a higher level of education. Private insurance/ workman's compensation was the primary payer for 96.7% of the LT group as opposed to 61.5% in UC; however, significance could not be calculated due to low cell counts. The LT group had longer inpatient rehabilitation hospitalization (p < .001), lower ISNCSCI Total Motor score (p = .001), lower FIM Transfer Index (p = .002), lower FIM Total Motor score (p = .002), and lower FIM Stairs (p = .012) than the UC group at inpatient rehabilitation discharge (see Table 1). The LT group completed a median of 60 (range = 30-230; IQR = 40-110) LT sessions.

Figure 3.

Subject allocation and matching. AIS = ASIA Impairment Scale; NRN = NeuroRecovery Network; SCIMS = SCI Model Systems.

Table 1.

Comparison of the continuous and categorical variables at discharge from inpatient rehabilitation

Continuous variable	LT		UC		p value
	Mean (SD)	Median (IQR)	Mean (SD)	Median (IQR)
Age at injury, years (Dis.)	38.8 (16.1)	39 (22.5-51.5)	38.7 (15.9)	38 (23-53)	.971a
Days hospitalized (Dis.)	75.7 (42.0)	65 (47-97)	52.7 (35.7)	49 (29-63)	<.001*b
ISNCSCI Total Motor score (Dis.)	57.0 (22.3)	56.5 (42-74.5)	69.1 (20.8)	76.5 (56-84)	.001*a
FIM Transfer Index (Dis.)	11.0 (5.0)	12 (7-15)	13.7 (5.3)	15.5 (12-18)	.002*a
FIM Transfer Index (Year 1)	16.7 (5.5)	18 (16-21)	16.6 (5.8)	18 (16-21)	0.929a
FIM Total Motor Score (Dis.)	50.3 (17.1)	51.5 (39-63.5)	60.2 (20.1)	69 (47.5-75.5)	0.002*a
FIM Total Motor Score (Year 1)	71.3 (19.3)	78 (69.83)	72.2 (21.3)	82 (67-87)	0.800a
FIM Locomotion (Dis.)	5.6 (1.0)	6 (6-6)	5.6 (1.1)	6 (6-6)	0.884b
FIM Locomotion (Year 1)	9.6 (2.9)	12 (11-14)	9.9 (3.2)	11 (7-12)	0.520b
FIM Stairs (Dis.)	1.7 (1.3)	1 (1-2)	2.6 (1.9)	1 (1-4)	0.012*b
FIM Stairs (Year 1)	3.8 (2.4)	5 (1-6)	4.0 (2.5)	5 (1-6)	0.356b
Categorical variable	n	%	n	%	p value
Sex (n = 144)					N/A
Male	57	79.2%	57	79.2%
Female	15	20.8%	15	20.8%
Marital status (n = 142)					.011*c
Married or in relationship	38	53.5%	23	32.4%
Single, divorced, widowed, other	33	46.5%	48	67.6%
Education (n = 135)					.001*c
High school or lower	38	55.9%	55	82.1%
More than high school	30	44.1%	12	17.9%
Primary payer (n = 56)					N/I
Private insurance/Work comp	29	96.7%	16	61.5%
Medicare/Medicaid/Veterans	1	3.3%	10	38.5%
AIS classification (n = 144)					N/A
AIS C	24	33.3%	24	33.3%
AIS D	48	66.7%	48	66.7%
Mode of mobility (n = 144)					N/A
Walking	9	12.5%	9	12.5%
Wheelchair	62	86.1%	62	86.1%
Both	1	1.4%	1	1.4%

Note: AIS = American Spinal Cord Injury Association Impairment Scale; Dis. = discharge from inpatient rehabilitation; IQR = interquartile range; ISNCSCI =International Standards for Neurological Classification of Spinal Cord Injury; LT = locomotor training; UC = usual care.

^aPaired t test.

^bWilkcoxon signed-rank test.

^cChi-square test.

^*Statistically significant difference (α = 0.05); N/A = significance not available due to matched sample; N/I = significance not interpretable due to low cell counts.

After adjusting for covariates (marital status, education level, inpatient hospitalization days, inpatient discharge FIM Total Motor score, and ISNCSCI Total Motor score), the LT group scored significantly higher on FIM Total Motor Score (difference = 11.59, 95% CI = 5.90, 17.28) and FIM Transfer Index (difference = 2.93, 95% CI = 0.99, 4.87) in comparison to UC at 1 year. LT participants were more likely to walk 150 feet (OR = 5.07, 95% CI = 1.44, 17.88) and to walk 1 flight of stairs (OR = 5.82, 95% CI = 1.42, 23.76) in comparison to UC participants. See Tables 2 and 3. However, after controlling for multiple comparisons only the adjusted differences in FIM Total Motor score and FIM Transfer Index remained statistically significant (ps < .0083).

Table 2.

Adjusted comparison of continuous mobility outcomes at 1 year follow-up

LT vs. UC	n	Mean difference	SE	95% CI	Effect size (d)	p value
Total FIM Motor score	111	11.589	2.812	(5.896, 17.282)	0.39	<.001*
FIM Transfer Index	112	2.930	0.958	(0.993, 4.866)	0.29	.004*
FIM Stairs	113	0.713	0.404	(-0.104, 1.530)	0.15	.085

Note: CI = confidence interval; LT = locomotor training; SE = standard error; UC = usual care.

^*Statistically significant difference (α = 0.05/6 = 0.0083).

Table 3.

Adjusted comparison of dichotomous outcomes at 1 year follow-up

LT vs. UC	N	Odds ratio	95% CI	p value
Walk 150 ft in the home	113	5.065	(1.435, 17.884)	.013
Walk one street block	113	2.933	(0.868, 9.910)	.082
Walk one flight of stairs	113	5.817	(1.424, 23.756)	.015

Note: CI = confidence interval; LT = locomotor training; UC = usual care. Statistically significant difference (α = 0.05/6 = 0.0083).

ISNCSCI Total Motor score at discharge from inpatient rehabilitation had significant positive associations with five outcomes at 1 year: FIM Total Motor score, FIM Stairs, walk 150 feet in the home, walk one street block, and walk one flight of stairs. FIM Total Motor score at discharge from inpatient rehabilitation had significant positive associations with three outcomes at 1 year: FIM Total Motor score, FIM Transfer Index, and FIM Stairs. The number of days spent in inpatient rehabilitation had significant negative associations with FIM Total Motor score and FIM Transfer Index at 1 year. High school or lower education had a significant positive association with FIM Total Motor score (see Table 4).

Table 4.

Covariate model results on primary and secondary outcomes

Models	Days hospitalizeda	FIM Total Motor scorea,b	ASIA Total Motor scorea,b	Educationc	Marital statusc
Total FIM Motor score	-0.141*	0.469*	0.246*	8.114*	1.191
FIM Transfer Index	-0.046*	0.108*	0.052	1.750	1.099
FIM Stairs	-0.009	0.033*	0.038*	0.288	-0.244
Walk 150 ft in the homea	0.985	1.011	1.051*	1.850	1.476
Walk one street blocka	0.996	1.029	1.052*	0.861	1.243
Walk one flight of stairsa	0.987	1.034	1.063*	1.426	0.583

Note: Education = high school or lower vs. more than high school; Marital status = married or in relationship vs. single, divorced, widowed, other.

^aSlope.

^bAt discharge from inpatient.

^cOdds ratio.

^*Statistically significant (α = 0.05).

Discussion

Individuals with miSCI who participated in the standardized OP LT program reported improved mobility at 1 year in comparison to a matched cohort who did not. A comparison of these matched groups at the time of discharge from inpatient rehabilitation revealed significant between-group differences. On average, individuals in the LT group spent a greater number of days in inpatient rehabilitation exhibiting lower motor function while also discharging from inpatient rehabilitation with less mobility and independence in comparison to those who would receive UC. Previous literature supports the positive association of severity of injury and comorbidities to length of stay in SCI.^26,27

These baseline differences were surprising and represent a potential selection bias. All individuals who presented with an AIS C or D injury at those facilities were eligible for participation in the LT program; it appears that across sites, those with the greatest functional and neurologic impairment were more likely to participate. This is potentially due to the intense staffing model (one physical therapist and three technicians) and high technology utilization (treadmill system with BWS) in the LT program in comparison to UC (generally one physical therapist with limited technology). The intensity and high-tech nature of the LT program may have been more attractive to individuals with less mobility at discharge, and therapists may have more strongly encouraged the LT program for these individuals, also thinking that LT might provide more improvement than UC. Those with longer lengths of stay, more education, a supportive spouse, or with greater motivation may have been more willing/able to participate in the 5 day per week LT protocol. Some or all of these factors (and more) may have produced selection bias or interacted with LT to produce better outcomes.

This retrospective analysis lacks randomization and rigid controls, so differences in outcomes must be interpreted with caution. However, LT at this frequency and duration has previously been reported to improve mobility in individuals with SCI,¹¹ and the LT group in this analysis had a lower mean baseline mobility score than the control group (p < .05). Therefore, it is reasonable to evaluate the association of 1-year mobility outcomes with and without the presence of LT without concluding cause. On average, the LT group scored 11.6 points higher on the FIM Total Motor score than matched participants at 1 year (d = 0.39). Unfortunately, a minimal clinically important difference has not been established for the FIM Total Motor score for individuals who have SCI. However, an 11.6-point difference is suggestive of a level that allows individuals to perform daily tasks with less assistance and may allow individuals with SCI to live more independently and require less caregiver assistance.²⁸

The FIM locomotor item could not be used to determine a difference in walking ability between groups as the reported numeric value does not distinguish between individuals walking or using a wheelchair.²⁹ A comparison of FIM Stairs demonstrated no statistically significant differences between groups, whereas the LT group reported higher scores than the UC group (0.713 points) with a small effect size (d = 0.15).

The LT group demonstrated significantly greater ability to transfer to a variety of surfaces (bed, shower, and toilet) compared to the UC group (d = 0.29). LT participants and families often report that even if walking is not recovered after LT, they achieve greater functional independence when transferring to a variety of surfaces. The inclusion of the FIM Transfer Index in this study was also influenced by motor learning literature and the principle of transference. Multiple authors³⁰ discuss transference as the likelihood that less difficult skills such as transfers become easier when training is focused on a more difficult skill such as walking. It is also important to note there could be other clinical and patient-related factors impacting the differences in these outcomes not controlled for in this analysis due to the lack of randomization, which should be assessed in future studies.

The odds of the LT group being able to walk 150 feet in their homes at 1 year was over five times greater than for those who received UC, yet this did not remain significant after controlling for multiple comparisons (p < .0083). The LT protocol generally focuses on improving independence with household mobility as this is a primary concern for insurance companies requiring resource justification. Therefore, it seems reasonable that the largest reported difference in walking ability between LT and UC participants would occur within the household environment.³⁰ The LT group reported nearly three times greater odds of being able to walk one street block in comparison to the UC group, which was not significant. The LT group was nearly six times more likely to walk up/down one flight of stairs in comparison to the UC group, yet this did not remain significant after controlling for multiple comparisons (p < .0083).

ISNCSCI Total Motor score and FIM Total Motor score at discharge from inpatient rehabilitation demonstrated the most significant impact on all mobility outcomes at 1 year post injury. Previous literature supports FIM scores at discharge from inpatient rehabilitation are predictive of mobility at 1 year follow-up,²⁸ and greater severity of injury (lower ISNCSCI Total Motor scores) has been associated with lower levels of mobility and independence at 1 year.^26,27

Based on literature previously discussed,^15,16 individuals receiving UC in five SCIMS centers received a median of 75 hours of OP PT during the first year of injury, while the LT group in this analysis received a median of 60 LT sessions (90 hours). The retrospective nature of this study limits a full understanding and quantification of UC therapy. Inpatient UC PT has been shown to cover a wide range of activities with only a minor focus on pre-gait and gait.³¹ Although OP UC PT may have a greater focus on mobility, it is likely that the LT group spent more time on mobility training than the UC group, but this quantity is unknown and likely varies across participant pairs. This alternative explanation for the study findings cannot be ruled out. Future investigations should be prospective and protocol-driven to evaluate whether the specificity of the LT intervention or the additional hours of therapy resulted in the between-group differences found in this analysis. These results should be generalizable to individuals with SCI as Ketchum et al.³² reported that adults in the SCIMS database are largely representative of the population of adults in the United States receiving rehabilitation for new onset traumatic SCI.

Limitations

There are several limitations to this investigation. Although we are using previously published estimations of OP therapy hours from SCIMS data, the inability to know or account for what types and dosing of interventions the UC group received is a limitation. This lack of categorization and quantification of UC interventions limits the interpretations that can be drawn from this research. Missing data are a limitation to this study, and mixed-effects models were used to maximize the amount of data that could be used from the pairs. Individuals with lumbar injuries were included in the control group (24%) to maximize power, even though only one individual in the LT group had a lumbar injury (1.5%). However, FIM Total Motor score and ISNCSCI Total Motor score at discharge from inpatient rehabilitation were included in all models to account for the variability associated with severity of injury. Primary payer was not included in models due to the large amount of missing data (66%) and should be included in future studies.

Participants in the NRN are encouraged to complete at least 40 sessions of LT; those participants may have had better insurance coverage than UC participants, which creates potential selection bias. Participants in this study have inpatient rehabilitation lengths of stay that are much longer than what has previously been reported (73 days LT group; 52 days UC group; and 34 days National Statistical Center), which may limit generalizability.¹⁷ Given that all outcomes for the SCIMS database are gathered through an interview, there is the potential of introducing interview and recall bias into this analysis.

Conclusion

This retrospective, matched cohort study provides a useful starting point for comparing the longer term effects (1 year) of intensive LT in individuals with miSCI in comparison to UC. The primary mobility outcome was highly significant (p < .001), with the LT group scoring 11.6 points higher on FIM Total Motor score than the UC group at 1 year post injury (d = .39). This was true even though the LT group scored 9.9 points lower than the UC group at inpatient rehabilitation discharge. Among secondary measures, transferring to multiple surfaces remained significantly better in the LT group after controlling for multiple comparisons. Future prospective studies should be designed with larger randomized samples and protocols specifying consistent types and quantities of interventions for both treatment and control groups to verify these conclusions and estimate group differences with better accuracy (smaller CIs, greater power).