Capítulo II Marco Teórico
2.1. Antecedentes teóricos
2.1.7. Canasta comercial transfronteriza
An overview of study recruitment and retention is presented in Figure 4.1. There were 217 patients referred to this study for screening from UNC Hospital and Duke Health (acute inpatient rehabilitation = 48; acute care = 169). Of these, 140 were excluded from the study based on eligibility criteria and 20 patients were unable to be screened between referral to the investigator and hospital discharge. Therefore, 56 patients (acute care = 22; acute inpatient rehabilitation = 34) were enrolled in the study. After providing informed consent, 6 participants did not complete the baseline assessments due either being excluded based on study criteria after consent (n = 4), participant withdrawal after consent (n = 2), or inability to complete baseline testing within 3 days of hospital discharge (n = 3). A total of 47 participants completed baseline testing. Following baseline testing, two participants did not complete the 3-month follow-up visit, one due to death and one being lost to follow-up. One additional participant completed part of her 3-month follow- up assessments (fall diary) but died prior to the completion of monitoring her daily walking activity and completion of questionnaires. Data from one day of walking activity were acquired from this participant prior to her hospitalization and subsequent death.
Participants Demographics
Participant characteristics are detailed in Table 4.1. The 47 study participants (57% male) were on average 59.8 (SD 11.7) years old with a mean of 14 (SD 2.8) years of education. The participants were of white (51%), black (43%), Asian (2%), or Hispanic/Latino (1%) race. A majority of the participants were married (66%) with a mean of 3 (SD 1) persons residing in their home at the time of hospital discharge. Participants homes were located in urban (23%), suburban (34%), and rural (43%) areas with 85% of the participants being discharged to their own home and 13% to a family member’s home. The gross annual household income of the majority of participants was less than $75,000 per year (76%) pre-stroke. Pre-stroke, 53% of the participants were working full-time and 32% were retired. Ninety-four percent of the participants were driving pre-stroke.
Table 4.1. Participant demographics (n = 47). Values are the count (proportion) or mean (SD).
Variable Count/Proportion or Mean/Median
(SD/IQR)
Number of participants missing data
Age (years) 59.8 (11.7) 0
Male sex – number (%) 27 (57) 0
Education (years) 14.0 (2.8) 0 Race – number (%) White/non-Hispanic 24 (51) 0 Black 20 (43) 0 Asian 2 (4) 0 Hispanic 1 (2) 0
Marital status – number (%)
Single 6 (13) 0
Married 31 (66) 0
Separated 3 (6) 0
Current living location – number (%)
Home 40 (85) 0
Family member’s home 6 (13) 0
Friend’s home 1 (2) 0
Geographic location of home – number (%)
Urban 11 (23) 0
Suburban 16 (34) 0
Rural 20 (43) 0
Persons residing in home – number 3 (1) 0
Gross household income – number (%)
< $25,000 11 (23) 0
$25,000 - $49,000 13 (28) 0
$50,000 - $74,999 11 (23) 0
$75,000 – $149,000 8 (17) 0
> $150,000 4 (9) 0
Driving status – yes, number (%) 44 (94) 0
Employment Status –
Employed full time 25 (53) 0
Employed part time 5 (11) 0
Unemployed 2 (4) 0
Retired 15 (32) 0
Retired on disability 0 (0) 0
Stroke Characteristics
Stroke characteristics are detailed in Table 4.2. The median time post stroke was 14 days (IQR 7, 21). Of the 47 participants, 36 (77%) had an ischemic stroke and 11 (23%) had a hemorrhagic stroke, with left-sided hemiplegia in 25 (53%). The locations of the stroke were predominantly subcortical (40%) and right hemispheric (26%). Twelve of the 36 participants (33%) with ischemic stroke received tissue plasminogen activator (tPA). The median admission NIHSS was 6 (IQR 3, 9) indicating moderate stroke severity. The median hospital length of stay was 17 days (IQR 9, 26).
Table 4.2. Participant stroke characteristics (n = 47). Values are the count (proportion) or mean/median (SD/IQR).
Variable
Count/Proportion or Mean/Median
(SD/IQR)
Number of participants with missing data
Time post stroke (days) 14 (7, 21) 0
Type of stroke – number (%)
Ischemic 36 (77) 0
Hemorrhagic 11 (23) 0
Location of stroke – number (%)
Right hemisphere 12 (26) 0
Left hemisphere 8 (15) 0
Subcortical 19 (40) 0
Bilateral 5 (11) 0
Brainstem 3 (6) 0
Side of hemiplegia, left – number (%) 25 (53) 0
tPA received, yes – number (%) 12 (26) 0
Admission NIHSS score (max 42)a 6 (3, 9) 13
Hospital length of stay (days) 17 (9, 26) 0
Abbreviations: tPA, Tissue plasminogen activator; NIHSS, National Institute of Health Stroke Scale
a 14 participants did not have an admission NIHSS score.
Language, cognition, vision, and comorbidity
Participant language, cognition, vision, and comorbidity levels are detailed in Table 4.3. Participants, on average, had language function within normal limits based on a mean aphasia quotient of 97.4% on the Bedside WAB.172 On average, participants had mild cognitive impairment, indicated by a mean MoCA score of 24.1 (SD 3.6), but no spatial neglect indicated by score on the Star Cancellation test laterality index of < 0.54. On average, participants had a mean of 6.2 (SD 2.2) comorbidities of moderate level based on a mean CIRS-G severity index of 0.92 (SD 0.35).173,174
Table 4.3. Participant (n = 47) language, cognition, vision, and comorbidity levels. Values are the count (proportion) or mean (SD).
Variable Mean (SD)
Number of participants with
missing data
Bedside WAB, Aphasia Quotient (%) 97.4 (10.8) 0
Montreal Cognitive Assessment (max. 30) 24.1 (3.6) 0
Star Cancellation Test, laterality index 0.50 (0.01) 0
Cumulative Illness Rating Scale for Geriatric (CIRS-G)
Total score (max=56) 11.9 (4.5) 0
Severity Index 0.92 (0.35) 0
Number of comorbidities (max=14) 6.2 (2.2) 0
Abbreviations: WAB, Western Aphasia Battery
Clinical self-report and physical performance outcomes
Participant self-report and physical performance outcomes are detailed in Table 4.4. Participants reported low (n = 7; 15%), moderate (n = 23, 49%), and high (n = 16, 34%) activity levels prior to their stroke. At hospital discharge, participants on average reported decreased balance confidence levels in performing functional activities that require balance with a mean ABC score of 65.1% (SD 21.7), moderate self-perceived impact of stroke on walking with a mean Walk- 12 score of 57.5% (SD 26.8), and mild depression with a median PHQ-9 score of 6 (IQR 2, 9). On average, comfortable walking speed was 0.62 m/s (SD 0.34) and 2-minute walk distance was 85.0 m (SD 48.5), which placed them, overall in a limited community ambulator category.27,175 The participants had diminished functional lower extremity strength with a median 5XSTS test of 22.9 s (IQR 16.0, 28.3).127 Seven participants (15%) could not complete the 5XSTS test due to an inability to transition from sitting to standing without using their upper extremities. Therefore, to account for all participants, we also calculated the repetitions completed per second for the 5XSTS, which produced a mean of 0.21 repetitions/s (SD 0.12). In addition, the participants were at risk for falls as evidence by a median Step Test score of less than 7 for their paretic and non-paretic limbs.9 At the time of discharge, 63.8% (n = 30) of the participants used an assistive device (rolling
walker = 11, standard walker = 1, quad cane = 9, and straight cane = 9), and 27.7% (n = 13) used bracing (AFO=10, AAFO = 1, ankle splint = 2) to walk.
Table 4.4. Participant (n = 47) self-report and physical performance outcomes. Values are the count (proportion) or mean/median (SD/IQR).
Variable Count/Proportion or Mean/Median (SD/IQR) Number of participants with missing data
Activities-specific Balance Confidence scale (%) 65.1 (21.7) 0
Walking Impact Scale (Walk-12) (%) 57.5 (26.8) 0
Patient Health Questionnaire-9 (max. 27) 6 (2, 9) 0
Pre-stroke IPAQ, activity category – number (%)a 1
Low 7 (15) -
Moderate 23 (49) -
High 16 (34) -
Pre-stroke IPAQa 1
Total met minutes 4044 (1418, 8558) -
Total walking met minutes 982 (297, 1559) -
5-meter gait speed (m/s) 0.61 (0.34) 0
2-minute walk distance (m) 85.0 (48.5) 2
5-times sit to stand test (s)b 22.9 (16.0, 28.3) 7
5-times sit to stand test (repetitions/s) 0.21 (0.12) 0
Step Test, limb (repetitions)
Non-paretic 6 (2, 9) 0
Paretic 5 (0, 8) 0
Combined 11 (5, 16) 0
Abbreviations: IPAQ, International Physical Activity Questionnaire
a 1 participant did not complete the pre-stroke IPAQ.
b 7 participants were unable to complete the 5-time sit to stand task without using their arms to assist.
Disability outcomes
Participant disability outcomes are detailed in Table 4.5. On admission, the participants (data from n = 33) had a mean AM-PAC raw score of 16.2 (SD 4.3) which indicates an average 54% impairment in mobility. The admission and discharge Functional Independence Measure (FIM) scores were available for 34 participants who were discharged from an acute inpatient rehabilitation facility. Based on the total FIM score, the participants were classified, on average, as moderately affected at admission (mean 62.3, SD 16.7) and mildly affected at discharge (mean 102.0, SD 9.5).176,177 The discharge mRS, as rated by the discharging physician, indicated that 32% (n = 15) of participants had “moderately severe disability”, 30% (n = 14) had “moderate disability”, 13% (n = 6) had “slight disability”, and 26% (n = 12) had “no significant disability”.
Table 4.5. Participant (n = 47) disability outcomes. Values are the count (proportion) or mean (SD).
Variable Count/Proportion or Mean (SD)
Number of participants missing data
Admission AM-PAC- Raw Score (max=24) 16.2 (4.3) 14
Functional Independence Measure
Total score (max=126)
Admission 62.3 (16.7) 0
Discharge 102.0 (9.5) 0
Total motor subscale (max=91)
Admission 36.6 (12.4) 0
Discharge 70.2 (7.5) 0
Total cognitive subscale (max=35)
Admission 25.7 (6.2) 0
Discharge 31.8 (3.3) 0
Discharge Modified Rankin Score – number (%)
0 – no symptoms at all 0 (0) -
1 – no significant disability despite symptoms 12 (26) 0
2 – slight disability 6 (13) 0
3 – moderate disability 14 (30) 0
4 – moderately severe disability 15 (32) 0
5 – severe disability 0 (0) -
Abbreviations: AM-PAC, Activity Measure for Post-Acute Care
Descriptive summary of participants at 3-month follow-up assessment Participant characteristic
At the 3-months post hospital discharge (Table 4.6), a majority of the participants were married (62%). The participants’ places of residence were located in urban (22%), suburban (27%), and rural (47%) areas with a majority living in their own home (71%) or a family members home (18%). A median of 3 persons total, including the participant, were residing in the place of residence. For 75% of the participants, the gross household income remained at less than $75,000
per year. Post stroke, 10% of the participants were working full-time and 24% were unemployed 3 months post hospital discharge, a significant decline compared to pre-stroke (P=0.001). At 3 months post discharge, 51% of the participants were driving compared to 94% pre-stroke (P<0.001).
Table 4.6. Participant demographic characteristics at 3-month post discharge follow-up (n = 45).
Variable Count/Proportion or Median (IQR) Missing Data
Marital Status – number (%)
Single 3 (7) 2
Married 28 (62) 2
Separated 2 (4) 2
Divorced 7 (16) 2
Widowed 3 (7) 2
Current Living Location – number (%)
Home 32 (71) 2
Family Member's Home 8 (18) 2
Friend's House 2 (4) 2
Independent Living Center 1 (2) 2
Current Geographic Area of Home – number (%)
Urban 10 (22) 2
Suburban 12 (27) 2
Rural 21 (47) 2
Number of persons in the home 3 (2, 4) 2
Gross Household income – number (%)
<$25,000 12 (27) 2
$25,000-$49,999 11 (24) 2
$50,000-$74,999 9 (20) 2
$75,000-$149,000 8 (18) 2
>$150,000 3 (7) 2
Current Driving Status – yes, number (%) 23 (51) 2
Current Employment Status – number (%)
Employed Full time 10 (22) 2
Employed part-time 2 (4) 2
Unemployed 11 (24) 2
Retired 15 (33) 2
Retired on disability 5 (11) 2
Phone Check-in
A phone check-in with each participant (n = 45) was completed every 2 weeks between the baseline and the 3-month follow-up visit. There were 222 total phone check-ins completed during
the study with an 82% completion rate based on the planned 270 phone check-ins. A mean of 4.9 (SD 1.2) calls were completed per participant.
Self-reported physical activity, social participation and community
At 3 months post hospital discharge, the participants’ perception of physical activity in the 7 days prior to the assessment date were determined using the Physical Activity Scale for Individuals with Physical Disabilities (PASIPD) with scores ranging from 0 (no activity) to > 100 MET-hours/day (very high activity). For our participants (n = 42), the median total score was 8.65 MET-hours/day (IQR 5.45, 18.29). Previous work using the SIPSO in participants 2-months post hospital discharge from inpatient rehabilitation found a mean score of 7.7 MET-hours/day (SD 7.1).178 The scores from our study and this study were low compared to healthy participants and compared with scores obtained from individuals with disabilities who self-report as “not active at all” (mean PASIPD score = 13.2).152
The participants’ perceptions of their ability to socially and physically reintegrate to a 'normal' lifestyle were assessed using the Subjective Index of Physical Function and Social Outcome (SIPSO) with a median score of 31 of a possible 40 (IQR 24, 36). Based on this score, we can assume that overall, the participants reported moderate physical and social reintegration. Previous work using the SIPSO in participants (n = 249) greater than 6-months living at home post stroke found a median score of 27.159
Obstacle-crossing exposure and fall status outcome results (Aim 1)
This section provides a descriptive summary of participant fall characteristics in the first 3 months post hospital discharge, as well as the performance characteristics during the obstacle-
crossing task at hospital discharge (i.e., obstacle pass/fail, spatiotemporal gait parameters, and visual tracking measures). This is followed by presentation of the analysis of differences between fallers and non-fallers and the bivariate associations between explanatory variables and the outcome, fall status. Next, differences between participants who passed or failed the obstacle- crossing task and the bivariate associations between explanatory variables and “the exposure” (obstacle pass or fail) are presented. A summary of correlations among potential explanatory variables (covariates) are also presented. Binary logistic regression models were performed with covariates found to have significant (1) differences between fallers and non-fallers and those who passed and failed the obstacle-crossing task and (2) bivariate associations with the exposure, obstacle-crossing, and the primary outcome, fall status. The results of the unadjusted binary logistic regression, adjusted binary logistic regression, and binary logistic regression model using a backward deletion method are subsequently presented.
Fall Characteristics
Of the 45 participants with fall data at 3 months post hospital discharge, 21 participants (47%) reported experiencing at least one fall. Of these, 8 participants (18% of entire sample) reported more than one fall (range 2-7). Of the 39 total falls reported by the participants, 5 resulted in injury and 2 required medical attention; none required hospitalization. Average time to first fall was 37.0 days (SD 25.4), with 52% of all reported falls occurring within the first month after hospital discharge. A majority of the falls occurred at the participant’s home (81%), either indoors (n=26) or outdoors (n=4), and 7 falls occurred in the community (19%). In addition, a majority of the falls occurred during walking or transfers (95%).
Obstacle-crossing performance characteristics
The average obstacle height use across all participants was 8.8 cm (SD 0.5). Of the 47 participants who performed the obstacle-crossing task at hospital discharge, 91% (n = 43) completed all 4 trials, with the other 9% (n = 4) completing at least 2 trials. The remaining 2 trials for each participant were not completed due to participant fatigue. Twenty-seven participants (60%) successfully passed all obstacle-crossing attempts. Eighteen participants (40%) failed the obstacle-crossing task on at least one attempt. Of the 21 fallers, 14 (67%) failed the obstacle- crossing task. There were no adverse events. Of the 24 non-fallers, 20 (83%) passed the obstacle- crossing task (Table 4.7). The incidence of fallers was significantly higher (incidence rate = 0.78) in the group that failed the obstacle-crossing task than in the group that passed (incidence rate = 0.26; P=.001). There was a moderately strong relationship (r=0.51, P=0.001) and a significant association (χ2 [1, n=45] = 11.67, P=0.001) between the exposure (obstacle fail/success) and the outcome (faller/non-faller). Participants who failed the obstacle-crossing task at hospital discharge were 10.00 (95% CI 2.45, 40.78) times more likely to fall in the first 3 months after discharge than those who passed the task. Participants who failed the obstacle-crossing task at hospital discharge had 3.00 (95% CI 1.51, 5.94) times the risk of falling (200% increase in risk) than those who passed the task.
The obstacle-crossing task was 67% sensitive (ability to classify faller) and 83% specific (ability to classify non-faller). The positive predictive value (PPV) was 78% (percentage of participants who fail the obstacle-crossing task and actually fall) and the negative predictive value (NPV) was 74% (percentage of participant who pass the obstacle-crossing task and do not fall). The positive likelihood ratio (+LR) was 4.00, indicating an increase in the probability of falls when a patient fails an obstacle-crossing task by approximately 25%.179 The negative likelihood ratio (-
LR) was 0.40, indicating a decrease in probability of falls when a patient passes the obstacle- crossing task by approximately 20%.179 The area under the receiver operating characteristics curve (AUC) was 0.76 (95% CI 0.61, 0.91), which means that the obstacle-crossing task is in the acceptable discrimination range.180
Table 4.7. Crosstabulation of obstacle crossing and fall status Faller Non-Faller Total
Fail Obstacle 14 4 18
Pass Obstacle 7 20 27
Total 21 24 45
Odds ratio: 10.00 (95% CI 2.45, 40.78) Risk ratio: 3.00 (95% CI 1.51, 5.94)
Obstacle-crossing spatial temporal parameters of gait during obstacle-crossing
Spatial temporal parameters of gait during obstacle crossing are presented in Table 4.8. There was a significant difference between the pre-obstacle (0.50 m/s, IQR 0.20, 0.83) and obstacle-crossing gait speed (0.46 m/s, IQR 0.20, 0.83; P<0.001). Further, there was a significant difference between the late and early phase pre-obstacle step length variability, with greater variability in late phase (33.00%, IQR 15.65, 63.76) compared to early phase (11.51%, IQR 5.21, 38.31; P<0.001). Pre-obstacle stride length also significantly decreased from the early (83.19 cm, SD 33.88) to late phase (78.13 cm, SD 36.00; P=0.001). There were no differences between early and late pre-obstacle step length or paretic and non-paretic lead and trail limb swing durations.
Table 4.8. Spatiotemporal parameters of gait during obstacle-crossing for all participants (n = 47). Values are mean/median (SD/IQR).
Variable Mean/Median (SD/IQR) Missing Data Gait speed
Pre-obstacle gait speed (m/s) 0.50 (0.26, 0.90) 0
Early Pre-obstacle 0.49 (0.26, 0.92) 0
Late Pre-obstacle 0.49 (0.20, 0.94) 0
Obstacle-crossing gait speed (m/s) 0.46 (0.20, 0.83) 2
Magnitude of change in pre-obstacle to obstacle-crossing gait speed (m/s)
-0.07 (0.06) 2
Step length
Pre-obstacle step-length (cm) 41.33 (16.71) 0
Early pre-obstacle 40.99 (17.31) 0
Late pre-obstacle 41.79 (17.10) 0
Variability and asymmetry
Pre-obstacle step-length COV (%) 22.07 (13.82, 49.98) 0
Early pre-obstacle 11.51 (5.21, 38.31) 0
Late pre-obstacle 33.00 (15.65, 63.76) 0
Pre-obstacle step-length ASI (%) 14.09 (7.62, 36.01) 0
Early pre-obstacle 13.60 (6.29, 50.09) 0
Late pre-obstacle 33.88 (13.06, 53.99) 0
Stride length
Pre-obstacle stride length (cm) 82.46 (32.62) 0
Early pre-obstacle 83.19 (33.88) 0
Late pre-obstacle 78.13 (36.00) 0
Swing duration over the obstaclea
Paretic lead limb (s) 0.74 (0.59, 1.18) 6
Paretic trail limb (s) 0.55 (0.51, 0.64) 19
Non-paretic lead limb (s) 0.66 (0.55, 0.95) 19
Non-paretic trail (s) 0.53 (0.48, 0.67) 6
Visual tracking during obstacle-crossing
During the obstacle-crossing task at hospital discharge, the participants spent more time fixating (ms) on the pre-obstacle walking path than on the post-obstacle walking path, the obstacle, or the non-walking path (Table 4.8). Specifically, participants maintained significantly longer visual fixation on the pre-obstacle walking path (4357 ms, IQR 1174, 7171) than on the non- walking path (141 ms, IQR 0, 1175; p<0.001); significantly longer fixation on the post-obstacle walking path (4034 ms, IQR 2977, 7708) than on the obstacle (2833, IQR 1338, 3877; P=0.016) and the non-walking path (141 ms, IQR 0, 1175; P<0.001); and significantly longer fixation on the
P<0.001). The median time to first fixation on the obstacle was 2969 ms (IQR 1624, 5910) and median total number of unique fixations on the obstacle was 4.4 (IQR 2.3, 9.6). Similarly, the participants spent significantly more of the total trial time (percent dwell time) on the post-obstacle walking path (37%, SD 16), pre-obstacle walking path (27%, SD 18), and obstacle (24%, SD 15) than on the non-walking path (1%, IQR 0, 6; P<0.001). Participants spent significantly more percentage of the total trial time fixating on the post-obstacle walking path (37%, SD 16) than on the obstacle (24%, SD 15; P=0.007).
Table 4.9. Visual tracking parameters during obstacle-crossing. Values are mean/median (SD/IQR). Variable Mean/Median (SD/IQR) Number of participants with missing dataa Fixation time on
Pre-obstacle walking path (ms) 4357 (1174, 7172) 5
Obstacle (ms) 2833 (1338, 3877) 5
Post-obstacle walking path (ms) 4034 (2977, 7708) 5
Non-walking path (ms) 141 (0, 1175) 5
Percent dwell time on
Pre-obstacle walking path (%) 27 (18) 5
Obstacle (%) 24 (15) 5
Post-obstacle walking path (%) 37 (16) 5
Non-walking path (%) 1 (0, 6) 5
Other
Time to first fixation on obstacle (ms) 2969 (1624, 5911) 5
Fixation count on obstacle (#) 4.4 (2.3, 9.6) 5
Analysis of differences for participants classified as fallers or non-fallers at 3 months post hospital discharge
Data for all participants (n = 47) and separately for faller (n = 21) and non-faller (n = 24) status at 3-months post hospital discharge for the analysis of differences are presented in tables grouped by similar content areas in Appendix 4 (Table A4.1 to Table A4.7). The P values in the respective tables represent the statistical significance of the differences between fallers and non- fallers for each variable.
Participants who were fallers were significantly more days post stroke at time of consent (19 days, IQR 13, 26) than non-fallers (9 days, IQR 5, 14; P=0.002), had significantly longer hospital lengths of stay (25, IQR 13, 26) than non-fallers (10, IQR 5, 14; P<0.001), and had better cognitive ability based on the Montreal Cognitive Assessment (MoCA) scores (26, IQR 24, 28) compared to non-fallers (24, IQR 20, 26; P=0.038). Although, based on the non-linear nature of the levels of the MoCA scoring and skewing of scores on the high end of the range of scoring, this difference is likely not clinically meaningful.181 There were no significant differences found between fallers and non-fallers for age, sex, years of education, race, type of stroke, side of hemiplegia, number treated with tPA, stroke severity, language function, unilateral spatial neglect, and presence and severity of comorbidities (Tables A4.1 to 4.3).
Fallers had a lower mean self-efficacy for balance at hospital discharge (57.3%, SD 18.1) than non-fallers (71.2%, SD 23.4; P=0.032) and a higher mean self-reported perceived impact of stroke on walking (72.1%, SD 20.5) than non-fallers (43.3%, SD 25.3, P<0.001). Fallers had a slower gait speed (0.47 m/s, SD 0.26) than non-fallers (0.74, SD 0.36; P=0.009), less walking endurance (64.92 m, SD 40.47) than non-fallers (102.77, SD 50.07, P=0.011), and more impaired balance and motor control in lower extremities as measured by median paretic, non-paretic and
combined lower extremity Step Test scores (paretic 0, IQR 0, 5) than non-fallers (paretic LE 7, IQR 5, 11; P=0.001). There were no significant differences found between fallers and non-fallers for severity of depression, pre-stroke activity levels, and lower extremity functional strength (Table A4.4).
Fallers had greater disability level and dependence in daily activities than non-fallers with significant differences in the number of participants with mRS score of 1 (“no significant disability”, mean of 2 fallers and 10 non-fallers; P=0.037) and mRS score of 4 (“moderately severe disability”, mean of 10 fallers and 4 non-fallers; P=0.037). There were no significant differences between fallers and non-fallers for the admission AM-PAC, admission and discharge total FIM, motor FIM subscale, and cognitive FIM subscale and discharge mRS levels 2 and 3 (Table A4.5). During the obstacle-crossing task at hospital discharge, compared to non-fallers, fallers had significantly slower pre-obstacle gait speed (0.40 m/s, IQR 0.19, 0.59 versus 0.67 m/s, IQR 0.37, 1.09; P=0.002). In support of our hypothesis, fallers had slower obstacle-crossing gait speed (0.25 m/s, IQR 0.12, 0.49 versus 0.59 m/s, IQR 0.27, 0.93; P=0.009) compared to non-fallers. There was no significant difference between fallers and non-fallers for the magnitude of change between pre- obstacle gait speed and obstacle-crossing gait speed. Compared to non-fallers, fallers had shorter pre-obstacle step length (35.93 cm, SD 13.14 versus 46.29 cm, SD 18.17; P=0.036), larger coefficient of variation (31.48%; IQR 20.93, 53.23 versus 14.26%, IQR 9.07, 27.93; P=0.028), and greater step length asymmetry (21.50%; IQR 14.94, 55.31 versus 7.76%, IQR 5.85, 13.74; P=0.001). The difference in pre-obstacle step length, step length variability, and step length asymmetry between fallers and non-fallers was driven by the late pre-obstacle phase. Lastly, fallers had shorter pre-obstacle stride length (70.64 cm, SD 64.33) than non-fallers (93.53 cm, SD 34.54; P=0.017), which was consistent for both the early pre-obstacle (P=0.015) and late pre-obstacle