Abstract
Background Obstacle crossing is impaired in people following stroke. It is not known whether people with stroke who fail an obstacle crossing task have more falls or whether the gait adjustments used to cross an obstacle differ from those used by people who pass the task.
Objective The purposes of this study were (1) to identify whether a group of people with stroke who failed an obstacle crossing task had a greater incidence of falling and (2) to determine whether people who fail an obstacle crossing task utilize different gait adjustments.
Design This was a prospective, observational study.
Methods Thirty-two participants with a recent stroke were recruited. Participants walked at self-selected speed and stepped over a 4-cm-high obstacle. Performance was rated as pass or fail, and spatiotemporal, center of mass (COM), and center of pressure (COP) data were collected. Prospective falls data were recorded for 20 participants over a 6-month period.
Results The incidence of fallers was significantly higher (incidence rate=0.833) in the group that failed the obstacle crossing task than in the group that passed the task (incidence rate=0.143). The group that failed the task had a slower walking speed and greater normalized separation between the trail heel (unaffected support limb) and COM as the affected lead toe cleared the obstacle. This group exhibited greater normalized times from affected lead toe clearance to landing, unaffected trail toe clearance to landing, and affected trail toe-off to toe clearance.
Limitations The sample size was small, and falls data were available for only 20 participants.
Conclusions Obstacle crossing is an important task to consider in people following stroke and may be useful in identifying those at risk of falls.
It is known that obstacle crossing performance is a difficult task for many people following stroke. Said et al1 found that 11 out of 24 people recovering from stroke made accidental foot contact or lost balance on at least 1 occasion during 12 attempts to step over a small obstacle. Investigations also have shown that people with stroke use different gait strategies to cross a 4-cm-high obstacle compared with individuals without impairments.2,3 People with stroke crossed the obstacle more slowly than those without impairments. The slower speed, however, did not account for all gait adjustments. Compared with individuals without impairments walking at matched speed, the gait adjustments made by people with stroke differed depending on which limb crossed the obstacle first. For example, the foot of the unaffected lead limb, affected trail limb, and unaffected trail limb were placed closer to the obstacle after crossing compared with participants without impairments at matched speed. Moreover, the body's center of mass (COM) was positioned closer to the heel of the support limb when the affected lead toe cleared the obstacle. This position would reduce the chance of losing balance forward if the lead limb inadvertently contacts the obstacle. These findings suggest that the gait adjustments during obstacle crossing are due to the stroke, and not simply secondary to reductions in gait speed following stroke.
To provide further evidence of the construct validity of the obstacle crossing task, the ability of the task to differentiate between different groups of people with stroke needs to be established. It would be of interest to determine whether obstacle crossing can differentiate between people with stroke who fall and people with stroke who do not fall. Falls are a major problem following stroke, with up to 70% of people with stroke falling after returning home.4–8 Many falls occur while walking,6,9,10 and there is evidence that people who are more active7 or less impaired11 after stroke are at greater risk of falls compared with those who are more dependent. This evidence suggests there is a need for tasks able to demonstrate deficits in higher-functioning people with stroke. Obstacle crossing has the potential to differentiate between fallers and nonfallers in people who can walk following stroke, as it provides a more challenging test of dynamic balance12–14 and locomotor control15–18 compared with level overground walking. To date, no studies have explored whether people with stroke who contact the obstacle or lose balance while crossing an obstacle have more falls than those who pass an obstacle crossing task. Establishing whether obstacle crossing performance differentiates between fallers and nonfallers following stroke would provide additional evidence that obstacle crossing is important to consider in people following stroke.
It also is not known whether the gait adjustments utilized by people with stroke who successfully clear an obstacle differ from the gait adjustments utilized by people who contact the obstacle or lose balance during the task. It is reasonable to propose that greater deviation from a “normal” gait pattern places people with stroke at risk of failing the task. Alternatively, some adjustments may reduce the risk of obstacle crossing failure. For example, it is reasonable to hypothesize that placement of the lead limb closer to the obstacle on landing (after the obstacle) will increase the risk of accidental heel contact. In contrast, reducing the distance between the COM and trail heel during affected lead toe clearance (ie, maintaining the COM within the base of support) may improve stability and reduce loss of balance. Understanding which gait modifications are associated with obstacle crossing success or failure will assist physical therapists with the development of strategies that will improve obstacle crossing performance in people with stroke.
The purpose of this study was twofold: (1) to determine whether the incidence of falling differs between people with stroke who fail an obstacle crossing task and those who pass and (2) to ascertain whether the gait adjustments utilized during obstacle crossing by people with stroke differ between those who fail and those who pass.
Method
This was a prospective, observational study. Ethics approval was obtained, and participants provided informed consent. Data were collected over 2 periods: March 2000 to February 2001 and April 2005 to May 2006. Data from all participants19 and participants recruited in the first time period2,3 have been reported previously.
Participants
Potential participants were identified by the treating physical therapists at Austin Health and La Trobe University. Thirty-two participants receiving physical therapy following a recent stroke and capable of walking 10 m without a gait aid or physical assistance were recruited. Participants had to be able to provide informed consent and follow instructions. Participants were excluded if they had other medical, musculoskeletal, or neurological conditions that might have an impact on walking. One participant had a previous stroke 8 years earlier with no residual neurological deficits (had returned to competitive tennis). One participant had no clinical history of stroke, but a computed tomography revealed an old lesion. This was the first stroke for the remaining 30 participants. Seventeen participants had left hemiplegia. The mean age of all participants was 62.6 years (SD=15.4), mean height was 168.5 cm (SD=9.3), and mean gait speed (no obstacle) was 81 cm/s (SD=36), and participants were tested a median of 53 days (interquartile range [IQR]=40) poststroke. No participants withdrew from the study. Additional data are provided in Table 1.
Characteristics of Participantsa
Apparatus
Data were collected in 2 laboratories. Data from the first laboratory (first test period) were recorded by a 6-camera VICON 512 3D motion system (Oxford Metrics Ltd, 14 Minns Estate, West Way, Oxford, OX2OJB, United Kingdom) and a Kistler forceplate (Performance System 9281B, Kistler Instrumente AG, Winterthur, Switzerland). Data from the second laboratory (second test period) were recorded by a 6- or 8-camera VICON 460 or 612 3D motion systems and an AMTI (Advanced Mechanical Technology Inc, Watertown, Massachusetts) forceplate. Data collection was overseen by the same investigator (C.M.S.) to ensure standardized procedures and data consistency. During the obstructed trials, a red balsa wood obstacle measuring 40 mm high, 1.5 mm thick, and 600 mm long was positioned after the forceplate, approximately 5 m from the starting position. The balsa wood obstacle was selected as previous work demonstrated it did not pose as great a safety threat as larger obstacles, but was large enough to challenge gait following stroke.1–3 Only 1 obstacle was selected, as additional obstacles would have increased the number of trials and potential for fatigue. Data processing utilized Vicon BodyBuilder versions 3.5 and 3.55 (build 136) (Oxford Metrics Ltd).
Procedure
Participants wore shorts, walking shoes, and any prescription eyewear. Anthropometric measurements were obtained (Vicon Plug-in Gait Product Guide) and used to calculate lower-limb joint centers.20,21 Twenty-one passive reflective markers were placed on the lower limbs, acromions, and obstacle as previously described.2 A static trial, in which knee markers were replaced with knee alignment devices (Oxford Metrics Ltd), allowed the identification of the flexion/extension knee axis and joint center. As a marker could not be placed on the toe, a series of static trials using a custom-made triangular device with 3 noncollinear markers and a known end point identified the distal point of the toe and heel on the sole of the shoe,2 allowing calculation of the distal point of the toe and heel in the walking trials.
All participants performed 4 unobstructed walking trials at self-selected speed to familiarize themselves with the testing situation. They then performed 8 trials with the obstacle. Eight trials provided sufficient opportunities for participants to lead with both limbs and to fail the task, as our previous study demonstrated fails were spread throughout a testing session.1 Participants were instructed to walk at self-selected speed and step over the obstacle without contacting it or losing balance. They were reminded to perform the task within their limits of safety and stop if they felt at risk. For safety, each participant was accompanied by a therapist, who walked behind and to the side of the individual, lightly holding a safety belt worn by participant. Assistance was only provided if required. Although the therapist's presence may have influenced performance, this assistance was an ethical requirement because of the risk of falls in this population. Participants rested between trials to minimize fatigue.
Performance on each trial was rated as successful or unsuccessful. A trial was classified as successful if the participant cleared the obstacle with both limbs and maintained balance. A trial was classified as unsuccessful if the participant contacted the obstacle with either limb or lost balance and required the therapist's assistance. Participants then were classified into 1 of 2 groups: pass or fail. Those who were successful on all trials were classified into the pass group. Those who were unsuccessful on one or more attempts were classified into the fail group.
The 20 participants who participated in the second testing period prospectively recorded any falls in a falls diary during the 6 months following testing. They were told that a fall was any event where they made unintentional contact with the ground or lower surface. Participants who had no falls were classified as nonfallers, and participants who had 1 or more falls were classified as fallers. Falls data were not obtained for the first data collection period (12 participants).
Gait Variables: Data Processing
For each participant, 1 trial leading with the affected limb and 1 trial leading with the unaffected limb were identified and analyzed. Trials were excluded from motion analysis if the participant required the therapist's assistance to maintain balance, as data may have been erroneous. The first trial with adequate data (minimal marker occlusion and clean forceplate strike) was selected. The trial selected was the first attempt to cross the obstacle with a given limb in 46% of trials analyzed. It has been reported previously that the gait pattern utilized by this sample of people with stroke to cross an obstacle is reasonably consistent over 3 attempts,19 so the selection of 1 trial for further analysis was justified.
Data from the first laboratory were filtered using a 3-point weighted average procedure provided by BodyBuilder. Data collected from the second laboratory were filtered using Woltring filtering routine with a predicted mean squared error value of 20. BodyBuilder was used to create COM and “virtual markers” on the shoe at the most distal point of the big toe and heel as reported previously.2,3 Center of pressure (COP) data also were obtained via BodyBuilder. Data then were exported to Microsoft Excel (Microsoft Corporation, Redmond, Washington) for further analysis.
Variable Selection
Variables that have already been shown to differentiate the obstacle crossing performance of people with stroke from controls who were healthy were collected.2,3 For example, obstacle crossing gait speed (calculated from lead heel contact pre-obstacle to trail heel contact post-obstacle) is known to be significantly slower following stroke.2 Table 2 lists other variables used in this study, and definitions for these variables are provided in the Figure. These variables were selected because they differentiate between the obstacle crossing performance of people with stroke and unimpaired people walking at matched speed.2,3 In addition, measurements of affected lead limb post-obstacle distance and affected trail limb vertical toe clearance were collected. Although differences in people with stroke have only been reported compared with people who are healthy walking at self-selected speed, these measures are of interest because reductions potentially increase the risk of obstacle contact.
Illustration of the phases of obstacle crossing. Solid line represents the lead limb (ie, the first limb to step over the obstacle). Dotted lines represent the trail limb. Phases identified are: (A) lead limb pre-obstacle distance, (B) trail limb pre-obstacle distance, (C) lead or trail limb toe clearance, (D) lead limb post-obstacle distance, and (E) trail limb post-obstacle distance. Pre-obstacle swing time is from toe-off to obstacle clearance. Post-obstacle swing time is from obstacle clearance to foot contact. COM=indicates the vertical projection of the center of mass at lead limb toe clearance, COP=indicates the position of the center of pressure at lead limb toe clearance. Adapted with permission of the American Physical Therapy Association from: Said CM, Goldie PA, Culham E, et al. Control of lead and trail limbs during obstacle crossing following stroke. Phys Ther. 2005;85:413–427.
Sample Size Estimation
It has previously been shown that 45% of people with stroke fall within 6 months of discharge from rehabilitation8 and that 1 in 3 people with stroke fail 1 or more trials on an obstacle crossing task.22 Assuming that 85% of people who fail will be classified as fallers and 20% of people who do not fail will be classified as fallers, a sample size of 20 would have 80% power to detect a significant difference (alpha=.05) between groups.
Statistical Analysis
All statistical analyses were undertaken using IBM SPSS Statistics version 19 for Windows (IBM Corporation, Armonk, New York). As falls increase with age, a Mann-Whitney U test was performed to determine whether people with stroke who fell were older than those who did not fall. To determine whether the incidence of fallers was higher in the fail group compared with the pass group, the incidence rate was calculated for each group and the incidence rate ratio was calculated. The Fisher exact test then was used to determine whether the incidence of falls differed between the 2 groups.
In addition to gait speed, 19 gait variables were identified as being of interest (Tab. 2). To reduce the number of comparisons, correlations between variables were examined. If a high correlation between variables was detected (r>.8), 1 variable was eliminated from further analysis, as shown in Table 2. All eliminated variables were highly correlated with affected lead limb crossing speed. In addition to affected lead limb crossing speed, 8 variables were included in further analysis, as shown in Table 2. As gait speed and leg length had an impact on spatial and temporal variables, data were normalized using the Froude number.23–27 Froude number (Fr) is calculated using the formula Fr = v2/gl, where v=velocity, g=gravity, and l=leg length. As the majority of data were not normally distributed, medians and interquartile ranges were calculated for the pass and fail groups. Mann-Whitney U tests then were performed to determine whether spatial and temporal variables differed between people who passed or failed the task. To reduce risk of type I errors, a Bonferroni adjustment corrected for the 9 tests resulting in a significance level of .006. To reduce risk of type II errors, results between the corrected and uncorrected significance level of .05 were interpreted as “suggestive of significance, but not definitive,”28(p7) allowing for the identification of areas that could be of interest for future investigation.29
Results
One participant was unable to complete all 8 obstructed trials due to fatigue. This participant had already failed an attempt to step over the obstacle and, therefore, was able to be classified and included in analysis. One participant died before the end of the 6-month follow-up period. This participant had recorded 2 falls and was able to be classified, so also was included in the analysis. Three participants did not lead with the unaffected limb for any trials, as indicated in Table 1; therefore, data for the unaffected lead limb were available for only 29 participants. There were no other missing data. One participant used a nonreciprocal gait pattern (the lead limb touched down before clearing the obstacle) for 1 trial leading with the affected limb and 1 trial with the unaffected limb, and 1 participant used a nonreciprocal gait pattern for 1 trial leading with the affected limb; however, data were available from other trials.
Ten of the 32 participants failed 1 or more attempts to cross the obstacle. Fails were spread throughout the session and could not be attributed to fatigue or task novelty. Seven of the 20 participants who completed falls diaries reported a fall and were classified as fallers (Tab. 3).1 It should be noted that 3 participants reported multiple falls, but having multiple falls did not alter their classification. The average age of participants who fell was 70.4 years (SD=8.7), compared with a mean age of 56.1 years (SD=15.4) in those who did not fall. The Mann-Whitney U test confirmed fallers were older than nonfallers (U=19.5, P=.039). The fallers' incidence rates were 0.143 for the pass group and 0.833 for the fail group, giving a faller incidence rate ratio of 5.83. The Fisher exact test confirmed that the fallers' incident rate was significantly higher for people in the fail group (P=.007).
Obstacle Crossing Performance (Pass/Fail) Versus Fall Status at 6 Months
Descriptive statistics for gait parameters for people who passed and failed the obstacle task are provide in Table 4. Crossing speed was significantly greater in people who passed the obstacle crossing task compared with those who failed (U=29.5, P=.001). When leading with the affected limb, the fail group had increased normalized lead (affected) post-obstacle swing time (U=35.0, P=.002), increased trail (unaffected) post-obstacle swing time (U=34.0, P=.002), and increased separation between the COM and the trail (stance) heel at lead toe clearance compared with people in the pass group (U=30.0, P=.001). When leading with the unaffected limb, the fail group had increased normalized trail (affected) pre-obstacle swing time compared with the pass group (U=24.0, P=.003). There was a trend for normalized trail (affected) toe clearance (U=41.0, P=.036) and the difference between the COM and COP at lead toe clearance to be increased in people who failed (U=24.0, P=.023); however, these results did not reach significance.
Gait Parameters for People With Stroke Who Passed and Failed an Obstacle Crossing Taska
Discussion
People with a subacute stroke who failed to safely step over a 4-cm-high obstacle on even 1 of 8 attempts had a higher incidence of fallers in the subsequent 6 months. Although the result was significant, a main limitation of the study is that falls data was only available for 20 participants. The results indicated that people who fell were older than those who did not fall. This finding was not surprising, given that associations between falls and age have been demonstrated in people with stroke7 and the general older population. It is possible that the difference in the incidence of fallers between people who failed and those who passed was influenced by age. However, studies using a similar protocol have reported no failures in older adults who were healthy.1,2 These studies suggest age alone does not lead to failure on an obstacle crossing task. Results should be prospectively verified in a larger sample, which would allow greater exploration of the interaction between stroke, age, obstacle crossing performance, and falls. Nonetheless, the results provide preliminary evidence that performance on an obstacle crossing task can differentiate between fallers and nonfallers following stroke.
Obstacle crossing may be a useful tool in predicting who is at risk of falls following stroke, particularly in people who are able to walk. There are multiple factors that differentiate between fallers and nonfallers living in the community following stroke, such as living with a spouse, poor health, time since first stroke, psychiatric problems, urinary incontinence, pain, motor impairment, history of falls, depression, disability, age, sensory impairment, visual impairment, and increased postural sway.5,7,11,30 People with stroke who had a fall or near fall in hospital, impaired upper-limb or lower-limb function, reduced balance (on the Berg Balance Scale or Step Test), or reduced independence in activities of daily living were more likely to have multiple falls after discharge from rehabilitation.4,8 However, as discussed earlier, risk factors may be different for people who are more mobile following stroke compared with a more dependent group. There is a need for tools to identify falls risk in people who are more functionally able, such as participants in our study who were able to walk unaided.
Furthermore, previous studies have focused on discriminating between those who had no falls or a single fall and those who had multiple falls following a stroke.4,8,30 Identification of people at risk of a single fall who may have more subtle deficits is more challenging than identifying those at risk of multiple falls. The obstacle crossing task was able to identify people at risk of even a single fall, which suggests it may be more discriminative for those at more mild risk of falls. There are other clinical tests used to assess walking and balance in high-functioning people with stroke, such as the Dynamic Gait Index31 or the mini-Balance Evaluation Systems test.32 These tests involve multiple tasks including obstacle crossing, although the obstacle (a shoe box) is substantially bigger than the obstacle in this study. It is not known whether these tests can differentiate between fallers and nonfallers poststroke. The results of this study provide a rationale for including tests with an obstacle crossing component in future studies evaluating falls risk in high-functioning people with stroke.
Several gait adjustments differed between people who passed and people who failed the obstacle crossing task. Although results do not indicate cause and effect, it is of interest to consider why certain movement patterns may be associated with failure. People with stroke who passed the obstacle crossing task crossed the obstacle more quickly than those who failed. People who walk slowly typically have greater impairments of balance and lower-limb strength,33 which may reduce their ability to make the adaptive gait adjustments required to safely negotiate an obstacle. Longer affected lead limb swing time from obstacle clearance to landing, when normalized for gait speed and leg length, was observed in people who failed. Similarly, people who failed had a longer affected trail limb swing time from toe-off to obstacle clearance. This observation may indicate greater difficulty controlling the affected limb during swing. People who failed also had increased normalized unaffected trail limb swing time from clearance to landing. This strategy would increase affected limb single support time, which may present an increased threat to balance control. There was a trend for people who failed to have higher normalized affected trail limb toe clearances. Higher toe clearances require greater modification of the swing limb trajectory, which places greater stress on the postural control system. Higher affected trail limb clearance than expected for a given gait speed may reflect caution, and be an attempt to reduce the chance of unwanted foot obstacle contact.
People who failed had greater separation between the COM and heel as the affected lead limb cleared the obstacle, once speed was taken into account. There also was a trend for people who failed to have greater separation between the COM and COP on the affected stance limb during unaffected lead limb clearance. Lead limb clearance is a particularly challenging phase in obstacle crossing,13 and maintaining the COM closer to the heel or COP would increase stability. Conversely, greater separation between the COM and the heel or COP would increase the risk of losing balance. Results of earlier research demonstrated that, when leading with the affected limb, the COM was positioned closer to the heel in people with stroke compared with individuals who were healthy.3 The authors hypothesized that this adaptation was a “safe strategy,” as it would reduce the risk of people with stroke losing balance. The previous study also showed that when people with stroke led with the unaffected limb, there was greater separation between the COM and the COP of the affected stance limb, which was hypothesized to increase instability. The results of this study provide additional support for the hypotheses that positioning the COM posteriorly and minimizing the separation between COM and COP may reduce loss of balance and risk of failure in people with stroke.
Once speed was accounted for, lead limb placement after the obstacle did not differ between people who failed and those who passed (Tab. 3). This finding was somewhat surprising, particularly given all contact failures were made by the lead limb on landing. One potential issue with normalizing for speed is that due to task demands, there is a limit as to how far people can reduce some spatial variables, even when walking very slowly. These participants had very large normalized values for post-obstacle distance, as reflected by the large IQRs shown in Table 4. This larger IQR may limit the ability of these variables to differentiate between those who failed and those who passed. There is evidence that placement of the lead limb closer to the obstacle on landing is partly related to obstacle crossing speed.2 This evidence is supported by our finding that a moderate correlation existed between crossing speed and affected and unaffected post-obstacle distance (r=.60, P<.001, and r=.75, P<.001, respectively). Therefore, although post-obstacle distance may not increase the odds of failing independent of gait speed, the reduction in post-obstacle distance associated with reduced crossing speed may contribute to an increased risk of lead heel contact with the obstacle. This finding may partly explain why a faster walking speed reduced the odds of failing on an obstacle crossing task. Another interesting finding was that 3 participants did not lead with the unaffected limb for any trials. Inspection of the characteristics of these participants does not provide any insight as to why participants did not lead with the unaffected limb. Two participants were relatively impaired, as illustrated by slow gait speeds (0.4 m/s and 0.2 m/s); however, 1 participant (participant 28) walked at a reasonable speed (0.8 m/s). The small number of participants limits further investigation of the contribution of other impairments to limb preference, but it would be of interest to further explore this issue in future studies.
Despite the small sample in this study, the results have clinical implications for the management of gait disorders following stroke. The results can be generalized only to people in the subacute phase of stroke; however, this population is more likely to be responsive to rehabilitation compared with survivors of chronic stroke. Participants had to be capable of following the testing procedure, so results may not be applicable to people with significant cognitive impairment. Despite these limitations, this is the first study to demonstrate that people who fail an obstacle crossing task are more likely to fall, and highlights the importance of considering obstacle crossing when retraining walking following stroke. The simplicity of classifying obstacle crossing as pass or fail means the task could be easily assessed in a clinical setting. Performance should be observed over several trials as fails do not always occur on the first trial and to provide an opportunity for the person to lead with both limbs. The results also provide preliminary evidence as to what features of obstacle crossing should be retrained. For example, it would be reasonable to investigate whether obstacle crossing improves as gait speed increases. Three of the remaining 6 movement deficits identified were associated with affected limb swing phase, so training focused on this phase of gait may be beneficial. Findings support the ongoing investigation of obstacle crossing following stroke, and exploration of issues such as how obstacle crossing changes with rehabilitation and whether specific training programs improve obstacle crossing following stroke is warranted.
The Bottom Line
What do we already know about this topic?
We know that some people with stroke have difficulty stepping over obstacles and that people with stroke utilize a different walking pattern compared with people without stroke. This walking pattern is different for all people with stroke.
What new information does this study offer?
The incidence of falls was nearly 6 times higher in people with stroke who contacted an obstacle or lost balance during obstacle crossing compared with those who were successful. People who failed the task crossed the obstacle more slowly and used a different walking pattern compared with those who passed the task.
If you're a patient or a caregiver, what might these findings mean for you?
People with stroke who fail an obstacle crossing task may be at risk for falls. For those in the subacute phase of stroke who do not have major cognitive deficits, it is important for the physical therapist to consider obstacle crossing as part of walking retraining. The therapist should observe performance over several attempts, note whether obstacle crossing improves as gait speed increases, and focus on specific phases of gait. More research is needed to find out how obstacle crossing changes with rehabilitation and whether specific training programs improve obstacle crossing following stroke.
Footnotes
-
Dr Said and Dr Lythgo provided concept/idea/research design, writing, and data collection and analysis. Dr Said provided project management, fund procurement, and study participants. Dr Said and Dr Galea provided institutional liaisons. Dr Lythgo provided facilities/equipment and consultation (including review of manuscript before submission). The first data collection period was conducted in partial fulfillment of Dr Said's doctoral studies and was supervised by Dr Patricia Goldie and Professor Aftab Patla. The authors acknowledge the therapists who assisted in participant recruitment and the people who volunteered to participate in the study.
-
Study approval was provided by the Human Research Ethics Committees at Austin Health and La Trobe University (Austin Health Human Research Ethics Committee H99/00804, H2004/01939, La Trobe University Human Ethics Committee 99/148).
-
The data were presented, in part, at the Third Australian and New Zealand Falls Prevention Conference; October 12–14, 2008; Melbourne, Victoria, Australia; and at Australian Physiotherapy Association Conference Week; October 1–5, 2009; Sydney, New South Wales, Australia.
-
Dr Said received salary support from NHMRC Health Professional Training Fellowship Grant 310612 and NHMRC project grant 385002 and a Career Interruption Fellowship from the University of Melbourne. Financial support for the first data collection period was provided by the Victorian Branch Australian Physiotherapy Association, School of Physiotherapy, La Trobe University and La Trobe University Faculty of Health Sciences.
- Received May 15, 2012.
- Accepted October 9, 2012.
- © 2013 American Physical Therapy Association
References
Issue highlights
