If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Human Engineering Research Laboratories, Veterans Affairs Pittsburgh Healthcare System, Pittsburgh, PADepartment of Physical Medicine and Rehabilitation, University of Pittsburgh, Pittsburgh, PA
Human Engineering Research Laboratories, Veterans Affairs Pittsburgh Healthcare System, Pittsburgh, PADepartment of Bioengineering, University of Pittsburgh, Pittsburgh, PA
Human Engineering Research Laboratories, Veterans Affairs Pittsburgh Healthcare System, Pittsburgh, PADepartment of Rehabilitation Science and Technology, University of Pittsburgh, Pittsburgh, PA
Human Engineering Research Laboratories, Veterans Affairs Pittsburgh Healthcare System, Pittsburgh, PADepartment of Physical Medicine and Rehabilitation, University of Pittsburgh, Pittsburgh, PADepartment of Bioengineering, University of Pittsburgh, Pittsburgh, PADepartment of Rehabilitation Science and Technology, University of Pittsburgh, Pittsburgh, PA
To determine the efficacy of a web-based transfer training module at improving transfer technique across 3 groups: web-based training, in-person training (current standard of practice), and a waitlist control group (WLCG); and secondarily, to determine subject factors that can be used to predict improvements in transfer ability after training.
Design
Randomized controlled trials.
Setting
Summer and winter sporting events for disabled veterans.
Participants
A convenience sample (N=71) of manual and power wheelchair users who could transfer independently.
Interventions
An individualized, in-person transfer training session or a web-based transfer training module. The WLCG received the web training at their follow-up visit.
Main Outcome Measure
Transfer Assessment Instrument (TAI) part 1 score was used to assess transfers at baseline, skill acquisition immediately posttraining, and skill retention after a 1- to 2-day follow-up period.
Results
The in-person and web-based training groups improved their median (interquartile range) TAI scores from 7.98 (7.18–8.46) to 9.13 (8.57–9.58; P<.01), and from 7.14 (6.15–7.86) to 9.23 (8.46–9.82; P<.01), respectively, compared with the WLCG that had a median score of 7.69 for both assessments (baseline, 6.15–8.46; follow-up control, 5.83–8.46). Participants retained improvements at follow-up (P>.05). A lower initial TAI score was found to be the only significant predictor of a larger percent change in TAI score after receiving training.
Conclusions
Transfer training can improve technique with changes retained within a short follow-up window, even among experienced wheelchair users. Web-based transfer training demonstrated comparable improvements to in-person training. With almost half of the United States population consulting online resources before a health care professional, web-based training may be an effective method to increase knowledge translation.
The average wheelchair user performs 15 to 20 transfers per day, resulting in significant stress to the upper limb, which is placed in positions known to increase the risk for shoulder and wrist injuries.
Paralyzed Veterans of America Consortium for Spinal Cord Medicine Preservation of upper limb function following spinal cord injury: a clinical practice guideline for health-care professionals.
Comparison of peak shoulder and elbow mechanical loads during weight-relief lifts and sitting pivot transfers among manual wheelchair users with spinal cord injury.
Because of the inability to rest an injured upper extremity without compromising independence, the onset of upper extremity pain and injury may lead to social isolation, dependence, and high medical expenditures in wheelchair users.
Association between mobility, participation, and wheelchair-related factors in long-term care residents who use wheelchairs as their primary means of mobility.
Transfers result in greater loads through the arm than pressure relief techniques and wheelchair propulsion. Therefore, it is imperative that wheelchair users perform appropriate and efficient transfers to minimize their risk of injury.
Relationship of physical therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project.
less than one quarter of individuals with thoracic-level spinal cord injuries (SCIs) report the ability to transfer independently at the time of discharge.
found that 63% did not use a proper handgrip, 50% did not set up their wheelchair at a proper angle, and 58% transferred without removing barriers. Deficits in transfer skills have been associated with nonergonomic biomechanics and measures of pain and pathology in the upper extremities,
Paralyzed Veterans of America Consortium for Spinal Cord Medicine Preservation of upper limb function following spinal cord injury: a clinical practice guideline for health-care professionals.
highlighting the continued need for utilization of appropriate transfer training techniques.
The Transfer Assessment Instrument (TAI) is an outcome measure that provides a standardized approach to allow clinicians to objectively and quantifiably evaluate the performance of a transfer.
Transfer training that results in increased TAI scores (better transfer ergonomics) has been associated with improvements in biomechanics of the upper extremity and trunk, which may reduce the risk of injury over time.
Current transfer evaluation and training methods rely on in-person interaction. In-person training offers the benefit of real-time feedback from a trainer. However, many physical and occupational therapists receive limited training in teaching appropriate transfer technique. Additionally, access to these medical professionals is not always possible, especially because of the mobility and participation barriers facing wheelchair users.
Only about 42% of wheelchair users claimed to have learned essential mobility skills from a professional, while the remaining 58% learned the skills on their own, from a peer, or never learned the skills.
Trust and sources of health information: the impact of the Internet and its implications for health care providers: findings from the first Health Information National Trends Survey.
using the web may be a way to disseminate transfer materials to end users more consistently and quickly. Internet access is a barrier to online training, yet with the use of smartphones and other mobile technologies on the rise it may be an additional way to reach a portion of those who would benefit from additional intervention.
comparing web-based to non–web-based training found comparable effect sizes and improved outcomes for both behavioral and educational interventions. Therefore, using an online medium may be an additional strategy for improving access to transfer training in wheelchair users.
The primary focus of this study was to determine the efficacy of a web-based transfer training module on the performance of independent wheelchair transfers immediately after training and after a 1- to 2-day delay. We hypothesized that after receiving web-based training, TAI scores would be comparable to those of a group that received in-person training and better than those of a group that received no training. To determine where resources should be the most focused, we also investigated the relationship between subject characteristics and change in TAI scores.
Methods
Participants
The study was approved by the University of PIttsburgh and the VA Pittsburgh Healthcare System Institutional Review Boards, Clinical Trials Record 15030691. Before beginning the study protocol, all participants provided informed consent. A convenience sample of individuals was enrolled at the 2015 and 2016 National Veterans Wheelchair Games (NVWG), as well as the 2016 National Disabled Veterans Winter Sports Clinic. Qualified participants were older than 18 years, used a wheelchair for their primary means of mobility (≥40h/wk), could independently transfer to and from a surface within 30 seconds, and spoke English. Participants were excluded if they had arm pain that limited their ability to transfer, bear weight through their arms, or used their legs while transferring. Participants were also excluded if they had any of the following: a neurologic condition that could impair learning, a history of heart or lung conditions that could be worsened by performing up to 6 back and forth transfers, or a history of pressure ulcers within the last 3 months.
To compare a web-based intervention to the current standard of practice, participants from the 2015 NVWG were randomly assigned to either in-person or web-based transfer training groups. To compare the efficacy of web-based training to no intervention, the remaining participants were randomly assigned to either web-based training or a waitlist control group (WLCG). At the 2016 NVWG we allowed participants to re-enroll; however, these individuals were excluded from data analysis. Randomization codes were generated before the start of the study for a 1:1 allocation ratio and stored in an Excel spreadsheet.a Randomization was completed using permutated blocks of 2 or 4. Raters and participants were not blinded to group assignment.
All groups were assessed at baseline and after a 1- to 2-day delay (fig 1). The intervention groups were also assessed immediately after training to evaluate skill acquisition, with the assessment after a 1- to 2-day delay evaluating skill retention. The WLCG received the web training after a 1- to 2-day delay, with an assessment before training (control condition) and immediately after training. Participants were asked to provide any feedback at the end of the study, and comments were recorded by study staff.
Fig 1Flowchart depicting the order of assessments and training for each of the 3 groups when repeated subjects were excluded.
Power analyses were performed based on a repeated-measures analysis of variance with 2-sided tests and a confidence level of .05, using previous data on in-person transfer training based on clinical practice guidelines.
Compared with the standard of care, subjects who received training had a mean ± SD TAI score of 8.7±1.2 versus 7.4±0.8, which leads to an estimated effect size of f=0.5. With this effect size, we would have at least 80% power to detect significant differences between the groups, with a total sample size of 38 subjects.
Assessments
All participants completed a demographic questionnaire regarding the participants' age, sex, race, diagnosis, hand and transfer preferences, use of assistive devices, and wheelchair type. At each assessment point, a trained rater evaluated the transfers using the TAI version 3.0. Because of the high correlation between parts 1 and 2 and the overlap of some components, only the TAI part 1 (supplemental appendix S1, available online only at http://www.archives-pmr.org/) was used in this study, with item 15 omitted because it relates to assisted transfers.
The TAI evaluates wheelchair preparation, body setup, and flight phases of the transfer. It has good face, content, and construct validity for community-dwelling mobility device users,
Each of the items in the TAI is scored “yes” (1 point), “no” (0 points), or “not applicable” (item not included in scoring), resulting in a minimum score of 0 and a maximum score of 10.
The in-person training consisted of an approximately 60-minute individualized transfer training session taught by a trained physical therapist to focus on each participant's transfer skill deficits. The training included 3 main components: (1) description and demonstration of all the component transfer skills as defined by the TAI; (2) discussion of the participant's deficits in skills, based on baseline TAI scores; and (3) blocked practice of component skills with knowledge of performance feedback. This training has been previously described and elicited improvements in transfer biomechanics (reduced loading and improved upper extremity positioning) among individuals with transfer deficits.
The web-based training group completed an approximately 60-minute online module based on the principles of the in-person training. The module is designed to educate participants about the impact of poor transfer techniques, as well as the value of proper technique. With the use of descriptions, pictures, and videos, the phases of a transfer are explained and demonstrated, while emphasizing the rationale behind each technique. The module takes approximately 1 hour to complete. Greater detail and excerpts from the training can be found in supplemental appendix S2 (available online only at http://www.archives-pmr.org/).
Analysis
SPSS Statistics 24.0b was used for all statistical analysis. The data were examined for normality using the Shapiro-Wilk test. Descriptive statistics were calculated and reported. The Kruskal-Wallis test was used to evaluate baseline differences between groups for noncategorical variables, while the Fisher exact test was used for categorical variables. Friedman tests were used to assess within-group changes in TAI scores, while Kruskal-Wallis tests were used for between-group differences. Dunn's test was performed if significance (P<.05) was indicated.
A multiple linear regression model was used to determine which factors may be useful in predicting the percent change in TAI score immediately after web-based training. The Spearman rank correlation coefficient between each factor and the percent change in TAI score was used to determine which noncategorical variables should be included in the regression. Similarly, the Kruskal-Wallis test was used to determine the significance of the percent change in TAI scores when grouped by the nominal variables; Mann-Whitney tests were used to determine differences between subcategories. A significance value of .20 was used for the aforementioned tests to ensure that any potentially influential factors in the regression were included, while the significance level of the regression was set to .05. The coding scheme for the regression factors is described in supplemental appendix S3 (available online only at http://www.archives-pmr.org/).
Results
Participants
Figure 2 describes enrollment and attrition in the study; data were analyzed for 71 participants. Eleven individuals repeated the study and were excluded from analysis. There were no injuries or incidents reported. Subject characteristics are presented in table 1. Sex distribution was the only significant difference between the intervention groups; however, including sex as a covariate did not alter the results of the TAI analysis. The percentage of minority participants was the only statistical difference between those who did not complete all the assessments of training and those who completed the whole study (supplemental appendix S4, available online only at http://www.archives-pmr.org/).
Fig 2Consolidated Standards of Reporting Trials (CONSORT) flowchart of study enrollment and retention. Abbreviation: NDVWSC, National Disabled Veterans Winter Sports Clinic.
NOTE. Values are mean (interquartile range), n (%), or as otherwise indicated. All variables were not normally distributed, so the median (interquartile range) is presented.
The median TAI scores are shown in figure 3. Thirty-six (50.7%) of the 71 participants had a baseline TAI score <7.36, the cutoff established by Tsai et al
indicating the participant might benefit from a training intervention. Immediately after training, participants significantly improved in their transfer technique (P<.01). Compared with the WLCG, at TAI 2 both intervention groups had significantly higher TAI scores (P<.01). The web and in-person groups also maintained improvements in TAI scores after the follow-up period, evidenced by no significant differences between TAI 2 and TAI 3 (P>.99 and P=.79, respectively).
Fig 3Box-and-whisker plots of TAI scores for each assessment. Median values are represented by horizontal lines, while boxes represent interquartile range (25th and 75th quartiles). T bars (whiskers) represent values that are 1.5 times the interquartile range. Outliers are shown as circles. Between group differences: ∗WLCG TAI 2 scores are significantly different than those of both intervention groups (P<.01). Within group differences: †Differences between TAI 1 and TAI 2 (P<.01). ‡Differences between TAI 1 and TAI 3 (P=.04, P<.01, and P=.01 for in-person training, web training, and WLCG, respectively). §Differences between TAI 2 and TAI 3 (P<.01).
Only 11 participants (15.5%) had a TAI score <7.36 after all received training. For their final TAI assessment, most of those participants failed the following TAI items: placing feet in a stable position before the transfer (n=7), using a correct handgrip for the leading (n=10) and trailing (n=7) extremities, and using the head-hips relationship (n=7). Qualitative participant feedback with rationale for their technique can be found in supplemental appendix S5 (available online only at http://www.archives-pmr.org/).
Regression
Four variables were found to be potentially significant in predicting the percent change in TAI score after training (table 2). Positive changes in TAI score were associated with a lower baseline TAI, a minority race, a diagnosis of multiple sclerosis (MS), and those who always transferred leading with the same arm. Only the baseline TAI score was found to be significant in the regression model predicting percent change in TAI score immediately after training (table 3).
Table 2Categorical factors to be entered into the regression model to predict percent change in TAI scores after training
NOTE. The overall model was significant (adjusted R2=.39, F5,54=8.52, P<.01).
Abbreviations: B, regression coefficients; B SE, SE of the unstandardized regression coefficients; β, beta coefficients; CI, confidence interval; Sig., significance.
The results supported the hypothesis that web-based transfer training would produce comparable improvements to in-person training in skill acquisition and retention as measured by TAI scores. In-person training is currently the standard of practice for transfer training
suggest that an online training module could produce similarly positive results. Online education has the potential to reduce barriers faced by wheelchair users, including program and transportation costs, as well as wheelchair accessibility difficulties.
have shown that nonindividualized web-based training can be beneficial in improving health outcomes in a real-life environment, which encourages the reproduction of web-based transfer training improvements in a nonresearch setting. However, without an assessment by a trained clinician, it is challenging to ensure participants are properly learning and carrying taught techniques through online learning; remote web-based assessment may be a future direction of research to address this point.
Improvement in TAI scores for the training groups was maintained after a few days, suggesting that improved transfer skills are present immediately after training and retained. It is possible that improvements could revert to baseline if skills are not reinforced; providing open access to transfer training materials for wheelchair users could help in skill retention.
Neither age nor years since diagnosis was found to be a significant predictor of percent change in TAI score in the regression model. These findings suggest that older adults and experienced wheelchair users can still benefit from training.
Our study found that 54.3% of users did not alternate the leading arm; this may increase their risk for impingement and rotator cuff–related dysfunctions because of uneven loading on their upper extremities.
Paralyzed Veterans of America Consortium for Spinal Cord Medicine Preservation of upper limb function following spinal cord injury: a clinical practice guideline for health-care professionals.
Participants with a higher baseline TAI score were not as likely to improve their scores with training as compared with those with lower baseline scores. This finding makes intuitive sense, as a person with a high initial score does not have as much room to improve. Participants of a minority race had significantly greater percent improvements in TAI score than those of the white race. Previous studies
Relationship of physical therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project.
have found that minorities are more likely to have a discharge location other than home, suggesting the necessity for additional treatment, increased medical costs, and decreased independence and quality of life. Knowing that transfer skills are related to overall independence among wheelchair users,
Relationship of physical therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project.
providing those most likely to require prolonged care with additional tools, such as the web-based transfer training module, may have a significant impact on the quality of life of those individuals.
Participants with the diagnosis of amputation, MS, or “other” had the lowest initial TAI scores; however, they also had greater percent improvements than the participants with SCI. A study of individuals with MS by Iezzoni et al
found that only 41.1% of power mobility device users and 17.7% of manual wheelchair users received any mobility-related training when they began using a wheelchair or scooter. Similarly, a study
of veterans with amputations who either used prostheses or wheeled mobility found that only 18% were formally instructed in the use of a manual wheelchair and propulsion. The lack of wheelchair skills training, including transfers, for people with MS and amputation likely contributed to the low TAI 1 scores, and the greater improvements after receiving training. By contrast, 42% of individuals with SCI are estimated to have received formal wheeled mobility training.
This study reinforced the necessity that all wheelchair users should receive training to improve their transfer technique, especially those with non-SCI diagnoses.
Study limitations
This study may have been affected by a sample bias secondary to enrollment of primarily active subjects who are participating in wheelchair sporting events. Wheelchair users who participate in athletics are more likely to be highly adept in wheelchair skills,
so it is possible that the results from this population may not be generalizable to all wheelchair users. However, this bias would most likely lead to an underestimation of impact. The in-person training group only consisted of participants with paraplegia. Although that distribution of diagnoses was not found to be significantly different from the other groups, it may have contributed to the higher initial TAI score of that group. Since participants with other diagnoses improved with training more so than participants with paraplegia, the effect of in-person training might have also been underestimated. Individuals who transferred with the use of their legs were excluded from this study; this limits the generalizability of our findings to individuals who use a sitting pivot or lateral scoot transfer technique. Future studies could also evaluate the impact of transfer training on individuals who require assistance to transfer, as well as the role of wheelchair setup.
Because of the short duration of the NVWG and National Disabled Veterans Winter Sports Clinic, the follow-up period to skill retention was limited. To fully understand how this training affects transfers, and thus the resulting effects on injury risk and quality of life, a longer follow-up period is needed. The inclusion of individuals who already completed the training without a separate randomization led to unbalanced groups, which may have influenced study results. The web-based training was conducted in a somewhat artificial setting, as subjects were reimbursed for their time and completed the module from a supplied computer. Additionally, it was not possible to achieve rater blinding during data collection, which may have biased the results. It is essential that web-based training programs be tested in home settings absent of investigators and other motivating factors to evaluate the true effect of the training.
Conclusions
Both web-based and in-person transfer skills training produced improvements in transfer technique among independent wheelchair users, compared with a group that had not received training. Poor baseline transfer methods were related to greater improvements in transfer technique after training. Using the web to provide access to transfer training materials may decrease the barriers to training and improve the quality of life of many wheelchair users.
Suppliers
a.
Excel spreadsheet; Microsoft Corp.
b.
SPSS Statistics 24.0 software; IBM Corp.
Supplemental Appendix S1 Transfer Assessment Instrument, Part 1
Supplemental Appendix S2 Web-Based Transfer Training Module Description
The module was developed as an alternative mechanism for presenting the content of the in-person training program by a group of clinicians, engineers, and rehabilitation scientists with feedback from wheelchair users. Additionally, education material development experts at the American Institutes for Research completed an external review, providing feedback on content, navigation, and accessibility of the language.
The module starts with a welcome landing page that orients the user to the 4 sections of the training: (1) description of upper limb pain and injury and how they are related to transfers; (2) introduction to the 3 phases of a transfer; (3) techniques that can be used during each phase to reduce stress on the upper limbs; and (4) methods to adapt the techniques for more challenging transfers such as car, shower, toilet, and floor transfers. Each transfer phase section includes an overview of key factors, a video describing components of correct technique for that phase, pages for each of these components with evidence-based rationale for the recommendations, videos and images of users demonstrating correct/incorrect techniques, and a page prompting the user to practice the techniques with bulleted key points. A quiz is completed at the end of each transfer phase section to reinforce the participants' understanding, and explanations are provided for incorrectly answered questions. A score of less than 75% directs the user back to the start of the section and requires them to retake the quiz before advancing.
Participants were provided with a laptop computer with a browser open to the web training and provided with access through either a touchpad or mouse. A study investigator explained how to navigate the training through the next/back buttons, listed the main sections of the training, and explained there would be 3 quizzes. Participants were provided with a mat table, shower chair, and transfer board for use during practice.
The full online transfer skills training module can be found at: http://www.upmc-sci.pitt.edu/book/independent-transfers-training. Representative screen shots can be found below of the welcome landing page, the body setup overview video, a component of correct technique during the flight phase (head-hips technique), a quiz question with feedback provided, and a prompt page for practicing correct flight and landing with navigation links shown at the bottom of the page.
0, Always uses the nondominant extremity to lead transfers; 1, always uses the dominant extremity to lead transfers; 2, uses either extremity to lead transfers
Demographic and TAI scores for individuals who did not complete the training or all of the TAI assessments. Each variable combined the groups and was compared with the rest of the study population that completed all the TAI assessments, using Mann-Whitney tests or the Fisher exact test.
Tabled
1
Demographics
In-Person Training (n=1)
Web Training (n=5)
WLCG (n=2)
P
TAI 1
7.86
7.14 (6.92–7.69)
6.54 (5.38–7.69)
.65
TAI 2
7.62
8.46 (8.46–8.57)
6.54 (5.38–7.69)
.22
Age (y)
66.0
58 (51.0–63.0)
37.5 (26.0–49.0)
.87
Sex: Female
1 (100)
0 (0)
0 (0)
.60
Race
.03
Minority
0 (0)
2 (40.0)
0 (0)
Diagnosis
.29
Paraplegia
1 (100)
2 (50.0)
1 (50.0)
Tetraplegia
0 (0)
1 (25.0)
0 (0)
Amputation
0 (0)
0 (0)
0 (0)
Multiple sclerosis
0 (0)
0 (0)
1 (50.0)
Other
0 (0)
1 (25.0)
0 (0)
Years since diagnosis
36.9
8.3 (7.0–11.0)
11.0 (7.0–20.9)
.52
Hand dominance
.72
Left
0 (0)
0 (0)
0 (0)
Right
1 (100)
5 (100)
2 (100)
Both/either
0 (0)
0 (0)
0 (0)
Transfer lead arm
.55
Left
0 (0)
2 (40.0)
0 (0)
Right
1 (100)
1 (20.0)
2 (100)
Both/either
0 (0)
2 (40.0)
0 (0)
AD
.35
Use an AD ≥1 transfer/d
0 (0)
0 (0)
0 (0)
Wheelchair type
.61
Manual wheelchair users
1 (100)
4 (80.0)
1 (50.0)
Level transfers (no.)
10.0
13.0 (5.0–15.0)
2.8 (2.0–3.5)
.85
Nonlevel transfers (no.)
3.0
5.0 (1.0–6.0)
2.5 (1.0–4.0)
.70
NOTE. Values are median, median (IQR), n (%), or as otherwise indicated.
Supplemental Appendix S5 Qualitative Participant Feedback on Training
Participants noted that they learned new skills with respect to how to set up their body for the transfer, how to achieve a smooth transfer, and how to improve toilet transfers. Many participants also reported being introduced to the head-hips technique for the first time. Additionally, they noted that watching videos of their peers completing transfers with both proper and improper technique helped them to refine their own techniques. Participants also articulated the motivation behind implementing changes in their technique such as decreasing numbness and tingling in the hands secondary to carpal tunnel syndrome.
Several participants commented on their preferences for transferring, as they differed from techniques taught in the training (n=10). Five preferred the use of a fist because of comfort, the need for added height, and feeling that a handgrip would not work because of worry that their hand would slip off the edge of the cushion, the wheel being too far away, and feeling the swing-away armrest would not support the user's body weight. Seven people reported preferring one foot on the ground and one on the footplate for either improved leverage or control of the trailing leg. One person felt it was not necessary to remove his clothing guard since it was lower than the wheel. Two people felt unsafe scooting forward because of a significant dump and a fear of falling. Upper extremity and trunk weakness or pathologies were reported by 3 of the 11 participants who did not achieve proficient transfers (TAI score >7.36) as barriers to modifying their transfer techniques.
Comparison of peak shoulder and elbow mechanical loads during weight-relief lifts and sitting pivot transfers among manual wheelchair users with spinal cord injury.
Association between mobility, participation, and wheelchair-related factors in long-term care residents who use wheelchairs as their primary means of mobility.
Relationship of physical therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project.
Trust and sources of health information: the impact of the Internet and its implications for health care providers: findings from the first Health Information National Trends Survey.
Supported in part by the Paralyzed Veterans of America Research Foundation, and by the Administration for Community Living (ACL), National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR) (grant no. 90SI501). NIDILRR is a Center within the ACL within the Department of Health and Human Services (HHS). The contents of this paper do not necessarily represent the policy of NIDILRR, ACL, or HHS, and you should not assume endorsement by the U.S. Government.