LAFERRIER et al. Mobility and AT in servicemembers with major traumatic limb loss
asked a broad range of questions on the number and types of prostheses ever received, currently used, rejected, or abandoned. Reasons for abandoning prostheses were also assessed. Wheelchair use and use of other mobility assis- tive devices were assessed. Survey questions on prosthe- ses and assistive-device use and abandonment were adapted from the Houghton Scale . Additional ques- tions were asked of survey participants who had aban- doned all prostheses and were using wheelchairs, including time until prostheses abandonment, reasons for abandonment, and years of wheelchair use. Questions on satisfaction with prostheses were adapted from the Pros- thesis Evaluation Questionnaire and the Orthotics and Prosthetic Users’ Survey [13–14]. Detailed analyses of the types of prosthetic-device use, rejection, and replace- ment patterns and the satisfaction with prosthesis and ser- vices are addressed in other articles in this issue [11,15– 17].
We examined other factors that may be associated with the use of mobility ATs, including demographic characteristics, comorbidities, quality of life, health sta- tus, combat-associated injuries, ambulatory function, and level of limb loss. Data on comorbidities included the presence of arthritis, depression, posttraumatic stress dis- order (PTSD), traumatic brain injury (TBI), phantom pain, residual-limb pain, chronic back pain, migraines, and stroke. Types of combat-associated injuries were also assessed and are described in detail by Epstein et al. in this issue . Self-rated quality of life and self-rated health status were rated as “excellent,” “very good,” “good,” “fair,” or “poor.” Cumulative trauma disorder, or worn leg syndrome, included arthritis, joint pain, heel pain, or plantar fasciitis on the contralateral limb. The number of surgeries before and after the initial amputa- tion was assessed. The survey collected data on seven graded levels of mobility function. For our article, we grouped mobility function into three levels: (1) nonam- bulatory (cannot walk), (2) ambulatory (household and community walkers), and (3) highly active (low- to high- impact recreational activities). The original survey col- lected data on 14 different levels of limb loss from shoul- der to partial foot amputations. Here, we focus on three different groups of lower-limb loss: unilateral lower limb, bilateral lower limb, and multiple limbs, including at least one lower limb (± upper limbs). For those with bilateral lower- and other multiple-limb loss, each limb
was analyzed separately, because each limb may have different prosthetic-device rejection and use patterns. We excluded upper-limb loss levels (unilateral upper-limb and bilateral upper-limb loss).
To describe univariate, bivariate, and multivariate findings, we analyzed the survey data using Stata 9.2 (StataCorp; College Station, Texas). For univariate analy- ses, statistical significance is based on Chi-square (cate- gorical data), Mann-Whitney U test (ordinal data), Student t-test (continuous data), and Fisher exact test, if cell size was <5. The level of significance is for a two- sided p < 0.05. Variables significant in univariate analy- ses are tested in logistic regression multivariate models. The outcome for the model is a bivariate outcome pre- dicting any wheelchair use (sole or supplementary) com- pared with no current use of wheelchairs. To avoid overfitting the model, we added variables significant in univariate analyses using forward stepwise selection based on the log likelihood ratio and significance of the coefficient. We compared the new model with the previ- ous model using the log likelihood ratio Chi-square test and kept the variable in the model if p < 0.05. The vari- able was removed from the model if p > 0.05 and if it was not a confounding factor. We also assessed potential interactions using the log likelihood ratio. Goodness of fit of the final model is assessed with the Hosmer-Lemeshow test statistic. A value of p > 0.05 indicates a well-fitted model [19–20].
Vietnam and OIF/OEF Groups
Of the 245 participants from the Vietnam group, data on mobility AT use and abandonment were collected from 178 participants with unilateral lower-limb loss, 50 with bilateral lower-limb loss (100 limbs), and 17 with multiple-limb loss (including at least one lower limb [41 limbs]), for a total of 319 limbs. The multiple-limb-loss group in the Vietnam group includes three subgroups:
loss of one upper and one lower limb (total 20 limbs),
loss of two upper limbs and one lower limb (total
6 limbs), and (3) loss of one upper and two lower limbs (total 15 limbs).
Of the 226 participants from the OIF/OEF group, 172 participants had unilateral lower-limb loss, 42 had