The Effect of Rehabilitation in a Comprehensive Inpatient Rehabilitation Unit on Mobility Outcome After Dysvascular Lower Extremity Amputation

Abstract

Czerniecki JM, Turner AP, Williams RM, Hakimi KN, Norvell DC. The effect of rehabilitation in a comprehensive inpatient rehabilitation unit on mobility outcome after dysvascular lower extremity amputation. Arch Phys Med Rehabil 2012;93:1384–91.

Objectives:

To (1) compare the total volume of rehabilitation therapy for patients ever attending a comprehensive inpatient rehabilitation unit (CIRU) versus never during the 12 months after amputation; (2) determine whether rehabilitation in a CIRU at any time in the first year after amputation results in greater mobility success compared with other types of rehabilitation environments of care; and (3) determine for those patients treated in a CIRU, which specific patient characteristics were associated with improved mobility outcome.

Design:

Prospective cohort study.

Setting:

Two Veterans Affairs medical centers.

Participants:

Patients (N=199) with peripheral vascular disease or diabetes undergoing a first unilateral major amputation were screened for participation between September 2005 and December 2008. Among these, 113 (57%) met study criteria; of these, 72 (64%) participated.

Intervention:

Ever attending a CIRU versus never attending a CIRU in first 12 months after amputation.

Main Outcome Measures:

Number of rehabilitation therapy visits, Locomotor Capability Index scores, and mobility success.

Results:

The mean number of all therapy visits for patients ever attending a CIRU was significantly greater than that for those never attending over a 12-month period (48.6 vs 22.6; P=.001). Mean total time per any rehabilitation visit was .83±.27 hours for those ever attending and .60±.20 hours for those never attending (P<.001). Patients who ever were treated in a CIRU were 17% more likely to achieve mobility success than those who were not, controlling for amputation level, major depressive episode, alcohol use, social support, total number of rehabilitation visits, and hospital site (risk difference=.17; 95% confidence interval, .09–.25; P<.001).

Conclusions:

Rehabilitation in a CIRU resulted in improved mobility success for veterans undergoing major lower extremity amputation secondary to peripheral vascular disease or diabetes. Among those admitted to a CIRU, younger patients with greater social support, healthy weight, and without chronic obstructive pulmonary disease had the greatest probability of mobility success.

THERE IS A CRITICAL need to better understand the effectiveness of rehabilitation interventions on functional outcomes after lower extremity amputation (LEA) and to determine the patient characteristics associated with improved outcome. As reflected in a review article by Sansam et al,¹ most of the scientific investigations of outcome after LEA have not evaluated the comparative effects of different rehabilitation environments. The effect of rehabilitation treatment factors, in particular the environment of rehabilitation care, has only recently been studied.^2–4 To our knowledge, there are no studies that have evaluated patient factors that may be associated with better outcomes in a specific rehabilitation environment. Knowledge of these factors will help inform health care providers and payers about the necessary rehabilitation treatment required to optimize outcome for each patient, and to ensure that treatment is provided in the most cost-effective environment.

Most individuals with LEA will be discharged from the acute surgical ward to either home or to a skilled nursing facility (SNF). Between 1987 and 1997, 40% of dysvascular amputee patients were discharged home without home care, 37% were discharged to an SNF, 9% were discharged home with home care, and only 10% were discharged to a comprehensive inpatient rehabilitation unit (CIRU).⁵ Similarly, in a subsequent study³ of 1996 Medicare claims data, only 14% received inpatient rehabilitation. It is uncertain which environment of care promotes the best functional outcome and, therefore, it is unclear whether this distribution of discharge destinations is based on sound health care policy.

The provision of rehabilitation therapies may be provided in a spectrum of environments of care. Rehabilitation in a CIRU typically includes an individualized coordinated combination of therapies (eg, physical, occupational, speech-language, and recreational therapies) and care (eg, medicine, psychology, social work, nursing) that is designed to meet specific functional goals. This differs from rehabilitation therapies provided in SNFs or in the outpatient environment in terms of the volume, time course, and the extent of team communication and coordination. Parsing out various components of a rehabilitation intervention to quantify their efficacy is challenging and fails to accurately reflect the less tangible aspects of interdisciplinary care coordination seen to varying degrees during the rehabilitation process. Initial efforts to evaluate the effects of rehabilitation on amputee outcome have therefore used the environment of care (eg, CIRU vs other rehabilitation therapy) as a proxy for the sum of various rehabilitation interventions provided in a specific environment of care. Although the number of studies is limited, most have made use of existing administrative datasets that are inherently subject to important limitations. First, although the samples are often large, the available data are often limited to what has been previously collected and do not include key factors such as comorbidities, demographics, and psychosocial conditions. Second, these large administrative datasets also often fail to include functional outcome measures (eg, mobility), which have specific clinical utility and relevance for individuals with amputation. Finally, evaluating and controlling for the effects of rehabilitation therapy volume while comparing rehabilitation environments is critical but very difficult in existing databases.

That said, these large administrative database studies have shown that admission to a CIRU is associated with a number of positive short-term clinical outcomes, including improved survival, reduced non-amputation-related hospital admissions, reduced subsequent amputations, and an increased likelihood of prosthesis fitting.³ Stineman et al⁴ developed a conceptual model of rehabilitation intervention that organizes the rehabilitation intervention under the broad constructs of timing, place, and type of rehabilitation. This framework was used to evaluate 2 components of the rehabilitation intervention: (1) whether the provision of rehabilitation in a CIRU was associated with a significant improvement in mobility outcome in dysvascular lower extremity amputees compared with consultative rehabilitation on another bed service; and (2) whether the timing of rehabilitation, either early or late, affected outcome. Mobility outcome was assessed by calculating the improvement in mobility from admission to discharge using the FIM motor score. Mobility was significantly improved in those patients receiving rehabilitation on the CIRU, but the timing of rehabilitation had no significant effect. This initial study provided important information about the effect of the rehabilitation environment of care on early postoperative outcome, but also raised important questions that remain to be answered. Most specifically, it captured a very brief window in the overall continuum of amputee care, and there is uncertainty about whether the benefits will be sustained across time. In addition, the outcome measure, the motor FIM, has been shown to have some limitations because of ceiling effects in the amputee population,^6,7 suggesting that some benefits of rehabilitation reflecting important and meaningful improvements in ambulation may not be adequately captured with this instrument.

In addition to comparing outcomes of dysvascular lower extremity amputees between different rehabilitation environments of care, it is important to determine whether all amputees benefit equally from the more intensive and expensive rehabilitation treatment in a CIRU, or whether there are specific subpopulations or patient characteristics that are associated with improved outcome. This information will allow patients, providers, and payers to establish more effective clinical care guidelines and a more consistent approach to care delivery for this population.

The present study was conducted to further our understanding of the relationship between rehabilitation environments of care and mobility outcome 1 year after lower extremity dysvascular amputation. The specific objectives were 3-fold: (1) to compare the total volume of rehabilitation therapy for patients ever receiving rehabilitation in a CIRU versus never during the 12 months after amputation; (2) to determine whether the provision of rehabilitation in a CIRU at any time in the first year after amputation results in greater mobility success compared with other types of rehabilitation environments of care after controlling for total volume of care; and (3) to determine for those patients treated in a CIRU, which specific patient characteristics were associated with improved mobility outcome.

METHODS

Study Design

This study used a subsample from a larger multisite trial that included 2 Veterans Health Administration (VHA) medical centers and 2 university hospitals.⁸ Our sample only included those subjects enrolled at the 2 VHA medical centers (Veterans Administration Puget Sound and Denver) because the rehabilitation environment of care data were more reliable and accessible. We performed a prospective cohort study of individuals undergoing major LEA because of complications of peripheral vascular disease (PVD) or diabetes. The decision to refer a patient to a particular rehabilitation environment and to perform a transmetatarsal (TM), transtibial (TT), or transfemoral (TF) amputation was made per usual care at the discretion of the participant’s surgeon and rehabilitation team. Participants were assessed via in-person or telephone interview presurgically and at 6 weeks, 4 months, and 12 months postsurgically. The time between the decision to amputate and surgery was often brief, and we were unable to enroll all participants before surgery. Participants enrolled before surgery (n=28; 39%) completed the presurgical status assessment at the time of enrollment, while those enrolled 6 weeks postsurgically completed a retrospective assessment of their presurgical status. Retrospective questions were identical to questions on the presurgical battery. The study coordinator prefaced the questions by saying, “In the following questions, we are asking you to tell us about your life and health in the week before your amputation.” All assessments were performed by a registered nurse-level trained study coordinator. This study was conducted in accordance with the procedures approved by local human subjects review boards.

Participants

A total of 199 individuals from the 2 VHA facilities were screened for participation between September 2005 and December 2008. Participants were considered eligible if (1) they were 18 years or older, (2) they were awaiting (or underwent in the last 6wk) a first major amputation (defined as TM level or higher), and (3) the primary cause of amputation was complications of diabetes or PVD. Participants were excluded if (1) they had inadequate cognitive or language function to consent or participate, defined by ≥6 errors on the Short Portable Mental Status Questionnaire; or (2) they were nonambulatory before the amputation for reasons unrelated to PVD or diabetes. Of the 199 individuals screened, 113 (57%) met study criteria; 72 of these participants (64%) were enrolled.

Participant Characteristics

Participant characteristics such as demographics (eg, age, marital status, race), information about the index amputation, and health factors (eg, body mass index, smoking, alcohol use) were assessed presurgically when possible (n=28), or at 6 weeks postsurgery (n=44). The index level of amputation was categorized as TM, TT, or TF as reported in the medical record and confirmed during the interview. The Charlson Comorbidity Index⁹ was used to determine the presence of presurgical comorbid conditions. Additional comorbid conditions hypothesized to be relevant in these populations were also assessed (table 1). With regard to health factors, smoking status was assessed by 3 standard questions from the Veterans Administration Large Health Survey. Participants were considered smokers if they endorsed smoking “every day” or “some days” before amputation, and nonsmokers if they endorsed the remaining category “not smoke at all.” A 3-item version of the Alcohol Use Disorders Identification Test was used to assess alcohol consumption patterns in the past year.¹⁰ Possible scores range from 0 to 12, with higher scores indicating greater alcohol use.

Table 1:

Baseline Sociodemographic and General Health Data by Subjects Who “Ever” Versus “Never” Attended a CIRU During the 12 Months After Amputation

Variable	Ever CIRU (n=35)	Never CIRU (n=37)	P*
Amputation level			<.001
TM	5 (14)	22 (59)
TT	24 (69)	14 (38)
TF	6 (17)	1 (3)
Age (y)	60.8±6.9	65.7±8.4	.01
Sex (male)	34 (97)	35 (95)	.59
BMI (kg/m2)	31.6±7.7	29.6±6.3	.23
Married/partner	18 (51)	25 (69)	.12
Race			.13
White	31 (89)	28 (76)
Black	1 (3)	7 (19)
Other	3 (8.6)	2 (5)
Employed	4 (11)	3 (8)	.66
Education level			.61
Some high school	1 (3)	3 (8)
High school grad	27 (77)	26 (72)
College grad	7 (20)	7 (19)
Living status			.89
Home alone	11 (31)	9 (25)
Home with spouse/other	21 (60)	23 (64)
SNF/nursing home	2 (6)	2 (6)
Other	1 (3)	2 (6)
Socioeconomic status			.65
≤$25,000	15 (43)	14 (39)
$25,001–$50,000	12 (34)	16 (44)
>$50,000	8 (23)	6 (17)
Self-perceived health (good/very good)	12 (34)	15 (42)	.52
Charlson			.05
Low	3 (9)	11 (31)
Moderate	8 (23)	11 (31)
High	14 (40)	8 (22)
Very high	10 (29)	6 (17)
Diabetes†	30 (86)	32 (86)	.93
Stroke†	7 (20)	7 (19)	.91
Heart attack†	12 (34)	14 (38)	.75
Dialysis†	2 (6)	3 (8)	.69
COPD†	6 (17)	3 (8)	.25
Arterial reconstruction	13 (37)	17 (47)	.39
Joint replacement	2 (6)	6 (17)	.15
Hypertension	24 (69)	25 (68)	.93
Smoker	17 (49)	9 (25)	.04
Alcohol consumption score (points)	1.6±3.0	1.8±2.9	.83
Modified social support score (6wk)	68.1±24.0	69.0±29.7	.89
Major depressive disorder (4mo)	11 (31)	3 (9)	.02
Premorbid LCI-5 score	47.2±10.1	49.3±9.4	.36

NOTE. Values are n (%), mean ± SD, or as otherwise indicated. Abbreviation: BMI, body mass index.

^*P value based on unpaired t tests for age and BMI, and on χ² test for the categorical variables.

^†Comorbidities obtained from the Charlson Comorbidity Index.

Mobility

Mobility was assessed using the well-validated Locomotor Capability Index (LCI-5): 14 items are graded on a 5-level ordinal scale ranging from “unable to perform the activity” (0 points) to “able to perform independently without assistance” (4 points).¹¹ Mobility was assessed at each follow-up, although for the purposes of this study we use only premorbid mobility and 12-month mobility. Premorbid mobility was defined as the level of mobility just before the development of disability (eg, ulcer, edema, associated pain) affecting the extremity undergoing amputation.⁸ To assess premorbid mobility, the LCI-5 instruction was modified slightly such that participants were asked to complete the items based on recall of function “immediately prior to developing any limitations in your leg awaiting amputation.” Premorbid mobility was assessed retrospectively 6 weeks after surgery for all enrolled participants. Additional details and justification for the retrospective measures used in this article are reported in a previous publication.⁸

Social Support

We assessed the degree of social support at 6 weeks postsurgery for all participants using the Multidimensional Scale of Perceived Social Support,¹² a 12-item self-report measure of perceived social support from 3 specific sources: family, friends, and significant other. For each item, participants are asked to rate their degree of agreement on a 7-point Likert scale. Possible total scores range from 12 to 84, with higher scores indicating greater perceived social support. Items primarily reflect perceived availability of emotional, informational, companionship, and affection support. Internal consistencies of the subscales and total scale in the current sample are all excellent (Cronbach α=.85–.91), and the scales have demonstrated strong test-retest stability over 2- to 3-month intervals (r=.72–.85).¹² In our previous work,¹³ social support (assessed using the Multidimensional Scale of Perceived Social Support) was an important predictor of mobility 6 months after amputation, controlling for demographic and amputation-related factors.

Major Depressive Episode

For assessing presence of a major depressive episode, we administered the Patient Health Questionnaire at 4 months postsurgery. The 4-month time point was selected to ensure that the diagnosis temporally preceded the 12-month outcome. The Patient Health Questionnaire is a well-validated self-report screening instrument designed to provide diagnoses of high-prevalence psychiatric disorders based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria.¹⁴ The module instructs participants to rate the degree to which they experienced each of 9 symptoms of depression over the past 2 weeks, ranging from 0 (not at all) to 3 (nearly every day). Major depressive episode is coded as present if individuals endorsed 5 or more symptoms on “more days than not” and 1 of the items endorsed was either depressed mood or anhedonia. Depression is included as a control variable because our previous work⁸ also suggests a relationship between mobility and treatment for anxiety or depression.

Rehabilitation Environment Assessment

Trained registered nurse-level study coordinators from each of the 2 participating VHA study sites searched the computerized patient record system to classify each participant’s rehabilitation experience over the 12-month period after surgery and to quantify time spent in several key therapies. The following items were recorded: referral to a CIRU, a transitional care unit, a private skilled nursing home or facility, and outpatient services; and minutes spent in physical therapy, occupational therapy, or rehabilitation psychology treatment. Those who participated in a CIRU may have also participated in another rehabilitation setting during the 12 months after amputation; however, our focus was on comparing those ever exposed to a CIRU, realizing few patients participate in rehabilitation care that is limited to a single environment. A CIRU was defined as a Commission on Accreditation of Rehabilitation Facilities-accredited rehabilitation bed unit providing goal-directed, interdisciplinary evaluation and treatment, including physical therapy, occupational therapy, rehabilitation psychology, social work, speech pathology, and recreation therapy, under the direction of a physical medicine and rehabilitation physician. The total number of rehabilitation visits and minutes (both inpatient and outpatient) documented for physical therapy, occupational therapy, and psychological services over the course of each participant’s 12-month experience was recorded. Visits that may have occurred outside a VHA facility would not have been captured; however, that is why we only selected VHA patients, assuming that most of the care would be performed in a VHA facility.

Outcomes

Mobility success was defined dichotomously. Success occurred when the level of mobility at 12 months was the same as or greater than the premorbid mobility level as measured by the LCI-5. The clinical utility and justification for this measurement were reported in a previous publication.⁸

Data Analysis

Descriptive statistics of presurgical and postsurgical variables are presented in table 1. For categorical and continuous variables, differences by whether subjects “ever” attended a CIRU versus “never” in the 12 months after amputation were made using Pearson chi-square tests and independent sample t tests, respectively. Rates of mobility success were computed by “ever” versus “never” attending a CIRU using chi-square tests. We performed bivariate analyses of all baseline variables with rehabilitation environment and with mobility success. Variables with an α level of .05 were considered statistically significant. Those variables that were significantly associated with rehabilitation environment or the outcome were considered candidate variables for the multivariate analyses. For the first objective, to compare total visits and total hours by environment of care, we used unpaired t tests. We reported total therapy visits and hours by ever versus never attending a CIRU.

For the second objective, to determine whether the provision of rehabilitation in a CIRU resulted in greater mobility success compared with not receiving care in a CIRU, we examined multivariate associations of explanatory factors with mobility success using negative binomial regression to report risk differences (RDs) and 95% confidence intervals (CIs). Negative binomial regression is a method suitable for count and rate data when relative risks and RDs are to be estimated.¹⁵ We evaluated the correlation between individual explanatory variables using Spearman ρ. We considered a correlation coefficient of ≥.61 as strong agreement and therefore highly correlated. For highly correlated variables, we selected the explanatory variable that had the greatest clinical significance and contributed to the most stable multivariate models. Only variables that were associated with the outcome at a .05 significance level or lower were kept in the model. We initially created a model with sociodemographic variables (see table 1). Next, the following general health variables considered clinically important were tested for addition to the model: self perceived health; body mass index category (underweight or normal, 24.9kg/m²; overweight, 25–29.4kg/m²; obese, >29.4kg/m²); smoking (coded yes or no); alcohol consumption score measured by Alcohol Use Disorders Identification Test; total burden of comorbid conditions from the Charlson Comorbidity Index (divided into low [0–3.0 points], moderate [3.1–5.9 points], high [5.1–7.9 points], and very high [≥8.0 points]); and the presence of any of the following conditions: history of cerebral vascular accident, myocardial infarction, hypertension, chronic obstructive pulmonary disease (COPD), asthma, posttraumatic stress disorder, total joint replacement, diabetes, hepatitis, dialysis, or metastatic cancer (coded yes for present or no for absent). In the third step, we added the 6-week social support score and major depressive episode (at 4mo). In the last step, we added the level of amputation, the VHA facility, and a measure of rehabilitation intensity (total visits or hours) to the model, and these were left in the model regardless of their strength of association. For our third objective, to determine which specific patient characteristics were associated with improved mobility outcome in a CIRU, we used the same steps only in patients who had ever attended a CIRU. All variables listed in table 1 were assessed. Stata 9.1^a was used to conduct all analyses.

RESULTS

Baseline Characteristics

Among the 72 participants enrolled, 3 participants (4%) formally withdrew, 2 were lost to follow-up (3%), and 6 participants (8%) died over the 12-month follow-up period. Sixty-one participants (85%) completed their 12-month interview. Most of the 72 participants were TT (n=38; 53%) amputees, followed by TM (n=27; 38%) amputees (see table 1). A greater proportion of TF (86%) and TT (63%) amputees had received rehabilitation in a CIRU over the 12-month period compared with TM (19%) amputees (P<.001). Patients who ever attended a CIRU were significantly younger (mean age ± SD, 60.8±6.9y) than those who never attended (65.7±8.4y) over the 12-month period (P=.01). Patients who had ever attended a CIRU were more likely to be smokers (49% vs 25%; P=.04) and to have had a major depressive disorder diagnosed at 4 months (31% vs 9%; P=.02) compared with those who never attended. There were no other statistically significant baseline differences between groups. It is noteworthy that premorbid mobility LCI-5 scores were not significantly different between rehabilitation environment groups (P=.36).

Visits and Hours of Specialty Rehabilitation by Environment of Care

The mean ± SD number of physical therapy visits for patients ever attending a CIRU was significantly greater (31.9±24.4) than for those never attending (13.9±21.9) over a 12-month period (P=.002) (table 2). A similar trend was observed for rehabilitation psychology visits (3.1 vs 1.4; P=.06) and total number of any therapy visits (48.6 vs 22.6; P=.001), despite fewer occupational therapy visits (7.7 vs 13.1; P=.03). The mean ± SD total time per any rehabilitation visit was .83±.28 hours for those ever attending and .60±.20 hours for those never attending (P<.001). Total number of visits and total hours of rehabilitation therapies were highly correlated (Spearman ρ=.96; P<.001); therefore, total visits were used for multivariate models because it led to more stable effect estimates.

Table 2:

Number of Rehabilitation Therapy Visits by Environment of Care

Variable	Ever CIRU (n=34)	Never CIRU* (n=37)	P†
Total PT visits	31.9±24.4	13.9±21.9	.002
Total OT visits	7.7±7.3	13.1±12.0	.03
Total RP visits	3.1±4.2	1.4±3.5	.06
Total rehabilitation visits	48.6±29.3	22.6±34.0	.001
Hours per rehabilitation visit	.83±.28	.60±.20	<.001

NOTE. Values are mean ± SD or as otherwise indicated.

Abbreviations: OT, occupational therapy; PT, physical therapy; RP, rehabilitation psychology.

^*Six participants in the never CIRU group did not receive any rehabilitation therapy visits at all (5 of these were TM amputees).

^†P value based on analysis of variance.

Mobility Success by Rehabilitation Environment of Care

At 12 months, 33% had achieved mobility success. Among those who achieved success (n=20), the mean change in LCI-5 scores from premorbid to 12 months after amputation was 5.5±7.8 points. For those who did not achieve success (n=41), the mean change in LCI-5 scores from premorbid to 12 months after amputation was −14.2±13.2 points. This difference between groups was statistically significant (P<.001). In a bivariate analysis, participants who were treated in a CIRU over the 12-month period achieved a higher rate of mobility success than those who never attended; however, the difference in rates was not statistically significant (36% vs 30%, respectively; P=.65). Despite the unequal distribution of amputation levels by rehabilitation environment of care, the rates of success were similar: 35%, 31%, and 33% of TM, TT, and TF amputees, respectively, achieved mobility success (P=.96). However, we included amputation level in the multivariate model because of the unequal distribution and importance of this exposure variable. In the multivariate model, patients who ever were treated in a CIRU were 17% more likely to achieve mobility success than those who were not, controlling for amputation level, major depressive episode, alcohol score, social support, total number of rehabilitation visits, and hospital site (RD=.17; 95% CI, .09–.25; P<.001) (table 3). Those who had a major depressive episode diagnosed at 4 months were 23% less likely to achieve mobility success (RD=−.23; 95% CI, −.12 to −.34; P<.001). Those with greater alcohol use were less likely to achieve mobility success (P<.001). Those with greater social support were more likely to achieve success (P=.003). The total number of rehabilitation visits was not significantly associated with mobility success, controlling for the rehabilitation environment (P=.80).

Table 3:

Multivariate Results for 12-Month Mobility Success

Risk Factor	RD* (95% CI)	P
CIRU	.17 (.09 to .25)	<.001
Amputation level†
TT	.05 (.02 to .13)	.006
TF	−.09 (−.41 to .24)	.60
Major depressive disorder at 4mo (yes)	−.23 (−.12 to −.34)	<.001
Alcohol score (points)	−.05 (−.03 to −.07)	<.001
Social support score (points)	.003 (.001 to .006)	.003
Total visits	−.0004 (−.004 to .003)	.78
Hospital site	.18 (.08 to .25)	.001

^*RDs generated from a negative binomial regression model represent an increase (or decrease if negative) in the success rate relative to reference category (site and total rehabilitation visits were also included in the model but were not statistically significant).

^†TM is the reference category.

Patient Characteristics Associated With Mobility Success in a CIRU

Among the 31 participants who were treated in a CIRU over the 12-month period, we sought to determine which participant characteristics were associated with 12-month mobility success, in an effort to identify ideal candidates for inpatient rehabilitation. In a multivariate model controlling for amputation level, hospital site, and total rehabilitation visits, the following characteristics were associated with achieving mobility success: younger age (RD=.03; 95% CI, .02–.04; P<.001), higher social support score (RD=.01; 95% CI, .002–.01; P=.009), no COPD (RD=.42; 95% CI, .19–.65; P<.001), and “underweight to normal” body mass index categories compared with “overweight” (RD=.57; 95% CI, .24–.90; P=.001) or “obese” (RD=.44; 95% CI, .10–.78; P=.01).

DISCUSSION

The primary goals of this investigation were (1) to compare the total volume of rehabilitation therapy for patients between rehabilitation environments of care; (2) to determine whether the provision of rehabilitation in a CIRU at any time in the first year after amputation results in greater mobility success compared with other types of rehabilitation environments of care; and (3) to determine whether specific factors among patients treated in a CIRU were associated with improved mobility outcome.

There is a spectrum of environments of care where rehabilitation treatments may be provided for individuals who have undergone LEA for diabetes/PVD including a CIRU, outpatient therapy in the home, center-based outpatient therapy, and therapy in an SNF. Few published data exist to guide clinical decision-making regarding the optimum environment of care for individuals with dysvascular LEA. This is confirmed by the variability in clinical care practices. Historically, most dysvascular amputee patients were treated in outpatient or SNF environments, with few being treated in CIRUs. From 1986 through 1987, Maryland Health Services data showed that discharge to a rehabilitation unit accounted for less than 10% of discharges, while discharges to SNFs and home each accounted for approximately 40%.⁵ In a follow-up study³ including a sample of national Medicare data, only 14% of individuals undergoing amputations from 1996 through 1997 were discharged to an inpatient rehabilitation unit, while 49% were discharged to home, and 37% were discharged to an SNF. Similarly, in a veteran population with dysvascular amputations between 2002 and 2004, only 20% were admitted to a specialized inpatient rehabilitation unit, and only 28% received any type of rehabilitation.¹⁶ A more recent study¹⁷ (2001–2006) has shown a greater proportion of patients being treated on CIRUs, with 55% of those that underwent major dysvascular LEA in 1 of 18 hospitals in Baltimore or Milwaukee between 2001 and 2006 being admitted to a CIRU. This is similar to our study population where 49% were admitted to a CIRU. The reasons for the increased proportion of patients being admitted to CIRUs in these 2 more recent studies are not known. It may be related to changes in health care practices, or differences in the inclusion and exclusion criteria of the participants in the 2 more recent studies. Prior data included all individuals undergoing amputation, while in both our study and that of Dillingham et al,¹⁷ there were more restrictive inclusion criteria that may have resulted in a population that had a higher level of function and a greater tendency for admission to a CIRU.

The fundamental question remains: “Does inpatient rehabilitation in a CIRU improve outcome, and what patient characteristics are associated with improved outcome in this environment of care?” Our study findings indicate that individuals with dysvascular LEA who were treated in a CIRU at any time in the first 12 months postamputation were significantly more likely to achieve mobility success than those who never attended a CIRU. Those who were treated in a CIRU environment of care were 17% more likely to achieve mobility success than those who never attended, controlling for a myriad of important potential confounding factors. They were exposed to a significantly greater volume of rehabilitation (visits and hours); however, the total number of visits was not associated with mobility success. This suggests that there are factors outside of rehabilitation treatment volume or intensity that result in enhanced mobility outcomes in those treated in a CIRU. Within the literature on stroke rehabilitation, there is evidence that 1 such factor may be the interdisciplinary coordination found in an inpatient unit. Existing data have shown that even among inpatient settings, teams with better structure and organization for treatment planning and communication who review and discuss their own outcome data produce higher motor FIM scores. Similarly, interventions targeting team processes, including communication and problem solving, have also been effective in improving rehabilitation outcomes.^18,19

Although the effect size is modest at 17%, the significant improvement in mobility outcome does suggest that inpatient rehabilitation may be advantageous for this population. Although a myriad of important potential confounding demographic, health, and psychosocial factors were controlled for in this analysis, we performed additional analyses to determine whether health care teams tended to select patients with certain traits for admission to a CIRU. In fact, there were few differences between those who had, and those who had not, been treated in a CIRU. The only factors that differed were that those admitted to a CIRU were significantly younger, and there were a greater proportion of smokers as well as individuals with a diagnosed major depressive episode. In addition, a greater proportion of TF (86%) and TT (63%) amputees had received rehabilitation in a CIRU compared with TM (19%) amputees. The distribution of amputation levels admitted to a CIRU is logical in that TM amputees are less likely to require as intense an interdiscipinary rehabilitation approach compared with higher levels of amputation. We included all levels to increase study power and were justified since the outcomes between amputation levels were not significantly different. The reasons for the increased frequency of those who had a history of smoking similarly is likely linked to the significant relationship between smoking and more proximal levels of amputation.⁸ The increased proportion of patients admitted to a CIRU with a history of major depression is an unlikely trait that is specifically selected for in the environment of the care decision process. Conjecturally, it may be the result of adverse impacts of this diagnosis on function and the implicit perceived need for a more intensive rehabilitation approach to enhance the level of function. It is important to note, however, that in terms of baseline mobility, there were no significant differences between the 2 populations. The differences in outcome between those admitted to a CIRU and those who were not admitted cannot be attributed to this factor. Individuals admitted to a CIRU were exposed to a significantly greater volume of rehabilitation (visits and hours), but once again, the total number of visits was not independently associated with mobility success. This suggests that there are factors outside of rehabilitation treatment volume or intensity that result in enhanced mobility outcome in those treated in a CIRU.

These research findings extend those of recent studies in which inpatient rehabilitation was shown to improve outcomes in dysvascular lower extremity amputees. Survival at 1 year, reduced nonamputation hospital admission, reduced reamputation, and increased probability of prosthetic prescription were all associated with an inpatient rehabilitation environment of care.³ Because of limitations in the available data, this analysis of the effect of inpatient rehabilitation was not able to control for psychosocial factors, premorbid mobility level, and the effects of rehabilitation intensity between those who were treated in an inpatient rehabilitation unit and those not treated in an inpatient rehabilitation unit. The rate of prosthetic prescription was used as a surrogate measure of mobility outcome, which the authors agree, may have significant limitations. Stineman⁴ showed that mobility outcome using the motor FIM as the outcome measure was improved in dysvascular lower extremity amputees who received rehabilitation in a CIRU compared with those who received rehabilitation in another bed service in the early postoperative period. The motor FIM is an acceptable measure for function in the early postoperative period; however, it is not suitable for long-term follow-up and is not amputee specific like the LCI-5.

The research design of our study has a number of strengths compared with previously published research in this area. One of the emphases of current medical research is comparative effectiveness research—that is, to determine the effect of treatment interventions through the use of observational studies that control for patient and system attributes. The challenge is to measure these factors accurately and account for them through rigorous statistical approaches.²⁰ One of the principal advantages of the prospective study design used in our study is that the variables that define both the patient characteristics and the system attributes were anticipated in advance. This is in contrast with large dataset analyses that are limited by the availability of a constrained set of variables that are often imperfect because they were not developed to address key research questions. We were able to define our study population including many aspects of the baseline demographic, health comorbidities, psychological factors, social factors, and premorbid mobility. In terms of other important mediating factors, we were able to define rehabilitation therapy treatment frequency in the multiple domains that encompass rehabilitation treatment as well as the environment of care of the rehabilitation intervention. The primary outcome measure, mobility, was quantified by a well-validated, amputation-specific mobility measure. The measure of mobility success has demonstrated a strong association with absolute mobility, satisfaction with mobility, and satisfaction with life.⁸

Another goal of this research was to determine whether certain patient characteristics were associated with a greater (or lesser) likelihood of success. This information may better arm the rehabilitation team in deciding which rehabilitation environment of care would be optimum for a patient. We found that patients who are younger, who have higher social support, who are not overweight or obese, and do not have a diagnosis of COPD are more likely to achieve mobility success in the comprehensive inpatient rehabilitation environment. It is important to note that 2 of these factors are modifiable. Although requiring additional study, these data suggest that optimizing body weight and strengthening social support systems may be important goals of the inpatient rehabilitation program.

In summary, this study enhances our understanding of the potential benefits of inpatient rehabilitation in a CIRU for patients with dysvascular amputation. It is, to our knowledge, the first prospective approach to addressing this question, and is especially relevant because historically, relatively few patients have been admitted to this environment of care. If these results are confirmed by additional study, it may suggest that rehabilitation in a CIRU should be considered more frequently in the plan of care of this patient population. We have also defined the key patient characteristics that will result in significant improvements in mobility in this environment of care. Although the intensity of the rehabilitation is greater in the CIRU, this does not seem to be the critical factor in the enhanced mobility success.

Study Limitations

As with all studies, there are limits to the generalizability of the findings. Specifically, these results are limited to a dysvascular population who are undergoing their first major LEA and who had some level of premorbid ambulatory function as well as adequate cognitive function.

One of the primary limitations of this study is the retrospective recall of premorbid mobility. In this patient population with underlying diabetes and PVD, there is a high incidence of comorbid medical conditions that can adversely affect mobility before the amputation. Therefore, any attempt to quantify the effect of treatment interventions on mobility outcome must simultaneously evaluate the impact of these comorbidities. As noted, premorbid mobility was measured by retrospective recall of function immediately before the onset of the disability that led to amputation. Thus, our measurement may be subject to recall bias or errors of recall. It is uncertain what the potential magnitude or direction of this bias may be because it has never been studied in this population at this point in their care continuum. There may be some reassurance, however, in previously published data related to disability and limb impairment. Only a 4-point response shift bias was noted in the preoperative recall of function in a population of total knee arthroplasty patients over a 6-month period using the Western Ontario and McMaster Universities Osteoarthritis Index.²¹ Sixty to seventy percent of these patients did not have recall bias and the results are similar in another orthopedic surgical study.²² Secondly, our intent was to enroll all subjects before their amputation; however, preoperative enrollment was challenging for many reasons. Included here are the administrative realities of admission dates relative to surgical dates plus the pragmatics of the clinical status of potential subjects who were often preoccupied with presurgical preparations. Our protocol was subsequently modified so that subjects could be recruited up to 6 weeks after surgery.

CONCLUSIONS

Rehabilitation in a CIRU resulted in improved mobility success for veterans undergoing major LEA secondary to PVD or diabetes. Success was not related to differences in baseline mobility, demographic factors, psychosocial factors, or number of total rehabilitation therapy visits. The CIRU environment appears to add a dimension to care that results in enhanced mobility outcome that is not captured by rehabilitation therapy volume alone. One of the challenges of patient care in this population is to try to identify which patients will best benefit from a more intensive rehabilitation environment. Unsurprisingly, of those admitted to a CIRU, younger patients with greater social support, healthy weight, and without COPD had the greatest probability of mobility success.

List of Abbreviations

CI

confidence interval

CIRU

comprehensive inpatient rehabilitation unit

COPD

chronic obstructive pulmonary disease

LCI-5

Locomotor Capability Index (5-level ordinal scale)

LEA

lower extremity amputation

PVD

peripheral vascular disease

RD

risk difference

SNF

skilled nursing facility

TF

transfemoral

TM

transmetatarsal

TT

transtibial

VHA

Veterans Health Administration