Rehospitalization in the First Year of Traumatic Spinal Cord Injury After Discharge From Medical Rehabilitation

Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2013;94(4 Suppl 2):S87-97

ORIGINAL ARTICLE

Rehospitalization in the First Year of Traumatic Spinal Cord Injury After Discharge From Medical Rehabilitation Gerben DeJong, PhD,a,b,c Wenqiang Tian, MD, PhD,a,b,c Ching-Hui Hsieh, PhD, OT,a,c Cherry Junn, MD,a Christopher Karam, MD,a Pamela H. Ballard, MD,a,b Randall J. Smout, MS,d Susan D. Horn, PhD,d Jeanne M. Zanca, PhD, MPT,e Allen W. Heinemann, PhD,f Flora M. Hammond, MD,g,h Deborah Backus, PT, PhDi From the aMedStar National Rehabilitation Hospital, Washington, DC; bGeorgetown University School of Medicine, Washington, DC; cMedStar Health Research Institute, Washington, DC; dInstitute for Clinical Outcomes Research, Salt Lake City, UT; eMount Sinai School of Medicine, New York, NY; fRehabilitation Institute of Chicago, Chicago, IL; gCarolinas Rehabilitation, Carolinas HealthCare System, Charlotte, NC; hIndiana University, Indianapolis, IN; and iShepherd Center, Atlanta, GA.

Abstract Objective: To determine rates of rehospitalization among discharged rehabilitation patients with traumatic spinal cord injury (SCI) in the first 12 months postinjury, and to identify factors associated with rehospitalization. Design: Prospective observational cohort study. Setting: Six geographically dispersed rehabilitation centers in the U.S. Participants: Consecutively enrolled individuals with new traumatic SCI (NZ951), who were discharged from participating rehabilitation centers and participated in a 1-year follow-up survey. Interventions: Not applicable. Main Outcome Measures: Occurrence of postrehabilitation rehospitalization within 1 year of injury, length of rehospitalization stays, and causes of rehospitalizations. Results: More than one third (36.2%) of participants were rehospitalized at least once in the 12-month follow-up period; 12.5% were rehospitalized at least twice. The average number of rehospitalizations among those rehospitalized at least once was 1.37 times, with an average length of stay (LOS) of 15.5 days across all rehospitalization episodes. The 3 most common health conditions associated with rehospitalization were those related to the genitourinary system (eg, urinary tract infection), respiratory system (eg, pneumonia), and skin and subcutaneous tissue (eg, pressure ulcer). Being a woman (95% confidence interval [CI], 1.034e2.279), having Medicaid as the main payer (95% CI, 1.303e2.936), and more severe case mix were associated with increased odds of rehospitalization. Those who had more intensive physical therapy (95% CI, .960e.981) had lower odds of rehospitalization. Some center-to-center variation in rehospitalization rates remained unexplained after case mix and practice differences were considered. The 6 SCI rehabilitation centers varied nearly 2-fold in rates at which their former SCI patients were rehospitalizeddfrom 27.8% to 50%. Center-to-center variation diminished when patient case mix was considered. Conclusions: Compared with earlier studies, rehospitalization rates among individuals with SCI in the first postinjury year remain high and vary by level and completeness of injury. Rehospitalization risk was associated with younger age, being a woman, unemployment and retirement, and Medicaid coverage. Those who had more intensive physical therapy had lower odds of rehospitalization. Future studies should examine center-tocenter variations in rehospitalization rates and availability of patient education and community resources. Archives of Physical Medicine and Rehabilitation 2013;94(4 Suppl 2):S87-97 ª 2013 by the American Congress of Rehabilitation Medicine

Supported in part by the National Institute on Disability and Rehabilitation Research, U.S. Department of Education (grant nos. H133A060103, H133N060005, H133N060028, H133N060009, H133N060014, H133N060027, H133N060009). The opinions contained in this publication are those of the authors and do not necessarily reflect those of the U.S. Department of Education. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated.

0003-9993/13/$36 - see front matter ª 2013 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2012.10.037

S88

G. DeJong et al

Individuals with traumatic spinal cord injury (SCI) have high rates of unplanned hospital readmissions, often for conditions that are potentially preventable, for example, urinary tract infections (UTIs), respiratory conditions, and pressure ulcers.1-7 Once readmitted, they may also require additional rehabilitation therapy to regain strength, endurance, and function lost while rehospitalized. Rehospitalization can be disruptive, undermine rehabilitation gains, and diminish an individual’s ability to live actively and independently. Hospital readmission rates are a key indicator of hospital quality, and mitigating rehospitalizations is considered an important clinical and policy objective.8-12 The U.S. Medicare program, for example, places hospitals at financial risk for higher than usual readmission rates for select diagnostic groups, especially for readmitting patients with conditions considered preventable. This article reports on the rehospitalization experience of individuals with traumatic SCI in the first 12 months after injury following their discharge from 6 geographically dispersed rehabilitation centers in the United States. It examines (1) the numbers of individuals rehospitalized, (2) the number of rehospitalizations, (3) lengths of stay (LOSs) associated with rehospitalization, (4) health conditions that lead to rehospitalization, and (5) patient characteristics and clinical practice features associated with rehospitalization. Our goal was to identify previously uncovered patient risk factors and clinical practices that could help reduce future rehospitalizations associated with SCI. Our study was governed by a conceptual framework that examined how rehabilitation outcomes varied with (1) demographic characteristics, (2) injury-related and severity-of-illness measures, (3) rehabilitation treatments and processes, and (4) the interaction of these variables.13 Given the study’s comprehensive scope and detailed characterization of patients and treatments, the study’s database was much richer than those seen in previous studies. Hence, it offered the possibility of obtaining a more nuanced and textured explanation for the rehospitalization phenomenon.

Methods Approach This analysis was conducted under the auspices of the 5-year, multicenter SCIRehab study (2007e2012), a prospective observational cohort study that sought to obtain a comprehensive and detailed view of SCI rehabilitation practices and their outcomes at discharge from rehabilitation and at 12 months after traumatic injury. The study used a practice-based evidence approach that examined the practice of rehabilitation by using clinical service and therapy taxonomies developed, tested, and validated by clinicians from all disciplines at each center. Study clinicians also collaborated

List of abbreviations: AIS CMG CMI CSI IRF LOS OR PHQ-9 SCI UTI

American Spinal Injury Association Impairment Scale case-mix group case-mix index Comprehensive Severity Index inpatient rehabilitation facility length of stay odds ratio Personal Health Questionnaire-9 spinal cord injury urinary tract infection

in the design of follow-up interview questionnaires that allowed the study team to examine 12-month outcomes, including rehospitalizations reported here. A more complete discussion of SCIRehab study methods can be found in Whiteneck et al.13

Study facilities The 6 participating rehabilitation centers comprise a convenience sample of SCI rehabilitation centers drawn from different parts of the U.S. Five of the 6 centers participated in the National Institute on Disability and Rehabilitation Research-sponsored SCI Model Systems program at the time of the study. SCI Model System centers report data in standardized ways that facilitate comparisons across centers. Four of the 6 centers are classified by the U.S. Medicare program as inpatient rehabilitation facilities (IRFs), and 2 centers are classified as long-term care hospitals. IRFs and longterm care hospitals follow different rules with respect to payment, minimum hours of therapy, LOSs, and scope of services. This leads to greater practice variation and allows one to examine how different practice patterns may be associated with outcomes. All 6 centers are nonprofit rehabilitation facilities.

Study participants Study patients consisted of individuals who (1) were 12 years and older, (2) had a first-time traumatic SCI, and (3) participated in the study’s 12-month postinjury follow-up survey in which they were queried about their rehospitalization experiences as well as other features of their postrehabilitation life. All eligible patients were enrolled consecutively after their admission to rehabilitation and obtaining their informed consent. Individuals who transferred to an acute care hospital but returned to the rehabilitation center remained in the study, but their acute hospitalizations days were not counted toward their rehabilitation LOS.13,14 We excluded (1) individuals who spent more than 2 weeks of their initial postacute rehabilitation in a nonstudy facility, (2) individuals who spent more than 1 week in a non-SCI unit of a study facility, and (3) individuals whose rehabilitation LOS was >300 days (nZ4). Each participating rehabilitation center obtained its own institutional review board approval for the enrollment and follow-up of study patients. We compared respondents and nonrespondents to the study’s follow-up survey to identify potential selection effects and to determine the generalizability of the findings to the larger study group from which the follow-up group originated.

Instruments and measures We used the International Standards for Neurological Classification of SCI and its American Spinal Injury Association Impairment Scale (AIS) to characterize neurologic level (eg, C1-4) and completeness of motor injury (ie, grades AeD).15 We categorized study participants into 1 of 4 impairment subgroups: (1) tetraplegia C1-4 grades AeC; (2) tetraplegia C5-8 grades AeC; (3) paraplegia grades AeC; and (4) all grade Ds, regardless of injury level.16 This categorization follows the convention used by other articles in this special issue of the Archives. Comprehensive Severity Index The Comprehensive Severity Index (CSI) provides a comprehensive, condition-specific measure of patient severity or acuity; it has www.archives-pmr.org

Rehospitalization after SCI rehabilitation been validated extensively in inpatient, ambulatory, rehabilitation, and long-term care settings since 1982.17-25 The CSI combines physiological, functional, and psychosocial complexity into a single overall continuous score (no upper limit) based on the extent and interaction of all the patient’s various health conditions. We calculated 3 continuous CSI scores for each study patient to represent the severity of his or her medical acuity at admission, during the rehabilitation stay, and on discharge. Higher CSI scores indicate greater severity of illness. An article in this series by Horn et al26 describes the CSI matrix in greater detail. FIM The FIM was used as the measure of functional status at admission, discharge, and at 1-year injury anniversary. This article uses both motor FIM (13 items) and cognitive FIM (5 items) subscores to characterize the study group. We used Rasch analysis to convert raw scores to equal-interval motor and cognitive FIM measures with scores ranging from 0 to 100 for ease of interpretation.27-29 Personal Health Questionnaire-9 The Personal Health Questionnaire-9 (PHQ-9) is a self-reported, 9-item measure of depressive symptomatology that was administered in the 12-month follow-up interview to evaluate severity of depressive symptoms. Each item evaluates frequency of a symptom using a scale that ranges from experiencing a symptom not at all (0) to nearly every day (3). PHQ-9 scores range from 0 to 27, from no depressive symptomatology to extensive symptomatology. The PHQ-9 has been found to be both valid and reliable.30,31 Case-mix group Case-mix groups (CMGs) are used by the Centers for Medicare and Medicaid Services to adjust IRF payments by taking into account patient functional status, age, and comorbidities. The Medicare payment system first assigns patients into 1 of 5 SCIrelated CMGs based on patient functional status at admission (ie, FIM score) and age, and then assigns the patient to 1 of 4 comorbidity tiers. A patient’s case weight is based on both the patient’s CMG and comorbidity tier.32 Case weights are determined for all IRF patients for whom the average case-mix weight across all facilities nationally is 1.0. A given facility’s case-mix index (CMI) is the average case weight for all patients in that facility, and a group CMI, such as SCI patients, is the average case weight for that group. We categorized SCI patients into their respective CMGs and tier groups in order to determine each patient’s case weight, as well as a facility’s CMI for its SCI patients in the study. Case weights for SCI patients in IRFs ranged from .702 (least severe) to 4.367 (most severe).

Data collection Chart abstraction Each study center identified 1 or more chart abstractors who were trained using charts from their own center. An experienced chart abstracter worked with the center abstracter in reviewing sample charts until they achieved 95% agreement on all data elements. Abstractors collected data on patient characteristics, medical history, comorbidities, employment status prior to injury, payers, use of durable medical equipment, use of orthotics and prosthetics, medications, and a variety of other variables related to patient status and outcome. The chart abstraction also provided www.archives-pmr.org

S89 data needed to determine patients’ severity of illness or acuity, as measured by the CSI. Once abstracted, we reviewed, edited, and entered the information into the study database. We conducted periodic reliability tests during the chart abstraction period. Point-of-care documentation To supplement data available in the medical record, clinicians recorded type and duration of therapy in which patients participated, using hand-held electronic devices at the point of care. Clinicians documented the time spent in each therapeutic activity, for example minutes spent in gait training or exercise. Participating clinicians included occupational therapists, physical therapists, recreational therapists, speech therapists, psychologists, social workers/case managers, and nurses. Nurses documented patient education and care management only. These data were uploaded periodically into the project database; data were checked regularly to identify potential irregularities that might undermine reliability. The study’s discipline-specific treatment taxonomies, training protocols, and reliability methods are described elsewhere.33-39 Therapists also rated patient participation using the Pittsburgh Rehabilitation Participation Scale for each therapy session.40 The Pittsburgh Participation Scale uses a 6-point Likert scale (eg, 1Zno participation, 3Zfair participation, 6Zexcellent participation). Follow-up data The SCI Model Systems program collects standardized patient data at discharge from rehabilitation (form I) and at specific injury anniversaries (form II). We collected form I data at discharge and form II follow-up data on the first anniversary of each patient’s injury. The centers collected form II data using either telephone or in-person interviews. We added study-specific questions to supplement form II. We also interviewed patients at 6 and 12 months. Follow-up interview questions were thoroughly vetted within the study team, among clinicians, and with patients from participating centers. Follow-up interview questions were pilot-tested prior to full implementation. We used all 3 sourcesd6-month interviews, 12-month interviews, and form II datadto evaluate the incidence, duration, and reason for rehospitalization in the interval from rehabilitation discharge to the patient’s 1-year postinjury date. We also used these data to evaluate patients’ vocational status, lifestyle, environmental supports, first postrehabilitation discharge destination, access to a physician knowledgeable about their SCI, and other issues up to the first anniversary of injury.

Data analysis We conducted a descriptive analysis of the rehospitalization data by examining the following: (1) whether an individual had been rehospitalized in the 12 months since injury, (2) the number of times an individual had been rehospitalized, (3) the number of days an individual had been rehospitalized, and (4) the health condition that prompted the rehospitalization. We limited the multivariate analyses only to whether the individual had, or had not, been rehospitalized. Because some patients were interviewed after their injury anniversary date, we considered only rehospitalizations that occurred up to the first anniversary of their injury. We divided the study group into the 4 AIS impairment groups previously outlined and then considered the patient’s demographic characteristics, health and functional status, health plan

S90 participation, rehabilitation experience (eg, therapy intensities, levels of patient engagement), life style, environmental supports, and how each were associated with whether the patient had been rehospitalized. We selected variables for the multivariate analysis based on the study’s conceptual framework, previous literature, recommendations of study clinicians, and the results of our bivariate analysis. We also considered how rehospitalization rates varied across the 6 participating study centers, taking into account each facility’s mix of SCI patients. We used both forced-entry and stepwise logistic regression analyses with rehospitalized/not rehospitalized as the dependent variable taking into account the measurement properties of independent variables. We used the c statistic, the area under the receiver operator characteristic curve, to evaluate how well the model discriminated between patients who were or were not rehospitalized. We report odds ratios (ORs) to identify patient characteristics, events, or practice features associated with increased or decreased odds of being rehospitalized. We also used logistic regression analyses to risk adjust center rehospitalization rates. We did so by determining the probability that a patient might be rehospitalized at least once in the interval from rehabilitation discharge to 12-month follow-up based on how the patient appeared on admission to rehabilitation. Hence, we considered only patient characteristics at the time of admission to rehabilitationdboth demographic characteristics and those related to health and functional status of the patient. We did not include rehabilitation practice variables or rehabilitation center, because we wanted to examine the predicted probability of rehospitalization for each patient regardless of what facilities might do to mitigate rehospitalizations, and then compare actual rehospitalization rates with predicted rates of rehospitalization. We validated our study findings using a separate validation sample. We divided the original cohort of 1376 enrolled patients into an analysis dataset with 75% of the cases (nZ1032) and a validation dataset with the remaining 25% (nZ344). We randomly assigned patients to 1 of these datasets, stratifying by level and completeness of injury, treatment center, and availability of follow-up interview data to ensure proportional representation within each stratum. There were no significant differences between data analysis and validation datasets on any dependent or independent variable used in the regression models. We then conducted logistic regression analyses using the analysis dataset minus those who did not participate in the 12-month follow-up interview. Once a reduced model was ascertained with only significant predictors, we validated the findings using the 25% sample (minus nonrespondents). We compared the HosmerLemeshow statistic for both samples to determine whether they had P values that exceeded .10, the threshold for determining whether the original logistic regression model validated well.

Results

G. DeJong et al The vast majority were working (66.8%) or in school (15.1%) at the time of their injury. This demographic profile is reflected in patient health-plan participation: 64.9% participated in a private health plan that corresponded to their employment status and 18.1% participated in Medicaid, the federal-state health plan for people with lower incomes and fewer assets, more typical of younger, school-age patients without a significant income or work history. As expected, those with higher-level injuries presented a more severe health and functional profile. Across the 4 AIS groups, respondents reported few differences in terms of home accessibility, access to transportation, exercise frequency, and participation in recreational and social activities.

Comparison with nonrespondents Fewer than 8% of eligible participants (nZ81) did not participate in the 12-month follow-up survey. We compared the 81 nonrespondents with the study’s 951 respondents. Respondents and nonrespondents did not differ in terms of sex, race, marital status, employment status at injury, English as primary language, or CSI scores. However, nonrespondents were older (41.6 vs 37.3y, t1030Z2.2, P<.05), more likely to have Medicare as a primary payer (13.6% vs 6.8%, c2Z10.2, P<.05), more overweight (body mass index, 26.7 vs 25.3, t1015Z2, P<.05), more likely to have less than a high school education (c2Z23.8, P<.001), and have lower motor FIM scores (15.1 vs 18.0, t1030Z2, P<.05) and cognitive FIM scores (32.2 vs 36.7, t1030Z3.2, P<.01) at admission.

Rehospitalization rates More than one third (36.2%) of the study group were rehospitalized at least once in the 12-month follow-up perioddfrom a high of 44.9% among those in the C1-4 grades AeC group to 23.8% among those in the American Spinal Injury Association grade D group (table 2); 12.5% were rehospitalized at least twice. The average number of rehospitalizations among those rehospitalized at least once was 1.37 with a combined average LOS of 15.5 days across all rehospitalization episodes.

Conditions leading to rehospitalization The 3 most common health conditions accounted for the majority of all reported reasons for rehospitalizations: those related to the genitourinary system (eg, UTI), respiratory system (eg, pneumonia), and skin and subcutaneous tissue (eg, pressure ulcers). Pressure ulcers, the third leading condition for rehospitalization accounted for a disproportionate number of rehospitalization days. Unfortunately, in 40.2% of rehospitalizations, interviewers did not capture a specific or primary reason for rehospitalization, and hence we do not report actual rates associated with reasons for rehospitalization (see study limitations section).

Study group characteristics The final study group consisted of 951 individuals with traumatic SCI, that is, the analysis dataset (nZ1032) minus the nonrespondents (nZ81) to the 12-month follow-up survey. Study participants were similar to those typically seen in a newly injured, American SCI cohort in terms of age, sex, and race (table 1). Compared with the other AIS impairment groups, those in the AIS grade D group were noticeably older (49y, P<.001).

Study group characteristics, environmental supports, lifestyle, and rehospitalization Those rehospitalized were disproportionately younger, more likely to be a woman, unemployed or retired, and more likely covered by Medicaid than those not rehospitalized (see table 1). We found no differences across racial groups and levels of environmental support (eg, marital status, home accessibility, and access to www.archives-pmr.org

Rehospitalization after SCI rehabilitation Table 1

S91

Study group characteristics by AIS neurologic category and rehospitalization

Characteristic Demographic characteristics Age (mean
SD) Female (%) Race (%) White Black Other Married (%) Education (%) High school Vocational status preinjury (%) Employed full- or part-time In school full- or part-time Unemployed/retired/other Health plan while in rehabilitation (%) Private insurance/payer Medicare Medicaid Worker’s compensation English is primary language (%) Health and functional status Admission CSI (mean
SD) Maximum CSI (mean
SD)* Discharge CSI (mean
SD) CMI (mean
SD) Admission motor FIM (mean
SD)y Discharge motor FIM (mean
SD)z 12-mo motor FIM (mean
SD) 12-mo PHQ-9 depression symptomatology score (mean
SD)x BMI group (%)jj Underweight (BMI<18.5) Healthy (18.5e24.9) Overweight (25e29.9) Obese (30) Lifestyle and environmental supports Home is accessible Has access to transportation Has access to physician knowledgeable about SCI Exercises regularly (2 times/wk) Participates in creative, recreational, or social activities

Paraplegia C1-4 Grade C5-8 Grade Grade AeC All Grade Total AeC (nZ263) AeC (nZ190) (nZ351) Ds (nZ147) (nZ951)

Yes (nZ344)

No (nZ607)

40.6
16.9 17.5

33.9
15.5 18.9

33.2
13.2 49.0
17.9 20.5 16.3

37.9
16.5 39.6
16.9 36.0
16.1# 18.7 22.1 16.8{

73.0 20.2 6.8 43.3

78.4 16.3 5.3 28.9

69.5 23.4 7.1 35.9

65.3 25.2 9.5 42.9

71.6 21.3 7.1 37.6

68.9 24.4 6.7 38.1

73.1 19.6 7.3 37.4

19.0 55.5 25.5

23.7 48.4 27.9

23.4 53.6 23.1

21.1 48.3 30.6

21.9 52.3 25.9

24.7 51.7 23.5

20.3 52.6 27.2

67.3 14.1 18.7

65.8 20.5 13.7

70.4 14.5 15.1

58.5 11.6 30.0

66.8 15.1 18.1

68.0 9.0 23.0

66.1 18.6 15.3

66.2 7.2 16.3 10.3 93.5

67.9 3.7 19.5 8.9 96.3

63.2 3.7 21.9 11.1 94.9

62.6 17.7 10.2 9.5 95.9

64.9 6.8 18.1 10.2 95.0

57.0 9.6 23.0 10.5 97.4

69.4 5.3 15.3 10.0 96.9

26.6
21.8 54.6
38.2 10.2
10.6 3.2
0.8 5.0
7.8 24.6
10.0 27.4
21.0 5.0
5.1

19.6
17.9 42.4
29.9 6.9
7.1 2.8
0.8 13.0
9.3 32.5
8.6 40.5
17.8 4.4
5.0

17.9
15.0 33.7
24.8 7.2
7.5 1.7
0.6 27.6
5.8 44.1
6.4 54.1
13.1 4.1
4.5

13.5
12.4 22.2
18.0 4.3
5.9 1.8
0.7 25.0
10.8 46.4
10.0 69.7
21.9 4.8
5.7

19.9
17.9 39.4
31.3 7.5
8.4 2.3
1.0 18.0
12.6 36.8
12.3 46.6
23.1 4.5
5.0

22.5
18.9 45.1
32.9 9.1
9.3 2.5
1.0 15.8
12.9 33.1
12.3 37.9
20.5 5.5
5.5

18.5
17.2# 36.2
29.9** 6.7
7.7** 2.2
0.9** 19.3
12.3** 38.9
11.8 51.4
23.1 4.0
4.6**

9.1 45.6 26.2 17.8

10.5 48.9 26.8 12.6

8.0 47.0 24.2 19.2

5.4 36.7 34.0 21.1

8.4 45.4 26.8 17.9

6.7 42.7 27.6 21.8

9.4 47.0 26.4 15.7

93.7 94.2 78.6

97.2 95.6 79.1

97.1 97.2 82.7

98.5 96.4 83.5

96.4 95.9 80.9

94.2 93.9 77.5

97.7 97.0 82.9

76.9 89.0

79.3 90.0

81.9 89.5

78.7 94.6

79.5 90.2

74.1 90.1

82.5# 89.3

Abbreviation: BMI, body mass index. * Highest CSI score recorded during the course of an inpatient rehabilitation stay. y Rasch-adjusted admission motor FIM score. z Rasch-adjusted discharge motor FIM score. x Rasch-adjusted discharge motor FIM score. jj Of the patients with SCI, 1.5% in the study group had missing BMI data. { P<.05. # P<.01. ** P<.001.

www.archives-pmr.org

Rehospitalized?

S92 Table 2

G. DeJong et al Rehospitalization during follow-up period

Rehospitalization Experience

C1-4 Grades AeC (nZ263)

Rehospitalized (%) 44.9 Number of rehospitalizations (%) 1 27.8 2 10.6 3 4.6 4 1.9 Average no. of rehospitalizations 1.45
1.02 (mean
SD)* Average rehospitalization LOS 19.9
33.6 (mean
SD)y Average LOS by no. of rehospitalizationsz 1 (mean
SD) 12.3
26.6 2 (mean
SD) 27.3
26.8 3 (mean
SD) 30.9
33.8 4 (mean
SD) 61.8
73.8

C5-8 Grades AeC (nZ190)

Paraplegia Grades AeC (nZ351)

All Grade Ds (nZ147)

All SCI (NZ951)

35.3

35.3

23.8

36.2

22.1 8.9 2.1 1.5 1.41
1.44

22.5 8.8 1.7 2.3 1.28
1.09

19.0 2.7 0.7 2.1 1.33
1.67

23.3 8.4 2.4 1.7 1.37
1.21

15.6
43.6

11.3
17.3

15.0
25.5

15.5
30.8

7.4
9.2 10.8
7.2 17.0
9.1 47.0
28.9

8.5
11.5 17.2
16.0 26.7
15.5 52.4
36.2

8.4
11.3 38.3
51.2 61.0 54.5
6.4

9.5
17.6 20.7
22.8 28.7
26.7 54.3
44.8

* Among those rehospitalized. y Among those rehospitalized. z Four patients who were rehospitalized 300 days during the follow-up period were excluded.

transportationdfactors commonly thought to be associated with risk for rehospitalization). Those who were rehospitalized were more likely to report that they had access to a physician who was knowledgeable about SCI.

Patient rehabilitation experience and rehospitalization We explored the patient’s rehabilitation experience to help identify potential features of the experience associated with rehospitalization (table 3). Those rehospitalized had a longer duration between onset of injury and rehabilitation admission (onset days) and less intensive physical, occupational, and recreational therapy than those not rehospitalized. Medical complications in rehabilitation were a harbinger of rehospitalization in the postrehabilitation period: those rehospitalized were more likely to have had a pressure ulcer or a UTI during rehabilitation (which may have been acquired while in acute care) or during an interrupted portion of the rehabilitation stay. The vast majority of patients with SCI reported that they had received adequate patient education in bowel, bladder, skin care, medications, nutrition, respiratory care, and other potential medical complications often associated with rehospitalization. We found no relation between quality of self-reported patient education and rehospitalization. We also examined the level of clinician-rated patient engagement in physical and occupational therapy using the 6-point Pittsburgh Rehabilitation Participation Scale. We found no significant differences in participation levels between those rehospitalized and not rehospitalized.

Patient discharge destination and rehospitalization The majority (nZ831, 87.4%) of SCI rehabilitation patients were discharged to home. Those discharged to home were less likely to be rehospitalized (33.8%), while those discharged to a skilled nursing facility (nZ67) were more likely to be rehospitalized

(52.2%; PZ.004). This difference does not take into account differences in patient severity or case mix.

Multivariate analysis The study sample for the logistic regression analysis was limited to 816 of the 951 patients. We excluded 135 patients who had missing data on candidate predictor variables (14.2%). Excluded patients were not significantly different from those remaining in the logistic regression analyses. The final logistic regression model identified 10 variables (P.05) associated with rehospitalization (table 4). Two of these were patient demographic characteristicsdbeing a woman and Medicaid as primary payer (a surrogate for lower income status). Four were related to the patient’s health and functional status at admission: those with higher severity scores and lower cognitive functional scores were more likely to be rehospitalized. Only 2 rehabilitation practice variables were associated with rehospitalization: patients who received less intensive physical therapy and had shorter LOSs had increased odds of rehospitalization. Our model validated well using the 25% validation sample; the Hosmer-Lemeshow test P values were .90 and .71 for the data analysis and validation datasets, respectively, well above the .10 threshold for the model to be considered well validated. The study’s 6 SCI rehabilitation centers varied nearly 2-fold in the rates at which their patients were rehospitalizeddfrom 27.8% to 50% (table 5). They also varied in the rates at which their patients were rehospitalized 2 timesdfrom 5.1% to 19.2%. In the logistic regression analyses, we sought to determine how much of the odds of rehospitalization were associated with differences in case-mix and practice patterns, and how much of the odds might be associated with the facility or the levels of community support in which centers were located, controlling for case-mix differences. Patients from 2 centers had higher ORs (ORZ2.721, PZ.007; ORZ2.085 PZ.044) than the reference center. We also sought to determine the predicted rates of rehospitalization for each rehabilitation center knowing only the patient www.archives-pmr.org

26.2 30.2 14.0 14.5
20.7 4.9
1.1

18.6* 22.9z 9.1z 9.3
9.8* 5.1
0.9

characteristics at rehabilitation admission (patient demographic, health, and functional profiles) without considering what rehabilitation center interventions or community resources may have mitigated readmissions. We then compared predicted and actual rehospitalization rates between centers (see table 5). Centers did not vary greatly in their rates of predicted rehospitalization; we observed a 16.1 percentage-point spread among centers in the differences between predicted and actual rates of rehospitalization. No center exceeded a 10 percentage-point difference between their actual and predicted rates of rehospitalization.

Discussion The 36.2% 12-month SCI rehospitalization rate reported here is similar to rates reported in annual reports of the National SCI Statistical Center (33.1%).41,42 Direct comparisons with other studies are difficult owing to differences in study selection criteria, case mix, duration of follow-up periods (1 to 6y), and differences in health delivery and financing systems across countries. Health systems in other countries do not differentiate between acute and postacute rehabilitation as sharply as those in the U.S. Also, rehabilitation centers in other countries typically have longer lengths of rehabilitation stay that shorten the duration of the postdischarge follow-up period when using the anniversary date of the patient’s injury as the endpoint. Still, 12-month rehospitalization rates approaching 50% are not uncommon.1,2,4-6

Reasons for rehospitalization

Abbreviation: RT, recreational therapy. * P<.01. y P<.001. z P<.05. x Average length of all interrupted stays for those who had an interrupted stay.

20.3 25.2 7.3z 5.8
3.9y 5.1
0.8 22.4 34.3 14.9 19.9
24.7 4.9
1.1 26.2z 26.9 12.4 8.1
6.7 4.9
1.0 41.5 36.4 17.8 9.7
9.1 4.8
1.1

www.archives-pmr.org

S93

5.7 22.9 11.4 11.0
12.4 5.3
0.7 19.8 25.1 9.3 12.9
13.8* 5.2
0.9 19.4 24.2 10.5 19.1
30.9 4.9
1.1

4.5 10.7 6.2 6.1
2.8* 5.3
0.8

67.5
18.3y 59.1
16.8 24.4
21.0* 57.4
15.9 58.0
16.0 20.7
20.7 62.9
20.3 57.8
18.3 11.1
12.7 71.3
21.4y 52.6
17.1 26.6
22.8 59.0
14.7 63.5
15.0 22.6
20.7 61.7
16.9* 59.4
13.8 22.5
17.0z 51.7
11.9 59.4
14.9 17.7
14.9

64.9
13.5* 69.0
14.7z 29.4
23.3z

60.3
17.0 53.8
16.0 25.1
25.5

70.2
15.3* 61.2
16.0 17.2
17.0z

27.8
25.6y 55.7
36.1 36.7
31.4 55.4
36.7 16.7
13.2 33.3
20.8 16.6
13.6 29.6
20.3 29.4
27.4 43.8
21.5 31.1
24.3 45.6
29.5 27.7
20.1y 72.0
41.4y 42.7
36.3 57.2
26.8 33.9
30.9* 77.8
40.5 45.2
34.9 72.4
43.9

Onset days (mean
SD) Rehabilitation LOS (mean
SD) Therapy intensity Physical therapy, min/d (mean
SD) Occupational therapy, min/d (mean
SD) RT, min/d (mean
SD) Medical complications Had pressure ulcer (%) Had UTI (%) Had interrupted stay (%) Length of interrupted stay (mean
SD)x Patient participation score (mean
SD)

Rehospitalized? Rehospitalized? Rehospitalized? Rehospitalized? Rehospitalized?

Yes (nZ118) No (nZ145) Yes (nZ67) No (nZ123) Yes (nZ124) No (nZ227) Yes (nZ35) No (nZ112) Yes (nZ344) No (nZ607)

All Grade Ds Paraplegia Grades AeC C5-8 Grades AeC C1-4 Grades AeC

Feature

Table 3

Rehospitalization by patient rehabilitation experience

All SCIs

Rehospitalization after SCI rehabilitation

While variation in rehospitalization rates exists across studies, the reasons for rehospitalization are nearly always the same and in the same rank order.1,2,4,6,7,41,42 As in this study, the 3 leading conditions are (1) genitourinary conditions (mainly UTIs), (2) respiratory conditions (mainly pneumonia), and (3) skin-related conditions (mainly pressure ulcers). These 3 conditions accounted for the majority of all rehospitalization days for which the reason was known, exclusive of days rehospitalized for additional rehabilitation. Pressure ulcers accounted for a disproportionate number of rehospitalization days. Because rehospitalization LOSs for pressure ulcer repair are typically much longer than LOSs for other conditions, such as UTI, the total rehospitalization days because of pressure ulcers in other studies sometimes exceed those for UTI, despite the lower incidence of pressure ulcers relative to UTIs.5,7 However, this was not the case in the present study. These 3 conditions are often considered potentially preventable. However, the high rate of rehospitalization among specialized SCI rehabilitation centers in this study underscores that we are still far from achieving success in preventing these conditions. Innovative care and surveillance are needed to reduce these rates. We did not distinguish between planned and unplanned readmissions, because the distinction is sometimes difficult to make. All 3 of the leading reasons for readmission, however, are almost always unplanned events.

Rehabilitation and rehospitalization We explored features of the rehabilitation experience to ascertain what might be associated with rehospitalization. We found that

S94

G. DeJong et al

Table 4

Logistic regression: risk of 1 rehospitalizations (nZ816)

Independent Variable

B
SE

Wald

P

Exp (B)*

95% CI for Exp (B)

Sex: female Payer: Medicaid Admission CSI Admission body mass index Admission cognitive FIMy Case weight Rehabilitation LOS Physical therapy (min/d) Center Xz Center Yz

0.428
.202 0.674
.207 0.012
.006 0.030
.014 0.012
.005 0.255
.100 0.007
.003 0.030
.006 1.001
.370 0.735
.365

4.513 10.479 3.945 4.704 5.704 6.550 4.435 27.750 77.328 4.056 cZ70.6%

.034 .001 .047 .030 .017 .010 .035 .000 .007 .044

1.535 1.956 1.012 1.030 0.988 1.290 0.993 0.963 2.721 2.085

1.034e2.279 1.303e2.936 1.000e1.024 1.003e1.059 0.978e0.998 1.061e1.568 0.986e1.000 0.960e0.981 1.318e5.617 1.020e4.264

NOTE. Variables included in the model are: age, sex, marital status, education level, English as the primary language, primary payer, employment status, admission CSI, admission body mass index, occupational therapy minutes per day, physical therapy minutes per day, RT minutes per day, therapy participation scale, admission motor FIM, admission cognitive FIM, access to transportation, access to physicians knowledgeable about SCI, home is accessible, pressure ulcers, UTI, CMI, discharge location, and sites. Of the 951 participants in the study, 816 participants were included in the logistic regression analyses; 293 were rehospitalized, 523 were not rehospitalized. Others were excluded because of missing data on key variables such as “had access to a physician knowledgeable about SCI.” Abbreviations: CI, confidence interval; RT, recreational therapy. * Odds ratio estimate. y Rasch-adjusted cognitive FIM. z Center unnamed.

those who had fewer hours of physical, occupational, and recreational therapy per day were more likely to be rehospitalized, but after adjusting for covariates, logistic regression analysis identified only fewer hours of physical therapy per day as being associated with increased odds of rehospitalization. We also found that longer rehabilitation stays were associated with decreased odds of rehospitalization. This finding suggests that rehabilitation provides more time for both patients and staff to manage the underlying medical vulnerabilities. However, a longer rehabilitation LOS narrows the window from rehabilitation discharge to 12-month follow-up during which a rehospitalization event may have occurred. Patient education in matters related to health maintenance is important in preventing medical complications leading to rehospitalization. The relation between patient education, medical complications, and rehospitalization, however, is not

Table 5

straightforward and presents a conundrum. For example, a patient who received more education with respect to skin care may have been more likely to have been rehospitalized, because the additional education may have come in the wake of an inpatient rehabilitation pressure ulcer that placed the patient at greater risk for a rehospitalization episode in the follow-up period. Thus, in the follow-up portion of the study, we queried respondents if they believed they had received the education they needed in the areas of bowel and bladder management, skin care, medications, nutrition, respiratory care, and risk for other SCI complications. Patients uniformly, and at high levels of agreement (90%), reported that they had received sufficient education in these domains. In the absence of being able to draw a direct line from lack of patient education to medical complication and rehospitalization, we are left with patient perceptions about the adequacy of their inpatient education experience,

Percent of patients with SCI rehospitalized in first 12 months since injury by rehabilitation center At Least 1 Rehospitalization*

Rehabilitation Center

Predicted (%)

Actual (%)

Percentage Point Difference

2 Rehospitalizations Actual (%)

Center Center Center Center Center Center Total

34.1 36.9 36.2 35.7 34.0 40.8 36.2

27.8 36.3 36.5 40.4 40.7 50.0 36.2

6.9 0.3 0.3 4.7 6.7 9.2 NA

5.1 13.1 14.8 8.7 19.2 14.1 11.2

A B C D E F

NOTE. In the interest of masking individual facilities, we will only provide percentages, not numbers of center patients, because one would otherwise be able to identify the center by volume. It is important to know how much variability there is from one facility to another. Differences among facilities do not necessarily reflect differences in facility quality as much as they may represent access to postrehabilitation health services in the local area. Abbreviation: NA, not applicable. * In ascending order of actual rehospitalization rate.

www.archives-pmr.org

Rehospitalization after SCI rehabilitation

S95

which may or may not have been adequately tested by the passage of time.

contextual variables are also difficult to measure and control statistically.

Postrehabilitation factors and rehospitalization

Implications for future research

We also explored postrehabilitation lifestyle and environmental supports, such as home accessibility, access to transportation, participation in regular exercise, and participation in creative, recreational, and social activities. We found high levels of environmental support and high rates of participation, which suggested no direct association with rehospitalization. We also asked about access to a physician who was knowledgeable about SCI, assuming that having such a physician might expose the respondent to more opportunities for interventions that could forestall a readmission. Unlike other studies, patients in this study reported high levels of access to physicians knowledgeable about SCI.43 However, the physician knowledgeable about SCI may not be the physician who was managing the patient’s primary medical care, but may have been the physician they saw in the rehabilitation center. Many primary care physicians are not knowledgeable about SCI and may not be as proactive in preventing secondary conditions. Delays in receiving primary care to address emergent issues can allow issues to escalate and require rehospitalization. Management of medical complications related to genitourinary, respiratory, and skin systems are a challenge even in the inpatient rehabilitation hospital environment. However, once a person is discharged from the hospital, multiple variables can influence how well even the most educated patient can navigate the health care system to help prevent secondary conditions. A person may have limited access to physical assistance, supplies, or equipment needed for pressure reliefs, respiratory treatment, or bladder care. Transportation and insurance issues can limit a person’s ability to receive care. Some health plans, such as Medicaid managed care plans, limit coverage for supplies and equipment or require special approvals leading to delays in receipt of needed devices or care. Moreover, some providers refuse to accept patients who participate in certain health plans because of low payments or policies that require extended approval processes.43 Our study identified patients who are at higher risk for complications leading to rehospitalization. That subset could possibly benefit from a health care navigator who can quickly identify potential medical issues and barriers to care and thus facilitate more timely interventions that can help manage emergent issues and avert rehospitalization. Prevention and surveillance directed at the top reasons for rehospitalization may have a high pay-off in reducing rehospitalization rates.44

Despite the study’s in-depth data, we uncovered few patient and practice variables associated with postrehabilitation rehospitalization in the first year of injury. Our findings suggest that we need to look beyond patient-centric variables and look more intensely at health system variables that are often more difficult to quantify and measure. Our follow-up survey did not adequately address features of the health care system that patients encountered in their local communities on their discharge from rehabilitation, and thus we could not adequately identify those features that may be associated with rehospitalizations. Yet, given the scope of the rehospitalization challenge, potential factors underlying center and community differences warrant closer examination. Future studies should examine more carefully the resources in the community (eg, availability of primary care, access to durable medical equipment) that enable an individual to manage more effectively the health challenges associated with living with an SCI, and thus avert rehospitalization. Various rehabilitation centers may want to compare their rehospitalization rates and compare discharge practices and local community variables that may affect differences in their respective risk-adjusted rehospitalization rates, and, on the basis of such comparisons, forge more effective continuity of care for patients with SCI with postdischarge providers and health systems. Those who were rehospitalized reported more depressive symptomatology at 12 months. This begs the question of whether depression leads to greater risk of rehospitalization or whether rehospitalization leads to higher depression scores. Future research should explore this dimension of the SCI experience as possibly contributing to a risk for postdischarge medical complications and rehospitalization.

Rehabilitation center differences in rehospitalization rates Finally, the study’s logistic regression analyses identified 2 centers that had higher than expected differences in rehospitalization rates after controlling for patient-related covariates. Other than more intensive physical therapy, we did not identify any practice variables that were associated with rehospitalization. Individual centers operate within larger health systems and communities that vary in their ability to respond to the needs of individuals with SCIdcapacities that are often beyond the ability of the individual center to control and yet are reflected in the center variable. These www.archives-pmr.org

Study limitations Several study features limit the generalizability of the results; 3 are mentioned here. While 92% of the eligible sample completed a 12-month follow-up interview, those who did not were older, more likely to be a Medicare beneficiary, more obese, more likely to have less than a high school education, and more likely to have lower FIM scores at admission. Thus, the findings may underrepresent rehospitalization among individuals with SCI with these characteristics. Another limitation was the high rate (40.2%) of unreported reasons for rehospitalizationda limitation that was confined mainly to 1 of the larger participating rehabilitation centers. Among other study centers, rehospitalization rates and reasons for rehospitalization mirror those reported in recent years to the National SCI Statistical Center.41,42,45 Even at the center with more unknown reasons for rehospitalization, patients with and without known reasons were very similar, and those with known reasons still mirrored the experience of other study centers and those reported by the National SCI Statistical Center. Finally, the study also did not capture the exact dates of rehospitalization, and hence we were unable to determine the time from rehabilitation discharge to first rehospitalization and any subsequent rehospitalizations.

S96

Conclusions Compared with earlier studies, rehospitalization rates among individuals with SCI in the first year postinjury remain high for conditions that are thought to be preventable, although their preventability is by no means certain given the altered physiology associated with acquiring an SCI. As in previous studies, the main reasons for rehospitalization are largely the sameddiseases of the genitourinary system, respiratory system, and diseases related to skin and subcutaneous tissue system. Being a woman, having Medicaid as the main payer, and medical comorbidities were associated with increased odds of rehospitalization. Those who had more intensive physical therapy had lower rates of rehospitalization. Some center-to-center variation in rehospitalization rates remains unexplained after case-mix and practice differences were considered. Future research on SCI postrehabilitation rehospitalization should examine more carefully the role of health system variables in meeting the health needs of individuals with SCI and averting rehospitalization.

Keywords Patient readmission; Rehabilitation; Spinal cord injuries

Corresponding author Gerben DeJong, PhD, Center for Post-acute Innovation & Research, MedStar National Rehabilitation Hospital, 102 Irving St NW, Washington, DC 20010. E-mail address: Gerben. [email protected].

References 1. Cardenas DD, Hoffman JM, Kirshblum S, McKinley W. Etiology and incidence of rehospitalization after traumatic spinal cord injury: a multicenter analysis. Arch Phys Med Rehabil 2004;85:1757-63. 2. Dryden DM, Saunders LD, Rowe BH, et al. Utilization of health services following spinal cord injury: a 6-year follow-up study. Spinal Cord 2004;42:513-25. 3. Jaglal SB, Munce SE, Guilcher SJ, et al. Health system factors associated with rehospitalizations after traumatic spinal cord injury: a population-based study. Spinal Cord 2009;47:604-9. 4. Middleton JW, Lim K, Taylor L, Soden R, Rutkowski S. Patterns of morbidity and rehospitalisation following spinal cord injury. Spinal Cord 2004;42:359-67. 5. Pagliacci MC, Celani MG, Spizzichino L, Zampolini M, Franceschini M. Hospital care of postacute spinal cord lesion patients in Italy: analysis of readmissions into the GISEM study. Am J Phys Med Rehabil 2008;87:619-26. 6. Savic G, Short DJ, Weitzenkamp D, Charlifue S, Gardner BP. Hospital readmissions in people with chronic spinal cord injury. Spinal Cord 2000;38:371-7. 7. Vaidyanathan S, Soni BM, Gopalan L, et al. A review of the readmissions of patients with tetraplegia to the Regional Spinal Injuries Centre, Southport, United Kingdom, between January 1994 and December 1995. Spinal Cord 1998;36:838-46. 8. Benbassat J, Taragin M. Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med 2000;160: 1074-81. 9. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009; 360:1418-28.

G. DeJong et al 10. Measure Applications Partnership of the National Quality Forum. Coordination strategy for post-acute care and long-term care performance measurement. Final report. Washington (DC): National Quality Forum; 2012. Available at: http://www.qualityforum.org/Publications/2012/02/ MAP_Coordination_Strategy_for_Post-Acute_Care_and_Long-Term_ Care_Performance_Measurement.aspx. Accessed August 12, 2012. 11. National Quality Forum. Coordination strategy for healthcare-acquired conditions and readmissions across public and private payers. Final report. Washington (DC): National Quality Forum; 2011. Available at: http://www.qualityforum.org/Publications/2011/10/MAP_Coordination_ Strategy_for_HealthcareAcquired_Conditions_and_Readmissions_Across_ Public_and_Private_Payers.aspx. Accessed August 12, 2012. 12. MedPAC Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy. Washington (DC): Medicare Payment Advisory Commission; March 2012. 13. Whiteneck G, Gassaway J, Dijkers M, Jha A. New approach to study the contents and outcomes of spinal cord injury rehabilitation: the SCIRehab Project. J Spinal Cord Med 2009;32:251-9. 14. Whiteneck G, Gassaway J, Dijkers M, et al. The SCIRehab project: treatment time spent in SCI rehabilitation. Inpatient treatment time across disciplines in spinal cord injury rehabilitation. J Spinal Cord Med 2011;34:133-48. 15. Marino R. Reference manual for the international standards for neurological classification of spinal cord injury. Chicago: American Spinal Injury Association; 2003. 16. Kirshblum S, Burns S, Biering-Sorensen F, et al. International standards for neurological classification of spinal cord injury (revised 2011). J Spinal Cord Med 2001;34:535-46. 17. Averill RF, McGuire TE, Manning BE, et al. A study of the relationship between severity of illness and hospital cost in New Jersey hospitals. Health Serv Res 1992;27:587-606. 18. Buckle JM, Horn SD, Oates VM, Abbey H. Severity of illness and resource use differences among white and black hospitalized elderly. Arch Intern Med 1992;152:1596-603. 19. Clemmer TP, Spuhler VJ, Oniki TA, Horn SD. Results of a collaborative quality improvement program on outcomes and costs in a tertiary critical care unit. Crit Care Med 1999;27:1768-74. 20. DeJong G, Hsieh CH, Gassaway J, et al. Characterizing rehabilitation services for patients with knee and hip replacement in skilled nursing facilities and inpatient rehabilitation facilities. Arch Phys Med Rehabil 2009;90:1269-83. 21. Gassaway J, Horn SD, DeJong G, Smout RJ, Clark C, James R. Applying the clinical practice improvement approach to stroke rehabilitation: methods used and baseline results. Arch Phys Med Rehabil 2005;86(12 Suppl 2):S16-33. 22. Horn SD, Sharkey PD, Buckle JM, Backofen JE, Averill RF, Horn RA. The relationship between severity of illness and hospital length of stay and mortality. Med Care 1991;29:305-17. 23. Horn SD, Torres A Jr, Willson D, Dean JM, Gassaway J, Smout R. Development of a pediatric age- and disease-specific severity measure. J Pediatr 2002;141:496-503. 24. Ryser DK, Egger MJ, Horn SD, Handrahan D, Gandhi P, Bigler ED. Measuring medical complexity during inpatient rehabilitation after traumatic brain injury. Arch Phys Med Rehabil 2005; 86:1108-17. 25. Willson DF, Horn SD, Smout R, Gassaway J, Torres A. Severity assessment in children hospitalized with bronchiolitis using the pediatric component of the Comprehensive Severity Index. Pediatr Crit Care Med 2000;1:127-32. 26. Horn SD, Smout RJ, DeJong G, et al. Association of various comorbidity measures with spinal cord injury rehabilitation outcomes. Arch Phys Med Rehabil 2013;94(4 Suppl 2):S75-86. 27. Mallinson T. Rasch analysis of repeated measures. Rasch Measurement Transactions 2011;25:1317. 28. Chen CC, Heinemann AW, Granger CV, Linn RT. Functional gains and therapy intensity during subacute rehabilitation: a study of 20 facilities. Arch Phys Med Rehabil 2002;83:1514-23.

www.archives-pmr.org

Rehospitalization after SCI rehabilitation 29. Whiteneck GG, Gassaway J. SCIRehab uses practice-based evidence methodology to associate patient and treatment characteristics with outcomes. Arch Phys Med Rehabil 2013;94(4 Suppl 2):S67-74. 30. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001;16:606-13. 31. Williams RT, Heinemann AW, Bode RK, Wilson CS, Fann JR, Tate DG. Improving measurement properties of the Patient Health Questionnaire-9 with rating scale analysis. Rehabil Psychol 2009;54: 198-203. 32. Federal Register/Vol. 76. No. 186/Monday, September 26, 2011/Rules and Regulations:59256-59263. Washington (DC): Government Printing Office. Available at: http://www.gpo.gov/fdsys/pkg/FR-2011-09-26/ pdf/2011-24671.pdf. Accessed February 10, 2012. 33. Abeyta N, Freeman ES, Primack D, et al. SCIRehab Project series: the social work/case management taxonomy. J Spinal Cord Med 2009;32: 336-42. 34. Cahow C, Skolnick S, Joyce J, Jug J, Dragon C, Gassaway J. SCIRehab Project series: the therapeutic recreation taxonomy. J Spinal Cord Med 2009;32:298-306. 35. Gordan W, Spivak-David D, Adornato V, et al. SCIRehab Project series: the speech language pathology taxonomy. J Spinal Cord Med 2009;32:307-18. 36. Johnson K, Bailey J, Rundquist J, et al. SCIRehab Project series: the supplemental nursing taxonomy. J Spinal Cord Med 2009;32: 329-35.

www.archives-pmr.org

S97 37. Natale A, Taylor S, LaBarbera J, et al. SCIRehab Project series: the physical therapy taxonomy. J Spinal Cord Med 2009;32:270-82. 38. Ozelie R, Sipple C, Foy T, et al. SCIRehab Project series: the occupational therapy taxonomy. J Spinal Cord Med 2009;32:283-97. 39. Wilson C, Huston T, Koval J, Gordon SA, Schwebel A, Gassaway J. SCIRehab Project series: the psychology taxonomy. J Spinal Cord Med 2009;32:319-28. 40. Lenze EJ, Munin MC, Quear T, et al. The Pittsburgh Rehabilitation Participation Scale: reliability and validity of a clinician-rated measure of participation in acute rehabilitation. Arch Phys Med Rehabil 2004;85:380-4. 41. National Spinal Cord Injury Statistical Center. Annual report for the Spinal Cord Injury Model Systems. Birmingham: National Spinal Cord Injury Statistical Center; 2009. 42. National Spinal Cord Injury Statistical Center. Annual report for the Spinal Cord Injury Model Systems. Birmingham: National Spinal Cord Injury Statistical Center; 2010. 43. DeJong G, Hoffman JM, Mead M, et al. Post-rehabilitative health care for individuals with SCI: extending health care into the community. Top Spinal Cord Inj Rehabil 2011;17:46-58. 44. Courage Center. Health care home outcomes. Available at: http:// www.couragecenter.org/ContentPages/hchoutcomes.aspx. Accessed June 11, 2012. 45. National Spinal Cord Injury Statistical Center. Annual report for the Spinal Cord Injury Model Systems. Birmingham: National Spinal Cord Injury Statistical Center; 2011.