Accepted Manuscript Rehabilitation Outcomes among Burns Patients with a Second Admission to an Inpatient Rehabilitation Facility Margaret A. DiVita , PhD, MS Jacqueline M. Mix , MPH Richard Goldstein , PhD Paul Gerrard , MD Paulette Niewczyk , PhD, MPH Colleen M. Ryan , MD Karen Kowalske , MD Ross Zafonte , DO Jeffrey C. Schneider , MD PII:
S1934-1482(14)00241-X
DOI:
10.1016/j.pmrj.2014.05.010
Reference:
PMRJ 1258
To appear in:
PM&R
Received Date: 29 August 2013 Revised Date:
12 May 2014
Accepted Date: 16 May 2014
Please cite this article as: DiVita MA, Mix JM, Goldstein R, Gerrard P, Niewczyk P, Ryan CM, Kowalske K, Zafonte R, Schneider JC, Rehabilitation Outcomes among Burns Patients with a Second Admission to an Inpatient Rehabilitation Facility, PM&R (2014), doi: 10.1016/j.pmrj.2014.05.010. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Rehabilitation Outcomes among Burns Patients with a Second Admission to an Inpatient Rehabilitation Facility
RI PT
Margaret A DiVita PhD, MS1,2, Jacqueline M Mix MPH1, Richard Goldstein PhD3, Paul Gerrard MD3, Paulette Niewczyk PhD, MPH1,4, Colleen M. Ryan MD5,6, Karen Kowalske MD7, Ross
1
SC
Zafonte DO3, Jeffrey C. Schneider MD3,6
Uniform Data System for Medical Rehabilitation, Amherst, NY 2Health Department, State
M AN U
University of New York at Cortland, Cortland, NY 3Dept. of Physical Medicine and Rehabilitation Spaulding Rehabilitation Hospital, Harvard Medical School, Boston MA 4Daemen College, Health Care Studies Dept., Amherst NY 5Sumner Redstone Burn Center, Surgical Services Massachusetts General Hospital, Harvard Medical School, Boston, MA 6Shriners
TE D
Hospitals for Children®-Boston, Boston, MA 7Department of Physical Medicine and
AC C
EP
Rehabilitation, University of Texas Southwestern Medical Center at Dallas, TX
ACCEPTED MANUSCRIPT
Abstract
1
Background: Burn survivors tend to have complex medical issues requiring rehabilitation to
3
improve overall function and quality of life. A subset of burn patients treated in inpatient
4
rehabilitation facilities (IRFs) may require more than one rehabilitation stay for the same injury.
5
Objective: To compare the rehabilitation outcomes among burn patients admitted to an IRF who
6
were discharged to acute care and then readmitted to an IRF with burn patients admitted to an
7
IRF only one time.
8
Design: Retrospective cohort study.
9
Setting: Inpatient rehabilitation facilities.
M AN U
SC
RI PT
2
Participants: Burn injury patients aged 18 years or above admitted to IRFs between 2002 and
11
2011.
12
Methods: We performed a secondary data analysis of data from Uniform Data System for
13
Medical Rehabilitation, a national data repository. Outcomes of the repeaters’ second stay (n =
14
188) were compared to the non-repeaters’ first and only stay (n = 6,855) utilizing linear
15
regression and logistic regression to determine whether repeater status was associated with
16
rehabilitation outcomes.
17
Main Outcome Measurements: Functional status (using the FIM® instrument) at admission,
18
discharge and change, length of stay, FIM® Efficiency (Total FIM® points gained per day) and
19
discharge disposition.
20
Results: Repeater status was inversely associated with discharge FIM® total (Coefficient = -
21
3.42, 95% CI -5.76, -1.07) and FIM® change (Coefficient = -4.05, 95% CI -6.34, -1.75) in linear
22
regression models. No other significant differences were found, and those differences in
23
discharge FIM® total and FIM® change were small.
AC C
EP
TE D
10
ACCEPTED MANUSCRIPT
Conclusions: Differences found in rehabilitation outcomes between the repeater and non-repeater
25
groups were small and may not reflect clinically meaningful differences. Burn injury patients
26
who required a second IRF admission had similar rehabilitation outcomes as burn injury patients
27
who did not require a second IRF admission, emphasizing the value of inpatient rehabilitation for
28
burn injury IRF readmissions.
RI PT
24
29
SC
30 31
M AN U
32 33 34 35
TE D
36 37 38
41 42 43 44
AC C
40
EP
39
45 46
1
ACCEPTED MANUSCRIPT
47
Introduction Due to recent advances in burn care, burn injury patients have improved survival
49
likelihoods.1-3 However, burn survivors generally have complex medical needs and may require
50
intensive rehabilitation to improve their quality of life by increasing their ability to perform basic
51
life activities. Inpatient rehabilitation facilities (IRFs) are one part of a multi-level post-acute
52
care continuum which provides intensive rehabilitation therapy services for patients who require
53
an extensive multidisciplinary team approach to rehabilitation therapy due to the complexity of
54
their nursing, medical management and rehabilitation needs. Patients admitted to an IRF are
55
required to participate in at least three hours of intensive therapy per day. The goal of is to return
56
patients to the community as a high-functioning individuals who require a minimum amount of
57
assistance from caregivers.
SC
M AN U
58
RI PT
48
IRFs often serve the needs of those with the most severe burns who have functional disabilities related to their injuries. Some burn injury patients may require more than one IRF
60
admission for the same acute injury, when subsequent visits to an acute care hospital are needed
61
between rehabilitation stays. Typically, readmissions back to an acute care venue after an IRF
62
admission are an unfavorable outcome and may be reflect unresolved medical problems from
63
acute care, poor transitions of care, or medically complex patients that were not ready to
64
participate in an intensive therapy program. In burn injury patients, it has been shown that
65
functional level at admission, age, and admission classification are significant predictors of acute
66
care transfer.4 Readmissions to acute care hospitals cost millions of Medicare dollars each year.5
67
There is a lack of knowledge concerning the trajectory of patients discharged to an acute
AC C
EP
TE D
59
68
care hospital from an IRF. Patients who are discharged from an IRF to an acute care facility may
69
subsequently be readmitted to an IRF for a second stay. Currently, empirical evidence does not
2
ACCEPTED MANUSCRIPT
exist regarding the rehabilitation outcomes of patients that require a second IRF stay. The aim of
71
this study was to compare the characteristics and outcomes among burn injury patients that were
72
discharged to an acute care hospital during their IRF stay and then required an additional IRF
73
admission (2.7%), with burn patients that had only one IRF stay (97.3%). Our goal in this study
74
was to determine if patients that require an IRF readmission have different rehabilitation
75
outcomes than those who only require one IRF stay. This will assist in determining the
76
rehabilitation needs of burn injury patients that have more complex medical needs.
SC
RI PT
70
78
Methods
79
Data Source
80
M AN U
77
We completed a secondary data analysis with data from the Uniform Data System for Medical Rehabilitation (UDSMR) for burn injury patients discharged between 2002 and 2011.
82
UDSMR is the world’s largest non-governmental medical rehabilitation data repository. Over 800
83
IRFs provide data to UDSMR, which accounts for over 70% of all IRFs in the United States. All
84
IRFs are required by the Centers for Medicare and Medicaid Services to complete the Inpatient
85
Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) to receive payment under the
86
federal prospective payment system. The IRF-PAI collects demographic, social, medical, and
87
functional data. Functional data is measured by the FIM® instrument.
EP
AC C
88
TE D
81
The FIM® instrument is an 18-item functional assessment tool which was designed to
89
evaluate the extent of functional disability and to monitor rehabilitation outcomes. The tool
90
assesses both motor and cognitive domains and includes items such as eating, grooming, bathing,
91
upper body dressing, lower body dressing, toileting, bowel management, bladder management,
92
bed/chair/wheelchair transfer, bed transfer, tub/shower transfer, walk/wheelchair, stairs,
3
ACCEPTED MANUSCRIPT
comprehension, expression, social interaction, problem solving, and memory. Each item in the
94
FIM® assessment is rated on a 7-level scale, with 1 corresponding to total assistance and 7
95
corresponding to complete independence. The individual items are summed to create a
96
composite measure which is utilized to determine the burden of care that is required from a
97
caregiver to assist the patient once they return to the community. The FIM® instrument has been
98
utilized to study functional outcomes in various IRF populations including patients with stroke,
99
traumatic brain injury, spinal cord injury and burn injuries, and has been shown to be a valid and reliable measure of functional status.6,7
101
Study Population
102
M AN U
100
SC
RI PT
93
Our study included adult patients with an age of 18 or greater with a primary diagnosis of burn injury. The identification of burn injury patients was done by impairment group code,
104
which is a code entered upon IRF admission indicating the primary reason for entering the
105
rehabilitation program. Burn injury impairment was an inclusion criterion for both initial and
106
repeat admission. Patients were excluded if they were discharged against medical advice (n= 40),
107
or came from a facility with a high frequency of zero onset day patients (n= 11). Onset to
108
rehabilitation is a variable collected by UDSMR which indicates the number of days from onset
109
of the injury to admission to an IRF. Admission to an IRF on onset day zero in the burn injury
110
population is clinically unexpected, as burn injuries typically require medical management in an
111
acute care setting prior to their transition to an IRF. It has been demonstrated in prior studies
112
utilizing this dataset that facilities with >5% of cases admitted on onset day zero have
113
characteristics that are not consistent with other facilities in the dataset.8,9 There were 2 facilities
114
in our dataset that demonstrated >5% of patients admitted on onset day zero, and cases from
115
these facilities were excluded from the analysis.
AC C
EP
TE D
103
4
ACCEPTED MANUSCRIPT
116
Patients who had more than one admission to any IRF but had the same date of burn injury were defined as repeater patients. In addition, the first visit for repeater patients ended in a
118
discharge to an acute care hospital. Date of birth and gender variables were also utilized to
119
confirm repeated admission to an IRF. The comparison population consists of “non-repeater”
120
patients with a single IRF admission.
RI PT
117
121
Outcome Variables
SC
122
Continuous rehabilitation outcome variables examined include admission FIM® total,
124
rehabilitation length of stay, discharge FIM® total, FIM® change, and FIM® efficiency. FIM®
125
change is defined as the difference in FIM® total rating from admission to discharge from the
126
IRF. FIM® efficiency is defined by the number of FIM® points gained per day and is
127
mathematically defined as the FIM® change divided by the length of stay (LOS) in the inpatient
128
rehabilitation facility. We also examined discharge disposition as an outcome variable, which
129
was categorized as community versus non-community settings.
TE D
M AN U
123
132
Statistical Analysis
Descriptive statistics including means and standard deviations for continuous variables
AC C
131
EP
130
133
and counts and percentages for categorical variables were calculated for the total population
134
together, and separately for the repeater and non-repeater groups. Independent sample t-tests
135
were performed for normal continuous variables (age and admission FIM® total), Kruskal-Wallis
136
tests were performed for non-normal continuous variables (onset days and length of stay), and
137
chi-square tests were performed for categorical variables. In addition, a separate analysis as
138
completed comparing the characteristics of the repeaters’ first admission to their second 5
ACCEPTED MANUSCRIPT
admission; particularly we compared the admission FIM® average, the discharge FIM® average,
140
FIM® gain, the length of stay, and the percentage of cases by comorbidity tier. We utilized paired
141
sample t-tests when comparing continuous variables between the first and second admission
142
among repeaters and the McNemar test for the categorical variable of comorbidity tier. A p-value
143
of less than .05 was considered statistically significant.
RI PT
139
Multivariable backwards linear regression analyses were used to calculate coefficients
144
and 95% confidence intervals for repeater status for continuous rehabilitation outcome variables.
146
Multivariable backwards logistic regression analysis was used to calculate odds ratios (OR) 95%
147
confidence intervals for repeater status for discharge disposition. Covariates were kept in the
148
models if the associated p value was less than .20 since using a traditional threshold of p < .05
149
can lead to the exclusion of important variables.10 Covariates included age, gender, race, marital
150
status, employment status prior to injury (employed, not employed, retired), primary payer
151
(Medicare, Medicaid, worker’s compensation, commercial, other), pre-hospital living setting
152
(community, other), pre-hospital living situation (alone or with others), pre-rehabilitation setting
153
(acute care hospital, other), and admission comorbidity tier (Tier A-none, Tier B-major, Tier C-
154
medium, Tier D-minor). The standard errors for all models were adjusted for clustering at the
155
facility level. All analyses were performed using SPSS version 21 and STATA version 12.1.
156 157 158
Results
AC C
EP
TE D
M AN U
SC
145
There were 7,094 adult patients from 2002 to 2011 with a primary diagnosis of burn
159
injury in the UDSMR® database. Of those subjects, 40 were discharged against medical advice,
160
and 11 were from zero onset facilities. Therefore, a total of 7,043 subjects meet inclusion
161
criteria. Among those patients, there were 188 who had more than one IRF stay for the same 6
ACCEPTED MANUSCRIPT
burns injury and were discharged to acute care after their first admission to the IRF (repeater
163
patients) and 6,855 patients who only had one IRF admission (non-repeater patients). Among the
164
repeater patients, 60% of the cases were readmitted to an IRF within one month (31 days) of
165
discharge from the first IRF admission (range 5-235 days). Demographic, medical and
166
rehabilitation characteristics of the study population are displayed in Table 1. The repeaters were
167
more functionally impaired at both admission and discharge of their first visit, compared to the
168
second visit; admission FIM® average 53.1 versus 65.0, and discharge FIM® average 68.2
169
versus 90.4, respectively. In addition, the first rehabilitation stay was shorter by an average of
170
over 8 days; rehabilitation length of stay average 13.4 versus 21.8. The repeaters were also more
171
likely to be in the most severe CMS comorbidity tier group (tier B) in their first visit when
172
compared to their second visit, 16% versus 10.1%, respectively. It is noted that part of the
173
inclusion criteria for repeaters was a discharge destination of acute care for their first visit,
174
explaining the large difference in discharged destination between the first and second visits. All
175
other differences between the first and second visit of the repeaters were negligible.
SC
M AN U
TE D
176
RI PT
162
There were significant differences in demographic and medical variables between repeaters and non-repeaters in onset days to rehabilitation, (87 days versus 32 days, p < .001),
178
employment status (30% retired versus 37% retired, p = .03), primary payer source (30%
179
Medicare versus 35% Medicare, p = .02), pre-hospital living situation (19% lived alone versus
180
28% lived alone, p = .01), and admission setting type (99% from acute care versus 77% from
181
acute care, p < .001). No other demographic or medical characteristics were significantly
182
different. All rehabilitation related characteristics were significantly different between repeaters
183
and non-repeaters except for FIM® change. Repeater patients had a significantly lower admission
184
FIM® total (65 versus 70, p < .001), a significantly lower discharge FIM® total (90 versus 96, p