- Email: [email protected]
Enteral Nutrition for Patients With Traumatic Brain Injury in the Rehabilitation Setting: Associations With Patient Preinjury and Injury Characteristics and Outcomes
Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2015;96(8 Suppl 3):S245-55
ORIGINAL RESEARCH
Enteral Nutrition for Patients With Traumatic Brain Injury in the Rehabilitation Setting: Associations With Patient Preinjury and Injury Characteristics and Outcomes Susan D. Horn, PhD,a Merin Kinikini, RD, FNP, DNP, CNSC,b Linda W. Moore, MS, RDN, CCRP,c Flora M. Hammond, MD,d,e Murray E. Brandstater, MD,f Randall J. Smout, MS,a Ryan S. Barrett, MSa From the aInstitute for Clinical Outcomes Research, International Severity Information Systems, Inc, Salt Lake City, UT; bNeuro Specialty Rehabilitation Unit, Intermountain Medical Center, Salt Lake City, UT; cHouston Methodist Hospital, Houston, TX; dCarolinas Rehabilitation, Charlotte, NC; eIndiana University, Indianapolis, IN; and fLoma Linda University Medical Center, Loma Linda, CA. Current affiliation for Horn, Department of Population Health Sciences, University of Utah School of Medicine, Salt Lake City, UT.
Abstract Objective: To determine the association of enteral nutrition (EN) with patient preinjury and injury characteristics and outcomes for patients receiving inpatient rehabilitation after traumatic brain injury (TBI). Design: Prospective observational study. Setting: Nine rehabilitation centers. Participants: Patients (NZ1701) admitted for first full inpatient rehabilitation after TBI. Interventions: Not applicable. Main Outcome Measures: FIM at rehabilitation discharge, length of stay, weight loss, and various infections. Results: There were many significant differences in preinjury and injury characteristics between patients who received EN and patients who did not. After matching patients with a propensity score of >40% for the likely use of EN, patients receiving EN with either a standard or a highprotein formula (>20% of calories coming from protein) for >25% of their rehabilitation stay had higher FIM motor and cognitive scores at rehabilitation discharge and less weight loss than did patients with similar characteristics not receiving EN. Conclusions: For patients receiving inpatient rehabilitation after TBI and matched on a propensity score of >40% for the likely use of EN, clinicians should strongly consider, when possible, EN for 25% of the rehabilitation stay and especially with a formula that contains at least 20% protein rather than a standard formula. Archives of Physical Medicine and Rehabilitation 2015;96(8 Suppl 3):S245-55 ª 2015 by the American Congress of Rehabilitation Medicine
The provision of adequate nutritional support for patients with moderate-to-severe traumatic brain injury (TBI) has been a clinical challenge for decades.1-3 Patients’ primary and secondary injuries create unique metabolic derangements that pose issues such as optimal timing and route of nutrition, appropriate fluid and Supported by the National Institutes of Health, National Center for Medical Rehabilitation Research (grant no. 1R01HD050439-01), the National Institute on Disability and Rehabilitation Research (grant no. H133A080023), and the Ontario Neurotrauma Foundation (grant no. 2007-ABI-ISIS-525). Publication of this article was supported by the American Congress of Rehabilitation Medicine. Disclosure: none.
electrolyte balance, drug administration, and dysphagia. In addition, it may be difficult to maintain tubes and lines in a confused or agitated patient, particularly in a rehabilitation setting. Individuals with TBI have a much higher resting metabolic expenditure (RME) than do patients without TBI.4 In fact, with severe TBI, RME has been found to range up to 240% of RME in patients without TBI; they are similar in metabolic response to patients with burns over 20% to 40% of their body surface area.4 The consequences of hypermetabolism, hypercatabolism, and altered immune function in patients with acute TBI are excessive
0003-9993/15/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.06.024
S246
S.D. Horn et al
protein breakdown, which can lead to malnutrition.5 However, patients with TBI requiring hospitalization often do not, or cannot, consume enough nutrition to support their increased requirements for recovery and rehabilitation.5 Enteral nutrition (EN) administered as early as possible has been established as the preferential route of nutritional support for this population versus total parenteral nutrition (TPN); some centers use a combination of EN and TPN in the early stages of injury if the patient does not tolerate adequate amounts of EN alone.6 Chourdakis et al7 recently reported that early EN may affect hormonal response to TBI and suggests that this may reduce catabolic and inflammatory processes induced by TBI. There appears to be a consensus on early initiation of EN, but less definitive are recommendations on advancement timing and formula components (eg, whether to use specialty formulas such as those containing immune-enhancing properties).8-12 The Institute of Medicine recommended inclusion of nutrient additives (eg, n-3 fatty acids, creatine, choline, and zinc) as potentially beneficial for recovery after TBI.3,5 Patients with TBI, similar to other trauma patients, likely require 2.0 to 2.5g of protein/kg at a minimum, especially during the early period after injury.6,13 Evaluation of the duration of a higher protein requirement has not been reported in the literature, but it likely correlates with metabolic status. If increased metabolic rates extend into the rehabilitation setting, then increased protein needs may also be present. Swallowing disorders and decreased behavioral/cognitive skills are frequently present in patients with severe brain injury and significantly affect oral intake.14 Persons who swallow abnormally take much longer to start eating and to achieve total oral feeding, and they require nonoral supplementation 3 to 4 times longer than those who swallow normally.14 Patients with severe TBI may also have intolerance to EN, which hampers survival and rehabilitation.15 Haddad and Arabi16 discuss proactive use of prokinetic agents, such as erythromycin and metoclopramide, as well as postpyloric feeding as ways to overcome problems of gastric distention and intolerance experienced by patients with TBI. Most reports11,13,17,18 regarding nutrition in patients with TBI address the route (TPN vs EN) and/or timing (early vs late) of initiation of nutritional support related to hospital admission and have addressed outcomes such as mortality or length of stay (LOS) in the acute care setting. We could not find any published reports that address the role of nutritional support during rehabilitation of patients with TBI. A practice-based evidence (PBE) study in stroke rehabilitation found that the use of EN support for 25% of the rehabilitation stay for patients with severe stroke was a significant factor in predicting higher discharge FIM total and motor scores, controlling for patient and other treatment differences.19,20 It is not known if these findings are applicable to the population with TBI receiving rehabilitation. This article describes nutritional support methods used for patients in a TBIPBE study during rehabilitation21 and examines associations of patient preinjury and injury characteristics with use and duration
List of abbreviations: CSI EN LOS PBE RME TBI TPN
Comprehensive Severity Index enteral nutrition length of stay practice-based evidence resting metabolic expenditure traumatic brain injury total parenteral nutrition
of EN support, as well as associations of EN with outcomes, controlling for patient differences.
Methods This comparative effectiveness PBE study examined the differential effects of a wide array of specific treatments administered in 10 acute inpatient rehabilitation facilities serving patients with TBI in a brain injury specialty unit who were enrolled from October 2008 to September 2011. The 10 participating centers constituted a convenience sample of adults with TBI on the basis of their willingness to participate in the research. The institutional review board at each center approved the study; each patient or his/her proxy gave informed consent.
Participants Inclusion criteria were as follows21: (1) Sustained a TBI, defined as damage to brain tissue caused by external force and evidenced by loss of consciousness, posttraumatic amnesia, skull fracture, or objective neurological findings. Diagnoses included International Classification of Diseases, 9th Revision, Clinical Modification codes consistent with the Centers for Disease Control and Prevention Guidelines for Surveillance of Central Nervous System Injury; (2) received inpatient care in a designated brain injury rehabilitation unit of one of the participating rehabilitation facilities; and (3) were 14 years or older and treated in an adult rehabilitation unit. Patient variables Patient characteristics, including demographic characteristics and injury characteristics, were recorded on the basis of clinicians’ suggestions as well as previous research indicating their importance in populations with TBI. In addition to patient data available on admission, we collected information on patients’ status during their rehabilitation stay, including the presence of aphasia and dysphagia. Table 1 lists the characteristics of study patients. Functional dependence We used admission FIM, an assessment of independent functioning consisting of 18 items in 2 domains: motor (13 items) and cognitive (5 items). Each FIM item was rated on a 7-category scale, ranging from 1 (total assistance required) to 7 (complete independence). To eliminate distortion in quantifying the status of patients whose capability is at the extremes of the instrument’s range, FIM motor and cognitive scores were recoded separately using tables published by Heinemann et al,22 which were based on Rasch analysis of FIM data of a large sample with brain injury. Injury severity and comorbidity The primary medical severity measure used was the Comprehensive Severity Index (CSI), which defines severity as the physiological and psychosocial complexity presented owing to the extent and interactions of a patient’s disease(s).21 The CSI is age- and disease-specific and is independent of treatments. It provides an objective, consistent method to operationalize patient severity of illness on the basis of >2100 individual signs, symptoms, and physical findings and >5600 disease-specific criteria sets related to all of a patient’s disease(s). More details about the CSI appear elsewhere.21 The CSI modification used here allowed separating severity of brain injury (called brain www.archives-pmr.org
Nutritional support for patients with TBI Table 1
S247
Characteristics of patients with and without EN
Characteristic Demographic characteristics Sex: male (%) Age at rehabilitation admission (y) Race/ethnicity Black White White Hispanic Other and unknown Highest education achieved Some high school, no diploma High school diploma Work toward or completed associate’s degree Work toward or completed bachelor’s degree Work toward or completed master’s/doctoral degree Unknown Marital status at injury Single/never married Married/common law Previously married Other/unknown Able to drive before injury No Yes Unknown Primary payer Medicare Medicaid Private insurance Worker’s compensation Self-pay/none MCO/HMO No-fault auto insurance Other/unknown Admission body mass index (kg/m2) <16 16e18.5 >18.5e25 >25e30 >30e35 >35e40 >40 Unknown Preexisting and coexisting conditions History of alcohol abuse before injury History of drug abuse before injury Anxiety before or during rehabilitation
EN (NZ451)
Table 1 (continued )
No EN (NZ1250) P
Characteristic
72.3 71.6 .807* 38.519.6 47.122.0 <.001y .012z 12.0 80.0 4.9 3.1
17.0 72.7 7.4 2.9 .268z
24.2
22.6
24.2 18.0
26.6 15.6
20.8
19.5
9.5
9.8
3.3
5.8 .114z
47.0 34.6 15.7 2.7
41.2 35.8 19.8 3.1 <.001z
5.5 81.4 13.1
12.6 68.8 18.6 <.001z
11.5 12.2 26.8 9.1 6.9 19.5 8.9 5.1
25.8 17.0 25.3 6.8 4.6 14.0 3.4 3.1 <.001z
1.3 12.4 54.3 20.4 6.4 1.6 1.3 2.2
1.4 6.3 47.0 26.2 9.0 2.9 1.7 5.5
34.8
36.9
.458*
20.0
22.8
.233*
22.0
21.1
.736*
(continued page) (continued in on nextnext column)
www.archives-pmr.org
Depression before or during rehabilitation Hypertension before or during rehabilitation Paralysis before or during rehabilitation Renal failure before or during rehabilitation Brain injury and severity information Cause of injury Fall Motor vehicle collision Sports Violence Other GCS score immediately after injury or after arrival in acute care (%) Mild (13e15) Moderate (9e12) Severe (3e8) Intubated/sedated Unknown Nature of brain injury Skull closed, contusion/ hemorrhage present Skull closed, no contusion/hemorrhage Skull open, contusion/ hemorrhage present Facial fracture Skull fracture Brain injury location Bilateral Left Right Midline shift (mm) >0e5 >5 Midline shift, mm not specified No midline shift Unknown Subdural hematoma Epidural hematoma Subarachnoid hemorrhage Intraventricular hemorrhage Brain stem involved Craniotomy during care episode Craniectomy during care episode Weight-bearing precaution during rehabilitation
EN (NZ451)
No EN (NZ1250) P
32.2
30.4
.513*
39.7
45.0
.053*
53.9
30.3
<.001*
8.6
8.9
.923*
21.3 69.6 1.6 5.1 2.4
36.8 50.0 2.0 7.3 3.9
<.001z
<.001z
8.4 5.8 44.6 14.9 26.4
18.2 8.2 23.8 11.0 38.8 .277z
67.6
70.4
23.9
23.3
8.4
6.3
18.2 31.3
12.5 24.6
65.9 17.3 16.9
61.8 19.0 19.2
9.8 13.1 11.8
13.6 11.2 10.2
22.6 42.8 43.0 10.0 66.3 24.6
34.3 30.6 47.5 7.4 53.2 14.0
.110* .086* <.001* <.001*
6.9 10.2
4.7 4.9
.086* <.001*
20.0
18.9
.626*
27.9
26.9
.666*
.003* .006* .300z
<.001z
(continued on next page)
S248
S.D. Horn et al by International Classification of Diseases, 9th Revision, Clinical Modification codes in the rehabilitation medical record.
Table 1 (continued ) Characteristic Days from injury to rehabilitation admission PTA clearance before rehabilitation admission Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Moderate-to-severe dysphagia on admission Moderate-to-severe aphasia on admission Functional status and LOS Admission FIM motor scoredraw Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Admission FIM cognitive scoredRasch-transformed Length of rehabilitation stay
EN (NZ451)
No EN (NZ1250) P
Data analyses
31.924.5 19.824.4 <.001y 17.3
47.8
61.820.1 3617.8
<.001* <.001y
22.616.1 15.213.1 <.001y
90.2
35.2
<.001*
49.9
43.7
.024*
21.511.3 38.616.4 <.001y 19.616.2 38.413.5 <.001y 10.55.4
16.76.6
<.001y
25.818.1 43.216.1 <.001y 32.215.0 18.310.0 <.001y
NOTE. Values are mean SD or as percentages. Abbreviations: GCS, Glasgow Coma Scale; MCO/HMO, managed care organization/health maintenance organization; PTA, posttraumatic amnesia. * Fisher exact test. y Two-sample t test. z c2 test.
injury component of the CSI) from severity of illness of all other injuries, complications, and comorbidities (called nonebrain injury component of the CSI). The brain injury component of the CSI allowed the detection of differences in the amount of brain damage in patients that might otherwise be hidden in an overall injury severity score. Treatment factors We collected treatment variables related to EN from a chart review. These included formula type and start and stop dates. EN was classified as either high protein or standard. High-protein formulas were defined by having at least 20% of calories coming from protein (most of these formulas actually had 25%), or standard formulas (usually having 14%e18% protein) with supplemental protein provided to increase the total protein amount to >20%. On the basis of the finding of better outcomes for EN support for 25% of the rehabilitation stay for patients with severe stroke,19 we explored the same level of EN support in this study of patients with TBI. Outcome variables Outcome variables examined to assess the association with EN were discharge FIM motor and cognitive scores, weight loss during rehabilitation, LOS, and various infections (aspiration pneumonia, pneumonia, urinary tract infections, sepsis) as defined
Analyses were performed using SAS version 9.2.a When data were missing, 1 adjustments were made, depending on the variable and its intended use in analyses. Sometimes we categorized values simply as “unknown” (and included an “unknown” category in the analysis as a dummy variable representing missingness); at other times we excluded patients with missing data from analyses. Because our sample had patients with a wide range of functional and cognitive disabilities and also had many patients who did not require EN, we needed to find a method to examine the effect of EN that was not biased by only the sickest patients or the most severely injured receiving EN. Our first step was to determine a subset of patients whom we predicted would benefit from EN, but may or may not have actually received it. To find this subset, we used the patient preinjury and injury characteristics data to develop a propensity score using a logistic regression model to predict receipt of EN.23 We then matched patients with the same propensity score, where one received EN and the other did not, and compared outcomes. Descriptive statistics were used to provide frequencies and percentages for categorical variables describing patients, treatments, and outcomes and means, medians, quartiles, and SDs to summarize continuous variables. Bivariate analyses were conducted to examine how patients who received EN differed from those who did not. For discrete variables, we used the chi-square test or McNemar test to determine the significance of associations. For continuous variables, we used t tests, paired t tests, analysis of variance, or Friedman rank sum test. A 2-sided P value of <.05 was considered statistically significant. Finally, we used ordinary least squares regression analyses to determine associations of EN with continuous outcome variables after controlling for other patient preinjury and injury characteristics. For independent variables with pairwise correlations r>.75, only one of the pair was allowed to enter the model. To help clinicians more easily decide which patients might benefit from EN, we also used recursive partitioning to build a classification model using a 2-stage procedure. First the single variable that best splits the data into 2 groups is found; the data are separated at this point, and then the process is repeated individually on the resultant subgroups until either they reach a minimum sample size (depending on the sample size of the data) or no improvement can be made in explaining an outcome. The original study sample included 2130 patients with TBI.21 We excluded patients who had TPN at any time during their rehabilitation stay (nZ127, 6.0%), because we did not collect details about TPN start and stop times; hence, we could not determine for how long they received TPN or any details about the content of TPN. We also excluded patients who had EN documented but no formula type was recorded or they had only fiber in their EN, which was not considered EN (nZ15, 0.7%). Next, patients receiving EN at the time of rehabilitation discharge and having a total CSI severity score at admission of >60 (indicating serious morbidity) were excluded (nZ107, 5.0%). In addition, we excluded patients with rehabilitation LOS >75 days (highest 1.5%; nZ31). The rationale for excluding these patients was that if they were discharged on EN and were also severely injured or stayed a long time in rehabilitation, it was unlikely that EN would have a significant effect on their outcomes. Patients with similar www.archives-pmr.org
Nutritional support for patients with TBI Table 2
S249
Propensity score models to predict patient’s need for EN Model 1
Intercept Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Moderate-to-severe aphasia on admission Moderate-to-severe dysphagia on admission Age on admission History of alcohol abuse before injury Skull fracture Days from injury to rehabilitation admission Cause of injury: miscellaneous Marital status: single Primary payer: Medicaid Primary payer: no-fault auto insurance Paralysis on admission Able to drive before injury Highest education achieved: high school diploma Race: white No. of observations used c statistic
Parameter Estimate
Odds Ratio Estimate
1.74 0.03 0.01
NA 1.03 1.01
Model 2 P (Wald c2)
Parameter Estimate
Odds Ratio Estimate NA NA NA
P (Wald c2)
11.27 44.31 4.71
.001 <.001 .030
0.36 NA NA
0.60 NA NA
.437 NA NA
0.04 0.96 60.62 NA NA NA 0.66 0.52 15.67 1.89 6.65 91.26 0.02 0.98 18.85 0.33 1.39 4.37 0.48 1.62 7.86 0.01 1.01 8.24 0.97 0.38 4.57 0.47 0.63 5.17 0.72 0.49 10.54 0.69 1.99 5.34 NA NA NA NA NA NA NA NA NA NA NA NA 1698 (yesZ449; noZ1249) .903
<.001 NA <.001 <.001 <.001 .037 .005 .004 .033 .023 .001 .021 NA NA NA NA
NA NA NA 0.13 0.88 87.79 0.46 0.63 8.83 2.36 10.60 157.33 0.02 0.98 25.18 NA NA NA NA NA NA 0.01 1.01 20.83 1.02 0.36 6.02 0.50 0.61 6.51 0.64 0.53 9.46 0.72 2.06 6.08 0.57 1.76 15.00 0.41 1.51 5.13 0.37 0.69 5.27 0.35 1.43 4.12 1698 (yesZ449; noZ1249) .883
NA <.001 .003 <.001 <.001 NA NA <.001 .014 .011 .002 .014 .000 .024 .022 .042
Abbreviation: NA, not applicable.
characteristics were excluded from the study of EN in the poststroke rehabilitation study.19 The 107 patients discharged on EN were significantly different (sicker) from those with EN during rehabilitation (nZ451, 21.2%) in the following ways: (1) older age (43.4y vs 38.2y; PZ.025); (2) longer time from injury to rehabilitation admission (55.2d vs 32.4d; P.001); (3) higher brain injury component of the CSI (78.6 vs 66.5; P<.001); (4) greater percent craniectomy (18.8% vs 9.9%; PZ.015); (5) lower admission Rasch-transformed FIM motor score (11.9 vs 19.5; P<.001); and (6) lower admission Rasch-transformed FIM cognitive score (18.4 vs 25.6; PZ.0002). Finally, we excluded patients from 1 site because of their practice of not using EN to treat their patients (nZ149, 7.0%).
Results The final study sample included 1701 (79.9%) patients (480 women and 1221 men). Of these patients, 451 received EN for >1 day and 1250 received either no EN or EN for only 1 day. Clinicians decided that 1 day of EN was too few to have any effect on outcomes and assigned these patients to the “no EN” category. As shown in table 1, patients with TBI who received EN were more likely to be white and less likely to be black, were more likely to be underweight or normal weight, have health maintenance organization or no-fault auto insurance, were driving before injury, were paralyzed, received their TBI in a motor vehicle collision, had a facial or skull fracture, had a subarachnoid hemorrhage, intraventricular hemorrhage, craniectomy, posttraumatic amnesia during rehabilitation, and more days from injury to rehabilitation admission. Patients receiving EN also had lower
www.archives-pmr.org
mean admission Rasch-transformed FIM motor and cognitive scores, and higher admission brain injury and nonebrain injury components of CSI scores. Of the 451 patients receiving EN, >90% had moderate-to-severe dysphagia and 50% had aphasia. The propensity score determined the combination of patient preinjury and injury characteristics that were significantly associated with use and duration of EN support. Significant variables are included in table 2. The strongest predictors of receiving EN included higher admission brain injury component of the CSI score, lower admission Rasch-transformed FIM motor score, and having moderate-to-severe dysphagia (c statisticZ.903). In table 2, we also present a second propensity score model that can be used if one does not have admission brain injury component of the CSI score, admission nonebrain injury component of the CSI score, or Rasch-transformed FIM motor score. The second model is almost as good as the first model that we used for our subsequent analyses and can be used to predict a patient’s probability of receiving EN when CSI and Rasch-transformed FIM measures are not available. If instead a recursive partitioning analysis was used to estimate a patient’s probability of receiving EN, our data found that a close approximation of a propensity score of >40% for the likely use of EN is an admission FIM motor score of 20, moderate-to-severe dysphagia, and >8 days from injury to rehabilitation admission. For each of the 1701 patients, we computed a propensity score for receiving EN. We decided to use a propensity score of >40% for receiving EN to indicate a need for EN. Of the 451 patients with EN, 335 (74.3%) had a propensity score of >40%; of the 1250 patients without EN, only 145 (11.6%) had a propensity score of >40%. The mean propensity score for the 335 patients with EN was 73.2%, whereas the mean propensity score for the 145 patients without EN
S250 Table 3
S.D. Horn et al Characteristics of propensity scoreematched patients with and without EN
Characteristic Demographic characteristics Sex: male (%) Age at rehabilitation admission (y) Race/ethnicity Black White White Hispanic Other and unknown Admission body mass index (kg/m2) Admission body mass index (kg/m2) <16 16e18.5 >18.5e25 >25e30 >30e35 >35e40 >40 Unknown Brain injury cause and severity Cause of injury Fall Motor vehicle collision Sports Violence Other Days from injury to rehabilitation admission PTA clearance before rehabilitation admission Admission brain injury component of the CSI score Maximum brain injury component of the CSI score Admission nonebrain injury component of the CSI score Maximum nonebrain injury component of the CSI score Moderate-to-severe dysphagia on admission Functional independence measures Admission FIM motor scoredraw Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Admission FIM cognitive scoredRasch-transformed Nutritional information Admission Braden total score Discharge Braden total score Lowest Braden nutrition Very poor Probably inadequate Adequate Excellent Unknown Lowest serum albumin during rehabilitation Lowest serum albumin category Very low (1.0e3.0g/dL) Low (>3.0e3.49g/dL) Normal (3.5g/dL) Missing Lowest serum transthyretin during rehabilitation Lowest serum transthyretin category Elevated (>29.6mg/dL) Normal (>15.5e29.6mg/dL)
EN (NZ145)
No EN (NZ145)
76.6 38.518.2
80.7 39.519.4
13.1 83.4 2.1 1.4 23.04.5
14.5 75.9 8.3 1.4 23.34.7
2.1 12.4 56.6 18.6 8.3 0.7 0.0 1.4
2.8 9.0 55.2 21.4 8.3 1.4 0.7 1.4
P .474* .646y .111z
.507y .921z
.658z 24.8 64.1 1.4 5.5 4.1 31.118.0 15.2 61.313.5 64.413.6 22.516.6 32.722.8 95.2
26.2 66.9 1.4 4.1 1.4 31.534.9 13.8 59.512.8 63.113.1 21.614.5 30.518.8 95.9
.896y .868* .246y .405y .635y .369y 1.000*
20.99.1 21.014.3 9.94.4 25.315.8
21.010.3 20.614.7 10.64.6 27.515.9
.900y .824y .192y .232y
14.82.4 18.72.3
14.92.5 18.42.5
.751y .470y .721z
2.1 28.3 26.2 0.7 42.8 3.20.6
4.8 28.3 25.5 1.4 40.0 3.20.6
37.2 21.4 31.7 9.7 25.26.4
36.6 20.0 29.0 14.5 24.47.6
10.3 32.4
8.3 24.8
.968y .646z
.535y .386z
(continued on next page)
www.archives-pmr.org
Nutritional support for patients with TBI
S251
Table 3 (continued ) Characteristic Mild malnutrition (10e15.5mg/dL) Missing
EN (NZ145)
No EN (NZ145)
2.8 54.5
4.1 62.8
P
NOTE. Values are mean SD or as percentages. Abbreviation: PTA, posttraumatic amnesia. * Fisher exact test. y Two-sample t test. z c2 test.
was only 60.5% (P<.001). Hence, we needed to create a matched and balanced sample before examining the association of EN with outcomes; we selected 1 patient with EN for each of the 145 patients without EN, matching on propensity scores. This resulted in mean propensity scores of 61% and 60.5% for the 145 patients with EN and 145 patients without EN, respectively. Table 3 compares the preinjury and injury characteristics of the 145 patients with EN and the 145 patients without EN. None of the characteristics were significantly different between the 2 groups irrespective of whether using paired or unpaired tests. EN duration ranged from 2 to 64 days (6%e100% of the rehabilitation stay), with a mean of 16.612.6 days and a median of 13 days. The mean percentage of the stay with EN was 52.4%26.9% and a median of 50.0%. Of the 145 patients with EN, 80 (55.0%) had a nothing-by-mouth order during the EN episodes that ranged from 2 to 33 days (5.6%e100.0% of the rehabilitation stay), with a mean of 5.17.4 days and a median of 2 days. The percentage of the stay with a nothing-by-mouth order was 18.7%24.0% with a median of 8.3%. Table 4 shows bivariate associations of each outcome with receipt of EN. Patients receiving EN had borderline higher discharge Rasch-transformed FIM motor and cognitive scores (PZ.055 and PZ.050, respectively), borderline longer LOS (PZ.062), and less weight change (PZ.075), but no significant differences in urinary tract infections, sepsis, pneumonia, or aspiration pneumonia. None of these bivariate analyses took into account how long the patient received EN or the type of formula used in EN. Hence, we conducted multiple regression analyses to account for duration and type of EN and additional potentially confounding outcome variables.
Table 4
Table 5 presents the results of ordinary least squares regression analyses to predict discharge FIM motor and cognitive scores, weight change during rehabilitation, and LOS. We found that EN with either a standard or a high-protein formula for 25% of the rehabilitation stay was associated with significantly higher discharge FIM subscale scores. The range of days for which patients in the propensity scoreematched sample received EN for 25% of their stay was 3 to 64 days, with a median of 15 days and a mean of 19 days. This is the level of EN support that was associated with better outcomes. This was not related to the site where the patient was treated, because no site variables entered the models significantly after patient and treatment variables entered. Also, EN with a high-protein formula for 25% of the stay was associated with an almost 2-lb weight gain from admission to discharge as compared with an almost 2-lb weight loss in patients with no EN. However, EN with either formula for 25% of the stay was not significant in predicting LOS or infections.
Discussion This study of the potential benefit of EN for patients with TBI requiring inpatient rehabilitation found that EN with either a highprotein or a standard formula for 25% of the rehabilitation stay was associated with better FIM motor and cognitive scores at discharge. Patients who received a high-protein formula had less weight loss and higher discharge FIM motor scores than those who received a standard formula. There are a number of possible explanations for these results. Other studies1,24 have found that patients with TBI can have an
Outcomes: Complications during rehabilitation and discharge FIM motor and cognitive scores
Characteristic
EN (NZ145)
No EN (NZ145)
P
Change in weight during rehabilitation Discharge FIM motor scoredraw Discharge FIM motor scoredRasch-transformed Discharge FIM cognitive scoredraw Discharge FIM cognitive scoredRasch-transformed LOS Pneumonitis/aspiration pneumonia during rehabilitation Pneumonitis/aspiration pneumonia before or during rehabilitation Pneumonia during rehabilitation Sepsis during rehabilitation Urinary tract infection during rehabilitation
0.59.0 60.913.4 53.19.2 21.05.9 52.512.6 31.214.0 0.7 5.5 24.1 2.1 7.6
2.812.1 57.116.8 50.910.7 19.65.5 49.712 28.412.2 1.4 5.5 22.1 4.8 8.3
.075* .032* .055* .047* .050* .062* 1.000y 1.000y .781y .335y 1.000y
NOTE. Values are mean SD or as percentages. * Two-sample t test. y Fisher exact test.
www.archives-pmr.org
S252
Table 5
Results of ordinary least squares regression analyses predicting discharge FIM motor and cognitive scores and weight loss during stay Discharge FIM Motor Scored Rasch-Transformed
Outcome Intercept Admission FIM motor scored Rasch-transformed Admission FIM cognitive scored Rasch-transformed Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Admission body mass index 18.5kg/m2 Admission body mass index >30kg/m2 Admission weight EN for 25% of the rehabilitation stay High-protein EN for 25% of the rehabilitation stay No. of observations used R2 Adjusted R2
Parameter Estimate
Standardized Parameter Estimate
P
44.9 0.28
.00 .40
<.001 <.001
Discharge FIM Cognitive Scored Rasch-Transformed Parameter Estimate
Standardized Parameter Estimate
P
47.1
.000
<.001
.350
<.001
0.27 0.12 0.10
2.59 5.65
.12 .16 290 .189 .181
.028 .003
4.93 6.37
.132
.026
.122
.020
.186 .150 290 .240 .227
.001 .005
Discharge Weight Parameter Estimate
Standardized Parameter Estimate
10.96 0.10
.00 .04
LOS
P
Parameter Estimate
Standardized Parameter Estimate
P
<.001 .012
28.22 0.34
.00 .38
<.001 <.001
5.75 0.90
.05 .93
.014 <.001
3.70
.03
.057
263 .935 .934
0.13
.13
.016
7.78
.20
<.001
290 .224 .216
NOTE. Independent variables allowed to enter in models include admission FIM motor and cognitive scores, admission brain injury and nonebrain injury components of the CSI score, admission body mass index categories of 8.5, 18.5e25, >25e30, and >30kg/m2, tube feeding for 25% of the rehabilitation stay, and high-protein tube feeding for 25% of the rehabilitation stay.
S.D. Horn et al
www.archives-pmr.org
Nutritional support for patients with TBI increased metabolic resting rate after their injury. An imaging study25 focusing on biopathology of TBI reported a brief period of hypermetabolic activity in the brain immediately after TBI, followed by several days of glucose metabolic depression. The period of metabolic depression typically lasts only a few days, but patients with TBI may require a higher metabolic expenditure both locally in the brain and more systemically. This increased metabolic need may persist for some time after the injury and could be positively affected by nutritional interventions such as EN. This could help to explain why our study found a greater improvement in FIM cognitive scores at discharge than in FIM motor scores, which was associated with EN and especially so with a high-protein formula. A challenge in studying the association of EN with outcomes after TBI rehabilitation is that patients who may receive EN are generally at the more severe end of the spectrum owing to altered awareness and arousal or severe dysphagia. Hence, when one includes an indicator variable for receipt of EN in analyses and does not have EN patients spread throughout the sample (a balanced sample), one finds that EN is associated with poorer outcomes. These poorer outcomes are not due to EN but due to the fact that it is only the more severely impaired patients who receive EN. Therefore, one needs to examine EN and its association with outcomes in a sample of patients with TBI, where patients receiving EN are balanced and spread more evenly throughout the sample. There are various ways of accomplishing this in observational studies. We used patient preinjury and injury characteristics to develop a propensity score for receiving EN and then matched patients with similar propensity scores, wherein one patient received EN and the other did not. For our original sample of 1701 patients, we were able to find only 145 patients without EN who had a propensity score of >40% for the likely use of EN, which was the probability that the project team felt indicated a minimal need for EN. This may be an indication that in our study centers, patients who needed EN usually received it. There were only 145 patients with this indication of needing EN who did not receive it, and it is with this sample and a matched sample with EN that we were able to assess the potential associations of EN with outcomes for patients with TBI requiring inpatient rehabilitation. A close approximation, in our data, of a propensity score of >40% for receiving EN is an admission FIM motor score of 20, moderate-to-severe dysphagia, and >8 days from injury to rehabilitation admission. The role that EN, and particularly protein, plays during rehabilitation of patients with TBI is rarely mentioned in the literature. In a recent observational study of patients with TBI treated at rehabilitation centers in Italy,26 the incidence of dysphagia on admission to the rehabilitation center was 42.3%, which decreased to 13.7% on discharge. Only 45.8% of patients with TBI admitted to rehabilitation were on oral feeding, which increased to 84.2% by discharge; however, no information on nutritional support during rehabilitation was provided.26 Similarly, a multicenter report27 on rehabilitation of patients with complicated mild TBI provided insight into the effect of admission FIM and Disability Rating Scale scores on LOS, but nutritional status or support required or used by these patients during rehabilitation was not mentioned. The present study indicates that EN, and particularly protein, in the rehabilitation setting may play a significant role in recovery of patients with TBI: receiving EN for at least 25% of the rehabilitation stay was associated with higher discharge FIM subscale scores for both motor skills and cognitive function. Supporting the hypothesis that protein needs continue to be high in patients with TBI during rehabilitation, the present study www.archives-pmr.org
S253 demonstrated that the strongest association with outcomes was of high-protein EN. Although this study did not assess nitrogen balance, clinical outcomes of improved discharge FIM subscale scores for both motor skills and cognitive function as well as less weight loss were greatest in patients receiving high-protein EN. These findings suggest that the protein catabolic rate may continue to increase during rehabilitation for TBI. Therefore, research assessing protein requirements in the rehabilitation setting or extended recovery period and not just the acute phase of recovery is greatly needed. For example, a randomized trial of a dose-ranging study of EN protein levels comparing standard protein (15% of calories) to high protein (20% of calories), high protein plus essential amino acids, low protein plus essential amino acids, and possible combinations with fish oil and other immune-modulating components within the formula would help to clarify whether higher protein would be beneficial in this population. Other aspects to study were as follows: metabolic needs during phases of TBI, best management practices for meeting nutritional intake goals, and a more detailed look at daily caloric and protein intake in those with diets and/or EN. Clinical studies should follow. As this study points to the possible benefits of EN, it is also important to note the potential risks of administering EN to a population with TBI. EN is invasive and generally not desired by patients. In addition, the agitation and confusion that often characterize the behavior of the individual with TBI receiving inpatient rehabilitation may make it difficult to administer or prevent EN. Restraints or a 1:1 sitter observation may be required to maintain tubes. The removal of a gastrostomy tube by the patient may place the patient at risk for serious infection. Duodenal tubes introduced nasally require an abdominal radiographic examination to confirm the placement. When a patient pulls the tube out or it is removed accidentally, rehabilitation therapies may be interrupted to allow time for the radiograph. The use of a nasal bridle system may be appropriate at times after TBI. The nasal bridal system has been used in patients with burns with some success, requiring fewer tube insertions and without sinusitis or aspiration pneumonia.28 The use of restraints and just the presence of a tube (gastric or nasal) itself may make patients more agitated, potentially prompting the use of additional medications (with risk of side effects of medications), and carry important risks that may outweigh the benefits of EN. Pneumonia and aspiration pneumonia are often cited as complications of EN.29,30 However, the present study suggests attenuated risks. For example, our study found no differences in rates of aspiration pneumonia or pneumonia between patients with EN and patients with similar characteristics who did not receive EN. This finding is in contrast to previous findings in acute care settings.7,11,29,30 In addition, our study found improved outcomes when EN was provided for at least 25% of the stay compared to patients with similar characteristics who did not receive EN, strongly suggesting a benefit that could be described to patients, staff, and family to support an EN decision.
Study limitations Some information that is relevant to this research was not available for our patient sample, including nutritional supplementation during acute care hospitalization, reason(s) for administration of EN, and total caloric intake during inpatient rehabilitation. Acute care records were not fully available, and thus, we do not know if EN was used there, how many days after injury it was started, what EN formula was used, and duration of EN before rehabilitation admission. We
S254 also did not have accurate calorie counts of oral intake to determine the total calories and protein a patient received during either the acute care or the rehabilitation hospital stay.
Conclusions For patients in rehabilitation for moderate-to-severe TBI and having a likelihood of using EN based on a propensity score of >40%, clinicians should strongly consider delivering EN for at least 25% of the rehabilitation stay. These findings also suggest a role of high-protein EN in the rehabilitation setting.
Supplier a. SAS Institute, Inc.
Keywords Brain injuries; Comparative effectiveness research; Enteral nutrition; Propensity score; Rehabilitation
Corresponding author Susan D. Horn, PhD, University of Utah School of Medicine, Department of Population Health Sciences, Health System Innovation and Research Program Williams Bldg, Rm 1N491, 295 Chipeta Way, Salt Lake City, Utah 84108. E-mail address: susan. [email protected].
Acknowledgments We gratefully acknowledge the contributions of clinical and research staff at each of the 10 inpatient rehabilitation facilities to designing, defining variables, collecting data, and helping with analyses and manuscripts for the TBIePractice Based Evidence study. The following study center directors were included: John D. Corrigan, PhD, and Jennifer Bogner, PhD (Department of Physical Medicine and Rehabilitation, Ohio State University, Columbus, OH); Nora Cullen, MD (Toronto Rehabilitation Institute, Toronto, Ontario, Canada); Cynthia L. Beaulieu, PhD (Brooks Rehabilitation Hospital, Jacksonville, FL); Flora M. Hammond, MD (Carolinas Rehabilitation, Charlotte, NC [now at Indiana University]); David K. Ryser, MD (Neuro Specialty Rehabilitation Unit, Intermountain Medical Center, Salt Lake City, UT); Murray E. Brandstater, MD (Loma Linda University Medical Center, Loma Linda, CA); Marcel P. Dijkers, PhD (Mount Sinai Medical Center, New York, NY); William Garmoe, PhD (Medstar National Rehabilitation Hospital, Washington, DC); James A. Young, MD (Physical Medicine and Rehabilitation, Rush University Medical Center, Chicago, IL); Ronald T. Seel, PhD (Brain Injury Research, Shepherd Center, Atlanta, GA).
References 1. Rapp RP, Young B, Twyman D, et al. The favorable effect of early parenteral feeding on survival in head-injured patients. J Neurosurg 1983;58:906-12.
S.D. Horn et al 2. Taylor SJ, Fettes SB, Jewkes C, Nelson RJ. Prospective, randomized, controlled trial to determine the effect of early enhanced enteral nutrition on clinical outcome in mechanically ventilated patients suffering head injury. Crit Care Med 1999;27:2525-31. 3. Doig GS, Heighes PT, Simpson F, Sweetman EA. Early enteral nutrition reduces mortality in trauma patients requiring intensive care: a metaanalysis of randomised controlled trials. Injury 2011;42: 50-6. 4. Pepe JL, Barba CA. The metabolic response to acute traumatic brain injury and implications for nutritional support. J Head Trauma Rehabil 1999;14:462-74. 5. Bistrian BR, Askew W, Erdman JW Jr, Oria MP. Nutrition and traumatic brain injury: a perspective from the Institute of Medicine. JPEN J Parenter Enteral Nutr 2011;35:556-9. 6. Dickerson RN, Pitts SL, Maish GO III, et al. A reappraisal of nitrogen requirements for patients with critical illness and trauma. J Trauma Acute Care Surg 2012;73:549-57. 7. Chourdakis M, Kraus MM, Tzellos T, et al. Effect of early compared with delayed enteral nutrition on endocrine function in patients with traumatic brain injury: an open-labeled randomized trial. J Parenter Enteral Nutr 2012;36:108-16. 8. Krakau K, Omne-Ponte´n M, Karlsson T, Borg J. Metabolism and nutrition in patients with moderate and severe traumatic brain injury: a systematic review. Brain Injury 2006;20:345-67. 9. Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons. Guidelines for the management of severe traumatic brain injury. XII. Nutrition. J Neurotrauma 2007;24(Suppl 1):S77-82. 10. Young B, Ott L, Twyman D, et al. The effect of nutritional support on outcomes from severe head injury. J Neurosurgery 1987;67:668-76. 11. Wang X, Dong Y, Han X, Qi XQ, Huang CG, Hou LJ. Nutritional support for patients sustaining traumatic brain injury: a systemic review and meta-analysis of prospective studies. PLoS One 2013;8:1-14. 12. Clifton GL, Robertson CS, Grossman RG, Hodge S, Foltz R, Garza C. The metabolic response to severe head injury. J Neurosurg 1984;60: 687-96. 13. Vizzini A, Aranda-Michel J. Nutritional support in head injury. Nutrition 2011;27:129-32. 14. Mackay LE, Morgan AS, Bernstein BA. Swallowing disorders in severe brain injury: risk factors affecting return to oral intake. Arch Phys Med Rehabil 1999;80:365-71. 15. Tan M, Zhu JC, Yin HH. Enteral nutrition in patients with severe traumatic brain injury: reasons for tolerance and medical management. Br J Neurosurg 2011;25:2-8. 16. Haddad SH, Arabi YM. Critical care management of severe traumatic brain injury in adults. Scand J Trauma Resusc Emerg Med 2012;20:12. 17. Justo Meirelles CM, de Aguilar-Nascimento JE. Enteral or parenteral nutrition in traumatic brain injury: a prospective randomised trial. Nutr Hosp 2011;26:1120-4. 18. Dickerson RN, Mitchell JN, Morgan LM, et al. Disparate response to metoclopramide therapy for gastric feeding intolerance in trauma patients with and without traumatic brain injury. J Parenter Enteral Nutr 2009;33:646-55. 19. James R, Gines D, Menlove A, Horn SD, Gassaway J, Smout R. Nutrition support (tube feeding) as a rehabilitation intervention. Arch Phys Med Rehabil 2005;86(12 Suppl 2):S82-92. 20. Horn SD, DeJong G, Smout R, Gassaway J, James R, Conroy B. Stroke rehabilitation patients, practice, and outcomes: is earlier and more aggressive therapy better? Arch Phys Med Rehabil 2005; 86(12 Suppl 2):S101-14. 21. Horn SD, Corrigan JD, Bogner J, et al. Traumatic Brain Injurye Practice Based Evidence study: design and patients, centers, treatments, and outcomes. Arch Phys Med Rehabil 2015;96(8 Suppl 3): S178-96. 22. Heinemann AW, Linacre JM, Wright BD, Hamilton BB, Granger C. Relationships between impairment and physical disability as measured by the Functional Independence Measure. Arch Phys Med Rehabil 1993;74:566-73.
www.archives-pmr.org
Nutritional support for patients with TBI 23. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399-424. 24. McEvoy CT, Cran GW, Cooke SR, Young IS. Resting energy expenditure in non-ventilated, non-sedated patients recovering from severe traumatic brain injury: comparison of prediction equations with indirect calorimetry values. Clin Nutr 2009;28:526-32. 25. Prins ML, Alexander D, Giza CC, Hovda DA. Repeated mild traumatic brain injury: mechanisms of cerebral vulnerability. J Neurotrauma 2013;30:30-8. 26. Zampolini M, Franceschini M. Rehabilitation of traumatic brain injury in Italy. Am J Phys Med Rehab 2011;90:79-82.
www.archives-pmr.org
S255 27. Kennedy RE, Livingston L, Marwitz JH, Gueck S, Kreutzer JS, Sander AM. Complicated mild traumatic brain injury on the inpatient rehabilitation unit: a multicenter analysis. J Head Trauma Rehabil 2006;21:260-71. 28. Parks J, Klaus S, Staggs V, Pena M. Outcomes of nasal bridling to secure enteral tubes in burn patients. Am J Crit Care 2013;22:136-42. 29. Hansen TS, Larsen K, Engberg AW. The association of functional oral intake and pneumonia in patients with severe traumatic brain injury. Arch Phys Med Rehabil 2008;89:2114-20. 30. Acosta-Escribano J, Ferna`ndez-Vivas M, Carmona TG, et al. Gastric versus transpyloric feeding in severe traumatic brain injury: a prospective, randomized trial. Intensive Care Med 2010;36:1532-9.
ORIGINAL RESEARCH
Enteral Nutrition for Patients With Traumatic Brain Injury in the Rehabilitation Setting: Associations With Patient Preinjury and Injury Characteristics and Outcomes Susan D. Horn, PhD,a Merin Kinikini, RD, FNP, DNP, CNSC,b Linda W. Moore, MS, RDN, CCRP,c Flora M. Hammond, MD,d,e Murray E. Brandstater, MD,f Randall J. Smout, MS,a Ryan S. Barrett, MSa From the aInstitute for Clinical Outcomes Research, International Severity Information Systems, Inc, Salt Lake City, UT; bNeuro Specialty Rehabilitation Unit, Intermountain Medical Center, Salt Lake City, UT; cHouston Methodist Hospital, Houston, TX; dCarolinas Rehabilitation, Charlotte, NC; eIndiana University, Indianapolis, IN; and fLoma Linda University Medical Center, Loma Linda, CA. Current affiliation for Horn, Department of Population Health Sciences, University of Utah School of Medicine, Salt Lake City, UT.
Abstract Objective: To determine the association of enteral nutrition (EN) with patient preinjury and injury characteristics and outcomes for patients receiving inpatient rehabilitation after traumatic brain injury (TBI). Design: Prospective observational study. Setting: Nine rehabilitation centers. Participants: Patients (NZ1701) admitted for first full inpatient rehabilitation after TBI. Interventions: Not applicable. Main Outcome Measures: FIM at rehabilitation discharge, length of stay, weight loss, and various infections. Results: There were many significant differences in preinjury and injury characteristics between patients who received EN and patients who did not. After matching patients with a propensity score of >40% for the likely use of EN, patients receiving EN with either a standard or a highprotein formula (>20% of calories coming from protein) for >25% of their rehabilitation stay had higher FIM motor and cognitive scores at rehabilitation discharge and less weight loss than did patients with similar characteristics not receiving EN. Conclusions: For patients receiving inpatient rehabilitation after TBI and matched on a propensity score of >40% for the likely use of EN, clinicians should strongly consider, when possible, EN for 25% of the rehabilitation stay and especially with a formula that contains at least 20% protein rather than a standard formula. Archives of Physical Medicine and Rehabilitation 2015;96(8 Suppl 3):S245-55 ª 2015 by the American Congress of Rehabilitation Medicine
The provision of adequate nutritional support for patients with moderate-to-severe traumatic brain injury (TBI) has been a clinical challenge for decades.1-3 Patients’ primary and secondary injuries create unique metabolic derangements that pose issues such as optimal timing and route of nutrition, appropriate fluid and Supported by the National Institutes of Health, National Center for Medical Rehabilitation Research (grant no. 1R01HD050439-01), the National Institute on Disability and Rehabilitation Research (grant no. H133A080023), and the Ontario Neurotrauma Foundation (grant no. 2007-ABI-ISIS-525). Publication of this article was supported by the American Congress of Rehabilitation Medicine. Disclosure: none.
electrolyte balance, drug administration, and dysphagia. In addition, it may be difficult to maintain tubes and lines in a confused or agitated patient, particularly in a rehabilitation setting. Individuals with TBI have a much higher resting metabolic expenditure (RME) than do patients without TBI.4 In fact, with severe TBI, RME has been found to range up to 240% of RME in patients without TBI; they are similar in metabolic response to patients with burns over 20% to 40% of their body surface area.4 The consequences of hypermetabolism, hypercatabolism, and altered immune function in patients with acute TBI are excessive
0003-9993/15/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.06.024
S246
S.D. Horn et al
protein breakdown, which can lead to malnutrition.5 However, patients with TBI requiring hospitalization often do not, or cannot, consume enough nutrition to support their increased requirements for recovery and rehabilitation.5 Enteral nutrition (EN) administered as early as possible has been established as the preferential route of nutritional support for this population versus total parenteral nutrition (TPN); some centers use a combination of EN and TPN in the early stages of injury if the patient does not tolerate adequate amounts of EN alone.6 Chourdakis et al7 recently reported that early EN may affect hormonal response to TBI and suggests that this may reduce catabolic and inflammatory processes induced by TBI. There appears to be a consensus on early initiation of EN, but less definitive are recommendations on advancement timing and formula components (eg, whether to use specialty formulas such as those containing immune-enhancing properties).8-12 The Institute of Medicine recommended inclusion of nutrient additives (eg, n-3 fatty acids, creatine, choline, and zinc) as potentially beneficial for recovery after TBI.3,5 Patients with TBI, similar to other trauma patients, likely require 2.0 to 2.5g of protein/kg at a minimum, especially during the early period after injury.6,13 Evaluation of the duration of a higher protein requirement has not been reported in the literature, but it likely correlates with metabolic status. If increased metabolic rates extend into the rehabilitation setting, then increased protein needs may also be present. Swallowing disorders and decreased behavioral/cognitive skills are frequently present in patients with severe brain injury and significantly affect oral intake.14 Persons who swallow abnormally take much longer to start eating and to achieve total oral feeding, and they require nonoral supplementation 3 to 4 times longer than those who swallow normally.14 Patients with severe TBI may also have intolerance to EN, which hampers survival and rehabilitation.15 Haddad and Arabi16 discuss proactive use of prokinetic agents, such as erythromycin and metoclopramide, as well as postpyloric feeding as ways to overcome problems of gastric distention and intolerance experienced by patients with TBI. Most reports11,13,17,18 regarding nutrition in patients with TBI address the route (TPN vs EN) and/or timing (early vs late) of initiation of nutritional support related to hospital admission and have addressed outcomes such as mortality or length of stay (LOS) in the acute care setting. We could not find any published reports that address the role of nutritional support during rehabilitation of patients with TBI. A practice-based evidence (PBE) study in stroke rehabilitation found that the use of EN support for 25% of the rehabilitation stay for patients with severe stroke was a significant factor in predicting higher discharge FIM total and motor scores, controlling for patient and other treatment differences.19,20 It is not known if these findings are applicable to the population with TBI receiving rehabilitation. This article describes nutritional support methods used for patients in a TBIPBE study during rehabilitation21 and examines associations of patient preinjury and injury characteristics with use and duration
List of abbreviations: CSI EN LOS PBE RME TBI TPN
Comprehensive Severity Index enteral nutrition length of stay practice-based evidence resting metabolic expenditure traumatic brain injury total parenteral nutrition
of EN support, as well as associations of EN with outcomes, controlling for patient differences.
Methods This comparative effectiveness PBE study examined the differential effects of a wide array of specific treatments administered in 10 acute inpatient rehabilitation facilities serving patients with TBI in a brain injury specialty unit who were enrolled from October 2008 to September 2011. The 10 participating centers constituted a convenience sample of adults with TBI on the basis of their willingness to participate in the research. The institutional review board at each center approved the study; each patient or his/her proxy gave informed consent.
Participants Inclusion criteria were as follows21: (1) Sustained a TBI, defined as damage to brain tissue caused by external force and evidenced by loss of consciousness, posttraumatic amnesia, skull fracture, or objective neurological findings. Diagnoses included International Classification of Diseases, 9th Revision, Clinical Modification codes consistent with the Centers for Disease Control and Prevention Guidelines for Surveillance of Central Nervous System Injury; (2) received inpatient care in a designated brain injury rehabilitation unit of one of the participating rehabilitation facilities; and (3) were 14 years or older and treated in an adult rehabilitation unit. Patient variables Patient characteristics, including demographic characteristics and injury characteristics, were recorded on the basis of clinicians’ suggestions as well as previous research indicating their importance in populations with TBI. In addition to patient data available on admission, we collected information on patients’ status during their rehabilitation stay, including the presence of aphasia and dysphagia. Table 1 lists the characteristics of study patients. Functional dependence We used admission FIM, an assessment of independent functioning consisting of 18 items in 2 domains: motor (13 items) and cognitive (5 items). Each FIM item was rated on a 7-category scale, ranging from 1 (total assistance required) to 7 (complete independence). To eliminate distortion in quantifying the status of patients whose capability is at the extremes of the instrument’s range, FIM motor and cognitive scores were recoded separately using tables published by Heinemann et al,22 which were based on Rasch analysis of FIM data of a large sample with brain injury. Injury severity and comorbidity The primary medical severity measure used was the Comprehensive Severity Index (CSI), which defines severity as the physiological and psychosocial complexity presented owing to the extent and interactions of a patient’s disease(s).21 The CSI is age- and disease-specific and is independent of treatments. It provides an objective, consistent method to operationalize patient severity of illness on the basis of >2100 individual signs, symptoms, and physical findings and >5600 disease-specific criteria sets related to all of a patient’s disease(s). More details about the CSI appear elsewhere.21 The CSI modification used here allowed separating severity of brain injury (called brain www.archives-pmr.org
Nutritional support for patients with TBI Table 1
S247
Characteristics of patients with and without EN
Characteristic Demographic characteristics Sex: male (%) Age at rehabilitation admission (y) Race/ethnicity Black White White Hispanic Other and unknown Highest education achieved Some high school, no diploma High school diploma Work toward or completed associate’s degree Work toward or completed bachelor’s degree Work toward or completed master’s/doctoral degree Unknown Marital status at injury Single/never married Married/common law Previously married Other/unknown Able to drive before injury No Yes Unknown Primary payer Medicare Medicaid Private insurance Worker’s compensation Self-pay/none MCO/HMO No-fault auto insurance Other/unknown Admission body mass index (kg/m2) <16 16e18.5 >18.5e25 >25e30 >30e35 >35e40 >40 Unknown Preexisting and coexisting conditions History of alcohol abuse before injury History of drug abuse before injury Anxiety before or during rehabilitation
EN (NZ451)
Table 1 (continued )
No EN (NZ1250) P
Characteristic
72.3 71.6 .807* 38.519.6 47.122.0 <.001y .012z 12.0 80.0 4.9 3.1
17.0 72.7 7.4 2.9 .268z
24.2
22.6
24.2 18.0
26.6 15.6
20.8
19.5
9.5
9.8
3.3
5.8 .114z
47.0 34.6 15.7 2.7
41.2 35.8 19.8 3.1 <.001z
5.5 81.4 13.1
12.6 68.8 18.6 <.001z
11.5 12.2 26.8 9.1 6.9 19.5 8.9 5.1
25.8 17.0 25.3 6.8 4.6 14.0 3.4 3.1 <.001z
1.3 12.4 54.3 20.4 6.4 1.6 1.3 2.2
1.4 6.3 47.0 26.2 9.0 2.9 1.7 5.5
34.8
36.9
.458*
20.0
22.8
.233*
22.0
21.1
.736*
(continued page) (continued in on nextnext column)
www.archives-pmr.org
Depression before or during rehabilitation Hypertension before or during rehabilitation Paralysis before or during rehabilitation Renal failure before or during rehabilitation Brain injury and severity information Cause of injury Fall Motor vehicle collision Sports Violence Other GCS score immediately after injury or after arrival in acute care (%) Mild (13e15) Moderate (9e12) Severe (3e8) Intubated/sedated Unknown Nature of brain injury Skull closed, contusion/ hemorrhage present Skull closed, no contusion/hemorrhage Skull open, contusion/ hemorrhage present Facial fracture Skull fracture Brain injury location Bilateral Left Right Midline shift (mm) >0e5 >5 Midline shift, mm not specified No midline shift Unknown Subdural hematoma Epidural hematoma Subarachnoid hemorrhage Intraventricular hemorrhage Brain stem involved Craniotomy during care episode Craniectomy during care episode Weight-bearing precaution during rehabilitation
EN (NZ451)
No EN (NZ1250) P
32.2
30.4
.513*
39.7
45.0
.053*
53.9
30.3
<.001*
8.6
8.9
.923*
21.3 69.6 1.6 5.1 2.4
36.8 50.0 2.0 7.3 3.9
<.001z
<.001z
8.4 5.8 44.6 14.9 26.4
18.2 8.2 23.8 11.0 38.8 .277z
67.6
70.4
23.9
23.3
8.4
6.3
18.2 31.3
12.5 24.6
65.9 17.3 16.9
61.8 19.0 19.2
9.8 13.1 11.8
13.6 11.2 10.2
22.6 42.8 43.0 10.0 66.3 24.6
34.3 30.6 47.5 7.4 53.2 14.0
.110* .086* <.001* <.001*
6.9 10.2
4.7 4.9
.086* <.001*
20.0
18.9
.626*
27.9
26.9
.666*
.003* .006* .300z
<.001z
(continued on next page)
S248
S.D. Horn et al by International Classification of Diseases, 9th Revision, Clinical Modification codes in the rehabilitation medical record.
Table 1 (continued ) Characteristic Days from injury to rehabilitation admission PTA clearance before rehabilitation admission Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Moderate-to-severe dysphagia on admission Moderate-to-severe aphasia on admission Functional status and LOS Admission FIM motor scoredraw Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Admission FIM cognitive scoredRasch-transformed Length of rehabilitation stay
EN (NZ451)
No EN (NZ1250) P
Data analyses
31.924.5 19.824.4 <.001y 17.3
47.8
61.820.1 3617.8
<.001* <.001y
22.616.1 15.213.1 <.001y
90.2
35.2
<.001*
49.9
43.7
.024*
21.511.3 38.616.4 <.001y 19.616.2 38.413.5 <.001y 10.55.4
16.76.6
<.001y
25.818.1 43.216.1 <.001y 32.215.0 18.310.0 <.001y
NOTE. Values are mean SD or as percentages. Abbreviations: GCS, Glasgow Coma Scale; MCO/HMO, managed care organization/health maintenance organization; PTA, posttraumatic amnesia. * Fisher exact test. y Two-sample t test. z c2 test.
injury component of the CSI) from severity of illness of all other injuries, complications, and comorbidities (called nonebrain injury component of the CSI). The brain injury component of the CSI allowed the detection of differences in the amount of brain damage in patients that might otherwise be hidden in an overall injury severity score. Treatment factors We collected treatment variables related to EN from a chart review. These included formula type and start and stop dates. EN was classified as either high protein or standard. High-protein formulas were defined by having at least 20% of calories coming from protein (most of these formulas actually had 25%), or standard formulas (usually having 14%e18% protein) with supplemental protein provided to increase the total protein amount to >20%. On the basis of the finding of better outcomes for EN support for 25% of the rehabilitation stay for patients with severe stroke,19 we explored the same level of EN support in this study of patients with TBI. Outcome variables Outcome variables examined to assess the association with EN were discharge FIM motor and cognitive scores, weight loss during rehabilitation, LOS, and various infections (aspiration pneumonia, pneumonia, urinary tract infections, sepsis) as defined
Analyses were performed using SAS version 9.2.a When data were missing, 1 adjustments were made, depending on the variable and its intended use in analyses. Sometimes we categorized values simply as “unknown” (and included an “unknown” category in the analysis as a dummy variable representing missingness); at other times we excluded patients with missing data from analyses. Because our sample had patients with a wide range of functional and cognitive disabilities and also had many patients who did not require EN, we needed to find a method to examine the effect of EN that was not biased by only the sickest patients or the most severely injured receiving EN. Our first step was to determine a subset of patients whom we predicted would benefit from EN, but may or may not have actually received it. To find this subset, we used the patient preinjury and injury characteristics data to develop a propensity score using a logistic regression model to predict receipt of EN.23 We then matched patients with the same propensity score, where one received EN and the other did not, and compared outcomes. Descriptive statistics were used to provide frequencies and percentages for categorical variables describing patients, treatments, and outcomes and means, medians, quartiles, and SDs to summarize continuous variables. Bivariate analyses were conducted to examine how patients who received EN differed from those who did not. For discrete variables, we used the chi-square test or McNemar test to determine the significance of associations. For continuous variables, we used t tests, paired t tests, analysis of variance, or Friedman rank sum test. A 2-sided P value of <.05 was considered statistically significant. Finally, we used ordinary least squares regression analyses to determine associations of EN with continuous outcome variables after controlling for other patient preinjury and injury characteristics. For independent variables with pairwise correlations r>.75, only one of the pair was allowed to enter the model. To help clinicians more easily decide which patients might benefit from EN, we also used recursive partitioning to build a classification model using a 2-stage procedure. First the single variable that best splits the data into 2 groups is found; the data are separated at this point, and then the process is repeated individually on the resultant subgroups until either they reach a minimum sample size (depending on the sample size of the data) or no improvement can be made in explaining an outcome. The original study sample included 2130 patients with TBI.21 We excluded patients who had TPN at any time during their rehabilitation stay (nZ127, 6.0%), because we did not collect details about TPN start and stop times; hence, we could not determine for how long they received TPN or any details about the content of TPN. We also excluded patients who had EN documented but no formula type was recorded or they had only fiber in their EN, which was not considered EN (nZ15, 0.7%). Next, patients receiving EN at the time of rehabilitation discharge and having a total CSI severity score at admission of >60 (indicating serious morbidity) were excluded (nZ107, 5.0%). In addition, we excluded patients with rehabilitation LOS >75 days (highest 1.5%; nZ31). The rationale for excluding these patients was that if they were discharged on EN and were also severely injured or stayed a long time in rehabilitation, it was unlikely that EN would have a significant effect on their outcomes. Patients with similar www.archives-pmr.org
Nutritional support for patients with TBI Table 2
S249
Propensity score models to predict patient’s need for EN Model 1
Intercept Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Moderate-to-severe aphasia on admission Moderate-to-severe dysphagia on admission Age on admission History of alcohol abuse before injury Skull fracture Days from injury to rehabilitation admission Cause of injury: miscellaneous Marital status: single Primary payer: Medicaid Primary payer: no-fault auto insurance Paralysis on admission Able to drive before injury Highest education achieved: high school diploma Race: white No. of observations used c statistic
Parameter Estimate
Odds Ratio Estimate
1.74 0.03 0.01
NA 1.03 1.01
Model 2 P (Wald c2)
Parameter Estimate
Odds Ratio Estimate NA NA NA
P (Wald c2)
11.27 44.31 4.71
.001 <.001 .030
0.36 NA NA
0.60 NA NA
.437 NA NA
0.04 0.96 60.62 NA NA NA 0.66 0.52 15.67 1.89 6.65 91.26 0.02 0.98 18.85 0.33 1.39 4.37 0.48 1.62 7.86 0.01 1.01 8.24 0.97 0.38 4.57 0.47 0.63 5.17 0.72 0.49 10.54 0.69 1.99 5.34 NA NA NA NA NA NA NA NA NA NA NA NA 1698 (yesZ449; noZ1249) .903
<.001 NA <.001 <.001 <.001 .037 .005 .004 .033 .023 .001 .021 NA NA NA NA
NA NA NA 0.13 0.88 87.79 0.46 0.63 8.83 2.36 10.60 157.33 0.02 0.98 25.18 NA NA NA NA NA NA 0.01 1.01 20.83 1.02 0.36 6.02 0.50 0.61 6.51 0.64 0.53 9.46 0.72 2.06 6.08 0.57 1.76 15.00 0.41 1.51 5.13 0.37 0.69 5.27 0.35 1.43 4.12 1698 (yesZ449; noZ1249) .883
NA <.001 .003 <.001 <.001 NA NA <.001 .014 .011 .002 .014 .000 .024 .022 .042
Abbreviation: NA, not applicable.
characteristics were excluded from the study of EN in the poststroke rehabilitation study.19 The 107 patients discharged on EN were significantly different (sicker) from those with EN during rehabilitation (nZ451, 21.2%) in the following ways: (1) older age (43.4y vs 38.2y; PZ.025); (2) longer time from injury to rehabilitation admission (55.2d vs 32.4d; P.001); (3) higher brain injury component of the CSI (78.6 vs 66.5; P<.001); (4) greater percent craniectomy (18.8% vs 9.9%; PZ.015); (5) lower admission Rasch-transformed FIM motor score (11.9 vs 19.5; P<.001); and (6) lower admission Rasch-transformed FIM cognitive score (18.4 vs 25.6; PZ.0002). Finally, we excluded patients from 1 site because of their practice of not using EN to treat their patients (nZ149, 7.0%).
Results The final study sample included 1701 (79.9%) patients (480 women and 1221 men). Of these patients, 451 received EN for >1 day and 1250 received either no EN or EN for only 1 day. Clinicians decided that 1 day of EN was too few to have any effect on outcomes and assigned these patients to the “no EN” category. As shown in table 1, patients with TBI who received EN were more likely to be white and less likely to be black, were more likely to be underweight or normal weight, have health maintenance organization or no-fault auto insurance, were driving before injury, were paralyzed, received their TBI in a motor vehicle collision, had a facial or skull fracture, had a subarachnoid hemorrhage, intraventricular hemorrhage, craniectomy, posttraumatic amnesia during rehabilitation, and more days from injury to rehabilitation admission. Patients receiving EN also had lower
www.archives-pmr.org
mean admission Rasch-transformed FIM motor and cognitive scores, and higher admission brain injury and nonebrain injury components of CSI scores. Of the 451 patients receiving EN, >90% had moderate-to-severe dysphagia and 50% had aphasia. The propensity score determined the combination of patient preinjury and injury characteristics that were significantly associated with use and duration of EN support. Significant variables are included in table 2. The strongest predictors of receiving EN included higher admission brain injury component of the CSI score, lower admission Rasch-transformed FIM motor score, and having moderate-to-severe dysphagia (c statisticZ.903). In table 2, we also present a second propensity score model that can be used if one does not have admission brain injury component of the CSI score, admission nonebrain injury component of the CSI score, or Rasch-transformed FIM motor score. The second model is almost as good as the first model that we used for our subsequent analyses and can be used to predict a patient’s probability of receiving EN when CSI and Rasch-transformed FIM measures are not available. If instead a recursive partitioning analysis was used to estimate a patient’s probability of receiving EN, our data found that a close approximation of a propensity score of >40% for the likely use of EN is an admission FIM motor score of 20, moderate-to-severe dysphagia, and >8 days from injury to rehabilitation admission. For each of the 1701 patients, we computed a propensity score for receiving EN. We decided to use a propensity score of >40% for receiving EN to indicate a need for EN. Of the 451 patients with EN, 335 (74.3%) had a propensity score of >40%; of the 1250 patients without EN, only 145 (11.6%) had a propensity score of >40%. The mean propensity score for the 335 patients with EN was 73.2%, whereas the mean propensity score for the 145 patients without EN
S250 Table 3
S.D. Horn et al Characteristics of propensity scoreematched patients with and without EN
Characteristic Demographic characteristics Sex: male (%) Age at rehabilitation admission (y) Race/ethnicity Black White White Hispanic Other and unknown Admission body mass index (kg/m2) Admission body mass index (kg/m2) <16 16e18.5 >18.5e25 >25e30 >30e35 >35e40 >40 Unknown Brain injury cause and severity Cause of injury Fall Motor vehicle collision Sports Violence Other Days from injury to rehabilitation admission PTA clearance before rehabilitation admission Admission brain injury component of the CSI score Maximum brain injury component of the CSI score Admission nonebrain injury component of the CSI score Maximum nonebrain injury component of the CSI score Moderate-to-severe dysphagia on admission Functional independence measures Admission FIM motor scoredraw Admission FIM motor scoredRasch-transformed Admission FIM cognitive scoredraw Admission FIM cognitive scoredRasch-transformed Nutritional information Admission Braden total score Discharge Braden total score Lowest Braden nutrition Very poor Probably inadequate Adequate Excellent Unknown Lowest serum albumin during rehabilitation Lowest serum albumin category Very low (1.0e3.0g/dL) Low (>3.0e3.49g/dL) Normal (3.5g/dL) Missing Lowest serum transthyretin during rehabilitation Lowest serum transthyretin category Elevated (>29.6mg/dL) Normal (>15.5e29.6mg/dL)
EN (NZ145)
No EN (NZ145)
76.6 38.518.2
80.7 39.519.4
13.1 83.4 2.1 1.4 23.04.5
14.5 75.9 8.3 1.4 23.34.7
2.1 12.4 56.6 18.6 8.3 0.7 0.0 1.4
2.8 9.0 55.2 21.4 8.3 1.4 0.7 1.4
P .474* .646y .111z
.507y .921z
.658z 24.8 64.1 1.4 5.5 4.1 31.118.0 15.2 61.313.5 64.413.6 22.516.6 32.722.8 95.2
26.2 66.9 1.4 4.1 1.4 31.534.9 13.8 59.512.8 63.113.1 21.614.5 30.518.8 95.9
.896y .868* .246y .405y .635y .369y 1.000*
20.99.1 21.014.3 9.94.4 25.315.8
21.010.3 20.614.7 10.64.6 27.515.9
.900y .824y .192y .232y
14.82.4 18.72.3
14.92.5 18.42.5
.751y .470y .721z
2.1 28.3 26.2 0.7 42.8 3.20.6
4.8 28.3 25.5 1.4 40.0 3.20.6
37.2 21.4 31.7 9.7 25.26.4
36.6 20.0 29.0 14.5 24.47.6
10.3 32.4
8.3 24.8
.968y .646z
.535y .386z
(continued on next page)
www.archives-pmr.org
Nutritional support for patients with TBI
S251
Table 3 (continued ) Characteristic Mild malnutrition (10e15.5mg/dL) Missing
EN (NZ145)
No EN (NZ145)
2.8 54.5
4.1 62.8
P
NOTE. Values are mean SD or as percentages. Abbreviation: PTA, posttraumatic amnesia. * Fisher exact test. y Two-sample t test. z c2 test.
was only 60.5% (P<.001). Hence, we needed to create a matched and balanced sample before examining the association of EN with outcomes; we selected 1 patient with EN for each of the 145 patients without EN, matching on propensity scores. This resulted in mean propensity scores of 61% and 60.5% for the 145 patients with EN and 145 patients without EN, respectively. Table 3 compares the preinjury and injury characteristics of the 145 patients with EN and the 145 patients without EN. None of the characteristics were significantly different between the 2 groups irrespective of whether using paired or unpaired tests. EN duration ranged from 2 to 64 days (6%e100% of the rehabilitation stay), with a mean of 16.612.6 days and a median of 13 days. The mean percentage of the stay with EN was 52.4%26.9% and a median of 50.0%. Of the 145 patients with EN, 80 (55.0%) had a nothing-by-mouth order during the EN episodes that ranged from 2 to 33 days (5.6%e100.0% of the rehabilitation stay), with a mean of 5.17.4 days and a median of 2 days. The percentage of the stay with a nothing-by-mouth order was 18.7%24.0% with a median of 8.3%. Table 4 shows bivariate associations of each outcome with receipt of EN. Patients receiving EN had borderline higher discharge Rasch-transformed FIM motor and cognitive scores (PZ.055 and PZ.050, respectively), borderline longer LOS (PZ.062), and less weight change (PZ.075), but no significant differences in urinary tract infections, sepsis, pneumonia, or aspiration pneumonia. None of these bivariate analyses took into account how long the patient received EN or the type of formula used in EN. Hence, we conducted multiple regression analyses to account for duration and type of EN and additional potentially confounding outcome variables.
Table 4
Table 5 presents the results of ordinary least squares regression analyses to predict discharge FIM motor and cognitive scores, weight change during rehabilitation, and LOS. We found that EN with either a standard or a high-protein formula for 25% of the rehabilitation stay was associated with significantly higher discharge FIM subscale scores. The range of days for which patients in the propensity scoreematched sample received EN for 25% of their stay was 3 to 64 days, with a median of 15 days and a mean of 19 days. This is the level of EN support that was associated with better outcomes. This was not related to the site where the patient was treated, because no site variables entered the models significantly after patient and treatment variables entered. Also, EN with a high-protein formula for 25% of the stay was associated with an almost 2-lb weight gain from admission to discharge as compared with an almost 2-lb weight loss in patients with no EN. However, EN with either formula for 25% of the stay was not significant in predicting LOS or infections.
Discussion This study of the potential benefit of EN for patients with TBI requiring inpatient rehabilitation found that EN with either a highprotein or a standard formula for 25% of the rehabilitation stay was associated with better FIM motor and cognitive scores at discharge. Patients who received a high-protein formula had less weight loss and higher discharge FIM motor scores than those who received a standard formula. There are a number of possible explanations for these results. Other studies1,24 have found that patients with TBI can have an
Outcomes: Complications during rehabilitation and discharge FIM motor and cognitive scores
Characteristic
EN (NZ145)
No EN (NZ145)
P
Change in weight during rehabilitation Discharge FIM motor scoredraw Discharge FIM motor scoredRasch-transformed Discharge FIM cognitive scoredraw Discharge FIM cognitive scoredRasch-transformed LOS Pneumonitis/aspiration pneumonia during rehabilitation Pneumonitis/aspiration pneumonia before or during rehabilitation Pneumonia during rehabilitation Sepsis during rehabilitation Urinary tract infection during rehabilitation
0.59.0 60.913.4 53.19.2 21.05.9 52.512.6 31.214.0 0.7 5.5 24.1 2.1 7.6
2.812.1 57.116.8 50.910.7 19.65.5 49.712 28.412.2 1.4 5.5 22.1 4.8 8.3
.075* .032* .055* .047* .050* .062* 1.000y 1.000y .781y .335y 1.000y
NOTE. Values are mean SD or as percentages. * Two-sample t test. y Fisher exact test.
www.archives-pmr.org
S252
Table 5
Results of ordinary least squares regression analyses predicting discharge FIM motor and cognitive scores and weight loss during stay Discharge FIM Motor Scored Rasch-Transformed
Outcome Intercept Admission FIM motor scored Rasch-transformed Admission FIM cognitive scored Rasch-transformed Admission brain injury component of the CSI score Admission nonebrain injury component of the CSI score Admission body mass index 18.5kg/m2 Admission body mass index >30kg/m2 Admission weight EN for 25% of the rehabilitation stay High-protein EN for 25% of the rehabilitation stay No. of observations used R2 Adjusted R2
Parameter Estimate
Standardized Parameter Estimate
P
44.9 0.28
.00 .40
<.001 <.001
Discharge FIM Cognitive Scored Rasch-Transformed Parameter Estimate
Standardized Parameter Estimate
P
47.1
.000
<.001
.350
<.001
0.27 0.12 0.10
2.59 5.65
.12 .16 290 .189 .181
.028 .003
4.93 6.37
.132
.026
.122
.020
.186 .150 290 .240 .227
.001 .005
Discharge Weight Parameter Estimate
Standardized Parameter Estimate
10.96 0.10
.00 .04
LOS
P
Parameter Estimate
Standardized Parameter Estimate
P
<.001 .012
28.22 0.34
.00 .38
<.001 <.001
5.75 0.90
.05 .93
.014 <.001
3.70
.03
.057
263 .935 .934
0.13
.13
.016
7.78
.20
<.001
290 .224 .216
NOTE. Independent variables allowed to enter in models include admission FIM motor and cognitive scores, admission brain injury and nonebrain injury components of the CSI score, admission body mass index categories of 8.5, 18.5e25, >25e30, and >30kg/m2, tube feeding for 25% of the rehabilitation stay, and high-protein tube feeding for 25% of the rehabilitation stay.
S.D. Horn et al
www.archives-pmr.org
Nutritional support for patients with TBI increased metabolic resting rate after their injury. An imaging study25 focusing on biopathology of TBI reported a brief period of hypermetabolic activity in the brain immediately after TBI, followed by several days of glucose metabolic depression. The period of metabolic depression typically lasts only a few days, but patients with TBI may require a higher metabolic expenditure both locally in the brain and more systemically. This increased metabolic need may persist for some time after the injury and could be positively affected by nutritional interventions such as EN. This could help to explain why our study found a greater improvement in FIM cognitive scores at discharge than in FIM motor scores, which was associated with EN and especially so with a high-protein formula. A challenge in studying the association of EN with outcomes after TBI rehabilitation is that patients who may receive EN are generally at the more severe end of the spectrum owing to altered awareness and arousal or severe dysphagia. Hence, when one includes an indicator variable for receipt of EN in analyses and does not have EN patients spread throughout the sample (a balanced sample), one finds that EN is associated with poorer outcomes. These poorer outcomes are not due to EN but due to the fact that it is only the more severely impaired patients who receive EN. Therefore, one needs to examine EN and its association with outcomes in a sample of patients with TBI, where patients receiving EN are balanced and spread more evenly throughout the sample. There are various ways of accomplishing this in observational studies. We used patient preinjury and injury characteristics to develop a propensity score for receiving EN and then matched patients with similar propensity scores, wherein one patient received EN and the other did not. For our original sample of 1701 patients, we were able to find only 145 patients without EN who had a propensity score of >40% for the likely use of EN, which was the probability that the project team felt indicated a minimal need for EN. This may be an indication that in our study centers, patients who needed EN usually received it. There were only 145 patients with this indication of needing EN who did not receive it, and it is with this sample and a matched sample with EN that we were able to assess the potential associations of EN with outcomes for patients with TBI requiring inpatient rehabilitation. A close approximation, in our data, of a propensity score of >40% for receiving EN is an admission FIM motor score of 20, moderate-to-severe dysphagia, and >8 days from injury to rehabilitation admission. The role that EN, and particularly protein, plays during rehabilitation of patients with TBI is rarely mentioned in the literature. In a recent observational study of patients with TBI treated at rehabilitation centers in Italy,26 the incidence of dysphagia on admission to the rehabilitation center was 42.3%, which decreased to 13.7% on discharge. Only 45.8% of patients with TBI admitted to rehabilitation were on oral feeding, which increased to 84.2% by discharge; however, no information on nutritional support during rehabilitation was provided.26 Similarly, a multicenter report27 on rehabilitation of patients with complicated mild TBI provided insight into the effect of admission FIM and Disability Rating Scale scores on LOS, but nutritional status or support required or used by these patients during rehabilitation was not mentioned. The present study indicates that EN, and particularly protein, in the rehabilitation setting may play a significant role in recovery of patients with TBI: receiving EN for at least 25% of the rehabilitation stay was associated with higher discharge FIM subscale scores for both motor skills and cognitive function. Supporting the hypothesis that protein needs continue to be high in patients with TBI during rehabilitation, the present study www.archives-pmr.org
S253 demonstrated that the strongest association with outcomes was of high-protein EN. Although this study did not assess nitrogen balance, clinical outcomes of improved discharge FIM subscale scores for both motor skills and cognitive function as well as less weight loss were greatest in patients receiving high-protein EN. These findings suggest that the protein catabolic rate may continue to increase during rehabilitation for TBI. Therefore, research assessing protein requirements in the rehabilitation setting or extended recovery period and not just the acute phase of recovery is greatly needed. For example, a randomized trial of a dose-ranging study of EN protein levels comparing standard protein (15% of calories) to high protein (20% of calories), high protein plus essential amino acids, low protein plus essential amino acids, and possible combinations with fish oil and other immune-modulating components within the formula would help to clarify whether higher protein would be beneficial in this population. Other aspects to study were as follows: metabolic needs during phases of TBI, best management practices for meeting nutritional intake goals, and a more detailed look at daily caloric and protein intake in those with diets and/or EN. Clinical studies should follow. As this study points to the possible benefits of EN, it is also important to note the potential risks of administering EN to a population with TBI. EN is invasive and generally not desired by patients. In addition, the agitation and confusion that often characterize the behavior of the individual with TBI receiving inpatient rehabilitation may make it difficult to administer or prevent EN. Restraints or a 1:1 sitter observation may be required to maintain tubes. The removal of a gastrostomy tube by the patient may place the patient at risk for serious infection. Duodenal tubes introduced nasally require an abdominal radiographic examination to confirm the placement. When a patient pulls the tube out or it is removed accidentally, rehabilitation therapies may be interrupted to allow time for the radiograph. The use of a nasal bridle system may be appropriate at times after TBI. The nasal bridal system has been used in patients with burns with some success, requiring fewer tube insertions and without sinusitis or aspiration pneumonia.28 The use of restraints and just the presence of a tube (gastric or nasal) itself may make patients more agitated, potentially prompting the use of additional medications (with risk of side effects of medications), and carry important risks that may outweigh the benefits of EN. Pneumonia and aspiration pneumonia are often cited as complications of EN.29,30 However, the present study suggests attenuated risks. For example, our study found no differences in rates of aspiration pneumonia or pneumonia between patients with EN and patients with similar characteristics who did not receive EN. This finding is in contrast to previous findings in acute care settings.7,11,29,30 In addition, our study found improved outcomes when EN was provided for at least 25% of the stay compared to patients with similar characteristics who did not receive EN, strongly suggesting a benefit that could be described to patients, staff, and family to support an EN decision.
Study limitations Some information that is relevant to this research was not available for our patient sample, including nutritional supplementation during acute care hospitalization, reason(s) for administration of EN, and total caloric intake during inpatient rehabilitation. Acute care records were not fully available, and thus, we do not know if EN was used there, how many days after injury it was started, what EN formula was used, and duration of EN before rehabilitation admission. We
S254 also did not have accurate calorie counts of oral intake to determine the total calories and protein a patient received during either the acute care or the rehabilitation hospital stay.
Conclusions For patients in rehabilitation for moderate-to-severe TBI and having a likelihood of using EN based on a propensity score of >40%, clinicians should strongly consider delivering EN for at least 25% of the rehabilitation stay. These findings also suggest a role of high-protein EN in the rehabilitation setting.
Supplier a. SAS Institute, Inc.
Keywords Brain injuries; Comparative effectiveness research; Enteral nutrition; Propensity score; Rehabilitation
Corresponding author Susan D. Horn, PhD, University of Utah School of Medicine, Department of Population Health Sciences, Health System Innovation and Research Program Williams Bldg, Rm 1N491, 295 Chipeta Way, Salt Lake City, Utah 84108. E-mail address: susan. [email protected].
Acknowledgments We gratefully acknowledge the contributions of clinical and research staff at each of the 10 inpatient rehabilitation facilities to designing, defining variables, collecting data, and helping with analyses and manuscripts for the TBIePractice Based Evidence study. The following study center directors were included: John D. Corrigan, PhD, and Jennifer Bogner, PhD (Department of Physical Medicine and Rehabilitation, Ohio State University, Columbus, OH); Nora Cullen, MD (Toronto Rehabilitation Institute, Toronto, Ontario, Canada); Cynthia L. Beaulieu, PhD (Brooks Rehabilitation Hospital, Jacksonville, FL); Flora M. Hammond, MD (Carolinas Rehabilitation, Charlotte, NC [now at Indiana University]); David K. Ryser, MD (Neuro Specialty Rehabilitation Unit, Intermountain Medical Center, Salt Lake City, UT); Murray E. Brandstater, MD (Loma Linda University Medical Center, Loma Linda, CA); Marcel P. Dijkers, PhD (Mount Sinai Medical Center, New York, NY); William Garmoe, PhD (Medstar National Rehabilitation Hospital, Washington, DC); James A. Young, MD (Physical Medicine and Rehabilitation, Rush University Medical Center, Chicago, IL); Ronald T. Seel, PhD (Brain Injury Research, Shepherd Center, Atlanta, GA).
References 1. Rapp RP, Young B, Twyman D, et al. The favorable effect of early parenteral feeding on survival in head-injured patients. J Neurosurg 1983;58:906-12.
S.D. Horn et al 2. Taylor SJ, Fettes SB, Jewkes C, Nelson RJ. Prospective, randomized, controlled trial to determine the effect of early enhanced enteral nutrition on clinical outcome in mechanically ventilated patients suffering head injury. Crit Care Med 1999;27:2525-31. 3. Doig GS, Heighes PT, Simpson F, Sweetman EA. Early enteral nutrition reduces mortality in trauma patients requiring intensive care: a metaanalysis of randomised controlled trials. Injury 2011;42: 50-6. 4. Pepe JL, Barba CA. The metabolic response to acute traumatic brain injury and implications for nutritional support. J Head Trauma Rehabil 1999;14:462-74. 5. Bistrian BR, Askew W, Erdman JW Jr, Oria MP. Nutrition and traumatic brain injury: a perspective from the Institute of Medicine. JPEN J Parenter Enteral Nutr 2011;35:556-9. 6. Dickerson RN, Pitts SL, Maish GO III, et al. A reappraisal of nitrogen requirements for patients with critical illness and trauma. J Trauma Acute Care Surg 2012;73:549-57. 7. Chourdakis M, Kraus MM, Tzellos T, et al. Effect of early compared with delayed enteral nutrition on endocrine function in patients with traumatic brain injury: an open-labeled randomized trial. J Parenter Enteral Nutr 2012;36:108-16. 8. Krakau K, Omne-Ponte´n M, Karlsson T, Borg J. Metabolism and nutrition in patients with moderate and severe traumatic brain injury: a systematic review. Brain Injury 2006;20:345-67. 9. Brain Trauma Foundation, American Association of Neurological Surgeons, Congress of Neurological Surgeons. Guidelines for the management of severe traumatic brain injury. XII. Nutrition. J Neurotrauma 2007;24(Suppl 1):S77-82. 10. Young B, Ott L, Twyman D, et al. The effect of nutritional support on outcomes from severe head injury. J Neurosurgery 1987;67:668-76. 11. Wang X, Dong Y, Han X, Qi XQ, Huang CG, Hou LJ. Nutritional support for patients sustaining traumatic brain injury: a systemic review and meta-analysis of prospective studies. PLoS One 2013;8:1-14. 12. Clifton GL, Robertson CS, Grossman RG, Hodge S, Foltz R, Garza C. The metabolic response to severe head injury. J Neurosurg 1984;60: 687-96. 13. Vizzini A, Aranda-Michel J. Nutritional support in head injury. Nutrition 2011;27:129-32. 14. Mackay LE, Morgan AS, Bernstein BA. Swallowing disorders in severe brain injury: risk factors affecting return to oral intake. Arch Phys Med Rehabil 1999;80:365-71. 15. Tan M, Zhu JC, Yin HH. Enteral nutrition in patients with severe traumatic brain injury: reasons for tolerance and medical management. Br J Neurosurg 2011;25:2-8. 16. Haddad SH, Arabi YM. Critical care management of severe traumatic brain injury in adults. Scand J Trauma Resusc Emerg Med 2012;20:12. 17. Justo Meirelles CM, de Aguilar-Nascimento JE. Enteral or parenteral nutrition in traumatic brain injury: a prospective randomised trial. Nutr Hosp 2011;26:1120-4. 18. Dickerson RN, Mitchell JN, Morgan LM, et al. Disparate response to metoclopramide therapy for gastric feeding intolerance in trauma patients with and without traumatic brain injury. J Parenter Enteral Nutr 2009;33:646-55. 19. James R, Gines D, Menlove A, Horn SD, Gassaway J, Smout R. Nutrition support (tube feeding) as a rehabilitation intervention. Arch Phys Med Rehabil 2005;86(12 Suppl 2):S82-92. 20. Horn SD, DeJong G, Smout R, Gassaway J, James R, Conroy B. Stroke rehabilitation patients, practice, and outcomes: is earlier and more aggressive therapy better? Arch Phys Med Rehabil 2005; 86(12 Suppl 2):S101-14. 21. Horn SD, Corrigan JD, Bogner J, et al. Traumatic Brain Injurye Practice Based Evidence study: design and patients, centers, treatments, and outcomes. Arch Phys Med Rehabil 2015;96(8 Suppl 3): S178-96. 22. Heinemann AW, Linacre JM, Wright BD, Hamilton BB, Granger C. Relationships between impairment and physical disability as measured by the Functional Independence Measure. Arch Phys Med Rehabil 1993;74:566-73.
www.archives-pmr.org
Nutritional support for patients with TBI 23. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399-424. 24. McEvoy CT, Cran GW, Cooke SR, Young IS. Resting energy expenditure in non-ventilated, non-sedated patients recovering from severe traumatic brain injury: comparison of prediction equations with indirect calorimetry values. Clin Nutr 2009;28:526-32. 25. Prins ML, Alexander D, Giza CC, Hovda DA. Repeated mild traumatic brain injury: mechanisms of cerebral vulnerability. J Neurotrauma 2013;30:30-8. 26. Zampolini M, Franceschini M. Rehabilitation of traumatic brain injury in Italy. Am J Phys Med Rehab 2011;90:79-82.
www.archives-pmr.org
S255 27. Kennedy RE, Livingston L, Marwitz JH, Gueck S, Kreutzer JS, Sander AM. Complicated mild traumatic brain injury on the inpatient rehabilitation unit: a multicenter analysis. J Head Trauma Rehabil 2006;21:260-71. 28. Parks J, Klaus S, Staggs V, Pena M. Outcomes of nasal bridling to secure enteral tubes in burn patients. Am J Crit Care 2013;22:136-42. 29. Hansen TS, Larsen K, Engberg AW. The association of functional oral intake and pneumonia in patients with severe traumatic brain injury. Arch Phys Med Rehabil 2008;89:2114-20. 30. Acosta-Escribano J, Ferna`ndez-Vivas M, Carmona TG, et al. Gastric versus transpyloric feeding in severe traumatic brain injury: a prospective, randomized trial. Intensive Care Med 2010;36:1532-9.