Evidence Tables

Chesnut RM, Carney N, Maynard H, et al. Rehabilitation for Traumatic Brain Injury. Rockville (MD): Agency for Health Care Policy and Research (US); 1999 Feb. (Evidence Reports/Technology Assessments, No. 2.) Evidence Tables.

Evidence table 1. Study - MacKay, 1992


Design	Patients	Setting	Sampling	Grouping	TBI severity measure	Confounding variables	Method of handling confounding variables	Definition of acute formalized rehabilitation

Retrospective	38 severe TBI patients (admission GCS 3-8)	Single inpatient rehabilitation facility	Consecutively discharged from inpatient rehabilitation between 1984 - 1990	Formalized, physiatrist driven multidisciplinary rehabilitation program at the acute care facility (N=17) vs "non formalized" acute care rehabilitation at 10 hospitals (N=21)	GCS, ISS, RLA on admission to rehabilitation, pupillary & pain responses, CT scan	TBI severity, associated injuries (ISS), surgical interventions. Admission GCS and RLA on admission to rehabilitation included in ANOVA analysis.	t-test	Evaluation and treatment on admission to the acute hospital by a physiatrist and specialists in PT, OT, and ST. Rehabilitative and preventive intervention involved structured multisensory stimulation, orientation, exercise, and positioning.

Independent variables	Dependent variables	Statistical methods
Formalized rehabilitation at acute care hospital vs "non formalized" rehabilitation (much less regular time of onset and intensity therapy [14% received only physical therapy, 65% received no speech therapy, and 14% received no rehabilitation at all])	Duration of coma, rehabilitation LOS, specific outcome parameters in terms of physical/motor, sensory/perceptual, cognitive/language outcome.	ANCOVA

Results	Analysis	Caveats
The formalized treatment group had approximately 1/3 the duration of coma and rehabilitation stays. Length of coma, rehabilitation length of stay, total length of stay, and RLA at acute and rehab discharge were significantly associated with type of acute rehabilitation when controlling for GCS using ANCOVA analysis. When controlling for admission RLA using ANCOVA analysis, only RLA at discharge from the acute facility remains significantly different between groups. Physical/motor, sensory/perceptual, cognitive/language outcome parameters were better for the formalized group.	Formalized multidisciplinary physiatrist-driven rehabilitation should be instituted as soon as possible following trauma center admission.	Lack of multivariate statistical handling of potential confounding variables. This particularly weakens the data for recovery of physical/motor, sensory/perceptual, cognitive/language outcome parameters where there was no control for confounding variables. Multivariate regression analysis would be a more powerful method than ANCOVA in analyzing the relative predictive power of type of acute rehabilitation program vs other confounding factors. Lack of long term followup

Evidence table 2. Study 1 - Aronow, 1987


Design	Patients	Setting	Sampling	Grouping	TBI severity measure	Confounding variables	Method of handling confounding variables	Definition of acute, formalized rehabilitation


Retrospective	Rehabilitation group N=68; Non-rehabilitation (control) group N=61. Selection = TBI (@1 hour of unconsciousness and @24 hours of altered consciousness), age at injury 5-80, acute hospital LOS @15 days, not comatose at acute hospital discharge.	Rehabilitation group=specialized inpatient brain injury rehabilitation program Non-rehabilitation (control) group=general neurosurgical unit at large teaching hospital in area of country with no comprehensive rehabilitation available for severe TBI	Rehabilitation group - 107 consecutive rehabilitation discharges. Non-rehabilitation (control) group - 1400 cases consecutively entered into an autonomous epidemiologic TBI database	Rehabilitation group inpatient, physiatrist driven comprehensive, multidisciplinary rehabilitation program Non-rehabilitation group - no comprehensive rehabilitation program (many received some informal, ancillary rehabilitation therapies)	PTA, acute hospital LOS, open brain injury, number of skull fractures	Age, sex, race, years post injury	Entered into regression analysis prior to independent variables; Chi square for PTA differences between groups	Inpatient, physiatrist driven comprehensive, multidisciplinary rehabilitation program


Independent variables	Dependent variables	Statistical methods

Rehabilitation vs. no rehabilitation	13 variable standardized measure (vocational status, living arrangement, number of recent inpatient treatment episodes, number of recent outpatient episodes, hours of daytime care required, functional status in self-care, mobility, and residential skills, number of home and outside social contacts, and number of physical cognitive, and emotional symptoms	Multivariate linear regression


Results	Analysis	Caveats

Rehabilitation was positively associated with outcome (total model R2=0.551, r=0.159 When patients were grouped according to severity as indexed by post traumatic amnesia, there was a trend toward better outcome for rehabilitation patients of comparable severity in terms of cost When average dollar costs were examined for patients with grouped post-traumatic amnesias there appeared to be an overall net savings favoring the rehabilitation group for a given level of post traumatic amnesia	When injury severity is considered, however, rehabilitation appeared to provide markedly better gains and the "unit cost" of these gains was less than for patients who received no rehabilitation	Injury severity indexed by PTA differed significantly between groups (70% of the rehabilitation group had PTAs@4 months while the non-rehabilitation group had PTAs @1 month in 74%) Little actual data presented in paper GCS was not used to index injury severity (due to missing data points) Timing of long term outcome (telephone interview) not specified The long-term outcome measurement tool was a unique, non-validated scale Poor correlation coefficient (0.159) for overall prediction model.

Evidence table 2. Study 2 - Heinemann, 1995


Design	Patients	Setting	Sampling	Grouping	TBI severity measure	Confounding variables	Method of handing confounding variables	Definition of acute, formalized rehabilitation

Prospectively collected data set (UDSMR)	140 patients with TBI	Eight UDSMR rehabilitation hospitals.	All patients admitted between 1990 1991.	None (observational study)	None	Age, admission FIM, rehabilitation LOS, interval between TBI and admission	Multivariate regression analysis	Average daily intensity=number of hours billed divided by LOS


Independent variables	Dependent variables	Statistical methods

Average daily intensity for PT, OT, ST, and psychological services Average daily intensity for all services combined	Discharge motor and cognitive FIM scores Achievement of motor or cognitive potential ([D/C FIM - admit FIM]/[100 admit FIM) Efficiency of change ([D/C FIM - admit FIM]/In[LOS])	Linear regression with forced entry of predictor variables (confirmed by backwards elimination linear regression analysis)


Results	Analysis	Caveats

Treatment intensity is not independent of functional status at admission, demographics, or medical characteristics Intensity measured as hours per day of PT, OT, or ST are not related to outcome as measured by the FIM at discharge from inpatient rehabilitation Intensity of psychological services (generally delivered as cognitive therapy) is positively related to cognitive FIM recovery as measured by the FIM at discharge from inpatient rehabilitation	The lack of interaction between treatment intensity (other than psychological services) and outcome may be due to the choice of outcome measures, lack of control of treatment quality or indications at the included hospitals, mismatch in timing of application of these therapies, or the HCFA mandate of a minimum 3 hours per day of treatment for each patient that was in place during this study. This mandate may have decreased the overall variation within therapies and weakened their correlations with outcome.	Unspecified definition of TBI No control for severity of injury No control over treatments at the involved hospitals Billing hours contain time not spent directly in patient care Lack of long-term followup

Evidence table 2. Study 3 - Spivak, 1992


Design	Patients	Setting	Sampling	Grouping	TBI severity measure	Confounding variables	Method of handling confounding variables	Definition of acute, formalized rehabilitation


Retrospective	95 patients with TBI	Single inpatient rehabilitation unit.	Patients with complete records admitted between 1988-1990	None (0bservational study)	GCS within 24 hours of acute admission, head AIS, duration of coma	TBI severity, severity of extracranial injuries (highest non head AIS), age, time since TBI	Unclear	Hours of treatment measured in 15 minute intervals of actual treatment performance


Independent variables	Dependent variables	Statistical methods

Intensity of treatment during the first month of rehabilitation = total hours of treatment during the first month. Patients split into high and low groups using median value (76 hours). Average daily intensity of treatment over entire LOS. Patients split into high and low groups using median value (4 hours/day). Corrected for off time by dividing all values by 0.66. LOS. Patients split into high and low groups using median value (58 days).	Clinical outcome (in-house rating scales for each discipline grouped along 3 axes [physical performance, higher-level cognitive skills, and cognitively-mediated physical skills]). Also RLA at discharge.	ANCOVA with multiple comparisons/ANOVA with multiple comparisons


Results	Analysis	Caveats

ANCOVA analysis with multiple comparisons of 2 independent variables (1 month treatment intensities or full stay treatment intensity plus LOS) on each of the 3 dependent variables: Statistically significant relation between discharge RLA and 1 month treatment intensity. Borderline significance (p=0.06) between higher-level cognitive skills and average treatment intensity. Borderline significance (p=0.07) for the triple interaction of RLA, LOS, and average treatment intensity. Further univariate ANOVA analysis of this borderline interaction revealed that there was a significant effect of high intensity treatment during the entire stay on Rancho scores in patients with long lengths of stays. This was not true for patients with short lengths of stays.	Authors argued that treatment intensity can play a significant role in improving outcome, especially in patient with poorer presentation and longer LOS's. Reason for weak role of treatment intensity could include: 1) ceiling effect due to most patients getting sufficient treatment such that more is not advantageous; 2) range of treatment too limited to show effect; 3) intensity results only relevant to higher cognitive functions. Rehabilitative treatment in TBI should include 5-6 hours/day once patients are out of coma.	Unspecified definition of TBI Clinical outcome dependent variables other than RLA measured using proprietary, non-standardized scale. Median split method of dividing independent variables into two groups may not provide physiologically meaningful groups. Unclear methods of controlling for confounding variables. Over emphasis on interactions that are of "borderline" significance (p values of 0.06 and 0.07) Lack of lack of multivariate statistical handling of potential confounding variables. Lack of long term follow-up

Evidence table 2. Study 4 - Blackerby, 1990


Design	Patients	Setting	Sampling	Grouping	TBI severity measure	Confounding variables	Method of handling confounding variables	Definition of acute, formalized rehabilitation

Retrospective	149 brain injury patients (97% with TBI)	Two commercial inpatient head injury rehabilitation provider units	All patients in program between 1986-1988	Number of patients pre vs post major programmatic changes including increase in mean daily therapy hours from 5.5 to 8 in patients in coma stimulation (54/14) and acute rehabilitation (44/37)	None	Age, level of function on admission (measure not specified), time post injury at admission	None (groups stated to not be different)	Mean daily therapy hours


Independent variables	Dependent variables	Statistical methods

Intensity of therapy	Inpatient rehabilitation LOS	T-statistics for pre vs post programmatic change LOS for coma treatment and acute treatment groups


Results	Analysis	Caveats

Statistically significant decrease in rehabilitation LOS in both groups after the increase in intensity of treatment	Decreased LOS correlated with increased intensity of treatment Projected an average cost savings of $16,950 for coma treatment and $18,504 for acute treatment patients.	Unspedified definition of TBI No control for confounding variables Change in treatment program vastly greater than simply the increased therapeutic intensity Treatment differences between the two hospitals involved. Lack of long term followup

Evidence table 3. Randomized controlled trials of cognitive rehabilitation - Health outcomes [arc 1]


Source	Intervention/treatment group	Comparison Group/2nd treatment group	Duration of intervention	Followup	Setting/population

Novack (1996)	Restorative and compensatory Hierarchical, structured CACR with therapist support and external cues.	Unstructured. Memory/reasoning tasks, games, computer games.	30 min. sessions 5 days/week 20 sessions 10 hours total	None	Acute inpatient rehabilitation Consecutive admissions over 3 years
Schmitter-Edgecombe (1995)	Compensatory Notebook training with wristwatch alarm cue	Group sessions for problem solving, discussion of social isolation, frustrations	60 min. sessions 2 days/week 8 weeks 16 sessions 16 hours total	6 months	Outpatient volunteers
Neistadt (1992)	Restorative Functional skills group (T1) trained in food preparation.	Perceptual skills group (T2) trained in parquetry block assembly. Each group trained in one skill, tested for both skills.	30 min. sessions 3 days/week 6 weeks 9 hours total	None	Boston School of Occupational Therapy, Tufts University. Subjects recruited from 10 head-injury programs, 8 residential, 2 outpatient
Ruff (1990)	Restorative and compensatory CACR and external aids (notebooks, calendars, schedules, timers, etc.)	Psychosocial functioning and ADLs	3 hours/day 4 days/week 8 weeks 96 hours total	None	University of San Diego Outpatients Population not specified
Helffenstein (1982)	Compensatory Interpersonal process recall (IPR) Videotape of social interaction, viewing of tape, feedback, corrections and practice	Nontherapeutic attention (with no feedback on interpersonal functioning)	1 hour/day 20 days 20 hours total	1 month (on 6 subjects)	Brain Injury Project Woodrow Wilson Rehabilitation Center


Source	N	Chronicity	Severity	Outcomes/analysis	Results	Level

Novack (1996)	T - 22* C - 22	T - 5.9 wks (3.3)** C - 6.4 wks (4.9)	21-GCS < 8 3-moderate GCS + positive CT 20-8 days coma + positive CT	Digit Span & Mental Control subtests of WMS, computer-based measures of reaction time (RT), Neuropsych. Batter, FIM on 24 of 44 subjects MANOVA/ANCOVA's, t-tests	No treatment effect [Delta]for FIM ADLs T = 29.3 C = 29.2	I
Schmitter-Edgecombe (1995)	T - 4 C - 4	T - 77.7 mos (46.8) C - 86.8 mos (67.7)	T = 139.3 DRS (2.2) C = 140.5 DRS (2.6)	Laboratory Recall (Index from WMS Logical Memory 1 & 2, Visual Reproduction 1 & 2) RBMT profile score Everyday Memory Questionnaire Observed Everyday Memory Failures (EMFs) Symptom Checklist 90 ANCOVAs	No treatment effect on 4 of 5 measures. Treatment group had fewer EMFs than control group at posttreatment. [Delta]for mean # of EMFs T = -23.37 C = -7.75 No treatment effect at followup	I
Neistadt (1992)	T1 - 23 T2 - 22	7.9 years (6.6)	< 10 WAIS-R Block Design scaled score Less than perfect score on pretest RKE-R	Parquetry Block Design Rabideau Kitchen Evaluation Revised (RKE-R) WAIS-R Block Design Subtest ANOVAs, t-tests	No treatment effect on RKE-R [Delta]for RKE-R T1 = 7.92 T2 = 2.68 Perceptual Skills Group (T2) scored significantly higher than Functional Skills Group (T1) on Parquetry Block Design	I
Ruff (1990)	T - 12 C - 12	T - 44.3 months (25.6) C - 52.2 months (19.2)	T - 25.5 coma days (16.4) C - 48.3 coma days (28.3)	Katz Adjustment Scale (KAS) Self-Report and Family-Report MANOVA/ANCOVA	No treatment effect [Delta]for means on KAS global scale scores ranged from: T = 4, C = 1 (Social Obstreperousness, Self-Report) to: T - -0.6, C = -0.4 (Acute Psychoticism, Family-Report)	I
Helffenstein (1982)	T - 8 C - 8	Not specified	Estimated to be mild to moderate	State Trait Anxiety Scale (STAS) Tennessee Self-Concept Scale (TSCS) Interpersonal Communication Inventory (ICI) Interpersonal Relationship Rating Scale (IRRS) Independent Observer Report Scale Videotape analysis ANCOVAs	Treatment effect on 1 variable from STAS and 3 variables from TSCS. Treatment effect on IRRS scale, and Independent Observer Report Scale (group means not provided)	II (a)

Evidence table 4. Comparative study of cognitive rehabilitation - Employment [arc 2]


Source	Intervention/target	Comparison	Duration of intervention	Followup	Setting/population

Prigatano (1984)	Restorative and compensatory Intensive, coordinated, multidisciplinary CR. Stresses awareness of deficits, compensatory skills development. Staff includes clinical neuropsychologist, speech pathologist, O.T., P.T., research psychologist.	No neuropsych. Rehab. progam (NRP) Other interventions not specified	4 days/week 6 hours/day 6 months 624 hours total	Between 33 months and 3 months, depending on when the person entered the program and was discharged	Treatment group: TBI clients who entered Presbyterian Hospital NRP between 2/80 and 8/82 who stayed at least 6 months. Control group: TBI files of referrals to NRP between 2/80 and 8/82 who did not enter the program were retrospectively examined.


Source	N	Chronicity	Severity	Outcomes/Analysis	Results	Level

Prigatano (1984)	T = 18 C = 17	T - 21.6 months C - 13.6 months	Not specified	1. WAIS-R Verbal IQ, Performance IQ, Vocab., Block Design, Digit Symbol. WMS Memory Quotient, Logical Memory, Visual Reproduction, Assoc. Learning. Halstead Reitan Trail Making Test, Finger Tapping, Tactual Performance Test. Russell-Neurenger Average Impairment Scale 2. KAS Relative Scale 3. Employment ANCOVAS	Treatment Effect: 1. WAIS-R Performance I.Q. [Delta]T = 8.7, C = 4.8 2. Block Design [Delta]T = 2, C = 1.4 3. WMS Memory Quotient [Delta]T = 9.5, C = 2	II(b)

Evidence table 5. Randomized controlled trials of cognitive rehabilitation - Intermediate outcomes [arc 3]


Source	Intervention/treatment group	Comparison group 2nd treatment group	Duration of intervention	Followup	Setting/population

Thomas-Stonell (1994)	Restorative and compensatory CACR with therapist intervention Teachware	Traditional therapy, community school programs	1 hour/day 2 days/week 8 weeks 16 hours total	None	Hugh MacMillan Rehab Centre - Toronto Population not specified
Twum (1994)	Restorative Imagery training specific to Verbal Task outcome measures; verbal labeling training specific to visual task outcome measures. Test stimuli presented until recall was perfect or until 6th trial. Delayed recall tested at 30 min.	Four groups: No training, verbal training, imagery training, both trainings	Single training session	None	Towson State University Referral sources not specified. All had neuropsych. evaluations by State DVR
Niemann (1990)	Restorative Computer and non-computer attention remediation	Restorative and compensatory memory training	2 hours/day 2 days/week 9 weeks 36 hours total	2 weeks	Outpatients -- U.C. San Diego Inf. Center. Contacted through hospitalsl, community colleges, and S.D. Head Injury Foundation
Ruff (1989)	Restorative and Compensatory CACR and external aids (notebooks, calendars, schedules, timers, etc.)	Computer and video games, coping skills, health, discussion, independent living, art	5 hours/day 4 days/week 8 weeks 160 hours total	None	University of San Diego Population not specified.
Ryan (1988)	Restorative and compensatory External mnemonics, encoding strategy practice, personalized emotional techniques, rehearsal, CACR, synthesis of all in group practice	Games, psychosocial support, art, group discussions, self-expression, relaxation exercises	5.5 hours/day 4 days/week 6 weeks 24 sessions 132 hours total	None	Univ. Virginia School of Medicine & Woodrow Wilson Rehab. Center Population not specified
Kerner (1985)	Restorative CACR Computer memory retraining group (CMRG)	Two comparison groups: 1. Computer control group (CCG) Used computers to create graphics. 2. No Exposure control group (NECG)	45 min. sessions 12 sessions 4.5 weeks 9 hours total	15 days CMRG only	DeAnza College Population not specified


Source	N	Chronicity	Severity	Outcomes/analysis	Results	Level

Thomas-Stonell (1994)	T - 6** C - 6	Ranged from 3 months to 4 years	Recovered to 7 or 8 on Rancho Scale	TeachWare Screening Module Standardized neuropsych. test Battery ANCOVAS used to test group differences	Treatment effect on 8 of 18 neuropsych. subtests. Group means not presented.	I
Twum (1994)	T1 - 15** T2 - 15 T3 - 15 T4 - 15	Average 13.2 months from return to consciousness to treatment	> 3 weeks coma Average WAIS-R IQ = 80	Verbal task mean # words Recalled, delayed recall, & trials to Perfect Visual task mean # words recalled, Delayed recall, & trials to perfect MANOVA	Treatment effect Imagery group scores higher than verbal on all verbal tasks. Verbal group scores higher than imagery on all imagery tasks. Difference in means between imagery and no imagery group on: 1. Immediate recall: +8 2. Delayed recall: +1.5 3. Trials to criterion: -2.5 Difference in means between Verbal Labeling and No Verbal Labeling Group on: 1. Immediate recall: +5 2. Delayed recall: +.75 3. Trials to criterion: -2.25	I
Niemann (1990)	T - 13 C - 13	T - 41.0 C - 37.1	Coma T - 15 days C - 20 days GOAT T - 94.4 (5.5)*** C - 90.7 (6.8) DRS T - 132.9 (9.0) C - 135.2 (7.0)	Attention test d2 PASAT Divided Attention Test Trails B RAVLT Block Span Learning SDNTB subtests MANOVA	No treatment effect	I
Ruff (1980)	T - 20 C - 20	Ranged from 1 to 7 years	T - 32.1 coma days (21.4) C - 48.8 coma days (26.4)	San Diego neuropsych. battery Forms A and B MANOVA	No treatment effect	I
Ryan (1988)	T - 10 C - 10	T - 54.5 months C - 57.3 months	Each group had 5 milds (DRS > 134/144) and 5 moderates (DRS < 134/144)	BVRT, Rey CFT, Taylor Complex Figure, Selective Reminding Test, Ruff Trail, WMS Logical Memory subtest MANOVA	No treatment effect	I
Kerner (1985)	CMRG - 12 CCG - 6 NECG - 6	> 3 months	Memory index rating severe to mild	Memory index (MI) scaled & standard scores Acquisition recall (AR) scaled & standard scores t-tests	Treatment effect on 5 of 12 measures ~; [Delta]AR (scaled) for CMRG = -4.33, for CCG = 0.00 [Delta]MI (scaled) for CMRG = -5.92, for CCG = 0.50, for NECG = 0.66 [Delta]MI (standard) for CMRG = -5.58, for CCG = 0.50, for NECG - 0.33 Significant increase for CMRG was not maintained at followup	I

Evidence table 6. Comparative studies of cognitive rehabilitation - Intermediate outcomes [arc 3]


Source	Intervention/treatment group	Comparison group/2nd treatment group	Duration of intervention	Followup	Setting/population

Gray (1992)	Restorative CACR Tasks selected that make demands on alerting, working memory, alternating attention and divided attention. Used feedback, reinforcement, visual stimuli, and cuing.	Recreational computing	T - 75 min. sessions 14 sessions over 3 - 9 weeks 17.5 hours total C - 60-90 min. sessions over 3 - 9 weeks Mean 12.7 hours total	6 months	Newcastle General Hospital, Cambridge Univ., U.K. Psychologists in outpatient clinics, staff of social services, and support groups in Edinburgh solicited for names of people with attention deficits due to brain injury.
Batchelor (1988)	Restorative CACR directed toward remediation in recent memory, attention/speed of information processing, and higher cognitive functioning	Restorative cognitive therapy, directed toward remediation consistent with Treatment Group, but delivered without computers	4 - 6 weeks 20 hours total	None	Westmead Hospital, Australia Consecutive referrals to rehabilitation medicine unit over 9 months
Wood (1987)	Restorative Visual training of information processing using CACR	C1 - Clients in same inpatient rehabilitation center as treatment group, who did not receive the CACR intervention C2 - Persons without TBI	1 hour/day 20 days 4 weeks 20 hours total	20 days	Inpatient rehabilitation center St. Andrew's Hospital, U.K.


Source	N	Chronicity	Severity	Outcomes/analysis	Results	Level

Gray (1992)	T - 17 C - 14	T - 79 weeks (151)** C - 84 weeks (152)	T - 8 mild to moderate, 9 severe C - 8 mild to moderate, 6 severe	PASAT # Correct, Longest String, and Information Processing Rate (IPR) WAIS-R subtests Neuropsych. Battery 22 Total Tests ANCOVAS used for analysis	Treatment effect on 2 of 22 tests at posttreatment Treatment effect on 6 of 22 tests at followup Group means not presented	II(b)
Batchelor (1988)	T - 17 C - 17	T - 72.7 days (66.5) C - 96.3 days (104.1)	T - 7.3 coma days (6.3) C - 7.0 coma days (8.7)	WAIS-R Russel's WMS Buschke Selective Reminding Test Taylor Figure PASAT Austin Maze ANCOVAs and t-tests used for analysis	No treatment effect	II(b)
Wood (1987)	T - 10 C1 - 10 C2 - 10	T - 27.5 months (5.8) C1 - 36.5 months (15.6)	PTA T - 2.4 months (0.5) C1 - 2.7 months (0.4) All required full time care	Pursuit Rotor, Digit Symbol, Choice Reaction Time, Simple Reaction Time, Visual and Choice Reaction Time for vigilance, Attention to Task, Attention Rating Scale	Treatment effect for attention to task and attention rating scale from baseline to first followup. Treatment effect for choice reaction time from baseline to second followup. Group means not presented.	II(b)

Evidence table 7. Studies of the relationship between intermediate tests and employment [arc 5]


Source	Setting/population	N	Chronicity	Severity

Girard (1996)	Current and former clients of a hospital-based, interdisciplinary outpatient TBI program.	Initial - 152 6 month followup - 114 12 month followup - 69	3 years (range 6 months to 12 years)	Not specified


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Girard (1996)	28 tests and subtests: Symbol Digit Written and Oral Halstead-Reitan Trails A & B WAIS-R Verbal subtests (6), Performance subtests (5) WMS-R Verbal Memory, Visual Memory, General Memory, Attention/Concentration, Delayed Recall Wisconsin Card Sorting Test The Booklet Category Test Wide Range Achievement Test (WRAT-R)	McAuley Outcome Scale Measures productive outcome in home, school, work. Scale ranges from 1 (not productive) to 6 (more productive than premorbid level).	Prospective case series. ANOVAs and Stepwise Multiple Regression used for analysis.	67% of sample was productive at followup, 33% not productive. 6 scores significantly related to outcome at first followup, 8 at second. With regression analysis, 9 test scores (WAIS-R Block Design, Digit Symbol, Picture Completion, and Full Scale IQ; Symbol Digit Written and Oral; Trails B; WAIS-R Arithmetic and WRAT-R Arithmetic) and 3 demographic characteristics (mechanism of injury, insurance funding, and premorbid substance abuse) accounted for 30% of the variance in outcome.	III


Source	Setting/population	N	Chronicity	Severity

Cicerone (1996)	Referrals to neuropsychology clinic of brain injury rehabilitation program. Selected on basis of having participated in neurorehabilitation, having neuropsych. evaluation, and being available for followup.	20	7.8 months (range 3 to 20 months)	Mild


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Cicerone (1996)	13 neuropsych. tests: Digit Span Forward & Backward Trails A & B Continuous Performance Test of Attention PASAT Logical Memory I & II CVLT Rey CFT Immediate and Delayed Recall WCST Perseveration Category Test Mazes Verbal Fluency	Good Outcome Group (GO): Clients who resumed responsibilities equivalent to premorbid status, or other, less demanding, productive activity (n = 10). Poor Outcome Group (PO): Clients unable to resume premorbid level of activity/responsibility (n = 10).	Retrospective t-tests	GO clients had significant improvement on 6 of 13 neuropsych. tests. PO clients had significant improvement on 1 neuropsych. test.	III


Source	Setting/population	N	Chronicity	Severity

Ip (1995)	Consecutive referrals for rehabilitation to Brain Injury Unit of a hospital between 1988 and 1994.	70	3.1 years	20% mild 27% moderate 53% severe (based on GCS)


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Ip (1995)	16 neuropsych. tests: WAIS-R Performance IQ, Verbal IQ, Full Scale IQ WMS-R Delayed Recall, Attention/Concentration, General Memory Index, Verbal Memory Index, Visual Memory Index Halstead-Reitan Trails A & B, Tapping dominant & non-dominant, Grip Strength dominant & non-dominant Grooved Pegboard Test dominant & non-dominant	Dichotomous measure of return to work or school (RTW/S), either full or part time, following hospital discharge	Retrospective evaluation of medical charts and followup phone calls to obtain RTW/S data. t-tests used for continuous scales chi-square used for nominal scales Wilcoxon rank sum test used for ordinal scales logistic regression used to evaluate association of neuropsych. tests with RTW/S	At followup, 42% had returned to work/school, 58% had not. 5 of 16 neuropsych. tests significantly related to RTW/S	III


Source	Setting/population	N	Chronicity	Severity

Fabiano (1995)	Consecutive referrals to 3 post-acute rehabilitation facilities. Minimum length of coma 24 hours, minimum chronicity 1 year.	94	59 months (42.6)*	Severe 20 days average length of coma (20.2)


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Fabiano (1995)	8 subtests of WAIS-R Picture Arrangement, Full Scale IQ, Performance IQ, Similarities, Block Design, Digit Symbol, Prediction Points, Verbal IQ, Length of Coma	Full time employment (n = 20), part time employment (n = 15), successful college attendance (n = 8), supported/sheltered employment (n = 11), unemployed (n = 40)	Retrospective evaluation of medical charts and followup phone calls to obtain current employment data. Discriminant Analysis used to specify strongest predictors of employment.	At followup, 8 subtests predicted employment status as follows: Full Time Employed: 62% accurate Part Time Employed: 58% accurate Not competitively employed: 67% accurate	III


Source	Setting/population	N	Chronicity	Severity

Ezrachi (1991)	Consecutive participants in NYU Head Trauma Program over 4 years. Sample chosen on basis of not being able to return to work for 1 year postinjury, and willingness to participate in program.	59	34.65 months (27.49)	Moderate to severe 36.2 days in coma (31.42)


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Ezrachi (1991)	38 neuropsych. subtests from 7 instruments: Orientation Remedial Module (ORM) Purdue Pegboard Visual measures from Rusk Institute of Rehabilitation Medicine (RIRM) Academic measures from MAT and WRAT WAIS-R Higher order verbal and conceptual skills from RIRM 19 tests of functional behavior - Behavioral Competence Index (BCI)	Vocational Status (VSTAT) 6 months after completion of program, rated on a 10 point scale from 1 (remains unemployable in any capacity) to 10 (returned to academic level position; no qualifications).	Prospective case series Factor analysis was used to reduce data for neuropsych. tests and BCI., with pretreatment scores to determine factors. Data reduction resulted in 12 neuropsych. factors and 4 BCI factors. After generating factor structures, factor scores were calculated separately from pretreatment and posttreatment data for each factor to serve as predictors for VSTAT. Stepwise multiple regressions were used to evaluate association of factors derived from neuropsych. tests and BCI with VSTAT.	At 6 month followup, factors from pretreatment data with strongest association to VSTAT were: Adaptation to community Verbal Aptitude Self-Appraisal Coma Factors from posttreatment data with strongest association to VSTAT were: Acceptance of program Verbal aptitude Involvement with others Psychomotor dexterity Self-appraisal Coma	III


Source	Setting/population	N	Chronicity	Severity

Fraser (1988)	Consecutive outpatient referrals who were employed prior to injury	48	Not specified	Average GCS for 35 who returned to work = 13 Average BCS for 13 who did not return to work = 11


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Fraser (1988)	9 neuropsych. tests: Tactual Performance Test (TPT), Category Test, Trails B, Halstead's Impairment Index, WAIS IQ (Verbal, Performance, Full Scale), WAIS Digit Symbol, and Name Writing Procedures	Dichotomous measure of returned to work or did not return to work.	Prospective case series Binary data analyzed with chi-square tests of association All other data analyzed with Wilcoxon rank sum tests	At 1 year followup, 35 had returned to work, 13 had not returned to work. 6 of 9 neuropsych tests significantly related to employment.	III


Source	Setting/population	N	Chronicity	Severity

Brooks (1987)	Consecutive admissions to Department of Neurosurgery, Institute of Neurological Sciences, Glasgow, UK	134	Ranged from 2 to 7 years	Coma duration > 6 hours, or PTA > 2 days, or surgery for intracranian hematoma


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Brooks (1987)	9 neuropsych tests: Progressive Matrices and Mill Hill Vocabulary Scale, WMS Logical Memory, Verbal Paired Associates, Buschke Procedure, Rey Complex Figure Copy, PASAT, Borkowski Word Fluency Test, Token Test, and a structured interview.	Dichotomous measure of employed or not.	Retrospective evaluation of medical charts and followup interview of client and informant 2 to 7 years posttrauma. t-tests used to compare neuropsych. tests scores between groups.	At followup, 34 clients were employed full or part time, 85 were unemployed, 13 (status other, i.e., homemaker, retired), 2 unknown Groups differed significantly on 6 of 9 neuropsychology tests	III


Source	Setting/population	N	Chronicity	Severity

Najenson (1980)	Consecutive discharges from Lowenstein Rehabilitation Hospital, Ra'anana, and Tel Aviv University School of Medicine, Israel, from 1/11/74 to 1/4/7	147	Not specified	Coma duration ranged from < 1 day (n = 21) to > 30 days (n = 21)


Source	Intermediate tests	Health/employment outcomes	Design/analysis	Results	Level

Najenson (1980)	2 neuropsych. tests: Raven Matrices Test and Wisconsin Card Sorting Test	Four employment categories: Skilled work, Unskilled Work, Sheltered Workshop, Unemployed	Prospective case series Frequencies reported, and statistical significance (p < .005) reported; method of analysis not reported.	At > 6 months followup (skilled and unskilled work groups were combined to constitute "working"): For Raven Matrices Test, 10 of 14 who scored above average were working (71%); 17 of 31 who scored average were working (55%); 17 of 48 who scored below average were working (35%). For Wisconsin Card Sorting Test, 31 of 40 who scored normal were working (78%); 18 of 62 who scored slight rigidity were working (29%); 3 of 16 who scored severe rigidity were working (19%).	III

Evidence table 8. Observational study of cognitive rehabilitation - Health outcomes [arc 1]


Source	Intervention	Duration of intervention	Followup	Setting/population	N

Wilson (1997)	Compensatory device programmed to provide reminders of daily activities neuropage	baseline 2 - 6 weeks treatment - 12 weeks baseline - 3 weeks	none	Cambridge Univ., U.K. referrals from hospital, therapists, psych. clinics, support group	15


Source	Chronicity	Severity	Outcomes/Analysis	Results	Level

Wilson (1997)	Ranged from 6 months to 13 years	RBMT ranges from severe to moderate or mild	Everyday Memory Failures (EMFs) maximum score individually determined, based on number of target reminders Mean % success used to measure effect	Significant increase in scores for all subjects during treatment. Score decreased for 11 of 15 during second baseline. Mean % Success: Baseline - 37.08% (24.86)* Treatment - 85.56% (18.58) Baseline - 74.46% (28.23)	III

Evidence table 9. Comparative studies of supported employment: Study characteristics

Source

Study rating/Comment

Design/Allocation

Outcome Measures

Followup

Haffey & Abrams,
1991

Class IIb study.
Highly biased assignment to groups under comparison. T1 group (WRP) contained clients selected as good job prospects by rehab. counselors. T2 & C full of poor job prospects.

Retrospective initial data collection for clients whose injury predated 1/88.
Prospective, non-randomized comparison of three groups:
T1: Supported employment (WRP)
T2:Day Treatment
C:Comparison Group

% Job placements
% Job retention
Monthly Employment Ratio (MER)

Every 6 months. Clients entering at the beginning followed up for 3 years, others less time
N Months followup
18 1 to 6
23 7 to 12
17 13 to 18
15 19 to 24
14 >24

Wehman, et al.,
1989 to 1993
(Cumulating sample, reported in 4 articles from 1989 to 1993)

Class III study.
Excellent measures and followup. Study focused on difficult population. Prospective data collection. Weakness of case-control study with no independent control groups and random assignment to conditions.

Prospective, case-control study with each consumer moving through 3 consecutive phases of employment history:
1. Pre-trauma employment
2. Post-trauma employment
3. Supported employment (SE)
(Post-trauma employment Phase designated as -SE)

% Job placements
Monthly Employment Ratio (MER)
Number of jobs held
Hourly wages
Monthly wages
Annual wages
Hours/week of work
Intervention hrs/wk required

Weekly job assessment on-site. Followup assessment every 6 months.

N	Years followup
5	5
15	4 to 5
33	>3
62	.5 to 3

Evidence table 10. Comparative studies of supported employment: Selection and allocation of clients


Source	Population/Selection	Allocation to comparison groups	N	Severity/Chronicity	Exclusions/Attrition


Haffey & Abrams, 1991	Supported employment (WRP) All referrals from 1/88 to 8/90 by State rehabilitation program or State worker's compensation as unable to find or keep jobs but suitable for work "under the right circumstances." Day treatment group Cases treated from 1/88 to 8/90 for whom competitive employment was not a current goal because of: medical problems, personal & family preference, economic disincentives, other activities (homemaker, student) Comparison group Consecutive discharges from inpatient rehabilitation program from 4/88 to 8/90 who did not enter WRP because: believed unnecessary, perception of severe functional limits, dependency reinforcement, threat to potential benefits, or substance abuse.	130 clients in WRP, all recommended by State vocational rehab as candidates for work. 35 clients in a day treatment program for whom competitive employment was not a goal for various reasons. Comparison group of 76 clients who received no formal post-acute rehabilitation.	T1: WRP: n = 130 T2: Day Treat. n = 35 Comparison: n = 76 N = 241	Group Md days coma T1 6 T2 7 C 7 All 7 Chronicity not specified.	Potential clients with current substance abuse or disabling conditions were referred to other treatment. Clients believed to have no potential for employment by State rehabilitation counselors, or by self or family, were excluded.

Wehman, et al., 1989 to 1993 (Cumulating sample reported in 4 articles from 1989 to 1993)	Referrals for supported employment from physicians (physiatrists), psychologists, rehabilitation counselors, and families. Inclusion criteria: 1. Between 18 and 64 year age. 2. Severe TBI (GCS ·8 for ·6h). 3. Strong evidence cannot work without job support. No exclusion merely for cognitive, physical, or psychosocial deficit.	3 work consecutive work phases compared for same group of 115 consumers: Phase 1: Pre-trauma employment Phase 2: Post-trauma employment Phase 3: Supported employment	Cumulating sample over 5 years, reported as follows: 1989 n = 5 1989: n = 20 1990: n = 53 1993: N = 115	All consumers severe TBI (GCS·8 for ·6h). Av. coma: 68 days. Range: 3-233 days. Mean scores on neuro-psych. tests below 50th percentile; range 10%-50%. Mean chronicity 7.75 years. Range 1-19 years.	Potential clients with current substance abuse were excluded.

Evidence table 11. Comparative studies of supported employment: Models of supported employment tested and outcomes


	Models of supported employment tested			Outcomes

Source	Setting	Intervention	Duration	Analysis	Results

Intervention hours required:
Haffey & Abrams, 1991	Community Re-entry services, Work re-entry program (WRP), Sharp Memorial Center, San Diego, CA	Transitional employment work entry program (WRP): Includes intake, vocational assessment, work hardening, job development, job analysis, transitional employment program, job placement, short-term support, and long-term followup. On-the-job support by a job coach for first 60 days-intense at first, then fading at the end.	2 hours intake26 hours initial assessment6 hrs job preparation12 hrs job development10 hrs job placement and support3 hrs post-placement employer contact60 hrs total of staff assistance60 days on-the-job support	Descriptive statistics Percentages	Placement as of 10/90:T1: 68% (88/130)T2: 39% C: 34%Retention at most recent followup:T1: 71% (62/88)T2: Not reported C: Not reported

Evidence table 12. Case management studies in traumatic brain injury rehabilitation: Design characteristics


Publication	Study purpose (target/problem-hypotheses/goals)	Design features	Intervention allocation method	Population characteristics	Group sizes	Group differences	Case management intervention model	Intervention duration	Measurement period	Co-interventions described




Ashley et al. (1994)	Target: pt progress & level of independence at end of rehab	Retrospective, comparison of two CM conditions, case-control study: Grp 1 selected then Grp 2 matched	Grp 1 subjects covered by same insurance carrier with a single CM; Grp 2 subjects covered by different insurance carriers with different CMs	~ 350 TBI patients of one post-acute rehab clinic treated from 1980-90, Disability Rating Scale score < 10 (moderately severe) & in vocational rehab Grp 1 Inclusions: admitted by a single carrier & single CM Grp 2 Matches: males; 20 yrs + at injury; treated in 1981 or in 1983-86; Workers' comp insurance coverage; not working; assigned as external CM	Grp 1 (same CM) = 21 Grp 2 (different CMs) = 18	Grp 1 = more pts w/ permanent disability claim advances & more autonomous CMs (authority to approve both disability payment & rehab services)	Insurance coverage models: Grp 1 = 1 CM & 1 insurer, Grp 2 = >1 CM & >1 insurer; CM variable = either or both disability-claims-approval authority & rehab authority	Inpatient rehab stay: means: Grp 1 = 230 days, Grp 2 = 276 days	At rehab discharge (8 - 9 mos)	Insurance coverages: number & type
Greenwood et al. (1994)	Problem: lack of rehab after acute period; Hypotheses: CM would: 1) decr hosp stay, 2) incr number of pts into rehab, 3) incr rehab duration, 4) not alter cognitive impairment, 5) improve employment & QOL, 6) reduce relatives' burden	Prospective, comparison of sequential pts in randomized matched settings (2 sets of 3 hospital units); intervention group (CM) = normal hospital services + case management; control group (NON-CM) = normal hospital services	By sequential hospital admissions	Inclusions: TBI treatment from 3/88 - 11/90; 16-60 yrs, < 7 days post-injury, > 6 hrs in coma or > 48 hrs amnesia = severe injuries; family consent;local resident Exclusions: hosp tx for substance abuse; psych or CNS disturbance in prior yr; homeless or unlikely to have followup	CM = 56 pts at entry 48 at 6 mos 37 at 12 mos 31 at 24 mos NON CM = 70 pts at entry	CM = more severity than NON CM: i.e. longer coma, lower Glasgow coma scores, more respiratory complications & tracheostomies, less conservative management, longer post-injury amnesia & DRS 18.3 compared with DRS 16.2 in NON CM Grp	Care-continuity model: advice, support, goal-planning, advocacy & referral; Tasks: assessed pt; developed rehab plan; facilitated cooperation & involvement of pt/fam & professionals; provided general info re head injury & informal counseling & support	Up to 2 yrs post-injury maximum	At 6, 12, 24 mos post-injury; 2-yr study	Care settings: neuro hospital, general hospital, rehab unit, outpt, daycare, home; services: physical therapy, occup therapy, speech therapy, psychology, social work, psychiatry, vocational rehab
Malec et al. (1995)	Goals: reduce time between brain injury & community reintgration from 7 yrs to 9 mos with 70% employment or education at level > 3 on Vocational Outcome Scale (VOS) & 45% employment at level 5 on VOS	Prospective, one setting, comparison of CM & non-CM via benchmarks	By sequential pt screenings	509 TBI patients of one emergency room treated from 10/94 - 10/95: Inclusions (n = 147): 18 - 55 yrs, primary dx of TBI & if receiving appropriate treatment for psychiatric or substance abuse comorbidities; Exclusions (n=350): a primary psychiatric or substance abuse dx; in residential care facility; no evidence of brain injury w/neuropsych exam; non-English speaking; out-of-state resident; developmentally disabled; followup refused against medical advice	CM = 147; 25 mild injury pts lost to followup; total = 122	CM Grp = 79% mild injury, 21% moderate or severe injury	Care-continuity + vocational coordination model: nurse case coordinator (NCC) tasks: screen; advise; support & reassure re emotional sx; monitor status; direct to medical, neuropsych & rehab services & to vocational case coordinator (VCC); VCC tasks: assess, plan, counsel pt/employer re job modifications & compensation techniques; support in job search; address emotional, behavioral, & social issues	No sx = 1 month w/ sx = 1 mo & 12 mos after subjects complete program	At 1 & 12 mos post-rehab; 2-yr study	Medical center services, home & community services

Evidence table 13. Case management studies in traumatic brain injury rehabilitation: Results


		Outcomes

Publication	Outcome measures	Functional status	Vocational	Living arrangements	Family	Other	Overall




Ashley et al. (1994)	Changes in disability rating scale (DRS), living status, occupational status, treatment costs	Grp 1 = more disability (DRS) improvement (moderate to mild); treatment effect: DRS mean improvement difference = .58; DRS at rehab discharge: Grp 1 = 1.14, Grp 2 = 1.94	Grp 1 = most improved occupational status (not working to part time at former job or equal position); treatment effect: occupational rating mean improvement difference = 7.04	Grp 1 = most improved living status (private qrtrs w/ professional help to private qrtrs w/o help); treatment effect: living rating mean improvement difference = 1.83	NA	Grp 1 = greater use of pre-settlement permanent disability advances; CM had both claims & rehab authority; lower rehab costs ($40,000 less per pt)	Both groups had significant differences in adm-discharge scores on disability, living & occupational status; Grp 1 made significantly more improvements in disability, living & occupational status
Greenwood et al. (1994)	Number of referrals to inpt/outpt facilities/services; time in rehab services; physical & cognitive impairment (including DRS); changes in affective, behavioral, social functioning & personality; global impairment; changes in pt's & relatives' housing, financial, vocational, leisure & medical needs; requirement for hours of care support & supervision; pt/family satisfaction with services & information; satisfaction w/CM	At 24 mos: DRS difference = 1.24 (higher for CM group), (not reported for 6 & 12 mos); no significant differences in other measures; DRS at 24 mos post-injury: 2.0 CM pts; 0.76 non-CM pts	NR	More CM patients needed a caretaker at 12 & 24 months	More CM families sought medical care, changed their use of leisure time, and rated the effect on the family as major; most (62%) families rates CM as helpful	NR	No significant differences in functional improvements; all trends favored control group, accounted for by difference in severity of injury; CM increased chance & range of contact with services (esp psychology, social work, speech therapy), but not duration of contact; more CM pts/families reported that the accident had had a major effect on the family; more CM pts required a care attendant.
Malec et al. (1995)	Change in brain-injury-related disability (Mayo-Portland Adaptability Inventory) per pt & staff; level of employment function (vocational outcome scale); job type, job setting, pay rate, type & cost of voc supports; independent living level; satisfaction of pts & family members	NA	At 1 yr: 34% went to nonsheltered work or training; 100% had retained a job for 90 days; 73% had been placed within 9 mos	NA	NA	Identified new service needs for adolescents, seniors & alcoholic pts	Focus is limited to vocational impact.