User Guide Series
Traumatic Brain Injury Model Systems National Database
Arun Karpur, MBBS, MPH
Research Faculty
Employment & Disability Institute
ILR School
Cornell University
16 E 34th St, New York, NY 10016
Prepared For:
The Center for Rehabilitation Research using Large Datasets
The University of Texas Medical Branch
301 University Blvd,
Galveston, TX
For additional information about this guide contact:
Arun Karpur, Research Faculty
Employment and Disability Institute
16 E. 34th Street
Cornell University
New York, New York 10016
(212) 240-2844 (Phone)
(212) 240-2822 (Fax)
Ak564@cornell.edu (email)
The user guide has been developed through the Center for Rehabilitation Research using
Large Datasets (CRRLD). The CRRLD is a collaboration between the University of
Texas Medical Branch (UTMB) and Cornell University’s Employment and Disability
Institute (EDI) and is funded through a grant from the National Institute of Health (NIH
grant # R24HD065702). The goal of the Center is to build rehabilitation research
capacity by increasing the quantity and quality of rehabilitation outcomes research using
large administrative and research datasets. The contents of this paper do not necessarily
represent the policy of the U.S. Department of Health and Human Services, and you
should not assume endorsement by the Federal Government (Edgar, 75.620 (b)).
Acknowledgement
Many thanks to the following individuals who provided valuable assistance with the development of this paper: Cindy Harrison-Felix, Gale Whiteneck and David C. Mellick at the Craig Hospital Traumatic Brain Injury Model Systems Data and Statistical Center, Jennifer Marwitz at Virginia Commonwealth University, Cate Miller at NIDRR, William Erickson and Margaret Waelder at Cornell University.
Disclaimer:
The Traumatic Brain Injury (TBI) Model Systems National Database is supported by the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (NIDRR) in collaboration with the TBI Model Systems Centers. However, these contents do not necessarily reflect the opinions or views of the TBI Model Systems Centers, NIDRR or the U.S. Department of Education
University at Texas-Medical Branch Project Team
Ken Ottenbacher, Ph.D., OTR (Principal Investigator)
James E. Graham, Ph.D.
Amol Karmarkar, Ph.D.
Employment and Disability Institute (EDI) Project Team
Susanne Bruyère, Ph.D., CRC, Director of EDI
William Erickson, Research Specialist
Arun Karpur, Extension Faculty
TBIMS ND Data Collection Process
Traumatic Brain Injury Model Systems National Database - Form I
Traumatic Brain Injury Model Systems - Form II
This User Guide presents information on the Traumatic Brain Injury Model Systems (TBIMS) National Database consisting of longitudinal information on the clinical course of recovery and rehabilitation outcomes for individuals with Traumatic Brain Injury (TBI). The purpose of the User Guide Series is to: (a) provide information on large existing administrative and survey datasets available for rehabilitation research; (b) describe their study designs, including methods and instruments for data collection, data structure, variable descriptions, as well as strengths and limitations; (c) provide descriptive analyses on key variables using the International Classification of Functioning, Disability and Health (ICF) framework; and (d) discuss practical aspects of missing data and data security. In departure from previous User Guides, disability statistics of key variables contained within this User Guide are not derived from nationally-representative survey data, but instead describe the nuances of a research dataset covering a specific population - i.e., people with TBI served within the TBI Model Systems Centers.
The TBIMS program, initiated in 1987, is funded by the National Institute on Disability and Rehabilitation Research (NIDRR), U.S. Department of Education. The TBIMS is a multi-center nationwide program, studying the course of recovery and rehabilitation among individuals with TBI served at centers with a coordinated system of acute neurotrauma, inpatient rehabilitation and lifetime follow-up care. As of 2011, NIDRR has funded 22 TBI centers nationally[1], 16 of which currently receive active funding support.[2] In addition to providing care and treatment within a comprehensive multi-disciplinary system of rehabilitation, these centers are a leading resource for generating new knowledge on effective rehabilitation intervention strategies for individuals with TBI through site-specific research. The centers also collaboratively engage in modular research projects. The focus of both site-specific and module research is in the areas of: (a) health and function, (b) employment, (c) participation and community living, and (d) technology for access and function. Furthermore, these centers also collaborate with the NIDRR-funded Model Systems Knowledge Translation Center for the dissemination of information regarding innovative and best practices originating from their research projects.
All TBIMS programs collect longitudinal follow-up data on individuals served by their centers who meet additional inclusion criteria to study the course of recovery and to document rehabilitation outcomes at one, two and five years post-injury and every five years thereafter for as long as the particular TBIMS center is funded. These data are a rich resource for researchers, enabling the evaluation of long-term recovery patterns for individuals served by these systems of care. The follow-up data are aggregated nationally at the TBIMS National Data and Statistical Center, located currently at the Craig Hospital in Englewood, Colorado. The National Data and Statistical Center coordinates and standardizes the process of data collection across the TBIMS programs forming the TBIMS National Database (TBIMS ND). This database is the only longitudinal long-term follow-up data set that documents pre-injury characteristics, acute care and rehabilitation services, and long-term rehabilitation outcomes for individuals with TBI in the United States. The primary purpose of this User Guide is to describe the TBIMS ND and discuss its potential uses in rehabilitation research by examining the data structure, key variables, and data trends. This User Guide does not provide specific interpretations of the observed trends, but encourages further exploration by researchers.
Before researchers consider utilizing the TBIMS ND, is it important to note some key limitations of this data source.
The International Classification of Functioning, Disability and Health (ICF), provides a conceptual framework to describe the state of functioning and health for individuals with disabilities. The ICF defines disability as a manifestation of interaction between the domains of person, environment, body structures and functions, activity limitation, and health condition (Figure 1). The theoretical framework of the ICF provides a method for grouping information on individuals with disabilities in order to understand their needs and assess their recovery and rehabilitation processes by the use of specific instruments. This User Guide employs the ICF framework in describing the variables for demonstrating conceptual connections across different sections of the TBIMS ND data.
Figure 1. ICF Model of Disability
This figure depicts the World Health Organization’s International Classification of Functioning, Disability, and Health, also called the ICF model. It depicts disability in the workplace as a complex interaction between a health condition, disorder, or disease and the contextual factors in which that condition exists. At the top of the figure, it shows three elements of disability under Health Condition. These are bodily functions and structure, activity, and participation which are on a horizontal plane with two headed arrows between Bodily Functions and Structure on the left and Activity in the middle and between activity and Participation on the right. All three are also connected to Health Condition with two headed arrows. At the very bottom of the figure is Contextual Factors with Environmental Factors and Personal Factors just above on a horizontal plane. Multiple two-directional arrows show that Health Condition, the three elements of disability, and the two contextual factors are all inter-related.
Source: Towards a Common Language for Functioning, Disability, and Health (ICF), World Health Organization, 2002, p. 9 (www.who.int/classifications/icf/trainings/icfbeginnersguide.pdf
End Figure 1
The following tables provide the groupings of variables under each of the ICF domains. Table 1 consists of variables collected in the pre-injury and acute phase (i.e., from injury through emergency acute hospital admission and inpatient rehabilitation) and Table 2 consists of variables in the post-acute phase (i.e., post-rehabilitation discharge follow-up) for TBIMS patients.
Table
1.
Pre-injury and acute variables distributed across the ICF Domains – Form I -
Traumatic Brain Injury Model Systems National Database
|
Personal |
Environmental |
Body
Structure and Function |
Activity
Limitation |
Participation
Restriction |
|
Age Sex Race/Ethnicity Education Pre-injury Marital Status |
TB Center Code Cause of Injury Payment source for
health expenses Pre-Injury
Incarceration Primary language Time spent in acute
care Time spent in rehab
care |
Pre-morbid conditions: blindness, deafness,
physical limitations Blood alcohol level in ER Intracranial CT diagnosis, including intracranial
compression, presence of extra-axial collection, and intraparenchymal
fragments Glasgow Coma Scale scores Revised Trauma Score in ER: Systolic BP Revised Trauma Score in ER: Respiratory Rate Neuropsychological Battery – O-log, California
Verbal Test-II, Reitan Trail Making Test Pre-morbid substance use, psychiatric, and alcohol Pre-injury behavioral disorders Associated spinal cord injury ICD-9 Diagnostic Codes Duration of post-traumatic amnesia Intracranial hypertension Cranial surgery Time between injury and ability to follow commands |
Pre-injury limitations: learning, dressing, going
outside home, working Pre-injury behavioral disorders Disability Rating Scale at rehabilitation
admission & discharge Functional Independence Measure (FIM) at
rehabilitation admission & discharge |
Employment one month prior to injury, including
hours and weeks of competitive work Living situation pre-injury and
post-rehabilitation discharge |
Table 2. Follow-up variables distributed across
the ICF Domains –Form II - Traumatic Brain Injury National Database
|
Personal |
Environmental |
Body
Structure and Function |
Activity
Limitation |
Participation
Restriction |
|
Age Marital Status
at Follow up |
TBI Center Code Incarceration
or arrests Post-injury
income |
Patient Health Questionnaire - 9 (PHQ-9) Generalized Anxiety Disorder - 7 Questionnaire (GAD-7) Substance use, psychiatric, and alcohol use Re-hospitalization |
Disability Rating Scale Functional Independence Measure (FIM) Glasgow Outcome Scale - Extended (GOS-E) Supervision Rating Scale |
PART - 18 – item questionnaire measuring participation restriction Employment post- injury Living situation at follow-up Satisfaction with life scale [actually quality of life measures are not
part of the ICF] |
For the purposes of the TBIMS ND, TBI is defined as "damage to brain tissue caused by an external mechanical force as evidenced by medically documented loss of consciousness or post-traumatic amnesia (PTA) due to brain trauma or by objective neurological findings that can be reasonably attributed to TBI on physical examination or mental status examination.[3]" Any individual with a moderate to severe TBI[4] who is 16 years of age or older at time of injury and presents at a TBIMS center within 72 hours of injury serves as a potential candidate for the TBIMS ND. Further, individuals receiving both acute hospital care and comprehensive rehabilitation services in designated TBIMS programs are eligible to be included in the TBIMS ND. Patients with concurrent injuries or pathologies are not excluded. Individuals, and in some cases their family members, are contacted to secure their consent to participate in the study. The TBIMS ND standard operating procedures offer further details in the process of eligibility determination, inclusion/exclusion criteria, and obtaining informed consent from the potential participants for the study[5]. Either members of the clinical care team or research team at each TBIMS center are responsible for recruiting patients and obtaining informed consent in accordance with the Health Insurance Portability and Accountability Act (HIPAA) and local/individual Institutional Review Board regulations. All TBIMS ND patients are recruited into the study during their inpatient rehabilitation stay (or at the time of IRF admission). After obtaining informed consent, data are collected from the patients using Form I of the data collection forms consisting of variables outlined in Table 1[6]. Broadly, these variables provide background information on patient demographics, their pre-morbid risk factors, clinical conditions at emergency department and/or inpatient rehabilitation admission as well as functional status and disability levels at inpatient rehabilitation admission and discharge.
The follow-up data are collected at the intervals of 1, 2, 5 years post-injury and every five years thereafter using Form II[7]. The follow-up in year 1 occurs within a four-month window – 10-to-14 months after the date of the injury; for year 2 this occurs in a six-month window – 21-to-27 months after the date of the injury; and for years 5 and every five years thereafter, it occurs within 6 months before or after the anniversary of injury. The follow-up is conducted by phone, in-person, or mail depending upon the availability of the participant. The TBIMS Standard Operating Protocol 105b[8] provides additional details on the process of data collection during the follow-up.
The TBIMS ND draws upon various sources of information. Specifically there are two data collection forms: I and II. Form I collects pre-injury data using The Pre-Injury History Questionnaire. This questionnaire collects socio-demographic characteristics, work and school participation, previous history of functional impairments or health conditions, and activity limitations prior to the injury. Form I also collects information from brain imaging studies, medical record data, neuropsychological tests, assessment of activity limitations and health status, as well as ICD-9 codes from acute hospital discharge records. Form II collects information on similar data elements as Form I with the exception of the brain imaging studies and neuropsychological assessments, but also includes several longer term outcome measures at intervals of 1, 2 and 5 years post-injury and every 5 years thereafter. Data, collected in real-time or retrospectively at each TBIMS center, are entered into the live web-based data management system, and are archived quarterly in TBIMS ND for the TBI MSND at the TBIMS National Data and Statistical Center.
A detailed description of the procedure for requesting access to the TBIMS ND is described within the TBI National Data and Statistical Center’s standard operating procedure number 602d [9]. This process requires the completion of a Data Request and User Agreement Form[10]. In addition to institutional affiliations, the individual requesting the data is required to provide a brief summary of the proposed project including study aims, research hypotheses, and methods. This information is reviewed by the TBIMS National Data and Statistical Center and TBIMS Research Committee for PI’s affiliation, scientific purpose, and scientific overlap with existing approved projects. Following initial review, the request is posted to the TBIMS Notification Listserv to ensure that the proposal does not duplicate already completed or proposed studies, to solicit collaborators if requestor is interested in collaborating with existing projects, and to invite further comments from TBIMS Project Directors. If the request is approved, data files are made available to the requestor in SAS, SPSS, or tab-delimited text formats.
Key variables from Forms I and II of the TBIMS ND are discussed in the following sections. The ICF framework will be used to describe these variables in order to provide conceptual continuity of the data elements. The TBIMS National Data and Statistical Center provides a complete online data dictionary, "TBIMS MS National Database Syllabus", that can be found at https://www.tbindsc.org/Syllabus.aspx. The syllabus provides a listing of all variables, variable names and their corresponding response values. Further, the syllabus also documents any change in the variables and their response categories over time.
These variables include personal characteristics that may impact the rehabilitation outcomes of people with TBI.
Participant Socio-Demographic Characteristics. The TBIMS ND has intake information (i.e., Form I) for 10,288 patients across the 20 TBIMS Centers from 1990-2010[11]. Three-fourths of the population were male, and more than two-thirds were Caucasians. Minority groups consisted primarily of African Americans (20%) and Hispanics (9%). Nearly half of the patients were never married and most were living independently (98%) at the time of injury. Some transition can be observed in living status between pre-injury and post-rehabilitation discharge. It can be observed in Table 3, that about 2% were living in other than private residences at the time of injury and 17% were discharged to other than private residences at discharge, indicating the impact of injury on independent living for TBIMS ND participants. About one-third of the patients were 16 – 25 years old;[12] 34% were 26 – 45 years old; 23% were 46 – 65 years old; and 12% were more than 65 years old at the time of injury. Nearly one third of the patients (29%) had post-secondary education experiences[13] (i.e., either they were working towards or had completed a post-secondary degree); 28% had completed high school or had a GED; and 22%had less than a high school degree.
Table 3. Transition in residence status pre-injury and post-rehabilitation discharge
|
|
At Injury |
At Discharge |
|
Private Home |
97.9 |
83.5 |
|
Institutional Setting |
0.8 |
15.8 |
|
Homeless |
0.8 |
0.1 |
This group of variables includes environmental factors that may impact the rehabilitation outcomes of patients.
Cause of Injury. Twenty one known causes of injury are identified by this variable[14]. These categories can be collapsed to eight: Assault (includes self-inflicted gunshot wounds), Fall, Motor Vehicle Accidents (MVA), Sports Injury, Striking Injury, Other, and Unknown. Of these groupings, the top three causes of injury include MVA (53%), Fall (23%), and Assault (13%). The trends in cause of injury by year of injury indicate an overall increasing trend in MVA up to the year 2000 and then this trend decreases from 2000 through 2010 (see Figure 2.a). The proportion of fall-related injuries increased over time, especially after year 2000. Further, trends vary by age groups, where trends for Falls increase with the increase in age group categories and trends for MVA decrease with an increase in age group categories (Figure 2. b). Trends for other causes of injury remain relatively stable across age group categories.
Figure 2.a. Trends in Causes of Injury for TBIMS ND individuals: 1990 through 2010
The figure is titled ‘Trends in Causes of Injury for TBIMS ND individuals: 1990 through 2010’. There are five lines in the chart, each representing the cause of injury leading to TBI. The blue line represents the category ‘Assaults’; red represents ‘Fall’; green represents ‘MVA’; lavender represents ‘Sports’; and light blue represents ‘Striking Injury’ category.
The horizontal axis is labeled ‘Year of Injury’ ranging from the year 1990 through 2010 and the vertical axis is labeled ‘Percentage’ ranging from 0 – 100 percent.
In the graph, the percentage of TBI due to Assaults begins at 30 percent in 1990, peaking to about 45 percent in 1991. This line then fluctuates between 20 – 30 percent for the years 1993 through 1997 and then declines and stabilizes at about 10 percent between the years 1998 through 2010.
The percentage of TBI due to Falls begins at 10 percent in 1990, increasing gradually to 30 percent through the year 2010. Prominent spikes to 20 percent and 25 percent are observed for the years 1993 and 1997.
The percentage of TBI due to MVA begins at about 55 percent in 1990 and dips to about 40 percent in the years 1991 and 1992. The trend increases over time peaking in the year 2000 to 70 percentage. After the year 2000, this line shows a declining trend ending at 50 percent mark for 2010.
The percentage of TBI due to Sports begins at near 0 percent mark in 1990 with only minor increase reaching to about 1 percent in 2010.
The percentage of TBI due to Striking Injury begins at 5 percentage mark in 1990, increasing to 12 percent mark in 1991 then reducing to 10 percent mark in 1992. This line gradually increases touching 20 percent mark in 1994 through 1997, and again reducing to 10 percent-level in 1998. This line remains at this level, without large variations through the year 2010.
End Figure 2.a
Figure 2. b. Trends in cause of injury by age category for TBIMS ND
This figure is titled ‘Trends in cause of injury by age category for TBIMS ND.’ There are five lines in the chart, each representing the cause of injury leading to TBI. The blue line represents the category ‘Assaults’; red represents ‘Fall’; green represents ‘MVA’; lavender represents ‘Sports’; and light blue represents ‘Striking Injury’ category.
The horizontal axis is labeled ‘Age Category’ that includes age-groups 16-25, 26 – 45, 46-65, and 65 and greater. The vertical axis is labeled ‘Percentage’ ranging from 0 – 80 percent.
In the graph, the percentage of TBI due to Assaults begins at about 10 percent for 16-25 year olds, increasing to about 18 percent for 26 – 45 year old. This line then shows a declining trend for the remaining age groups accounting for about 15 percent of TBI among 46 – 65 year old and 4 percent among 65 years and older.
The percentage of TBI due to fall begins at about 7 percent for 16-25 year olds and demonstrates an increasing trend across all the remaining age groups. The trend increases to 15 percent for 26-45 year old, about 33 percent for 46-65 year old and spikes to 70 percent for 65 years and older individuals.
The percentage of TBI due to MVA begins at about 70 percent for 16-25 year olds and demonstrates a declining trend across all the remaining age groups. The trend decreases to about 50 percent for 26-45 year old, 40 percent for 46-65 year old, and 20 percent for 65 years and older individuals.
The percentage of TBI due to sports begins at about 3 percent for 16-25 year old and shows a declining trend to 1 percent for 26-45 year old. This line then plateaus at this level for the rest of age groups.
The percentage of TBI due to striking injury begins at about 6 percent mark for individuals in the 16-25 year age group, spikes to about 10 percent for 26-45 year old and 46-65 year old individuals. This line then dips to about 6 percent for 65 years and older individuals.
End Figure 2.b
Payer Source for Acute Hospital care and Inpatient Rehabilitation. This variable indicates the primary and secondary payer sources paying for acute care and inpatient rehabilitation services for the participant. This information is collected from the hospital billing/business office[15]. This variable was re-categorized in 2011 and four codes were removed. The dataset provided for this analysis reflected the most recent codes for this variable. More than half of the individuals indicated their primary source as private insurance; nearly 37% indicated public insurance as their primary source, and about 6% indicated workers compensation as their primary source of care for acute hospital care and inpatient rehabilitation.
Pre-injury incarceration. Information on previous history of any penal incarcerations with conviction for felony was also collected. About 9% of the patients had a history of incarceration. This variable was added in 1997, however data prior to 1997 are available for some cases as the centers were encouraged to collect data retrospectively.
Inpatient rehabilitation Length of stay (LOS). Inpatient rehabilitation LOS is a calculated variable from admission to discharge from the inpatient rehabilitation facilities. Inpatient rehabilitation LOS is indicated as a marker of underlying brain injury pathology and health status (Arango-Lasparilla et al., 2010). The mean LOS in rehabilitation for TBIMS ND patients is 27.3 days (95% CI: 26.8 – 27.8). This varies substantially by the year of injury where individuals with injuries in years earlier than 1995 had about 40 days of rehabilitation stay; individuals with injuries between 1995 through 2002 had about 30 days of rehabilitation stay; and individuals with injuries after 2002 had about 25 days of rehabilitation stay (Figure 3).
Figure 3. Length Of Stay in Rehabilitation by Year of Injury: 1990 – 2010
This line graph is titled "LOS in Rehabilitation by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the mean Length Of Stay (LOS), in number of days, in inpatient rehabilitation.
In the graph, the mean LOS begins at about 50 days in 1990 and dips to less than 40 days in 1991. This line picks up to cross 40 days in 1993 and again declines to less than 40 days in 1994 and 30 days in 1995. The line then stabilizes at about 30 days level for the rest of the years.
End Figure 3
These variables indicate impairments in body structure and functions that may impact rehabilitation outcomes among TBIMS ND patients.
Pre-injury conditions/limitations. This variable was added in 2005 and is based on questions from the long form of the 2000 Decennial Census. Pre-injury conditions visual impairment, hearing impairment, or conditions that limit physical functioning such as walking, climbing stairs, reaching, lifting or carrying. The pre-injury limitations contain information about difficulty in the last 6 months in: (a) learning, remembering or concentration; (b) dressing, bathing, or getting around inside the home; (c) going outside home alone to shop or visit doctor’s office; and (d) working at a job or business. Another set of questions inquire about the pre-injury psychiatric history and specifically asks if the participant: (a) received treatment for mental health problems, (b) was hospitalized for a psychiatric problem, or (c) ever attempted suicide. This set of questions in combination with the pre-injury conditions and limitations could be used to identify participants who had disabilities prior to incurring a TBI. Utilizing this approach, 29% of TBIMS ND participants indicate a prior disability, with most indicating mental health and physical disabilities (7% each) followed by sensory and cognitive disabilities (6% each). About 2% indicate work-related disabilities before injury. Users must exercise caution as having a pre-injury condition does not indicate that a person had a disability. Information on activity limitation and participation limitations is important in determining if a person has a disability based on the ICF framework.
Intracranial CT diagnosis. Extensive information on intracranial CT diagnosis is available for each participant in the TBIMS ND. These data are collected from the CT scans performed within seven days of injury. Only trained TBIMS personnel (in doing CT scans) are authorized to complete this section of the Form I data, Further, it is advisable to use specific brain pathology-related information in determining if an individual had a CT scan-related information rather than the global item variable (CT information available – yes/no) as the latter was added only in 2007 and it could be misleading if one were to examine the global variable only. Information is available on the extent of intracranial compression, presence of intracranial hemorrhage/contusions, subarachnoid hemorrhage, intraventricular hemorrhage, focal cortical/non-cortical parenchymal contusions/hemorrhage, presence of any extra-axial collection, and presence of intraparenchymal fragments. In addition to the intracranial CT diagnosis, information is also collected on the presence of Intracranial Hypertension using data from intracranial pressure (ICP) monitors. However, 43% of the data for this field is marked "unknown" due to patients not having an ICP monitor, limiting its role in evaluating the prevalence of intracranial hypertension.
Pre-injury Alcohol or Drug use. The set of questions about an individual’s alcohol consumption are modeled after the Centers for Disease Control and Prevention’s Behavioral Risk Factor Surveillance System. These variables were added to the database in 1997. A comprehensive indicator of drug and alcohol problem use is established in the TBIMS ND through a set of items on alcohol use and one item about drug use. It is important to note that these are self-reported variables. Based on this constructed variable, nearly 38% of individuals had problems with alcohol or drug use within one year prior to the date of injury. A trend is noted, especially after year 1998, in the proportion of individuals who report pre-injury alcohol or drug use, as indicated in Figure 4.
Figure 4. Trends in Problem Alcohol or Drug Use among TBIMS ND Patients: 1990 – 2010
This line graph is titled "Trends in Problem Alcohol or Drug Use among TBIMS ND Patients: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the percentage of TBIMS ND patients with alcohol or drug use problem.
This line begins at about 25 percent mark for the year 1990; increases to about 30 percent in 1992 and dips below 20 percent in 1993. This line then fluctuates between 20 to 30 percent mark for the years 1994 to 1998. This line then shows an upward trend in 1999 and increases to reach 40 percent in 1999 and fluctuates at this level for the remainder of the years (i.e., 2000 through 2010).
End Figure 4
Measures of severity of injury and loss of cognitive functions. These include a battery of instruments meant to evaluate severity of injury as well as the extent of loss of cognitive or higher-order brain functions following TBI. These specifically include: (a) Glasgow Coma Scale (GCS), and (b) Neuropsychological battery. The GCS is administered in the Emergency Department while conducting a post-injury assessment. Neuropsychological batteries, administered between 2 – 4 weeks post injury in rehabilitation facility, are used to identify and monitor cognitive function loss in patients with TBI. Sub-groups of the TBI population can be constructed based on the severity of loss of cognitive function and injury using these metrics. Further, GCS is correlated with long-term outcomes (Zafonte et al., 1996; Cowen et al., 1995).
Duration of Post-traumatic amnesia (PTA). This variable captures the duration of time from the date of injury to the date on which the patient regained orientation and memory. PTA has been used to classify severity of TBI (Rees, 2003) and has been used as a predictor of long-term functional outcomes (Zafonte et al., 1997). The mean duration of PTA was 24.4 days (95% CI: 23.9 – 24.9). When examined across the year of injury, an overall declining trend in the length of PTA can be observed from 1990 through 2000, with a sharp increase in 1993. The trend stabilizes between 2000 and 2010 (Figure 5)[16].
Figure 5. Duration of Post-Traumatic Amnesia by Year of Injury: 1990 – 2010
This line graph is titled "Duration of PTA by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the mean length of post-traumatic amnesia (PTA) in number of days.
This line begins at about 40 days in 1990, dips to about 30 days for the years 1992 and 1993, picks back up to 40 days in 1993. This line then again declines to 30 days in 1994 and dips below 30 days mark in 1995. The line then fluctuates at this level for years 1996 through 1999 and dips to 20 days mark in 2001. This line then fluctuates between 20 and 25 days for the remainder years.
End Figure 5.
Among those with brain injury after 2003, nearly 25% had cranial surgeries. More than half of patients with Assault injuries receive cranial surgery (52%), followed by one-third of patients with Fall and Sports injuries (31% and 32%, respectively). Patients with MVA were least likely to have cranial surgeries (17%).
These variables document and measure the extent of activity limitations and participation restrictions in inpatient rehabilitation settings for TBIMS ND patients. Two variables/scales comprise the variables in this domain, the Disability Rating Scale (DRS) and Functional Independence Measure (FIMTM)[17].
Disability Rating Scale (DRS). In brief, the Disability Rating Scale was developed to monitor and track level of disability of individuals with TBI from injury through rehabilitation into community-based settings (Rappaport et al., 1982). The DRS is an eight item scale measuring impairment (3 items), cognitive ability (3 items), level of functioning (1 item), and employability (1 item). The DRS has been studied extensively for its predictive validity and its sensitivity to various severities of TBI (Rao & Kilgore, 1992). Limitations of DRS are discussed elsewhere (Hammond et al., 2001). It is important to note that higher DRS scores indicate greater disability and lower scores (i.e., zero) indicate close to normal function.
Mean DRS score at admission was 12.4 (95% CI: 12.3 – 12.5) and at discharge the mean was 6.3 (95% CI: 6.2 – 6.4). The difference in DRS scores between admission and discharge remains stable for the most part (except for year 1992) across years of injury (Figure 6). Variation in DRS values (as well as other measures) across key demographic variables are provided in Table 4.
Figure 6. Mean Disability Rating Scale scores at Admission and Discharge by Year of Injury: 1990 – 2010
This graph is titled "Mean DRS scores at Admission and Discharge by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the average DRS scores. This is a line graph with three lines; the blue colored line represents mean DRS score at admission; red color line represents DRS score at discharge; and the green color line represents the difference in DRS score between admission and discharge.
In the graph, the DRS score at admission begins at about 12 points in 1990, rising to 15 points in 1992 and declining to 13 points in 1995. This further declines to 12 point-level in 1998 and fluctuates at this mark till 2003 with a slight increase in 2004 to 13 point-level. This line again declines to 12 point mark for the remainder of the years.
The DRS at discharge begins at about 6 point mark in 1990 and increases to 7 points in 1994. The line slightly dips for 1995 and 1996 and picks back up to 7 points-level in 1997. The line then fluctuates slightly at this level for the remainder of the years.
The difference in DRS scores begins at about 7 points mark in 1990 spiking to 9 points in 1992 and dipping back to 7 points in 1995. This line slightly increases again to 8 points in 1996 and fluctuates between 6 and 7 points for the remainder of the years.
End Figure 6.
Functional Independence Measure (FIMTM). The FIMTM is an instrument developed for planning and monitoring inpatient rehabilitation services and outcomes related to functional independence (Wright, 2000). This 18 item instrument measures cognitive (5 items) and motor functioning (13 items). Each item is rated on a scale of 1 to 7, with 1 indicating "total assistance" and 7 indicating "complete independence" in performing tasks. Though the ability of the FIMTM to predict long-term outcomes continues to be a matter of further research (Whitlock and Hamilton, 1995), several studies have shown the predictive ability of the FIMTM on return to employment one to three years post-injury (Webb et al., 1995). It is important to note that the TBIMS ND dataset consists of individual item-level as well as summation scores across all of the cognitive and motor domains of the FIMTM and also provides the FIMTM total score.
The mean FIMTM Cognitive score at admission was 15.5 (95% CI: 15.4 – 15.7) and at discharge was 23.9 (23.8 – 24.0). The FIMTM Motor score at admission was 36.2 (35.8 – 36.5) and at discharge was 67.8 (67.4 – 68.1). The difference in FIMTM Motor score between admission and discharge is much higher compared to FIMTM Cognitive score. The mean FIMTM total score at admission was 51.7 (95% CI: 51.2 – 52.2) and at discharge was 91.8 (95% CI: 91.3 – 92.2). Figures 7 a, b, and c, illustrate variations in FIM scores across cohorts of individuals injured in a particular year. Variation in FIMTM scores across key demographic variables is provided in Table 4.
Table 4.
Functional Independence MeasureTM and Disability Rating Scale scores across Key Variables
|
Key Variables |
FIMTM Cognitive (admission) |
FIMTM Cognitive (discharge) |
FIMTM Motor (admission) |
FIMTM Motor (discharge) |
DRS (Admission) |
DRS (Discharge) |
|
Female |
16.2 |
24.4 |
35.0 |
65.4 |
12.2 |
6.4 |
|
Male |
15.3 |
23.8 |
36.9 |
68.6 |
12.5 |
6.3 |
|
|
|
|
|
|
|
|
|
White |
15.6 |
24.1 |
36.2 |
67.7 |
12.3 |
6.2 |
|
Black |
15.3 |
23.1 |
37.0 |
67.1 |
12.6 |
6.7 |
|
Asian/Pacific Islander |
16.3 |
25.2 |
38.0 |
68.8 |
11.9 |
6.0 |
|
Native American |
17.7 |
26.0 |
41.9 |
72.6 |
11.4 |
5.8 |
|
Hispanic |
15.1 |
23.7 |
35.3 |
68.3 |
12.5 |
6.2 |
|
Other |
16.0 |
25.0 |
38.6 |
71.1 |
12.4 |
5.6 |
|
|
|
|
|
|
|
|
|
16 – 25 |
15.3 |
24.5 |
36.5 |
69.3 |
12.7 |
6.2 |
|
26 - 45 |
15.4 |
24.1 |
37.8 |
69.7 |
12.4 |
6.1 |
|
46 - 65 |
15.4 |
23.4 |
35.7 |
67.1 |
12.3 |
6.5 |
|
65 > |
16.4 |
22.9 |
33.6 |
59.2 |
11.8 |
7.1 |
|
|
|
|
|
|
|
|
|
< HS Diploma |
15.2 |
23.2 |
35.4 |
66.4 |
12.6 |
6.7 |
|
GED |
14.7 |
23.5 |
34.7 |
67.3 |
12.6 |
6.5 |
|
HS Diploma |
15.6 |
23.7 |
35.5 |
66.3 |
12.4 |
6.5 |
|
College |
15.8 |
24.5 |
37.5 |
69.4 |
12.2 |
6.0 |
|
|
|
|
|
|
|
|
|
Assault |
15.1 |
23.2 |
40.3 |
70.8 |
12.4 |
6.3 |
|
Fall |
15.8 |
23.4 |
37.1 |
66.3 |
11.8 |
6.5 |
|
MVA |
15.5 |
24.3 |
35.2 |
67.6 |
12.6 |
6.2 |
|
Sports |
16.5 |
25.8 |
42.8 |
75.2 |
11.3 |
5.4 |
|
Striking Injury |
15.5 |
23.7 |
34.1 |
66.1 |
12.6 |
6.6 |
Figure 7.a. Functional Independence Measure (FIM)TM Cognitive scores at Admission and Discharge by Year of Injury: 1990 – 2010
The graph is titled "FIMTM Cognitive scores at Admission and Discharge by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the average FIM cognitive score. This is a line graph with three lines; mean FIM cognitive score at admission; FIM cognitive score at discharge; and the difference in FIM cognitive score between admission and discharge.
In the graph, the FIM cognitive scores begin at 15 points in 1990, increasing to 16 points in 1995 and dipping back to 15 point-level for years 1996 and 1997. This line then shows an increasing trend with a spike to 17 point-level in 2001and declines gradually to end at 14 points in 2010.
The FIM cognitive scores at discharge begin at 25 points in 1990, dips to about 22 point in 1991 and 1992 and picks back up to 25 points in 1995. This line fluctuates between the 22 and 25 points for the rest of the years.
The difference in the FIM cognitive scores between admission and discharge begin at about 10 points in 1990. The line then dips below the 10 points-level in 1991 and 1992 and picks up in 1993. This line then fluctuates between 8 and 10 points for the remainder years.
End Figure 7.a.
Figure 7.b. FIMTM Motor scores at Admission and Discharge by Year of Injury: 1990 – 2010
The graph is titled "FIMTM Motor scores at Admission and Discharge by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the average FIM motor score. This is a line graph with three lines; mean FIM motor score at admission; FIM motor score at discharge; and the difference in FIM motor score between admission and discharge.
The FIM motor scores at discharge begin at 75 points in 1990 and increases to 80 in 1992. This dips to about 75 points for the years 1993 through 1996. The line dips below 70 points in 1997 and picks up to stabilize the 70 points-level for the remaining years.
The FIM motor scores at admission begins at 40 points in 1990, spikes to 45 points in 1992, and fluctuates between 35 and 40 points for the rest of the years.
The difference in FIM scores at admission and discharge varies between 35 – 40 points for all the years.
End Figure 7.b
Figure 7.c. Functional Independence MeasureTM Total scores at Admission and Discharge by Year of Injury: 1990 -2010
This graph is titled "FIMTM Total scores at Admission and Discharge by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and the vertical axis represents the average FIM total score. This is a line graph with three lines; mean FIM total score at admission; FIM total score at discharge; and the difference in FIM total score between admission and discharge.
The FIM total score at admission begins at 55 points in 1990 and fluctuates between 55-60 points for all the years. The line dips slightly to 50 points for the years 1993 and 1996.
The FIM total score at discharge begins at 100 points in 1990 and shows a gradual declining trend to end at about 90 points in 2010.
The difference in FIM total scores between admission and discharge begins at 50 points in 1990 and dips to 40 points in 1991. This picks up to 50 points in 1993 and dips to below 40 points in 1997. The line stays at this level for years 1997 to 2003 and picks up to 40 points-level for the remainder of the years.
End Figure 7.c
The difference in FIM total scores between admission and discharge begins at 50 points in 1990 and dips to 40 points in 1991. This picks up to 50 points in 1993 and dips to below 40 points in 1997. The line stays at this level for years 1997 to 2003 and picks up to 40 points-level for the remainder of the years.
Employment and Earnings at Injury. The TBIMS ND collects information on employment status of patients in the month prior to the date of injury. Prior to 1994, three employment variables were collected (i.e., primary, secondary and tertiary). In 1994, the tertiary employment status was dropped and the current dataset has information on primary and secondary employment status for patients in the month prior to injury. Specific categories of responses for employment status variables are provided in the syllabus[18]. Employment, besides indicating financial capital, indicates the extent of physical activities and productive engagement prior to the injury. Employment prior to injury has also been demonstrated to predict post-injury rehabilitation outcomes (Crepaeu & Scherzer, 1993; Shames et al., 2007).
In addition to the employment status, the TBIMS ND collects information on the average number of hours of paid work (minimum wages and above) across all jobs as well as specific occupation categories (the major occupational category using 1990 Census Occupation Codes) for primary employment. Annual earnings[19] from all jobs and the number of weeks of competitive employment for the year prior to the injury were added to the data base beginning in 2001.
In the TBIMS ND data, 51% of individuals were employed (includes individual who were competitively employed, volunteered, looking for work and special employment), 6% were unemployed, and 5% were students one month prior to injury. Data were missing for nearly one quarter of individuals. Among those competitively employed, 78% worked full-time and 17% worked part-time one month prior to injury. Also, among those competitively employed, the majority (60%) earned less than $40,000 a year prior to injury. Earnings data were missing for 11% of individuals who were competitively employed. This variable was added to the TBIMS ND in 2001.
Form II is utilized for collecting follow-up data on rehabilitation outcomes for TBIMS ND patients at 1, 2, and 5 years post-injury and every 5 years thereafter as long as the TBIMS center is funded to do so. Table 5 provides the percentage of TBIMS ND participants having follow-up information across all follow-up time-points to date where either the person with TBI or their proxies (typically family members) were interviewed. Initial examinations indicate that follow-up data are available for a large proportion of TBIMS ND participants 5 years post-injury (ranging from 80 – 60%). However, the number of missing records increases with each follow-up time-point due in part to sample attrition and as data for later years for many patients are yet to be collected. As is evident in Table 7, researchers may have the ability to conduct research on longitudinal follow-up 5 to 10 years post rehabilitation discharge with a reasonable sample size. However, the heterogeneity in missingness of the specific outcome variable could potentially limit the overall power of the study to detect differences or test hypotheses.
Table 5.
Percentage of Follow-up Records available by Year of Injury
|
Injury Year |
N |
Y1 |
Y2 |
Y5 |
Y10 |
Y15 |
Y20 |
|
1990 |
93 |
60.6 |
42.6 |
58.5 |
73.4 |
67 |
59.6 |
|
1991 |
105 |
54.3 |
57.1 |
64.8 |
70.5 |
76.2 |
15.2 |
|
1992 |
126 |
47.6 |
51.6 |
64.3 |
69.8 |
66.7 |
0 |
|
1993 |
97 |
63.9 |
65 |
62.9 |
85.6 |
58.8 |
0 |
|
1994 |
97 |
61.2 |
65.3 |
60.2 |
70.4 |
59.2 |
0 |
|
1995 |
140 |
69.5 |
63.1 |
70.9 |
68.1 |
60.3 |
0 |
|
1996 |
149 |
52.4 |
50.3 |
72.5 |
73.2 |
14.1 |
0 |
|
1997 |
134 |
64.2 |
59.7 |
78.4 |
67.2 |
0 |
0 |
|
1998 |
237 |
80.2 |
74.3 |
84.4 |
69.6 |
0 |
0 |
|
1999 |
577 |
84.1 |
80 |
80.2 |
70.2 |
0 |
0 |
|
2000 |
592 |
83.1 |
81.9 |
80.2 |
59.6 |
0 |
0 |
|
2001 |
674 |
83.2 |
90.1 |
80 |
16.5 |
0 |
0 |
|
2002 |
559 |
94.1 |
83.2 |
80.1 |
0 |
0 |
0 |
|
2003 |
737 |
83.6 |
78.7 |
78.7 |
0 |
0 |
0 |
|
2004 |
768 |
82.2 |
81 |
77.1 |
0 |
0 |
0 |
|
2005 |
811 |
81.5 |
80.3 |
63 |
0 |
0 |
0 |
|
2006 |
872 |
85 |
81.2 |
20 |
0 |
0 |
0 |
|
2007 |
863 |
86.2 |
84.6 |
0 |
0 |
0 |
0 |
|
2008 |
872 |
86.2 |
74.4 |
0 |
0 |
0 |
0 |
|
2009 |
890 |
81.8 |
13.5 |
0 |
0 |
0 |
0 |
|
2010 |
802 |
12.1 |
0 |
0 |
0 |
0 |
0 |
It must also be noted that 926 patients died during the course of the follow-up and the percentage of deaths reported across the follow-up years indicates a higher percentage of deaths in the later follow-up time points (Figure 8.a.). One can also observe that proportion of deaths varies between 2% to 4% for most cohorts and increases markedly for those injured in more recent years. This variation could be attributable to higher risk of deaths in post acute phase of TBI (Figure 8.b.). It must be noted that the percentage values displayed in the charts represent average trends across the follow-up timepoints and do not represent individual-level changes. The higher proportion of deaths in the Y20 cohort as well as those in 2010 are primarily due to the overall lower sample size of the populations (N = 141 and 110 respectively).
Figure 8.a. Percentage of Deaths across Follow-up Time Points
This graph is titled "Percentage of Deaths across Follow-up Time Points." The horizontal axis represents the follow-up years of data collection and vertical axis represents the percentage of deaths.
The percentage of deaths is about 3 percent in Year 1 of follow-up (Y1) and spikes to 5 percent in Y5 and dips below 2 percent in Y9. This line spikes again to reach 8 percent in Y10 and dips to 2 percent in Y12. This line then rises steeply for Y13 to 12 percent and ends at 16 percent mark in Y20.
End Figure 8.a
Figure 8.b. Percentage of Deaths by Year of Injury: 1990 – 2010
This graph is titled "Percentage of Deaths by Year of Injury: 1990 – 2010." The horizontal axis represents the year of TBI injury and vertical axis represents percentage of deaths.
The percentage of deaths begins at about 3 percent in 1990, rising to 3.5 percent in 1992 and 4 percent in 1994. This line then dips to 3 percent in 1995 and rises to 5 percent in 1996. This line oscillates between 3 and 5 percent from 1997 to 2002 with a slight dip in 2003. The trend then increases to reach 9 percent in 2010.
End Figure 8.b
These variables include any change in an individual’s personal characteristics that may have an impact on rehabilitation outcomes. Only those variables with a likelihood of changing/varying over time are presented in the following sections.
Marital status. Figure 9 presents trends in the change of marital status of the TBIMS ND follow-up participants compared with their pre-injury status. In the first year of follow-up, more than one-third of individuals indicated that they were divorced; nearly one-quarter were widowed; 11 % were divorced and married and about 3% to 4% indicated they were divorced, widowed, and married (events in no specific order). These personal-level changes in family structure could have an impact on long-term rehabilitation outcomes (Bay, Blow, & Yan, 2011).
Figure 9. Marital Status across Follow-up Time Points
This graph is titled "Marital Status across Follow-up Time Points." The horizontal axis represents the follow-up year and vertical axis represents percentage of individuals with different marital status.
This is a column chart comprising of six set of columns each representing a marital status; No Change in marital status; divorced status; widowed status; divorced and married status; widowed and married status; and divorced, married and widowed status.
The column representing percentage of individuals without any change in marital status begins at about 20 percent in Y1 and increases to 25 percent mark in Y2. This trend peaks to 40 percent mark in Y15 and declines to 30 percent in Y20.
The column representing percentage of individuals divorced begins at about 35 percent mark in Y1 and shows a declining trend reaching to 20 percent in Y20.
The column representing percentage of individuals widowed begins at about 25 percent declining across the follow-up years to reach at 18 percent in Y15 and rises back up to 25 percent in Y20.
The column representing percentage of individuals divorced and married begins at 11 percent increasing over time to reach at 20 percent in Y20.
The column representing percentage of individuals widowed and married begins at about 3 percent in Y1 and fluctuates at that level across the different follow-up years.
The yellow column representing percentage of individuals divorced, married or widowed begins at 2 percent in Y1, increasing to 4 percent in Y2 and dips to 3 percent in Y5. A declining trend is observed for this column for the rest of the follow-up years.
End Figure 9.
These variables are comprised of indicators of loss or improvement in specific body structure and/or functions that impact long-term recovery and rehabilitation outcomes for individuals with TBI These include self-rated measures of psychiatric symptoms as well as alcohol and substance use history.
Patient Health Questionnaire 9 (PHQ-9) and Generalized Anxiety Disorder – 7 (GAD-7). PHQ-9 is a nine-item self-rated depression screening instrument. Each item in PHQ-9 is rated on a scale from 0 – 3. The scores can be used to indicate minimal, mild, moderate, moderately severe and severe depression (for details see Kroenke, Spitzer, & Williams, 2001). This variable was added to the TBIMS ND data base in 2007.
The Generalized Anxiety Disorder – 7 (GAD-7) is a seven-item self-rated questionnaire developed to identify Generalized Anxiety Disorders (Spitzer et al., 2006). The GAD-7 scores can be used to indicate minimal, mild, moderate, and sever anxiety disorders. This variable was added to the TBIMS ND database in 2010. Figures 10.a. and 10.b. provide trends in depression and generalized anxiety disorders for TBIMS ND patients across the follow-up years.
Figure 10.a. Prevalence of Depression by Follow-up Time Points
This graph is titled "Prevalence of Depression by Follow-up Time Points." The horizontal axis represents the year of follow-up and vertical axis represents percentage of individuals with depression.
This is a column graph with five sets of columns. Blue column represents minimal depression; red color represents mild depression; green color represents moderate depression; purple color represents moderately severe depression; and light blue color represents severe depression.
Percentage of minimal depression begins at 35 percent in Y1, reducing to 30 percent in Y5 and increasing up to 40 percent in Y20.
Percentage of mild depression begins at 12 percent in Y1 reducing slightly to 10 percent in Y5 and increasing to 20 percent in Y20.
Percentage of moderate depression begins at 7 percent in Y1, reducing to 5 percent for the remainder of the follow-up years.
Percentage of moderately severe depression begins at 3 percent in Y1and fluctuates between 3 to 4 percent for the remainder of the follow-up years ending at 5 percent in Y20.
Percentage of severe depression begins at 2 percent in Y1 and remains at about this level till Y10. This then dips to 1 percent in Y15 and Y20.
End Figure 10.a.
Figure 10.b. Prevalence of Generalized Anxiety Disorder by Follow-up Time Points
This graph is titled "Prevalence of Generalized Anxiety Disorder by Follow-up Time Points." The horizontal axis represents the follow-up years and the vertical axis represents the percentage of individuals with anxiety.
This is a column chart with four columns: minimal anxiety; mild; moderate and severe anxiety.
The column representing minimal anxiety begins at 6 percent in Y1 and increases to 7 and 9 percent for Y5 and Y10 respectively. This dips to 8 percent in Y15 and increases up to 10 percent in Y20.
The column representing mild anxiety begins at 2 percent in Y1 and oscillates at this level for Y2, and Y5. This increases to 3 percent in Y10 and ends at 12 percent in Y20.
The column representing moderate anxiety begins at less than 1 percent in Y1, increasing gradually to end at 2 percent in Y20.
The column representing severe anxiety begins at less than 1 percent in Y1, increasing gradually to end at 2 percent in Y20.
End Figure 10.b.
History of Drug and Alcohol Use. Figure 11 displays the prevalence of drug and alcohol use among the TBIMS ND patients. A moderately increasing trend in prevalence of drug or alcohol use can be observed across the follow-up years. The observed increase in the trends for Y20 could be also due to lower Ns.
Figure 11. Prevalence of Drug/Alcohol Use across Follow-up Time Points
This graph is titled "Prevalence of Drug/Alcohol Use across Follow-up Time Points." This is a column graph with horizontal axis representing the follow-up years and the vertical axis representing the percentage of individuals with drug/alcohol use.
The column beings at 15 percent in Y1, and increases across the follow-up years to reach 17 percent in Y10. This slightly dips to 15 percent in Y15 and picks up to 20 percent in Y20.
End Figure 11.
Re-hospitalization. The TBIMS ND also collects information on re-hospitalization in the year prior to the follow-up time point of data collection (prior to 2004 the timeframe was "since the last evaluation" instead of "last year"). The prevalence of re-hospitalization decreases over the follow-up period until the 10th year post-injury, when the prevalence of re-hospitalizations increases before trending downward again after the 15th year (Figure 12).
Figure 12. Prevalence of Re-hospitalization by Follow-up Time Points
This graph is titled "Prevalence of Re-hospitalization by Follow-up Time Points." This is a bar graph with horizontal axis representing the follow-up time and vertical axis representing percentage of individuals with re-hospitalization.
The column begins at 25 percent in Y1 and demonstrates a declining trend across the follow-up years to end at 12 percent in Y20.
End Figure 12.
These variables include the standardized assessment of activity limitation for the TBIMS ND patients using the Disability Rating Scale, Functional Independence Measure, Glasgow Outcome Scale – Extended, and Supervision Rating Scale. These also include variables indicating restrictions in participation in home, community, school, work and leisure activities among TBIMS ND patients.
Disability Rating Scale (DRS), Glasgow Outcome Scale – Extended (GOS-E), and Supervision Rating Scale (SRS). The DRS, described within the Form I section, is an eight-item instrument that helps in tracking the improvement in rehabilitation outcomes. Lower scores indicate less disability, while higher scores indicate greater disability (DRS scores range from 0 – 29). The GOS-E, is an eight-level measure intended to measure functional recovery among TBI patients (Jennett, Snoek, & Bond, 1981; Pettigrew, Wilson, & Teasdale, 1998). Contrary to DRS, lower scores in GOS-E indicate worse outcomes compared to higher scores (GOS-E scores range from 1 – 8). The SRS is a 13-level measure that bases measurement of functional recovery on the extent of supervision one may need (SRS scores range from 1 – 13). A higher score indicates dependence and lower scores indicating independence (Boake, 1996).
Figure 13 illustrates the mean DRS, GOS-E and SRS scores across the follow-up years for the TBIMS ND patients. Changes in the DRS, GOS-E, and SRS scores remain comparatively stable across the follow-up years with a slightly declining trend.
Figure 13. DRS, GOS-E and SRS Scores across Follow-up Time Points
This graph is titled "DRS, GOS-E and SRS Scores across Follow-up Time Points." This is a line graph with horizontal axis representing the follow-up years and vertical axis representing mean scores for DRS, GOS-E, and SRS scores.
The mean DRS scores line begins at 3 in Y1 with a declining trend; reaching 2 percent in Y10. This line again increases to about 3 in Y20.
The GOS-E line begins at the mean score of 5, increasing to 6 for Y2 and Y5 and decreases to 5 for the remainder of the follow-up years.
The SRS mean score line begins at about 4 in Y1 and shows a declining trend ending at 3 in Y20.
End Figure 13.
Functional Independence Measure (FIMTM). The FIMTM (described within the Form I section) is another instrument measuring functional recovery based on activities of daily living among TBIMS ND patients. Figure 14 illustrates the patterns in mean FIMTM Total scores.
Figure 14. FIMTM Total Scores across Follow-up Time Points
This graph is titled "FIMTM Total Scores across Follow-up Time Points." This is a line graph with horizontal axis representing follow-up years and vertical axis representing mean total FIM scores.
The line begins at about 113 in Y1 and increases across the follow-up years to reach at 117 in Y10. This line dips slightly and ends at 116 in Y20.
End Figure 14.
PART-O. This 18-item instrument was developed by the Participation Special Interest Group within TBIMS to understand participation restrictions using telephone interview methodology. This instrument is a combination of other instruments measuring participation in TBI (Whiteneck et al., 2011). The TBIMS ND provides mean scores across the domains of productivity, social relations and community involvement (also referred to as being "out and about"). Figure 15 illustrates the mean scores across the three domains of the PART-O and displays the summary score. In general, TBIMS ND patients had high scores for the social domain, followed by "out and about," and productivity. It is important to note that this variable was introduced in the dataset in 2007. A general increasing trend exists across all of the domains in participation which indicates an improved situation for TBIMS patients measured at the successive time points. There is a drop in participation scores between year 10 and year 15, with an increase in year 20.
Figure 15. Participation Scores by Follow-up Time Points
This graph is titled "Participation Scores by Follow-up Time Points." This is a line graph with horizontal axis representing year of follow-up and vertical axis representing mean scores.
There are four lines in this graph –the mean scores for participation out & about, scores for social, productivity, and summary scores.
The mean scores for participation out & about begins at 1.5 in Y1. This increases steadily across the follow-up years reaching to about 2 in Y20.
Scores for social begin at 2 in Y1, dips slightly in Y15 and ends at 2 in Y20.
Productivity begins at 1 in Y1 and increases steadily to 1.5 in Y10. This line dips slightly to 1 in Y15 and ends at 1.5 in Y20.
Summary scores begin at 2 in Y1, declines to 1.5 in Y2. This line increases to about 2 in Y10 again dipping slightly in Y15 and ends at 1.5 in Y20.
End Figure 15.
Productive Engagement. This variable measures productive engagement (i.e., participation in work and/or education) for TBIMS ND participants. We observe a declining trend in employment (includes competitive employment, special employment, unemployed looking for work, volunteering, and on unpaid leave from work) across all the follow-up time points with the exception of 20 years post-injury and trends in retirement increase substantially at 15 and 20 years post-injury (Figure 16). It is important to note that data were missing for about 15-35% of individuals in each follow-up year and, therefore, should be used with caution.
Figure 16. Productive Engagement across Follow-up Time Points
This graph is titled "Productive Engagement across Follow-up Time Points." This a column graph with horizontal axis representing the year of follow-up and vertical axis representing percentage engaged productively.
This graph has three set of columns –employed, student, and retired.
Employed begins at 53 percent in Y1 and decreases across the follow-up years to 40 percent in Y15. This ends at 42 percent in Y20.
Student begins at 5 percent in Y1. This decreases to 1 percent in Y10, Y15 and Y20.
Retired begins at 25 percent in Y1 and remains at this level across all the follow-up years. This ends at 28 percent in Y20.
End Figure 16.
Annual Earnings. This variable collects information on annual income from competitive employment across all jobs during the previous 12 months from the date of interview for TBIMS ND patients. This variable was added to the data collection system in 2001. In preliminary analysis, 75 - 80% of the data were missing, despite restricting the sample to follow-up data collected after 2001. This lack of data may be due to the fact that earnings are only reported for those who are competitively employed. No results of preliminary analysis are reported here and users are cautioned about the extent to which this variable is missing data.
Satisfaction With Life Scale (SWLS). Developed by Diener and colleagues in 1985, the SWLS is an instrument used to measure one’s satisfaction with life based on one’s comparison of one’s own life circumstances to self-established standards. Pavot and Diener (1993) provide a full review of psychometric properties of SWLS. Satisfaction with life is not only a marker of subjective sense of recovery among individuals with TBI (Cicerone et al., 2004), but also forms a key construct of an individual’s quality of life after TBI (Dijkers, 2004). The SWLS was added to the TBIMS ND in 1998.
Figure 18 illustrates trends in the SWLS for TBIMS NDC participants during the follow-up years. The trends in SWLS have remained consistent across the time points around the average score of 20 - a neutral point where respondents are equally satisfied or dissatisfied.
Figure 17. Satisfaction With Life Scale Mean Scores by Follow-up Time Points
This graph is titled "Satisfaction With Life Scale Mean Scores by Follow-up Time Points." This is a line chart with horizontal axis representing follow-up years and vertical axis representing mean satisfaction with life scores.
This line begins at 20 percent in Y1 and spikes slightly at Y20 to 23.
End Figure 17.
TBIMS ND offers a rich array of follow-up data for researchers to track short-term and long-term clinical, psychosocial, and functional outcomes post-injury. However, appropriate caution must be exercised before its usage and interpretation. Users are encouraged to review this User’s Guide to consider the scope of their research questions. Further, they must refer to the TBIMS ND online syllabus document (https://www.tbindsc.org/Syllabus.aspx) for specific information on variables and to understand changes in the patterns of response overtime. Users are encouraged to contact the TBI Model Systems National Data and Statistical Center and to seek their feedback on the ability and feasibility of using the TBIMS ND to investigate their research questions.
References
Arango-Lasprilla, J. C., Ketchum, J. M., Cifu, D., Hammond, F., Castilo, C., Nicholls, E., Wantnabe, T., & Lequerica, A. (2010). Predictors of extended rehabilitation length of stay after traumatic brain injury. Archives of Physical Medicine & Rehabilitation, 91, 1495 – 1504.
Bay, E. H., Blow, A. J., & Yan, X. (2011). Interpersonal relatedness and psychological functioning following traumatic brain injury: Implications for marital and family therapists. Journal of Marital & Family Therapy, 38 (3), 556 – 567.
Boake, C. (1996). Supervisory Rating Scale: A measure of functional outcome from brain injury. Archives of Physical Medicine & Rehabilitation, 77, 765 – 772.
Cicerone, K. D., Mott, T., Azulay, J., & Freil, J. C. (2004). Community integration and satisfaction with functioning after intensive cognitive rehabilitation after traumatic brain injury. Archives of Physical Medicine & Rehabilitation, 85 (6), 943 – 950.
Corrigan, J., D., Cuthbert, J., P., Whiteneck, G., G., Dijkers, M., P., Coronado, V., Heinemann, A., W., Harrison-Felix, C., & Graham, J., E.(2012). Representativeness of the Traumatic Brain Injury Model Systems National Database. Journal of Head Trauma Rehabilitation, 27(6), 391 – 403.
Cowen, T. D., Meythaler, J. M., DeVivo, M. J., Ivie, C. S. 3rd, Lebow, J., & Novack, T. A. (1995) Influence of early variables in traumatic brain injury on functional independence measure scores and rehabilitation length of stay and charges. Archives of Physical Medicine & Rehabilitation, 76, 797-803.
Crepaeu, F. & Scherzer, P. (1993). Predictors and indicators of work status after traumatic brain injury: A meta-analysis. Neuropsychological Rehabilitation, 3 (1), 5 – 35.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction With Life Scale, Journal of Personality Assessment, 49, 71 – 75.
Dijkers, M. P. (2004). Quality of life after traumatic brain injury: A review of research approaches and findings. Archives of Physical Medicine & Rehabilitation, 85 (S2), 21 – 35.
Hammond, F. M., Grattan, K . D., Sasser, H., Corrigan, J. D., Bushnik, T., & Zafonte, R. D. (2001). Long-term recovery course after traumatic brain injury: A comparison of the Functional Independence Measure and Disability Rating Scale. Journal of Head Trauma & Rehabilitation, 16 (4), 318 – 329.
Jennett, B., Snoek, J., & Bond, M. R. (1981). Disability after severe head injury: Observations on the use of the Glasgow Outcomes Scale. Journal of Neurology, Neurosurgery & Psychiatry, 44, 285 – 293.
Kreonke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16 (9), 606 – 613.
Pavot, W. & Diner, E. (1993). Review of the Satisfaction With Life Scale. Psychology Assessment, 5 (2), 164 – 172.
Pettigrew, L.E., Wilson, J.T., & Teasdale, G.M. (1998). Assessing disability after head injury: improved use of the Glasgow Outcome Scale. Journal of Neurosurgery, 6, 939–943.
Rao, N., Kilgore, K. M. (1992) Predicting return to work in traumatic brain injury using assessment scales. Archives of Physical Medicine & Rehabilitation, 73, 911–916.
Rappaport, M., Hall, K. M., Hopkins, K., Belleza, T., & Cope, D. N. (1982) Disability rating scale for severe head trauma: Coma to community. Archives of Physical Medicine & Rehabilitation, 63, 118–123.
Rees, P. M. (2003). Contemporary issues in mild traumatic brain injury. Archives of Physical Medicine & Rehabilitation, 84 (12), 1885 – 1894.
Shames, J., Treger, I., Ring, H., & Giaquinto, S. (2007). Return to work following traumatic brain injury: Trends and challenges. Disability and Rehabilitation, 29 (17), 1387 – 1395.
Spitzer, R. L., Kroenke, K., Williams, J. B. & Lowe, B. (2006). A brief measure for assessing generalized anxiety disorder. Archives of Internal Medicine, 166, 1092 – 1097.
Webb, C. R., Wrigley, M., Yoels, W., Fine, P. R. (1995) Explaining quality of life for persons with traumatic brain injuries 2 years after injury. Archives of Physical Medicine & Rehabilitation, 76, 1113–1119.
Whiteneck, G. G., Dijkers, M. P., Heinemann, A. W., Bogner, J. A., Bushnik, T. , Cicerone, K. D., . . . Mills, S. R. (2011). Development of participation assessment with recombined tools – Objective for use after traumatic brain injury. Archives of Physical Medicine & Rehabilitation, 92, 542 – 551.
Whitlock, J. A., & Hamilton, B. B. (1995) Functional outcome after rehabilitation for severe traumatic brain injury. Archives of Physical Medicine & Rehabilitation,76, 1103–1112.
Wright, J. (2000). The FIM ™. The Center for Outcome Measurement in Brain Injury. http://www.tbims.org/FIM (accessed on January 30, 2013).
Zafonte, R. D., Hammond, F. M., Mann, N. R., Black, K. L., & Mills, S. R. (1996). Relationship between Glasgow coma scale and functional outcome. American Journal of Physical Medicine & Rehabilitation, 75 (5), 364 – 369.
Zafonte, R. D., Mann, N. R., Mills, S. R., Black, K. L., Wood, D. L., & Hammond, F. (1997) Posttraumatic amnesia: Its relation to functional outcome. Archives of Physical Medicine & Rehabilitation, 78 (10), 1103 – 1106.
Zafonte, R., Ricker, J., Lobmard, L., Mann, N., & Black, K. (1996). Does the requirement of craniotomy predict outcomes? A preliminary investigation. Brain Injury, 13 (1), 31 – 38.
[1] An interagency agreement between the VA and NIDRR continues to support the development, implementation and management of a mirror database to the TBIMS ND in which the 4 VA Polytrauma Rehabilitation Centers capture similar data in order to support comparative studies of civilian and military populations with TBI.
[2] For a list of TBIMS centers that are currently funded, please visit https://www.tbindsc.org/Centers.aspx.
[3] Definition for TBI, including specific inclusion and exclusion criteria can be downloaded from : https://www.tbindsc.org/SOP/101a%20-%20Identification%20of%20Subjects.pdf
[4] Moderate to severe TBI is defined as a person with TBI having post-traumatic amnesia for greater than 24 hours, loss of consciousness of 30 minutes or greater, intracranial abnormalities in brain imaging studies, or a value of 13 or less on the Glasgow Coma Scale at the time of admission in the emergency department.
[5] Detailed guidelines for obtaining informed consent can be found at: https://www.tbindsc.org/SOP/102a%20-%20Guidelines%20and%20Strategies%20for%20National%20Database%20Recruitment%20and%20Consent.pdf
[6] Please see https://www.tbindsc.org/SOP.aspx for Form I questionnaire items
[7] Please see https://www.tbindsc.org/SOP.aspx for Form II questionnaire items.
[8] Please see https://www.tbindsc.org/SOP.aspx for additional information.
[9] Please visit https://www.tbindsc.org/SOP/602d%20-%20External%20Use%20TBIMS%20National%20Database%20Notification.pdf for additional information.
[10] Please visit https://www.tbindsc.org/SOP/602df%20-%20External%20Data%20Request%20and%20Use%20Agreement%20Form.pdf for additional details.
[11] Note that TBIMS ND includes data from 1988 to present; only data from 1990 to 2010 were analyzed for the purpose of this user’s guide.
[12] It is important to note that the dataset provides users with calculated age at injury and it is therefore possible for users to meaningfully categorize age.
[13] It is important to note that the dataset collects information on the number of years of education up to high school degree and has categories for higher education where individuals can specifically indicate if they were working towards a degree program or have already attained a post-secondary education degree.
[14] See https://www.tbindsc.org/SyllabusDetail.aspx?MOD=1&ID=CSEINJ for detailed codes for cause of injury. Also, note that codes 10, 11, & 12 were used to create "Assault" Category; code 19 was designated "Fall" Category; codes 1, 2, 3, 4, & 5 were grouped for "MVA" Category; and codes 13, 14, 15, 16, 17 & 18 were used to create "Sports" Category.
[15] See https://www.tbindsc.org/SyllabusDetail.aspx?MOD=1&ID=PAY for detailed categories of Payor Source.
[16] Note that while computing the mean PTA one needs to set the values 999 and 888 to missing. Results could be erroneous without this minor data manipulation. Further, PTA is tracked only until inpatient rehabilitation discharge and therefore has a truncated distribution. One can potentially address this by calculating PTA as duration from injury to rehabilitation discharge for those cases which were coded 888=still in PTA at rehabilitation discharge.
[17] Users are urged to review the TBIMS ND syllabus https://www.tbindsc.org/Syllabus.aspx to ensure that they set the values such as 66, 77, 888, 999 as missing to avoid errors in computing the averages. The data set does not follow a consistent approach for indicating missing values for variables. Missing values differ across variables and also depend on the variable length (e.g. in some cases, is , and 88 is used instead of 888, etc.)
[18] See https://www.tbindsc.org/SyllabusDetail.aspx?MOD=1&ID=EMP for employment status categories.
[19] Annual earnings were reported as categories.