Chapter 2: Analysis of the Colorado Traumatic Brain Injury Registry
Introduction
The Colorado Traumatic Brain Injury Registry and Follow-Up System (CTBIRFS) was a joint surveillance activity of the Colorado Department of Public Health and Environment, Craig Hospital, and the Centers for Disease Control and Prevention. The purpose of the system was to identify outcomes associated with traumatic brain injury including quality of life, reintegration into the community, return to work and school, functional status, service use, and secondary complications. A random sample of traumatic brain injury cases was drawn from the State of Colorado’s Hospital Discharge File for 1966 through 1999. Patients selected were Colorado residents 16 and older, who had been discharged alive from a Colorado hospital from an acute care hospitalization with TBI. A complex questionnaire was developed which drew upon established interview scales and prior instruments. Sources of scales and questions that were incorporated into the interviews include, among others, the Craig Handicap Assessment and Reporting Technique (CHART); FIM score; Health Status Questionnaire (HSQ); and the Health-Related Quality of Life (SF-36) form. Patients selected for inclusion in the system (or their proxies) were interviewed one year after discharge and on subsequent anniversaries of their discharge.
The content of the survey changed somewhat over time. For example, shorter versions of established health scales were substituted for longer ones, allowing for continued calculation of important measures of health status and/or disability, though decreasing the length of successive surveys. Finally, some questions were not asked in the fourth year of follow-up.
Methods
The database contained a wealth of information directly relevant to its purpose of characterizing sequellae and recovery from TBI but was lacking in some aspects. For example, though broad cause-and-intent categorical codes were included in the database, external cause-of-injury codes were absent. As a consequence, we recoded the cause-and-intent data into the following categories: Motor Vehicle Crashes (includes motorcycle crashes); Assault or Suicide (non-firearm); Unintentional Falls; Other Unintentional (includes sports injuries); and Other and Unknown (includes firearms-related injuries).
Using data collected from 1990 to 2002, we examined the distribution of cases across the four follow-up years. As one of the goals of this report is to provide longitudinal information on health and disability trends in survivors of traumatic brain injury, we linked cases across each of the four follow-up years and identified a subset of patients who had been interviewed in each of the four years. As noted previously, some variables were only available for the first three years of the study. Basic demographic characteristics, case mix, and severity of the first year interviewees (N=1603) were compared with the group that had completed all four interviews (N=380). Tables 1 and 2 show the basic age distribution of initial and long-term interviewees. Over 60 percent of patients in the database were 16 to 44, and patients 65 or older accounted for approximately 15 percent of the cases.
Table 1. Age Distribution for Patients Completing the 1st Interview |
Table 2. Age Distribution for Patients Completing All Interviews |
389 |
24.3 |
587 |
36.6 |
345 |
21.5 |
270 |
16.8 |
12 |
0.7 |
1,603 |
100.0 |
|
95 |
25.0 |
143 |
37.6 |
86 |
22.6 |
56 |
14.7 |
0 |
0.0 |
380 |
100.0 |
|
Tables 3 and 4 show the distribution of males and females in the database. Relative to the initial interviewees, women represent a slightly higher proportion of the patients who completed all interviews.
Table 3. Distribution of Males and Females for Patients Completing the 1st Interview |
Table 4. Distribution of Males and Females for Patients Completing All Interviews |
1,028 |
64.1 |
563 |
35.1 |
12 |
0.8 |
1,603 |
100.0 |
|
235 |
61.8 |
145 |
38.2 |
380 |
100.0 |
|
Tables 5 and 6 show the breakdown by race of patients in the study. The patients completing the first and follow-up interviews were predominantly white. A somewhat greater proportion of minorities were represented among initial interviewees than among long-term interviewees.
Table 5. Breakdown by Race for Patients Completing the 1st Interview |
Table 6. Breakdown by Race for Patients Completing All Interviews |
17 |
1.1 |
36 |
2.2 |
8 |
0.5 |
1,335 |
83.3 |
2 |
0.1 |
205 |
12.8 |
1,603 |
100.0 |
|
4 |
1.1 |
9 |
2.4 |
3 |
0.8 |
341 |
89.7 |
0 |
0 |
23 |
6.1 |
380 |
100 |
|
Tables 7 and 8 show the distribution of injury cause for patients included in the study.
Table 7. Injury Cause Distribution for Patients Completing the 1st Interview |
Table 8. Injury Cause Distribution For Patients Completing All Interviews |
814 |
50.8 |
110 |
6.9 |
434 |
27.1 |
215 |
13.4 |
30 |
1.9 |
1,603 |
100.0 |
|
208 |
54.7 |
26 |
6.8 |
92 |
24.2 |
49 |
12.9 |
5 |
1.3 |
380 |
100.0 |
|
Motor vehicle crashes predominate, followed by unintentional falls, other unintentional, assault/suicide, and other and unknown. Neither the rankings nor relative proportions of the injury cause vary dramatically between first and long-term interviewees. Unfortunately the file does not differentiate motorcycle crashes from other motor vehicle crashes.
Finally, Tables 9 and 10 show average age and length of stay in days by injury cause.
Table 9. Average Age and Length of Stay (LOS) for Patients Completing the 1st Interview
814 |
35.1 |
7.8 |
110 |
35.1 |
6.1 |
434 |
59.1 |
6.1 |
215 |
37.6 |
4.0 |
18 |
34.6 |
13.6 |
12 |
N/A |
N/A |
Table 10. Average Age and Length of Stay for Patients Completing All Interviews
208 |
35.7 |
7.6 |
26 |
34.6 |
6.5 |
92 |
55.5 |
6.0 |
49 |
38.2 |
3.0 |
5 |
30.8 |
10.4 |
The most noticeable difference between first and long-term interviewees is the somewhat lower average age of patients discharged after unintentional falls. Otherwise, the two groups are quite comparable.
In conclusion, based on the tabular analysis of demographics, injury cause and length of stay, the subgroup of patients who completed all interviews is not substantially different from the initial interviewees.
Payer Source by Injury Category
These data also show primary payer by injury cause in table 11. “Other liability insurance” -- presumably automobile insurers — dominated the payers for motor vehicle crashes (51.4%) and “Self-pay,” which includes the uninsured, accounted for 14.4 percent of payers for motor vehicle crashes.
Table 11. Analysis of Payer Source by Injury Category – Number and (Percent)
7
(3.4) |
4
(15.4) |
2
(2.1) |
3
(6.1) |
0
(0) |
16
(4.2) |
11
(5.2) |
2
(7.7) |
22
(23.9) |
17
(34.7) |
2
(40.0) |
54
(14.2) |
107
(51.4) |
0
(0) |
0
(0) |
2
(4.1) |
0
(0) |
109
(28.8) |
6
(2.9) |
3
(11.5) |
24
(26.1) |
4
(8.2) |
0
(0) |
37
(9.7) |
2
(1.0) |
2
(7.7) |
2
(2.1) |
0
(0) |
0
(0) |
6
(1.6) |
5
(2.4) |
2
(7.7) |
15
(16.3) |
7
(14.3) |
0
(0) |
29
(7.6) |
7
(3.4) |
3
(11.5) |
18
(19.6) |
4
(8.2) |
0
(0) |
32
(8.4) |
1
(0.5) |
0
(0) |
1
(1.1) |
0
(0) |
0
(0) |
2
(0.5) |
0
(0) |
0
(0) |
0
(0) |
0
(0) |
1
(20.0) |
1
(0.3) |
30
(14.4) |
5
(19.2) |
4
(4.4) |
6
(12.2) |
0
(0) |
45
(11.8) |
0
(0) |
1
(3.9) |
0
(0) |
0
(0) |
0
(0) |
1
(0.3) |
32
(15.4) |
4
(15.4) |
4
(4.4) |
6
(12.2) |
2
(40.0) |
48
(12.6) |
208
(100) |
26
(100) |
92
(100) |
49
(100) |
5
(100) |
380
(100) |
Analysis of Outcomes, Long-Term Disability, and Recovery
We first examined reported head injury severity as a function of injury cause, presented here in table 12. Two findings are clear:
- Moderate head injuries represented a relatively greater proportion of cases (37%) among motor vehicle crash (MVC) patients than among other causes, and
- Critical and severe injuries represented a relatively greater proportion of cases (54.1%) among unintentional fall patients than among other causes.
Table 12. Analysis of Head Injury Severity by Injury Cause – Frequency and (Percent)
77
(37.0) |
4
(15.4) |
17
(18.5) |
9
(18.4) |
0
(0.0) |
107
(51.4) |
61
(29.3) |
10
(38.5) |
28
(30.4) |
14
(28.6) |
4
(1.9) |
117
(56.3) |
41
(19.7) |
7
(26.9) |
22
(23.9) |
20
(40.8) |
0
(0.0) |
90
(43.3) |
28
(13.5) |
5
(19.2) |
25
(27.2) |
6
(12.2) |
1
(0.5) |
65
(31.3) |
1
(0.5) |
0
(0.0) |
0
(0.0) |
0
(0.0) |
0
(0.0) |
1
(0.5) |
208
(100.0) |
26
(100.0) |
92
(100.0) |
49
(100.0) |
5
(100.0) |
380
(100.0) |
The database offers a large number of scale scores and individual questions about outcomes. We screened many of these measures and present representative findings.
The Craig Handicap Assessment and Reporting Technique was designed to quantify the extent of handicap in individuals. CHART consists of six dimensions, each with a maximum score of 100. The dimensions are occupation, cognitive independence, physical independence, mobility, social integration, and economic self-sufficiency. A total CHART score of 600 indicates no handicap at all. Table 13 displays available CHART occupation scores for each of the four one-year anniversary follow-ups. These scale scores measure participation in work and related matters. CHART occupational scores tended to be higher for those involved in motor vehicle crashes as compared to All Others in Years 1 and 2, though larger number of Missing or Not Asked in Years 3 and 4 preclude further interpretation.
Table 13. CHART Occupational Scores by Year and Injury Cause, Mean and (N), Perfect = 100
81.6
(198) |
76.6
(164) |
(18) |
86.0
(197) |
79.3
(163) |
(20) |
78.1
(138) |
70.7
(110) |
(132) |
75.3
(117) |
66.5
(94) |
(169) |
Table 14 presents CHART social integration scores. These scores, available only for the Year 1 and Year 2 follow-up, show that relative to patients involved in MVCs, all other patients were more socially isolated.
Table 14. CHART Social Integration Scores by Year and Injury Cause – Mean and (N)
85.8
(201) |
81.7
(?167 ?) |
86.1
(204) |
84.6
(?168?) |
Finally, Table 15 displays total CHART scores for the four years of follow-up. The analysis reveals few differences across the injury categories and the four years of follow-up. Scores declined in the third and fourth follow-up years for victims of falls, many of them elderly.
Table 15. Total CHART Scores by Year and Injury Cause = Mean and (N),
Perfect Health = 600
531.7
(188) |
527.0
(151) |
539.6
(187) |
537.0
(153) |
523.2
(132) |
509.0
(102) |
516.9
(109) |
495.00
(86) |
We compared FIM scores for motor and cognitive scales (Tables 16 and 17). These scale totals are a composite of a series of questions scaled from 1 (Total Assistance) to 7 (Complete Independence). The scores show little variation across injury cause and little variation across Years 1 and 2, the two years for which relatively complete data are available.
Table 16. FIM Motor Scores by Injury Cause for Years 1 and 2 – Mean and (N), Perfect = 91
88.2
(193) |
88.8
(156) |
88.0
(192) |
88.9
(158) |
Table 17. FIM Cognitive Scores by Injury Cause for Years 1 and 2 – Mean and N, Perfect = 35
32.1
(197) |
32.2
(162) |
31.9
(191) |
32.4
(158) |
Tables 18 and 19 display analyses of elements and total scores from the Health Status Questionnaire (HSQ). The HSQ is an outcomes measurement tool that yields scores on eight health attributes and can be used to measure the risk of a depressive disorder. The attributes are health perception, physical functioning, role limitations/physical health, role limitations/ emotional problems, social functioning, mental health, bodily pain, and energy/fatigue. HSQ physical function scores were somewhat lower for patients involved in motor vehicle crashes than for all others, as were HSQ mental health scores (Table 19). Scores for victims of motor vehicle crashes and Other and Unknown tended to be slightly lower in the first-year follow-up. Scores tended to increase in Year 2. HSQ total scores were lower for patients involved in motor vehicle crashes than for all others. Scores generally improved in Year 2 (Table 20).
Table 18. HSQ Physical Function Scores by Injury Cause for Years 1 and 2 – Mean and (N)
77.0
(206) |
81.0
(171) |
82.9
(208) |
85.4
(168) |
Table 19. HSQ Mental Health Scores by Injury Cause for Years 1 and 2 – Mean and (N)
62.7
(204) |
71.7
(169) |
82.9
(208) |
72.2
(169) |
Table 20. Total HSQ Scores by Injury Cause for Years 1 and 2 – Mean and (N)
339.1
(201) |
370.6
(169) |
357.0
(208) |
382.2
(165) |
Employment Impacts
We probed the impacts of head injury on employment by examining several sets of measures. Respondents were asked to report their current job hours worked at each of the four anniversaries of injury. Table 21 reports simple employment trends by year and injury cause. Overall reported levels of employment increased through Year 3 and declined slightly in Year 4 (Table 22).
Table 21. Jobs Hours Worked per Week by Injury Cause and Year – Mean and (N)
21.6 |
21.4 |
(204) |
(169) |
25.3 |
23.2 |
(202) |
(167) |
26.6 |
25.7 |
(207) |
(172) |
25.8 |
24.5 |
(208) |
(172) |
Table 22. Are You Employed? By Injury Cause and Year – Frequency and (Percent)
119 |
130 |
140 |
135 |
(57.5) |
(62.5) |
(67.6) |
(64.9) |
88 |
78 |
67 |
73 |
(42.5) |
(37.5) |
(32.4) |
(35.1) |
|
213 |
225 |
246 |
237 |
(56.3) |
(59.2) |
(64.9) |
(62.9) |
165 |
155 |
133 |
140 |
(43.7) |
(40.8) |
(35.1) |
(37.1) |
|
Table 23 displays the distribution of stated reasons for non-employment by victims at each of the anniversary follow-ups. The ‘Other’ category includes miscellaneous categories such as students and homemakers. This table suggests that unplanned or premature retirement may be associated with injury outcomes. Permanent or temporary disability and extended medical treatment also clearly prevent some victims from re-entering the workforce, even as much as five years after injury.
Table 23. Why Are You Not Working? By Injury Cause and Year – Frequency and (Percent)
17 |
19 |
19 |
22 |
(19.1) |
(24.7) |
(28.8) |
(30.6) |
9 |
4 |
1 |
1 |
(10.1) |
(5.2) |
(1.5) |
(1.4) |
18 |
17 |
22 |
25 |
(20.2) |
(22.1) |
(33.3) |
(34.7) |
45 |
37 |
24 |
24 |
(50.6) |
(48.1) |
(36.4) |
(33.3) |
89 |
77 |
66 |
72 |
(100.0) |
(100.0) |
(100.0) |
(100.0) |
|
54 |
61 |
57 |
67 |
(32.0) |
(40.1) |
(43.5) |
(47.9) |
16 |
8 |
3 |
1 |
(9.5) |
(5.3) |
(2.3) |
(0.7) |
27 |
29 |
40 |
40 |
(16.0) |
(19.1) |
(30.5) |
(28.6) |
72 |
54 |
31 |
32 |
(42.6) |
(35.5) |
(23.7) |
(22.9) |
169 |
152 |
131 |
140 |
(100.0) |
(100.0) |
(100.0) |
(100.0) |
|
Discussion
This chapter analyzes perhaps the most comprehensive disability outcome database for traumatic brain injury. The cases in the database are primarily white males 16 to 64; individuals 65 or older account for approximately 15 percent of the cases. The average initial hospital length of stay was greatest for motor vehicle crashes.
Head injuries of moderate severity were more commonly associated with motor vehicle crashes than with other injury categories. Life-threatening head injuries were most commonly associated with unintentional falls, though this may in part reflect the average older age of this group. One shortcoming of the analysis is the small cell frequencies for some injury categories; comparisons between causes should be interpreted with caution.
Interestingly, different scales revealed different views. For example, CHART scores tended to be higher for patients involved in motor vehicle crashes, whereas FIM scores were relatively similar and HSQ scores tended to be lower for the those patients.
Finally, our analyses suggest that although many TBI victims return to work after their injuries, permanent disability, lengthy temporary disability, or extended medical care prevents many others from returning to a productive life. (Whiteneck, Charlifue, Gerhart, Overholser, and Richardson, 1992)