Volume 46 Number 8, 2009
Pages 1003 — 1010
Abstract — In an attempt to increase understanding regarding the nonacute healthcare needs of veterans with traumatic brain injury (TBI), we examined the outpatient utilization and cost patterns of 72 patients with TBI who were at least 4 years postinjury. We selected participants from a clinical database of veterans receiving care at a western Department of Veterans Affairs (VA) medical center. We extracted data from national utilization databases maintained by the VA and examined data from primary care and internal medicine, psychiatry and substance use, rehabilitation, and other services (e.g., ancillary, diagnostic, prosthetic, dental, nursing home, and home care). We extracted data for fiscal years 2002 to 2007. In addition to descriptive statistics, we modeled visits per year as a function of time since injury. The data show that this sample of patients with TBI consistently used a wide array of outpatient services over time with considerable variation in cost. Further study regarding economic aspects of care for patients with TBI is warranted.
Key words: aging, costs, Department of Veterans Affairs, healthcare, nonacute care, outpatient services, rehabilitation, traumatic brain injury, utilization, veterans.
Planning for the healthcare needs of veterans with traumatic brain injury (TBI) is a difficult task that requires knowledge of short- and long-term outcomes. While a modest amount of information is available regarding the former, less is known about the long-term outcomes. Highlighting the need for research in this area is the number of TBIs sustained in the current military conflicts , as well as the growing population of individuals aging with TBI . In an attempt to understand the nonacute healthcare needs of veterans with TBI, we examined the outpatient utilization and cost patterns of 72 patients with TBI who were at least 4 years postinjury. This descriptive study focuses on veterans who are beyond the acute phase of recovery.
The number of individuals with TBI is increasing, and social systems are grappling with how to support such individuals as they age . Two-thirds of those who sustain a TBI before age 30 will likely live for another 30 to 40 years . Lending support to this finding is research by Harrison-Felix et al., who demonstrated that sustaining a TBI resulted in only a 7-year reduction of life expectancy . While the effects of aging with TBI are not yet clearly understood, research suggests the need for ongoing, long-term follow-up care.
Whereas a modest literature exists on the utilization and costs of acute TBI care [4-6], long-term follow-up data are scarce. Cifu et al. analyzed a group of individuals with moderate to severe TBI and found that half of all hospital admissions 3 years postinjury were for orthopedic surgery, reconstructive surgery, or both . These surgeries were performed to reduce pain, enhance function, and improve appearance . Of the hospitalizations for psychiatric purposes, the majority were for depression, while other reasons included behavioral dyscontrol and substance use . At 5 years postinjury, hospitalization for orthopedic and reconstructive surgery declined while the use of specialized mental health care (e.g., psychiatric hospitalization) increased . Hodgkinson et al. noted similar findings and found that patients with TBI who were 6 months to 4 years postinjury used services related to restoration of function, while those between 6 and 17 years postinjury used services related to life changes, such as the loss of a relationship or employment . With regard to cost, most published data have been collected to understand the acute phase of care, and a strength of several such studies is that they have adjusted for severity of injury. One study found that costs increase as TBI severity increases , while another found an inverse relationship between cost and injury severity . An early study determined that patients with moderate TBI required the most extensive and costly resources, postulating that the least injured did not require intensive services and the most severely injured did not survive long enough to consume extensive resources .
To date, literature focused on veterans has noted the importance of studying the economics of TBI. These studies have suggested strategies such as determining the most advantageous geographic placement of future Department of Veterans Affairs (VA) TBI treatment units, using telemedicine to meet the demand of TBI-related services, and establishing a TBI database across the VA and Department of Defense that would allow researchers to study the effects of TBI severity and psychiatric comorbidities on healthcare utilization [13-15]. This article furthers research in this area by describing healthcare utilization and cost for veterans with TBI 4 to 40 years postinjury, taking into account age, years since injury, and severity of TBI.
We selected participants from a clinical database of 408 veterans receiving care at a western VA medical center (VAMC). We confirmed history of TBI through chart reviews using criteria set forth by the Centers for Disease Control and Prevention . We assessed injury severity (mild, moderate, or severe) by a continuing medical education program developed by the VA . We used the following specific criteria: mild TBI-altered or loss of consciousness (LOC) of <30 minutes with negative imaging, Glasgow Coma Scale (GCS) of 13 to 15, or posttraumatic amnesia (PTA) of <24 hours; moderate TBI-LOC of <6 hours with positive imaging, GCS of 9 to 12, or PTA of <7 days; and severe TBI-LOC >6 hours with abnormal imaging, GCS of <9, or PTA of >7 days .
Of this highly diverse group of 408 veterans, we excluded 12 who died and 73 who had neurologic disorders (e.g., stroke, anoxic brain injury, spinal cord injury, multiple sclerosis, Parkinson disease, Alzheimer disease) in addition to their history of TBI. Because we did not have access to non-VA healthcare utilization, we excluded patients with <1 VA visit of any type per year since the beginning of the VA electronic medical record system in 1998 (131 individuals). We also excluded patients <4 years postinjury (116 individuals) and patients who had multiple TBIs identified within the medical record (4 individuals). The remaining 72 veterans comprised the analytical sample.
For our sample, we extracted data from national utilization databases (National Patient Care Database and Patient Treatment Files) maintained by the VA. We used these databases to identify the outpatient clinic stop. We used the method created by Yu et al.  and Phibbs et al.  to categorize these codes into primary care and internal medicine, psychiatry and substance use, rehabilitation, and other services (e.g., ancillary, diagnostic, prosthetic, dental, nursing home, and home care). We used Health Economics Resource Center (HERC) Average Cost data instead of Decision Support System data; the two sources are highly correlated when annualized, and HERC provides an annualized data set that is easy to use . We extracted data for fiscal years 2002 to 2007, resulting in a median of 6 years (range: 4-6) of data per patient. Although we extracted data from fiscal years 2002 to 2007, not all patients had 6 complete years of data (eight had 5 years of data and one had 4 years). The HERC Average Cost files provided the cost data , and we inflated all costs to 2007 dollars .
In addition to descriptive statistics, we used regression models to estimate the mean number of visits per year for 4 to 40 years since injury while controlling for age. For each of four categories of outpatient healthcare utilization (primary care and internal medicine, psychiatric and substance use, rehabilitation, and other services), we ran models for all patients (n = 68), for those with mild TBI only (n = 17), and for those with moderate or severe TBI (n = 51). The regressions did not include four patients with moderate TBI >40 years postinjury, as too few data points were available to estimate a stable mean. We assumed that the mean number of visits per year would vary smoothly over time since injury (i.e., on average, utilization would not change drastically year-to-year). Therefore, we performed a natural cubic B-spline transformation  of the variable "years since injury" and used this as a predictor in the regression models. This method smooths the raw data and allows for nonlinear trajectories. Additionally, to estimate the mean utilization at any time point, this method incorporates information from surrounding time points, which is consistent with the assumption of smoothly varying utilization.
Of the 72 veterans, 60 were service connected (10%-100%) for various issues, including their history of brain injury. Sixteen veterans sustained their TBI during military service (between 1965 and 1993). Table 1 shows descriptive information for our sample. Most (76%) of our sample had either a moderate or severe TBI. Almost all the veterans were male.
Table 1.Characteristics of veterans with traumatic brain injury (TBI) Š4 years postinjury, fiscal years 2002 to 2007.
Characteristic All Subjects(n = 72) Mild TBI(n = 17) Moderate/Severe TBI(n = 55) Mean (Median) ± SD Range Mean (Median) ± SD Range Mean (Median) ± SD Range
Age* 46.6 (47.5) ± 11.2 26-73 48.5 (52.0) ± 8.3 33-59 46.1 (45.0) ± 11.9 26-73 Years Since Injury* 18.4 (15.0) ± 12.2 4-50 19.7 (17.0) ± 10.9 4-39 18.0 (13.0) ± 12.7 4-50 Male (n) 70 - 17 - 53 -
*As of 2001.SD = standard deviation.
Table 2 shows the estimated annual outpatient visits. These patients with TBI used a wide array of services. Primary care and internal medicine use for those with mild TBI was higher than for those with moderate or severe TBI. For psychiatry and substance use services as well as rehabilitation, those with moderate or severe TBI had higher utilization than those with mild TBI. With the exception of patients 10 years postinjury, those with mild TBI used more "other" healthcare services than patients with moderate or severe TBI.
Table 2.Estimated mean number of outpatient visits per year and 95% confidence intervals for veterans with traumatic brain injury (TBI) Š4 years postinjury, fiscal years 2002 to 2007.*
Healthcare Service Years Since Injury Mild TBI(n = 17) Moderate/Severe TBI(n = 51)
Primary Care and Internal Medicine 10 5.7 (3.2, 9.7) 4.3 (3.3, 5.5) 20 8.2 (5.8, 11.6) 6.7 (5.2, 8.6) 30 11.7 (7.6, 17.7) 6.2 (4.4, 8.6) 40 16.5 (8.4, 31.4) 4.4 (2.0, 8.7) Psychiatry and Substance Use 10 1.4 (0.4, 3.2) 4.1 (2.4, 6.8) 20 2.1 (0.9, 3.9) 2.5 (1.4, 4.2) 30 2.9 (0.9, 7.2) 3.3 (1.7, 5.7) 40 4.0 (0.6, 14.2) 5.9 (1.8, 16.1) Rehabilitation 10-40† 0.9 (0.5, 1.2) 1.3 (0.8, 2.0) Other‡ 10 4.9 (3.2, 7.2) 5.8 (4.3, 7.7) 20 8.1 (6.1, 10.7) 5.7 (4.6, 7.1) 30 13.0 (8.4, 19.8) 5.7 (3.9, 8.2) 40 20.6 (10.5, 39.6) 5.6 (3.0, 9.9)
*Estimates are from B-spline regression models, which assume that utilization varies smoothly over time since injury. All models controlled for age and were estimated at mean age for cohort (mild = 48.5 years, moderate/severe = 46.1 years)†Estimated mean utilization was constant over time, therefore estimates are same for all four time points (10, 20, 30, and 40 years postinjury).‡Excludes primary care and internal medicine, psychiatry and substance use, and rehabilitation.
Within severity levels at 10 and 20 years postinjury, patients with mild TBI used more primary care and internal medicine than all other services (for example, at 10 years postinjury, 5.7 annual visits to primary care and internal medicine vs 1.4 psychiatry and substance abuse visits, 0.9 rehabilitation visits, and 4.9 other healthcare service visits). Beyond 20 years postinjury, those with mild TBI used more "other" healthcare services than the remaining three healthcare categories. Patients with moderate or severe TBI used more primary care and internal medicine than psychiatry and substance use or rehabilitation services with the exception of the 40-year time point, when they used more psychiatry and substance use services than the remaining three healthcare categories. With regard to trends, all visits increased over time for patients with mild TBI, but the pattern for patients with moderate or severe TBI varied.
Table 3 shows outpatient costs to VA. In any given year, patients may have had visits or costs in just one category or in all four categories; therefore, not all sample size subtotals add to 72. The average visit cost across the four categories ranged from $604 to $1,800. Notably, costs varied considerably.
Table 3.Outpatient costs for veterans with traumatic brain injury (TBI) Š4 years postinjury, fiscal years 2002 to 2007 (2007 dollars).*
Healthcare Service Year n* Mean ± SD Median Range
Primary Care and Internal Medicine 2002 64 1,349 ± 1,138 1,018 30-4,018 2003 69 1,251 ± 1,103 932 115-6,821 2004 69 1,808 ± 1,827 951 170-10,068 2005 66 1,620 ± 1,420 1,071 171-8,184 2006 66 1,757 ± 1,932 1,120 17-10,322 2007 65 2,273 ± 3,079 1,383 144-20,758 Total No. - 10,058 - - Mean (Median) - 1,676 (1,689) - - Psychiatry and
2002 50 1,207 ± 1,259 905 80-7,593 2003 47 1,237 ± 1,141 1,035 12-5,291 2004 44 1,983 ± 1,925 1,247 15-8,134 2005 50 2,932 ± 3,650 1,408 32-18,687 2006 49 2,225 ± 2,962 1,012 13-11,760 2007 49 1,216 ± 1,313 760 11-6,388 Total No. - 10,800 - - Mean (Median) - 1,800 (1,610) - - Rehabilitation 2002 31 529 ± 517 397 19-2,625 2003 27 516 ± 637 294 50-3,235 2004 40 657 ± 788 362 34-3,058 2005 41 785 ± 1,519 337 27-7,801 2006 38 894 ± 1,909 266 29-8,903 2007 37 899 ± 1,505 414 22-6,976 Total No. - 4,280 - - Mean (Median) - 604 (721) - - Other† 2002 67 831 ± 907 593 11-4,769 2003 69 1,152 ± 1,478 655 13-7,991 2004 67 1,966 ± 3,130 760 23-21,971 2005 69 1,958 ± 2,786 871 54-16,066 2006 64 1,283 ± 1,230 1,117 22-4,949 2007 61 1,307 ± 1,387 756 3-7,520 Total No. - 8,497 - - Mean (Median) - 1,416 (1,295) - -
*Patients may have had visits or costs in just one category or all four categories, therefore not all subtotals add to 72.†Excludes primary care and internal medicine, psychiatry and substance use, and rehabilitation.SD = standard deviation.
Figures 1 to 4 display graphs of all subjects' outpatient visits. We modeled visits per year as smooth functions of time since injury. This method provides meaningful estimates by smoothing the raw data, using surrounding information to estimate utilization at any one time point, and staying consistent with the assumption that, on average, utilization will vary smoothly over time since injury. There was generally consistent use of all outpatient services over time.
Despite considerable variation in cost, healthcare use among veterans in this sample remained relatively constant many years postinjury. Estimates indicate that for this cohort, primary care and internal medicine utilization increased from approximately three visits per year to approximately seven visits per year, remained constant for 10 years, and then tailed off slightly. Hodgkinson et al. found that their sample of patients with TBI also used approximately seven primary care visits during a 12-month time frame . With the exception of the 10-year time point, patients with mild TBI used more visits (8.2-16.5) than the Hodgkinson et al. sample . This difference may not be unusual because the prevalence of chronic medical and psychological conditions in veterans receiving care at VAMCs is much higher than in the general U.S. population . Service connection and the use of automated pharmacy refills may also account for this difference. For the entire sample, the model estimated a constant mean utilization over time (10- to 40-year time points) of approximately four outpatient visits per year for psychiatry and substance use services and a constant mean of about two visits per year for rehabilitation services. Mean utilization of other healthcare services was estimated at approximately seven visits per year at 10 years postinjury, increasing to about nine visits per year at 40 years postinjury. No comparable published data exist with which to compare these findings.
Our results are descriptive but quite robust to outliers. Figures 1 to 4 show the mean visits per year from models using a B-spline method that assumes healthcare utilization will, on average, vary smoothly over time since injury and that is flexible enough to provide data-driven estimates. In all the analyses there were fluctua-tions over time, but use of services never tailed off completely.
Limitations to our study exist that warrant discussion. First, the data represent a single VAMC. Also, the final analytical sample comprised only 18 percent of the original sample of 408. Furthermore, some veterans sustained their injuries while in the military while others had civilian-acquired injuries, further complicating generalizability. Patients with TBI who had one or more healthcare visits per year may be different than those who did not use services as frequently. Additionally, some veterans may have had other forms of insurance (Medicare, Medicaid, or employer-based), and we did not observe their non-VA healthcare use. Therefore, for our analytical group, we selected patients who consistently used VA providers (i.e., at least one visit of any type per year). This does not mean that they did not use non-VA providers, but we wanted to ensure that our conclusions reflected what we observed (i.e., VA utilization).
Little knowledge exists regarding utilization and cost of healthcare services by patients with TBI who are many years postinjury. In light of this, we demonstrated that, despite variability in cost, healthcare use remains relatively constant over time for this group of veterans. While we can draw no conclusions regarding the marginal effect of TBI on utilization and cost of healthcare, this offers data up to 40 years postinjury, filling a gap in the literature since the only other published study examining long-term utilization after TBI provided data up to 17 years postinjury . Examining long-term healthcare utilization and associated costs allows VA healthcare professionals as well as those in administrative positions to plan for the sufficient allocation of financial and other resources for this group of patients. Future studies designed to longitudinally examine the patterns and trajectories of utilization and cost for this growing group of individuals are warranted.
Go to TOP
Last Reviewed or Updated Wednesday, March 31, 2010 10:05 AM