OVER A MILLION PEOPLE sustain a traumatic brain

ORIGINAL ARTICLE Change in Inpatient Rehabilitation Admissions for Individuals With Traumatic Brain Injury After Implementation of the Medicare Inpatient Rehabilitation Facility Prospective Payment System Jeanne M. Hoffman, PhD, Elena Donoso Brown, MS, Leighton Chan, MD, MPH, Sureyya Dikmen, PhD, Nancy Temkin, PhD, Kathleen R. Bell, MD 1305 ABSTRACT. Hoffman JM, Donoso Brown E, Chan L, Dikmen S, Temkin N, Bell KR. Change in inpatient rehabilitation admissions for individuals with traumatic brain injury after implementation of the Medicare inpatient rehabilitation facility prospective payment system. Arch Phys Med Rehabil 2012;93: 1305-12. Objective: To evaluate the impact of Medicare s inpatient rehabilitation facility (IRF) prospective payment system (PPS) on use of inpatient rehabilitation for individuals with traumatic brain injury (TBI). Design: Retrospective cohort study of patients with TBI. Setting: One hundred twenty-three level I and II trauma centers across the U.S. who contributed data to the National Trauma Data Bank. Participants: Patients (N 135,842) with TBI and an Abbreviated Injury Score of the head of 2 or greater admitted to trauma centers between 1995 and 2004. Interventions: None. Main Outcome Measure: Discharge location: IRF, skilled nursing facility, home, and other hospitals. Results: Compared with inpatient rehabilitation admissions before IRF PPS came into effect, demographic characteristics of admitted patients changed. Those admitted to acute care trauma centers after PPS was enacted (January 2002) were older and nonwhite. No differences were found in rates of injury between men and women. Over time, there was a significant drop in the percent of patients being discharged to inpatient rehabilitation, which varied by region, but was found across all insurance types. In a logistic regression, after controlling for patient characteristics (age, sex, race), injury characteristics (cause, severity), insurance type, and facility, the From the Departments of Rehabilitation Medicine (Hoffman, Donoso Brown, Dikmen, Temkin, Bell), Neurological Surgery (Dikmen, Temkin), Psychiatry and Behavioral Medicine (Dikmen) and Biostatistics (Temkin), University of Washington, Seattle, WA; and Rehabilitation Medicine Department, Clinical Research Center, National Institutes of Health, Bethesda, MD (Chan). Presented in part to the State of the Science Symposium on Post Acute Rehabilitation: Setting and Agenda and Developing an Evidence Base for Practice and Public Policy, February 12-13, 2007, Washington, DC. Supported by the Department of Education, National Institute on Disability and Rehabilitation Research Traumatic Brain Injury Model System Funding (grant no. H133A070032); and resources from the intramural program of the National Institutes of Health. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated. Correspondence to Jeanne M. Hoffman, PhD, Dept of Rehabilitation Medicine, Box 356490, University of Washington, Seattle, WA 98195-6490, e-mail: jeanneh@u.washington.edu. Reprints are not available from the author. 0003-9993/12/9308-00039$36.00/0 http://dx.doi.org/10.1016/j.apmr.2012.04.030 odds of being discharged to an IRF after a TBI decreased 16% after Medicare s IRF PPS system was enacted. Conclusions: The enactment of the Medicare PPS appears to be associated with a reduction in the chance that patients receive inpatient rehabilitation treatment after a TBI. The impact of these changes on the cost, quality of care, and patient outcome is unknown and should be addressed in future studies. Key Words: Brain injuries; Medicare; Rehabilitation. 2012 by the American Congress of Rehabilitation Medicine OVER A MILLION PEOPLE sustain a traumatic brain injury (TBI) annually, with 275,000 of these people hospitalized. 1 Rehabilitation services provided early after injury have been found to decrease costs, length of stay, and improve functional outcomes for persons with TBI. 2 Changes in service delivery or reimbursement have the potential to impact functional outcomes for persons after TBI. 3 In 2002, inpatient rehabilitation facilities (IRFs) experienced a major change in their reimbursement system. As part of the Balanced Budget Act of 1997, a prospective payment system (PPS) was enacted for several forms of postacute care, including IRFs. 4 This initiative aimed to contain costs, as well as improve care by providing payment based on need, taking into account diagnosis, age, functional status, and comorbidities. Up until 2002, IRFs were reimbursed for the cost per average discharge. 5 Despite the initiative s intent to improve care, concerns regarding the PPS were raised, including: the potential for premature acute care discharge, decreased rehabilitation access, the variability of functional evaluations, and the poor accounting of comorbidities, affecting the calculations of reimbursement. 6 Since the implementation of the PPS for IRFs, researchers have begun to assess the impact of the PPS on inpatient rehabilitation services, and results appear to be mixed. For example, a study by Paddock et al 7 looked at changes in a case-mix among patients with stroke, hip fracture, and lower AIS GCS ICD-9 IRF ISS NTDB PPS SNF TBI List of Abbreviations Abbreviated Injury Scale Glasgow Coma Scale International Classification of Diseases inpatient rehabilitation facility Injury Severity Score National Trauma Data Bank prospective payment system skilled nursing facility traumatic brain injury

1306 PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman extremity joint replacement. Results suggested that there was no difference in the severity of patients admitted to IRFs preversus post-pps implementation. In addition, even without a change in the case-mix, IRFs in most studies have demonstrated a decrease in length of stay. 7-10 However, rather than lead to a decreased need for services, Zorowitz 11 notes that the focus on length of stay has increased discharges to skilled nursing facilities (SNFs) prior to the completion of rehabilitation, as well as increased the number of transfers between facilities during care. In addition to looking at overall service delivery outcomes, researchers have attempted to examine the impact on patient outcomes after PPS implementation. A PPS simulation study for stroke patients 12 suggested that while group therapy treatments might increase, no change in rehabilitation outcomes would be expected. This is consistent with the findings by Sood et al, 13 who found no change in outcomes when looking at broad measures like discharge to community. However, some studies have indicated that across a variety of rehabilitation diagnoses, patient outcomes have declined when looking at discharge FIM scores. 7,8,14 Despite reduced functional improvement, Shah et al 14 noted that inpatient rehabilitation patients from 4 facilities reported increased satisfaction. While these studies have examined changes that occurred within IRFs, few studies have examined how the PPS has impacted the use of inpatient rehabilitation care. Buntin et al 15 looked at the trends in all postacute care after implementation of the PPS. They found that the implementation of the IRF PPS resulted in decreased probability of the use of multiple sources of rehabilitation care, including SNF, home health, and inpatient rehabilitation. In specific patient populations, however, the changes varied. For example, after PPS implementation, those patients with joint replacement had a higher likelihood of SNF transfer, while those with hip fracture had a lower probability of SNF transfer but a higher probability of inpatient rehabilitation admission. To our knowledge, no studies have directly examined the impact that the enactment of the IRF PPS has had on use of rehabilitation care after TBI. Changes in care were anticipated given our previous study 3 that examined the potential impact prior to implementation. Results of that study suggested that access to inpatient rehabilitation for persons with TBI might be hindered because of decreased reimbursement and the financial needs of IRFs. We hypothesized that implementation of the IRF PPS would lead to reduced access (fewer admissions) to inpatient rehabilitation for individuals with TBI. METHODS Study Design The National Trauma Data Bank (NTDB), which contains information on nearly 1 million trauma patient admissions at 568 participating hospitals from 1994 through early 2005, was used for this study. We queried the NTDB sample for all patients with a TBI (based on reported International Classification of Diseases [ICD-9] codes 800 804 and 850 854), who were discharged to an IRF, SNF, other hospital, private residence, or other known location after their acute-care hospital stay. Only those hospitals that contributed a minimum number of patients (at least 100 prior to January 1, 2002, and 100 after) were chosen to ensure that any changes seen were not because of differences in individual hospital practices. The ICD-9 diagnoses were then converted to Abbreviated Injury Scale (AIS) scores and only individuals with an AIS head score of 2 were included. The AIS rates the severity of injuries to 9 different body regions from minor (1) to unsurvivable (6), with a moderate injury rated as 2. The resulting dataset consisted of 135,842 patient admissions from 123 different hospitals. This study is exempt from institutional review board approval, because the dataset does not contain any identifying information. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology statement. 16 Measures All measures are variables collected in the NTDB. Demographic and injury information. Age is presented as the mean and SD in the description of the sample. To test the impact of age on the discharge destination, age was categorized into 10 groups with 10-year increments from 0 to 9, 10 to 19, up to 80 and above. Race/ethnicity was divided into 6 categories: (1) white, (2) black, (3) Hispanic, (4) Asian/Pacific Islander, (5) Native American, and (6) other. The cause of TBI was derived from e-codes describing the mechanism or external factors that caused the trauma. These were grouped in 5 categories including: (1) motor vehicle collision, (2) fall, (3) assault, (4) other, and (5) unknown. The Glasgow Coma Scale (GCS) 17 score from emergency department admission was examined. The GCS is measured acutely after brain trauma to classify severity, and includes responsiveness in motor, verbal, and eye modalities. The scores range from 3 to 15, and a low score represents worse neurologic function. However, if the patient was sedated or paralyzed, interfering with the ability to obtain accurate GCS scores, a score could not be calculated. To examine the impact on discharge destination, GCS scores were categorized into mild (13 15), moderate (8 12), severe (3 7), and paralyzed. The AIS head score, as mentioned above, had to be greater than 2 (minimum to be included in the study). The 3 most severely injured body regions have their AIS score squared and added together to produce the Injury Severity Score (ISS). 18 In this study we specifically created the ISS using the 3 most severely injured body regions other than the head. To examine the impact on discharge destination, ISS nonhead scores were categorized into 9 groups based on 5-point increments (0 4, 5 9, 10 14, to 40 and above). Hospital and discharge information. Acute length of stay was measured in days from 0 to 364. Days in the intensive care unit were also recorded in full day increments and must be shorter than the total length of stay in the hospital. Payer source was grouped into 8 categories: (1) Health Maintenance Organization, (2) Preferred Provider Option, (3) Medicaid, (4) Medicare, (5) self-pay, (6) military, (7) liability, and (8) other. Other included other as well as Crippled Children s Fund funding and no charge. The NTDB groups facilities into 4 regions: (1) South, (2) Midwest, (3) West, and (4) Northeast. Acute care charges were adjusted to 2006 dollars. 19 The main outcome variable, discharge from acute hospital, was coded into 5 categories (1) SNF, (2) home (community residence), (3) other hospital (burn unit, acute specialty unit, or another hospital), (4) IRF, and (5) other. The category of other included an option of other, as well as jail or psychiatric facility. Statistical Analysis Bivariate comparison of variables pre- and postenactment of the PPS was conducted using t tests for continuous variables, Mann-Whitney U tests for ordinal variables, and chi-square tests for categorical variables. Discharge location was modeled as a dichotomous outcome (IRF vs other) using a multivariate logistic regression model. The model was adjusted for all categorical effects because of age, sex, race/ethnicity, nature of incident, GCS, AIS head, ISS nonhead, and insurance type.

PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman 1307 Table 1: Demographics and Injury Information All Subjects Pre-PPS Post-PPS Change P n 71,561 64,281 Age (y) (mean SD) 35.0 22.5 37.0 23.5 2.0.001 Male 70.0 69.8 0.2.359 Race.001 White 64.4 58.7 5.7 Black 16.9 13.9 2.9 Hispanic 10.5 15.1 4.6 Asian/Pacific Islander 1.3 2.2 0.9 Native American 0.4 0.4 0.0 Other 1.5 2.1 0.6 Unknown 5.0 7.6 2.6 Cause of injury.001 Motor vehicle collision 45.5 55.1 9.5 Fall 17.2 21.7 4.5 Assault 9.2 11.7 2.5 Other 5.8 5.7 0.1 Unknown 22.2 5.8 16.4 GCS (mean SD) 12.7 3.8 13.2 3.4 0.5.001 ISS nonhead (mean SD) 6.24 7.81 6.12 7.76 0.12.004 AIS head (mean SD) 3.21 0.94 3.23 0.93 0.02.001 NOTE. Values are percentages unless otherwise indicated. Calculations of significance exclude the unknown category. The admitting facility was also modeled as a fixed effect to account for correlated data from each site (df 122). As a sensitivity analysis, the model was further adjusted for all 2-way interactions among the demographic and injury-related variables. The PPS was modeled as a simple dichotomous predictor (pre vs post), and the results are reported as odds ratios. In order to examine whether the change in admission to inpatient rehabilitation occurring at the time of the PPS was different in subgroups, we also ran additional models with the interaction between the PPS and each variable, including region. Given the number of analyses, a Bonferroni correction was used requiring a P value below.006 to be considered meaningful. All missing covariate values (unknown) were modeled as separate categories, and therefore no subjects would be excluded during the modeling process. Statistical calculations were carried out in SAS (version 9.2). a RESULTS Tables 1 and 2 display demographics of the sample and characteristics of their hospital stay, as well as comparisons between those injured before (pre) and after (post) the IRF PPS was enacted. Results indicate that, both before and after the PPS, there were equal proportions of men, and acute care length of stay was the same for those who experienced TBIs. All other variables showed significant change; there were fewer whites and blacks admitted to acute care in the years after PPS implementation, and patients tended to be older. While GCS scores show a reduction in severity (higher scores reflecting lower severity), AIS head scores increased on average from 3.21 to 3.23, demonstrating a slight increase in severity of injury. In contrast, a slight decrease in the severity of the overall injury, other than the head, is indicated by a decrease in ISS. Change in Discharge Destination The change in discharge destination over time is displayed in figure 1. The reduction of discharges to inpatient rehabilitation appears to have occurred in the years leading up to 2002. Figure 2 displays change in discharge location by payer source. While the largest change in discharge to inpatient rehabilitation occurs for individuals who are covered under Medicare, there is a trend for a reduction in inpatient rehabilitation discharges across all payers. Individual discharge location appeared to differ across payer source, with some payers showing increases in discharge to home, others to SNF, and still others to alternative facilities. Finally, all regions showed a reduction in admissions to inpatient rehabilitation after PPS enactment. Table 2: Hospital and Discharge Information Hospital and Discharge Variables Pre-PPS Post-PPS Change P Acute LOS (mean SD) 8.5 12.6 8.5 13.4 0.0.783 ICU days (mean SD) 3.8 7.3 4.8 8.2 0.9.001 Payer source.001 HMO 14.5 14.3 0.2 Preferred Provider Option 15.7 12.0 3.7 Medicaid 12.0 11.6 0.4 Medicare 9.3 10.4 1.1 Self-pay 18.0 18.5 0.5 Military 0.7 0.4 0.3 Liability 8.9 7.0 1.8 Other 5.1 6.2 1.1 Unknown 15.9 19.6 3.7 Adjusted charges ($1000s) Mean 43.8 95.6 51.5 109.1 7.7.001 Median 18.4 22.0 3.6.001 Discharge location.001 SNF 5.6 6.9 1.3 Home 70.5 71.8 1.4 Other hospital 4.5 4.4 0.0 Rehabilitation 18.1 14.1 3.9 Other 1.4 2.7 1.3 NOTE. Values are percentages unless otherwise indicated. Calculations of significance exclude the unknown category. Abbreviations: HMO, Health Maintenance Organization; ICU, intensive care unit; LOS, length of stay.

1308 PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman Fig 1. Change in discharge location before and after implementation of the IRF PPS in 2002. Abbreviations: Hosp., hospital; Rehab, rehabilitation. Figure 3 shows the change in discharge location by region, where individuals in the Northeast had a 50% reduction in inpatient rehabilitation admissions for those who had a diagnosis of TBI and the largest increase in discharge directly to home. Impact of the PPS on Admission to Inpatient Rehabilitation In order to evaluate whether the enactment had an independent impact on admissions to inpatient rehabilitation for indi- Fig 2. Change in discharge location by insurance type and before/after enactment of the IRF PPS. Abbreviations: HMO, Health Maintenance Organization; Hosp., hospital; Rehab, rehabilitation.

PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman 1309 Fig 3. Change in discharge location by region and before/after enactment of the IRF PPS. Abbreviations: Hosp., hospital; Rehab, rehabilitation. viduals with TBI, a logistic regression model was evaluated. The model s outcome was either to inpatient rehabilitation or any other location. The results of the logistic regression model with all variables displayed can be seen in table 3. Results indicate that, given all other variables being equal, the odds of being discharged to an IRF after a TBI decreased 16% after Medicare s IRF PPS was enacted. The estimated reduction was unchanged when interactions were included in the model. When the interaction between the PPS and each variable was included, the only significant interaction was found between PPS and region. Results of the interaction effect are consistent with those displayed in figure 3, with significantly fewer individuals being admitted to inpatient rehabilitation after the PPS was enacted in the Northeast region compared with the other regions. DISCUSSION This study used information from the NTDB to look at the impact of the IRF PPS on access to care for persons with TBI. We found that there was a change in the proportion of persons after TBI that were admitted to rehabilitation. While the largest decrease in admissions occurred in persons who had Medicare as their primary form of insurance, smaller decreases also occurred in other insurance provider populations. This could indicate a potential follow along effect for other insurers. Regional differences were seen as well, with the Northeast region having the largest reduction in rehabilitation admissions and the largest increase in home discharges. It should be noted that the Northeast region had the highest percentage of IRF admissions prior to the PPS, and the decrease after PPS implementation brought this region s practice in line with others (see fig 3). Changes in the other regions were not as dramatic and followed the overall decrease seen over time with the implementation of the PPS (see fig 1). In the years since the implementation of the IRF PPS, rehabilitation admissions have been stable. While there were some demographic differences in the population being seen in trauma centers during the pre- versus post-pps implementation, these differences did not completely explain the effect that the IRF PPS had on the reduction of rehabilitation admissions. This finding is in agreement with Zorowitz, 11 who notes that the focus on tight management of length of stay after PPS implementation has increased discharges to SNFs prior to the completion of rehabilitation, as well as increased the number of transfers between facilities during care. However, there may have been an impact on inpatient rehabilitation admissions because of the severity of injury. Injury severity differed based on which measure was used. Both GCS and ISS nonhead scores suggest less severe injuries after implementation, but the AIS head score suggesting increasing severity. The reduction in the severity of other body injuries could lead to a decrease in admissions to inpatient rehabilitation. However, given that those individuals injured after the IRF PPS was implemented were older, the increased severity of AIS head scores may be related to the increased frequency of mass lesions, such as subdural hematomas. 20 Our findings are consistent with previous studies that looked at the IRF PPS implementation and access to rehabilitation care. The finding of decreased access to care for persons with TBI is in agreement with the prediction of Hoffman et al, 3 who suggested that access would be limited after PPS implementation. From this study alone, we are unable to say if these changes were because of financial considerations for IRFs after PPS implementation or a result of some other factor. Other researchers have found that initially after the implementation of the PPS, IRFs experienced increased profits, 9,10 and up until 2004 had not experienced any significant changes in the number of IRF openings or closings. 21 However, since 2005, the number of IRFs has been slowly declining. 22 Nevertheless, our finding of decreased use of inpatient rehabilitation is also consistent with the findings of Buntin et al, 15 because the likelihood of receiving postacute care in any form decreased

1310 PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman Table 3: Logistic Regression Predicting Discharge to Inpatient Rehabilitation Predictors n % P Odds Ratio 95% CI Age (vs 20 29y) 15,595 11.5.001* 0 9y 22,647 16.7.001 0.40 0.36 0.44 10 19y 25,553 18.8.212 0.96 0.91 1.02 30 39y 19,308 14.2.003 1.09 1.03 1.16 40 49y 18,058 13.3.001 1.22 1.15 1.29 50 59y 11,451 8.4.001 1.44 1.34 1.54 60 69y 7,412 5.5.001 1.74 1.61 1.88 70 79y 7,931 5.8.001 1.87 1.72 2.04 80y 7,458 5.5.001 1.46 1.33 1.60 Missing 429 0.3.178 1.22 0.91 1.64 Male (vs female) 94,933 69.9.022* 0.96 0.92 0.99 Race (vs white) 83,756 61.7.001* Black 21,015 15.5.411 0.98 0.92 1.03 Hispanic 17,263 12.7.001 0.72 0.67 0.77 Other 5,332 3.9.001 0.82 0.74 0.91 Missing 8,476 6.2.136 0.91 0.80 1.03 Cause of injury (vs MVC) 67,980 50.0.001* Fall 26,260 19.3.001 0.81 0.76 0.85 Assault 14,118 10.4.001 0.70 0.66 0.76 Other 7,866 5.8.001 0.77 0.71 0.84 Missing 19,618 14.4.001 0.88 0.84 0.93 GCS (vs mild) 84,770 62.4.001* Moderate 9,636 7.1.001 3.02 2.85 3.20 Severe 15,155 11.2.001 4.82 4.59 5.06 Paralyzed 5,521 4.1.001 4.13 3.86 4.43 Missing 20,760 15.3.001 1.78 1.68 1.88 AIS head (vs 2) 35,748 26.3.001* 3 46,619 34.3.001 2.39 2.24 2.55 4 41,538 30.6.001 4.09 3.83 4.37 5 11,937 8.8.001 7.95 7.37 8.57 ISS nonhead (vs 0 4) 77,807 57.3.001* 5 9 25,339 18.7.001 1.57 1.50 1.65 10 14 16,048 11.8.001 2.42 2.30 2.55 15 19 6,762 5.0.001 3.12 2.91 3.33 20 24 4,879 3.6.001 3.63 3.37 3.91 25 29 2,989 2.2.001 4.74 4.33 5.19 30 34 982 0.7.001 5.36 4.61 6.23 35 39 467 0.3.001 4.35 3.50 5.39 40 569 0.4.001 4.58 3.75 5.58 Insurance (vs Medicare) 13,350 9.8.001* HMO 19,544 14.4.048 0.92 0.85 1.00 Preferred Provider Option 18,923 13.9.538 0.98 0.90 1.06 Medicaid 16,066 11.8.177 0.94 0.87 1.03 Self-pay 24,792 18.3.001 0.55 0.50 0.60 Military 759 0.6.040 0.79 0.63 0.99 Liability 10,844 8.0.001 1.19 1.10 1.30 Other 7,604 5.6.001 0.73 0.66 0.81 Missing 23,960 17.6.001 0.78 0.71 0.85 Facility (fixed effect, 123 sites).001* Post-PPS (vs pre-pps) 64,281 47.3.001* 0.84 0.81 0.88 NOTE. The reference category for each categorical variable is shown in parentheses with the number of individuals and the percent of the entire group they represent shown in the columns marked n and % on the row with the variable name. Odds ratios greater than 1 indicate those in the given category are more likely to go to rehabilitation than those in the reference category. Abbreviations: CI, confidence interval; HMO, Health Maintenance Organization; MVC, motor vehicle collision. *Overall comparison among all categories. for stroke, total joint replacement, and hip fracture patients after the implementation of the PPS. In addition to agreeing with previous research, this study expands on the current literature by examining changes in practice that occurred prior to admission to inpatient rehabilitation. The majority of the literature on this topic has focused on changes (or lack thereof) in outcomes with persons that have been admitted to IRFs. What is lacking in that research is outcomes for those that are not admitted at all to inpatient rehabilitation. Our findings indicate that IRFs may have altered their admission criteria prior to the onset of the PPS in anticipation of payment changes. While we know the minimum amount of therapy services received on

PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman 1311 inpatient rehabilitation (which is described by Centers for Medical Services regulation), the type, amount, and quality of services persons with TBI received if they were discharged to another setting remains unknown. There are 2 potential policy implications from the findings of this research. Most notable is the impact of Medicare policy changes on the behavior of private insurance providers and other funders. Even though the change for the PPS applied only to those persons with TBI funded by Medicare, those funded by all other funding sources experienced a similar reduction in inpatient rehabilitation admissions. There are 3 possible reasons for this: (1) acute hospital care providers altered their referrals to IRFs based on the Medicare PPS regardless of the funding source; (2) funding sources adopted the PPS used by Medicare; or (3) IRFs that accept Medicare patients must treat all patients under Medicare guidelines regardless of funding source, and therefore the IRFs changed their admission criteria for all patients. No matter the cause, changes in government funding have the potential to change not only the governmental programs but also the coverage and funding provided by private and other insurance. In an era where health care reform is at the forefront, policymakers need to be aware of how changes at the governmental regulatory level may impact insurance coverage broadly. In addition to the impact on private insurance, this study illustrates the need for outcome measurement to be consistent across postacute facilities and care providers. Previous studies have demonstrated that, for persons with TBI, admission soon after injury to an IRF was related to improved outcomes. 23 Our study illustrates that persons with TBI who would have been admitted to inpatient rehabilitation prior to PPS implementation are being discharged to locations that likely have lower levels of rehabilitation interventions. This group and their functional and medical outcomes are not being included in the current PPS research. Therefore, we do not know if differences in discharge destination had any effect on outcomes for persons with TBI. A unified system for measuring medical and functional outcomes would provide a more complete assessment of outcome after TBI compared by type of postacute care. With an increased focus on quality of care and patient outcomes, both government and private insurers may find that potentially higher initial costs for inpatient rehabilitation or more intense outpatient rehabilitation may be more cost-effective over time, as has been seen in the literature on recovery after stroke. 24 Study Limitations There are a few limitations to this study that should be noted. There were missing data in the NTDB, especially on the severity of injury. Another limitation is that these analyses could be underestimating the impact of the PPS, given that facilities were preparing for the enactment of the PPS prior to 2002. This is evident in the reduction of IRF admissions for persons with TBI in the years approaching the final implementation date. Also, we used January 1, 2002 as the enactment date for the PPS for the purposes of this study. However, some facilities had phased in the PPS over the prior year. This could add to the underestimation of the impact on access to inpatient rehabilitation after PPS implementation. Finally, the NTDB does not provide information on the quality of care. This additional data would be helpful in determining the full impact of PPS implementation on persons with TBI. CONCLUSIONS This study identified that the enactment of the PPS for IRFs was associated with a reduction in the chance of admission to IRFs for persons with TBI. This impacted not only persons covered by Medicare, but also those covered by private insurance or other funders. The ability to measure care and compare outcomes for persons with TBI who are treated in different postacute rehabilitation facilities or not treated at all is necessary to fully understand the effects of rehabilitation health care reform. The current emphases on quality measures and comparative effectiveness may be helpful in deciphering the effects of future health care policy changes if similar outcome measures are required for use in all levels of postacute rehabilitation, including outpatient rehabilitation. References 1. Centers for Disease Control and Prevention. Traumatic brain injury. 2008. Available at: http://www.cdc.gov/ncipc/tbi/tbi.htm. Accessed January 15, 2009. 2. Frankel J, Marwitz J, Cifu D, Kreutzer J, Englander J, Rosenthal M. A follow-up study of older adults with traumatic brain injury: taking into account decreasing length of stay. Arch Phys Med Rehabil 2006;87:57-62. 3. Hoffman J, Doctor J, Chan L, Whyte J, Jha A, Dikmen S. Potential impact of the new Medicare prospective payment system on reimbursement for traumatic brain injury inpatient rehabilitation. Arch Phys Med Rehabil 2003;84:1165-72. 4. Buntin MB. Access to postacute rehabilitation. Arch Phys Med Rehabil 2007;88:1488-93. 5. Chan L, Koepsell TD, Deyo RA, et al. The effect of Medicare s payment system for rehabilitation hospitals on length of stay, charges, and total payments. N Engl J Med 1997;337:978-85. 6. Stineman MG. Prospective payment, prospective challenge. Arch Phys Med Rehabil 2002;83:1802-5. 7. Paddock SM, Escarce JJ, Hayden O, Buntin MB. Did the Medicare inpatient rehabilitation facility prospective payment system result in changes in relative patient severity and relative resource use? Med Care 2007;45:123-30. 8. Gillen R, Tennen H, McKee T. The impact of the inpatient rehabilitation facility prospective payment system on stroke program outcomes. Am J Phys Med Rehabil 2007;86:356-63. 9. McCue MJ, Thompson JM. Early effects of the prospective payment system on inpatient rehabilitation hospital performance. Arch Phys Med Rehabil 2006;87:198-202. 10. Thompson JM, McCue MJ. Performance of freestanding inpatient rehabilitation hospitals before and after the rehabilitation prospective payment system. Health Care Manage Rev 2010;35:36-45. 11. Zorowitz RD. Inpatient rehabilitation facilities under the prospective payment system: lessons learned. Eur J Phys Rehabil Med 2009;45:259-63. 12. Dobrez D, Lo Sasso A, Heinemann A. The effect of prospective payment on rehabilitative care. Arch Phys Med Rehabil 2004;85:1909-14. 13. Sood N, Buntin MB, Escarce JJ. Does how much and how you pay matter? Evidence from the inpatient rehabilitation care prospective payment system. J Health Econ 2008;27:1046-59. 14. Shah P, Heinemann A, Manheim L. The effect of Medicare s Prospective Payment System on patient satisfaction: an illustration with four rehabilitation hospitals. Am J Phys Med Rehabil 2007;86:169-75. 15. Buntin MB, Colla CH, Escarce JJ. Effects of payment changes on trends in post-acute care. Health Serv Res 2009;44:1188-210. 16. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Epidemiology 2007;18:800-4. 17. Teasdale G, Jennett B. Assessment of coma and impaired consciousness. A practical scale. Lancet 1974;2:81-4. 18. United States Department of Labor. Consumer Price Index. 2006. Available at: http://www.bls.gov/cpi/. Accessed May 13, 2011.

1312 PROSPECTIVE PAYMENT SYSTEM AND TBI REHABILITATION, Hoffman 19. Baker SP, O Neill B. The injury severity score: an update. J Trauma 1976;16:882-5. 20. Stocchetti N, Paternò R, Citerio G, Beretta L, Colombo A. Traumatic brain injury in an aging population. J Neurotrauma 2012;29:1119-25. 21. Mallinson TR, Manheim LM, Almagor O, Demark HM, Heinemann AW. Trends in the supply of inpatient rehabilitation facilities services: 1996 to 2004. Arch Phys Med Rehabil 2008;89:2066-79. 22. Inpatient rehabilitation facility services. In: Medicare Payment Advisory Commission, editor. Report to the Congress: Medicare payment policy. Washington (DC): MedPac; 2011. p 201-27. 23. Kunik C, Flowers L, Kazanjian T. Time to rehabilitation admission and associated outcomes for patients with traumatic brain injury. Arch Phys Med Rehabil 2006;87:1590-6. 24. Wang H, Camicia M, Terdiman J, Hung YY, Sandel ME. Time to inpatient rehabilitation hospital admission and functional outcomes of stroke patients. PM R 2011;3:296-304. Supplier a. SAS Inc, 100 SAS Campus Dr, Cary, NC 27513-2414.