ON JANUARY 1, 2002, the Centers for Medicare and

1165 Potential Impact of the New Medicare Prospective Payment System on Reimbursement for Traumatic Brain Injury Inpatient Rehabilitation Jeanne M. Hoffman, PhD, Jason N. Doctor, PhD, Leighton Chan, MD, MPH, John Whyte, MD, PhD, Amit Jha, MD, MPH, Sureyya Dikmen, PhD ABSTRACT. Hoffman JM, Doctor JN, 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. Objective: To evaluate the potential impact of the new Medicare prospective payment system (PPS) on traumatic brain injury (TBI) rehabilitation. Design: Retrospective cohort study of patients with TBI. Patients were assigned to their appropriate case-mix group (CMG) based on Medicare criteria. Setting: Fourteen urban rehabilitation facilities throughout the United States. Participants: Patients with TBI admitted to inpatient rehabilitation and enrolled in the Traumatic Brain Injury Model Systems from 1998 to 2001 (N 1807). Interventions: Not applicable. Main Outcome Measures: Cost of inpatient rehabilitation admission, length of stay (LOS), and functional outcomes. Results: The median cost of inpatient rehabilitation for patients with TBI exceeded median PPS payments for all TBI CMGs by 16%. Only 3 of the 14 hospitals received reimbursement under PPS that exceeded costs for their TBI patients. Conclusions: Compared with current costs, the new Medicare payment system may reimburse facilities significantly less than their costs for the treatment of TBI. To maintain their current financial status, facilities may have to reduce LOS and/or reduce resource use. With a decreased LOS, inpatient rehabilitation services will have to improve FIM efficiency or discharge patients with lower discharge FIM scores. Key Words: Brain injuries; Health care economics and organizations; Medicare; Prospective payment system; Rehabilitation. From the Departments of Rehabilitation Medicine (Hoffman, Doctor, Chan, Dikmen), Medical Education and Biomedical Informatics (Doctor), Health Services (Doctor), Neurosurgery (Dikmen), and Psychiatry and Behavioral Sciences (Dikmen), University of Washington, Seattle, WA; Moss Rehabilitation Research Institute (Whyte) and Department of Rehabilitation Medicine, Thomas Jefferson University (Whyte), Philadelphia, PA; Craig Hospital, Englewood, CO (Jha); University of Colorado, Denver, CO (Jha). Supported by the National Center for Medical Rehabilitation Research (grant no. T32 HD07424) and National Institute on Disability and Rehabilitation Research through the Traumatic Brain Injury Model Systems project. The views expressed herein are those of the authors and not necessarily of the University of Washington, Moss Rehabilitation Research Institute, or Craig Hospital. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Correspondence to Jeanne M. Hoffman, PhD, Dept of Rehabilitation Medicine, University of Washington, Box 356490, Seattle, WA 98195-6490, e-mail: jeanneh@u.washington.edu. Reprints are not available. 0003-9993/03/8408-7820$30.00/0 doi:10.1016/s0003-9993(03)00232-6 2003 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation ON JANUARY 1, 2002, the Centers for Medicare and Medicaid Services (CMS) implemented a prospective payment system (PPS) for inpatient rehabilitation facilities. 1 The rehabilitation PPS now provides a single payment for all hospital services including room and board, nursing care, therapy, radiology, medications, and laboratory billing. The new rehabilitation PPS is based on case-mix groups (CMGs) that provide payment to facilities for individual patient stays based on the principal rehabilitation diagnosis, functional status as measured by the FIM TM instrument, and age. Each rehabilitation patient is classified into a CMG at the time of admission based on these 3 variables. There are 95 separate CMGs, with 5 for traumatic brain injury (TBI). The 5 CMGs for TBI vary depending on the patient s motor and cognitive functioning levels. Each CMG is assigned a relative weight that reflects the level of resources required to treat that class of patient. Those patients with more functional deficits who need more rehabilitation will receive higher weights. The weights are then multiplied by a conversion factor ($11,838 for 2002) to obtain a base payment for the patient. The base payment is then adjusted for the patient s most severe qualifying comorbid condition (fig 1). Finally, each facility adjusts its reimbursement by 4 additional factors: (1) the wage rates in its region, (2) percentage of low-income patients, (3) rural adjustment if applicable, and (4) high-cost outlier payments. The adjustment for wage rates for each facility reflects the varying wage levels in different parts of the United States and is related to the labor-related share of the PPS payment. The low-income payment adjustment reflects the increase in costs of treating low-income patients and is based on the percentage of individuals served by the rehabilitation facility who are on Medicaid. In development of the PPS, CMS research indicated that rural facilities had significantly higher costs per case; therefore, a rural adjustment was also included. Finally, a special outlier payment above the PPS was developed to protect facilities from extremely high cost cases. For those cases in which costs exceed an adjusted threshold amount, a payment of 80% of the amount above the threshold is paid in addition to the CMG payment. However, CMS limits these outlier payments to 3% of total payments for each year. Concerns have been raised that the new PPS system will not compensate appropriately for cases in higher acuity CMGs based on CMS assigning insufficient weights to these groups. 2 Inappropriate underweighting for higher acuity CMGs, such as patients recovering from severe trauma, could lead to undercompensation for these cases. Poor compensation may reduce

1166 PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman Fig 1. PPS flow chart when LOS equals or exceeds average Medicare LOS. *Medicare limits outlier payments to 3% of total yearly payments. the number of high-acuity patients that inpatient rehabilitation facilities will be willing to admit, possibly leading to a significant number of underserved patients. Furthermore, the burden of this underreimbursement may fall disproportionately on facilities, such as academic medical centers, where patients are often treated despite the financial risk that may be associated with increased acuity. 3 Finally, because other payers may implement Medicare funding rules, widespread adoption of the PPS system by other insurers could have a dramatic effect on inpatient rehabilitation facilities, compounding any inequities built into the Medicare PPS system. 4,5 Given the concerns about the possible impact of inpatient rehabilitation PPS and the possibility that some diagnoses may result in the loss of money under the new system, we sought to evaluate the potential impact on a particularly high-acuity condition, TBI. Patients with severe TBI are some of the most costly patients seen in inpatient rehabilitation 6 and thus are at risk for inadequate reimbursement. Poor reimbursement for these patients may decrease length of stay (LOS), thus compromising patient functional outcomes. METHODS Data Source The US Department of Education, National Institute on Disability and Rehabilitation Research, has provided funding for Traumatic Brain Injury Model Systems (TBIMS) since 1987, with expanded funding to add additional centers in 1997. 7 The TBIMS project is a prospective, longitudinal multicenter study that examines the course of recovery and outcomes of persons after TBI. Information in the database is collected during initial hospitalization and annually up to year 5 and at subsequent 5-year intervals up to year 20. We analyzed 1807 patients from 14 sites who were enrolled in the TBIMS from 1998 to 2001. Entry into the TBIMS requires patients to have a diagnosis of TBI, to be at least 16

PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman 1167 years of age, to arrive at the acute care hospital within 24 hours of injury, to be admitted from acute care directly to inpatient rehabilitation, and to provide informed consent. Analyses were conducted to compare the 152 patients who were at or over the age of 65 years to those under age 65 (n 1655) to determine if they differed significantly. No differences were found between these 2 groups in assignment to CMG, cost, or LOS. Therefore, only aggregate results are presented. Measures TBIMS database. The TBIMS collect information on each patient enrolled, including the actual amount charged for each patient s inpatient rehabilitation stay, the actual LOS for each patient, and the FIM at admission and discharge. The FIM includes assessment of basic activities of daily living in motor and cognitive functioning. Eighteen items from 6 domains, including self-care, sphincter control, mobility, locomotion, communication, and social cognition, are scored based on the patient s level of independence or dependence in each area. Ratings are made on a scale from 1 to 7, with 1 being completely dependent to 7 being independent. 8 In the new assessment tool for inpatient rehabilitation PPS, the same FIM instrument is used. Transfer to tub or shower is scored but not used in the determination of the CMG. Facility data. Estimated low-income patient percentage and Medicare cost-to-charge data were collected directly from each inpatient rehabilitation facility in the study. Information on low-income patient percentages was calculated from hospital disproportionate share hospital values (DSH). Disproportionate share hospital values are calculated by the following formula: Medicare SSI Days DSH Total Medicare Days Medicaid Non-Medicare Days (1) Total Days where SSI is Supplemental Security Income. The low-income patient percentage (LIP) is then calculated: LIP (1 DSH).4838 (2) For those facilities in our study, low-income patient percentage values ranged from 1.029 to 1.147. The Medicare cost-to-charge ratios were obtained from each facility and were specific to inpatient rehabilitation units. Costto-charge ratios were gathered from Medicare cost reports where the ratio is the total yearly costs to the total yearly charges for rehabilitation. The cost-to-charge ratio is computed from total patient costs and total patient charges. However, because the number of cases drops out of the ratio, it also represents a ratio of average patient costs to average patient charges. Hence, cost-to-charge ratios can be used to estimate the facility cost associated with any patient, x i, when only the actual individual charges are available by using the following equation: Estimated Patient Cost (x i ) Costs Charges Charges(x i) (3) Summing across all TBI patients treated gives the estimated cost of TBI care. Even though only estimates of individual costs can be calculated, this method improves on the biases found when using only hospital charges. 9 One facility (facility 7) did not provide their Medicare cost-to-charge ratio; therefore, an average of all other facilities data was substituted for that rehabilitation unit. Comorbidity data. Under the PPS, additional diagnoses that require treatment during inpatient rehabilitation and increase the cost of rehabilitation are defined as comorbidities. The full list of diagnoses allowed under the new system is published in the Final Rule. 1 In the current study, 3 of the 14 facilities provided detailed information on comorbid diagnoses treated during each patient s rehabilitation stay. In addition to the information from the 3 facilities, an estimate of the frequency of comorbidities in patients with TBI was taken from a Medicare billing database of 2046 patients treated for TBI. These patients represent all Medicare beneficiaries admitted to an inpatient rehabilitation freestanding hospital from 1987 to 1994 in the United States for treatment of a TBI. 6 Area wage and rural adjustments. Area wage indexes for each facility were taken from the published list in CMS s Final Rule. 1 Wage indexes range from a high of 1.4483 for New York City, a 45% increase, to a low of 0.4651 for Puerto Rico, a 46% reduction. The area wage adjustment applies to the labor portion of the payment, which has been determined to be 72.395% of the CMG payment. Multiplying the area wage index by the labor portion of the payment gives the adjustment. In addition to the area wage adjustments, facilities are divided into urban versus rural locations. Urban locations are defined as metropolitan statistical areas or counties in which there is a city with a population greater than 50,000. Rural locations are everywhere else. Facilities that are rural receive an additional rural adjustment that increases the total CMG payment by 19%. None of the 14 TBIMS sites was considered rural. Assignment of CMGs and Comorbidity Tiering Patients were assigned to 1 of the 5 CMGs for TBI based on their admission FIM score. Comorbidity tiers (0, none; 1, severe/high cost; 2, moderate/medium cost; 3, mild/low cost) were randomly assigned to individual patients based on the estimated percentage of occurrence of each tier from the actual data collected from 3 sites. Random assignment to patients was chosen based on the lack of pattern for frequency or severity of tiers to CMGs in these data. The Medicare average LOS (the amount of time that Medicare estimates the patient should stay in inpatient rehabilitation) and payment were then adjusted based on the tier assigned. The standard payment for each CMG was then adjusted for area wage index and percentage of low-income patients to determine the final PPS payment. In addition to randomly assigning tiers based on data from the 3 TBIMS sites, we also used 2 other methods to assess the impact of comorbid illnesses. First, the tiers were randomly assigned based on the percentages calculated on the 1987 94 data from Medicare patients with TBI. The distribution of the comorbidities in this group differed significantly from those admitted to the TBIMS, and the results represent another reference point in our analysis. Finally, it is possible that coding of comorbidities will increase with the new system, because payment was not contingent on regular coding of additional comorbid diagnoses prior to the PPS. 10 Therefore, we conducted a sensitivity analysis to test the financial impact of 20% to 70% increases in comorbidities. This was done by increasing the 3 levels of comorbid diagnoses based on their frequency of occurrence from the comorbidity data provided by the 3 Model Systems. For example, in this data, 1% of patients had tier 1 diagnoses, 19% had tier 2 diagnoses, 10% had tier 3 diagnoses, and 70% had no comorbid diagnoses. With a 20% increase in comorbidities,

1168 PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman Table 1: Demographics for Entire Sample Age (y) 37.5 16.8 Injury severity (lowest GCS motor score) 3.6 2.0 FIM (range, 17 119) Admission 53.9 23.9 Discharge 89.0 21.4 Male (%) 74.0 White (%) 65.8 Level of education (%) High school 32.8 High school 30.4 High school 32.9 Mechanism of injury (%) Motor vehicle crash 59.7 Fall 16.5 Assault 13.7 Other 10.1 NOTE. Values are mean standard deviation or as otherwise indicated. Abbreviation: GCS, Glasgow Coma Scale. Table 2: Percentage in Each CMG CMG* 5 4 3 2 1 n(%of sample) Facility 13 63 13 15 6 2 82 (5) Facility 1 49 13 27 4 7 238 (13) Facility 6 48 21 25 3 3 119 (7) Facility 4 46 17 13 14 10 156 (9) Facility 9 46 13 26 5 9 99 (5) Facility 7 45 20 19 3 12 118 (6) Facility 8 45 13 37 0 5 106 (6) Facility 5 37 12 40 3 9 140 (8) Facility 2 34 19 35 2 10 222 (12) Facility 11 32 17 34 3 14 88 (5) Facility 14 31 8 43 10 8 49 (3) Facility 10 25 17 33 12 13 89 (5) Facility 12 23 22 34 10 12 83 (5) Facility 3 8 12 47 8 24 201 (11) All facilities 38 16 31 6 10 1790 *Values are rounded and may exceed 100%. 1.8% of patients were then randomly assigned to tier 1, 31.6% were then assigned to tier 2, 16.6% were then assigned to tier 3, and 50% were then assigned to tier 0. Cost of Cases and Outliers We applied the facility-specific cost-to-charge ratio to the total inpatient rehabilitation charges to calculate the cost of each case. The difference between this cost and the new PPS payment was calculated. We identified outliers and calculated payment according to the method defined by CMS. 1 In its simplest form, outliers are defined as cases that exceed a threshold amount (the PPS CMG payment plus $11,211), adjusted for area wage index and low-income patient percentage. The final payment is then calculated as the CMG payment plus 80% of the cost beyond the threshold amount. Cost-to-Charge Analyses Our sensitivity analysis of cost-to-charge ratios involves the systematic variation of the structural assumptions within the cost estimation model to determine at what point hospitals begin to lose money. We conducted a sensitivity analysis on cost-to-charge ratios because these values change over time and are estimates of actual cost of rehabilitation for all diagnoses and not specifically for TBI rehabilitation. Cost-tocharge ratios used in the sensitivity analysis represented the range of values collected from the set of facilities. These values were then multiplied by the actual charges for rehabilitation and divided by the actual LOS for each patient to create a cost per day for each patient. These values were then compared with the average payment per day under PPS for all TBI CMGs. RESULTS Demographics Demographic data for the entire sample are presented in table 1. The percentage of patients assigned to each CMG and sample sizes for the entire sample and by site is displayed in table 2. Data are presented from the highest (CMG 5) to the lowest (CMG 1) level of independence. Potential Financial Impact Table 3 shows the difference between the median cost per case and the PPS payment for each facility. The expected payment from PPS will not cover the cost of rehabilitation for TBI patients, with losses ranging from 5% to 41% of costs. Of the 14 sites, 3 have PPS payments that exceed their median cost per case. When the differences between the cost and PPS payment are broken down by individual CMG, the median cost and PPS payment are equal for those patients in CMG 3. However, the differences for the more severely injured patients (CMGs of 4 and 5) are large (16% to 27% loss), whereas the differences for the less severely injured patients (CMGs of 1 and 2) are smaller (both approximately 7% loss). Length of Stay The difference between the actual LOS for each patient was compared with the LOS necessary to break even. This break- Table 3: Difference Between Median Cost Per Case and PPS Payment Cost ($) PPS Payment ($) Loss per Case ($)* (% of cost) Facility 4 25,583 15,066 10,517 ( 41) Facility 5 23,578 16,887 6691 ( 28) Facility 9 23,274 16,886 6388 ( 27) Facility 8 23,729 17,593 6136 ( 26) Facility 6 25,441 20,018 5423 ( 21) Facility 3 17,906 14,906 3209 ( 18) Facility 7 20,415 18,100 2315 ( 11) Facility 2 18,551 16,470 2081 ( 11) Facility 14 15,941 14,474 1467 ( 9) Facility 10 16,345 15,347 998 ( 6) Facility 12 17,455 16,540 915 ( 5) Facility 13 31,150 31,693 543 ( 2) Facility 11 16,400 17,204 804 ( 5) Facility 1 15,434 19,854 4420 ( 29) All facilities 19,949 16,837 3112 ( 16) *Negative values indicate gain with PPS payment.

PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman 1169 Table 4: Difference in Median LOS Across CMGs Actual LOS Break-Even LOS Under PPS Difference in LOS* Facility 5 22.0 14.8 7.2 Facility 6 25.0 18.9 6.1 Facility 4 16.0 9.9 6.1 Facility 8 26.0 20.9 5.1 Facility 9 23.0 18.8 4.2 Facility 13 25.0 21.1 3.9 Facility 7 26.0 23.6 2.4 Facility 2 18.0 15.7 2.3 Facility 3 15.0 12.9 2.1 Facility 12 16.0 14.8 1.2 Facility 10 23.5 22.4 1.1 Facility 14 16.0 15.5 0.5 Facility 11 22.0 22.4 0.4 Facility 1 17.0 22.5 5.5 All facilities 20.0 17.3 2.7 *Negative values indicate gain with PPS payment. even LOS was calculated by dividing the PPS payment by the ratio of the cost to the actual LOS. Assuming no change in the cost per day, results indicate that the median LOS required to break even was on average 2.7 days less (13.5%) than the median actual LOS (table 4). In other words, facilities would need to shorten their LOS by almost 3 days to maintain their current financial status for patients with TBI. Results broken down across the 5 CMGs are similar to those found when analyzing cost. Those patients in CMG 3 have a LOS that matches the break-even LOS. All other CMGs have an actual LOS that exceeds the break-even LOS, ranging from 1 to 7 days (7% to 24% difference). The PPS includes the Medicare average LOS that corresponds to each CMG and is also adjusted based on tiering. These Medicare average LOSs are used to determine whether a facility will receive their full payment under the PPS or a per diem payment. If a patient is discharged to another facility (eg, hospital, skilled nursing facility, other rehabilitation) before reaching his/her Medicare average LOS, the facility will receive a per diem payment. If the patient is discharged after the Medicare average LOS to another facility, the rehabilitation facility will receive the entire payment under the PPS. In our study, these Medicare average LOSs are similar to the actual LOS (except for the most severely injured patients) and are greater than the LOS required to break even. Comorbidities The impact of comorbidities on payment and LOS was evaluated by comparing the separate methods of tiering. The first method is described earlier by using percentages based on data collected from 3 TBIMS sites. The second is based on percentages from Medicare patients with TBI. The final methods are based on a sensitivity analysis with incremental increases from 20% to 70% upcoding from the first method of tiering. Our results indicate that the assignment of comorbidities had only a minor effect on overall reimbursement. Even when all patients have some type of comorbid diagnosis (70% increase), the cost of rehabilitation still exceeds PPS payment by 8% (fig 2). Functional Outcomes The impact of the PPS on functional improvement was evaluated assuming that attempts are made to achieve the same Fig 2. Sensitivity analysis to evaluate the PPS payment increase with incremental upcoding of comorbid conditions. functional outcomes. For this evaluation, we compared each patient s actual functional improvement to the expected functional improvement (which was estimated by dividing the actual functional improvement by the LOS and multiplying by the LOS required to break even). Table 5 shows the differences in levels of actual improvement and the potential loss of functional improvement expected under the PPS for each facility. Results range from an average of a small 0.1-point (.01%) decrease in FIM score to a large 11.8-point (36%) reduction in FIM score under the new system. To achieve similar functional outcomes under the new system, the efficiency of treatment would have to increase. Table 6 displays the average actual FIM improvement per day compared with the FIM improvement per day required to achieve the same level of improvement. Twelve of the 14 sites would be required to increase their FIM efficiency to achieve the same level of discharge function. Outliers To determine the potential impact on payment for patients with TBI, we determined the PPS payment for outlier cases for Table 5: Difference of Median FIM During Inpatient Rehabilitation Actual FIM Estimated FIM Under PPS Estimated Decrease in Functional * Facility 9 33.0 21.2 11.8 Facility 4 27.0 17.4 9.6 Facility 6 40.0 31.2 8.8 Facility 5 33.0 24.4 8.6 Facility 13 40.5 33.2 7.3 Facility 7 43.0 36.0 7.0 Facility 3 30.0 23.6 6.4 Facility 8 37.0 31.0 6.0 Facility 2 34.0 29.4 4.6 Facility 11 38.0 35.2 2.8 Facility 12 27.0 24.3 2.7 Facility 10 26.0 23.9 2.1 Facility 14 29.0 28.9 0.1 Facility 1 37.0 46.1 9.1 All facilities 34.0 28.9 5.1 *Negative values indicate gain.

1170 PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman Table 6: Difference of Median FIM per Day During Inpatient Rehabilitation Actual FIM per Day Estimated FIM per Day Under PPS Difference in Functional per Day* Facility 4 2.86 2.00 0.86 Facility 3 2.69 1.94 0.75 Facility 5 2.25 1.55 0.70 Facility 9 1.94 1.28 0.66 Facility 13 2.10 1.61 0.49 Facility 8 1.76 1.40 0.36 Facility 6 2.05 1.78 0.27 Facility 2 2.18 1.93 0.25 Facility 12 1.97 1.73 0.24 Facility 14 1.90 1.76 0.14 Facility 10 1.11 1.00 0.11 Facility 7 1.95 1.87 0.08 Facility 11 1.49 1.50 0.01 Facility 1 1.69 2.26 0.57 All facilities 1.98 1.73 0.25 *Negative values indicate gain. Fig 4. Sensitivity analysis to evaluate robustness of difference between PPS payment and cost of rehabilitation per day. each site. Figure 3 shows the payment ratio for high-cost outliers to the total PPS payment. Results suggest that overall outlier payments make up 40% of total payments for TBI: of the over $40 million paid to rehabilitation facilities for the treatment of TBI patients in our study, $16 million would be considered outlier payments. Outliers were not specific to any group and existed in all CMGs. Cost of Rehabilitation Figure 4 depicts the comparison of average PPS payment per day for all CMGs compared with the cost of rehabilitation per day given the range of cost-to-charge ratios (.45 to.95). Results indicate that once the cost-to-charge ratio exceeds.50, the cost of rehabilitation exceeds the PPS payment. As depicted in figure 4, at.67, the average cost-to-charge ratio for the 14 sites, the average loss per day is $266, or 25% of the costs. Ratio of payment for high-cost outliers to total PPS pay- Fig 3. ment. DISCUSSION Results from the present study suggest that the PPS may significantly decrease funding for patients with TBI and may potentially lead to decreased LOSs to compensate for these reductions. Despite attempts by the PPS to adjust for individual facility factors and additional diagnoses for each patient treated the new system does not appear to fully capture the financial costs related to TBI. Even if every patient had a qualifying comorbid condition, the overall median PPS payment may still not meet the cost of inpatient rehabilitation. Results from the sensitivity analysis suggest that when the cost-to-charge ratio exceeds approximately.50, the cost of inpatient rehabilitation per day will be above the PPS payment. Two of the 3 facilities that appear to benefit from the PPS have cost-to-charge ratios below.50. The third facility that benefits most from the PPS has a cost-to-charge ratio below the average of the 14 facilities, although above.50. This facility differs from the others in that it has the lowest median cost with a large percentage of patients with severe TBI in the most dependent CMGs with relatively short LOSs. The changes in LOS expected for facilities to break even are likely to lead to either decreases in functional outcomes for patients or expectations of increased efficiency. However, increased efficiency may not be possible in patients with TBI. The changes in LOS can also have a financial impact beyond that of reimbursement for hospital charges. With decreased LOS, physician and psychologist charges may also be reduced for some inpatient rehabilitation facilities, which could lead to even greater losses. These losses can be significant for hospitals that rely on inpatient profits to overcome losses typically incurred from outpatient services. Forty percent of all the dollars spent on the study patients would have qualified as outlier payments. Given that Medicare limits such payments to 3% of total payments for all patients during the year, it is unlikely that the outlier system would mitigate the losses we have projected. Outliers occur in all CMGs, suggesting that increased injury severity is not the primary factor associated with increased cost and LOS.

PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman 1171 Limitations The current study has several limitations. First, the results of the present study apply only to the CMGs for patients with TBI. Thus, it is possible that other rehabilitation diagnoses may fit more accurately with the PPS system and may compensate for any losses incurred for patients with TBI. Furthermore, the TBIMS facilities are all associated with academic medical centers, and the impact of the PPS for TBI in other facilities may differ, especially those in rural areas. Besides TBI being a diagnosis that may lead to losses under the PPS, it may also be a diagnosis with a high rate of outliers. Other CMGs may have few or no outliers under the new payment system and the 3% limit, which is based on all discharges and is not CMG specific, may be appropriate. However, given the large number of outliers, up to 45% for 1 facility (accounting for 66% of total payments), it is unlikely that the rates would not be above the 3% limit. FIM improvement and efficiency was calculated per day for the purposes of the current study. Although recovery in TBI is unlikely to be linear and stable, the FIM is only completed at admission and discharge and daily fluctuations are not available. Although a minority of patients with TBI are funded by Medicare, we believe that these results are relevant because other payers are likely to follow the PPS as they have followed implementation of the inpatient hospital PPS. 4 Further evidence suggesting that the patients funded by Medicare have levels of injury, cost, and LOS similar to those of persons not funded by Medicare comes from analyses we conducted comparing patients age 65 years or older with those under age 65. Results suggest that no differences between groups exist in CMG assignment or in the impact of the PPS leading to decreased reimbursement or LOS. Possible Solutions Several solutions may have to be considered to reduce the negative financial impact of the PPS on patients with TBI. Theoretically, 1 possibility is to increase the efficiency of rehabilitation to decrease LOS. This would require the use of rehabilitation methods that would not dramatically increase costs. Attempts to increase the number of hours of rehabilitation per day may lead to increased staff costs and may not be beneficial to patients who may not be able to benefit from increased intensity of rehabilitation. On the other hand, improving services on the weekend might be cost effective. A second possible option is to select different patients for rehabilitation and admit only those patients who can benefit from short rehabilitation stays or who have less severe injuries or comorbidities that would increase payment but not LOS. The number and length of admissions to rehabilitation increased when the PPS was enacted for acute care inpatient hospitalization, 2 suggesting that patients came to rehabilitation earlier and often were sicker. With the PPS for inpatient rehabilitation, there will likely be a change in referral patterns for inpatient rehabilitation. Decisions will have to be made weighing the cost of sending patients to rehabilitation compared with lengthening their acute inpatient stay. Rehabilitation facilities will also need to balance the cost of early discharges to other facilities where they would receive per diem payments against keeping patients who may exceed their PPS payment. In addition to selection of patients, facility selection may also be an avenue for reducing cost. Facilities with a lower cost base may be able to provide treatment that would result in similar outcomes without decreasing LOS or intensity of treatment. Other options that should be considered include preparing patients and their families early in the admission process for a short LOS if they are admitted directly from inpatient hospitalization, or to defer rehabilitation until patients can benefit maximally for rehabilitation. Delaying admission to inpatient rehabilitation or considering more than 1 admission to address different rehabilitation issues may also need consideration. To protect small facilities from the financial cost of outliers, facilities may consider investing in reinsurance, 11 which is a method of managing financial risk without institutions being required to maintain a reserve to assure financial solvency. Under the PPS, outlier payments are limited; therefore, facilities may avoid treating costly patients or may transfer cases to providers who cannot avoid treating such patients, which can lead to cost variations not accounted for in PPS. Voluntary reinsurance, therefore, would allow those facilities with more outliers, facilities that are required to treat patients despite the financial risk, to manage the risk and maintain solvency. Rehabilitation for TBI would benefit from the development of new, efficient, and efficacious rehabilitation methods that decrease the cost of inpatient rehabilitation and the LOS required. In addition, with the continuing decrease in LOS that has been occurring over the years and the fact that TBI is associated with long-term deficits, more aggressive efforts are necessary to develop effective community interventions and support for patients with TBI. It is likely that modification will have to be made to the current inpatient rehabilitation PPS. The CMG weights for TBI may have to be increased, and the adjustment for other factors, such as rural status, may have to be refined. Those who developed the comorbidity tiering system suggest that more severely injured patients may have multiple comorbid conditions that may significantly increase cost and suggest that further research is needed to address how comorbidities are weighted. 12 The current list of comorbid diagnoses may not include all diagnoses that lead to increased costs in rehabilitation populations. Such diagnoses, including depression and substance abuse, may influence the LOS and course of rehabilitation, especially in patients with TBI, which would lead to increasing costs. CONCLUSIONS The new Medicare payment system may lead to significant underreimbursement for patients with TBI. These payment reductions could have a profound, negative effect on the treatment of this type of patient. Many rehabilitation facilities will have to become more efficient when treating these patients or risk financial losses. Also, Medicare should monitor its payments for patients with TBI and should modify its reimbursement patterns to ensure access to inpatient rehabilitation for those patients who might benefit from it. Acknowledgments: We are indebted to Anthony Stringer, PhD, Georgia Model Brain Injury System, Emory University, Atlanta, GA, and Tom Novack, PhD, Spain Rehabilitation Center, University of Alabama, Birmingham, AL, for assisting with the project. References 1. Department of Health and Human Services Center for Medicare and Medicaid Services. 66 Federal Register 41316 (2001). 2. Reynolds M. Final inpatient rehabilitation PPS rule improves on proposed rule. Healthc Financ Manage 2001;55(10):68-70.

1172 PROSPECTIVE PAYMENT AND TBI REHABILITATION, Hoffman 3. Taheri PA, Butz DA, Dechert R, Greenfield LJ. How DRGs hurt academic health systems. J Am Coll Surg 2001;193:1-8; discussion 8-11. 4. Scheffler RM, Clement DG, Sullivan SD, Hu TW, Sung HY. The hospital response to Medicare s Prospective Payment System: an econometric model of Blue Cross and Blue Shield plans. Med Care 1994;32:471-85. 5. Rosenberg M, Browne M. The impact of the inpatient prospective payment system and diagnosis-related groups: a survey of the literature. North Am Actuarial J 2001;5:84-94. 6. 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. 7. TBI Model System Grants. J Head Trauma Rehabil 1999;14(2): 189-200. 8. Hamilton BB, Granger CV, Sherwin FS, Zielezny M, Tashman JS. A uniform national data system for medical rehabilitation. In: Fuhrer MJ, editor. Rehabilitation outcomes: analysis and measurement. Baltimore: Brookes; 1987. p 137-47. 9. Drummond M, O Brien B, Stoddart G, Torrance G. Method for the economic evaluation of health care programmes. 2nd ed. Oxford: Oxford Univ Pr; 1997. 10. Grimaldi P. Inpatient rehabilitation facilities are now paid prospective rates. J Health Care Finance 2002;28(3):32-4. 11. Dada M, White W. Evaluating financial risk in the Medicare prospective payment system. Manage Sci 1999;45:316-29. 12. Carter G, Beeuwkes Buntin M, Hayden O, et al. Analyses for the initial implementation of Medicare s inpatient rehabilitation prospective payment system. Santa Monica: RAND; 2001.