Volume 87, Issue 12 , Pages 1590-1596, December 2006
Time to Rehabilitation Admission and Associated Outcomes for Patients With Traumatic Brain Injury
Article Outline
Abstract
Kunik CL, Flowers L, Kazanjian T. Time to rehabilitation admission and associated outcomes for patients with traumatic brain injury.
Objective
To examine the association between time from injury to rehabilitation admission and outcomes for patients with traumatic brain injuries (TBIs).
Design
Retrospective chart review.
Setting
One hundred–bed inpatient rehabilitation facility with a 20-bed brain injury unit.
Participants
Patients with TBIs discharged from initial inpatient rehabilitation between 2003 and 2004 (N=158).
Interventions
Not applicable.
Main Outcome Measures
Outcomes examined were functional independence at discharge (motor, cognitive, total FIM scores), rehabilitation length of stay (LOS), and rehabilitation cost.
Results
Significant linear trends were observed for time to admission and motor FIM scores, total FIM scores, rehabilitation LOS, and cost. All linear regression models contained time to admission as a significant predictor of rehabilitation outcomes. Over half of the variability in outcomes was explained by predictors including time to admission and case-mix group or individual FIM scores with the exception of discharge motor FIM score, for which only 45% of the variability was explained.
Conclusions
Patients who progress to rehabilitation earlier do better functionally and have lower costs and shorter LOSs. Furthermore, the time to rehabilitation admission is easily calculated and could be used by rehabilitation providers in adjunct with admission FIM scores to estimate resource utilization.
Key Words: Brain injuries, Length of stay, Outcome assessment (health care), Rehabilitation
RESEARCHERS HAVE SUGGESTED that acute medical management affects rehabilitation outcomes in patients with traumatic brain injuries (TBIs).1, 2 These associations have led researchers to suggest that acute length of stay (LOS) may be an indirect measure of severity. Horn et al3 observed that more severe stroke patients had longer times between the onset of stroke symptoms and admission to rehabilitation. The results regarding the association between acute LOS and rehabilitation LOS, however, are mixed.4, 5
Time to rehabilitation admission has been examined within freestanding rehabilitation centers, and early admission was shown to correlate with improved neurologic outcomes in patients with TBI.6 Significant associations have also been observed between long acute hospital LOSs and higher rehabilitation charges1 and rehabilitation LOS and charges.5 The most powerful predictor of charges was the admission FIM motor score.1 This finding supports the well-coordinated patient-centered approach during the acute management phase, including medical stabilization and preventive measures to improve outcomes in rehabilitation.
Other factors that have been associated with outcomes are age7, 8 and the presence of comorbidities such as pneumonia and recent surgery.9 Deshpande et al9 observed significant associations between history of pneumonia and history of recent surgery and transfer to acute care during rehabilitation. Cifu et al7 found a positive association between age and both rehabilitation LOS and cost of FIM point change. Frankel et al8 observed a significant positive association between age and acute rehabilitation LOS and age and total rehabilitation charges.
The purpose of this study was to examine the association between time to rehabilitation admission and outcomes for patients with TBIs. The outcomes examined were motor, cognitive, and total FIM scores, rehabilitation LOS, and rehabilitation cost. We hypothesized that patients admitted earlier after injury would have better outcomes compared with patients admitted longer after injury. Specifically, patients admitted earlier would have greater function, as indicated by higher discharge FIM scores, and decreased rehabilitation LOS and cost.
Methods
Subject Selection
A retrospective study design was conducted using data from all patients discharged from inpatient rehabilitation for TBIs between January 2003 and December 2004. Rehabilitation was provided at a catastrophic care hospital that specializes in treating patients with TBIs and spinal cord injuries. Rehabilitation was provided in a 20-bed unit according to individualized care plans based on injury and severity. The care plans are based on expected FIM outcomes and LOSs and were developed based on experience learned from the center’s own patient population. The facility provides a full continuum of services including intensive care unit (ICU), acute medical care, and postacute services. Furthermore, a pre-rehabilitation and education program (PREP) is available for patients who are semicomatose or minimally conscious (Rancho Los Amigos Level of Cognitive Functioning Scale [RLA] levels I−III). Admission criteria require that patients have an acquired brain injury, are medically stable, are not on a ventilator, and are 15 years of age or older.
Only patients admitted directly to inpatient rehabilitation for their initial inpatient rehabilitation were included in the analysis. To facilitate the generalizability of the results, patients who were admitted to the ICU, were admitted to the acute medical care unit, or participated in the PREP before receiving their inpatient rehabilitation were excluded from the sample. Initial rehabilitation was defined according to the definition established by the Center for Medicare and Medicaid Services (CMS) as a patient’s first admission to any comprehensive medical rehabilitation program for this impairment.10 Patients were also excluded if they were not living at home before the injury, were not admitted directly from an acute care setting, or had an interruption of longer than 3 days during their inpatient rehabilitation.
A total of 83 patients were excluded from the analysis for reasons as follows: 66 were admitted to ICU, acute medical care, or PREP before inpatient rehabilitation; 7 because of noninitial rehabilitation; 1 was not living at home before injury; 7 were not admitted from an acute care setting; and 2 because of interruptions longer than 3 days. Patients excluded from the analysis were an average of 29.8 years of age, and 75% were admitted to rehabilitation 4 weeks or more postinjury. The average time from injury to admission was 54 days. The population of excluded patients differs from the sample of included patients (54.00d from injury to admission vs 23.57d, respectively); however, the sample included more closely resembles the average time reported by the Traumatic Brain Injury Models Systems: a mean acute LOS of 20 days for the period from 2001 to 2005.11 A total of 158 patients were included in this analysis.
Data Collection
Outcomes are monitored and submitted voluntarily to Uniform Data Systems for Medical Rehabilitation (UDSmr). These data were examined for this study. Data submitted to UDSmr include demographics (age at admission to rehabilitation, sex, marital status, race, primary insurance carrier), injury characteristics (type of injury, onset days [time to rehabilitation admission]), rehabilitation LOS, and rehabilitation functional measures (FIM instrument scores on admission and discharge, case-mix group [CMG], tier).
The FIM is a widely recognized measure of functional ability. Higher FIM scores indicate greater function and independence, whereas lower scores indicate less function and more dependence. FIM scores are based on 18 items rated on an ordinal Likert scale from 1 to 7. Motor function is assessed in 13 items, and cognitive function is assessed in 5 items. The motor and cognitive scales are computed as individual subscale scores and collectively as the total FIM score. The FIM was administered by FIM-credentialed nurses and physical, occupational, and speech therapists within 72 hours of admission to rehabilitation and within 72 hours of discharge from rehabilitation.
CMGs were also used as a measure of functional ability at admission. CMGs are determined based on motor and cognitive admission FIM scores: CMG 0201 (motor, 52−84; cognitive, 24−35), CMG 0202 (motor, 40−51; cognitive, 24−35), CMG 0203 (motor, 40−84; cognitive, 5−23), CMG 0204 (motor, 30−39), and CMG 0205 (motor, 12−29). Generally, as the CMG increases, a patient’s functional ability decreases. The CMS developed weighting factors, referred to as tiers, to account for the effect that certain comorbidities have on resource use. Tier A indicates that no comorbidities are present; tier D, tier C, and tier B, respectively, have comorbidities that require more resources. These comorbidities can be present before a patient’s injury (eg, morbid obesity) or can coexist (eg, multiple fractures).
Other variables including RLA levels on admission, etiology of traumatic injury, and rehabilitation costs were extracted from medical records and the hospital financial information system. The RLA was originally developed to categorize behavior levels for patients with brain injuries.12 The scale ranges from 1 (no response) to 10 (purposeful, appropriate). Patients are generally unable to tolerate rehabilitation based on the CMS guidelines if they are at an RLA level below 3. Rehabilitation cost includes all costs incurred during the rehabilitation LOS.
Data Analysis
We analyzed the data using the SPSSa for Windows statistical package. Time to admission was analyzed as a continuous variable in days and also categorized into 5 groups based on weeks to admission. The outcomes examined for association with time to admission were discharge motor, cognitive, and total FIM scores, LOS (in days), and rehabilitation cost.
We assessed the sample distribution of outcomes for normality using estimates of skewness, kurtosis, and histograms. Kurtosis is an estimate of the degree in which the peak of a distribution deviates from the distribution of a normal curve. A kurtosis between −1 and 1 typically indicates a peak that approximates a normal distribution. We defined a normal distribution as a distribution with skewness between ±1, kurtosis less than 1, and a histogram that visually did not grossly deviate from the distribution of a normal curve.
Scatterplots were generated to visually assess the relation between time to admission in days and all outcomes. Correlations were computed for time to admission in days and all outcome variables, and F tests for linearity were computed for time to admission based on weeks and outcome variables. Statistical significance was established at a P value of less than .05.
We used linear regression with stepwise variable selection to identify predictors of outcomes. This type of regression model is generated by evaluating variables at each step for inclusion if their probability of F is sufficiently small; variables are removed or excluded if their probability of F becomes sufficiently large. Time to admission, age, CMG tier, and RLA level on admission were among the predictors selected for possible inclusion in all regression models, along with admission motor and/or cognitive FIM scores or CMG. The criterion for the inclusion of variables was a P value less than or equal to .05 for the F test. The criterion for elimination of variables from the regression models was a P value greater than or equal to .10 for the F test. Differences in models were examined by comparing the values of the adjusted R2 for each model. Analysis of variance F tests were used to evaluate statistical significance. Statistical significance was established at a P value of less than .05.
We examined all final regression models for goodness of fit and violation of assumptions for linear regression modeling using residual analyses. We used a histogram of standardized residuals and a quantile-quantile plot of observed versus theorized quantiles from normally distributed residuals to graphically assess the assumption that residuals were approximately normally distributed. We used scatterplots of standardized residuals against predicted values to assess the assumptions of constant variance of residuals and to identify potential outliers. We also examined tolerance and variance inflation values to assess multicolinearity of the predictors included in the final models.
Results
Subject Description and Injury Characteristics
Demographic characteristics of the 158-patient sample are presented in table 1. The age distribution was not normal (mean, 28.13y; median, 23y), and most patients were white, single men. Patients were admitted for rehabilitation on average 23.57±14.43 days after the onset of their injury or illness. Time to admission ranged from 3 to 86 days.
Table 1. Demographic Characteristics of Sample
| Characteristic | n | % | Cumulative % |
|---|---|---|---|
| Age group (y) | |||
 14−20 | 59 | 37.3 | 37.3 |
| 21−30 | 45 | 28.5 | 65.8 |
| 31−40 | 24 | 15.2 | 81.0 |
| 41−50 | 19 | 12.0 | 93.0 |
| 51−60 | 9 | 5.7 | 98.7 |
| 61−70 | 2 | 1.3 | 100.0 |
| Sex | |||
| Male | 118 | 74.7 | 74.7 |
| Female | 40 | 25.3 | 100.0 |
| Marital status | |||
| Single | 105 | 66.5 | 66.5 |
| Married | 43 | 27.2 | 93.7 |
| Divorced | 7 | 4.4 | 98.1 |
| Other | 3 | 1.9 | 100.0 |
| Race | |||
| White | 116 | 73.4 | 73.4 |
| Black | 31 | 19.6 | 93.0 |
| Other | 11 | 7.0 | 100.0 |
| Primary insurance carrier | |||
| Blue Cross | 58 | 36.7 | 36.7 |
| MCO or HMO | 40 | 25.3 | 62.0 |
| Medicaid | 38 | 24.1 | 86.1 |
| Workers’ compensation | 21 | 13.3 | 99.4 |
| Commercial | 1 | 0.6 | 100.0 |
Table 2 presents injury characteristics of the sample. Most patients had closed brain injuries, and over half of the injuries were caused by automobile collisions. The CMG distribution was skewed toward more dependent patients; over half of the patients were categorized in the 2 highest CMG levels, 0204 and 0205. Approximately half of the patients were categorized in CMG tier A, indicating they had no significant comorbidities likely to increase the cost or burden of care. However, 27% of patients were categorized in CMG tier B, the highest tier. Over half of the patients in CMG tier D had the comorbidity International Classification of Diseases, Ninth Revision (ICD-9) code 342.90 (hemiplegia, unspecified), and over half of the patients in CMG tier C had the comorbidity ICD-9 code 787.2 (dysphagia). Most patients (88%) in CMG tier B had the comorbidity ICD-9 code V44.0 (tracheostomy status). RLA levels on rehabilitation admission ranged from 4 to 9; however, most patients were admitted at an RLA level of 7 or below.
Table 2. Injury Characteristics of the Sample
| Characteristic | n | % | Cumulative⁎ |
|---|---|---|---|
| Weeks to admission | |||
| <1 | 9 | 5.7 | 5.7 |
| 1 | 31 | 19.6 | 25.3 |
| 2 | 41 | 25.9 | 51.2 |
| 3 | 32 | 20.3 | 71.5 |
| ≥4 | 45 | 28.5 | 100.0 |
| Type of injury | |||
| Closed injury | 48 | 93.7 | 93.7 |
| Open injury | 10 | 6.3 | 100.0 |
| Cause of illness/injury | |||
| Automobile collision | 92 | 58.2 | 58.2 |
| Other | 41 | 25.9 | 84.1 |
| Work related | 21 | 13.3 | 97.4 |
| Crime victim | 3 | 1.9 | 99.3 |
| CMG | |||
| 0201 (motor FIM, 52−84; cognitive FIM, 24−35) | 13 | 8.2 | 8.2 |
| 0202 (motor FIM, 40−51; cognitive FIM, 24−35) | 1 | 0.6 | 8.8 |
| 0203 (motor FIM, 40−84; cognitive FIM, 5−23) | 49 | 31.0 | 39.8 |
| 0204 (motor FIM, 30−39) | 36 | 22.8 | 62.6 |
| 0205 (motor FIM, 12−29) | 59 | 37.3 | 99.9 |
| CMG tier | |||
| A | 81 | 51.3 | 51.3 |
| D | 16 | 10.1 | 61.4 |
| C | 18 | 11.4 | 72.8 |
| B | 43 | 27.2 | 100.0 |
| RLA | |||
| 4 – Confused/agitated | 41 | 25.9 | 25.9 |
| 5 – Confused, inappropriate nonagitated | 58 | 36.7 | 62.6 |
| 6 – Confused, appropriate | 34 | 21.5 | 84.1 |
| 7 – Automatic, appropriate | 21 | 13.3 | 97.4 |
| 8 – Purposeful, appropriate | 2 | 1.3 | 98.7 |
| 9 – Purposeful, appropriate | 1 | 0.6 | 99.3 |
⁎Totals may not equal 100 due to missing data or rounding. |
Functional Independence Measures
Table 3 presents potential predictors and outcomes by time to rehabilitation admission in weeks. All discharge FIM scores were normally distributed. Significant differences were found based on time to admission and all FIM scores at both admission and discharge. Correlations between time to admission and all outcome variables were greater than .40 and were significant, with the exception of discharge cognitive scores. Furthermore, analysis of linearity indicated significant linear trends for each of these variables; however, a scatterplot of time to admission and discharge cognitive FIM scores did not indicate a linear relation. Patients admitted 3 weeks compared with patients admitted 4 weeks or more after onset had similar mean values for potential predictors and outcomes. In the case of admission motor FIM and all outcomes, the mean value was smaller for patients admitted 4 weeks or more postinjury compared with patients admitted 3 weeks postinjury.
Table 3. Potential Predictors and Outcomes by Weeks to Rehabilitation Admission
| Variables | Weeks to Admit | Total | F Statistic | Corr With Days to Admit⁎ | ||||
|---|---|---|---|---|---|---|---|---|
| <1 | 1 | 2 | 3 | ≥4 | ||||
| Potential predictors | ||||||||
| Admit motor FIM score | 51.67±9.22 | 44.61±14.37 | 35.73±15.09 | 31.63±17.33 | 31.60±15.48 | 36.37±16.32 | 21.54† | −.38† |
| Admit cognitive FIM score | 16.78±6.40 | 18.45±7.76 | 15.05±5.72 | 13.41±6.24 | 14.24±6.46 | 15.25±6.67 | 7.36‡ | −.23† |
| Admit total FIM score | 72.33±15.70 | 66.16±20.96 | 53.32±19.52 | 47.59±22.73 | 48.29±21.21 | 54.33±22.09 | 19.82† | −.37† |
| RLA | 5.78±1.30 | 5.43±1.22 | 5.05±0.92 | 5.09±1.09 | 5.44±1.01 | 5.29±1.07 | 0.098 | −.04 |
| Outcomes | ||||||||
| Rehabilitation LOS (d) | 11.33±2.40 | 16.68±10.03 | 22.02±9.17 | 26.81±11.37 | 26.56±13.04 | 22.63±11.68 | 26.68† | .41† |
| Total rehabilitation cost ($) | 12,416.37±22,60.71 | 18,514.10±11,119.34 | 25,353.59±10,972.77 | 31,728.74±31,728.74 | 31,532.65±15,718.54 | 26,325.76±14,222.68 | 30.94† | .44† |
| Discharge motor FIM score | 76.22±4.63 | 71.61±7.25 | 69.00±9.57 | 64.13±10.83 | 61.13±12.47 | 66.70±11.08 | 32.29† | −.44† |
| Discharge cognitive FIM score | 24.89±3.76 | 26.55±5.33 | 24.07±4.76 | 23.06±5.31 | 22.67±5.52 | 24.00±5.29 | 9.17† | −.26† |
| Discharge total FIM score | 107.11±6.07 | 103.65±10.47 | 98.49±11.21 | 92.03±14.75 | 88.47±15.48 | 95.83±14.30 | 36.42† | −.45† |
⁎Spearman ρ calculated because of the nonnormal distribution of days to admission. |
†P<.01. |
‡P<.05. |
Rehabilitation LOS and Cost
The rehabilitation LOS ranged from 4 to 58 days with an average of 22.63±11.68 days, and the rehabilitation cost ranged from $5292 to $69,397 with an average of $26,326±$14,223. Both rehabilitation LOS and cost were approximately normally distributed based on estimates of skewness and kurtosis. Skewness estimates for rehabilitation LOS and cost ± standard error (SE) were .61±.19 and .71±.19, respectively. Kurtosis estimates for LOS and cost ± SE were −.33±.38 and −.09±.38.
Rehabilitation LOS and cost are presented in table 3 by weeks to admission. Significant linear associations were found between groups based on weeks to admission for rehabilitation LOS and cost. Correlations between time to admission and rehabilitation LOS and cost were significant. Patients admitted 3 weeks compared with patients admitted 4 weeks or more after onset had similar mean values, and the mean values were smaller for patients admitted 4 weeks or more compared with patients admitted 3 weeks postinjury.
Linear Regression Analyses
Linear regression models were generated using time to admission, CMG or FIM scores, CMG tier, RLA, and age as predictors of outcomes. Model results and adjusted R2 values are presented in Table 4, Table 5. Table 4 presents the final regression models generated using CMG as a potential predictor, and table 5 presents the final regression models generated using FIM scores as potential predictors. Time to admission was a significant predictor of all outcomes excluding discharge cognitive scores, which were not analyzed because of the nonlinear relation shown in the scatterplot. When compared with the overall explanation of outcome variance, however, admission cognitive and/or motor FIM scores were better predictors than CMG, as indicated by the higher adjusted R2 values shown by all models (see table 5). Regression diagnostics did not show anything problematic.
Table 4. Linear Regression Models Using CMG as a Predictor With Significance Values and Adjusted R2
| Outcome and Predictors | Unstandardized β | SE | Standardized β | P | Adjusted R2 (SE) |
|---|---|---|---|---|---|
| Rehabilitation LOS | 0.56(7.72) | ||||
| (Constant) | 7.42 | 5.42 | .17 | ||
| Days to admission | 0.18 | 0.05 | .22 | .00 | |
| CMG | 5.43 | 0.67 | .56 | .00 | |
| RLA | −1.80 | 0.70 | −.17 | .01 | |
| Total rehabilitation costs | 0.56(9346.68) | ||||
| (Constant) | 5137.47 | 6562.50 | .44 | ||
| Days to admission | 237.76 | 55.98 | .24 | .00 | |
| CMG | 6688.57 | 813.27 | .56 | .00 | |
| RLA | −1855.59 | 852.33 | −.14 | .03 | |
| Discharge motor FIM score | 0.45(8.18) | ||||
| (Constant) | 90.36 | 2.23 | .00 | ||
| Days to admission | −0.23 | 0.05 | −.30 | .00 | |
| CMG | −4.81 | 0.58 | −.51 | .00 | |
| Discharge total FIM score | 0.51(10.023) | ||||
| (Constant) | 128.06 | 2.73 | .00 | ||
| Days to admission | −0.29 | 0.06 | −.29 | .00 | |
| CMG | −6.42 | 0.71 | −.56 | .00 |
Table 5. Linear Regression Models Using FIM Scores as Predictors With Significance Values and Adjusted R2
| Outcome and Predictors | Unstandardized β | SE | Standardized β | P | Adjusted R2 (SE) |
|---|---|---|---|---|---|
| Rehabilitation LOS | 0.67(6.70) | ||||
| (Constant) | 40.88 | 2.01 | .00 | ||
| Days to admission | 0.13 | 0.04 | .16 | .00 | |
| Admission motor FIM score | −0.47 | 0.04 | −.65 | .00 | |
| Admission cognitive FIM score | −0.28 | 0.10 | −.16 | .01 | |
| Total rehabilitation costs | 0.68(8058.66) | ||||
| (Constant) | 47,489.04 | 2412.59 | .00 | ||
| Days to admission | 181.19 | 47.57 | .18 | .00 | |
| Admission motor FIM score | −564.97 | 50.59 | −.65 | .00 | |
| Admission cognitive FIM score | −318.45 | 119.45 | −.15 | .01 | |
| Discharge motor FIM score | 0.52(7.65) | ||||
| (Constant) | 59.22 | 2.25 | .00 | ||
| Days to admission | −0.21 | 0.05 | −.28 | .00 | |
| Admission motor FIM score | 0.40 | 0.04 | .5 | .00 | |
| Discharge total FIM score | 0.55(9.49) | ||||
| (Constant) | 80.76 | 2.84 | .00 | ||
| Days to admission | −0.27 | 0.06 | −.27 | .00 | |
| Admission motor FIM score | 0.43 | 0.06 | .50 | .00 | |
| Admission cognitive FIM score | 0.37 | 0.14 | .17 | .01 |
Discussion
The purpose of this study was to examine the association between time to rehabilitation admission and outcomes for patients with TBIs. Time to rehabilitation admission is significantly related to functional ability (discharge motor, total FIM scores) at discharge and rehabilitation LOS and cost. We hypothesized that patients admitted earlier after injury would have better outcomes compared with patients admitted longer after injury. Data support the hypothesis with regard to functional ability at discharge, LOS, and cost.
Time to admission is a significant predictor of discharge motor and total FIM scores. The total FIM scores and motor FIM scores appear to decrease approximately 1 point for every 4 to 5 days of delayed admission. These findings are similar to findings recently observed in the stroke population. Horn et al3 observed that earlier rehabilitation admission and higher-level activities administered earlier in the rehabilitation process are associated with better outcomes, specifically discharge total and motor FIM scores.
The effect of time to admission is relatively small for cognitive FIM scores, which do not appear to be linearly related to time to admission. The relatively limited impact of time to admission on discharge cognitive FIM scores is not unforeseen given the lack of sensitivity inherent in the scale. The cognitive FIM comprises only 5 items that assess comprehension, expression, social interaction, problem solving, and memory. The total scale ranges from 5 to 35, and the scale is not sensitive enough in some cases to capture subtle but clinically meaningful cognitive improvement. Therefore, this small effect is likely due to the limitations of the scale, and caution should be used in interpreting these results as indicating a lack of an association with time to admission.
CMG appears to be the strongest single predictor of both rehabilitation LOS and cost, as evidenced by the relatively large standardized β estimates compared with other predictors within the same model. Because CMGs were developed for this purpose, this finding was anticipated and has been replicated in other research.2 However, admission motor and cognitive FIM scores appear to predict better independently than the CMG alone. This is evidenced by the increase in adjusted R2 values for models containing individual FIM scores compared with models containing CMGs. This holds true in the prediction of both LOS and cost. Statistically this finding is not unexpected, because continuous measures are known to generally predict better than any categorization of the same measures; the findings are likely due to the loss of precision and power due to the categoric nature of CMGs.13 CMG is simply a categorization of the motor and cognitive FIM scores, and therefore one would expect that the continuous variables—the admission motor and cognitive FIM scores—would predict better than the CMG. Furthermore, Cowen et al1 also observed that models containing the admission motor FIM score and acute LOS contributed significantly to the variance in discharge FIM cognitive (71.2%) and discharge FIM motor scores (69.5%). These values are higher than the adjusted R2 values observed in the present study for discharge motor scores (54%). Although the models in the present study explained less variability, the evidence still suggests that time to admission plays a significant role in the ultimate recovery of function during rehabilitation and suggests that early admission to rehabilitation results in higher function at discharge.
As time to rehabilitation admission increases, the rehabilitation LOS also increases in a linear manner. Based on the estimates obtained, LOS appears to increase approximately 1 day for every 5 to 7 days of delayed admission to rehabilitation. Cowen et al1 observed similar findings, showing that for every 10 days of delayed admission to rehabilitation, rehabilitation LOS increases approximately 1 day.
Similar to rehabilitation LOS, as time to rehabilitation admission increases the rehabilitation cost also increases in a linear manner. Cost appears to increase between $181 and $238 for every day of delayed admission to rehabilitation. Cowen1 also observed that charges increased approximately $212 for each day of delayed admission. Although the present study did not examine charges, this estimate does not seem too different than the observed values, considering the differences in study timeframes (1990s vs 2000s).
An interesting finding is the contribution of the RLA level as a significant predictor of rehabilitation LOS and cost when combined with CMG. The RLA has become widely accepted and is simple to administer; however, the authors are aware of very little research that examines the Rancho as a predictor of outcomes.14 Mayer et al,15 in an analysis of TBI inpatient rehabilitation charges, found a significant, negative correlation between RLA level on admission and average daily rehabilitation charges. However, RLA level on admission strongly and significantly correlated with medical and surgical supply charges and respiratory charges. In addition, from a clinical perspective, understanding the cognitive issues such as memory, attention, insight, and safety awareness is paramount to effectively engaging patients in rehabilitation and improving outcomes.
It is also remarkable that CMG tiers were not significant predictors in any of the models because the tiers were designed to ensure that health care organizations were reimbursed for costly conditions and complications. The lack of significance may be due to the fact that tiers do not accurately measure severity. In patients with brain injury, comorbidities that are likely to affect rehabilitation outcomes include pain, multiple fractures, spasticity, peripheral nerve injuries, skin alterations, and contractures. However, these are not the comorbidities that drive CMG tiering. As Stineman et al16 noted in their discussion of tiering, ICD-9 comorbidities may not be exhaustive, and based on comorbidity coding alone, one cannot distinguish between preexisting conditions and complications. It is plausible that time to admission is a surrogate for severity of both medical complications and/or the severity of the brain injury and therefore is more likely to reflect the presence of these medical complications. Horn et al3 observed that the addition of the Comprehensive Severity Index (CSI) and related components to models predicting stroke outcomes did not considerably increase the variability explained. Horn3 reasoned that other predictor variables correlated sufficiently with CSI to explain outcome variability. This finding is encouraging given that many rehabilitation facilities may not have access to the many variables that compose the CSI but may have access to other predictor variables such as time to admission.
Unlike other studies,7, 8 age was not a significant factor in any of the models. This finding may be due to the age distribution in the present study. The distribution was skewed toward younger patients, and therefore lack of variability in the sample would preclude age from being a factor in the outcomes. In the present study it was not possible to measure the impact of comorbidities such as pneumonia and infection on outcomes because of the low prevalence of these factors. It is possible that patients with these factors were subject to selection bias because of the design of the study. Research has suggested that patients with pneumonia and infection are more likely to be transferred from rehabilitation to acute care and therefore subject to interruptions in rehabilitation.9 If the transfer took longer than 3 days, patients were excluded from the present study. If selection bias is present, the results of the present study may be less generalizable to patients with these conditions.
Study Limitations
A limitation of the present study is the method for cost calculation. The sample contained costs incurred during 2003 and 2004, and no adjustments were made to standardize costs to account for annual increases or adjust cost to current value. However, given the short study timeframe, lack of adjustment should not have had a significant impact on cost estimates.
Conclusions
Time to admission is an important predictor of rehabilitation outcomes in patients with TBIs. Patients who progress to rehabilitation earlier do better functionally and have lower costs and shorter LOSs. Time to admission is easily obtained and could be used by rehabilitation providers in adjunct with admission FIM scores to estimate resource utilization.
Supplier
References
- . Influence of early variables in traumatic brain injury on functional independence measure scores and rehabilitation length of stay and charges. Arch Phys Med Rehabil. 1995;76:797–803
- . Measuring medical complexity during inpatient rehabilitation after traumatic brain injury. Arch Phys Med Rehabil. 2005;86:1108–1117
- . Stroke rehabilitation patients, practice, and outcomes: is earlier and more aggressive therapy better?. Arch Phys Med Rehabil. 2005;86(12 Suppl 2):S101–S114
- . Functional outcome after rehabilitation for severe traumatic brain injury. Arch Phys Med Rehabil. 1995;76:1103–1112
- . The relationship between therapy intensity and rehabilitative outcomes after traumatic brain injury: a multicenter analysis. Arch Phys Med Rehabil. 2003;84:1441–1448
- . Preliminary outcome analysis of a long-term rehabilitation program for severe acquired brain injury. Arch Phys Med Rehabil. 2000;81:1447–1456
- . Functional outcomes of older adults with traumatic brain injury: a prospective, multicenter analysis. Arch Phys Med Rehabil. 1996;77:883–888
- . 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
- . Risk factors for acute care transfer among traumatic brain injury patients. Arch Phys Med Rehabil. 1997;78:350–352
- . Medicare program; prospective payment system for inpatient rehabilitation facilities: proposed rule. 2000;65 Federal Register 66410
- In: Traumatic brain injury facts and figures. Vol 12(1:Birmingham: Traumatic Brain Injury Model Systems National Data Center; 2006;
- . Rancho Los Amigos Scale of Functioning. Downey: Rancho Los Amigos Hospital; 1972;
- . Generalized linear models. London: Chapman & Hall; 1989;
- . Reliability and validity of the Disability Rating Scale and the Levels of Cognitive Functioning Scale in monitoring recovery from severe head injury. Arch Phys Med Rehabil. 1987;68:94–97
- . Characterization and correlates of medical and rehabilitation charges for traumatic brain injury during acute rehabilitation hospitalization. Arch Phys Med Rehabil. 2003;84:242–248
- . Diagnostic coding and medical rehabilitation length of stay: their relationship. Arch Phys Med Rehabil. 1998;79:241–248
- a Version 12.0; SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
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.
PII: S0003-9993(06)01320-7
doi:10.1016/j.apmr.2006.09.001
© 2006 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation. Published by Elsevier Inc. All rights reserved.
Volume 87, Issue 12 , Pages 1590-1596, December 2006
