Relationship Between Duration of Therapy Services in a Comprehensive Rehabilitation Program and Mobility at Discharge in Patients With Orthopedic Problems.

Patients following surgery or those with traumatic orthopedic conditions often enter comprehensive inpatient rehabilitation programs for intensive therapy to restore their functional abilities and improve their chances of returning home. One criterion for admission to an inpatient rehabilitation program is the patient's ability to participate in 3 to 4 hours of therapy per day. According to guidelines set forth by Medicare with respect to inpatient rehabilitation facilities, patients should receive a minimum of 3 hours of therapy per day.[1] Those patients who could tolerate more hours of therapy may receive more therapy if it is available.

Rehabilitation facilities are reimbursed as TEFRA facilities under the Tax Equity and Fiscal Responsibility Act of 1982, rather than under the inpatient prospective payment system (PPS). For patients covered by Medicare, TEFRA facilities receive payments on the basis of reasonable costs per case up to a maximum limit, regardless of the diagnosis, the length of stay, or the amount of therapy a patient receives. This maximum amount is determined by the hospital's cost per discharge in a base year. This cost per discharge is influenced by adjustments for geographical differences such as rural or urban setting and geographical location. The TEFRA system is in contrast to the PPS, in which reimbursement is differentiated based on diagnosis. Hospitals that consistently fall below their maximum TEFRA allowable amount receive bonus payments. Therefore, from the facility's perspective, the least costs occur when therapy services are provided to a Medicare-funded population by treating a large volume of patients for short stays for the minimum allowable hours per day. In this way, the personnel-related costs associated with providing therapy can be minimized while maximizing the Medicare reimbursement. One way to minimize costs to the facility is to decrease the hours of therapy rendered. This can be done by decreasing the amount of therapy per day, the days per week that therapy is received, or the length of stay. Several researchers have demonstrated that the amount of therapy a patient receives is related to outcomes such as independence in ambulation and ambulation distance,[2] independence in activities,[3,4] and chances of discharge to home.[3]

Several researchers have examined the impact of the amount of therapy on rehabilitation outcomes. The amount of therapy can be measured in different ways. For example, it could mean the number of sessions during the course of rehabilitation, the number of days therapy is received, or the amount of time in minutes that therapy is rendered. A study of people with hip fractures in the United States showed that, when the number of acute care physical therapy sessions decreased from 7.6 to 6.3 with implementation of the PPS, both the proportion of patients ambulating independently and the average distance ambulated at the time of discharge were reduced.[2] These researchers, however, did not separate the effect of a decreased length of stay from that of decreased duration of physical therapy services.[2]

In an investigation of the changes that occurred after implementation of PPS, outcomes of patients with acute orthopedic disorders who received 5- and 7-day physical therapy coverage were compared.[5] Increasing the number of days that a patient received physical therapy treatments did not decrease the patients' length of stay, defined as the number of days from admission or surgery to discharge from physical therapy.[5] This study was limited by the fact that although the number of consecutive treatments was increased when 7-day coverage was implemented, the average number of treatments remained the same as for patients treated only during weekdays.

Two studies have demonstrated that the amount of physical therapy a patient receives is related to gains in independence in mobility in an acute care orthopedic population. One study of people with hip fracture showed that patients who received more than one physical therapy treatment per day had 3.98 times greater odds of achieving independence in important functional milestones such as bed mobility, transfers, and ambulation with a walker than patients who received only one session per day and 5.7 times greater odds of being discharged to the home.[3] Another study demonstrated that duration of physical therapy in minutes was an important predictor of functional status at discharge as measured by the Acute Care Index of Function when other variables such as age, length of stay, number of diagnoses, and initial functional status were controlled.[4] This was the only study found in the literature that examined the effect of duration of physical therapy in minutes on discharge outcomes.

The relationship between duration of physical therapy and improved outcomes has been shown to occur in acute care settings,[4] but it is unclear whether this relationship remains after patients are transferred to a comprehensive rehabilitation program. The purpose of our study was to examine the relationship between the total hours of physical therapy and occupational therapy received by patients with orthopedic problems enrolled in a comprehensive rehabilitation facility and their mobility at discharge. Data were available for other variables that might have an effect on mobility at the time of discharge, such as age, mobility at the time of admission, length of stay in rehabilitation, sex, and whether the patient was living independently before injury. These data allowed us to examine the impact of amount of physical therapy while adjusting for the influence of other variables.

Method

Design and Subjects

We conducted a retrospective cohort study of patients with orthopedic problems who were seen by physical therapists and occupational therapists during a comprehensive rehabilitation hospitalization. Most subjects were admitted to this program from a 350-bed acute care hospital located in a culturally diverse metropolitan area. Although our study involved review of existing medical records, informed consent was obtained from the subjects prior to discharge from the rehabilitation program. Data were collected from social service records, quality assurance records, and patient information sheets for 116 patients who were consecutively admitted to the rehabilitation program between August 1996 and February 1997. Patients were included in this study if they: (1) signed an informed consent form and (2) were admitted with a primary orthopedic diagnosis.

Procedure

The physical therapy and occupational therapy staff used the Functional Independence Measure (FIM) to measure physical disability at admission to and discharge from the rehabilitation program. The FIM is part of the Uniform Data Set (UDS),[6] which was developed to enable documentation of the severity of physical disability and measure the outcomes of rehabilitation.[7] Measurements obtained with the FIM have been shown to have good interrater reliability (intraclass correlation coefficient [ICC]=.90-.99) and test-retest reliability (ICC=.90-.93) in a group of patients with mixed diagnoses enrolled in rehabilitation programs.[8] Both the motor subscale and the cognitive subscale of the FIM have been shown to have good internal consistency and have been used to predict length of stay[9] and disability[10] in a group of patients with mixed diagnoses enrolled in rehabilitation programs. Only the mobility subscale was used as an outcome measure in this study because we did not want to include cognitive changes that might have taken place during the stay, although few cognitive changes would be expected in this population. We used the cognitive subscale score in an attempt to control for cognitive ability when investigating the change in mobility subscale scores.

The mobility subscale contains 6 self-care items, 3 transfer items, 2 locomotion items, and 2 bowel and bladder items. These items are scored on a scale from 1 to 7, with 7 representing independence and 1 representing complete dependence or inability to perform a task. Subscale scores were calculated by adding the points for all items. There were no cases of individual item scores being absent. All members of the rehabilitation team involved in scoring the patients at admission and discharge met UDS criteria for administration and scoring of the FIM at the beginning of their employment in the comprehensive rehabilitation unit.

Duration of physical therapy and occupational therapy. The quality assurance records of the rehabilitation department contained information about the total hours of physical therapy and occupational therapy for each subject. The total hours were recorded in quarter-hour increments in the quality assurance records. In the rehabilitation program involved in this study, physical therapists and occupational therapists worked together to meet team goals. The total hours of therapy were recorded only as a compilation of physical therapy and occupational therapy in minutes and were recorded in quarter-hour increments. For subjects with orthopedic problems enrolled in this program, occupational therapy generally involved transfer and self-care training in simulated home environments. Physical therapy generally involved gait training and transfer training with an assistive device and exercises to address strength deficits. All of these activities were measured by the FIM mobility subscale. Because both physical therapy and occupational therapy emphasized goals related to mobility tasks, we concluded that it was appropriate to use the compiled physical therapy and occupational therapy hours as a measure of duration of therapy.

Medical and demographic information. Each subject's age, sex, length of stay, and primary diagnosis were obtained from quality assurance records. These records also contained information on discharge destination. Social service records were used to obtain information on the patient's functional status before injury, which was coded as "independent" or "not independent" based on patient and family reports to the case manager. The UDS records contain information on secondary diagnoses by International Classification of Diseases (ICD-9) codes.[11] All data were located in the social service (case manager) records and quality assurance records.

Descriptive statistics were generated to characterize the subjects. The subjects had a mean age of 72.6 years (SD= 12.0, range=21-99) (Tab. 1). Sixty-nine percent of the subjects were female, 38.0% were married, 81% were of Hispanic origin, and 75% spoke Spanish as a first language. Sixty-nine percent were functioning in the home independently prior to acute care admission (Tab. 2).

Table 1. Subject Characteristics (N= 116): Mean, Standard Deviation, and Range

Characteristic                     bar[X]    SD    Range

Age (y)                             72.6    12.0   21-99

Hours of physical therapy/
occupational therapy                40.5    13.8   14-83

Length of stay (d)                  14.0     4.2    6-27

Number of comorbidities              3.1     1.9    0-9

Initial FIM(a) mobility subscale
score (total possible
score = 91)                         47.1    10.3   23-72

Initial FIM cognitive subscale
score (total possible
score=35)                           31.1     5.6   12-35

Discharge FIM mobility subscale
score-all subjects (total
possible score=91)                  71.6    11.9   23-87

Discharge FIM mobility subscale
score-subjects discharged
home (n=88)                         75.0     8.9   41-87

Discharge FIM mobility subscale
score- subjects not
discharged home (n=25)              59.8    13.8   23-85

Mean change in FIM mobility
subscale scores                     24.5     8.7    0-46

(a) FIM = Functional Independence Measure.

Table 2. Subject Characteristics: Frequency and Percentage(a)

                                  Frequency   Percentage

Female                                79         68.7
Male                                  36         31.3
Married                               41         38.0
Not married                           67         38.5
Country of origin
  North-American                      19         16.8
  Hispanic                            92         81.4
  Other                                2          1.8
Primary language
  English                             26         23.0
  Spanish                             85         75.2
  Other                                2          1.8
Premorbid status
  Independent                         79         68.7
  Not independent                     36         31.3
Premorbid environment
  Home                               109         95.6
  Adult community-living
    facility or boarding               5          4.4
Primary diagnosis
  Femur fracture                      44         38.2
  Total hip replacement               19         16.5
  Total knee replacement              37         32.2
  Below-knee amputation                3          2.6
  Multiple fractures                   3          2.6
  Other orthopedic problem             9          8.1

(a) Some data not available for all subjects.

Fracture of the femur was the most common diagnosis (38.2%), followed by elective total knee replacement (32.2%), elective total hip replacement (16.5%), and other lower-extremity or systemic orthopedic pathology (13.3%) (Tab. 2). The subjects had an average of 3.1 secondary diagnoses.

Subjects were hospitalized for an average of 14.0 days, with the length of hospitalization ranging from 6 to 27 days. The subjects received an average of 40.5 hours of physical therapy and occupational therapy, with a range from 14 to 83 hours (Tab. 1).

Data Collection

We developed a standardized data collection form, operational definitions for all variables, and decision rules for categorizing variables. The forms were used to record and code the data from the quality assurance records, the social service records, and the UDS and FIM records. Data were collected by a group of researchers, all of whom were present at each data collection session, and any discrepancies or questions regarding coding of information were discussed and agreed on by the group.

Data Analysis

All data were managed and analyzed using SAS Version 6.08 statistical software(*) on a VAX mainframe computer. Univariate correlational statistics were calculated to determine relationships among mobility subscale scores at the time of admission, cognitive subscale scores at the time of admission, age, total hours of therapy, number of secondary diagnoses, length of stay, and mobility subscale scores at the time of discharge. Paired t tests were used to determine whether the subjects showed a change between admission and discharge mobility subscale scores. A multiple linear regression was used to explore the relationship between the duration of physical therapy and occupational therapy and mobility at the time of discharge. A regression model was built, including variables found to be related to the dependent variable by univariate correlation using the backward deletion process.

Results

The subjects demonstrated an average mobility subscale score of 47.1 and an average cognitive subscale score of 31.1 at the time of admission. The average mobility subscale score at the time of discharge was 71.6. The subjects showed an average change in mobility subscale scores of 24.5 (P=.0001) (Tab. 1). Subjects who were discharged to their home were different from subjects who were discharged to an institution with respect to their mobility subscale score at the time of discharge (75.0 versus 59.8, P=.0001) (Tab. 1).

A number of variables other than the duration of physical therapy and occupational therapy appeared to influence the subjects' mobility at the time of discharge. The univariate correlation demonstrated that mobility at the time of discharge from the hospital was related to mobility at the time of admission to the rehabilitation program, age, length of stay, baseline cognitive status, and number of secondary diagnoses (Tab. 3). Therefore, these variables were included in the initial regression models to examine and control for their effect. In the multiple linear regression model, we used mobility at the time of discharge as the dependent variable, total hours of physical therapy and occupational therapy as the independent variable, and the confounding variables listed earlier. In addition, we included sex in the original model. An independent t test also demonstrated a difference (P=.0002) in mobility at discharge between subjects who were living at home independently prior to hospitalization ([bar]X=74.7, SD=10.0) and subjects who needed assistance prior to hospitalization ([bar]X=64.5, SD=13.2). A dummy coded variable representing premorbid living situation was also included in the model.

[TABULAR DATA 3 NOT REPRODUCIBLE IN ASCII]

Hours of physical therapy and occupational therapy explained some of the variance in the full regression model (partial correlation=.069, P=.006), as did mobility at the time of admission, cognitive function at the time of admission, length of stay, and number of secondary diagnoses. Sex, premorbid status, and age did not contribute to the model and explained very little of the variance (Tab. 4).

Table 4. Relationship Between Hours of Physical Therapy and Mobility at Discharge (Original Model)(a)

                         Standardized             Partial
Variable                 Coefficient      P     Correlation

Intercept                   42.5        .0001
Mobility at admission        0.62       .0001      .299
No. of secondary
  diagnoses                 -0.92       .0647      .033
Cognitive status at
  admission                  0.34       .0178      .053
Age                         -0.08       .2410      .013
Total hours of therapy       0.25       .0038      .078
Length of stay              -0.59       .0375      .041
Sex                         -0.17       .9157      .000
Premorbid status            -2.73       .1279      .022

(a) [R.sup.2] = .6183.

Using the backward deletion process, we constructed a final model consisting of the variables achieving statistical significance. This model accounted for 60.5% of the variance in mobility at the time of discharge ([R.sup.2]=.605). Duration of physical therapy was a predictor of mobility at the time of discharge (P=.006) and predicted 6.9% of the variance in mobility at the time of discharge when mobility at the time of admission, cognitive status at the time of admission, length of stay, and number of secondary diagnoses were taken into account (Tab. 5). Multicollinearity diagnostics were completed to ensure that correlations between the covariates did not violate the assumptions of the multiple linear regression model.

Table 5. Relationship Between Hours of Physical Therapy and Mobility at Discharge (Final Model)(a)

                         Standardized             Partial
Variable                 Coefficient      P     Correlation

Intercept                   29.2        .0001
Mobility at admission        0.68       .0001      .362
Total hours of therapy       0.26       .0059      .069
Cognitive status at
  admission                  0.18       .0049      .071
Length of stay              -0.56       .0403      .038
No. of secondary
  diagnoses                 -1.06       .0273      .045

(a) [R.sup.2] = .6053.

Discussion

We were able to develop a predictive model that explains 61% of the variance in mobility at the time of discharge in this population. As would be expected, the majority of this variance was explained by mobility at the time of admission (36%), cognitive status at the time of admission (7.1%), and number of comorbid conditions (4.5%). These variables presumably were not modifiable by the rehabilitation team. However, total hours of physical therapy and occupational therapy (6.9%) and length of stay (3.8%) also contributed to the model and were under the control of the rehabilitation team. The results of our study provide evidence to support increasing the number of hours a patient receives both by increasing minutes per day and by increasing length of stay.

The results of our study support other studies of recovery of abilities following orthopedic injury or surgery. Our study demonstrated that total minutes of therapy and length of stay were predictors of mobility at discharge when other important covariates were included in the model. This finding agrees with those of a previous study by Roach et al[4] of patients with orthopedic conditions who received therapy in the acute care setting in which minutes of therapy was an important predictor of mobility at the time of discharge from acute care when length of stay was included in the model. Magaziner and colleagues[12] found with long-term recovery from hip fracture that length of hospital stay and number of therapy sessions were predictors of ability to perform activities of daily living and walking ability 1 year after discharge.

Jette and Jette[13] found that, for patients who received outpatient therapy for knee impairments, total duration of physical therapy in days was not a predictor of health outcomes as measured by all subscales of the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) at completion of therapy. It is difficult to relate findings from all of these studies because the study populations had different levels of acuity, were at different stages of the rehabilitation process, and had different impairments. Perhaps the intensity of therapy, as measured by duration in minutes, is more important in subjects with more severe impairments in an earlier stage of rehabilitation than later in the rehabilitation process. In the cited studies, different measurement tools were used to measure the dependent variable and may have been measuring different constructs.

Cognitive status and comorbidities were 2 other important covariates in our predictive model. We found that cognitive status as measured by the cognitive subscale of the FIM was a predictor of mobility at the time of discharge. This finding is similar to those of studies of long-term recovery at 1 year of patients with hip fractures[12,14] and outcomes at completion of rehabilitation.[13] The number of comorbidities was also a predictor of mobility in our model. This finding is in agreement with those of researchers who found that the number of comorbidities predicts mobility scores at the time of discharge from the hospital, SF-36 physical function scores at completion of rehabilitation, and activities of daily living and walking ability 1 year after discharge.[4,13,15]

In contrast to the findings of other studies, age was not a predictor in our model. Age has been found to be a predictor of long-term health outcomes after hip fracture,[12,14] of short-term outcomes after outpatient rehabilitation for knee rehabilitation,[13] and of short-term outcomes of patients who received therapy in acute care for orthopedic conditions.[4] It is unclear why age was not a predictor of mobility in our group of subjects.

In a randomized, controlled clinical trial comparing 2 groups of subjects with total hip or knee arthroplasty,[14] one group began intensive rehabilitation on day 3 postoperatively and one group began intensive rehabilitation on day 7. Subjects in the early rehabilitation group had a shorter length of stay by 2.8 days, received more therapy sessions as evidenced by higher physical therapy costs by $393 and higher occupational therapy costs by $159, had lower total hospital costs by $1,871, and had higher FIM mobility subscale scores at the end of the hospital stay than subjects in the late rehabilitation group. These findings suggest that more therapy earlier in the rehabilitation process can improve health outcomes and reduce costs. Although our study had dissimilar findings with respect to length of stay, our results suggest that more therapy improves health outcomes.

All except one of the subjects who participated in our study were admitted to acute care from the community. The subjects who returned home had higher mobility subscale scores at the time of discharge ([bar]X=75.03, SD=8.94; n=88) than those who were discharged to nursing homes, assisted-living facilities, or acute care ([bar]X=69.8, SD=13.8; n=25) (Tab. 6). The average score of those subjects returning home (75.0) was still well below the maximum possible mobility subscale score (91). This finding suggests that patients still are quite dependent when they are discharged to the home environment. The lower mean mobility subscale score at the time of discharge of those subjects who did not return home (59.8) suggests that their limitations were so severe that they were unable to safely return to the community.

Table 6. Mobility Scores by Discharge Destination

Discharge
Destination          Frequency   Percentage   [bar]X    SD

Home alone              24          21.7       76.1    11.0
Home with family        64          56.5       74.6     8.1
Adult
  community-living
  facility               5           4.3       71.4    10.4
Skilled nursing
  facility              14          12.2       58.5    13.5
Acute care               6           5.2       53.0    12.6

One recent study of variables related to functional decline for older individuals living in the community identified low exercise level, low social function, and slow gait as predictors of functional decline, defined as a loss of ability to perform activities of daily living at 4 years.[16] Other researchers[17] found reduced lower-extremity performance, low physical activity, and low frequency of social contact to be risk factors for functional decline. Individuals who are discharged from rehabilitation with functional limitations presumably will have decreased activity and exercise levels, decreased lower-extremity performance, and slow gait. They may also have less social contact because participation in social activities may be more difficult.

Our study had several limitations. The population under consideration had a very high proportion of Hispanic subjects (81.4%), and they may differ from patients seen in other hospitals in other regions of the country. Cultural differences in social support structures, such as availability of family members to provide physical support, may also have an impact on the patient's ability to return to the home environment. The results of our study are difficult to generalize to more culturally varied patient populations. The mean length of time from onset of the medical event to rehabilitation admission and the mean length of stay in rehabilitation have shown a steady decrease in the last decade. In 1990, the mean length of time from onset to admission was 26 days, and the mean length of stay was 28 days. In 1996, when these data were collected, the mean length of time from onset to admission was 15 days, and the mean length of stay was 18 days. In 1998, these values were further reduced to a mean of 13 days from onset to admission and 16 days for mean length of stay.[18] These findings suggest that this trend is continuing, and the results of this study and previous studies must be interpreted in the context of this trend.

Conclusion

The results of this study indicate that, controlling for initial level of functioning, the hours of physical therapy and occupational therapy a patient with orthopedic diagnoses receives during a comprehensive rehabilitation program predicts mobility at the time of discharge as measured by the mobility subscale of the FIM. Other variables that were important predictors of mobility were cognitive status at admission, length of stay, and number of secondary diagnoses. Additionally, patients who returned to their homes had higher FIM mobility subscale scores than patients who were discharged to nursing homes, adult community-living facilities, and acute care settings. These results suggest that increasing the amount of therapy patients with orthopedic diagnoses receive in comprehensive rehabilitation may increase the chances of discharge to the home by improving mobility skills.

(*) SAS Institute Inc, PO Box 8000, Cary, NC 27511.

References

[1] Coverage of services (chapter 2). In: Medicare Intermediary Manual, Claims Process, Part 3. Baltimore, Md: US Dept of Health and Human Services, Health Care Financing Administration. HCFA publication 13-3.

[2] Fitzgerald RF, Moore PS, Dittus RS. The care of elderly patients with hip fracture: changes since implementation of the prospective payment system. N Engl J Med. 1988;319:1392-1397.

[3] Guccione AA, Fagerson TL, Anderson JJ. Regaining functional independence in the acute care setting following hip fracture. Phys Ther. 1996;76:818-826.

[4] Roach KE, Ally D, Finnerty B, et al. The relationship between duration of physical therapy services in the acute care setting and change in functional status in patients with lower-extremity orthopedic problems. Phys Ther. 1998;78:19-24.

[5] Holden MK, Daniele CA. Comparison of seven- and five-day physical therapy coverage in patients with acute orthopedic disorders. Phys Ther. 1987;67:1240-1246.

[6] Guide for the Uniform Data Set for Medical Rehabilitation, Version 5.0. Buffalo, NY: State University of New York at Buffalo; 1996.

[7] Granger CV. The emerging science of functional assessment: our tool for outcomes analysis. Arch Phys Med Rehabil. 1998;79:235-240.

[8] Ottenbacher KF, Ysu Y, Granger CV, Fiedler RF. The reliability of the Functional Independence Measure: a quantitative review. Arch Phys Med Rehabil. 1996;77:1226-1232.

[9] Heinemann AW, Linacre JM, Wright BD, et al. Prediction of rehabilitation outcomes with disability measures. Arch Phys Med Rehabil. 1994;75:133-143.

[10] Linacre JM, Heinemann AW, Wright BD, et al. The structure and stability of the Functional Independence Measure. Arch Phys Med Rehabil. 1994;75:127-132.

[11] International Classification of Diseases, Ninth Revision. Geneva, Switzerland: World Health Organization; 1980.

[12] Magaziner J, Simonsick EM, Kashner TM, et al. Predictors of functional recovery one year following discharge for hip fracture: a prospective study. J Gerontol. 1990;45:M101-M107.

[13] Jette DU, Jette AM. Physical therapy and health outcomes in patients with knee impairments. Phys Ther. 1996;76:1178-1187.

[14] Munin MC, Rudy TE, Glynn NW, et al. Early inpatient rehabilitation after elective hip and knee arthroplasty. JAMA. 1998;279:847-852.

[15] Koval KJ, Skovron ML, Aharanoff GB, Zuckerman JD. Predictors of functional recovery after hip fracture in the elderly. Clin Orthop. 1998;348:22-28.

[16] Sarkisian CA, Liu H, Gutierrez PR, et al. Modifiable risk factors predict functional decline among older women: a prospectively validated clinical prediction toll. The Study of Osteoporotic Fractures Research Group. J Am Geriatr Soc. 2000;48:170-178.

17 Stuck AE, Walthert JM, Nikolaus T, et al. Risk factors for functional status decline in community-living elderly people: a systematic literature review. Soc Sci Med. 1999;48:445-469.

[18] Update on change in length in stay for medical rehabilitation patients and impact on outcomes. AMRPA Magazine. 2000;3(10): 39-40.

NJ Kirk-Sanchez, PT, MSPT, is Assistant Professor, Department of Physical Therapy, School of Health, College of Health and Urban Affairs, Florida International University, University Park, Miami, FL 33199 (USA) ([e-mail]). She was a doctoral student, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, Coral Gables, Fla, when this work was completed. Address all correspondence to Ms Kirk-Sanchez.

KE Roach, PT, PhD, is Associate Professor, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine.

Both authors provided concept/research design, writing, and data analysis. Ms Kirk-Sanchez provided subjects, data collection, project management, and institutional liaisons.

This study was approved by the University of Miami Medical Sciences Subcommittee for the Protection of Human Subjects in Research and the institutional review board of the hospital from which the cohort was drawn.

This article was submitted November 8, 1999, and was accepted September 11, 2000.