- Email: [email protected]
Ratings of physical function obtained by interview are legitimate for patients hospitalized after stroke
Ratings of Physical Function Obtained by Interview are Legitimate for Patients Hospitalized After Stroke Ketan S. Sadaria, BS, Richard W. Bohannon, MS, EdD, Nora Lee, MD, and Rose Maljanian, BS, MBA
This study was conducted to examine the legitimacy of Barthel Index (BI) scores and subset of Functional Independence Measure (FIM) scores obtained by interview from patients hospitalized after stroke. This study included 82 patients with stroke. Interviews by a trained study coordinator were conducted to obtain BI and FIM subset (transfer, locomotion, feeding) scores from patients or a knowledgeable proxy. In addition, therapists’ observational FIM scores on subset items recorded in the medical record were retrieved. The reliability, responsiveness, and validity of the BI and FIM subset were examined using Cronbach’s alpha, effect size calculations, Wilcoxon tests, Spearman (rS) correlations, and regression analysis. Statistical analysis indicated that the post-stroke interview BI (.952) and FIM subset (.939) scores had excellent internal consistency. Large effect sizes and Wilcoxon test results between the prestroke and poststroke interview scores for the BI and FIM subset (z, ⫺3.739 to ⫺6.168) indicated that these 2 instruments were responsive to changes in patients’ physical function status that accompanied stroke. Excellent correlations were found between BI and FIM interview scores poststroke (rs, .913 to .971). The FIM subset interview scores and therapists’ scores showed correlations that were moderate to good (rs, .508 to .754). Length of stay correlated significantly with both the interview BI score (rs, ⫺.604) and the interview FIM subset score (rs, ⫺.583). Length of stay was correlated also with the National Institutes of Health Stroke Scale scores (rs, .484). The regression analysis depicted the BI interview scores as the best predictor of the length of stay. Results provide support for the use of functional scores obtained by interview from hospitalized stroke patients. Both the BI and the FIM subset examined in this study are legitimate for this purpose. Key Words: Activities of daily living—Measurement—Outcomes—Stroke. Copyright © 2001 by National Stroke Association
Cerebrovascular accident (stroke) is the leading cause of serious long-term disability in the United States.1 Each year, approximately 550,000 people experience stroke in
From the Department of Physical Therapy, School of Allied Health, University of Connecticut, Storrs; and Institute of Outcomes Research, Department of Neurology, Hartford Hospital, Hartford, CT. Received February 16, 2000; accepted March 9, 2001. Address reprint requests to Richard W. Bohannon, MS, EdD, PT, NCS, Department of Physical Therapy, School of Allied Health, U-2101, University of Connecticut, Storrs, CT 06269-2101. Copyright © 2001 by National Stroke Association 1052-3057/01/1002-0008$35.00/0 doi:10.1053/jscd.2001.24656
the US, 150,000 die, and 300,000 people experience disability.2 Although the mortality from stroke has decreased significantly in the past 30 years, the number of strokes has not decreased. As a result, the number of stroke survivors with chronic disability has increased.3,4 Approximately 3 million Americans are currently living with varying degree of disability because of stroke. The annual cost for stroke in the US is estimated to be $30 billion.2 This economic burden is primarily attributed to disabilities that prevent patients with stroke from returning to their previous level of independence.1,2 Numerous measures of disability exist. However, only 2 were recommended for use with patients with stroke by the Agency for Health Care Policy and Research (AH-
Journal of Stroke and Cerebrovascular Diseases, Vol. 10, No. 2 (March-April), 2001: pp 79-84
79
K.S. SADARIA ET AL.
80 2
CPR). The 2 recommended by the AHCPR were the Barthel Index (BI)5 and the Functional Independence Measure (FIM).6 We searched the Medline and Cumulative Index of Nursing and Allied Health (CINHAL) databases and found that these 2 instruments have been widely used for research between 1982 and 1999. Using the terms “stroke” and the “Barthel Index,” we found 234 studies listed in Medline and 248 studies listed in CINAHL. Simultaneously, using the terms “stroke” and “FIM,” we found 70 studies listed in Medline and 178 studies listed in CINAHL. The BI was initially developed in 1955 for use with chronic diseases.5 However, the BI was not published until 1965. Since then, the BI has been used extensively to document the physical function of patients with stroke. The BI scale consists of 10 items (feeding, chair/bed transfer, grooming, toilet use, bathing, ambulation, stair climbing, dressing, bowel control, and bladder control). In the original BI scale, item scores ranged from 0 to 5, 0 to 10, or 0 to 15, depending on the items. A score of 100 indicated full competence in all areas. Collin et al6 proposed an alternative scoring system in which BI item scores range from 0 to 1, 0 to 2, or 0 to 3, depending on item, with maximum total score of 20. The FIM7 was first described in 1984. It was designed to be used as a universal assessment tool in the Uniform Data System for medical rehabilitation. The FIM was developed in an effort to provide comprehensive assessment of both physical and cognitive disability. Thus, the FIM has 2 main domains, motor and cognitive. The motor domain has 4 subscales (self-care, sphincter control, transfers, and locomotion), whereas the cognitive domain has 2 subscales (communication and social cognition). The FIM scale comprises 18 items (13 motor and 5 cognitive), each of which is scored on a 7-point scale (1 being completely dependent and 7 being fully independent). Thus, total FIM score ranges from 18 to 126. The BI and the FIM both have been found to be valid and reliable in assessing patients’ functional status.6,8-11 Available literature indicates that BI correlates very well with the FIM.12,13 However, Kidd et al.9 found that FIM is more sensitive to changes in functional status than the BI. Shah et al.14 revised the original BI to improve its limited sensitivity. Comparison of these functional scales indicates that BI can be derived from the motor items of the FIM.15 Both the BI and the FIM have been applied using a variety of methods. Methods used with the BI include direct observation, postal questionnaire (self-report), telephone interview, and interview of patients and their care takers or nurses. These different application methods have been found to be reliable and valid.16,17 Scoring of function with the FIM has been achieved using observation, self-reported questionnaire, and telephone interview methods, which also have been found to be reliable
18,19
and valid. However, available literature reveals very little about the legitimacy of functional scores derived from interviews in an acute care setting. In acute care, the time that can be devoted to activities such as outcome assessment is limited; outcome assessments of patients with stroke are no exception. The time required to apply instruments such as the BI and FIM, if applied in their entirety, may be problematic. Obtaining information by interview is one way of reducing time requirements and using a subset of items is another. These approaches, however, cannot be undertaken lightly or recommended without testing their legitimacy. To be considered legitimate in a particular application, a measure must be reliable, responsive, and valid. The validity of a measure can be established in several ways but is often showed by its correlation with another measure of the same construct or with a different measure of known importance. One of the critical aspects of the acute care hospitalization is length of stay in the hospital. In today’s costconscious health care environment, hospital length of stay is an important rehabilitation outcome from both a clinical and an administrative point of view. Brosseau et al.20 noted that predictions of rehabilitation length of stay are helpful in managing admissions, planning stroke programs according to patient characteristics, and projecting discharges; predictions of length of stay also provide a basis for payment for rehabilitation services. Research conducted several years ago indicated that an inverse relationship exists between patients’ functional status quantified using FIM scores and length of stay on an acute rehabilitation unit.21 Research has also shown patient age to be a predictor of the length of stay.20 With the potential of managed care to drive length of stay, it is unknown whether patients’ function remains a predictor of this important variable. Therefore, the relationship between functional status and acute care length of stay needs to be investigated. In addition to the BI and FIM, other scales such as National Institutes of Health Stroke Scale (NIHSS),22 which is known to predict other outcomes,23,24 should also be investigated for their ability to predict length of stay. The purpose of this study was to examine the legitimacy of BI and FIM scores obtained by interview. Specifically, the purpose was to examine reliability, responsiveness, and validity of BI scores and a subset of FIM scores obtained by interview. Two aspects of the validity were tested, convergent construct validity and predictive validity. This last aspect was conducted in conjunction with the NIHSS.
Methods This study was a part of a larger prospective outcome study of patients with stroke admitted to the neurologic
RATINGS OF PHYSICAL FUNCTION AFTER STROKE
81
therapist and had received his/her first therapy treatments. This interview included documentation of patients’ functional status before stroke (prestroke) and after first visit (poststroke) using both the BI and the FIM subset. The project coordinator used a standardized data form to record information regarding patients’ functional status. The scores for the FIM transfer, locomotion, and feeding were also obtained from the medical record where therapists’ direct observation of patients’ functional status was recorded. For this study, the NIHSS was completed within 24 hours of admission by a resident physician who first participated in a training and certification process for use of the instrument.
unit of the Hartford Hospital, Hartford, CT. This study, which was approved by the hospital’s institutional review board, was primarily descriptive and correlational. Subjects of the study included all the stroke patients whose mental and emotional status allowed participation, who provided informed consent, who completed the BI and FIM interviews, and who were scored using the NIHSS. Of 169 patients admitted to the unit over a 6-month period, 82 subjects met these criteria. Instruments The BI and FIM were the main instruments used in this study. For the BI, Wade and Collin’s version8 was used. A total BI score was derived by adding scores from individual items; the potential range of the scores was 0 to 20. Of the 18 FIM items, a subset of 3 physical items that corresponded with BI items was selected. Specifically, the FIM subset included the bed-to-chair transfer, locomotion, and feeding items. A total of these individual subset items was derived that had a potential range from 3 to 21. The NIHSS was used to quantify the neurologic deficits of the acute stroke patients. The NIHSS consists of 15 items, which includes consciousness, vision, extraocular movement, facial palsy, limb strength, ataxia, sensation, speech, and language. It is an ordinal scale whose item scores range from 0 to 2, 0 to 3, or 0 to 4, depending the items. A total score was derived by adding individual scores; this yielded a potential total score of 0 or “normal” to 42. The NIHSS has been found to be valid and reliable.25,26
Statistical Analysis Descriptive and inferential statistics were calculated using the Systat software (Chicago, IL) program. Cronbach’s alpha was used to check the reliability (internal consistency) of the functional measures. The responsiveness of the 2 scoring instruments was described using effect size27 and Wilcoxon matched pairs signed-ranks tests. The relationships between the length of stay and BI, FIM, and NIHSS scores were determined using Spearman correlations (rs). Regression analysis was performed to determine the relative value of the BI, FIM subset, and NIHSS scores as predictors of outcome (length of stay in the hospital). The critical level of significance for this study was P ⱕ .05.
Results Procedure
The patients’ mean (⫾ standard deviation) age was 70.8 (⫾ 13.4) years. Table 1 summarizes BI and FIM (subset) scores for prestroke and poststroke physical function obtained by poststroke interview. Table 1 also presents data obtained from therapists’ notes, which indicate patients’ observed functional status. The values of
A trained project coordinator interviewed patients (or if not possible, a knowledgeable proxy) regarding each patient’s physical functioning. This interview was conducted within 48 hours of admission after each patient had been evaluated by a physical and an occupational
Table 1. Descriptive statistics of BI and FIM scores
BI (interview) Prestroke No. of patients (n) Range Minimum (total) Maximum (total) Mean Median Standard deviation
82 0-20 6 20 19.3 20 2
FIM: subset (interview)
Poststroke
Prestroke
Poststroke
82 0-20 0 20 14.9 18 6.5
82 3-21 11 21 20.4 21 1.7
82 3-21 3 21 16.3 20 6.2
FIM: subset (clinical observation) Prestroke
Poststroke 47 3-21 3 21 12.7 14 5.7
Difference: prestrokepoststroke (interview) BI 82 0-20 0 17 4.4 1 4.8
FIM 82 0-18 0 18 4.1 1 5.6
K.S. SADARIA ET AL.
82
Table 2. Results of Wilcoxon matched pairs signed-ranks tests between prestroke and poststroke scores for BI and FIM
Table 4. Spearman correlations between length of stay and BI and FIM subset items BI
Scoring item
BI Wilcoxon Z
FIM Wilcoxon Z
Transfer Mobility Feeding Stairs Grooming Bowel Bladder Toilet use Dressing Bathing Total
⫺4.543 ⫺4.940 ⫺4.235 ⫺5.543 ⫺4.472 ⫺3.739 ⫺3.906 ⫺4.179 ⫺5.161 ⫺5.568 ⫺6.163
⫺4.878 ⫺5.692 ⫺4.472
⫺6.168
FIM
Item
rs
P
rs
P
Transfer Mobility Feeding Stairs Grooming Bowel Bladder Toilet use Dressing Bathing Total
⫺0.584 ⫺0.540 ⫺0.555 ⫺0.356 ⫺0.464 ⫺0.631 ⫺0.612 ⫺0.642 ⫺0.478 ⫺0.353 ⫺0.604
⬍.001 ⬍.001 ⬍.001 ⬍.002 ⬍.001
⫺0.550 ⫺0.528 ⫺0.577
⬍.001 ⬍.001 ⬍.001
⬍.001 ⬍.001 ⬍.001 ⬍.002 ⬍.001
⬍.001
⫺0.583
⬍.001
NOTE. All Z statistics are significant at P ⬍ .001.
the reliability (internal consistency) of the instruments (measures) demonstrated by the Cronbach’s alpha were as follows: prestroke interview BI ⫽ 0.828, prestroke interview FIM (subset) ⫽ 0.660, poststroke interview BI ⫽ 0.952, poststroke interview FIM (subset) ⫽ 0.939 and poststroke observation FIM (subset) ⫽ 0.903. Calculated effect sizes and the results of the Wilcoxon tests supported the responsiveness of the BI and the FIM subset obtained by interview. The effect sizes were large, 2.2 for the BI and 2.4 for the FIM subset. The results of the Wilcoxon tests between prestroke and poststroke BI and FIM scores showed that all individual items as well as total scores differed significantly. Table 2 summaries results of the Wilcoxon tests. These values clearly indicate that both the BI and FIM are capable of detecting change in the patients’ physical function status between baseline and post stroke. The Spearman correlations between BI and the FIM scores are presented in the Table 3. The correlation between the prestroke total scores of these instruments was
Table 3. Spearman correlations between BI and FIM scores obtained by interview Activity
Correlation*
Prestroke Total BI score v FIM subset Poststroke Total BI score v FIM subset Transfer Locomotion Feeding * All correlations are significant at P ⬍ .001.
0.645 0.971 0.913 0.933 0.939
moderate (0.645). However, the correlations between the poststroke BI and FIM scores were excellent (0.913 to 0.971). The Spearman correlation between FIM scores for interview and therapist observation indicated a moderate to good correlation between these 2 methods of administration. These correlations ranged from .508 to .753 for individual items. The correlation between the total FIM subset scores obtained by interview and observation was .751. All the correlations were significant at P ⬍ .001. Correlations between the hospital length of stay and the BI as well as the FIM subset interview scores are shown in the Table 4. The correlation had a range of ⫺.353 to ⫺.642. All were significant at P ⬍ .001 or P ⬍ .002. The correlations between hospital length of stay and FIM subset scores obtained by interview (⫺.583) were higher than those obtained by the therapists and recorded in the medical record (⫺ .502). The NIHSS was also moderately correlated with the length of stay (.484). Regression analysis, with length of stay as dependent variable, identified the BI score as the best predictor. It showed the FIM subset and NIHSS scores to add no further explanation of the length of stay.
Discussion This study of patients with stroke was designed to examine legitimacy of physical function ratings obtained by interview in acute care. The BI and FIM subset were the primary instruments used to check patients’ physical function as outcome measure. Consistent with the research on the reliability of the BI and FIM,6,9,11 this study found excellent internal consistency in the both scoring systems. The excellent internal consistency of interview method of the BI and the FIM subset provides support for reliability and advocates use of these time efficient scor-
RATINGS OF PHYSICAL FUNCTION AFTER STROKE
ing system to measure patient’s physical function in the acute care setting. The results comparing prestroke and poststroke values indicate that both the BI and FIM subset are responsive measures. The effect size values as well as individual item Wilcoxon test results indicate that the FIM subset possesses slightly more responsiveness than BI among the acute care patients. This greater responsiveness of FIM is consistent with the available research.6,12 Although only a subset of the FIM (3 items) was used, it remained more responsive than the BI. We would attribute this responsiveness to its reliance on a scoring system with a broader range of scores (1 to 7) than the BI for each item. The significant correlations in this study between BI and FIM scores of the specific activities (transfer, locomotion, and eating) are consistent with published research on BI and total FIM scores.9,12,28 The significant correlations provide support for the convergent construct validity of these instruments and the use of interviews to obtain BI and the FIM subset scores for patients hospitalized after stroke. However, moderate correlations between FIM scores obtained by interview and clinical observation raise some concern regarding previously published research.29 Further research needs to be conducted to evaluate the relationship between scores obtained using the interview and observation methods in the acute care setting. This study of patients hospitalized after stroke also yielded results similar to those of previous studies of patients’ functional status and length of stay.20,21 Patients’ functional status was inversely related to length of stay in this study. That is, those with lower BI or FIM sunset scores had longer lengths of stay. Severity of neurologic involvement as measured by the NIHSS was positively related and served as a moderate predictor of length of stay. Thus, patients who were more involved neurologically had longer lengths of stay. It is widely accepted that the length of stay can be affected and predicted by using many factors.20,21,30-32 Further research is warranted to evaluate and establish a definite relationship between patients’ physical function and length of stay in the hospital. The baseline performance of patients with stroke is typically documented by interview of a patient or a significant other. Such information is important as an indicator of potential for recovery and is therefore useful for setting goals and for planning the rehabilitation process.33 Documentation of patients’ physical function after stroke is a necessity for quantifying outcome of the rehabilitation process as well as for reimbursement. Searches of the medical records conducted in conjunction with this study suggest that attempts to judge acute outcomes on the basis of recorded observations will be hampered by a
83
lack of complete data, even when a small set of items is used to characterize function. Although the importance of the observation method is widely accepted in the rehabilitation profession, the data and statistical analysis in this study reinforced the value of data from interviews and use of a subset of instrument items. The established reliability, responsiveness, and validity clearly advocate use of interview and subset scoring system (for BI and FIM) as legitimate outcome measures for rating the physical function of patients with stroke in acute care. However, further research for interview method needs to be performed using BI and FIM scale in its entirety as well as an alternative subsets of items.
References 1. National Stroke Association: Recovery and Rehab. www.stroke.org/NS804.0, 2001 2. Post-Stroke Rehabilitation. Clinical Practice Guidelines. Rockville, MD: U.S. Department of Health and Human Services. Public Health Service, Agency for Health Care Policy and Research, 1995. 3. Shahar E, McGovem PG, Sparfka JM, et al. Improved survival of stroke patients during 1980s. Stroke 1995;26: 1-5. 4. Modam B, Wagener DK. Some epidemiological aspects of stroke: Mortality/morbidity trends, age, sex, race, socioeconomic status. Stroke 1992;23:1230-1236. 5. Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Maryland State Med J 1965;14:61-65. 6. Collin C, Wade DT, Davies S, et al. The Barthel ADL index: A reliability study. Inter Disabil Stud 1988;10:6163. 7. Granger CV, Hamilton BB, Keith R. Advance in functional assessment for medical rehabilitation. Top Geriat Rehabil 1986;1:59-74. 8. Wade DT, Collin C. The Barthel ADL Index: A standard measure of disability? Inter Disabil Stud 1988;10:64-67. 9. Kidd D, Stwart G, Baldry J, et al. The functional independence measure: A comparatative validity and reliability study. Disabil Rehabil 1995;17:10-14. 10. Hamilton BB, Laughlin JA, Fiedler RC, et al. Interrater reliability of the 7-level functional independence measure (FIM). Scand J Rehabil Med 1994;26:115-119. 11. Granger CV, Cotter AC, Hamilton BB, et al. Functional assessment scales: A study of persons after stroke. Arch Phys Med Rehabil 1993;74:133-138. 12. Bohannon RW. Scoring transfer and locomotion independence of home care patients: Barthel versus Functional Independence Measure. Int J Rehabil Res 1999;22: 65-66. 13. Sharrack B, Hughes RAC, Soudain S, et al. Psychometric properties of clinical rating scales used in multiple sclerosis. Brain 1999;122:141-159. 14. Shah S, Frank V, Cooper B. Improving the sensitivity of the Barthel Index for stroke rehabilitation. J Clin Epidemiol 1989;42:703-709. 15. Nyein K, McMichael L, Turner-Stokes L. Can a Barthel score be derived from the FIM? Clin Rehabil 1999;13:5663.
84 16. Gompertz P, Pound P, Ebrahim S. A postal version of Barthel Index. Clin Rehabil 1994;8:233-239. 17. Yeo D, Faleiro R, Lincoln NB. Barthel ADL Index: A comparison of administration methods. Clin Rehabil 1995;9:34-39. 18. Smith PM, Illing SB, Feilder RC, et al. Intermodal agreement of follow-up telephone functional assessment using the Functional Independence Measure in patients with stroke. Arch Phys Med Rehabil 1996;77:431-435. 19. Hoeing H, Branch LG, McIntyre L, et al. The validity in person with spinal cord injury of self-reported functional measure derived from the Functional Independence Measure. Spine 1999;24:539-544. 20. Brosseau L, Philippe P, Potvin L, et al. Post-Stroke inpatient rehabilitation: Predicting length of stay. Am J Phys Med Rehabil 1996;75:422-430. 21. Bohannon RW, Cooper J. Admission function portends discharge function and length of stay after stroke. J Stroke Cerebrovasc Dis 1994;4:235-237. 22. Brott T, Adams HP, Olinger CP, et al. Measurement of acute cerebral infarction: A clinical examination scale. Stroke 1989;20:864-870. 23. Adams HP, Davis PH, Leira EC, et al. Baseline NIH stroke scale score strongly predicts outcome after stroke. Neurology 1999;53:126-131. 24. Roth EJ, Heinemann AW, Lovell LL, et al. Impairment and disability: Their relation during stroke rehabilitation. Arch Phys Med Rehabil 1998;79:329-335.
K.S. SADARIA ET AL. 25. Wtyyk RJ, Pessin MS, Kaplan RF, et al. Serial assessment of acute stroke patient using NIH stroke scale. Stroke 1994;25:362-365. 26. Goldstein LB, Bertels C, Davis JN. Interrater reliability of the NIH stroke scale. Arch Neurol 1989;46:660-662. 27. Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status. Medical Care 1989;27: S178-189. 28. Fricke J, Unsworth CA. Inter-rater reliability of the original and modified Barthel Index, and comparison with the Functional Independence Measure. Aust Occup Ther J 1996;43:22-29. 29. Brosseu L, Philippe P, Boulanger YL. The Functional Independence Measure: Validity of selected method applied to stroke survivors. Top Geriat Rehabil 1995;11:75-85. 30. Brock K, Robinson P, Simondson J, et al. Prediction of hospital length of stay following stoke, J Qual Clin Pract 1997;17:37-476. 31. Hakim EA, Bakheit AMO. A study of the factors which influence the length of stay of stroke patients. Clin Rehabil 1998;12:151-156. 32. Tirschwell DL, Kukull WA, Longstreth WT. Medical complications of ischemic stroke and length of stay: Experience in Seattle, Washington. J Stroke Cerebrovasc Dis 1999;8:336-343. 33. Cushman LA, Poduri KR, Palenski C. Predicting discharge functional status and rehabilitation efficiency from preadmission functional assessments. J Stroke Cerebrovasc Dis 1995;5:33-38.
This study was conducted to examine the legitimacy of Barthel Index (BI) scores and subset of Functional Independence Measure (FIM) scores obtained by interview from patients hospitalized after stroke. This study included 82 patients with stroke. Interviews by a trained study coordinator were conducted to obtain BI and FIM subset (transfer, locomotion, feeding) scores from patients or a knowledgeable proxy. In addition, therapists’ observational FIM scores on subset items recorded in the medical record were retrieved. The reliability, responsiveness, and validity of the BI and FIM subset were examined using Cronbach’s alpha, effect size calculations, Wilcoxon tests, Spearman (rS) correlations, and regression analysis. Statistical analysis indicated that the post-stroke interview BI (.952) and FIM subset (.939) scores had excellent internal consistency. Large effect sizes and Wilcoxon test results between the prestroke and poststroke interview scores for the BI and FIM subset (z, ⫺3.739 to ⫺6.168) indicated that these 2 instruments were responsive to changes in patients’ physical function status that accompanied stroke. Excellent correlations were found between BI and FIM interview scores poststroke (rs, .913 to .971). The FIM subset interview scores and therapists’ scores showed correlations that were moderate to good (rs, .508 to .754). Length of stay correlated significantly with both the interview BI score (rs, ⫺.604) and the interview FIM subset score (rs, ⫺.583). Length of stay was correlated also with the National Institutes of Health Stroke Scale scores (rs, .484). The regression analysis depicted the BI interview scores as the best predictor of the length of stay. Results provide support for the use of functional scores obtained by interview from hospitalized stroke patients. Both the BI and the FIM subset examined in this study are legitimate for this purpose. Key Words: Activities of daily living—Measurement—Outcomes—Stroke. Copyright © 2001 by National Stroke Association
Cerebrovascular accident (stroke) is the leading cause of serious long-term disability in the United States.1 Each year, approximately 550,000 people experience stroke in
From the Department of Physical Therapy, School of Allied Health, University of Connecticut, Storrs; and Institute of Outcomes Research, Department of Neurology, Hartford Hospital, Hartford, CT. Received February 16, 2000; accepted March 9, 2001. Address reprint requests to Richard W. Bohannon, MS, EdD, PT, NCS, Department of Physical Therapy, School of Allied Health, U-2101, University of Connecticut, Storrs, CT 06269-2101. Copyright © 2001 by National Stroke Association 1052-3057/01/1002-0008$35.00/0 doi:10.1053/jscd.2001.24656
the US, 150,000 die, and 300,000 people experience disability.2 Although the mortality from stroke has decreased significantly in the past 30 years, the number of strokes has not decreased. As a result, the number of stroke survivors with chronic disability has increased.3,4 Approximately 3 million Americans are currently living with varying degree of disability because of stroke. The annual cost for stroke in the US is estimated to be $30 billion.2 This economic burden is primarily attributed to disabilities that prevent patients with stroke from returning to their previous level of independence.1,2 Numerous measures of disability exist. However, only 2 were recommended for use with patients with stroke by the Agency for Health Care Policy and Research (AH-
Journal of Stroke and Cerebrovascular Diseases, Vol. 10, No. 2 (March-April), 2001: pp 79-84
79
K.S. SADARIA ET AL.
80 2
CPR). The 2 recommended by the AHCPR were the Barthel Index (BI)5 and the Functional Independence Measure (FIM).6 We searched the Medline and Cumulative Index of Nursing and Allied Health (CINHAL) databases and found that these 2 instruments have been widely used for research between 1982 and 1999. Using the terms “stroke” and the “Barthel Index,” we found 234 studies listed in Medline and 248 studies listed in CINAHL. Simultaneously, using the terms “stroke” and “FIM,” we found 70 studies listed in Medline and 178 studies listed in CINAHL. The BI was initially developed in 1955 for use with chronic diseases.5 However, the BI was not published until 1965. Since then, the BI has been used extensively to document the physical function of patients with stroke. The BI scale consists of 10 items (feeding, chair/bed transfer, grooming, toilet use, bathing, ambulation, stair climbing, dressing, bowel control, and bladder control). In the original BI scale, item scores ranged from 0 to 5, 0 to 10, or 0 to 15, depending on the items. A score of 100 indicated full competence in all areas. Collin et al6 proposed an alternative scoring system in which BI item scores range from 0 to 1, 0 to 2, or 0 to 3, depending on item, with maximum total score of 20. The FIM7 was first described in 1984. It was designed to be used as a universal assessment tool in the Uniform Data System for medical rehabilitation. The FIM was developed in an effort to provide comprehensive assessment of both physical and cognitive disability. Thus, the FIM has 2 main domains, motor and cognitive. The motor domain has 4 subscales (self-care, sphincter control, transfers, and locomotion), whereas the cognitive domain has 2 subscales (communication and social cognition). The FIM scale comprises 18 items (13 motor and 5 cognitive), each of which is scored on a 7-point scale (1 being completely dependent and 7 being fully independent). Thus, total FIM score ranges from 18 to 126. The BI and the FIM both have been found to be valid and reliable in assessing patients’ functional status.6,8-11 Available literature indicates that BI correlates very well with the FIM.12,13 However, Kidd et al.9 found that FIM is more sensitive to changes in functional status than the BI. Shah et al.14 revised the original BI to improve its limited sensitivity. Comparison of these functional scales indicates that BI can be derived from the motor items of the FIM.15 Both the BI and the FIM have been applied using a variety of methods. Methods used with the BI include direct observation, postal questionnaire (self-report), telephone interview, and interview of patients and their care takers or nurses. These different application methods have been found to be reliable and valid.16,17 Scoring of function with the FIM has been achieved using observation, self-reported questionnaire, and telephone interview methods, which also have been found to be reliable
18,19
and valid. However, available literature reveals very little about the legitimacy of functional scores derived from interviews in an acute care setting. In acute care, the time that can be devoted to activities such as outcome assessment is limited; outcome assessments of patients with stroke are no exception. The time required to apply instruments such as the BI and FIM, if applied in their entirety, may be problematic. Obtaining information by interview is one way of reducing time requirements and using a subset of items is another. These approaches, however, cannot be undertaken lightly or recommended without testing their legitimacy. To be considered legitimate in a particular application, a measure must be reliable, responsive, and valid. The validity of a measure can be established in several ways but is often showed by its correlation with another measure of the same construct or with a different measure of known importance. One of the critical aspects of the acute care hospitalization is length of stay in the hospital. In today’s costconscious health care environment, hospital length of stay is an important rehabilitation outcome from both a clinical and an administrative point of view. Brosseau et al.20 noted that predictions of rehabilitation length of stay are helpful in managing admissions, planning stroke programs according to patient characteristics, and projecting discharges; predictions of length of stay also provide a basis for payment for rehabilitation services. Research conducted several years ago indicated that an inverse relationship exists between patients’ functional status quantified using FIM scores and length of stay on an acute rehabilitation unit.21 Research has also shown patient age to be a predictor of the length of stay.20 With the potential of managed care to drive length of stay, it is unknown whether patients’ function remains a predictor of this important variable. Therefore, the relationship between functional status and acute care length of stay needs to be investigated. In addition to the BI and FIM, other scales such as National Institutes of Health Stroke Scale (NIHSS),22 which is known to predict other outcomes,23,24 should also be investigated for their ability to predict length of stay. The purpose of this study was to examine the legitimacy of BI and FIM scores obtained by interview. Specifically, the purpose was to examine reliability, responsiveness, and validity of BI scores and a subset of FIM scores obtained by interview. Two aspects of the validity were tested, convergent construct validity and predictive validity. This last aspect was conducted in conjunction with the NIHSS.
Methods This study was a part of a larger prospective outcome study of patients with stroke admitted to the neurologic
RATINGS OF PHYSICAL FUNCTION AFTER STROKE
81
therapist and had received his/her first therapy treatments. This interview included documentation of patients’ functional status before stroke (prestroke) and after first visit (poststroke) using both the BI and the FIM subset. The project coordinator used a standardized data form to record information regarding patients’ functional status. The scores for the FIM transfer, locomotion, and feeding were also obtained from the medical record where therapists’ direct observation of patients’ functional status was recorded. For this study, the NIHSS was completed within 24 hours of admission by a resident physician who first participated in a training and certification process for use of the instrument.
unit of the Hartford Hospital, Hartford, CT. This study, which was approved by the hospital’s institutional review board, was primarily descriptive and correlational. Subjects of the study included all the stroke patients whose mental and emotional status allowed participation, who provided informed consent, who completed the BI and FIM interviews, and who were scored using the NIHSS. Of 169 patients admitted to the unit over a 6-month period, 82 subjects met these criteria. Instruments The BI and FIM were the main instruments used in this study. For the BI, Wade and Collin’s version8 was used. A total BI score was derived by adding scores from individual items; the potential range of the scores was 0 to 20. Of the 18 FIM items, a subset of 3 physical items that corresponded with BI items was selected. Specifically, the FIM subset included the bed-to-chair transfer, locomotion, and feeding items. A total of these individual subset items was derived that had a potential range from 3 to 21. The NIHSS was used to quantify the neurologic deficits of the acute stroke patients. The NIHSS consists of 15 items, which includes consciousness, vision, extraocular movement, facial palsy, limb strength, ataxia, sensation, speech, and language. It is an ordinal scale whose item scores range from 0 to 2, 0 to 3, or 0 to 4, depending the items. A total score was derived by adding individual scores; this yielded a potential total score of 0 or “normal” to 42. The NIHSS has been found to be valid and reliable.25,26
Statistical Analysis Descriptive and inferential statistics were calculated using the Systat software (Chicago, IL) program. Cronbach’s alpha was used to check the reliability (internal consistency) of the functional measures. The responsiveness of the 2 scoring instruments was described using effect size27 and Wilcoxon matched pairs signed-ranks tests. The relationships between the length of stay and BI, FIM, and NIHSS scores were determined using Spearman correlations (rs). Regression analysis was performed to determine the relative value of the BI, FIM subset, and NIHSS scores as predictors of outcome (length of stay in the hospital). The critical level of significance for this study was P ⱕ .05.
Results Procedure
The patients’ mean (⫾ standard deviation) age was 70.8 (⫾ 13.4) years. Table 1 summarizes BI and FIM (subset) scores for prestroke and poststroke physical function obtained by poststroke interview. Table 1 also presents data obtained from therapists’ notes, which indicate patients’ observed functional status. The values of
A trained project coordinator interviewed patients (or if not possible, a knowledgeable proxy) regarding each patient’s physical functioning. This interview was conducted within 48 hours of admission after each patient had been evaluated by a physical and an occupational
Table 1. Descriptive statistics of BI and FIM scores
BI (interview) Prestroke No. of patients (n) Range Minimum (total) Maximum (total) Mean Median Standard deviation
82 0-20 6 20 19.3 20 2
FIM: subset (interview)
Poststroke
Prestroke
Poststroke
82 0-20 0 20 14.9 18 6.5
82 3-21 11 21 20.4 21 1.7
82 3-21 3 21 16.3 20 6.2
FIM: subset (clinical observation) Prestroke
Poststroke 47 3-21 3 21 12.7 14 5.7
Difference: prestrokepoststroke (interview) BI 82 0-20 0 17 4.4 1 4.8
FIM 82 0-18 0 18 4.1 1 5.6
K.S. SADARIA ET AL.
82
Table 2. Results of Wilcoxon matched pairs signed-ranks tests between prestroke and poststroke scores for BI and FIM
Table 4. Spearman correlations between length of stay and BI and FIM subset items BI
Scoring item
BI Wilcoxon Z
FIM Wilcoxon Z
Transfer Mobility Feeding Stairs Grooming Bowel Bladder Toilet use Dressing Bathing Total
⫺4.543 ⫺4.940 ⫺4.235 ⫺5.543 ⫺4.472 ⫺3.739 ⫺3.906 ⫺4.179 ⫺5.161 ⫺5.568 ⫺6.163
⫺4.878 ⫺5.692 ⫺4.472
⫺6.168
FIM
Item
rs
P
rs
P
Transfer Mobility Feeding Stairs Grooming Bowel Bladder Toilet use Dressing Bathing Total
⫺0.584 ⫺0.540 ⫺0.555 ⫺0.356 ⫺0.464 ⫺0.631 ⫺0.612 ⫺0.642 ⫺0.478 ⫺0.353 ⫺0.604
⬍.001 ⬍.001 ⬍.001 ⬍.002 ⬍.001
⫺0.550 ⫺0.528 ⫺0.577
⬍.001 ⬍.001 ⬍.001
⬍.001 ⬍.001 ⬍.001 ⬍.002 ⬍.001
⬍.001
⫺0.583
⬍.001
NOTE. All Z statistics are significant at P ⬍ .001.
the reliability (internal consistency) of the instruments (measures) demonstrated by the Cronbach’s alpha were as follows: prestroke interview BI ⫽ 0.828, prestroke interview FIM (subset) ⫽ 0.660, poststroke interview BI ⫽ 0.952, poststroke interview FIM (subset) ⫽ 0.939 and poststroke observation FIM (subset) ⫽ 0.903. Calculated effect sizes and the results of the Wilcoxon tests supported the responsiveness of the BI and the FIM subset obtained by interview. The effect sizes were large, 2.2 for the BI and 2.4 for the FIM subset. The results of the Wilcoxon tests between prestroke and poststroke BI and FIM scores showed that all individual items as well as total scores differed significantly. Table 2 summaries results of the Wilcoxon tests. These values clearly indicate that both the BI and FIM are capable of detecting change in the patients’ physical function status between baseline and post stroke. The Spearman correlations between BI and the FIM scores are presented in the Table 3. The correlation between the prestroke total scores of these instruments was
Table 3. Spearman correlations between BI and FIM scores obtained by interview Activity
Correlation*
Prestroke Total BI score v FIM subset Poststroke Total BI score v FIM subset Transfer Locomotion Feeding * All correlations are significant at P ⬍ .001.
0.645 0.971 0.913 0.933 0.939
moderate (0.645). However, the correlations between the poststroke BI and FIM scores were excellent (0.913 to 0.971). The Spearman correlation between FIM scores for interview and therapist observation indicated a moderate to good correlation between these 2 methods of administration. These correlations ranged from .508 to .753 for individual items. The correlation between the total FIM subset scores obtained by interview and observation was .751. All the correlations were significant at P ⬍ .001. Correlations between the hospital length of stay and the BI as well as the FIM subset interview scores are shown in the Table 4. The correlation had a range of ⫺.353 to ⫺.642. All were significant at P ⬍ .001 or P ⬍ .002. The correlations between hospital length of stay and FIM subset scores obtained by interview (⫺.583) were higher than those obtained by the therapists and recorded in the medical record (⫺ .502). The NIHSS was also moderately correlated with the length of stay (.484). Regression analysis, with length of stay as dependent variable, identified the BI score as the best predictor. It showed the FIM subset and NIHSS scores to add no further explanation of the length of stay.
Discussion This study of patients with stroke was designed to examine legitimacy of physical function ratings obtained by interview in acute care. The BI and FIM subset were the primary instruments used to check patients’ physical function as outcome measure. Consistent with the research on the reliability of the BI and FIM,6,9,11 this study found excellent internal consistency in the both scoring systems. The excellent internal consistency of interview method of the BI and the FIM subset provides support for reliability and advocates use of these time efficient scor-
RATINGS OF PHYSICAL FUNCTION AFTER STROKE
ing system to measure patient’s physical function in the acute care setting. The results comparing prestroke and poststroke values indicate that both the BI and FIM subset are responsive measures. The effect size values as well as individual item Wilcoxon test results indicate that the FIM subset possesses slightly more responsiveness than BI among the acute care patients. This greater responsiveness of FIM is consistent with the available research.6,12 Although only a subset of the FIM (3 items) was used, it remained more responsive than the BI. We would attribute this responsiveness to its reliance on a scoring system with a broader range of scores (1 to 7) than the BI for each item. The significant correlations in this study between BI and FIM scores of the specific activities (transfer, locomotion, and eating) are consistent with published research on BI and total FIM scores.9,12,28 The significant correlations provide support for the convergent construct validity of these instruments and the use of interviews to obtain BI and the FIM subset scores for patients hospitalized after stroke. However, moderate correlations between FIM scores obtained by interview and clinical observation raise some concern regarding previously published research.29 Further research needs to be conducted to evaluate the relationship between scores obtained using the interview and observation methods in the acute care setting. This study of patients hospitalized after stroke also yielded results similar to those of previous studies of patients’ functional status and length of stay.20,21 Patients’ functional status was inversely related to length of stay in this study. That is, those with lower BI or FIM sunset scores had longer lengths of stay. Severity of neurologic involvement as measured by the NIHSS was positively related and served as a moderate predictor of length of stay. Thus, patients who were more involved neurologically had longer lengths of stay. It is widely accepted that the length of stay can be affected and predicted by using many factors.20,21,30-32 Further research is warranted to evaluate and establish a definite relationship between patients’ physical function and length of stay in the hospital. The baseline performance of patients with stroke is typically documented by interview of a patient or a significant other. Such information is important as an indicator of potential for recovery and is therefore useful for setting goals and for planning the rehabilitation process.33 Documentation of patients’ physical function after stroke is a necessity for quantifying outcome of the rehabilitation process as well as for reimbursement. Searches of the medical records conducted in conjunction with this study suggest that attempts to judge acute outcomes on the basis of recorded observations will be hampered by a
83
lack of complete data, even when a small set of items is used to characterize function. Although the importance of the observation method is widely accepted in the rehabilitation profession, the data and statistical analysis in this study reinforced the value of data from interviews and use of a subset of instrument items. The established reliability, responsiveness, and validity clearly advocate use of interview and subset scoring system (for BI and FIM) as legitimate outcome measures for rating the physical function of patients with stroke in acute care. However, further research for interview method needs to be performed using BI and FIM scale in its entirety as well as an alternative subsets of items.
References 1. National Stroke Association: Recovery and Rehab. www.stroke.org/NS804.0, 2001 2. Post-Stroke Rehabilitation. Clinical Practice Guidelines. Rockville, MD: U.S. Department of Health and Human Services. Public Health Service, Agency for Health Care Policy and Research, 1995. 3. Shahar E, McGovem PG, Sparfka JM, et al. Improved survival of stroke patients during 1980s. Stroke 1995;26: 1-5. 4. Modam B, Wagener DK. Some epidemiological aspects of stroke: Mortality/morbidity trends, age, sex, race, socioeconomic status. Stroke 1992;23:1230-1236. 5. Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Maryland State Med J 1965;14:61-65. 6. Collin C, Wade DT, Davies S, et al. The Barthel ADL index: A reliability study. Inter Disabil Stud 1988;10:6163. 7. Granger CV, Hamilton BB, Keith R. Advance in functional assessment for medical rehabilitation. Top Geriat Rehabil 1986;1:59-74. 8. Wade DT, Collin C. The Barthel ADL Index: A standard measure of disability? Inter Disabil Stud 1988;10:64-67. 9. Kidd D, Stwart G, Baldry J, et al. The functional independence measure: A comparatative validity and reliability study. Disabil Rehabil 1995;17:10-14. 10. Hamilton BB, Laughlin JA, Fiedler RC, et al. Interrater reliability of the 7-level functional independence measure (FIM). Scand J Rehabil Med 1994;26:115-119. 11. Granger CV, Cotter AC, Hamilton BB, et al. Functional assessment scales: A study of persons after stroke. Arch Phys Med Rehabil 1993;74:133-138. 12. Bohannon RW. Scoring transfer and locomotion independence of home care patients: Barthel versus Functional Independence Measure. Int J Rehabil Res 1999;22: 65-66. 13. Sharrack B, Hughes RAC, Soudain S, et al. Psychometric properties of clinical rating scales used in multiple sclerosis. Brain 1999;122:141-159. 14. Shah S, Frank V, Cooper B. Improving the sensitivity of the Barthel Index for stroke rehabilitation. J Clin Epidemiol 1989;42:703-709. 15. Nyein K, McMichael L, Turner-Stokes L. Can a Barthel score be derived from the FIM? Clin Rehabil 1999;13:5663.
84 16. Gompertz P, Pound P, Ebrahim S. A postal version of Barthel Index. Clin Rehabil 1994;8:233-239. 17. Yeo D, Faleiro R, Lincoln NB. Barthel ADL Index: A comparison of administration methods. Clin Rehabil 1995;9:34-39. 18. Smith PM, Illing SB, Feilder RC, et al. Intermodal agreement of follow-up telephone functional assessment using the Functional Independence Measure in patients with stroke. Arch Phys Med Rehabil 1996;77:431-435. 19. Hoeing H, Branch LG, McIntyre L, et al. The validity in person with spinal cord injury of self-reported functional measure derived from the Functional Independence Measure. Spine 1999;24:539-544. 20. Brosseau L, Philippe P, Potvin L, et al. Post-Stroke inpatient rehabilitation: Predicting length of stay. Am J Phys Med Rehabil 1996;75:422-430. 21. Bohannon RW, Cooper J. Admission function portends discharge function and length of stay after stroke. J Stroke Cerebrovasc Dis 1994;4:235-237. 22. Brott T, Adams HP, Olinger CP, et al. Measurement of acute cerebral infarction: A clinical examination scale. Stroke 1989;20:864-870. 23. Adams HP, Davis PH, Leira EC, et al. Baseline NIH stroke scale score strongly predicts outcome after stroke. Neurology 1999;53:126-131. 24. Roth EJ, Heinemann AW, Lovell LL, et al. Impairment and disability: Their relation during stroke rehabilitation. Arch Phys Med Rehabil 1998;79:329-335.
K.S. SADARIA ET AL. 25. Wtyyk RJ, Pessin MS, Kaplan RF, et al. Serial assessment of acute stroke patient using NIH stroke scale. Stroke 1994;25:362-365. 26. Goldstein LB, Bertels C, Davis JN. Interrater reliability of the NIH stroke scale. Arch Neurol 1989;46:660-662. 27. Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status. Medical Care 1989;27: S178-189. 28. Fricke J, Unsworth CA. Inter-rater reliability of the original and modified Barthel Index, and comparison with the Functional Independence Measure. Aust Occup Ther J 1996;43:22-29. 29. Brosseu L, Philippe P, Boulanger YL. The Functional Independence Measure: Validity of selected method applied to stroke survivors. Top Geriat Rehabil 1995;11:75-85. 30. Brock K, Robinson P, Simondson J, et al. Prediction of hospital length of stay following stoke, J Qual Clin Pract 1997;17:37-476. 31. Hakim EA, Bakheit AMO. A study of the factors which influence the length of stay of stroke patients. Clin Rehabil 1998;12:151-156. 32. Tirschwell DL, Kukull WA, Longstreth WT. Medical complications of ischemic stroke and length of stay: Experience in Seattle, Washington. J Stroke Cerebrovasc Dis 1999;8:336-343. 33. Cushman LA, Poduri KR, Palenski C. Predicting discharge functional status and rehabilitation efficiency from preadmission functional assessments. J Stroke Cerebrovasc Dis 1995;5:33-38.