Authors: Feng-Hang Chang, ScD, MPH, OTR Pengsheng Ni, MD, MPH Alan M. Jette, PhD, MPH, PT, FAPTA
Management
Affiliations: From the Health and Disability Research Institute, Boston University School of Public Health, Boston, Massachusetts.
ORIGINAL RESEARCH ARTICLE
Correspondence:
Does Activity Limitation Predict Discharge Destination for Postacute Care Patients?
All correspondence and requests for reprints should be addressed to: Feng-Hang Chang, ScD, MPH, OTR, Health and Disability Research Institute, Boston University School of Public Health, 715 Albany St, T5W, Boston, MA 02118-2526.
ABSTRACT
Disclosures: Funded in part by the Department of Education (National Institute on Disability and Rehabilitation Research) Advanced Rehabilitation Research & Training Grant H133P120001 (to A.M.J.). Dr Jette holds stock interest in CREcare, LLC, a small business he started that distributes outcome instruments. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
0894-9115/14/9309-0782 American Journal of Physical Medicine & Rehabilitation Copyright * 2014 by Lippincott Williams & Wilkins DOI: 10.1097/PHM.0000000000000097
Chang F-H, Ni P, Jette AM: Does activity limitation predict discharge destination for postacute care patients? Am J Phys Med Rehabil 2014;93:782Y790.
Objective: This study aimed to examine the ability of different domains of activity limitation to predict discharge destination (home vs. nonhome settings) 1 mo after hospital discharge for postacute rehabilitation patients.
Design: A secondary analysis was conducted using a data set of 518 adults with neurologic, lower extremity orthopedic, and complex medical conditions followed after discharge from a hospital into postacute care. Variables collected at baseline include activity limitations (basic mobility, daily activity, and applied cognitive function, measured by the Activity Measure for Post-Acute Care), demographics, diagnosis, and cognitive status. The discharge destination was recorded at 1 mo after being discharged from the hospital.
Results: Correlational analyses revealed that the 1-mo discharge destination was correlated with two domains of activity (basic mobility and daily activity) and cognitive status. However, multiple logistic regression and receiver operating characteristic curve analyses showed that basic mobility functioning performed the best in discriminating home vs. nonhome living.
Conclusions: This study supported the evidence that basic mobility functioning is a critical determinant of discharge home for postacute rehabilitation patients. The Activity Measure for Post-Acute CareYbasic mobility showed good usability in discriminating home vs. nonhome living. The findings shed light on the importance of basic mobility functioning in the discharge planning process. Key Words: Discharge Destination, Rehabilitation, Receiver Operating Characteristic Curve, International Classification of Functioning, Disability and Health
782
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014 Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
P
rediction of discharge destination for patients after an acute illness is important to healthcare providers, patients, and families. Even though patients who receive rehabilitation care are more likely to be discharged home compared with those who do not receive rehabilitation care, a proportion of patients cannot return to a community setting and need further care in a postacute care institutional setting.1,2 Knowing the expected destination at the time of discharge from acute care can assist practitioners to set an optimistic plan and a realistic goal for the patient. Accordingly, patients and their families can be prepared for what they need, such as home modification, nursing home search, outpatient services, and community supports.3 Determining discharge destination is complicated and usually involves multiple factors such as the severity of impairment, ability to function independently in daily life, family and social supports, availability of funding and resources, and home environment.4,5 Of these factors, activity limitations (as defined by the World Health Organization’s International Classification of Functioning, Disability and Health Framework) have been noted as a significant predictor of discharge destination.6 Jackson et al.5 found that cognitive and physical functional performance override diagnosis and demographic variables to predict discharge destination in inpatient rehabilitation patients. Other studies also confirmed that, with better physical, cognitive, and activities of daily living performance, individuals are more likely to live in a home environment and maintain community living.7Y9 However, limited research has examined the discriminative abilities of specific domains of activity limitation to predict living situation after discharge from hospital into postacute care. In addition, most of the existing studies focused only on an illness-specific population, such as patients after stroke or traumatic brain injury. Very few studies examined the impact of activity limitations on discharge destination in a diverse sample, which may lead to the concern about whether the findings can be generalized to other rehabilitation populations. A number of functional measures have been used for determining the discharge destination after a hospitalization. The most commonly used instruments include the Functional Independence Measure and the Barthel Index.5,9Y11 Other functional measures such as the Motor Assessment Scale,12 the Cognitive Screening Test,13 and the Berg Balance Scale,14,15 which assess a specific functional area, have also been used to predict discharge destination. www.ajpmr.com
However, there are limitations in using these measures to predict outcomes after discharge from inpatient settings. For example, measures such as the Functional Independence Measure and the Barthel Index assess only a limited number of basic activities and fail to cover a broader range of activity limitations required for community function. This drawback leads to significant ceiling effects particularly in predicting long-term outcomes in postacute rehabilitation patients.16 Other measures that focus only on one functional area (e.g., motor skills, balance, or cognitive functioning) tend to be illness specific and unable to provide a full picture of an individual’s activity limitations. Other issues such as response burden and floor effect have also been identified with these instruments.12 Beyond these challenges, most instruments for determining discharge destination in the existing literature were applied at the phase of admission to the acute care hospital, in which the patient’s impairment and functioning may still be changing enormously.17 Thus, a measure that can be used to predict discharge destination while the disease or injury is more stable can be useful in the discharge planning process. The Activity Measure for Post-Acute Care (AM-PAC) is an outcome measure developed based on the International Classification of Functioning, Disability and Health construct of activity. It was designed to measure a wide range of activity limitations for a diverse group of postacute care patients across inpatient and community settings.18,19 It covers three primary activity limitation domains: applied cognitive, basic mobility, and daily activity functioning. Considering both feasibility (low response burden) and comprehensiveness, Haley et al.18 developed the short form of the AM-PAC based on item response theory and computer-adaptive versions. The AM-PAC, as compared with the Functional Independence Measure, has demonstrated better item precision, sensitivity of change, and fewer floor effects.16 Nevertheless, the ability of different AM-PAC functional domains to predict living location after acute hospitalization has not been investigated. The primary aim of this study was to extend previous studies on evaluating the ability of different domains of activity limitation to predict living location (home vs. nonhome settings) 1 mo after discharge from an acute care hospital, what the authors termed ultimate discharge destination, for postacute rehabilitation patients. Because previous studies have shown that multiple demographic and impairment variables (e.g., age, sex, and diagnosis) may influence discharge destination, these factors Predicting Discharge Destination for Patients
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
783
were taken into account in the analyses. The study objectives included (1) investigating whether level of mobility, applied cognitive, and daily activity functioning can predict 1-mo discharge destination and (2) evaluating the discriminative abilities of these activity limitations to differentiate between home vs. institutional settings compared with demographic and other clinical factors.
METHODS Participants Overall, 518 participants were recruited at discharge from a large acute care hospital or on admission to one of two rehabilitation hospitals in a Northeastern metropolitan city. The primary diagnoses of the participants were (1) neurologic disorder, (2) lower extremity orthopedic trauma, and (3) medically complex conditions. The inclusion criteria for this study were as follows: participants must (1) be 18 yrs or older; (2) understand and speak English; (3) have a prognosis for survival of 1 yr; (4) be able to give informed consent; and (5) be receiving postacute care in an inpatient rehabilitation, skilled nursing, outpatient clinic, or home care setting. Individuals were ineligible for the study if the recruiter deemed that the patient had memory, judgment, or severe communication impairments that could interfere with his/her ability to accurately answer questions regarding his/her daily activities.
Data Collection This study was part of a prospective, longitudinal cohort study. Data were collected at four time points: discharge from acute care (baseline), 1 mo after baseline (T1), 6-mo follow-up (T2), and 12-mo follow up (T3). For the current study, only baseline and 1 mo data were used because the variations of living location at T2 and T3 were too small to be used for analysis (96.49% of the participants lived at home at T2 and 96.73% of the participants lived at home at T3).
Outcome Variable The dependent variable, living location at T1, was categorized dichotomously: home and nonhome (including transitional living, intermediate care, skilled nursing, subacute care, rehabilitation facility, and another hospital).
Independent Variables The primary independent variables were three domains of activity limitations: basic mobility, daily activity, and applied cognitive functioning. Other independent variables included demographics (age,
784
Chang et al.
sex, and race) and impairment variables (primary diagnosis and cognitive status). All of these independent variables were collected at baseline. The selection of these independent variables was based on the literature and available in the study data set. Data on activity limitations were collected using the short form of the AM-PAC, which comprises ten items in each of the three subscales (applied cognitive, basic mobility, and daily activity). Items were phrased, BHow much difficulty do you currently have (without help from another person or device) with the following activitiesI?[ (5-point rating) or BHow much help from another person do you need with the following activitiesI?[ (6-point rating). Raw summary scores in each subscale were transformed to interval-level scores along the same difficulty continuum using Rasch partial-credit methods.20 The final scores range from 0 to 100 (the higher scores reflect greater function). Test-retest reliability estimates for the AM-PAC ranged from 0.91 to 0.97.21 The participants’ medical records were abstracted to retrieve their primary diagnoses, which were categorized into three main categories: neurologic, lower extremity orthopedic, and complex medical. Neurologic impairments referred to central nervous system impairments, such as Parkinson disease, cerebrovascular diseases, multiple sclerosis, and traumatic brain injury. Lower extremity orthopedic impairments included lower extremity fracture, amputation, and replacement (e.g., hip replacement). Complex medical conditions were defined as conditions that were not immediately life threatening but placed the person at risk for debility and/or functional limitations (e.g., cardiopulmonary, amputations, diabetes, liver disease, oncology, or other chronic neuromuscular or musculoskeletal conditions). Cognitive status was measured by the Short Portable Mental Status Questionnaire, which assessed the presence and the degree of impairment in cognitive functions such as orientation, memory, and mental operations. The score ranges from 0 to 10.
Data Analysis Bivariate analyses (including Pearson correlation coefficients and W2 test) were used to examine whether activity limitations and other independent variables were significantly correlated with the outcome variable. The significant variables with P value of less than 0.05 were then included in the multivariate logistic regression analysis to predict 1-mo discharge destination to home or an institutional setting.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
After logistic regression, the receiver operating characteristic curves of the AM-PAC were drawn to demonstrate the discriminative ability of the three activity limitation domains in predicting the two discharge destinations. The receiver operating characteristic curves demonstrated the true-positive and false-positive rates by plotting sensitivity and 1specificity along the y-axis and the x-axis. The areas under the curve (AUCs) were computed for the comparison of the effectiveness in discriminating discharge home or not. AUC ranges from 0.5 to 1; greater than 0.9 is highly discriminative, 0.7Y0.9 is moderately discriminative, 0.5Y0.7 is lowly discriminative, and 0.5 is not discriminative. A P value of less than or equal to 0.05 was considered statistically significant. The cut point of each subscale ratings that best discriminated the two locations was determined by the Youden Index. The Youden Index is defined as maximum (sensitivity [c] + specificity [c] j 1), where c is the cut point. It ranges from 0 to 1, with a value of 1 indicating that the AM-PAC score distribution in the home group is completely separated from that in the nonhome group, and a value of 0 means complete overlap of two distributions.22,23 The Kernel method was used to smooth the empirical cumulative density functions of AM-PAC scores in the home and nonhome groups. Iterative numeric
method was used to find the Youden Index and the corresponding cut point. The standard error of cut point was estimated by bootstrap method (500 bootstrap samples).24
RESULTS The baseline demographic and impairment characteristics of this sample and the bivariate analyses results are listed in Table 1. The mean age of the sample was 68.11 yrs. Approximately half of the participants were women, and most of the participants were white. Approximately 44% of the participants were diagnosed with complex medical conditions; 32% had lower extremity orthopedic conditions, and 23% had neurologic impairments. Of these 518 participants enrolled in the study, 417 (81%) completed the 1-mo follow-up evaluation, and the rest dropped out or died. Those lost to follow-up showed no difference from those who completed the study on demographics except for sex. There are significantly more women in the dropout sample (64% women) than in the T1 completion sample (50% women) (W2 = 6.13, P = 0.01). The two samples also showed no significant difference on baseline AM-PACYbasic mobility and AM-PACYdaily activity scores but showed difference on the AMPACYapplied cognitive score. On average, those who
TABLE 1 Characteristics of the participants and bivariate analyses of independent variables and discharge destination Total (N = 416)
Variable
T1 Destination Home (n = 339) Nonhome Settings (n = 77)
t Test
Mean (SD) AM-PAC (range, 0Y100)
Applied cognitive Basic mobility Daily activity
Age Cognitive status (range, 0Y10)
65.42 (11.88) 46.89 (14.05) 57.19 (12.62) 67.60 (14.34) 9.66 (0.95)
65.68 (11.96) 48.96 (12.92) 58.75 (11.96) 67.50 (14.48) 9.73 (0.80)
64.31 (11.53) 37.89 (15.24) 50.46 (13.24) 68.03 (13.79) 9.33 (1.44)
Male Female Main diagnosis Neurologic Lower extremity Complex medical Race White Nonwhite Habitation before Home hospitalization Nonhome settings Baseline living Rehabilitation facility location Another hospital Other
www.ajpmr.com
207 209 102 139 175 377 39 397 7 254 129 9
(49.88) (50.12) (24.52) (33.41) (42.07) (90.63) (9.38) (98.27) (1.73) (64.80) (32.91) (2.30)
172 (50.74) 167 (49.26) 78 (23.01) 113 (33.33) 148 (43.66) 311 (91.74) 28 (8.26) 324 (98.48) 5 (1.52) 206 (64.78) 103 (32.39) 9 (2.83)
0.37 G0.001 G0.001 0.77 G0.001 W2 Test
n (%) Sex
P
35 (45.45) 42 (54.55) 24 (31.17) 26 (33.77) 27 (35.06) 66 (85.71) 11 (14.29) 73 (97.33) 2 (2.67) 48 (64.86) 26 (35.14) 0 (0)
0.40 0.25 0.10 0.49 0.33
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
785
TABLE 2 Results from the multivariate logistic regression analysis of discharge destination Variables AM-PAC Basic mobility Daily activity Cognitive status
A
OR (95% CI)
Wald, P
0.04 0.02 0.24
1.04 (1.02Y1.07) 1.02 (0.99Y1.05) 1.27 (0.97Y1.65)
G0.001 0.21 0.08
CI, confidence interval; OR, odds ratio.
dropped out had lower score (mean, 61.63) than those who completed T1 assessments (mean, 65.41) (t = 2.86, P = 0.004). Of all independent variables, only basic mobility function, daily activity function, and cognitive status were significantly correlated with the 1-mo discharge destination. The participants with higher scores on the AM-PACYbasic mobility, AMPACYdaily activity, and cognitive status were more likely to be discharged home. The logistic regression analyses were performed with the significant variables identified from the bivariate analyses. The model containing three variables was significant (W2 = 39.79, P G 0.001), and the coefficient and odds ratio of each variable are shown in Table 2. Only basic mobility function significantly predicted the 1-mo discharge destination (P G 0.05), indicating that greater mobility was associated with an individual being discharged home. The detailed statistical values of AUC, cut points, sensitivity, and specificity of the three AMPAC functional scales are displayed in Table 3. The AUCs of applied cognitive function were insignificant, whereas the AUCs for basic mobility and daily activity functioning were significant, with basic mobility showing the largest discriminative ability (AUC, 0.74) (see Fig. 1). On the basis of the Kernel smoothed method, the cut point of basic mobility (44.15) yielded a sensitivity value of 0.67 and a specificity value of 0.63 for determining the discharge destination, whereas the cut point of daily activity (47.8) yielded high sensitivity (0.87) but lower spec-
ificity (0.38). Because the AUC does not show the cut point, the authors presented the empirical and Kernel smoothed cumulative density functions of basic mobility and daily activity scores along with the sensitivity and specificity of cut point in Figures 2 and 3.
DISCUSSION The authors’ analyses of a postacute rehabilitation sample demonstrated the ability of different domains of activity limitations to predict the 1-mo discharge destination after hospitalization. Of the three activity areas, basic mobility was the main functional area that predicted discharge home after hospitalization. Daily activity function was significantly correlated with the 1-mo discharge destination in the bivariate analysis but was not significant in the multivariate regression model. Applied cognitive function was not significantly associated with the 1-mo discharge destination in either bivariate or regression analysis. Part of this result is consistent with previous research in which functional performance was found to be a strong predictor of home discharge.9,25,26 The findings of this study provided more details specifying that basic mobility overweighed daily activity in determining the 1-mo discharge destination. As noted earlier, most of the studies measured activity limitations with the Functional Independence Measure or the Barthel Index,5,9,25,26 in which physical function was assessed in a general way with a restricted selection of activities. Hence, even though these studies suggest physical performance to be a good predictor of the 1-mo discharge destination, they provide little information on which domain or activities are stronger than the others. On the other hand, the AM-PAC includes a wider range of items and split physical function into basic mobility (e.g., walking several blocks) and daily life (e.g., tying shoes). Compared with other studies that simply contend physical function as a predictor, this study further explicated the importance of basic mobility in determining a postacute patient’s living location. The
TABLE 3 The AUC (95% confidence interval), cut points (standard error), sensitivity, and specificity of AM-PAC in discriminating discharge destination
AM-PAC Daily activity Basic mobility
AUC
95% CI
Cut Point
Sensitivity
Specificity
0.68a 0.74a
0.59Y0.78 0.66Y0.82
47.8 (3.44) 44.15 (1.84)
0.87 0.67
0.38 0.63
a
Statistically significant from 0.5. CI, confidence interval.
786
Chang et al.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FIGURE 1 Receiver operating characteristic curves comparing the AM-PACYapplied cognitive, AM-PACYdaily activity, and AM-PACYbasic mobility.
result implies that those who had difficulties in performing basic physical activities were more likely to stay at institutions or other nonhome settings. The other critical finding in this study was that applied cognitive performance was not significantly correlated with the 1-mo discharge destination. Even though cognitive status was significantly correlated with the 1-mo discharge destination, it was not significant in the regression model. This finding was somewhat surprising because previous
studies have suggested cognitive functioning to be a significant determinant of the 1-mo discharge destination.5,13,26 A possible explanation of this discrepancy is that most of the participants in this study had minimal cognitive impairment (mean [SD] score on cognitive status [range, 0Y10], 9.66 [0.95]). Accordingly, physical functional limitations were more likely to be the main concern for the participants and may influence their decision on living location. Further investigation may be needed
FIGURE 2 The empirical (points) and Kernel smoothed (curved lines) cumulative density functions (CDFs) in nonhome and home groups and the sensitivity and specificity of cut point (Basic Mobility Scale). www.ajpmr.com
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
787
FIGURE 3 The empirical (points) and Kernel smoothed (curved lines) cumulative density functions (CDFs) in nonhome and home groups and the sensitivity and specificity of cut point (Daily Activity Scale).
to examine the effect of cognitive activity limitations in people with a wider range of cognitive status. To explicate the ability of activity limitations to discriminate home vs. nonhome living, the receiver operating characteristic curves of the three activity areas measured by the AM-PAC were identified. The findings suggested that AM-PACYbasic mobility and AM-PACYdaily activity can significantly differentiate the probability of discharge to home vs. nonhome setting, and AM-PACYbasic mobility showed the largest discriminative ability. However, the AMPACYapplied cognitive was not able to significantly discriminate the living location. In addition, the authors identified the ideal threshold of AM-PACYbasic mobility and AMPACYdaily activity for discrimination as 44.16 of 100 and 48.83 of 100, respectively. Both cut points are lower than the median score, 50. That is, patients who scored more than 44.16 on basic mobility and more than 48.83 on daily activity are more likely to return home. These cut points represent the specific values that maximize information from the AM-PAC for characterizing discharge setting and can be a useful reference for discharge planning. None of the other impairment variables were significantly associated with the 1-mo discharge destination. The finding of no correlation between diagnosis and discharge destination is in agreement with the report by Jackson et al.5 that diagnosis was not a strong predictor of discharge destination. Both the literature and the finding of this study imply that the impairment variables may not be the most important predictor of discharge location.
788
Chang et al.
Other demographic variables such as age, race, and sex were also not correlated with the 1-mo discharge destination. This finding agrees with a great proportion of past research even though inconsistent evidence can be found in the literature. Several studies suggested that these demographic variables do not have a strong impact on discharge destination.5,7,26 However, several studies argued that patients who are younger, who are nonwhite, or who are male are more likely to be discharged home.4,27 With a wide range of age and approximately equal proportion of men and women, this study is helpful in clarifying the inconsistency in literature.
Limitations This study has several limitations. First, the study participants were recruited from hospitals in one region in the United States. Moreover, the impairment groups in this sample included only patients with neurologic, complex medical, and lower extremity impairments, and most of the participants had limited cognitive impairment. The authors also noted that those who were lost to follow-up had somehow lower applied cognitive scores. Thus, it remains unknown whether the results can be generalized to patients with more severe cognitive impairment or other diagnoses. Future studies should be replicated in a larger and more diverse sample to see whether the same results can be found. Second, even though this study has a longitudinal design with 1-, 6-, and 12-mo follow-ups, only baseline data could be used to predict living location at 1 mo because the variations of living location at
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
6 and 12 mos were too small to be used for analysis. Further investigation on whether patients’ living location can retain over time and whether the baseline factors can predict long-term dispositions can be conducted. Finally, several variables such as depression, length of stay in hospital, length of rehabilitation, transitional care, insurance coverage, and marital status may be relevant to the outcome but were not available in this data set.7,24,27 Future longitudinal studies will be needed to address the impact of all these potential factors for the prediction of discharge destination.
CONCLUSIONS This study supported the evidence that activity limitation is a critical determinant of the 1-mo discharge destination in postacute rehabilitation patients. Particularly, basic mobility demonstrated significant impact and the greatest discriminative ability to differentiate home vs. nonhome living. Cognitive performance, however, showed a minor impact. The two subscales, basic mobility and daily activity, of the AM-PAC showed good usability in discriminating the living location. The findings provide important information for discharge planning in rehabilitation.
REFERENCES 1. Jupp BJ, Mallela SK, Kwan J, et al: Development and evaluation of the GEMS (gait, eyesight, mental state, sedation) tool as an aid to predict outcome after hospitalization. Geriatr Gerontol Int 2011;11:8Y15. doi:10.1111/j.1447-0594.2010.00626.x 2. Murray PK, Singer M, Dawson NV, et al: Outcomes of rehabilitation services for nursing home residents. Arch Phys Med Rehabil 2003;84:1129Y36. doi:10.1016/ S0003-9993(03)00149-7 3. de Guise E, LeBlanc J, Feyz M, et al: Prediction of outcome at discharge from acute care following traumatic brain injury. J Head Trauma Rehabil 2006;21:527Y36 4. Cuthbert JP, Corrigan JD, Harrison-Felix C, et al: Factors that predict acute hospitalization discharge disposition for adults with moderate to severe traumatic brain injury. Arch Phys Med Rehabil 2011; 92:721Y30.e3. doi:10.1016/j.apmr.2010.12.023 5. Jackson JP, Whisner S, Wang EW: A predictor model for discharge destination in inpatient rehabilitation patients. Am J Phys Med Rehabil 2013;92:343Y50. doi:10.1097/PHM.0b013e318278b1df 6. World Health Organization: International Classification of Functioning, Disability and Health (ICF). Geneva, Switzerland, World Health Organization, 2001 www.ajpmr.com
7. Young Y: Factors associated with permanent transition from independent living to nursing home in a continuing care retirement community. J Am Med Dir Assoc 2009;10:491Y7. doi:10.1016/j.jamda.2009.03.019 8. Cohen-Mansfield J, Wirtz PW: Predictors of entry to the nursing home: Does length of follow-up matter? Arch Gerontol Geriatr 2011;53:309Y15. doi:10.1016/ j.archger.2010.12.009 9. McKenna K, Tooth L, Strong J, et al: Predicting discharge outcomes for stroke patients in Australia. Am J Phys Med Rehabil 2002;81:47Y56 10. Reistetter TA, Graham JE, Deutsch A, et al: Utility of functional status for classifying community versus institutional discharges after inpatient rehabilitation for stroke. Arch Phys Med Rehabil 2010;91:345Y50. doi:10.1016/j.apmr.2009.11.010 11. van Baalen B, Odding E, Stam HJ: Cognitive status at discharge from the hospital determines discharge destination in traumatic brain injury patients. Brain Inj 2008;22:25Y32. doi:10.1080/02699050701810662 12. Brauer SG, Bew PG, Kuys SS, et al: Prediction of discharge destination after stroke using the Motor Assessment Scale on admission: A prospective, multisite study. Arch Phys Med Rehabil 2008;89:1061Y5. doi:10.1016/j.apmr.2007.10.042 13. van der Zwaluw CS, Valentijn SAM, NieuwenhuisMark R, et al: Cognitive functioning in the acute phase poststroke: A predictor of discharge destination? J Stroke Cerebrovasc Dis 2011;20:549Y55. doi:10.1016/ j.jstrokecerebrovasdis.2010.03.009 14. Wee JYM, Bagg SD, Palepu A: The berg balance scale as a predictor of length of stay and discharge destination in an acute stroke rehabilitation setting. Arch Phys Med Rehabil 1999;80:448Y52. doi:10.1016/ S0003-9993(99)90284-8 15. Blum L, Korner-Bitensky N: Usefulness of the Berg Balance Scale in stroke rehabilitation: A systematic review. Phys Ther 2008;88:559Y66. doi:10.2522/ptj.20070205 16. Coster WJ, Haley SM, Jette AM: Measuring patientreported outcomes after discharge from inpatient rehabilitation settings. J Rehabil Med 2006;38:237Y42. doi:10.1080/16501970600609774 17. Lysack CL, MacNeill SE, Lichtenberg PA: The functional performance of elderly urban African-American women who return home to live alone after medical rehabilitation. Am J Occup Ther 2001;55:433Y40 18. Haley SM, Coster WJ, Andres PL, et al: Activity outcome measurement for postacute care. Med Care 2004; 42(suppl 1):I49Y61. doi:10.1097/01.mlr.0000103520. 43902.6c 19. Haley SM, Andres PL, Coster WJ, et al: Short-form Activity Measure for Post-Acute Care. Arch Phys Med Rehabil 2004;85:649Y60. doi:10.1097/01.mlr. 0000103520.43902.6c 20. Andrich D: Rash Models for Measurement. Beverly Hills CA, Sage Publications, 1998 21. Andres PL, Haley SM, Ni PS: Is patient-reported function reliable for monitoring postacute outcomes? Am J
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
789
Phys Med Rehabil 2003;82:614Y21. doi:10.1097/01. PHM.0000073818.34847.F0 22. Youden WJ: Index for rating diagnostic tests. Cancer 1950;3:32Y5 23. Bewick V, Cheek L, Ball J: Statistics review 13: Receiver operating characteristic curves. Crit Care 2004;8:508Y12. doi:10.1186/cc3000 24. Dagher JH, Richard-Denis A, Lamoureux J, et al: Acute global outcome in patients with mild uncomplicated and complicated traumatic brain injury. Brain Inj 2013;27:189Y99. doi:10.3109/02699052.2012.729288 25. Campbell SE, Seymour DG, Primrose WR, et al: A multi-centre European study of factors affecting the
790
Chang et al.
discharge destination of older people admitted to hospital: Analysis of in-hospital data from the ACMEplus project. Age Ageing 2005;34:467Y75. doi:10.1093/ ageing/afi141 26. Frank M, Conzelmann M, Engelter S: Prediction of discharge destination after neurological rehabilitation in stroke patients. Eur Neurol 2010;63:227Y33. doi:10.1159/000279491 27. Pohl PS, Billinger SA, Lentz A, et al: The role of patient demographics and clinical presentation in predicting discharge placement after inpatient stroke rehabilitation: Analysis of a large, US data base. Disabil Rehabil 2013;35:990Y4. doi:10.3109/09638288.2012.717587
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Management
Affiliations: From the Health and Disability Research Institute, Boston University School of Public Health, Boston, Massachusetts.
ORIGINAL RESEARCH ARTICLE
Correspondence:
Does Activity Limitation Predict Discharge Destination for Postacute Care Patients?
All correspondence and requests for reprints should be addressed to: Feng-Hang Chang, ScD, MPH, OTR, Health and Disability Research Institute, Boston University School of Public Health, 715 Albany St, T5W, Boston, MA 02118-2526.
ABSTRACT
Disclosures: Funded in part by the Department of Education (National Institute on Disability and Rehabilitation Research) Advanced Rehabilitation Research & Training Grant H133P120001 (to A.M.J.). Dr Jette holds stock interest in CREcare, LLC, a small business he started that distributes outcome instruments. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
0894-9115/14/9309-0782 American Journal of Physical Medicine & Rehabilitation Copyright * 2014 by Lippincott Williams & Wilkins DOI: 10.1097/PHM.0000000000000097
Chang F-H, Ni P, Jette AM: Does activity limitation predict discharge destination for postacute care patients? Am J Phys Med Rehabil 2014;93:782Y790.
Objective: This study aimed to examine the ability of different domains of activity limitation to predict discharge destination (home vs. nonhome settings) 1 mo after hospital discharge for postacute rehabilitation patients.
Design: A secondary analysis was conducted using a data set of 518 adults with neurologic, lower extremity orthopedic, and complex medical conditions followed after discharge from a hospital into postacute care. Variables collected at baseline include activity limitations (basic mobility, daily activity, and applied cognitive function, measured by the Activity Measure for Post-Acute Care), demographics, diagnosis, and cognitive status. The discharge destination was recorded at 1 mo after being discharged from the hospital.
Results: Correlational analyses revealed that the 1-mo discharge destination was correlated with two domains of activity (basic mobility and daily activity) and cognitive status. However, multiple logistic regression and receiver operating characteristic curve analyses showed that basic mobility functioning performed the best in discriminating home vs. nonhome living.
Conclusions: This study supported the evidence that basic mobility functioning is a critical determinant of discharge home for postacute rehabilitation patients. The Activity Measure for Post-Acute CareYbasic mobility showed good usability in discriminating home vs. nonhome living. The findings shed light on the importance of basic mobility functioning in the discharge planning process. Key Words: Discharge Destination, Rehabilitation, Receiver Operating Characteristic Curve, International Classification of Functioning, Disability and Health
782
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014 Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
P
rediction of discharge destination for patients after an acute illness is important to healthcare providers, patients, and families. Even though patients who receive rehabilitation care are more likely to be discharged home compared with those who do not receive rehabilitation care, a proportion of patients cannot return to a community setting and need further care in a postacute care institutional setting.1,2 Knowing the expected destination at the time of discharge from acute care can assist practitioners to set an optimistic plan and a realistic goal for the patient. Accordingly, patients and their families can be prepared for what they need, such as home modification, nursing home search, outpatient services, and community supports.3 Determining discharge destination is complicated and usually involves multiple factors such as the severity of impairment, ability to function independently in daily life, family and social supports, availability of funding and resources, and home environment.4,5 Of these factors, activity limitations (as defined by the World Health Organization’s International Classification of Functioning, Disability and Health Framework) have been noted as a significant predictor of discharge destination.6 Jackson et al.5 found that cognitive and physical functional performance override diagnosis and demographic variables to predict discharge destination in inpatient rehabilitation patients. Other studies also confirmed that, with better physical, cognitive, and activities of daily living performance, individuals are more likely to live in a home environment and maintain community living.7Y9 However, limited research has examined the discriminative abilities of specific domains of activity limitation to predict living situation after discharge from hospital into postacute care. In addition, most of the existing studies focused only on an illness-specific population, such as patients after stroke or traumatic brain injury. Very few studies examined the impact of activity limitations on discharge destination in a diverse sample, which may lead to the concern about whether the findings can be generalized to other rehabilitation populations. A number of functional measures have been used for determining the discharge destination after a hospitalization. The most commonly used instruments include the Functional Independence Measure and the Barthel Index.5,9Y11 Other functional measures such as the Motor Assessment Scale,12 the Cognitive Screening Test,13 and the Berg Balance Scale,14,15 which assess a specific functional area, have also been used to predict discharge destination. www.ajpmr.com
However, there are limitations in using these measures to predict outcomes after discharge from inpatient settings. For example, measures such as the Functional Independence Measure and the Barthel Index assess only a limited number of basic activities and fail to cover a broader range of activity limitations required for community function. This drawback leads to significant ceiling effects particularly in predicting long-term outcomes in postacute rehabilitation patients.16 Other measures that focus only on one functional area (e.g., motor skills, balance, or cognitive functioning) tend to be illness specific and unable to provide a full picture of an individual’s activity limitations. Other issues such as response burden and floor effect have also been identified with these instruments.12 Beyond these challenges, most instruments for determining discharge destination in the existing literature were applied at the phase of admission to the acute care hospital, in which the patient’s impairment and functioning may still be changing enormously.17 Thus, a measure that can be used to predict discharge destination while the disease or injury is more stable can be useful in the discharge planning process. The Activity Measure for Post-Acute Care (AM-PAC) is an outcome measure developed based on the International Classification of Functioning, Disability and Health construct of activity. It was designed to measure a wide range of activity limitations for a diverse group of postacute care patients across inpatient and community settings.18,19 It covers three primary activity limitation domains: applied cognitive, basic mobility, and daily activity functioning. Considering both feasibility (low response burden) and comprehensiveness, Haley et al.18 developed the short form of the AM-PAC based on item response theory and computer-adaptive versions. The AM-PAC, as compared with the Functional Independence Measure, has demonstrated better item precision, sensitivity of change, and fewer floor effects.16 Nevertheless, the ability of different AM-PAC functional domains to predict living location after acute hospitalization has not been investigated. The primary aim of this study was to extend previous studies on evaluating the ability of different domains of activity limitation to predict living location (home vs. nonhome settings) 1 mo after discharge from an acute care hospital, what the authors termed ultimate discharge destination, for postacute rehabilitation patients. Because previous studies have shown that multiple demographic and impairment variables (e.g., age, sex, and diagnosis) may influence discharge destination, these factors Predicting Discharge Destination for Patients
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
783
were taken into account in the analyses. The study objectives included (1) investigating whether level of mobility, applied cognitive, and daily activity functioning can predict 1-mo discharge destination and (2) evaluating the discriminative abilities of these activity limitations to differentiate between home vs. institutional settings compared with demographic and other clinical factors.
METHODS Participants Overall, 518 participants were recruited at discharge from a large acute care hospital or on admission to one of two rehabilitation hospitals in a Northeastern metropolitan city. The primary diagnoses of the participants were (1) neurologic disorder, (2) lower extremity orthopedic trauma, and (3) medically complex conditions. The inclusion criteria for this study were as follows: participants must (1) be 18 yrs or older; (2) understand and speak English; (3) have a prognosis for survival of 1 yr; (4) be able to give informed consent; and (5) be receiving postacute care in an inpatient rehabilitation, skilled nursing, outpatient clinic, or home care setting. Individuals were ineligible for the study if the recruiter deemed that the patient had memory, judgment, or severe communication impairments that could interfere with his/her ability to accurately answer questions regarding his/her daily activities.
Data Collection This study was part of a prospective, longitudinal cohort study. Data were collected at four time points: discharge from acute care (baseline), 1 mo after baseline (T1), 6-mo follow-up (T2), and 12-mo follow up (T3). For the current study, only baseline and 1 mo data were used because the variations of living location at T2 and T3 were too small to be used for analysis (96.49% of the participants lived at home at T2 and 96.73% of the participants lived at home at T3).
Outcome Variable The dependent variable, living location at T1, was categorized dichotomously: home and nonhome (including transitional living, intermediate care, skilled nursing, subacute care, rehabilitation facility, and another hospital).
Independent Variables The primary independent variables were three domains of activity limitations: basic mobility, daily activity, and applied cognitive functioning. Other independent variables included demographics (age,
784
Chang et al.
sex, and race) and impairment variables (primary diagnosis and cognitive status). All of these independent variables were collected at baseline. The selection of these independent variables was based on the literature and available in the study data set. Data on activity limitations were collected using the short form of the AM-PAC, which comprises ten items in each of the three subscales (applied cognitive, basic mobility, and daily activity). Items were phrased, BHow much difficulty do you currently have (without help from another person or device) with the following activitiesI?[ (5-point rating) or BHow much help from another person do you need with the following activitiesI?[ (6-point rating). Raw summary scores in each subscale were transformed to interval-level scores along the same difficulty continuum using Rasch partial-credit methods.20 The final scores range from 0 to 100 (the higher scores reflect greater function). Test-retest reliability estimates for the AM-PAC ranged from 0.91 to 0.97.21 The participants’ medical records were abstracted to retrieve their primary diagnoses, which were categorized into three main categories: neurologic, lower extremity orthopedic, and complex medical. Neurologic impairments referred to central nervous system impairments, such as Parkinson disease, cerebrovascular diseases, multiple sclerosis, and traumatic brain injury. Lower extremity orthopedic impairments included lower extremity fracture, amputation, and replacement (e.g., hip replacement). Complex medical conditions were defined as conditions that were not immediately life threatening but placed the person at risk for debility and/or functional limitations (e.g., cardiopulmonary, amputations, diabetes, liver disease, oncology, or other chronic neuromuscular or musculoskeletal conditions). Cognitive status was measured by the Short Portable Mental Status Questionnaire, which assessed the presence and the degree of impairment in cognitive functions such as orientation, memory, and mental operations. The score ranges from 0 to 10.
Data Analysis Bivariate analyses (including Pearson correlation coefficients and W2 test) were used to examine whether activity limitations and other independent variables were significantly correlated with the outcome variable. The significant variables with P value of less than 0.05 were then included in the multivariate logistic regression analysis to predict 1-mo discharge destination to home or an institutional setting.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
After logistic regression, the receiver operating characteristic curves of the AM-PAC were drawn to demonstrate the discriminative ability of the three activity limitation domains in predicting the two discharge destinations. The receiver operating characteristic curves demonstrated the true-positive and false-positive rates by plotting sensitivity and 1specificity along the y-axis and the x-axis. The areas under the curve (AUCs) were computed for the comparison of the effectiveness in discriminating discharge home or not. AUC ranges from 0.5 to 1; greater than 0.9 is highly discriminative, 0.7Y0.9 is moderately discriminative, 0.5Y0.7 is lowly discriminative, and 0.5 is not discriminative. A P value of less than or equal to 0.05 was considered statistically significant. The cut point of each subscale ratings that best discriminated the two locations was determined by the Youden Index. The Youden Index is defined as maximum (sensitivity [c] + specificity [c] j 1), where c is the cut point. It ranges from 0 to 1, with a value of 1 indicating that the AM-PAC score distribution in the home group is completely separated from that in the nonhome group, and a value of 0 means complete overlap of two distributions.22,23 The Kernel method was used to smooth the empirical cumulative density functions of AM-PAC scores in the home and nonhome groups. Iterative numeric
method was used to find the Youden Index and the corresponding cut point. The standard error of cut point was estimated by bootstrap method (500 bootstrap samples).24
RESULTS The baseline demographic and impairment characteristics of this sample and the bivariate analyses results are listed in Table 1. The mean age of the sample was 68.11 yrs. Approximately half of the participants were women, and most of the participants were white. Approximately 44% of the participants were diagnosed with complex medical conditions; 32% had lower extremity orthopedic conditions, and 23% had neurologic impairments. Of these 518 participants enrolled in the study, 417 (81%) completed the 1-mo follow-up evaluation, and the rest dropped out or died. Those lost to follow-up showed no difference from those who completed the study on demographics except for sex. There are significantly more women in the dropout sample (64% women) than in the T1 completion sample (50% women) (W2 = 6.13, P = 0.01). The two samples also showed no significant difference on baseline AM-PACYbasic mobility and AM-PACYdaily activity scores but showed difference on the AMPACYapplied cognitive score. On average, those who
TABLE 1 Characteristics of the participants and bivariate analyses of independent variables and discharge destination Total (N = 416)
Variable
T1 Destination Home (n = 339) Nonhome Settings (n = 77)
t Test
Mean (SD) AM-PAC (range, 0Y100)
Applied cognitive Basic mobility Daily activity
Age Cognitive status (range, 0Y10)
65.42 (11.88) 46.89 (14.05) 57.19 (12.62) 67.60 (14.34) 9.66 (0.95)
65.68 (11.96) 48.96 (12.92) 58.75 (11.96) 67.50 (14.48) 9.73 (0.80)
64.31 (11.53) 37.89 (15.24) 50.46 (13.24) 68.03 (13.79) 9.33 (1.44)
Male Female Main diagnosis Neurologic Lower extremity Complex medical Race White Nonwhite Habitation before Home hospitalization Nonhome settings Baseline living Rehabilitation facility location Another hospital Other
www.ajpmr.com
207 209 102 139 175 377 39 397 7 254 129 9
(49.88) (50.12) (24.52) (33.41) (42.07) (90.63) (9.38) (98.27) (1.73) (64.80) (32.91) (2.30)
172 (50.74) 167 (49.26) 78 (23.01) 113 (33.33) 148 (43.66) 311 (91.74) 28 (8.26) 324 (98.48) 5 (1.52) 206 (64.78) 103 (32.39) 9 (2.83)
0.37 G0.001 G0.001 0.77 G0.001 W2 Test
n (%) Sex
P
35 (45.45) 42 (54.55) 24 (31.17) 26 (33.77) 27 (35.06) 66 (85.71) 11 (14.29) 73 (97.33) 2 (2.67) 48 (64.86) 26 (35.14) 0 (0)
0.40 0.25 0.10 0.49 0.33
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
785
TABLE 2 Results from the multivariate logistic regression analysis of discharge destination Variables AM-PAC Basic mobility Daily activity Cognitive status
A
OR (95% CI)
Wald, P
0.04 0.02 0.24
1.04 (1.02Y1.07) 1.02 (0.99Y1.05) 1.27 (0.97Y1.65)
G0.001 0.21 0.08
CI, confidence interval; OR, odds ratio.
dropped out had lower score (mean, 61.63) than those who completed T1 assessments (mean, 65.41) (t = 2.86, P = 0.004). Of all independent variables, only basic mobility function, daily activity function, and cognitive status were significantly correlated with the 1-mo discharge destination. The participants with higher scores on the AM-PACYbasic mobility, AMPACYdaily activity, and cognitive status were more likely to be discharged home. The logistic regression analyses were performed with the significant variables identified from the bivariate analyses. The model containing three variables was significant (W2 = 39.79, P G 0.001), and the coefficient and odds ratio of each variable are shown in Table 2. Only basic mobility function significantly predicted the 1-mo discharge destination (P G 0.05), indicating that greater mobility was associated with an individual being discharged home. The detailed statistical values of AUC, cut points, sensitivity, and specificity of the three AMPAC functional scales are displayed in Table 3. The AUCs of applied cognitive function were insignificant, whereas the AUCs for basic mobility and daily activity functioning were significant, with basic mobility showing the largest discriminative ability (AUC, 0.74) (see Fig. 1). On the basis of the Kernel smoothed method, the cut point of basic mobility (44.15) yielded a sensitivity value of 0.67 and a specificity value of 0.63 for determining the discharge destination, whereas the cut point of daily activity (47.8) yielded high sensitivity (0.87) but lower spec-
ificity (0.38). Because the AUC does not show the cut point, the authors presented the empirical and Kernel smoothed cumulative density functions of basic mobility and daily activity scores along with the sensitivity and specificity of cut point in Figures 2 and 3.
DISCUSSION The authors’ analyses of a postacute rehabilitation sample demonstrated the ability of different domains of activity limitations to predict the 1-mo discharge destination after hospitalization. Of the three activity areas, basic mobility was the main functional area that predicted discharge home after hospitalization. Daily activity function was significantly correlated with the 1-mo discharge destination in the bivariate analysis but was not significant in the multivariate regression model. Applied cognitive function was not significantly associated with the 1-mo discharge destination in either bivariate or regression analysis. Part of this result is consistent with previous research in which functional performance was found to be a strong predictor of home discharge.9,25,26 The findings of this study provided more details specifying that basic mobility overweighed daily activity in determining the 1-mo discharge destination. As noted earlier, most of the studies measured activity limitations with the Functional Independence Measure or the Barthel Index,5,9,25,26 in which physical function was assessed in a general way with a restricted selection of activities. Hence, even though these studies suggest physical performance to be a good predictor of the 1-mo discharge destination, they provide little information on which domain or activities are stronger than the others. On the other hand, the AM-PAC includes a wider range of items and split physical function into basic mobility (e.g., walking several blocks) and daily life (e.g., tying shoes). Compared with other studies that simply contend physical function as a predictor, this study further explicated the importance of basic mobility in determining a postacute patient’s living location. The
TABLE 3 The AUC (95% confidence interval), cut points (standard error), sensitivity, and specificity of AM-PAC in discriminating discharge destination
AM-PAC Daily activity Basic mobility
AUC
95% CI
Cut Point
Sensitivity
Specificity
0.68a 0.74a
0.59Y0.78 0.66Y0.82
47.8 (3.44) 44.15 (1.84)
0.87 0.67
0.38 0.63
a
Statistically significant from 0.5. CI, confidence interval.
786
Chang et al.
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
FIGURE 1 Receiver operating characteristic curves comparing the AM-PACYapplied cognitive, AM-PACYdaily activity, and AM-PACYbasic mobility.
result implies that those who had difficulties in performing basic physical activities were more likely to stay at institutions or other nonhome settings. The other critical finding in this study was that applied cognitive performance was not significantly correlated with the 1-mo discharge destination. Even though cognitive status was significantly correlated with the 1-mo discharge destination, it was not significant in the regression model. This finding was somewhat surprising because previous
studies have suggested cognitive functioning to be a significant determinant of the 1-mo discharge destination.5,13,26 A possible explanation of this discrepancy is that most of the participants in this study had minimal cognitive impairment (mean [SD] score on cognitive status [range, 0Y10], 9.66 [0.95]). Accordingly, physical functional limitations were more likely to be the main concern for the participants and may influence their decision on living location. Further investigation may be needed
FIGURE 2 The empirical (points) and Kernel smoothed (curved lines) cumulative density functions (CDFs) in nonhome and home groups and the sensitivity and specificity of cut point (Basic Mobility Scale). www.ajpmr.com
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
787
FIGURE 3 The empirical (points) and Kernel smoothed (curved lines) cumulative density functions (CDFs) in nonhome and home groups and the sensitivity and specificity of cut point (Daily Activity Scale).
to examine the effect of cognitive activity limitations in people with a wider range of cognitive status. To explicate the ability of activity limitations to discriminate home vs. nonhome living, the receiver operating characteristic curves of the three activity areas measured by the AM-PAC were identified. The findings suggested that AM-PACYbasic mobility and AM-PACYdaily activity can significantly differentiate the probability of discharge to home vs. nonhome setting, and AM-PACYbasic mobility showed the largest discriminative ability. However, the AMPACYapplied cognitive was not able to significantly discriminate the living location. In addition, the authors identified the ideal threshold of AM-PACYbasic mobility and AMPACYdaily activity for discrimination as 44.16 of 100 and 48.83 of 100, respectively. Both cut points are lower than the median score, 50. That is, patients who scored more than 44.16 on basic mobility and more than 48.83 on daily activity are more likely to return home. These cut points represent the specific values that maximize information from the AM-PAC for characterizing discharge setting and can be a useful reference for discharge planning. None of the other impairment variables were significantly associated with the 1-mo discharge destination. The finding of no correlation between diagnosis and discharge destination is in agreement with the report by Jackson et al.5 that diagnosis was not a strong predictor of discharge destination. Both the literature and the finding of this study imply that the impairment variables may not be the most important predictor of discharge location.
788
Chang et al.
Other demographic variables such as age, race, and sex were also not correlated with the 1-mo discharge destination. This finding agrees with a great proportion of past research even though inconsistent evidence can be found in the literature. Several studies suggested that these demographic variables do not have a strong impact on discharge destination.5,7,26 However, several studies argued that patients who are younger, who are nonwhite, or who are male are more likely to be discharged home.4,27 With a wide range of age and approximately equal proportion of men and women, this study is helpful in clarifying the inconsistency in literature.
Limitations This study has several limitations. First, the study participants were recruited from hospitals in one region in the United States. Moreover, the impairment groups in this sample included only patients with neurologic, complex medical, and lower extremity impairments, and most of the participants had limited cognitive impairment. The authors also noted that those who were lost to follow-up had somehow lower applied cognitive scores. Thus, it remains unknown whether the results can be generalized to patients with more severe cognitive impairment or other diagnoses. Future studies should be replicated in a larger and more diverse sample to see whether the same results can be found. Second, even though this study has a longitudinal design with 1-, 6-, and 12-mo follow-ups, only baseline data could be used to predict living location at 1 mo because the variations of living location at
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
6 and 12 mos were too small to be used for analysis. Further investigation on whether patients’ living location can retain over time and whether the baseline factors can predict long-term dispositions can be conducted. Finally, several variables such as depression, length of stay in hospital, length of rehabilitation, transitional care, insurance coverage, and marital status may be relevant to the outcome but were not available in this data set.7,24,27 Future longitudinal studies will be needed to address the impact of all these potential factors for the prediction of discharge destination.
CONCLUSIONS This study supported the evidence that activity limitation is a critical determinant of the 1-mo discharge destination in postacute rehabilitation patients. Particularly, basic mobility demonstrated significant impact and the greatest discriminative ability to differentiate home vs. nonhome living. Cognitive performance, however, showed a minor impact. The two subscales, basic mobility and daily activity, of the AM-PAC showed good usability in discriminating the living location. The findings provide important information for discharge planning in rehabilitation.
REFERENCES 1. Jupp BJ, Mallela SK, Kwan J, et al: Development and evaluation of the GEMS (gait, eyesight, mental state, sedation) tool as an aid to predict outcome after hospitalization. Geriatr Gerontol Int 2011;11:8Y15. doi:10.1111/j.1447-0594.2010.00626.x 2. Murray PK, Singer M, Dawson NV, et al: Outcomes of rehabilitation services for nursing home residents. Arch Phys Med Rehabil 2003;84:1129Y36. doi:10.1016/ S0003-9993(03)00149-7 3. de Guise E, LeBlanc J, Feyz M, et al: Prediction of outcome at discharge from acute care following traumatic brain injury. J Head Trauma Rehabil 2006;21:527Y36 4. Cuthbert JP, Corrigan JD, Harrison-Felix C, et al: Factors that predict acute hospitalization discharge disposition for adults with moderate to severe traumatic brain injury. Arch Phys Med Rehabil 2011; 92:721Y30.e3. doi:10.1016/j.apmr.2010.12.023 5. Jackson JP, Whisner S, Wang EW: A predictor model for discharge destination in inpatient rehabilitation patients. Am J Phys Med Rehabil 2013;92:343Y50. doi:10.1097/PHM.0b013e318278b1df 6. World Health Organization: International Classification of Functioning, Disability and Health (ICF). Geneva, Switzerland, World Health Organization, 2001 www.ajpmr.com
7. Young Y: Factors associated with permanent transition from independent living to nursing home in a continuing care retirement community. J Am Med Dir Assoc 2009;10:491Y7. doi:10.1016/j.jamda.2009.03.019 8. Cohen-Mansfield J, Wirtz PW: Predictors of entry to the nursing home: Does length of follow-up matter? Arch Gerontol Geriatr 2011;53:309Y15. doi:10.1016/ j.archger.2010.12.009 9. McKenna K, Tooth L, Strong J, et al: Predicting discharge outcomes for stroke patients in Australia. Am J Phys Med Rehabil 2002;81:47Y56 10. Reistetter TA, Graham JE, Deutsch A, et al: Utility of functional status for classifying community versus institutional discharges after inpatient rehabilitation for stroke. Arch Phys Med Rehabil 2010;91:345Y50. doi:10.1016/j.apmr.2009.11.010 11. van Baalen B, Odding E, Stam HJ: Cognitive status at discharge from the hospital determines discharge destination in traumatic brain injury patients. Brain Inj 2008;22:25Y32. doi:10.1080/02699050701810662 12. Brauer SG, Bew PG, Kuys SS, et al: Prediction of discharge destination after stroke using the Motor Assessment Scale on admission: A prospective, multisite study. Arch Phys Med Rehabil 2008;89:1061Y5. doi:10.1016/j.apmr.2007.10.042 13. van der Zwaluw CS, Valentijn SAM, NieuwenhuisMark R, et al: Cognitive functioning in the acute phase poststroke: A predictor of discharge destination? J Stroke Cerebrovasc Dis 2011;20:549Y55. doi:10.1016/ j.jstrokecerebrovasdis.2010.03.009 14. Wee JYM, Bagg SD, Palepu A: The berg balance scale as a predictor of length of stay and discharge destination in an acute stroke rehabilitation setting. Arch Phys Med Rehabil 1999;80:448Y52. doi:10.1016/ S0003-9993(99)90284-8 15. Blum L, Korner-Bitensky N: Usefulness of the Berg Balance Scale in stroke rehabilitation: A systematic review. Phys Ther 2008;88:559Y66. doi:10.2522/ptj.20070205 16. Coster WJ, Haley SM, Jette AM: Measuring patientreported outcomes after discharge from inpatient rehabilitation settings. J Rehabil Med 2006;38:237Y42. doi:10.1080/16501970600609774 17. Lysack CL, MacNeill SE, Lichtenberg PA: The functional performance of elderly urban African-American women who return home to live alone after medical rehabilitation. Am J Occup Ther 2001;55:433Y40 18. Haley SM, Coster WJ, Andres PL, et al: Activity outcome measurement for postacute care. Med Care 2004; 42(suppl 1):I49Y61. doi:10.1097/01.mlr.0000103520. 43902.6c 19. Haley SM, Andres PL, Coster WJ, et al: Short-form Activity Measure for Post-Acute Care. Arch Phys Med Rehabil 2004;85:649Y60. doi:10.1097/01.mlr. 0000103520.43902.6c 20. Andrich D: Rash Models for Measurement. Beverly Hills CA, Sage Publications, 1998 21. Andres PL, Haley SM, Ni PS: Is patient-reported function reliable for monitoring postacute outcomes? Am J
Predicting Discharge Destination for Patients Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
789
Phys Med Rehabil 2003;82:614Y21. doi:10.1097/01. PHM.0000073818.34847.F0 22. Youden WJ: Index for rating diagnostic tests. Cancer 1950;3:32Y5 23. Bewick V, Cheek L, Ball J: Statistics review 13: Receiver operating characteristic curves. Crit Care 2004;8:508Y12. doi:10.1186/cc3000 24. Dagher JH, Richard-Denis A, Lamoureux J, et al: Acute global outcome in patients with mild uncomplicated and complicated traumatic brain injury. Brain Inj 2013;27:189Y99. doi:10.3109/02699052.2012.729288 25. Campbell SE, Seymour DG, Primrose WR, et al: A multi-centre European study of factors affecting the
790
Chang et al.
discharge destination of older people admitted to hospital: Analysis of in-hospital data from the ACMEplus project. Age Ageing 2005;34:467Y75. doi:10.1093/ ageing/afi141 26. Frank M, Conzelmann M, Engelter S: Prediction of discharge destination after neurological rehabilitation in stroke patients. Eur Neurol 2010;63:227Y33. doi:10.1159/000279491 27. Pohl PS, Billinger SA, Lentz A, et al: The role of patient demographics and clinical presentation in predicting discharge placement after inpatient stroke rehabilitation: Analysis of a large, US data base. Disabil Rehabil 2013;35:990Y4. doi:10.3109/09638288.2012.717587
Am. J. Phys. Med. Rehabil. & Vol. 93, No. 9, September 2014
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
