In the United States, stroke is the fifth leading cause of death, with almost 130 000 people dying each year.1 An estimated 6.6 million American adults 20 years and older have experienced a stroke, and projections show that by 2030, the incidence will increase by 3.4 million.2 With the aging Baby Boomer generation, the prevalence of stroke survivors is projected to increase.2 Despite an increase in the prevalence of stroke, advances in health care technology have led to higher survival rates and greater demand for interdisciplinary rehabilitation services along the continuum of care. These rehabilitative services have become a crucial component in assisting patients toward their prestroke abilities and lifestyle, as well as informing educated decisions on optimal discharge placement for patients.
Currently, 20% of stroke survivors require institutionalized care beyond 3 months.2 Eventually, 75% of stroke survivors are discharged home whereas approximately half of those 75% continue to require assistance with various activities of daily living (ADL).3 As such, the rehabilitative community may wonder if these patients receive care that best fosters rehabilitating toward their prestroke abilities. When it comes to discharge planning, numerous factors are considered when patients are transitioned from one setting of care to the next. What is more, these factors are not consistent patient to patient in how they influence each individual's care transition circumstance. Therefore, determining an optimal discharge destination according to a patient's unique set of circumstances is a complex and multifactorial process. Furthermore, future budget constraints will make efficient care for patients following stroke even more important, as national spending on stroke management reaches an estimated $34 billion each year in the United States1 and is anticipated to reach an annual $240 billion by 2030.4
A previously published systematic review assessed the influence of demographic and socioeconomic factors on poststroke management, and prior systematic reviews have discussed the influence of patient factors on discharge destination.3 As medical and rehabilitative professions continue to follow evidence-based practice for best patient care, the demand for quantitative data to inform clinical decision-making is increasing in importance. Outcome measures are a reliable and valid way to assess patients' body structure function, activity, and participation abilities and may significantly impact clinical decision-making for appropriate discharge destination in patients following a stroke.5 Several outcome measures are specifically validated to assess the functional abilities or clinical presentation of this patient population.6,7 What remains unknown is whether patient score, as measured by outcome measures, is related to discharge destination. Therefore, the purpose of this systematic review and series of meta-analyses is to identify the association between outcome measure score at admission to acute care hospital or rehabilitation facility with discharge destination in adults following acute or subacute stroke in the United States.
Protocol and Registration
A review protocol, Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), was used as a prospective checklist to write this systematic review. To ensure the clarity and transparency of the research provided, the PRISMA statement consists of a 27-item checklist and a 4-phase flow diagram.8 This systematic review was registered in the Prospero database with the registration number 42015029357.
Articles were included if they met the following criteria: (1) sample of patients with a diagnosis of a stroke; (2) cohort studies; (3) the articles used patient's score on an outcome measures as the independent variable; (4) the articles used more than 1 discharge destination as the dependent variable; and (5) the articles were published after 2004. Articles were excluded if (1) the patient was readmitted to a previous level of care (ie, inpatient rehabilitation facility [IRF] to acute care); (2) patients included were younger than 18 years; (3) they were not written in English; and (4) studies were not conducted in the United States.
Computer-based search strategies using PubMed, SCOPUS, and CINAHL databases were developed in November 2015 by the research team and a professional biomedical research librarian to identify all potentially eligible articles for the review by title. Each database was searched using a comprehensive strategy that included search terms related to “stroke,” “clinical rehabilitation predictors,” and the “continuum of care.” All 3 databases were searched using comparable strategies (see Appendix, Supplemental Digital Content 2, http://links.lww.com/JNPT/A195, for individual database search details). There were no limits applied. A hand search was then performed by reviewing the references of included articles to gather studies not revealed through the electronic searches. A successive search was performed in July 2016 following the same process for inclusion of newly published studies since the initial search and new literature was not identified.
Studies were included in this systematic review and meta-analyses based on a strategic process outlined in Figure 1. By title, abstract, and then full-text, articles were independently screened by a combination of 2 of 5 members of the research team, based on predetermined inclusion and exclusion criteria, to reduce selection bias. Titles were reviewed by 2 assigned members (K.G., A.S.), abstracts by E.T. and A.S., and full-texts by E.T. and K.G. After independent screening, the 2 assigned members at each stage of study selection compared results and made a finalized list of studies to screen for the next stage. In cases of disagreement between the 2 assigned reviewers, a remaining team member was appointed for consultation; however, all disputes were resolved within the assigned reviewers.
Data Collection Processes
The data were collected from the articles by 3 team members (E.T., A.S., and K.G.). All data were pulled by A.S. and then verified by E.T. and K.G. on separate occasions. Disagreements were resolved by consulting with A.S. and reaching a consensus between the 3 team members. On the basis of the type of data (continuous or categorical), effect size was calculated in 1 of 2 manners. For continuous data, Cohen's d was calculated using the equation: d = M1 − M2/spooled, where spooled = √[(s12 + s22)/2].9 For categorical data, odds ratios (ORs) and confidence intervals (CIs) were calculated for each category at each discharge destination using the number of participants in each group compared with the contrasting group with a 2 × 2 table. ORs were calculated for each discharge destination by the study's established outcome measure score ranges. The OR of each discharge location was calculated against every other possible discharge location. In addition, in the case that needed information was not presented in the original article, effort was made to contact the authors.
The study population referred to the type of stroke, which included ischemic attack, hemorrhagic attack, or transient ischemic attack (TIA) cerebral vascular accidents. The time poststroke was the length of time (days, weeks, etc) since the stroke episode. The starting point on the continuum of care was defined as the initial inpatient setting before transitioning to a lower acuity of care. These starting points included acute hospitalization, rehabilitation unit, and IRF. The outcome measures were defined as standardized clinical instruments, as found within the StrokeEDGE Taskforce of the Academy of Neurologic Physical Therapy from the APTA, used to screen patients and monitor progress.5 It should be noted, however, that Bland et al10 did not strictly include a standardized clinical instrument but utilized the Functional Independence Measure (FIM) to create their own severity clusters based on other measurements of sensorimotor, cognition, and language impairments and activity levels.
The discharge destination was the setting the patient was sent to after rehabilitation ceased. This included IRF, skilled nursing facility (SNF), subacute care facility, home with services, home without services, assisted living, transfer to another acute care center, and death. For continuity in comparing data across various locations, we categorized discharge destinations into “home” and general “institutionalized care” except where IRF or SNF was specifically delineated. Standardized mean differences of change scores were referenced using home attendance as the comparator group.
Finally, it should be noted that outcome measure scores were examined only from measures administered at admission in acute care hospital and inpatient rehabilitative settings. This decision was based on the following 2 clinical tendencies: (1) the time frame between patients' hospital stay and IRF admission is relatively short (∼1 week),11–13 and (2) patients' deficits following a stroke are not drastically improved until intensive rehabilitation therapies, typically provided after hospital discharge.14
Risk of Bias in Individual Studies
The Quality Assessment Tool for Quantitative Studies15 was used to assess the methodology of the 9 included studies. This tool can be used for any qualitative study, with 8 domains assessing selection bias, study design, confounders, blinding, data collection methods, withdrawal/dropouts, intervention integrity, and analyses. Intervention integrity and analyses were not applicable to the included study designs and were not assessed. Two members of the research team (E.T., K.G.) completed the individual review and scoring with the assessment tool. Conflicts arose at this stage of quality assessment, requiring a third assessor to remedy disputes. For example, within the confounders domain, disagreement resulted when trying to determine the presence of important differences between groups (ie, those with home discharge vs institutionalized care) and for what percentages of confounding factors were controlled. Opinions of the members differed between “yes, there are differences” and “can't tell.” Resolving these disagreements involved a thorough review of each article's study design and the advisement of one of the mentoring authors on this article, whom we counted as our third assessor in this case. Each article was given a global rating of strong, moderate, or weak based on the following instructions per the assessment tool: strong = no constituent items with weak ratings; moderate = only 1 constituent item with weak rating; weak = 2 or more constituent items with weak ratings.15
Outcome measures assessed in this research study included the FIM and FIM-related tools, NIH Stroke Scale (NIHSS), and Mobility Scale for Acute Stroke (MSAS). The total FIM scale ranges from 18 (lowest) to 126 (highest), indicating levels of motor and cognitive functions, with a minimally clinically important difference (MCID) of 22 for this patient population.16 The NIHSS score ranges from 0 to 42, with higher scores indicating greater severity of stroke symptoms. The MCID or minimal detectable change (MDC) has yet to be established for this tool. The MSAS score ranges from 6 to 36, with lower scores indicating that more assistance is required for mobility, with no established MCID or MDC at this time.
The 5 articles for the meta-analyses, as selected per their data congruency with our available statistical software, utilized either an FIM-related tool or the NIHSS to measure patient presentation. On these measures, patient presentation was categorized as “poor,” “average,” or “above-average,” based on the score obtained on the FIM-related tool or the NIHSS. Categories were defined as follows: poor: ≤39, ≥14; average: 40-79, 6-13; and above-average: ≥80, ≤5 for the FIM and NIHSS, respectively. FIM categories were derived from the work of Bottemiller et al,17 and those for the NIHSS were from the study by Schlegel et al.18 In the study by Bland et al,10 who utilized the FIM to create severity clusters, Cluster A had full intact motor and sensation testing, dementia not present, language deficits not present, modified or independent ADL, minimal to moderate balance impairment whereas Cluster D had impaired motor and sensation testing, major impairment of dementia and neglect, major impairment of language, maximum assistance with ADL, unable to ambulate independently, and major balance impairment. Clusters B and C followed within this continuum, respectively. Cluster A data were allocated to the FIM above-average category, Clusters B and C data were allocated to the FIM average category, and Cluster D data were allocated to the FIM poor category. Where patient presentation could not be categorized per the aforementioned criteria, the data were presented as the odds of home discharge for every 1-point increase on the outcome measure.
Synthesis of Results
The series of meta-analyses were completed using Comprehensive Meta-Analysis, version 3.0. Discharge destinations were categorized into 2 primary domains: home and institutionalized care. Home was defined as the patient's prestroke long-term residence, and institutionalized care was defined as a change in location (toward increased acuity) from a patient's prestroke residence. Reports of home included home, assisted living residences, and transitional living. Reports of institutionalized care included subacute care facility, rehabilitation, SNF, or long-term nursing facility.
A total of 4 meta-analyses were performed: 1 presenting continuous data on FIM-related data for home discharge, and 3 presenting categorical data on FIM-related and NIHSS data for home or institutionalized care discharge. It should be noted that Bottemiller et al17 denoted their nonhome discharge locations as “facility” whereas Bland et al10 and Schlegel et al18 differentiated their nonhome discharge locations into SNF and IRF. For the sake of continuity, data sets from IRF and SNF were condensed into a “combined” destination to remain comparable with the “facility” data set; however, all reference institutionalized care. Also for continuity, patient presentations were grouped into poor, average, or above-average categories. For Bland et al,10 who presented data in 4 FIM-related severity clusters, Cluster A data were grouped with the above-average presentation, clusters B and C were grouped with the average presentations (denoted as AverageB and AverageC, respectively), and Cluster D was grouped with poor presentation.
To perform the series of meta-analyses, researchers calculated the I2 index, a value representing the degree of variation between studies included for analysis. This value provides a percentage of variation between studies that is due to heterogeneity, rather than chance, and is necessary when combining several formats of data. The I2 index is more useful than a Q test because it provides an indication of variability across studies, not just the presence or absence of heterogeneity. Also, I2 is independent of the number of studies reviewed.19 Higgins and Thompson20 proposed percentages that were used to quantify the magnitude of heterogeneity for this study: 0% = none; 25% = low; 50% = medium; and 75% = high heterogeneity. If I2 was less than 50%, a fixed-effects model was used, and if I2 was more than 50%, a random-effects model was used when running the statistics. The presence of publication bias was assessed using a funnel plot of the precision of the effect sizes for the included studies by plotting the log of the ORs against the standard error for each analysis (Figure 3).
The first meta-analysis (Figure 2A) was completed for the studies reporting the FIM as a continuous predictor of discharge to home. In this analysis, cognitive and motor FIM scores were reported independently for a single study and therefore treated as separate measures to predict recovery.21 A second group of meta-analyses was completed for categorical data, both FIM and NIHSS, using the categories poor, average, and above-average (described earlier) as predictors of discharge destination (Figures 2B-D).
The initial search yielded a total of 2954 publications. Following review of titles and abstracts, 69 full-text articles were identified, with 9 publications eligible for this review and 5 studies used for meta-analysis per statistical software analysis data acceptance (4 studies could not be combined with any other studies for the meta-analysis phase of the project). See Figure 1 for a flow diagram of study selection and reasons for exclusion.
Outcome measures used in the studies for systematic review included the FIM-total,13,16,22 FIM-motor and FIM-cognition,14,21 AcuteFIM,12 FIM-related clusters,10 NIHSS,18,22 and the MSAS.13 For clarification of outcome measurement scaling direction, all variations of the FIM operate as measurements of function, with higher scores indicating greater functional independence, whereas the NIHSS operates as a measurement of symptom severity, with higher scores indicating greater sensory, motor, and cognitive impairments.
Home nomenclature across studies were “home,”13,17,18,21,22 “home or assisted living residences,”12,14,23 “board and care or transitional living,”23 and “home with services and home with no services.”10 Institutionalized care nomenclature were “subacute care facilities,”22,23 “rehabilitation and long-term nursing facility,”18,23 “facility alone,”17 “SNF,”21,23 “acute unit, IRF, and SNF,”14 “IRF and SNF,”23 “nursing home including subacute rehabilitative services,”14 “acute care unit own/other facility, chronic hospital, alternate level of care, and other,”23 and “not-home.”14
Starting point on the continuum of care for the studies were as follows: “acute hospitalization,”11,12,17,18 “rehabilitation unit,”16,23 “inpatient rehabilitation center,”21,23 and “acute inpatient rehabilitation unit.”14 Additional study characteristics for the 9 articles included for systematic review are listed in Table 1.
Risk of Bias Within Studies
The Quality Assessment Tool for Quantitative Studies15 scores are outlined in Table 2. Of the 9 studies, 6 were determined to be of moderate and 3 were determined to be of weak methodological quality.
Results of Individual Studies for Systematic Review
Of the 5 studies that reported associated FIM or FIM-related scores with discharge destination, all found a significant positive association with above-average FIM score (≥80) and home discharge (OR >1)10,17 or having a moderate to large effect size (0.5-0.7 to 0.8-2.0+), respectively.14,21,23 In addition, all average (40-79) and poor FIM (≤39) score associations each with institutionalized care discharge were found significant (OR >1)10,17 or having a moderate to large effect size.14,21,23 Here, “institutionalized care” denotes discharge placement to an SNF, IRF, or, simply, facility per nomenclature used in each of the studies for systematic review.
For the 2 studies that associated NIHSS score with discharge destination, both found a significant positive association with above-average NIHSS score (≤5) and home discharge (OR >1)17 or having a large effect size.22 In addition, all average (6-13) and poor NIHSS score (≥14) associations each with institutionalized care discharge were found significant (OR >1)18 or having a large effect size.22
For the 2 remaining studies that described outcome measure score on a continuous scale, home discharge was found significantly favorable by 1.06 for every 1 point patients' scored on the AcuteFIM12 (OR >1) and for patients scoring greater than 26 on the MSAS (OR >1).
Synthesis of Results
Main findings of the systematic review revealed significant associations between above-average presentation and home discharge, average presentation and institutionalized care discharge, and poor presentation and institutionalized care discharge.
For the 2 studies that explored an FIM-related score as a continuous predictor of home discharge, one of these studies reported on the AcuteFIM as a total scale21 and the other reported FIM Motor and Cognitive subscales separately.12 To run this analysis, a random-effects model was utilized because I2 was high (85.8). Here, the summary measure indicated that for every 1-point increase on the FIM, a patient is nearly 1.08 times more likely to be discharged home than to institutionalized care (OR = 1.074; 95% CI, 1.045-1.103).
In 2 other meta-analyses, we explored poor and average FIM-related and NIHSS scores as categorical predictors of discharge to institutionalized care. For the meta-analysis evaluating poor scores, a fixed-effects model was utilized because I2 was low (14.8). The summary measure indicated that patients who present poorly on FIM-related or NIHSS measures (FIM ≤39; NIHSS score ≥14) are 3.4 times more likely to discharge to institutionalized care (OR = 3.385; 95% CI, 2.591-4.422). For the meta-analysis evaluating average scores, a random-effects model was utilized because I2 was high (83.9). The summary measure indicated that patients who present average on FIM-related or NIHSS measures (FIM = 40-79; NIHSS score = 6-13) are 1.9 times more likely to discharge to institutionalized care (OR = 1.879; 95% CI, 1.227-2.877).
The fourth meta-analysis explored above-average FIM-related and NIHSS scores as categorical predictors of home discharge. Within the meta-analysis exploring high scores, again a random-effects model was utilized, given a very high I2 (95.5). The summary measure here suggested that patients who perform above-average on FIM-related or NIHSS measures (FIM ≥80; NIHSS score ≤5) are 12 times more likely to discharge home (OR = 12.08; 95% CI, 3.550-41.07).
Risk of Bias Across Studies
Funnel plots where used to assess bias across studies plotting the log of the ORs against the standard error for each analysis (Figure 3). There were no biases to report for continuous data assessing home discharge (Figure 3A) or low-range scores assessing facility discharge (Figure 3B). Potential bias exists for mid-range scores assessing facility discharge (Figure 3C) and high-range scores assessing home discharge (Figure 3D).
No additional analyses were completed.
Summary of Evidence
Findings from these meta-analyses are consistent with common sense practice: the better a patient's outcome measure score (on FIM, AcuteFIM, NIHSS, etc), the greater the likelihood of home discharge. Our results, however, reveal the quantitative impact of this likelihood: for every 1-point increase on the AcuteFIM at admission of acute hospitalization12 or FIM at admission of IRF,21 patients are nearly 8% more likely to go home as opposed to institutionalized care. Likewise, patients scoring above-average on FIM-related and NIHSS measures at hospitalization and IRF admission are roughly 12 times more likely to discharge home. In contrast, patients who score poorest on these measures are approximately 3.4 times more likely to discharge to institutionalized care. For those scoring somewhere in the middle, discharge to institutionalized care is still almost 2 times that of home discharge. In addition, systematic review results revealed significance for home discharge for patients scoring higher than 26 on the MSAS.
Where score on an outcome measure is of greatest consideration for home discharge, the difficulty lies in determining most appropriate discharge destination for patients with stroke who score somewhere in between above-average and poor. The use of outcome measure to point toward a specific discharge setting for patients following a stroke is unrealistic. These tools are not designed to determine optimal discharge destination only to aid in decision making. In fact, outcome measures are traditionally designed and intended to measure a patient's body structure function, activity level, and/or participation in a way that is standardized and evidence-based for progress assessment. Those highlighted in this review, FIM and NIHSS, are measures of activity/participation and body structure function, respectively. The FIM encompasses level of assistance needed with physical and occupational abilities, not specific to any one clinical population, and is often used in acute care hospital settings. The NIHSS, on the contrary, is specific to patients following a stroke and is commonly used across all care settings. Therefore, given these outcome measures' validities in quantifying patient status at baseline or with change, it seems reasonable that they would be strongly influential in determining the care setting best suited to the patient's status. Regardless, decision making for discharge planning within this population is complex and multifactorial and always will be.
Other literature reviews have investigated determinants of discharge destination in terms of patient factors4 or have investigated patient performance in assessing readiness for “early supported discharge” (an alternative to conventional hospitalization).24 These and other literature studying the stroke population and their care transitions have concluded that further research is needed to prioritize outcome measure results for discharge planning.
Currently, clinicians across the US health care system choose optimal discharge destinations for patients with stroke, based on a presumably collaborative and educated insight, when not superseded by insurer influence. Having evidence-based data upon which to draw for this critical decision making can only further justify and objectify these recommendations. In addition, in a health care culture such as within the United States where affordability and coverage are often at the forefront of patient, provider, and third-party consideration, there is a progressive demand for validation in directing optimal patient care. This is especially true amongst a patient population that continues to grow in survival rating and its utilization of medical and rehabilitative resources.25 As evident in the results of this review, patients' outcome measure presentations have the robustness to serve this need for validation in continuum of care placement. Furthermore, we have provided evidence to support advanced discharge planning based on the predictive ability of an acute or subacute admission score. Regardless of stroke severity and despite the strength of functional performance, outcome measure scores cannot be used in isolation of the patient's biopsychosocial factors and must continue to be studied as influential to this process.
Future directions of research may include studies of epidemiologic data utilizing a multivariate analysis to account for at least some of the numerous personal factors that naturally influence the relationship between outcome measure score and discharge planning recommendations. This would include relevant personal, social, and clinical factors of patients in the acute care setting in conjunction with discharge destinations of SNF, IRF, home health, outpatient services, and home with no services. Although Nguyen et al21 had a similar design to this ideal model, more studies of equal proportion but varying outcome measures are needed to better supply a comprehensive review of the literature to determine how patients are best served per functional and medical status after a stroke event. In addition, further research efforts could assess the validity of other stroke-specific outcome measures for discharge planning and analyze the predictive strength across these tools. Finally, follow-up research needs to assess the success of patients in their respectively sent locations to verify these decisions. Ultimately, standardized outcome measures should be further used and studied among the poststroke population to improve health care policy and complement clinical judgment in the task of recommending discharge destinations for patients to receive the necessary care for achieving their optimal function.
The assessment of only the FIM, AcuteFIM, and FIM-related clustering by Bland et al10 and the NIHSS among the 5 studies included for the meta-analyses limited our scope of the overall use of outcome measures to determine discharge destination. Also, only one of the studies, that of Ngyuen et al,21 ran a logistic regression in which biopsychosocial factors were assessed for their influence on discharge destinations. The studies included in this review that were conducted in the acute care setting included participants with TIA as well as patients with a stroke. Because of the transient nature of TIA, it is possible that outcome measure score was determined prior to or after resolution of neurologic function. As such, TIA may be a confounder to the relationship between outcome measure score and discharge destination. None of the included studies completed analyses on this relationship, and the presentation of the results in the original research reports did not permit us to assess this relationship further. It should be noted that the data extracted from the 5 studies were all of retrospective type, in which we simply reported the characteristics of patients within each discharge destination, not truly analyzing the “predictive” nature of these tools. Finally, although the quality of our articles is a limitation based on their assessment of moderate and weak scores shown in Table 1, it was concluded that ratings were universally lower because study designs of “other” were allocated as having “weak” scores in that domain, increasing likelihood of weak ratings, overall.
Patient scores poststroke, as measured by the FIM (including its variations) and the NIHSS, can be useful in predicting home versus institutionalized care upon discharge. Despite the applicability of these instruments to potentially determine optimal discharge destinations among patients with stroke, patient-specific biopsychosocial factors may supersede isolated results of functional outcome measures. Nevertheless, these results provide a framework with which to start the plan of care and discharge process in acute and subacute care settings.
The researchers acknowledge the assistance of Heidi Beke-Harrigan, a Walsh University librarian, who assisted in the development of a comprehensive search strategy and advised in database selection.
1. US Department of Health & Human Services, National Center for Chronic Disease Prevention and Health Promotion. Stroke facts. http://http://www.cdc.gov
/stroke/facts.htm. Published March 2015. Accessed February 27, 2017.
2. American Heart Association American Stroke Association. Impact of stroke (stroke statistics). http://http://www.strokeassociation.org
/STROKEORG/AboutStroke/Impact-of-Stroke-Stroke-statistics_UCM_310728_Article.jsp-.WLRQbRjGxPU. Published June 2016. Accessed February 27, 2017.
3. Van der Cruyssen K, Vereeck L, Saeys W, Remmen R. Prognostic factors for discharge destination after acute stroke: a comprehensive literature review. Disabil Rehabil. 2015;37(14):1214–1227.
4. Ovbiagele B, Goldstein LB, Higashida RT, et al Forecasting the future of stroke in the United States: a policy statement from the American Heart Association and American Stroke Association. Stroke. 2013;44(8):2361–2375.
5. Pinto Zipp G, Sullivan JE, Rose DK, et al Academy of Neurologic Physical Therapy. Outcome measures recommendations. StrokEDGE Taskforce recommendations. http://http://www.neuropt.org
/docs/stroke-sig/strokeedge_taskforce_summary_document.pdf?sfvrsn=2. Published 2011. Accessed February 27, 2017.
6. Adams HP Jr, Davis PH, Leira EC, et al Baseline NIH Stroke Scale
score strongly predicts outcome after stroke: a report of the Trial of Org 10172 in Acute Stroke Treatment (TOAST). Neurology. 1999;53(1):126–131.
7. Ward I, Pivko S, Brooks G, Parkin K. Validity of the stroke rehabilitation assessment of movement scale in acute rehabilitation: a comparison with the Functional Independence Measure
and stroke Impact Scale-16. PMR. 2011;3(11):1013–1021.
8. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg. 2010;8(5):336–341.
9. Soper DS. Effect size (Cohen's calculator for a Student t
-test [software]. http://http://www.danielsoper.com
/statcalc. Published 2017. Accessed February 27, 2017.
10. Bland MD, Whitson M, Harris H, et al Descriptive data analysis examining how standardized assessments are used to guide post-acute discharge recommendations for rehabilitation services after stroke. Phys Ther. 2015;95(5):710–719.
11. Bohannon RW, Lee N, Maljanian R. Postadmission function best predicts acute hospital outcomes after stroke. Am J Phys Med Rehabil. 2002;81(10):726–730.
12. Roberts PS, Mix J, Rupp K, et al Using functional status in the acute hospital to predict discharge destination for stroke patients. Am J Phys Med Rehabil. 2016;95(6):416–424.
13. Tinl ML, Kale MK, Doshi S, Guarino AJ, Beninato M. The Mobility Scale for Acute Stroke predicts discharge destination after acute hospitalization. J Rehabil Med. 2014;46(3):219–224.
14. Lutz BJ. Determinants of discharge destination for stroke patients. Rehabil Nurs. 2004;29(5):154–163.
15. Thomas BH, Ciliska D, Dobbins M, Micucci S. A process for systematically reviewing the literature: providing the research evidence for public health nursing interventions. Worldviews Evid Based Nurs. 2004;1(3):176–184.
16. Keith RA, Granger CV, Hamilton BB, Sherwin FS. The Functional Independence Measure
: a new tool for rehabilitation. Adv Clin Rehabil. 1987;1:6–18.
17. Bottemiller KL, Bieber PL, Basford JR, Harris M. FIM scores, FIM efficiency, and discharge disposition following inpatient stroke rehabilitation. Rehabil Nurs. 2006;31(1):22–25.
18. Schlegel D, Kolb SJ, Luciano JM, et al Utility of the NIH Stroke Scale
as a predictor of hospital disposition. Stroke. 2003;34(1):134–137.
19. Huedo-Medina TB, Sánchez-Meca J, Marín-Martinez F, Botella J. Assessing heterogeneity in meta-analysis: Q statistic or I2
index? Psychol Methods. 2006;11(2):193–206.
20. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539–1558.
21. Nguyen VQ, PrvuBettger J, Guerrier T, et al Factors associated with discharge to home versus discharge to institutional care after inpatient stroke rehabilitation. Arch Phys Med Rehabil. 2015;96(7):1297–1303.
22. Elwood D, Rashbaum I, Bonder J, et al Length of stay in rehabilitation is associated with admission neurologic deficit and discharge destination. PMR. 2009;1(2):147–151.
23. Reistetter TA, Graham JE, Deutsch A, Granger CV, Markello S, Ottenbacher KJ. Utility of functional status for classifying community versus institutional discharges after inpatient rehabilitation for stroke. Arch Phys Med Rehabil. 2010;91(3):345–350.
24. Mas MÀ, Inzitari M. A critical review of early supported discharge for stroke patients: from evidence to implementation into practice. Int J Stroke. 2012;10(1):7–12.
25. Wissel J, Olver J, Sunnerhagen KS. Navigating the poststroke continuum of care. J Stroke Cerebrovasc Dis. 2013;22(1):1–8.
clinical prediction rules; functional independence measure; human movement system; NIH Stroke Scale
Supplemental Digital Content
© 2018 Academy of Neurologic Physical Therapy, APTA