Observational Study Medicine ® OPEN Predicting length of stay in patients admitted to stroke rehabilitation with severe and moderate levels of functional impairments ∗ Alejandro García-Rudolph, PhDa,b,c, , Blanca Cegarra, MSca,b,c,d, Eloy Opisso, PhDa,b,c, Josep María Tormos, PhDa,b,c, Montserrat Bernabeu, MDa,b,c, Joan Saurí, PhDa,b,c Abstract Severe stroke patients are known to be associated with larger rehabilitation length of stay (LOS) but other factors besides severity may be contributing. We aim to identify LOS predictors within a population of mostly severe patients and analyze the impact of socioeconomic situation in functionality at admission. A retrospective observational cohort study was conducted including 172 inpatients admitted to a rehabilitation center between 2007 and 2019. Associations with LOS were examined among 30 potential predictor variables using bivariate correlations. Significantly correlated (P < .002, Bonferroni adjustment) variables were entered into 9 different multiple linear regression models. No mild participants were included, 63.37% severe and 36.63% moderate. Most significant LOS determinants were: 1) total functional independence measure (FIM) (P < .001) and hemiparesis (P = .0108) (adjusted R2 = 0.24), 2) cognitive FIM (P = .002) and severity (P = .001) (adjusted R2 = 0.22), and 3) home accessibility (P = .043) and hemiparesis (P = 0.032) (adjusted R2 = 0.19). Known LOS predictors (e.g., depression, ataxia) within the full stroke severities were not found significant in our dataset. Socioeconomic situation was found moderately correlated with total FIM (r = 0.32, P < .0001). When stratifying the patients’ socioeconomic situation into mild, important, and severe social risk, their respective median total FIM at admission were 61.5, 50, and 41, with significant differences between the mild and important group (P < .001); also significant differences were found between mild and severe groups (P < .001). A few of the variables identified in the literature as significant predictors of LOS within the full stroke population were also significant for our dataset (National Institutes of Health Stroke Scale, FIM, home accessibility) explaining less than 25% of the LOS variance. Most of the 30 analyzed known predictors were not significant (e.g., depression, age, recurrent stroke, ataxia, orientation, verbal communication, etc) suggesting that factors outside functional, socioeconomic, medical, and demographics not included in this study (e.g., rehabilitation sessions intensity) have important influences on LOS for severe patients. Patients at mild social risk obtained significantly higher total FIM at admission than patients at important and severe social risk. The importance of socioeconomic situation has been scarcely studied in the literature in relation to functionality at admission; our results suggest that it requires to be considered. Abbreviations: C-FIM = cognitive FIM, ECO = economic, ENV = environmental, FAM = family, FIM = functional independence measure, LOS = length of stay, M-FIM = motor FIM, NIHSS = National Institutes of Health Stroke Scale, PACI = partial anterior circulation infarcts, POCI = posterior circulation infarcts, SEQ = socioeconomic questionnaire, SOC = social, SUP = support, T-FIM = total FIM. Keywords: inpatients, length of stay, regression analysis, rehabilitation, socioeconomic status, stroke Editor: Fadi T. Khasawneh This research was partially funded by Barriers and facilitators to social participation (PARTICIPA) and Personalised Medicine by Predictive Modeling in Stroke for better Quality of Life (PRECISE4Q) H2020 Grant Nr 777107. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The authors have no conflicts of interest to disclose. Supplemental Digital Content is available for this article. The datasets generated during and/or analyzed during the current study are not publicly available, but are available from the corresponding author on reasonable request. a Department of Research and Innovation, Institut Guttmann, Institut Universitari de Neurorehabilitació adscrit a la UAB, Badalona, b Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès), c Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Badalona, d Universitat de Barcelona, Barcelona, Spain. ∗ Correspondence: Alejandro García-Rudolph, Department of Research and Innovation, Institut Guttmann—Hospital de Neurorehabilitació, Cami Can Ruti s/n 08916— Badalona, Barcelona, Spain (e-mail: alejandropablogarcia@gmail.com). Copyright © 2020 the Author(s). Published by Wolters Kluwer Health, Inc. This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial License 4.0 (CCBY-NC), where it is permissible to download, share, remix, transform, and buildup the work provided it is properly cited. The work cannot be used commercially without permission from the journal. How to cite this article: García-Rudolph A, Cegarra B, Opisso E, Tormos JM, Bernabeu M, Saurí J. Predicting length of stay in patients admitted to stroke rehabilitation with severe and moderate levels of functional impairments. Medicine 2020;99:43(e22423). Received: 12 April 2020 / Received in final form: 2 August 2020 / Accepted: 26 August 2020 http://dx.doi.org/10.1097/MD.0000000000022423 1 García-Rudolph et al. Medicine (2020) 99:43 Medicine 1. Introduction the full stroke population will also emerge for this sample, a different composite of predictors will best account for the Stroke rehabilitation length of stay (LOS) is one of the most variance associated with LOS for patients admitted to stroke relevant quantitative indexes that measure health service rehabilitation with severe and moderate functional impairments. utilization within a hospital. LOS is the principal predictive factor of medical expenses among variables that affect the total costs during hospitalization.[1] The ability to accurately predict 2. Materials and methods which stroke patients are likely to require longer inpatient care is desirable for both budgetary planning and healthcare providers’ 2.1. Study design considerations as well as to manage emotional expectations when A quantitative, longitudinal, retrospective review of health communicating with patients and families.[2] records was conducted for patients who completed the inpatient Many factors have been shown to influence subacute rehabili- stroke rehabilitation program at an urban rehabilitation hospital, tation LOS, including stroke severity measured with the National between September 2007 and November 2019. institute of Health Stroke Scale (NIHSS),[3] ability to perform This study conforms to the STROBE guidelines.[24] activities of daily living,[4] or admission Functional Independence Measure (FIM) score.[5] The presence of ataxia may increase LOS,[6] dysphagia,[7] as well as aphasia,[8] diabetes,[9] obesity,[10] 2.2. Participants and hypertension.[11] Besides, recurrent stroke patients have been The inclusion criteria for the study were adult patients with the previously reported requiring longer LOS.[12] diagnosis of stroke, both ischemic or hemorrhagic, who were Furthermore, there is evidence that motor[13] and cognitive[14] admitted within 3 weeks of the onset of symptoms, without any rehabilitation after stroke should be started as early as possible. previous comorbidities leading to disability. Exclusion criteria Nevertheless, time since stroke onset to rehabilitation admission were diagnosis of stroke in the context of another concomitant has been scarcely included as covariate in LOS predictive models. comorbidity (e.g., traumatic brain injury) and a previous history Falls are common post-stroke (12%–47%) and may extend of another disabling condition. Cases of transient ischemic attack inpatient stroke rehabilitation LOS[15] as well as depression.[16] or subarachnoid hemorrhage were not included. Persons were In terms of social factors, there are conflicting reports about only included in the current study if they were 18 to 85 years of whether living alone predicts LOS, for example, Tan et al (longer age at the time of stroke, and the database included complete data LOS),[17] Saxena et al (shorter LOS).[18] Besides, inadequate within a week of admission and discharge. family support[19] and environmental factors (e.g., home modifications) may delay LOS.[20] 2.3. Measurements A 2015 Lancet review[21] reports that socioeconomic status (SES) is reflected in short-term and long-term outcomes after Patients were classified into RPGs as described in the related stroke. Studies have demonstrated an association between lower research.[23] RPG severity is based on age and functional ability SES and having more severe deficits after stroke assessed by (evaluated using the FIM at admission). NIHSS at admission.[22] To our best knowledge there is a lack of Medical complications and comorbidities (reported using similar studies addressing associations between functional ICD9 codes) were collected from the participants’ electronical independence, for example, total FIM(T-FIM), motor FIM (M- health records (EHRs). After a frequency analysis, less frequent FIM), and cognitive FIM (C-FIM) at admission and SES. were discarded (presented in SupMaterial) and the following Although several researchers have previously examined the were included as predictor variables: diabetes, ataxia, aphasia, prediction of LOS within the full spectrum of stroke rehabilita- dysphagia, depression, hypertension, dyslipidemia, and atrial tion patients (mild, moderate, and severe), different variables may fibrillation (all of them recorded as yes/no). Missing values have different impact in LOS when excluding the population with were completed by means of the specific internal or external mild functional impairments. For example, while age has reports. previously been identified as a significant contributor of LOS, Demographics (age, sex, education, marital status), stroke this variable may not have the same impact for severe and milder characteristics (type and location), functionality (T-FIM, M-FIM, patients as the latter group tends to be younger.[5] To classify and C-FIM) scores at admission and discharge, time since stroke stroke severity at admission as mild, moderate, or severe, in this onset to rehabilitation admission after discharge from an acute work, we apply the RPG benchmark (Rehabilitation Patient stroke care facility (in days), were also obtained from EHRs. Groups), as in similar previous research.[23] The socioeconomic situation is assessed since 2007 in Institut The objectives of the present study are to analyze the Guttmann hospital by means of the Socioeconomic Question- associations between functional independence (T-FIM, M-FIM, naire (SEQ).[25] It involves 5 items described in Table 1: home and C-FIM) at admission and SES within a population of family (SEQ-FAM), economy (SEQ-ECO), home physical (SEQ- ischemic and hemorrhagic (moderate-RPG and severe-RPG) ENV), family support (SEQ-SUP), and need of social support stroke patients admitted to an inpatient rehabilitation hospital (SEQ-SOC). and predict their LOS from a wide range of potential predictors, A fall is defined according to the Falls Management protocol including the aforementioned demographics, clinical, and social following the Joint Commission International standards and state-of-the-art variables. managed according to the hospital protocols. It is hypothesized that M-FIM, C-FIM, and T-FIM at FIM gain was computed by calculating the difference between admission will have a stronger association (negative correlation) FIM scores at admission and at discharge. The FIM efficiency is with SES than NIHSS. FIM gain divided by LOS in days. It is also hypothesized that, while some of the same variables The LOS for each patient was determined by calculating the that have been identified as significant predictors of LOS within number of days between the date of admission and discharge. 2 García-Rudolph et al. Medicine (2020) 99:43 www.md-journal.com Table 1 Description of items of the Socioeconomic Questionnaire (SEQ). Item Description Score HOME FAMILY (LIVING WITH, CORE Ranges from: Lives with family/core of coexistence or stable partner (1 point) to lives alone, no relatives Range 1–5 OF COEXISTENCE) (FAM-SEQ) close (5 points) ECONOMY (ECO-SEQ) Ranges from: Own and/or family/core of coexistence with sufficient and stable incomes (1 point) to own Range 1–5 and/or family/core of coexistence with no fix incomes received HOME PHYSICAL (ENV-SEQ) Ranges from: Appropriate to your needs (1 point) to cannot return home (4 points) and no home (5 points) Range 1–5 FAMILY SUPPORT (SUP-SEQ) Ranges from: Autonomous or no support needed from family/core of coexistence (1 point) to Rejected or Range 1–5 abandoned by family or by core of coexistence (4 points) and no family/no core of coexistence (5 points) NEED SOCIAL SUPPORT Ranges from autonomous or with enough informal support (1 point) to Needs public institutional alternative Range 1–5 (INFORMAL/FORMAL) (SOC-SEQ) (e.g., long-term sociosanitary center or assisted (4 points) and cannot access to public support (e.g., foreigner without residence card) (5 points) TOTAL (TOT-SEQ) The overall score, determines 4 categories: no social risks (5 points), mild social risks (6–9 points), Range 5–25 important social risks (10–14), and severe social risks (≥15 points) 2.4. Statistical analyses under study (September, 2007 to November, 2019). After All statistical analyses were performed in R-v3.5.1 (64 bits), level excluding 243 patients with more than 3 weeks since stroke of significance was set at P = .05. Patients were stratified into 2 onset to admission, 247 with more than 1 week since admission RPG groups (moderate-RPG and severe-RPG). Descriptive to FIM assessment and 244 with more than 1 week since statistics were used for demographic characteristics of partic- admission to SEQ assessment a total of 1421 were assessed for ipants as well as functional and socioeconomic assessments. The discharge conditions. After excluding 326 with more than 1 week 2 RPG groups were compared using the x2 test for categorical since FIM assessment to discharge, 1095 patients were assessed variables and the Kruskal–Wallis test for continuous and ordered for data quality constraints. After removing incomplete records variables. (289 missing M-FIM at admission, 258 C-FIM at admission, 91 Associations with LOS were examined among 30 potential T-FIM at discharge, 209 SEQ at admission, 42 with missing predictor variables using bivariate (Pearson or Spearman as demographic data) and 4 outliers with LOS ≥ to the 98th appropriate) correlations. Using Cohen criteria, associations percentile, of the 202 remaining records, 29 with no NIHSS in were considered weak below 0.10, moderate between 0.10 and acute phase were removed, leaving the final 172 records (details 0.49, and strong between 0.50 and 1.00.[26] To maintain the in Fig. 1). experiment-wise error rate at P < .05, a Bonferroni adjustment Table 2 presents the characteristics of the study sample. Of the was used (0.05/30 = 0.0017). Significantly correlated variables total 172 included patients, 63.37% are severe-RPG and 36.63% were included in a multiple regression analysis using the enter are moderate-RPG. method to predict LOS. The percentage of male patients was 66.7%. The mean Multicollinearity of independent variables is tested by the (standard deviation) age was 51.72 (11.1) years, no significant variance inflation factor (VIF) and the tolerance (1/VIF). differences were found between both groups in relation to age. Tolerance is associated with each independent variable and The mean (standard deviation) time since stroke onset to ranges from 0 to 1. A tolerance below 0.40 and/or a VIF of 5 or admission across severities was 17.89 (4.36) days (no significant 10 and above indicates a multicollinearity problem.[27] differences between groups), similar to related research (e.g., The assumption of independent errors is evaluated using McClure et al[5] with 16.04 (15.21)). The percentage of ischemic the Durbin–Watson. The closer to 2 that the value is, the patients is 70.3%. better. As a conservative rule it is suggested that for values less than 1 or greater than 3 the assumption of independence is 3.1. FIM and SES assessments at admission not met.[27] In relation to functionality, at admission severe-RPG patients had significantly lower C-FIM (P = .007), M-FIM (P < .001), and 2.5. Ethical considerations T-FIM (P < .001) than moderate-RPG patients; similarly, at The authors assert that all procedures contributing to this work discharge significantly lower M-FIM (P < .001) and T-FIM comply with the ethical standards of the relevant national and (P < .001). institutional committees on human experimentation and with the But severe-RPG patients had significantly higher scores Helsinki Declaration of 1975, as revised in 2008. The study than moderate-RPG when considering FIM Gain (P < .001), was approved by the hospital’s Ethics Committee of Clinical ENV-SEQ (P < .001), and TOT-SEQ (P = .007), as shown in Research. The participants were anonymized and nonidentifi- Table 3. able. Participants did not provide written informed consent to be When stratifying the TOT-SEQ score into mild social risk, included specifically in this study, they provide informed consent important social risk, and severe social risk using the criteria for to be included in research studies addressed by the hospital. stratification presented in Table 1, the median TOT-SEQ value of the moderate RPG group is 9.00 (7.00, 10.00) therefore the moderate-RPG group belongs to the mild social risk level. 3. Results Meanwhile, the median TOT-SEQ value of the severe-RPG group A total of 2135 ischemic and hemorrhagic stroke patients were is 10.00 (8.00, 12.00), therefore the severe-RPG group belongs to admitted to the hospital’s rehabilitation unit during the period the important social risk level (P = .007). 3 García-Rudolph et al. Medicine (2020) 99:43 Medicine Records idenfied through Idenficaon database searching with primary diagnosis of stroke (n =2135) -More than 3 weeks since stroke onset to admission (n=243) -More than one week since admission to FIM assessment (n=247) Records aer me constraints -More than one week since admission applied at admission to SEQ assessment (n=224) (n =1421) Screening -More than one week since discharge to FIM assessment (n=326) Records aer me constraints applied at discharge (n =1095) -RPG calculaon: Missings in Motor FIM at admission (n=289) -RPG calculaon: Missings in Cognive FIM at admission(n=258) Eligibility -FIM Gain calculaon: Missing TOT FIM items at Records aer data quality discharge (n=91) constraints applied -Missing SEQ items at admission (n=209) (n =202) -Missing demographics data at admission (n=42) -LOS ≥ to the 98th percenle (n=4) -No NIHSS at acute phase, (n =29) Included Records included for study analysis (n =172) Figure 1. Patient selection flowchart. 3.2. Bivariate correlation analysis: NIHSS, SES, and FIM at 3.3. The impact of socioeconomic situation in FIM at admission and discharge admission and NIHSS Several strong correlations were found between stroke severity In this section we take as a starting point the stratification of the and functionality: NIHSS and T-FIM-Adm are strongly correlat- TOT-SEQ score into mild social risk, important social risk, and ed (r = 0.51, P < .0001) as well as T-FIM-Adm and T-FIM-Dis severe social risk using the criteria for stratification presented in (r = 0.51, P < .0001). Table 1. In relation to socioeconomic situation, TOT-SEQ correlated T-FIM and NIHSS scores at admission are presented in moderately with C-FIM-Adm (r = 0.22, P = .01), with M-FIM- Figure 2 and Table 5. The median T-FIM at admission Adm (r = 0.27, P < .001) and slightly stronger with T-FIM-Adm for the mild, important, and severe groups were 61.5, (r = 0.32, P < .0001), being weaker the correlation with NIHSS 50, and 41, with significant differences between the (r = 0.19, P < .05) (Table 4). mild and important group (P < .001); also significant Other moderate associations involving socioeconomic items differences were found between mild and severe group are presented in sup material, the most relevant are: ENV-SEQ (P < .001). correlated positively with FIM gain (r = 0.24, P = .01) and The median NIHSS for the mild, important, and severe groups negatively with T-FIM-Adm (r = 0.33, P < .001); meanwhile, were 13, 14, and 16. significant differences were found between SOC-SEQ also correlated negatively with T-FIM-Adm (r = mild and severe groups (P < .05). 0.25, P < .001). Notes: Box plot (minimum-minimum-lower quartile-median- Age is associated with worse FIM gain (r = 0.23, P = .01) and upper quartile-maximum); the numbers in the box indicate the ∗∗ with worse T-FIM-Dis (r = 0.30, P < .001). median; P < .01. 4 García-Rudolph et al. Medicine (2020) 99:43 www.md-journal.com Table 2 Characteristics at admission for the total patients included in the study. Moderate-RPG (N = 63) Severe-RPG (N = 109) Total (N = 172) P Male 42 (66.7%) 71 (65.1%) 113 (65.7%) .839 Age in years, mean (SD) 49.58 (10.8) 52.95 (11.1) 51.72 (11.1) .065 Ischemic stroke 47 (74.6%) 74 (67.9%) 121 (70.3%) .353 TACI 37 (58.7%) 66 (60.6%) 103 (59.9%) .111 PACI 8 (12.7%) 3 (2.8%) 11 (6.4%) POCI 2 (3.2%) 4 (3.7%) 6 (3.5%) Hemorrhagic primary 9 (14.3%) 26 (23.9%) 35 (20.3%) Hemorrhagic secondary 7 (11.1%) 9 (8.3%) 16 (9.3%) Hemiparesis LEFT 23 (36.5%) 55 (50.5%) 78 (45.3%) .077 TSO mean (SD) 17.52 (4.57) 18.10 (4.24) 17.89 (4.36) .461 NIHSS acute phase 11.68 (5.20) 14.68 (4.7) 13.58 (5.1) < .001 Hypertension 41 (65.1%) 83 (76.1%) 124 (72.1%) .119 Dislipidemia 23 (36.5%) 38 (34.9%) 61 (35.5%) .828 Dysphagia 13 (20.6%) 46 (42.2%) 59 (34.3%) .004 Diabetes 14 (22.2%) 26 (23.9%) 40 (23.3%) .807 Atrial fibrillation 4 (6.3%) 15 (13.8%) 19 (11.0%) .135 Ataxia 2 (3.2%) 5 (4.6%) 7 (4.1%) .651 Aphasia 26 (41.3%) 50 (45.9%) 76 (44.2%) .558 Depression 30 (47.6%) 75 (68.8%) 105 (61.0%) .006 Recurrent stroke 9 (14.3%) 7 (6.4%) 16 (9.3%) .087 BMI mean (SD) 26.05 (5.33) 26.89 (5.26) 26.58 (5.29) .24 Primary level of education 28 (44.4%) 53 (48.6%) 81 (47.1%) .865 Secondary level of educ 21 (33.3%) 33 (30.3%) 54 (31.4%) Higher education 14 (22.2%) 23 (21.1%) 37 (21.5%) Marital status Married 40 (63.5%) 69 (63.3%) 109 (63.4%) .299 Single 16 (25.4%) 27 (24.8%) 43 (25.0%) Divorced 4 (6.3%) 5 (4.6%) 9 (5.2%) Widow 0 (0.0%) 6 (5.5%) 6 (3.5%) Separated 3 (4.8%) 2 (1.8%) 5 (2.9%) Smoking habits Current 15 (23.8%) 19 (17.4%) 34 (19.8%) .599 Former 6 (9.5%) 11 (10.1%) 17 (9.9%) Non 42 (66.7%) 79 (72.5%) 121 (70.3%) Inpatient falls 15 (23.8%) 40 (36.7%) 55 (32.0%) .081 LOS mean (SD) 49.82 (19.12) 72.39 (27.86) 64.12 (27.23) < .001 OPTIMAL LOS 5 (7.9%) 19 (17.4%) 24 (14.0%) .083 BMI = body mass index, LOS = length of stay, NIHSS = National Institutes of Health Stroke Scale, PACI = partial anterior circulation infarcts, POCI = posterior circulation infarcts, SD = standard deviation, TACI = total anterior circulation infarcts, TSO = time since stroke onset to admission. Significant P values (P < .05) should be in bold. Table 3 Functional independence measures (at admission and discharge) and Socioeconomic Questionnaire at admission. Moderate-RPG (N = 63) Severe-RPG (N = 109) Total (N = 172) P C-FIM Adm 23.36 (7.44) 19.05 (9.74) 20.63 (9.18) .007 M-FIM Adm 50.57 (10.87) 23.57 (8.25) 33.46 (16.00) < .001 T-FIM Adm 73.93 (10.76) 42.63 (15.73) 54.09 (20.66) < .001 C-FIM Dis 27.41 (6.84) 24.94 (8.34) 25.84 (7.89) .086 M-FIM Dis 73.03 (9.44) 57.79 (19.08) 63.37 (17.79) < .001 T-FIM Dis 100.44 (11.40) 82.74 (23.80) 89.22 (21.86) < .001 FIM Gain 26.50 (12.39) 40.11 (22.34) 35.12 (20.34) < .001 FIM efficiency mean (SD) 0.60 (0.41) 0.62 (0.39) 0.62 (0.40) .358 FIM efficiency Med (P1 P3) 0.53 (0.34, 0.64) 0.63 (0.34, 0.87) 0.56 (0.34, 0.84) FAM-SEQ 1.31 (0.94) 1.45 (1.00) 1.40 (0.98) .249 ECO-SEQ 1.49 (0.84) 1.68 (1.05) 1.61 (0.98) .231 ENV-SEQ 1.66 (0.80) 2.23 (0.92) 2.02 (0.91) < .001 SUP-SEQ 2.46 (0.61) 2.58 (0.62) 2.54 (0.62) .167 SOC-SEQ 2.36 (0.97) 2.57 (0.98) 2.49 (0.98) .182 TOT-SEQ mean (SD) 9.30 (2.98) 10.52 (3.26) 10.07 (3.20) .007 TOT-SEQ Med (P1 P3) 9.00 (7.00, 10.00) 10.00 (8.00, 12.00) 9.50 (8.00, 11.00) C-FIM = cognitive FIM, ECO = economic, ENV = environmental, FAM = family, FIM = functional independence measure, Med = median, M-FIM = motor FIM, SD = standard deviation, SEQ = socioeconomic questionnaire, SOC = social, SUP = support, TOT = total, T-FIM = M-FIM + C-FIM, T-FIM = total FIM. Significant P values (P < .05) should be in bold. 5 García-Rudolph et al. Medicine (2020) 99:43 Medicine Table 4 Correlations of FIM at admission and discharge with SEQ and NIHSS. NIHSS M-FIM Adm C-FIM Adm T-FIM Adm TOT-SEQ LOS T-FIM Dis FIM Gain NIHSS 1 ∗∗∗ M-FIM Adm 0.43 1 ∗∗∗ ∗∗∗ C-FIM Adm 0.47 0.39 1 ∗∗∗ ∗∗∗ ∗∗∗ T-FIM Adm 0.51 0.93 0.64 1 ∗ ∗∗∗ ∗∗ ∗∗∗ TOT-SEQ 0.19 0.27 0.22 0.32 1 ∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗∗ LOS 0.25 0.50 0.25 0.51 0.26 1 ∗∗∗ ∗∗∗ ∗ ∗∗∗ ∗∗ ∗∗ T-FIM Dis 0.42 0.44 0.17 0.51 0.20 0.25 1 ∗∗∗ ∗∗∗ ∗∗ ∗∗∗ FIM Gain ns 0.42 0.14+ 0.42 0.13+ 0.20 0.49 1 FIM gain = total FIM at discharge—total FIM at admission, C-FIM Adm = cognitive FIM at admission, LOS = length of stay, M-FIM Adm = motor FIM at admission, NIHSS = National Institutes of Health Stroke Scale, ns = non-significant, SEQ = socio economic questionnaire, T-FIM Adm = total FIM at admission, T-FIM Dis = total FIM at discharge. ∗ P  0.05. ∗∗ P  0.01. ∗∗∗ P  0.001. Figure 2. Comparison of functional independence measure at admission and NIHSS among the 3 social risk groups. NIHSS = National Institutes of Health Stroke Scale. Table 5 SEQ, NIHSS, and FIM for each social risk level. Mild (N = 86) Important (N = 71) Severe (N = 15) Total (N = 172) P TOT-SEQ < .001 Mean (SD) 7.73 (1.06) 11.31 (1.29) 17.67 (2.69) 10.08 (3.20) Median 8.00 (7.00, 9.00) 11.00 (10.00, 12.00) 17.00 (16.00, 18.00) 9.50 (8.00, 11.00) Range 6.000–9.000 10.00–14.00 15.00–24.00 6.00–24.00 T-FIM Adm .002 Mean (SD) 59.06 (20.29) 50.87 (19.90) 40.93 (18.63) 54.09 (20.67) Median 61.50 (47.00, 74.00) 50.00 (35.00, 68.00) 41.00 (23.50, 58.00) 56.00 (36.00, 71.00) Range 18.000–97.000 18.000–85.000 18.00–70.00 18.00–97.00 NIHSS .049 Mean (SD) 12.94 (5.09) 13.69 (4.93) 16.800 (5.240) 13.587 (5.124) Median 13.00 (9.00, 16.75) 14.00 (10.00, 17.50) 16.00 (12.50, 21.50) 13.00 (9.75, 17.00) Range 2.000–25.000 3.000–23.000 10.000–25.000 2.000–25.000 FIM = functional independence measure, NIHSS = National Institutes of Health Stroke Scale, SEQ = socio economic questionnaire, T-FIM Adm = total FIM at admission. Significant P values (P < .05) should be in bold. 6 García-Rudolph et al. Medicine (2020) 99:43 www.md-journal.com Multiple lineal regression models were performed on signifi- Table 6 cant variables from the univariate analyses. Given the sample size Correlations with LOS. limitation, a maximum of 5 variables were used per model (as in r P R2 similar previous research[28]). Nine significant models emerged as Severity (moderate/severe) 0.4003865 < .000001 0.1602 presented in Table 7, highest R2 values were obtained in model Gender male 0.0484579 .5279 #2 adjusted R2 = 0.2442 (R2 = 26.19) with side of paresis and Age 0.0662903 .3876 T-FIM-Adm contributing significantly. Marital status (married/not) 0.037076 .6292 FIM scores significantly contributed to model #3, #4, and #5; Type ischemic/hemorr 0.01053576 .8909 meanwhile, depression did not contribute to any of them. Side of paresis (left/right) 0.1855227 .01483 Model #6 is composed by 3 variables, the 3 of them contribute Time since onset to Adm 0.0221928 .7726 significantly: NIHSS, severity, and hemiparesis, adj R2 = 0.1885 NIHSS 0.2471814 .001079 0.0576 and R2 = 0.2027. M-FIM Adm 0.5006698 < .000001 0.2506 C-FIM Adm 0.2472141 .001078 0.0611 Models #7, #8, and #9 show significant contributions of the T-FIM Adm 0.5063567 < .000001 0.2563 socioeconomic variables, highest R2 values are obtained in model Depression 0.221513 .003497 #8 which includes ENV-SEQ instead of TOT-SEQ, adj R2 = Current smoker 0.04964352 .5178 0.1952, R2 = 0.2141. BMI 0.08178256 .2862 Hypertension 0.09270506 .2264 3.6. Bivariate correlation analysis in a subsample: Dysphagia 0.1333635 .08114 Dyslipidemia 0.07571433 .3236 cognition variables with LOS Ataxia 0.0649127 .3975 The initial n = 172 patients were analyzed in relation to cognition Diabetes 0.007660164 .9206 assessments at admission. Therefore, we identified a subset of Atrial fibrillation 0.08089877 .2914 almost 70% of the initial sample (n = 118 with 44 moderate-RPG Aphasia 0.01757197 .819 Recurrent stroke 0.1761855 .02078 and 74 severe-RPG) who were assessed in cognition items FAM-SEQ 0.1443192 .05892 according to the inclusion criteria. ECO-SEQ 0.1791327 .01871 In the initial sample 63.37% were severe-RPG and 36.63% ENV-SEQ 0.2449986 .001199 0.0599 were moderate-RPG, similarly, in this subset 62.71% were SUP-SEQ 0.174516 .02204 severe-RPG and 37.29% were moderate-RPG. SEC-SEQ 0.1443091 .05893 Measurements of cognition included: temporal orientation, TOT-SEQ 0.2649201 .0004447 0.07017 spatial orientation, personal orientation, verbal comprehension, Education low/med-high 0.007391863 .9233 verbal denomination, and verbal repetition, as well as 5 C-FIM Fallers 0.1078687 .159 items (auditory comprehension, verbal expression, social inter- BMI = body mass index, C-FIM Adm = cognitive FIM at admission, ECO = economic, ENV = action, memory, and problem solving). environmental, FAM = family, FIM = functional independence, LOS, length of stay, M-FIM Adm = As shown in supplementary material (http://links.lww.com/ motor FIM at admission, NIHSS = National Institutes of Health Stroke Scale, SEQ = socio economic MD/E966), neither orientation nor verbal fluency items correlat- questionnaire, SOC = social, SUP = support, TOT = total, T-FIM Adm = total FIM at admission. Variables with P < .002 should be in bold. ed with LOS. Two of the 5 C-FIM items were significantly correlated with LOS (social interaction and problem resolution) but C-FIM was already included in 1 of the models presented in 3.4. Bivariate correlation analysis: demographics, clinical, the previous section (model #4) and in this work we did not and social variables with LOS further analyze the specific contribution of C-FIM items. Potential associations with LOS were examined among 30 variables using bivariate correlations. The following variables 4. Discussion (Table 6) were significantly correlated with LOS: Previous studies have mostly focused on investigating the severity (moderate-RPG and severe-RPG), NIHSS, M-FIM-Adm, prediction of LOS within the full severity spectrum of stroke C-FIM-Adm, T-FIM-Adm, ENV-SEQ, and TOT-SEQ. rehabilitation patients. Such studies involved a wide range of No other variables were found to be significant at our number of participants, e.g., n = 586,[29] n = 11,983,[2] n = conservative a level (0.05/30 = 0.0017); however, it should be 117,[19] n = 3839,[30] n = 151,[31] n = 165,[23] n = 353.[32] noted that side of paresis, depression, recurrent stroke, and 2 To our best knowledge, very few previous studies targeted a socioeconomic items (ECO-SEQ and SUP-SEQ) were all specific severity group, for example McClure et al[5] analyzed significantly correlated with LOS at P < .05. predictors of LOS in patients (n = 134) admitted to stroke rehabilitation with high levels of functional independence (T- FIM > 100). According to their findings, the 2 most predictive 3.5. Multivariate regression analysis variables were M-FIM score at admission and orientation to As shown in sup material M-FIM-Adm, C-FIM-Adm, T-FIM- person, place, and time, although verbal communication ability Adm are strongly correlated, as well as ENV-SEQ and TOT-SEQ; also contributed significantly to the model, collectively account- therefore, we included the following variables in model #1: ing for 60% of the variance associated with LOS.[5] Our results, severity, NIHSS, T-FIM-Adm, and TOT-SEQ (presented in similar to McClure in number of participants and in the set of Table 7). A significant model emerged (P < .00001) with an considered variables, also identified FIM score (motor, cognitive, adjusted R2 = 0.1982 (R2 = 0.2176), T-FIM-Adm contributed and total) as the most predictive variables, but neither orientation significantly to the model; nevertheless, it presents a multi- to person, place, and time, nor verbal communication ability collinearity problem (tolerance = 0.35). were found significant. 7 García-Rudolph et al. Medicine (2020) 99:43 Medicine Table 7 Multivariate lineal regressions, nonstandard beta with CIs, standard beta, Durbin test, VIF, R2, and adjusted R2. Model Variables b (95% CI) t test Std b VIF Tol P R2 Adj R2 1 Severity 9.33 ( 1.97, 20.65) 1.63 0.16 2.20 0.45 .105 0.2176 0.1989 NIHSS 0.028 ( 0.81, 0.87) 0.067 0.005 1.37 0.72 .946 T-FIM-Adm 0.39 ( 0.69, 0.09) 2.57 0.29 2.83 0.35 .0108 TOT-SEQ 0.73 ( 0.46, 1.93) 1.20 0.086 1.09 0.91 .229 Durbin test D-W = 1.81, P = .216 2 Hemiparesis 9.48 ( 16.75, 2.22) 2.579 0.17 1.02 0.97 .0108 0.2619 0.2442 Recurrent 9.26 ( 21.74, 3.20) 1.467 0.09 1.03 0.96 .1443 T-FIM-Adm 0.54 ( 0.71, 0.36) 6.019 0.41 1.05 0.94 < .0001 Depression 6.67 ( 0.82, 14.17) 1.758 0.11 1.05 0.95 .0806 Durbin test D-W = 1.87, P = .386 3 Hemiparesis 9.54 ( 16.91, 2.17) 2.55 0.17 1.05 0.95 .011 0.256 0.2382 Severity 5.11 ( 6.04, 16.28) 0.90 0.09 2.25 0.44 .366 T-FIM-Adm 0.47 ( 0.73, 0.21) 3.60 0.35 2.21 0.45 < .0001 Depression 6.36 ( 1.18, 13.91) 1.66 0.11 1.05 0.94 .098 Durbin test D-W = 1.90 P = .59 4 Hemiparesis 12.55 ( 20.60, 4.50) 3.07 0.23 1.22 0.81 .0024 0.241 0.2232 Severity 16.26 (8.19, 24.34) 3.97 0.28 1.15 0.86 .00010 C-FIM-Adm 0.69 ( 1.13, 0.24) 3.08 0.23 1.26 0.78 .002384 Depression 7.36 ( 0.23, 4.96) 1.91 0.13 1.05 0.95 .057 Durbin test D-W = 1.86, P = .35 5 Hemiparesis 5.85 ( 13.34, 1.63) 1.54 0.10 1.05 0.95 .122 0.2308 0.2124 Severity 5.81 ( 7.24, 18.88) 0.88 0.10 2.99 0.33 .380 M-FIM-Adm 0.54 ( 0.94, 0.13) 2.66 0.31 3.10 0.32 .0085 Depression 6.24 ( 1.45, 13.95) 1.60 0.11 1-06 0.93 .111 Durbin test D-W = 1.957, P = .78 6 Hemiparesis 9.86 ( 17.55, 2.17) 2.53 0.18 1.074 0.93 .0122 0.2027 0.1885 Severity 18.69 (10.53, 26.84) 4.52 0.331 1.13 0.88 < .000001 NIHSS 0.83 (0.05, 1.60) 2.12 0.156 1.14 0.87 .0352 Durbin test D-W = 1.87, P = .394 7 Hemiparesis 8.35 ( 15.86, 0.85) 2.19 0.15 1.02 0.97 .0293 0.201 0.1867 Severity 19.92 (12.03, 27.80) 4.98 0.353 1.05 0.94 < .000001 TOT-SEQ 1.21 (0.034, 2.39) 2.03 0.12 1.03 0.96 .0437 Durbin test D-W = 1.83, P = .266 8 Hemiparesis 8.25 ( 15.81, 0.69) 2.15 0.15 1.04 0.95 .0326 0.2141 0.1952 Severity 17.88 (9.68, 26.09) 4.30 0.31 1.15 0.86 < .000001 Recurrent 12.20 ( 25.18, 0.76) 1.85 0.13 1.04 0.95 .065 ENV-SEQ 4.56 (0.28, 8.85) 2.10 0.15 1.13 0.87 .0369 Durbin test D-W = 1.79, P = .188 9 Hemiparesis 7.64 ( 15.16, 0.13) 2.00 0.14 1.03 0.96 .0462 0.2135 0.1946 Severity 19.16 (11.26, 27.06) 4.78 0.34 1.07 0.93 < .000001 Recurrent 10.59 ( 23.45, 2.27) 1.62 0.11 1.03 0.96 .1059 TOT-SEQ 1.23 (0.05, 2.40) 2.07 0.14 1.03 0.96 .0399 Durbin test D-W = 1.81, P = .188 C-FIM = cognitive FIM, ECO = economic, ENV = environmental, FAM = family, FIM = functional independence measure, M-FIM = motor FIM, NIHSS = National Institutes of Health Stroke Scale, SEQ = socioeconomic questionnaire, SUP = support, SOC = social, TOT = total, T-FIM = total FIM, T-FIM = M-FIM + C-FIM. Our main contributing factor is T-FIM. As presented in M-FIM is our second main contributing factor, found as main Table 7, for model #2, T-FIM-Adm standardized Beta is 0.41, factor in Grant et al[2] according to their results, each 1-point therefore, as T-FIM-Adm increases by 1 standard deviation improvement in the M-FIM score predicted a reduction in LOS by (20.67 points) the LOS decreases by 0.41 standard deviations, the 0.6 days. In our case, for model #5 M-FIM-Adm standardized standard deviation of LOS is 27.23, so this constitutes a change of Beta is 0.31; therefore, each 1 point improvement in M-FIM 11.16 days (0.4127.23 = 11.16); therefore, each 1 point Adm score predicted a reduction in LOS by 0.52 days. In improvement in T-FIM-Adm score predicted a reduction in McClure et al[5] regression model M-FIM Adm standardized Beta LOS by 0.54 days. In Yang et al[19] T-FIM-Adm standardized is 0.29, each 1 point improvement in M-FIM-Adm score Beta is 0.375, therefore, as T-FIM-Adm increases by 1 standard predicted a reduction in LOS by 0.51 days. deviation (21.84 points) the LOS decreases by 0.375 standard Each 1 point improvement in M-FIM score predicted a deviations, the standard deviation of LOS is 22.11, so this reduction in LOS of half a day in both McClure model (mild constitutes a change of 8.29 days (0.37522.11 = 8.29); patients)[5] and in our model (mostly severe patients); meanwhile, therefore, each 1 point improvement in T-FIM-Adm score Grant model (mild, moderate, and severe patients) predicted a predicted a reduction in LOS by 0.38 days according to Yang slightly larger reduction in LOS (0.6 days).[2] As shown in Table 7 et al.[19] our model #5 may present some multicollinearity problem, 8 García-Rudolph et al. Medicine (2020) 99:43 www.md-journal.com tolerance is slightly below 0.40, but VIF is quite below 5; As presented in Table 1, higher values of TOT-SEQ indicate therefore, our values are acceptable.[27] higher social risks, the negative correlations remark that higher In McClure study,[5] social factors, such as informal support or social risks are moderately associated with lower cognitive, family living arrangement, were not found significant contrib- motor and total functionality at admission. utors to LOS. Lai et al[23] identified some of them (e.g., living We also identified correlations between TOT-SEQ and NIHSS, alone) as predictors of exceeding target LOS. Their study in this case, the higher the social risk the higher the NIHSS (r = included all 3 severity levels (12.3% mild-RPG, 21% severe- 0.19, P < .05); therefore, higher severity levels are also associated RPG, and 55% moderate-RPG). Of the n = 165 included patients, with higher social risks (though in this case the correlation is 34% exceeded the optimal LOS; meanwhile in our case, applying weaker than with FIM). the same criteria for optimal LOS, 86% of our patients exceeded When stratifying TOT-SEQ score into mild social risk, it, as shown in Table 2. In Lai’s sample, median LOS for important social risk, and severe social risk using the criteria moderate-RPG and severe-RPG patients were 29 and 48 days.[23] for stratification presented in Table 1, we identified significant The median LOS reported in our study are 47 days for moderate- between-groups differences in functionality. RPG and 67 days for severe-RPG. Lai’s mean FIM gain were 23.1 The median T-FIM at admission for the mild, important, and (14.6) and 27.4(14.2),[23] in our case mean FIM gain were 26.50 severe groups were 61.5, 50, and 41, with significant differences (12.39) and 40.11 (22.4) respectively. This may indicate that between the mild and important groups (P < . 001); also larger LOS in severe-RPG in our sample may lead to larger FIM significant differences were found between mild and severe gain. But when considering FIM efficiency (defined as FIM gain groups (P < .001). divided by LOS) as presented in Table 3, we reported an FIM The median NIHSS for the mild, important, and severe groups efficiency median of 0.53 for moderate-RPG and 0.63 for severe- were 13, 14, and 16. Significant differences were found between RPG; meanwhile, Lai’s are 0.8 and 0.6 respectively.[23] Therefore mild and severe groups (P < .05). Lai’s shows remarkable higher efficiency in moderate-RPG and we in severe-RPG. 4.1. Study limitations In relation to specific socioeconomic variables, models #7, #8, and #9 in Table 7 include significant SEQ items. Our best adj r2 is One of the main limitations of this study is a consequence of the obtained in model #8 which includes ENV-SEQ, as described in data being collected in 1 single urban center that covered Table 1, ENV-SEQ refers to environmental, a.k.a. physical rehabilitation care, suggesting that the generalization of these conditions at home, it ranges from Appropriate to your needs results should be considered carefully. These findings need to be (1 point), Architectural barriers with possibilities for adaptation replicated in larger samples to determine whether they are (2 points), or Architectural barriers without possibility of generalizable. adaptation (3 points). As shown in Table 7 ENV-SEQ Another limitation is that our models explained 24.42% of the standardized Beta is 0.15, therefore each 1 point increase in total variance in LOS, which means that approximately 75% of ENV-SEQ score predicted an increase in LOS by 457 days. the variation in LOS was influenced by other factors. Regarding recurrent stroke, Lai et al[23] conclude that patients The median LOS reported in our study, 47 days for moderate- with recurrent stroke require a longer LOS during inpatient RPG patients and 67 days for severe-RPG patients, are quite long, rehabilitation than patients without recurrent stroke, indepen- especially when compared with LOS in the United States or dent of stroke severity. In our case recurrent stroke was included Canada, for example, Grant et al[2] reported a median stroke in 3 models in Table 7, but did not contribute significantly in any rehabilitation LOSs of 35 days. Thus, our findings may not be of them. generalizable to rehabilitation facilities in countries where LOS is In relation to the explained variance, our best is model #2 (adj significantly shorter. Nevertheless, a recent study in southern R2 = 24.42%), remarkably lower than McClure’s. Our results in Ontario, Canada, involving n = 117 patients reported a median that sense are similar to Grant et al[2] their study includes all LOS of 45 days[19] quite similar to our moderate–RPG patients. stroke severity levels and their multivariable regression model Previous studies have shown that higher severity as measured, explains 20% of the variation of LOSs. Grant et al[2] considered a for example, by NIHSS increases LOS, providing a possible similar set of variables but a notably larger dataset (n = 11,983). explanation for our larger LOS. For example Appelros’ multiple Therefore, our results suggest that factors outside functional, regression analysis showed that each point on the NIHSS socioeconomic, medical, and demographic patient characteristics increased the total LOS by 3.4 days.[3] Similarly, in our case, as have important influences on LOSs, being such factors more presented in Table 7, for model #6, NIHSS standardized Beta is relevant in our dataset (which includes 63.37% severe-RPG) than 0.156; therefore, each 1 point increase in NIHSS score predicted in similar studies including mild patients or the full severity an increase in LOS by 0.83 days. spectrum. During the past few years, progress has been made toward The second main aspect addressed in our study is the association identifying the roles of important inflammatory signaling between socioeconomic variables with stroke severity (NIHSS) and molecules, cells, and proteins in the process of initiation and functionality at admission (T-FIM, M-FIM, and C-FIM). development of poststroke inflammation. Clinically, the suscep- The association between stroke severity and socioeconomic tibility of the patients to stroke and the subsequent prognosis are situation has been scarcely reported in previous research, for influenced by such inflammatory processes.[33] As stroke patients example, lower SES and having more severe deficits after stroke with systemic inflammation have been reported to exhibit assessed by NIHSS at admission.[22] clinically poorer outcomes,[34] it is an important element to According to our results, TOT-SEQ correlated moderately consider in future work. Specifically, increasing evidence shows with C-FIM-Adm (r = 0.22, P = .01), with M-FIM-Adm (r = that inflammation plays an important role in the progression of 0.27, P < .001) and slightly stronger with T-FIM-Adm (r = intracerebral hemorrhage inflammation.[35] Therapeutic target- 0.32, P < .0001). ing of inflammation in acute stroke has gained interest as a 9 García-Rudolph et al. Medicine (2020) 99:43 Medicine potential neuroprotective strategy. For example, matrix metal- Author contributions loproteinases are part of the neuro-inflammatory process that AGR, BC and JMT conceived the study, AGR and BC collected, occurs during intracerebral hemorrhage and are, thus, also ideal selected, and cleaned the data. AGR, EO and JS performed the as biomarkers and therapeutic targets in intracerebral hemor- statistical analysis of the data. AGR and BC drafted the initial rhage treatment.[35] Notably, over recent decades C-reactive manuscript. EO, JMT, MB and JS revised the manuscript protein has been the focus of an intense investigation to explore critically for important intellectual content and approved the final its role in the setting of intracerebral hemorrhage and currently is manuscript. proposed as a risk assessment tool and prognostic marker[36] and can also be considered in future research as an LOS potential References predictor. Furthermore, the neutrophil-to-lymphocyte ratio[37] integrates information on both the innate and adaptive compart- [1] Hung C-Y, Wu W-T, Chang K-V, et al. Predicting the length of hospital ments of the immunity and represents a reliable measure of the stay of post-acute care patients in Taiwan using the Chinese version of the continuity assessment record and evaluation item set. PLoS One 2017;12: inflammatory burden that could contribute to increasing e0183612. knowledge about some of the mechanisms involved in the [2] Grant C, Goldsmith CH, Anton HA. Inpatient stroke rehabilitation spontaneous intracerebral hemorrhage-induced injury and yield lengths of stay in Canada derived from the national rehabilitation information on the disease course[37,38] and therefore can also be reporting system, 2008 and 2009. Arch Phys Med Rehabil. 2014;95: 74–78. considered LOS predictor in future work. [3] Appelros P. Prediction of length of stay for stroke patients. Acta Neurol Scand 2007;116:15–9. [4] Chung L, Wang YH, Chen TJ, et al. The predictive factors for length of 5. Conclusions stay for stroke patients in Taiwan using the path model. Int J Rehabil Res In this study, we analyzed the associations between FIM at 2006;29:137–43. [5] Mcclure J, Andrew Salter, Katherine Meyer, et al. Predicting length of admission and SES within a population of ischemic and stay in patients admitted to stroke rehabilitation with high levels of hemorrhagic stroke rehabilitation patients (nonmild according functional independence. Disability Rehabilitation 2011;33:2356–61. to RPG benchmark). We found that Motor FIM, Cognitive FIM, [6] Ng YS, Stein J, Ning M, et al. Comparison of clinical characteristics and and TOTAL FIM at admission are moderately correlated with functional outcomes of ischemic stroke in different vascular territories. SES. Besides, we found that NIHSS is moderately correlated Stroke 2007;38:2309–14. [7] Patel DA, Krishnaswami S, Steger E, et al. Economic and survival burden with SES, being this association weaker than with functional of dysphagia among inpatients in the United States. Dis Esophagus independence. 2018;31:1–7. When stratifying the patients’ socioeconomic situation into [8] Mamman Watila M, Abdullahi Balarabe S. Factors predicting post- mild social risk, important social risk, and severe social, we stroke aphasia recovery. J Neurol Sci 2015;352:12–8. [9] Lau LH, Lew J, Borschmann K, et al. Prevalence of diabetes and its effects identified significant between-groups differences in functionality on stroke outcomes: a meta-analysis and literature review. J Diabetes between the mild social risk and important social risk group and Investig 2019;10:780–92. between the mild social risk and severe social risk group. [10] Oesch L, Tatlisumak T, Arnold M, et al. Obesity paradox in stroke— Our results indicate an association between stroke functionali- myth or reality? A systematic review. PLoS One 2017;12:e0171334. ty at admission and socioeconomic situation, confirming our first [11] Kim SM, Hwang SW, Oh EH, et al. Determinants of the length of stay in stroke patients. Osong Public Health Res Perspect 2013;4:329–41. hypothesis. [12] Hardie K, Hankey GJ, Jamrozik K, et al. Ten year risk of first recurrent We analyzed 30 state-of-the art predictors of LOS and found stroke and disability after first-ever stroke in the Perth community stroke that a few of the same variables that have been identified as study. Stroke 2004;35:731–5. significant predictors of LOS within the full stroke population, [13] Quinn TJ, Paolucci S, Sunnerhagen KS. Evidence-based stroke rehabili- tation: an expanded guid-ance document from the European Stroke were also significant predictors in our sample (FIM, home Organisation (ESO) guidelines for management of ischaemic stroke and accessibility, NIHSS). transient ischaemic attack. J Rehabil Med 2008;41:99–111. Nevertheless, most of LOS predictors found significant in the [14] Prokopenko SV, Mozheyko EY, Petrova MM, et al. Correction of post- literature, were not for our sample, for example, depression, falls, stroke cognitive impairments using computer programs. J Neurol Sci recurrent stroke, ataxia, orientation, verbal communication. 2013;325:148–53. [15] Wong JS, Brooks D, Mansfield A. Do falls experienced during inpatient In relation to the explained variance (24%), our results suggest stroke rehabilitation affect length of stay. Functional status, and that factors outside functional, socioeconomic, medical, and discharge destination? Arch Phys Med Rehabil 2016;97:561–6. demographic patient characteristics have important influences on [16] Gainotti G, Antonucci G, Marra C, et al. Relation between depression LOSs, being such factors more relevant in our dataset (which after stroke, antidepressant therapy, and functional recovery. J Neurol Neurosurg Psychiatry 2001;71:258–61. includes 63.37% severe-RPG) than in similar studies including [17] Tan WS, Heng BH, Chua KF, et al. Factors predicting inpatient mild patients or the full severity spectrum. Administrative and rehabilitation length of stay of acute stroke patients in Singapore. Arch therapy-specific variables (intensity, therapy type, inpatient Phys Med Rehabil 2009;90:1202–7. services, etc) may be important factors that influence LOS. [18] Saxena SK, Koh GC, Ng TP, et al. Determinants of length of stay during These factors were neither measured nor evaluated in this study post-stroke rehabilitation in community hospitals. Singapore Med J 2007;48:400–7. offering opportunities for future work. [19] Yang C, Finestone H, Chen P. Psychosocial determinants of longer length of stay on a Canadian stroke inpatient rehabilitation unit. Int J Phys Med Rehabil 2015;3:281. Acknowledgment [20] New PW, Jolley DJ, Cameron PA, et al. A prospective multicentre study of barriers to discharge from inpatient rehabilitation. Med J Aust Special thanks go to Jaume Lopez Carvallo from Institut 2013;198:104–8. Guttmann’s Research and Innovation Department for his [21] Marshall I, Wang Y, Crichton S, et al. The effects of socioeconomic status support in data access. on stroke risk and outcomes. Lancet Neurol 2015;14:1206–18. 10 García-Rudolph et al. Medicine (2020) 99:43 www.md-journal.com [22] Kerr GD, Higgins P, Walters M, et al. Socioeconomic status and transient [31] Nkemdirim O, Arinze R, Colleen A. Predictors of hospital length of stay ischaemic attack/stroke: a prospective observational study. Cerebrovasc and readmissions in ischemic stroke patients and the impact of inpatient Dis 2011;31:130–7. medication management. J Stroke Cerebrovasc Dis 2016;25:1939–51. [23] Lai W, Buttineau M, Harvey JK, et al. Clinical and psychosocial [32] Gaspari AP, Cruz EDA, Batista J, et al. Predictors of prolonged hospital predictors of exceeding target length of stay during inpatient stroke stay in a comprehensive stroke unit. Rev Lat Am Enfermagem 2019;27: rehabilitation. Top Stroke Rehabil 2017;24:510–6. e3197. [24] STROBE. Available at: https://strobe-statement.org/index.php?id=avail [33] Di Raimondo D, Tuttolomondo A, Buttà C, et al. Effects of ACE able-checklists. Accessed March 2020. inhibitors and angiotensin receptor blockers on inflammation. Curr [25] Cabrera González A, Menenéndez Caicoya A, Salamea García D. Pharm Des 2012;18:4385–413. Evaluation of the reliability and validity of a scale of social evaluation of [34] Elkind MS, Cheng J, Rundek T, et al. Leukocyte count predicts outcome the elderly. Aten Primaria 1999;23:434–40. after ischemic stroke: the Northern Manhattan Stroke Study. J Stroke [26] Cohen J. A power primer. Psychol Bull 1992;112:155–9. Cerebrovasc Dis 2004;13:220–7. [27] O’Brien RM. “A caution regarding rules of thumb for variance inflation [35] Lattanzi S, Di Napoli M, Ricci S, et al. Matrix metalloproteinases in acute factors”. Quality Quantity 2007;41:673–90. intracerebral hemorrhage. Neurotherapeutics 2020;17:484–96. [28] Rakesh N, Boiarsky D, Athar A, et al. Post-stroke rehabilitation: factors [36] Di Napoli M, Slevin M, Popa-Wagner A, et al. Monomeric C-reactive predicting discharge to acute versus subacute rehabilitation facilities. protein and cerebral hemorrhage: from bench to bedside. Front Immunol Medicine 2019;98:e15934. 2018;9:1921. [29] Lin CL, Lin PH, Chou LW, et al. Model-based prediction of length of stay [37] Lattanzi S, Brigo F, Trinka E, et al. Neutrophil-to-lymphocyte ratio in for rehabilitation stroke patients. J Formos Med Assoc 2009;108: acute cerebral hemorrhage: a system review. Transl Stroke Res 2019; 653–62. 10:137–45. [30] Kang J-H, Bae H-J, Choi Y-A, et al. Length of hospital stay after [38] Zangari R, Zanier ER, Torgano G, et al. Early ficolin-1 is a sensitive stroke: a Korean Nationwide Study. Ann Rehabil Med 2016;40: prognostic marker for functional outcome in ischemic stroke. J 675–81. Neuroinflammation 2016;13:16. 11