UBC Faculty Research and Publications

Adverse outcomes following hospitalization in acutely ill older patients Wong, Roger Y; Miller, William C May 14, 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12877_2008_Article_104.pdf [ 244.33kB ]
JSON: 52383-1.0132594.json
JSON-LD: 52383-1.0132594-ld.json
RDF/XML (Pretty): 52383-1.0132594-rdf.xml
RDF/JSON: 52383-1.0132594-rdf.json
Turtle: 52383-1.0132594-turtle.txt
N-Triples: 52383-1.0132594-rdf-ntriples.txt
Original Record: 52383-1.0132594-source.json
Full Text

Full Text

ralssBioMed CentBMC GeriatricsOpen AcceResearch articleAdverse outcomes following hospitalization in acutely ill older patientsRoger Y Wong*1,4 and William C Miller2,3,4Address: 1Division of Geriatric Medicine, Department of Medicine, University of British Columbia, Canada, 2Department of Occupational Science and Occupational Therapy, University of British Columbia, Canada, 3GF Strong Rehabilitation Research Laboratory, Vancouver, British Columbia, Canada and 4Vancouver Coastal Health Research Institute, Vancouver, British Columbia, CanadaEmail: Roger Y Wong* - rymwong@interchange.ubc.ca; William C Miller - bcmiller@telus.net* Corresponding author    AbstractBackground: The longitudinal outcomes of patients admitted to acute care for elders units (ACE)are mixed. We studied the associations between socio-demographic and functional measures withhospital length of stay (LOS), and which variables predicted adverse events (non-independent living,readmission, death) 3 and 6 months later.Methods: Prospective cohort study of community-living, medical patients age 75 or over admittedto ACE at a teaching hospital.Results: The population included 147 subjects, median LOS of 9 days (interquartile range 5–15days). All returned home/community after hospitalization. Just prior to discharge, baseline timedup and go test (TUG, P < 0.001), bipedal stance balance (P = 0.001), and clinical frailty scale scores(P = 0.02) predicted LOS, with TUG as the only independent predictor (P < 0.001) in multipleregression analysis. By 3 months, 59.9% of subjects remained free of an adverse event, and by 6months, 49.0% were event free. The 3 and 6-month mortality was 10.2% and 12.9% respectively.Almost one-third of subjects had developed an adverse event by 6 months, with the highest riskwithin the first 3 months post discharge. An abnormal TUG score was associated with increasedadjusted hazard ratio [HR] 1.28, 95% confidence interval [CI] 1.03 to 1.59, P = 0.03. A higherFMMSE score (adjusted HR 0.89, 95% CI 0.82 to 0.96, P = 0.003) and independent living beforehospitalization (adjusted HR 0.42, 95% CI 0.21 to 0.84, P = 0.01) were associated with reduced riskof adverse outcome.Conclusion: Some ACE patients demonstrate further functional decline following hospitalization,resulting in loss of independence, repeat hospitalization, or death. Abnormal TUG is associatedwith prolonged LOS and future adverse outcomes.BackgroundAcute care for elders units (ACE) focus on early rehabilita-cally independent or terminally ill are less likely to benefit[5]. Since the publication of the original randomized trialPublished: 14 May 2008BMC Geriatrics 2008, 8:10 doi:10.1186/1471-2318-8-10Received: 10 January 2008Accepted: 14 May 2008This article is available from: http://www.biomedcentral.com/1471-2318/8/10© 2008 Wong and Miller; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 9(page number not for citation purposes)tion, discharge planning, and delivering functionally ori-ented, patient-centered care [1,2]. ACE can improveoutcomes [3,4], although some patients who are physi-[3], implementation of ACE has not been widespread[6,7]. Possible explanations include financial costs [6],mixed findings from subsequent studies [8-11], difficultyBMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10in predicting outcomes, and a lack of data on whether anybenefit is sustainable. The prediction of ACE outcomes isinfluenced by multiple factors [12-15]. Recently, a system-atic review showed that physical function, illness severity,cognition, comorbidity, presenting medical diagnosis,multiple medication use, and age could affect hospitallength of stay (LOS), readmissions, discharge destinationand mortality [16]. However, these findings have notbeen reproduced in prospective studies. In most hospitals,ACE patients are selected by age [3], although there maybe benefits to selecting patients based on pre-morbidfunctional status. It remains unclear whether any of thecommonly used clinical parameters, such as mobility andbalance scores, cognitive and depression scores, illnessseverity and comorbidity scores, and clinical frailty scores,are useful in predicting the outcome of ACE patients. Ide-ally some of these standardized measurements mightproactively identify individuals in ACE who are at risk forprolonged hospitalization, frequent hospital readmis-sion, loss of independent residence following hospitaliza-tion, and death, so that specific strategies to optimize ACEoutcomes can be targeted.The overall goal of this article is to describe the outcomesof a population of older patients admitted to ACE by 3 to6 months post discharge. Specifically, we report on theassociations between socio-demographic and perform-ance-based variables in mobility, balance, cognition,depression and activities of daily living (ADL) function atdischarge with LOS during hospitalization (primary out-come), and which variables best predict mortality,readmission to acute care, and living disposition at 3 and6 months after hospital discharge (secondary outcomes).MethodsSettingThis is a prospective cohort study that recruited a sampleof older adults admitted to 2 ACE units at the VancouverGeneral Hospital, which comprise 44 medical beds underthe care of either internal medicine (clinical teachingunits) or family medicine hospitalists. Detailed descrip-tion of ACE was previously described [17]. We receivedapproval from the institutional research ethics board toconduct this study.SubjectsWe included ACE patients who consented to participate.They were eligible if they were 75 years or over (our hos-pital used this age cut-off as a surrogate marker to deter-mine ACE eligibility), lived in the community pre-hospitalization, and could comprehend simple three-stepcommands in English. We also included ACE patientswho transitioned through a separate sub-acute medicalnesses have stabilized but require time for functionalrecovery before returning home. We recognize the hetero-geneity between ACE and SAM patients, but decided toinclude the ACE-SAM patients because they represented acommon hospital trajectory post ACE stay, while increas-ing our sample size and power of the study. We did notinclude subjects if they were transferred from/to criticalcare or palliative care because these populations are notnormally serviced by ACE; residing at a long term carefacility prior to hospitalization; residing outside the catch-ment of the hospital (greater than 100 km distance); ordeemed medically unstable by 1 of 2 internal medicineresidents who reviewed each subject's physiologic param-eters prior to any performance-based measure.We calculated a priori the required sample size to be 150,using an alpha = 0.01, beta = 0.20 and over sampling for20% attrition, which enabled modelling of up to 10 inde-pendent variables including interaction terms [18] at whatCohen [19] defined as a moderate to large effect size. Torecruit 150 eligible subjects who would consent, we endedup screening ACE patients between October 2004 andOctober 2005 until the target sample size was reached.Three subjects declined to continue before baseline test-ing, therefore leaving 147 subjects from whom we col-lected baseline data (Table 1). The ACE subjects in thisstudy were representative of the typical ACE populationwho survived hospitalization, at least based on age, sex,medical diagnosis, medication number and mobilityindependence when compared to the previously reportedconsecutive patient series from our ACE [10]. In the cur-rent cohort, there was no age and sex difference betweenstudy subjects and non-participants. Unfortunately we didnot have authorized access to other baseline data on thenon-participants.Data collection and outcomesAll data was collected in an unblinded fashion. We pre-screened consecutive admissions to ACE and identifiedpotential subjects. Informed consent was obtained fromsubjects, all of whom were capable of granting consent.We used uniform definitions and obtained the followingbaseline data from the hospital health records: socio-demographic data, medical diagnosis as defined by pre-established categories, and number of prescription medi-cations at admission. We used the cumulative illness rat-ing scale (CIRS) to measure medical complexity andcomorbidity (score range 0–56), with higher scores indi-cating greater disease burden [20]; the geriatric prognosticindex (GPI) to estimate the 1-year mortality risk after hos-pital discharge (score range 0–26), with higher scores pre-dictive of higher mortality risk [21]; and the clinical frailtyscale (CFS) to estimate the degree of fitness and frailtyPage 2 of 9(page number not for citation purposes)(SAM) unit after their initial stay in ACE. The SAM unitcomprises of 32 medical beds for patients whose acute ill-(score range 1–7), with higher scores indicating moresevere frailty [22]. We also counted the number of inde-BMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10pendent ADL among 5 activities (eating, continence and/or functional ability to toilet, dressing, transferring, andbathing) prior to hospitalization.We selected a number of validated performance-basedtests that could be implemented without adding substan-tial burden to patients or workload to clinicians. Subjectsunderwent baseline testing in ACE just prior to antici-pated discharge (median 2 days before discharge). Weopted to obtain the functional measures just before dis-charge instead of at admission to allow for clinical stabili-zation of the patients, and to study how function couldpredict future adverse outcomes post hospitalization.Testing began with the 2-minute walk test (2MWT) toassess walking endurance and as a proxy of communityambulation potential [23], followed by a rest station dur-ing which the Folstein mini-mental state examinationto assess walking skill/speed [23], followed by anotherrest station when the short form geriatric depression scale(GDS) was done to screen for major depression [25]. Test-ing finished with a battery of balance tests of lowerextremity function (sitting to standing, standing with 2feet together, standing on 1 leg, standing in a semi-tan-dem position, and standing in a full tandem position)that have been shown to be predictive of subsequent dis-ability, nursing home admissions and mortality [26]. Alltests were explained and demonstrated to the subjects,and conducted according to standardized protocols. Sub-jects who demonstrated excessive fatigue were offered thepossibility to terminate or interrupt testing, with theopportunity to resume at a later time or on another day.No one requested this opportunity.We scheduled the 3 and 6 months follow up home visitsTable 1: Characteristics of study subjects at baseline, 3-month and 6-month follow up. Characteristic Baseline 3-month follow up 6-month follow upN = 147 N = 88 N = 72Age in years 83.9 ± 5.7 83.3 ± 5.2 83.3 ± 5.2Female sex (%) 78 (53.1) 41 (46.6) 36 (50.0)Marital status (%)Married 54 (36.7) 35 (39.8) 27 (37.5)Widowed 67 (45.6) 38 (43.2) 34 (47.2)Single 26 (17.7) 15 (17.1) 11 (15.3)Residence before hospitalization (%)Community independent living 59 (40.1) 43 (48.9) 36 (50.0)Other 88 (59.9) 45 (51.1) 36 (50.0)Medical diagnosis (%)Cardiac disease 21 (14.3) 13 (14.8) 11 (15.3)Pulmonary disease 4 (2.7) 4 (4.6) 2 (2.8)Gastrointestinal disease 31 (21.1) 21 (23.9) 18 (25.0)Infection 33 (22.5) 20 (22.7) 17 (23.6)Neurologic disease 15 (10.2) 7 (8.0) 5 (6.9)Diabetes mellitus 2 (1.4) 1 (1.1) 1 (1.4)Functional decline 5 (3.4) 3 (3.4) 2 (2.8)Cancer 3 (2.0) 0 (0.0) 0 (0.0)Other 32 (21.8) 18 (20.5) 15 (20.8)Number of prescription medications at admission 4.9 ± 3.0 4.7 ± 2.8 4.6 ± 2.8Cumulative illness rating scale score 23.9 ± 4.2 23.9 ± 4.2 23.5 ± 4.2Geriatric prognostic index score 1.7 ± 2.2 1.8 ± 2.0 1.7 ± 2.0Clinical frailty scale score 4.7 ± 0.8 5.1 ± 0.9 4.8 ± 1.0Number of independent ADL 4.8 ± 0.5 4.8 ± 0.5 4.9 ± 0.4Folstein mini-mental state examination score 25.5 ± 3.9 26.2 ± 3.2 26.7 ± 2.8Geriatric depression scale score 3.4 ± 2.6 3.2 ± 2.5 3.0 ± 2.5Distance travelled in 2-minute walk test (meters) 62.9 ± 31.6 69.4 ± 33.8 73.1 ± 33.5Time to complete timed up and go test (seconds) 31.6 ± 22.8 28.3 ± 20.5 26.6 ± 18.1Sitting to standing balance test score 0.8 ± 1.1 0.8 ± 1.1 0.8 ± 1.1Standing with 2 feet together balance test score 2.4 ± 1.2 2.5 ± 1.2 2.6 ± 1.2Standing on 1 leg balance test score 1.2 ± 1.2 1.3 ± 1.3 1.4 ± 1.3Plus-minus values are mean ± SD. Categorical data are reported as number of subjects and percentage in parentheses.Page 3 of 9(page number not for citation purposes)(FMMSE) was done to screen for cognitive impairment[24]. The timed up and go test (TUG) was then completedfor all subjects at the time of hospital discharge providingreminders by mail one week and a telephone call 3 daysBMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10prior to the actual visits. During each home visit, we con-ducted a structured personal interview to collect self-reported information on the current living dispositionand the number of hospital readmissions in the timeelapsed since the last point of data collection. A readmis-sion was defined as a return to any hospital for at least 1overnight stay. Subjects were allowed as much time asnecessary to complete the interview, and could interrupttesting at any time if rest was needed.The primary study outcome was LOS during hospitaliza-tion, defined as the number of hospital days spent duringan entire hospital encounter based on hospital records.The secondary outcomes were self-reported living disposi-tion, readmissions to any hospital since the index hospi-talization in ACE, and death at 3 and/or 6 months postdischarge. In addition, we defined an adverse event (acomposite outcome) as non-independent living after hos-pitalization (that is, any living arrangement other than liv-ing in own or rental home), hospital readmission, ordeath at follow up.Statistical analysisDescriptive statistics were computed for all variables:median and interquartile range for LOS, means and stand-ard deviations for other continuous variables, and fre-quencies and percentages for categorical variables. Weused multiple linear regression modeling to estimate theeffect of variables on the primary outcome of LOS. Theindependent variables entered into the model includeddemographic variables, namely age, sex, marital status,independent living before hospitalization, clinical meas-ures, such as CIRS, GPI, CFS at baseline, number of med-ications taken, number of independent ADL pre-admission, and functional measures including FMMSEscores, GDS scores, TUG scores, and the balance testscores while standing with 2 feet together. To satisfy nor-mality assumptions, a logarithmic transformation wasused. Standardized regression coefficients, their 95% con-fidence intervals and P values were reported. For the pur-pose of modeling, living disposition was collapsed intocommunity independent living versus others. Due to co-linearity between independent variables, TUG wasentered into the model instead of 2MWT, the standingwith 2 feet together score was the only one entered amongthe balance tests performed, and standardized indices ofCRIS, GPI and CFS were entered into the model instead ofmedical diagnoses because the former 3 were compositemeasures. While the TUG and 2MWT are highly corre-lated, we selected the TUG because it was performed moreoften in the current sample and also most highly corre-lated with the outcome on a bivariate level. Medical insta-bility was not entered as a separate independent variablewise fashion, and the same results were obtained fromstepwise, forward, and backward regression approaches.In analyzing the secondary outcomes, the probabilities ofany adverse event by 3 and 6 months were calculatedusing Kaplan-Meier survival analysis. A discrete-time Coxproportional-hazards model was used to estimate theunadjusted and adjusted hazard ratios for the compositeadverse outcome. The same independent variables fromthe multiple linear regression analysis were used, exceptwe also included CFS at 3 months as a time-dependentvariable. Hazard ratios, their 95% confidence intervals,and P values are reported. Patients who declined to con-tinue or could not be contacted were considered right-cen-sored, assuming their follow-up times would otherwise belonger, and all contributed to time at risk in the Coxregression analysis. We accepted a level of significancewhen P < 0.05 for all analyses. All data analysis was com-pleted using SAS software, Version 9.1 of the SAS Systemfor Windows, SAS Institute Incorporation, Cary, NC, USA.ResultsOf the 147 ACE patients who had baseline data, 88 sub-jects (59.9%) who remained in the study by 3 months hadnot developed an adverse event of non-independent liv-ing, readmission or death, while 72 (49.0%) were free ofan adverse event by 6 months (Figure 1). There were 46subjects (31.3%) who had developed an adverse event by6 months (see below), and 29 subjects (19.7%) eitherdeclined to continue or could not be contacted. Thesesubjects did not differ in terms of demographics from therest of the sample. If subjects moved to a long term carefacility, they were not interviewed as they were consideredto have experienced an adverse outcome. The characteris-tics of our subjects at each study time point are summa-rized in Table 1. Of note, all 72 subjects who participatedat 6 months were included in the 3-month follow-up. Thepopulation seen at 6 months was similar to those who didnot follow up, except the former had higher FMMSE (26.7± 2.8 vs. 24.2 ± 4.4, P < 0.0001), faster TUG (26.6 ± 18.1vs. 36.3 ± 25.8, P = 0.009), and shorter LOS (10.6 ± 7.9 vs.16.9 ± 20.9, P = 0.017).Data on the primary outcome (LOS) was available for allsubjects (N = 147), who spent a median time of 9 days(interquartile range 5–15 days). We used simple linearregression to estimate the standardized regression coeffi-cients of various characteristics that influenced LOS(Table 2). Specifically, 3 characteristics resulted in signifi-cant bivariable associations: TUG (P < 0.001), balance testscore while standing with both feet together (P = 0.001),and CFS score at baseline (P = 0.02). However, in multipleregression analysis, TUG was the only independent varia-Page 4 of 9(page number not for citation purposes)because of overlap with CIRS and GPI. The independentvariables were entered into the regression model in a step-ble that was significantly associated with LOS (standard-BMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10ized regression coefficient 0.33, 95% confidence interval(95% CI) 0.16 to 0.49, P < 0.001).The probability of an adverse event occurring within thefirst 3 and 6 months post discharge was shown in the Kap-lan-Meier survival analysis (Figure 2). The unadjustedhazard ratios of various risk factors for an adverse eventusing the Cox proportional-hazards model are shown inTable 3. Higher FMMSE score was associated with reducedhazard, whereas longer TUG time and older age were asso-ciated with increased hazard for an adverse event. In mul-and independent living before hospitalization (adjustedHR 0.42, 95% CI 0.21 to 0.84, P = 0.01) were associatedwith reduced hazard, whereas longer TUG times led toincreased hazard ratios (adjusted HR 1.28, 95% CI 1.03 to1.59, P = 0.03 for 20 seconds longer).In addition, we added LOS to the Cox regression analysison adverse events. The hazard of an adverse eventincreased by 4% for each additional day in hospital(unadjusted HR 1.04, 95% CI 1.02 to 1.06, P < 0.0001).After re-running the multiple Cox proportional-hazardsStudy flow diagramFigure 1Study flow diagram. Of the original 150 subjects recruited, 88 participated at the 3-month and 72 participated at the 6-month follow up after discharge from ACE. LTCF = long term care facility.150 ACE subjects granted consent      3 Declined to continue  147 Participated in baseline testing      88 Participated in 3-month follow up       15 Died 15 Moved to LTCF   6 In hospital at 3 months   9 Unable to contact 14 Declined to continue      72 Participated in 6-month follow up   4 Died 3 Moved to LTCF 3 In hospital at 6 months6 Declined to continue        Page 5 of 9(page number not for citation purposes)tiple Cox proportional-hazards analysis higher FMMSEscore (adjusted HR 0.89, 95% CI 0.82 to 0.96, P = 0.003)analysis, LOS replaces TUG. This final model (Table 4)shows higher FMMSE score (adjusted HR 0.87, 95% CIBMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/100.80 to 0.95, P = 0.0011) and independent living beforehospitalization (adjusted HR 0.34, 95% CI 0.15 to 0.74, P= 0.0063) were associated with reduced hazard, whereaslonger LOS led to increased hazard ratios (adjusted HR1.06, 95% CI 1.03 to 1.09, P < 0.0001).DiscussionIn this prospective cohort study we followed olderpatients admitted to ACE longitudinally for 6 monthspost discharge. Of this sample 59.9% remained free of anadverse event of non-independent living, readmission ordeath by 3 months, and 49.0% were still event free by 6months. The 3 and 6-month mortality rate was 10.2% and12.9% respectively. Although all subjects were initiallyable to return home, almost one-third had developed anadverse event by 6 months, with the highest probability ofan adverse event occurring within the first 3 months. Anabnormal TUG score was associated with increased risk ofan adverse event, whereas a higher FMMSE score andindependent living before hospitalization were associatedwith reduced risk. The baseline TUG, bipedal stance bal-ance test, and CFS scores showed significant associationswith LOS during hospitalization, with the TUG score asthe only independent predictor in multiple regressionanalysis.Our study captured ACE patients who survived hospitali-zation and were able to return to independent living atdischarge. Their CFS scores put them in the mildly frailcategory. We did not capture the moderately or severelyfrail patients, who were ineligible or did not consent. Werecognize a number of our study subjects could not be fol-lowed up by 3 and 6 months due to an adverse eventdefined as non-independent living, readmission or death.We used a composite adverse event rather than looking ateach adverse event separately due to relatively small num-bers of events. While our finding might be intuitive for amoderately or severely frail group, it is nonethelessintriguing to see this substantial dropout rate for ourmildly frail cohort.Our findings extend and support the literature on the out-comes and their predictors in older patients followingadmission to ACE. Beyond the immediate benefits of ACEKaplan-Meier survival curve at 3 months and 6 monthsFigure 2Kaplan-Meier survival curve at 3 months and 6 months. The estimated survival probability is 0.76 and 0.67 Table 2: Linear regression analyses to estimate the standardized regression coefficients of various characteristics on hospital length of stay (the primary outcome). Characteristic Standardized regression coefficient (95% CI) P ValueAge 0.03 (-0.14, 0.20) 0.73Male sex 0.05 (-0.13, 0.22) 0.60Married 0.06 (-0.29, 0.41) 0.75Independent living before hospitalization 0.10 (-0.07, 0.27) 0.24Number of medications at admission -0.14 (-0.31, 0.03) 0.10Cumulative illness rating scale score 0.08 (-0.09, 0.25) 0.37Geriatric prognostic index score -0.06 (-0.23, 0.11) 0.52Clinical frailty scale score 0.20 (0.04, 0.37) 0.02Number of independent ADL 0.06 (-0.11, 0.23) 0.46Folstein mini-mental state examination score 0.01 (-0.16, 0.18) 0.94Geriatric depression scale score 0.06 (-0.12, 0.23) 0.52Timed up and go test score 0.33 (0.16, 0.49) <0.001Standing with 2 feet balance test score -0.28 (-0.44, -0.12) 0.001Characteristics with positive coefficients imply increased values were associated with longer length of stay, whereas negative coefficients imply vice versa. CI = confidence interval.Page 6 of 9(page number not for citation purposes)care [3,8,10], ACE patients are at risk for adverse out-comes, especially in the immediate 3 months after hospi-respectively.BMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10talization. This is unlikely due to premature dischargeduring the index hospital stay, for the mean LOS in thissample is actually higher than other published data [10].Nor is this likely an independent effect of age, sex, maritalstatus, comorbidity, polypharmacy or depression basedon our analysis, although we cannot exclude the possibil-ity that other important confounding variables could bemissing. Rather it likely reflects the overall frailty of ACEpatients (clinical frailty scale scores 4.7 ± 0.8 at baseline,5.1 ± 0.9 at 3 months, and 4.8 ± 1.0 at 6 months). Meas-ures of physical function have shown to be predictive ofoutcomes in hospitalized older adults [27-29], and in par-ticular, mobility has been found to be predictive of ADLfunction [30]. The TUG is a proxy of household mobilityand has been found to be reliable and valid in a variety ofolder populations [31,32]. In addition, previous studieson geriatric evaluation and management units have longestablished that pre-morbid function was one of the keypredictors for hospital outcomes [33], at least within thesetting of highly selective patients who were neither toowell nor too frail to benefit. Since our ACE took "all com-ers" age 75 or over (our hospital made this operationaldecision deliberately, not to violate the fundamental pre-cept of targeting appropriate patients, but rather for qual-ity improvement and ethical reasons so that no olderadults would be excluded from receiving ACE care whichwe considered best clinical practice), a functional measurelike TUG would likely be predictive of future outcomes.We observe that the TUG alone accounted for 11% of thevariance in the multiple regression model for LOS (that is,R-squared 0.11), and was the highest and only statisticallysignificant contributor among the other 12 independentvariables selected in our attempt to predict LOS. This hasimportant implications, for the TUG is easy to do, inex-pensive, and does not require extensive training or specialequipment. The other predictors of LOS (standardizedbalance test with 2 feet standing together, the CFS) andpredictors of adverse event (the FMMSE, self reported res-idence before hospitalization), also have potential foreasy implementation clinically. The psychometric proper-ties of the performance based measures are well estab-lished in the literature, therefore allowing uniforminterpretation across different ACE units. We recognizethat other non-clinical, non-functional factors mightimpact on LOS, such as social work availability for dis-charge planning, day of admission during the week, etc.The first 3 months following discharge from ACE repre-sents a high-risk period of non-independent living,readmission or death, and the risk thereafter appears totaper somewhat. This raises the question of whether deliv-ering timely post-discharge interventions within thisperiod will reduce or even eliminate this risk. The natureTable 4: Statistically significant risk factors and the associated adjusted hazard ratios for an adverse event (defined as non-independent living, hospital readmission, or death at follow up) using the multiple Cox proportional-hazards model.Characteristic Hazard ratio (95% CI) P ValueHigher mini-mental state examination score 0.87 (0.80, 0.95) 0.0011Independent living before hospitalization 0.34 (0.15, 0.74) 0.0063Longer length of stay in hospital 1.06 (1.03, 1.09) <0.0001Table 3: Risk factors and the associated unadjusted hazard ratios for an adverse event (defined as non-independent living, hospital readmission, or death at follow up) using the Cox proportional-hazards model. Characteristic Hazard ratio (95% CI) P ValueAge (years) 1.05 (1.00, 1.11) 0.05Male sex 0.97 (0.54, 1.74) 0.92Married 0.89 (0.48, 1.63) 0.69Independent living before hospitalization 0.56 (0.30, 1.05) 0.07Number of medications at admission 0.99 (0.89, 1.09) 0.80Higher cumulative illness rating scale score 1.03 (0.96, 1.10) 0.46Higher geriatric prognostic index score 1.01 (0.89, 1.16) 0.86Higher clinical frailty scale score at baseline 1.25 (0.88, 1.79) 0.22Higher clinical frailty scale score at 3 months 1.25 (0.59, 2.64) 0.56More independent ADL 0.92 (0.52, 1.64) 0.78Higher mini-mental state examination score 0.91 (0.85, 0.98) 0.01Higher geriatric depression scale score 1.08 (0.98, 1.20) 0.14Time to complete timed up and go test (seconds)* 1.28 (1.03, 1.59) 0.03Higher balance test score while standing with 2 feet 0.86 (0.67, 1.10) 0.23* The hazard ratio associated with the timed up and go test is for a 20 second increase.CI = confidence interval.Page 7 of 9(page number not for citation purposes)CI = confidence interval.BMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/10of such hospital-based outreach interventions has notbeen well defined, although there is some evidence thathome programs might improve outcomes [34,35]. Fur-ther studies are warranted.Our findings should be interpreted within the context oftheir limitations. The results might not be generalizable toall older adults admitted to hospital, such as those fromnursing homes or who speak minimal/no English. ACEPatients who consented in our study were likely more fitand healthy as compared with the majority of those whodid not/could not consent, and therefore might not berepresentative of the main ACE population. While sub-group analyses to identify the best predictors in males andfemales might be interesting, unfortunately our study wasnot adequately powered to do so. Our subjects were gen-erally cognitively intact, and we recognize that functionalattributes of cognitively impaired individuals might bedifferent and need to be studied separately. Specifically itwould be helpful to identify delirium cases and theirimpact in a future study. Hindsight would suggest exclu-sion of SAM patients from the analysis in an attempt toreduce heterogeneity, although this would imply screen-ing for more ACE patients before recruiting the requirednumber, thereby raising question on the representation ofthe cohort. There might be seasonal effect on the type ofACE patients and outcomes since the study was conductedfrom July to January, and findings might differ if the studywas done during the winter period. Although we havemade efforts within the available resources to extend thefollow up duration in this study to 6 months post dis-charge, ideally the frequency and length of follow upshould be greater. Our follow up duration nonethelessexceeds current published knowledge on ACE patientsand contributes to our understanding of this population.We acknowledge the attrition of the initial patient cohortcould make the interpretation of the secondary objectivefindings (predicting functional decline) challenging. Welost 19.7% of our subjects to follow up, which matchedwith our original 20% over-sampling to prevent type-2error. The dropouts should not affect the analysis on LOSas all independent variables included were measured atbaseline. Furthermore, dropouts still contributed to timeat risk for an adverse event. Finally, we did not conductthe regression analyses on another independent group ofpatients to assess the validity of the analyses.ConclusionFor many older adults discharged from ACE programs,independent functioning at home and in the communityis of major concern. The challenges faced by these individ-uals do not end once their acute medical problem isaddressed. In fact, many experience significant functionalalization), repeat hospitalization, or death. Our studyfindings identify TUG as potentially useful for identifyingacutely ill elderly patients who are at risk of adverse out-comes after hospitalization in a selected sample of ACEpatients who are able to return to independent living afterdischarge. The goal of this work is to develop founda-tional data for interventions that may influence func-tional impairments and alter future outcomes, althoughthe exact nature of such interventions is yet to be deter-mined and requires further studies.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsBoth authors contributed to the development of the con-ceptualization and design of the study. RYW was primarilyresponsible supervising the data analyses and interpretingand preparing the manuscript. WCM supervised the datacollection and provided assistance with data analyses andediting the final manuscript. All authors read andapproved the final draft of the manuscript.AcknowledgementsThe authors would like to recognize the subjects who participated in the study and the staff from ACE who facilitated subject recruitment and Ms Lisa Kuramoto for her assistance with the analyses. Salary support for WCM is provided by the Canadian Institutes of Health Research. This work was supported by a Vancouver Coastal Health Research Institute "In it for life" grant.References1. Palmer RM, Landefeld CS, Kresevic DM, Kowal J: A medical unit forthe acute care of the elderly.  J Am Geriatr Soc 1994, 42:545-52.2. Covinsky KE, Palmer RM, Kresevic DM, Kahana E, Counsell SR, Fort-insky RH, Landefeld CS: Improving functional outcomes in olderpatients: lessons from an acute care for elders unit.  Jt CommJ Qual Improv 1998, 24(2):63-76.3. Landefeld CS, Palmer RM, Kresevic DM, Fortinsky RH, Kowal J: Arandomized trial of care in the hospital medical unit espe-cially designed to improve the functional outcomes ofacutely ill older patients.  N Engl J Med 1995, 332:1338-4.4. Covinsky KE, King JT Jr, Quinn LM, Siddique R, Palmer R, KresevicDM, Fortinsky RH, Kowal J, Landefeld CS: Do acute care for eldersunits increase hospital costs? A cost analysis using the hospi-tal perspective.  J Am Geriatr Soc 1997, 45(6):729-34.5. Ellis G, Langhorne P: Comprehensive geriatric assessment forolder hospital patients.  Br Med Bulletin 2005, 71:45-59.6. Jayadevappa R, Bloom BS, Raziano DB, Lavizzo-Mourey L: Dissemi-nation and characteristics of acute care for elders (ACE)units in the United States.  Intern J Technology Assessment in HealthCare 2003, 19(1):220-7.7. Rockwood K: Commentary: What does an ACE unit trump.Can J Geriatr 2006, 9(3):102.8. Counsell SR, Holder CM, Liebenauer LL, Palmer RM, Fortinsky RH,Kresevic DM, Quinn LM, Allen KR, Covinsky KE, Landefeld CS:Effects of a multicomponent intervention on functional out-comes and process of care in hospitalized older patients: arandomized controlled trial of acute care for elders (ACE) ina community hospital.  J Am Geriatr Soc 2000, 48(12):1572-81.9. Asplund K, Gustafson Y, Jacobsson C, Bucht G, Wahlin A, Peterson J,Blom JO, Angquist KA: Geriatric-based versus general wardsPage 8 of 9(page number not for citation purposes)decline during and following hospitalization, which inturn leads to loss of independence (including institution-for older acute medical patients: a randomized comparisonof outcomes and use of resources.  J Am Geriatr Soc 2000,48(11):1381-8.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Geriatrics 2008, 8:10 http://www.biomedcentral.com/1471-2318/8/1010. Wong RY, Chittock DR, McLean N, Wilbur K: Discharge out-comes of older medical in-patients in a specialized acutecare for elders unit compared with non-specialized units.Can J Geriatr 2006, 9(3):96-101.11. Flamer HE, Christophidis N, Margetts C, Ugoni A, McLean AJ:Extended hospital stays with increasing age: the impact of anacute geriatric unit.  Med J Aust 164(1):10-3. 1996 Jan 112. Rochon PA, Katz JN, Morrow LA, McGlinchey-Berroth R, AhlquistMM, Sarkarati M, Minaker KL: Comorbid illness is associatedwith survival and length of hospital stay in patients withchronic disability. A prospective comparison of threecomorbidity indices.  Med Care 1996, 34(11):1093-101.13. Inouye SK, Bogardus ST Jr, Vitagliano G, Desai MM, Williams CS,Grady JN, Scinto JD: Burden of illness score for elderly persons:risk adjustment incorporating the cumulative impact of dis-eases, physiologic abnormalities, and functional impair-ments.  Med Care 2003, 41(1):70-83. Erratum in: Med Care. 2003;41(3): 446.14. Di Iorio A, Longo A, Mitidieri Costanza A, Palmerio T, Benvenuti E,Giardini S, Bavazzano A, Guizzardi G, Senin U, Bandinelli S, FerrucciL, Abate G: Factors related to the length of in-hospital stay ofgeriatric patients.  Aging (Milano) 1999, 11(3):150-4.15. Covinsky KE, Justice AC, Rosenthal GE, Palmer RM, Landefeld CS:Measuring prognosis and case mix in hospitalized elders. Theimportance of functional status.  J Gen Intern Med 1997,12(4):203-8.16. Campbell SE, Seymour DG, Primrose WR: ACMEPLUS Project. Asystematic literature review of factors affecting outcome inolder medical patients admitted to hospital.  Age Ageing 2004,33(2):110-5.17. Wong RYM, Shaw M, Acton C, Wilbur K, McMillan M, Breurkens E,Sowden C, Trautman SM, Chan N: An interdisciplinary approachto optimize health services in a specialized acute care forelders (ACE) unit.  J Can Geriatr Soc 2003, 6(3):177-86.18. How to Use G*Power   [http://www.psycho.uni-duesseldorf.de/aap/projects/gpower]19. Cohen J: Statistical power analysis for the behavioural sci-ences (2nd ed.).  Hillsdale, NY: Lawrence Erlbaum Associates; 1988. 20. Linn BS, Linn MW, Gurel L: Cumulative illness rating scale.  J AmGeriatr Soc 1968, 16:622-6.21. Walter LC, Brand RJ, Counsell SR, Palmer RM, Landefeld CS, Fortin-sky RH, Covinsky KE: Development and validation of a prog-nostic index for 1-year mortality in older adults afterhospitalization.  JAMA 2001, 285(23):2987-94.22. Rockwood K, Song X, MacKnight C, Bergman H, Hogan DB, McDow-ell I, Mitnitski A: A global clinical measure of fitness and frailtyin elderly people.  CMAJ 2005, 173(5):489-95.23. Connelly DM, Stevenson TJ, Vandervoort AA: Between-andwithin-rater reliability of walking test in a frail elderly popu-lation.  Physiotherapy Canada 1996, 48(1):47-51.24. Folstein MF, Folstein SE, McHugh PR: Mini-Mental State: a practi-cal method for grading the cognitive state of patients for theclinician.  J Psych Res 1975, 12:189-98.25. Yesavage JA: Development and validation of a GeriatricDepression Scale.  J Psych Res 1983, 7(1):37-49.26. Guralnick JM, Ferrucci L: Lower extremity function in personsover the age of 70 years as a predictor of subsequent disabil-ity.  N Engl J Med 1995, 322(9):556-61.27. Alarcón T, Bárcena A, González-Montalvo JI, Penãlosa C, Salgado A:Factors predictive of outcome on admission to an acute ger-iatric ward.  Age Ageing 1999, 28(5):429-32.28. Espaulella J, Arnau A, Cubí D, Amblàs J, Yánez A: Time-dependentprognostic factors of 6-month mortality in frail elderlypatients admitted to post-acute care.  Age Ageing 2007,36(4):407-13.29. Rozzini R, Sabatini T, Cassinadri A, Boffelli S, Ferri M, Barbisoni P,Frisoni GB, Trabucchi M: Relationship between functional lossbefore hospital admission and mortality in elderly personswith medical illness.  J Gerontol A Biol Sci Med Sci 2005,60(9):1180-3.30. Suzuki T, Yoshida H, Kim H: Walking speed as a good predictorfor maintenance of I-ADL among the rural community eld-erly in Japan: a 5 year follow-up study from TMIG-LISA.  Ger-iatrics and Gerontology Int 2003, 3:1-6.31. Podsiadlo D, Richardson S: The timed up and go test: a test ofbasic functional mobility for frail elderly persons.  J Am GeriatrSoc 1991, 39(2):142-8.32. Shumway-Cook A, Brauer S, Woollacott M: Predicting the proba-bility for falls in community-dwelling older adults using thetimed up and go test.  Phys Ther 2000, 80(9):896-903.33. Stuck AE, Siu AL, Wleland D, Adams J, Rubenstein LZ: Comprehen-sive geriatric assessment: a meta-analysis of controlled tri-als.  Lancet 1993, 342:1032-6.34. Caplan GA, Williams AJ, Daly B, Abraham K: A randomized, con-trolled trial of comprehensive geriatric assessment andmultidisciplinary intervention after discharge of elderly fromthe emergency department – the DEED II study.  J Am GeriatrSoc 2004, 52(9):1417-23.35. Cunliffe AL, Gladman JR, Husbands SL, Miller P, Dewey ME, HarwoodRH: Sooner and healthier: a randomized controlled trial andinterview study of an early discharge rehabilitation servicefor older people.  Age Ageing 2004, 33(3):246-52.Pre-publication historyThe pre-publication history for this paper can be accessedhere:http://www.biomedcentral.com/1471-2318/8/10/prepubyours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 9 of 9(page number not for citation purposes)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items