UBC Faculty Research and Publications

Mobility predicts change in older adults’ health-related quality of life: evidence from a Vancouver falls… Davis, Jennifer C; Bryan, Stirling; Best, John R; Li, Linda C; Hsu, Chun L; Gomez, Caitlin; Vertes, Kelly A; Liu-Ambrose, Teresa Jul 15, 2015

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

Item Metadata


52383-12955_2015_Article_299.pdf [ 1MB ]
JSON: 52383-1.0223642.json
JSON-LD: 52383-1.0223642-ld.json
RDF/XML (Pretty): 52383-1.0223642-rdf.xml
RDF/JSON: 52383-1.0223642-rdf.json
Turtle: 52383-1.0223642-turtle.txt
N-Triples: 52383-1.0223642-rdf-ntriples.txt
Original Record: 52383-1.0223642-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessMobility predicts change in older adults’health-related quality of life: evidence froma Vancouver falls prevention prospectivecohort studyJennifer C. Davis1, Stirling Bryan1, John R. Best4,2,5,6, Linda C. Li2,3, Chun Liang Hsu4,2,5,6, Caitlin Gomez4,Kelly A. Vertes4 and Teresa Liu-Ambrose4,2,5,6*AbstractBackground: Older adults with mobility impairments are prone to reduced health related quality of life (HRQoL) ishighly associated with mobility impairments. The consequences of falls have detrimental impact on mobility.Hence, ascertaining factors explaining variation among individuals’ quality of life is critical for promoting healthyageing, particularly among older fallers. Hence, the primary objective of our study was to identify key factors thatexplain variation in HRQoL among community dwelling older adults at risk of falls.Methods: We conducted a longitudinal analysis of a 12-month prospective cohort study at the Vancouver FallsPrevention Clinic (n = 148 to 286 depending on the analysis). We constructed linear mixed models whereassessment month (0, 6, 12) was entered as a within-subjects repeated measure, the intercept was specified as arandom effect, and predictors and covariates were entered as between-subjects fixed effects. We also included thepredictors by sex and predictor by sex by time interaction terms in order to investigate sex differences in the relationsbetween the predictor variable and the outcome variable, the EQ-5D.Results: Our primary analysis demonstrated a significant mobility (assessed using the Short Performance PhysicalBattery and the Timed Up and Go) by time interaction (p < 0.05) and mobility by time by sex interaction(p < 0.05). The sensitivity analyses demonstrated some heterogeneity of these findings using an imputed and acomplete case analysis.Conclusions: Mobility may be an important predictor of changes in HRQoL over time. As such, mobility is acritical factor to target for future intervention strategies aimed at maintaining or improving HRQoL in late life.Keywords: Mobility, Quality of life, Falls, Older adultsBackgroundPoor quality of life is universally acknowledged as anadverse health outcome [1]. A more recent shift isrecognizing that poor health related quality of life(HRQoL) may be a critical marker of other adversehealth outcomes [1]. It may be that poor HRQoL isan indicator of underlying conditions including pain,disability, depression, polipathology and frailty [1–7].Several older adult populations (i.e., heart failure, is-chaemic heart disease, type 2 diabetes, metatstaticprostate cancer, chronic kidney disease, lung cancerand those awaiting movement to residential care)have demonstrated the prognostic importance ofHRQoL of life as independent predictors of death andclinical complications [4, 5, 8–10]. Hence, HRQoL isan important outcome measure in the context ofhealthy aging.Impaired mobility is associated with lower HRQoL[11]; however, little research has investigated factors* Correspondence: teresa.ambrose@ubc.ca4Aging, Mobility, and Cognitive Neuroscience Lab, University of BritishColumbia, 2211 Wesbrook Mall, Vancouver, BC V6T 2B5, Canada2Department of Physical Therapy, University of British Columbia, 2177Wesbrook Mall, Vancouver, BC V6T 2B5, CanadaFull list of author information is available at the end of the article© 2015 Davis et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License(http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium,provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Davis et al. Health and Quality of Life Outcomes  (2015) 13:101 DOI 10.1186/s12955-015-0299-0that explain variation in HRQoL over time in thispopulation of fallers at risk for poor HRQoL. Fallsare a common geriatric syndrome and are the thirdleading cause of chonic disability worldwide [12]with approximately 30 % of community-dwellingadults aged 65 and older experiencing one or morefalls annually [13]. In particular, of the 30 % ofcommunity-dwelling seniors who fall, half fall recur-rently and are at significant risk for hospitalization,institutionalization, and even death [14–16]. Theconsequences of falls have a large and potentiallydetrimental impact on mobility [17]. Importantly,HRQoL is highly associated with mobility impair-ments in older adults [18–20]. Specifically, functionalabilities such as walking are associated with changesin both physical and mental HRQoL [21]. Given thatmobility is a key predictor of HRQoL it is critical toassess factors that explain changes in HRQoL overtime in fallers.To date, much of the insight into factors that ex-plain variation in HRQoL is based on cross-sectionaldata. As such, the literature is relatively devoid ofunderstanding the determinants of change in HRQoLamong older adults among a population at high riskfor HRQoL decline—fallers. Hence, our primary ob-jective was to determine what factors were signifi-cant predictors of change in HRQoL, as measured bythe EQ-5D-3 L, over time (i.e., from baseline to 6 to12 months) among older men and women presentingto the Vancouver Falls Prevention Clinic. Under-standing factors that explain variation in HRQoLover time will help guide future intervention strat-egies that are aiming to improve HRQoL amongolder adults at high risk of falls. Women consistentlyreport lower HRQoL compared with men [22].Hence, our secondary objective was to examine whetheror not there was a significant sex by time interaction oncekey predictors in HRQoL were identified.MethodsStudy designWe conducted a longitudinal analysis of a 12-monthprospective cohort study at the Vancouver Falls Preven-tion Clinic (www.fallclinic.com) from June 7, 2010through October 24, 2013. Participants received a com-prehensive assessment at the Vancouver Falls PreventionClinic at baseline and 12-months. No intervention be-yond the comprehensive assessment at the VancouverFalls Prevention Clinic was received by participants.However, some participants may receive additional fol-lowup by the geriatrician or other health care profes-sionals based on recommendations and referrals fromthe initial assessment.ParticipantsThe sample consisted of community dwelling womenand men who lived in the lower mainland region ofBritish Columbia were eligible for study entry if they: were adults ≥ 70 years of age referred by a medicalprofessional to the Falls Prevention Clinic as a resultof seeking medical attention for a non-syncopal fallin the previous 12 months; understood, spoke, and read English proficiently; had a Physiological Profile Assessment (PPA)[23] score of at least 1.0 SD above age-normativevalue or Timed Up and Go Test (TUG) [24]performance of greater than 15 s or oneadditional non-syncopal fall in the previous12 months (a fall was defined as “Unintentionallycoming to the ground or some lower level and otherthan as a consequence of sustaining a violent blow,loss of consciousness, sudden onset of paralysis as instroke or an epileptic seizure” [25]; were expected to live greater than 12 months (basedon the geriatricians’ expert opinion); were able to walk 3 m with or without an assistivedevice; and were able to provide written informed consent.We excluded those with a neurodegenerative dis-ease (e.g., Parkinson’s disease) or dementia, patientswho recently had a stroke, those with clinically sig-nificant peripheral neuropathy or severe musculoskel-etal or joint disease, and anyone with a historyindicative of carotid sinus sensitivity (i.e., syncopalfalls). We highlight that exclusions for this studywere based on clinical grounds. The Falls PreventionClinic is targeting treatment of older adults at risk ofimpaired mobility and functional decline specifically.Thus individuals with neurodegenerative disease ordementia are referred to alternate clinics.Ethical approval was obtained from the VancouverCoastal Health Research Institute and the Universityof British Columbia’s Clinical Research Ethics Board(H09-02370). All participants provided written in-formed consent.Vancouver falls prevention clinic measuresA comprehensive set of measurements relating to mobil-ity and cognitive function that were collected at baselineare described below.Comorbidity, activities of daily living and depressionFunctional comorbidity index (FCI) was calculated to es-timate the degree of comorbidity associated with phys-ical functioning [26]. This scale’s score is the totalnumber of comorbidities. We used the 15-item GeriatricDavis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 2 of 10Depression Scale (GDS) [27, 28] to indicate the presenceof depression; a score of ≥ 5 indicates depression [29].Balance and mobilityMobility and balance were assessed using the ShortPhysical Performance Battery (SPPB) [30] and theTimed-Up-and-Go Test (TUG) [31]. For the Short Phys-ical Performance Battery, participants were assessed onperformances of standing balance, walking, and sit-to-stand performance. Each component is rated out of fourpoints, for a maximum of 12 points; a score < 9/12 pre-dicts subsequent disability [32]. For the TUG, partici-pants rose from a standard chair, walked a distance ofthree meters, turned, walked back to the chair and satdown [31]. We recorded the time (in seconds) tocomplete the TUG, based on the average of two separatetrials. A TUG performance time of ≥ 13.5 s correctlyclassified persons as fallers in 90 % of cases [31].Physiological falls riskPhysiological falls risk was assessed using the short formof the Physiological Profile Assessment (PPA). The PPAis a valid and reliable [33] measure of falls risk. Based ona participant’s performance in five physiological domain-s—postural sway, reaction time, strength, propriocep-tion, and vision—the PPA computes a falls risk score(standardized score) that has a 75 % predictive accuracyfor falls in older people [34, 35]. A PPA Z-score of ≥ 0.60indicates high physiological falls risk [36].Global cognitive functionWe assessed global cognition using the Mini MentalState Examination (MMSE) and the Montreal CognitiveAssessment (MoCA). The MMSE is a widely used andwell-known questionnaire used to screen for cognitiveimpairment (i.e., MMSE <24) [37]. It is scored on a 30-point scale with a median score of 28 for healthy com-munity dwelling octogenarians with more than 12 yearsof education [37]. The MMSE may underestimate cogni-tive impairment for frontal system disorders because it hasno items specifically addressing executive function [37].The Montreal Cognitive Assessment (MoCA), a briefscreening tool for MCI [38] with high sensitivity andspecificity, was used to categorise participants as with,or without, possible MCI. It is more sensitive than theMMSE in detecting mild cognitive impairment [38]. It isa 30-point test covering eight cognitive domains: 1) at-tention and concentration; 2) executive functions; 3)memory; 4) language; 5) visuo-constructional skills; 6)conceptual thinking; 7) calculations; and 8) orientation.Scores below 26 are considered to be indicative of pos-sible MCI. A bonus point is given to individuals withless than 12 years of education.Primary outcome measureThe primary outcome variable of interest is the EQ-5D-3L. Patients completed the EQ-5D-3L using paper ver-sions that were given to them upon presentation to theFalls Prevention Clinic. Telephone interviews were usedto complete the EQ-5D-3L at 6 and 12 months. Nocards were used to aid interpretation.EQ-5D-3LWe assessed HRQoL using the EQ-5D three level ver-sion (EQ-5D-3L). The EQ-5D-3Lis a preference basedutility instrument that captures 243 health states [39] toascertain an individual’s HRQoL according to five do-mains: mobility, self-care, usual activities, pain and, anx-iety or depression. Each domain has three possibleresponse options indicating no problems, some prob-lems or severe problems. The EQ-5D-3 L health stateutility values (HSUVs) at each time point are boundedfrom −0.54 to 1.00 where a score of less than zero is in-dicative of a health state worse than death. Individuals’preferences for the scoring of the EQ-5D-3L were esti-mated using the time trade off technique on a randomsample of adults taken from the population living in theYork (UK) region (N = 3000) [40]. Thus, the EQ-5D re-flects societal norms of individuals’ preferences for a dis-tinct set of health states. The EQ-5D is the most widelyused generic instrument that uses a utility-based scoringapproach, yielding a single summary score (i.e., health-state utility value) on a common scale to facilitate com-parison across different health conditions and patient pop-ulations [39]. The HSUV is anchored at zero—a healthstate equivalent to death, and 1.0—a state of “full health.”Health-state utility values less than zero are defined ashealth states worse than death. Health-state utility valuesare an essential outcome for economic evaluations.Handling of missing dataMissing data were handled in three ways. First, using therestricted maximum likelihood estimator, all individualswith baseline data for the variables in the model (i.e.,available case set) were included (ML analysis). Specific-ally, an available case set only includes data where re-sults are known, using as a denominator the totalnumber of people who had data recorded a particularvariable of interest, models were restricted to those indi-viduals with Escore data at baseline and 12-monthfollow-up (i.e., complete case analysis). Of note, acomplete case analysis deletes all participant IDs withincomplete data (in the variables involved) from theanalysis. Third, multiple imputation using the ICE(Imputation by Chained Equations) procedure in STATA10.0 was using to create five complete data sets (MI ana-lysis). We followed recommendations by Oostenbrink[41, 42] and Briggs [43, 44] for multiple imputation ofDavis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 3 of 10missing effectiveness data. We imputed missing EQ-5Dvalues at each time point (i.e., 6 and 12 months). Foreach missing value, we generated five possible valuesusing multiple linear regression. Covariates included age,FCI, TUG, PPA and baseline EQ-5D utility score, andthe weight and value of the missing variable in the pre-ceding period. The final imputed value was the meanvalue from the five data sets created.Statistical analysesWe report the available case set as our base case ana-lysis. Data were initially examined using visual analysisof histograms and computation of skew and kurtosis.The TUG and SPPB variables departed from normality(skew > |1|) and underwent log10 transformation. Theoutcome variable, EQ-5D-3 L HSUVs, also showed skewat each time point (ranging between −1.5 and −1.4);therefore, analyses were conducted on the transformedEQ-5D-3 L HSUVs.For the main analyses, linear mixed models were con-structed using the SPSS 22.0 MIXED procedure (IBMCorporation, 2013). Assessment month (0, 6, 12) was en-tered as a within-subjects repeated measure, the interceptwas specified as a random effect, and predictors (i.e., SPPBor TUG, depending on the model) and covariates (i.e., sex,age, and their interactions with time (i.e., sex*time, age*-time)) were entered as between-subjects fixed effects. Afirst-order auto-regressive covariance matrix provided su-perior model fit compared to an unstructured covariancematrix (based on the Bayesian Information Criterion) andallowed for model convergence across the models. De-nominator degrees of freedom were calculated from theSatterthwaite approximation [45].A separate linear mixed model was constructed foreach predictor variable examined. In addition to the spe-cific predictor and its interaction with time, models in-clude participant age and sex and their interactions withtime. We also included the predictor X sex and predictorX sex X time interaction terms in order to investigatesex differences in the relations between the predictorvariable and the outcome variable, EQ-5D-3 L HSUVs. Ifnot statistically significant, these terms were dropped.Additionally, in the examination of SPPB and TUG askey mobility related predictors, the use of armrest wasincluded as a covariate, along with its interaction withtime. The use of armrest did not interact with the mainvariables of interest the model, and therefore these inter-action terms were excluded. In the text, we report theunstandardized beta estimate (B), its 95 % confidenceinterval, and its significance value. Given a significantinteraction with sex, we stratified the data and ran themodels separately for males and females. To visualizesignificant interaction effects, we used model-based esti-mated marginal means at low (−1 SD), average (0 SD),and high (+1 SD) levels of the predictor [46]. When ahigher-order interaction was significant (e.g., 3-wayinteraction), we do not report significant lower-order in-teractions (2-way interactions) or main effects.ResultsTwo-hundred and forty three (for the SPPB) or 244 (forthe TUG) participants are included in the Maximum Like-lihood Models using the available case analysis. Our sensi-tivity analyses include the complete case analysis (n = 148)and the imputed case analysis (n = 286). Further, we alsoconducted all of the above analyses using the log10 trans-formed EQ-5D-3 L HSUVs.ParticipantsTable 1 reports descriptive statistics of the complete caseanalysis at baseline for our variables of interest for thiscohort. At baseline, this cohort of community-dwellingsenior women has a mean (SD) EQ-5D-3 L HSUV of0.78 (0.22), a mean SPPB of 7.3 (2.5) and a mean TUGof 19.7 (10.5). On average, participants had at least twoexisting co-morbidities and were 82 ± 7 years of age. Par-ticipants were classified as having high falls risk with amean PPA score of 1.6 ± 1.0. Further, the mean MMSEscore was 26 ± 3 and the mean MoCA score was 22 ± 4.Table 1 Baseline characteristics of the Vancouver falls preventioncohort (available case analysis)Variables at baseline Mean (SD) or number (%) or median (IQR)Age (years) (n = 315) 82.5 (6.5)Sex (Male/Female) (n = 308) 112 (36.4)/196 (63.4)Living status (n = 253)Lives alone 100 (39.5)Lives with others 122 (48.2)Assisted living 31 (12.3)Education (n = 299)< Grade 9 33 (11.0)Grades 9–13, no diploma 59 (19.7)High school with diploma 58 (19.4)Trades school 23 (7.7)Some university 36 (12.0)University 90 (30.1)FCI (n = 320) 2.5 (1.9)GDS (n = 315) 3.1 (2.6)EQ-5D (n = 245) 0.778 (0.217) or 0.8 (0.27)SPPB (n = 303) 7.3 (2.5)TUG (n = 296) 19.7 (10.5)PPA (n = 311) 1.7 (1.1)MMSE (n = 315) 26.4 (3.2)MoCA (n = 303) 22.1 (4.6)Davis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 4 of 10A cut-off of 26 or lower on the MoCA is used to classifyindividuals with mild cognitive impairment.Available case analysis (non-transformed EQ-5D)The base case analysis is presented in Table 2.Short performance physical batteryThe Maximum Likelihood Model for the available caseanalysis (n = 243) demonstrated that baseline SPPB was as-sociated with baseline EQ-5D HSUVs. Further, a significantSPPB by time interaction (p < 0.05) and SPPB by time bysex interaction (p < 0.05) were also observed (Fig. 1a and b).When the analyses were run separately for males and fe-males using the complete case set, we found that formales (n = 51), baseline SPPB was not associated withbaseline EQ-5D HSUVs (B = .02, p = .295) but therewas a trend for SPPB to predict change in EQ-5DHSUVs over time (B = .04, p = .081). Alternatively, forfemales (n = 97), baseline SPPB was associated withbaseline EQ-5D HSUVs (B = .03, p = .034) but did notpredict change in EQ-5D HSUVs over time (B =−.02,p = .127). These effects among men and women aregraphed in Fig. 2a and b.Timed up and goThe Maximum Likelihood Model for the available caseanalysis (n = 244) demonstrated that baseline TUG wasassociated with baseline EQ-5D HSUVs. Further, a sig-nificant TUG by time interaction (p < 0.05) and TUG bytime by sex interaction (p < 0.05) were also observed forthe non-transformed EQ-5D data. When the analyseswere run separately for males and females using thecomplete case set, we found that for males (n = 57),baseline TUG was not associated with baseline EQ-5DHSUVs (B = -0.10, p = .671) nor change in EQ-5DHSUVs over time (p = 0.139). For females (n = 91),baseline TUG was associated with baseline EQ-5DHSUVs (B = -0.33, p = .015) and there was a trend forTUG predicting change in EQ-5D HSUVs over time(B = 0.28, p = .073). These effects among men andwomen are graphed in Fig. 3a and b.Sensitivity analysisAll sensitivity analyses were conducted on the trans-formed and non-transformed EQ-5D data are presentedin Table 3.Complete case analysisShort performance physical batteryThe Mixed Linear Model for the complete case ana-lysis (n = 148) demonstrated that baseline SPPB wasassociated with baseline EQ-5D HSUVs for the log-transformed EQ-5D data only. Further, a significantSPPB by time interaction (p < 0.05) and SPPB by timeby sex interaction (p < 0.05) were not observed for thetransformed or non-transformed data. For the EQ-5Ddata (non-transformed), a significant SPPB by sex bytime interaction was observed (p < 0.05).Timed up and goThe Mixed Linear Model for the complete case analysis(n = 148) demonstrated that baseline TUG was associatedwith baseline EQ-5D HSUVs. Further, a significant TUGby time by sex interaction (p < 0.05) was also observed forthe transformed and non-transformed EQ-5D data.Imputed case analysisShort performance physical batteryThe Mixed Linear Model for the imputed case analysis(n = 286) demonstrated that baseline SPPB was associ-ated with baseline EQ-5D HSUVs for both the trans-formed and non-transformed data. No significant SPPBby time interaction (p > 0.05) and SPPB by time by sexinteraction (p > 0.05) were observed for the transformedand non-transformed EQ-5D data.Timed up and goThe Mixed Linear Model for the imputed case analysis(n = 286) demonstrated that the baseline TUG was asso-ciated with baseline EQ-5D HSUVs. No significant TUGby time interaction (p > 0.05) and TUG by time by sexinteraction (p > 0.05) were observed for the transformedand non-transformed EQ-5D data.DiscussionHRQoL is an essential component that contributed tohealthy ageing [47]. Given the demonstrated associationbetween mobility and HRQoL [18–20], it is critical tounderstand key measures that explain variations inTable 2 The maximum likelihood model for the available caseanalyses for the SPPB and TUGMaximum likelihoodPredictor Non-transformed Log-transformedSPPB, N = 243 B (p value) B (p value)SPPB .04 (<.001)** .12 (<.001)**SPPB*time −.03 (.045)* −.08 (.041)*SPPB*sex −.01 (.239) −.04 (.328)SPPB*sex*time .04 (.036)* .10 (.077)TUG, N = 244TUG −.41 (<.001)** −1.29 (<.001)**TUG*time .33 (.032)* .94 (.053)TUG*sex .19 (.329) .50 (.414)TUG*sex*time −.57 (.040)* −1.48 (.084)*p < 0.05**p < 0.01Davis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 5 of 10HRQoL among high risk groups such as older fallers. Inthis study, we found that two valid and reliable measuresof mobility—the SPPB and the TUG—predicted HRQoLover time and this relationship was dependent on sex ina population of older fallers. As such, mobility may be acritical measure to consider to maintain or improveHRQoL among older fallers and promote healthy ageing.This study extends our previous cross-sectional find-ings that the SPPB, a valid and reliable measure of bal-ance and mobility, explained a large and significantamount of variation in HRQoL at baseline [48]. We nowdemonstrate, for males, there was a significant SPPB bytime interaction indicating that the change in SPPB overtime explains significant variation in HRQoL. Con-versely, this was not observed for females. Specifically,the average trajectories for males of low, medium andhigh SPPB scores all demonstrated a trend of declineover the 12 month period with the high SPPB group ex-periences the slowest rate of decline. For females, thereappeared to be a regression to the mean effect (i.e., re-gardless of baseline function, over time, all females dem-onstrated a trend toward the average), with the highSPPB group declining over time and the low and averageSPPB groups improving over time. Examining theunderlying reasons for these differences in balanceand mobility and their change over time betweenmales and females is essential to appropriate targetintervention strategies. One hypothesis is that individ-uals with low and average SPPB scores may be morecompliant with the recommendations received at theFalls Prevention Clinic.The intricacies of sex specific relationships betweenmobility and HRQoL over time are largely understudied.One previous cross-sectional study demonstrated that alow SPPB score was significantly associated with thelowest quartile of EQ-5D index score for men and low-est and second lowest quartiles for women [49]. Gaitspeed was significantly associated with the EQ-5D indexfor participants of both sexes, however, standup timewas associated with the EQ-5D for men only [49]. Assuch, cross-sectional data have previously demonstratedsex effects in the relationship of balance and mobilitywith HRQoL [47]. Despite these associations, there re-mains a gap in the understanding of longitudinalchanges in mobility and the unique contribution toHRQoL among men and women.Fig. 1 a SPPB by time interaction among men and women over 12 months. b TUG by time interaction among men and women over 12-monthsDavis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 6 of 10This study provides a critical first step for future longi-tudinal studies and intervention studies to explore thetemporal relationships of mobility and HRQoL amongmen and women and to consider the targeting of futureintervention strategies aimed at improving or maintain-ing HRQoL differently among men and women. Specif-ically, we need to better understand why the observedtrajectories among men and women demonstrate differ-ent temporal trends and how this impacts preventionand treatment strategies delivered by clinicians. For ex-ample, compliance to recommendations (a behaviouralpattern) between men and women may be an explana-tory factor. If so, this would highlight that cliniciansneed to tailor treatment and management for men andwomen differently.We note the following limitations of our study. Al-though this was not an intervention study, it could bepossible that the management of balance and mobilitymay confound the outcomes. The presence of missingdata could influence the interpretation of the results. Assuch, we conducted sensitivity analyses with the multipleimputed case set and the complete case set. Further, theEQ-5D data were significantly skewed. Although aresample size was large enough that the analyses shouldbe robust to departure from normality, in our base caseand sensitivity analyses, we report the result of the trans-formed and non-transformed data.ConclusionsThis study confirms the critical role that mobility playsin HRQoL at baseline among older fallers. Further ithighlights key differences in this relationship betweenmen and women over time. Specifically, men demon-strate decline over time regardless of mobility status;Fig. 2 a Model-based estimated marginal means for low (−1 SD), average and high (+1 SD) SPPB scores for males. b Model-based estimatedmarginal means for low (−1 SD), average and high (+1 SD) SPPB scores for femalesDavis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 7 of 10Fig. 3 a Model-based estimated marginal means for low (−1 SD), average and high (+1 SD) TUG scores for males. b Model-based estimatedmarginal means for low (−1 SD), average and high (+1 SD) TUG scores for femalesTable 3 Mixed linear models for the multiply imputed and complete case sets for the SPPB and TUGMultiple imputation Complete caseN = 286 N = 148Predictor Non-transformed Log-transformed Non-transformed Log-transformedSPPB B (p value) B (p value) B (p value) B (p value)SPPB .03 (<.001)** .10 (<.001)** .02 (.052) .08 (.042)*SPPB*time −.01 (.076) −.03 (.087) −.01 (.388) −.04 (.312)SPPB*sex −.01 (.348) −.03 (.462) −.002 (.926) .001 (.979)SPPB*sex*time .01 (.304) .02 (.422) .04 (.036)* .12 (.067)TUG N = 290 N = 148TUG −.36 (<.001)** −1.13 (.001)** −.27 (.042)* −.89 (.039)*TUG*time .11 (.075) .31 (.111) .23 (.158) .64 (.209)TUG*sex .07 (.716) .11 (.852) .11 (.652) .20 (.792)TUG*sex*time −.16 (.169) −.41 (.274) −.70 (.015)* −1.94 (.032)**p < 0.05**p < 0.01Davis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 8 of 10whereas women in the highest tertile of mobility onlydemonstrate a declining trend in HRQoL over time. Onepotential explanation that needs investigation is thatwomen may be more compliant with recommendationsreceived at the Falls Prevention Clinic. The importantmessage at a clinical level is that men and women’streatment and prevention strategies need to tailoredtreatment and prevention strategies.Competing interestsThe authors declare that they have no competing interests.Authors’ contributionsTLA was principal investigator for the Vancouver Falls Prevention ClinicCohort study. TLA and JCD were responsible for study concept and design,acquisition of data, data analysis and interpretation, writing and reviewing ofthe manuscript. JCD and JB were responsible for data analysis. JCD, TLA, JRB,SB, CLH, LL, CG, and KAV drafted and revised the manuscript. JCD, TLA andSB acquired and interpreted the data. All authors read and approved thefinal manuscript.AcknowledgmentsWe thank the Vancouver Falls Prevention Cohort study participants. TheCanadian Institute for Health Research Emerging Team Grant (CIHR,MOB-93373 to Karim Khan, TLA, LL) provided funding for this study. TLAis a Canada Research Chair in Physical Activity, Mobility, and CognitiveNeuroscience, a Michael Smith Foundation for Health Research (MSFHR)Scholar, a Canadian Institutes of Health Research (CIHR) New Investigator,and a Heart and Stroke Foundation of Canada’s Henry JM Barnett’sScholarship recipient. JCD and JB are funded by a CIHR and MSFHRPostdoctoral Fellowship. LL is a MSFHR Scholar and a Canada ResearchChair. CLS is a CIHR Doctoral Trainee. These funding agencies did notplay a role in study design. We obtained approval for the VancouverFalls Prevention Clinic Cohort study from UBC Clinical Ethics ReviewBoard.Author details1Centre for Clinical Epidemiology and Evaluation, University of BritishColumbia and Vancouver Coastal Health Research Institute (VCHRI), 828 West10th Avenue, Vancouver, BC V6T 2B5, Canada. 2Department of PhysicalTherapy, University of British Columbia, 2177 Wesbrook Mall, Vancouver, BCV6T 2B5, Canada. 3Arthritis Research Centre of Canada, 5591 No. 3 Road,Richmond, BC V6X 2C7, Canada. 4Aging, Mobility, and CognitiveNeuroscience Lab, University of British Columbia, 2211 Wesbrook Mall,Vancouver, BC V6T 2B5, Canada. 5Djavad Mowafaghian Centre for BrainHealth, University of British Columbia & VCHRI, 2215 Wesbrook Mall,Vancouver, BC V6T 1Z3, Canada. 6Center for Hip Health and Mobility,University of British Columbia & VCHRI, 828 West 10th Avenue, Vancouver, BCV5Z 1E2, Canada.Received: 26 November 2014 Accepted: 6 July 2015References1. Bilotta C, Bowling A, Nicolini P, Case A, Pina G, Rossi SV, et al. Older People’sQuality of Life (OPQOL) scores and adverse health outcomes at a one-yearfollow-up. A prospective cohort study on older outpatients living in thecommunity in Italy. Health Qual Life Outcomes. 2011;9:72.2. Masel MC, Graham JE, Reistetter TA, Markides KS, Ottenbacher KJ. Frailty andhealth related quality of life in older Mexican Americans. Health Qual LifeOutcomes. 2009;7:70.3. Masel MC, Ostir GV, Ottenbacher KJ. Frailty, mortality, and health-relatedquality of life in older Mexican Americans. J Am Geriatr Soc.2010;58:2149–53.4. Puts MT, Shekary N, Widdershoven G, Heldens J, Lips P, Deeg DJ. What doesquality of life mean to older frail and non-frail community-dwelling adultsin the Netherlands? Qual Life Res. 2007;16:263–77.5. Zaninotto P, Falaschetti E, Sacker A. Age trajectories of quality of life amongolder adults: results from the English Longitudinal Study of Ageing. QualLife Res. 2009;18:1301–9.6. Stubbs B, Binnekade T, Eggermont L, Sepehry AA, Patchay S, Schofield P.Pain and the risk for falls in community-dwelling older adults: systematicreview and meta-analysis. Arch Phys Med Rehabil. 2014;95:175–87. e179.7. Stubbs B, Schofield P, Patchay S. Mobility limitations and fall-related factorscontribute to the reduced health-related quality of life in older adults withchronic musculoskeletal pain. Pain Pract. 2014. doi:10.1111/papr.12264.8. Movsas B, Moughan J, Sarna L, Langer C, Werner-Wasik M, Nicolaou N, et al.Quality of life supersedes the classic prognosticators for long-term survivalin locally advanced non-small-cell lung cancer: an analysis of RTOG 9801.J Clin Oncol. 2009;27:5816–22.9. Pedersen SS, Martens EJ, Denollet J, Appels A. Poor health-related quality oflife is a predictor of early, but not late, cardiac events after percutaneouscoronary intervention. Psychosomatics. 2007;48:331–7.10. Tsai YC, Hung CC, Hwang SJ, Wang SL, Hsiao SM, Lin MY, et al. Quality oflife predicts risks of end-stage renal disease and mortality in patients withchronic kidney disease. Nephrol Dial Transplant. 2010;25:1621–6.11. Chang NT, Chi LY, Yang NP, Chou P. The impact of falls and fear of fallingon health-related quality of life in Taiwanese elderly. J Community HealthNurs. 2010;27:84–95.12. Murray C, Lopez A. Global and regional descriptive epidemiology ofdisability: incidence, prevalence, health expectancies, and years lived withdisability. In: Murray C, Lopez A, editors. The global burden of disease.Boston: The Harvard School of Public Health; 1996. p. 201–46.13. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderlypersons living in the community. N Engl J Med. 1988;319:1701–7.14. Pluijm S, Smit J, Tromp E, Stel V, Deeg D, Bouter L, et al. A risk profile foridentifying community-dwelling elderly with a high risk of recurrent falling: resultsof a 3-year prospective study. Osteoporosis International. 2006;17:417–25.15. Formiga F, Navarro M, Duaso E, Chivite D, Ruiz D, Perez-Castejon JM,et al. Factors associated with hip fracture-related falls among patientswith a history of recurrent falling. Bone. 2008;43:941–4.16. Rockwood K, Stolee P, McDowell I. Factors associated withinstitutionalization of older people in Canada: testing a multifactorialdefinition of frailty. Journal of the American Geriatrics Society.1996;44:578–82.17. Dai B, Ware WB, Giuliani CA. A structural equation model relating physicalfunction, pain, impaired mobility (IM), and falls in older adults. ArchGerontol Geriatr. 2012;55:645–52.18. Fagerstrom C, Borglin G. Mobility, functional ability and health-relatedquality of life among people of 60 years or older. Aging Clin Exp Res.2010;22:387–94.19. Davis JC, Marra CA, Najafzadeh M, Liu-Ambrose T. The independentcontribution of executive functions to health related quality of life in olderwomen. BMC Geriatr. 2010;10:16.20. Davis JC, Marra CA, Liu-Ambrose TY. Falls-related self-efficacy is independentlyassociated with quality-adjusted life years in older women. Age Ageing.2011;40:340–6.21. Heydarnejad S, Dehkordi AH. The effect of an exercise program on thehealth-quality of life in older adults. A randomized controlled trial. Dan MedBull. 2010;57:A4113.22. Redondo-Sendino A, Guallar-Castillon P, Banegas JR, Rodriguez-Artalejo F.Gender differences in the utilization of health-care services among theolder adult population of Spain. BMC Public Health. 2006;6:155.23. Lord S, Sherrington C, Menz H. A physiological profile approach for fallsprevention. In: Falls in older people Risk factors and strategies forprevention. Cambridge: Cambridge University Press; 2001. p. 221–38.24. Whitney JC, Lord SR, Close JC. Streamlining assessment and intervention ina falls clinic using the timed up and go test and physiological profileassessments. Age Ageing. 2005;34:567–71.25. Gibson MJ, Andres RO, Isaacs B, Radebaugh T, Worm-Petersen J. Theprevention of falls in later life. A report of the Kellogg International WorkGroup on the prevention of falls by the elderly. Danish Medical Bulletin.1987;34:1–24.26. Groll DL, To T, Bombardier C, Wright JG. The development of acomorbidity index with physical function as the outcome. J Clin Epidemiol.2005;58:595–602.27. Yesavage JA. Geriatric depression scale. Psychopharmacol Bull.1988;24:709–11.Davis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 9 of 1028. Yesavage JA, Brink TL, Rose TL, Lum O, Huang V, Adey M, et al.Development and validation of a geriatric depression screening scale: apreliminary report. J Psychiatr Res. 1982;17:37–49.29. van Marwijk HW, Wallace P, de Bock GH, Hermans J, Kaptein AA, Mulder JD.Evaluation of the feasibility, reliability and diagnostic value of shortenedversions of the geriatric depression scale. Br J Gen Pract. 1995;45:195–9.30. Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremityfunction in persons over the age of 70 years as a predictor of subsequentdisability. N Engl J Med. 1995;332:556–61.31. Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability for fallsin community-dwelling older adults using the Timed Up & Go Test. PhysTher. 2000;80:896–903.32. Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremityfunction in persons over the age of 70 years as a predictor of subsequentdisability. N Engl J Med. 1995;332:556–62. doi:10.1056/NEJM199503023320902.33. Lord SR, Menz HB, Tiedemann A. A physiological profile approach to fallsrisk assessment and prevention. Phys Ther. 2003;83(3):237–52.34. Lord S, Clark R, Webster I. Physiological factors associated with falls in anelderly population. Journal of American Geriatrics Society. 1991;39:1194–200.35. Lord S, Ward J, Williams P, Anstey K. Physiological factors associated withfalls in older community-dwelling women. Journal of American GeriatricsSociety. 1994;42:1110–7.36. Delbaere K, Close JCT, Brodaty H, Sachdev P, Lord SR. Determinants ofdisparities between perceived and physiological risk o falling among elderlypeople: cohort study. British Medical Journal. 2010;341:c4165.37. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practicalmethod for grading the cognitive state of patients for the clinician. JPsychiatr Res. 1975;12:189–98.38. Nasreddine ZS, Phillips NA, Bedirian V, Charbonneau S, Whitehead V, Collin I,et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool formild cognitive impairment. J Am Geriatr Soc. 2005;53:695–9.39. Marra CA, Esdaile JM, Guh D, Kopec JA, Brazier JE, Koehler BE, et al. Acomparison of four indirect methods of assessing utility values inrheumatoid arthritis. Med Care. 2004;42:1125–31.40. Dolan P. Modeling valuations for EuroQol health states. Med Care.1997;35:1095–108.41. Oostenbrink JB, Al MJ. The analysis of incomplete cost data due to dropout.Health Econ. 2005;14:763–76.42. Oostenbrink JB, Al MJ, Rutten-van Molken MP. Methods to analyse cost dataof patients who withdraw in a clinical trial setting. Pharmacoeconomics.2003;21:1103–12.43. Briggs A, Clark T, Wolstenholme J, Clarke P. Missing… presumed at random:cost-analysis of incomplete data. Health Econ. 2003;12:377–92.44. Briggs AH, Gray AM. Handling uncertainty when performing economicevaluation of healthcare interventions. Health Technol Assess. 1999;3:1–134.45. Satterthwaite FE. An approximate distribution of estimates of variancecomponents. Biometrics Bulletin. 1946;2:110–4.46. Wolitzky-Taylor KB, Arch JJ, Rosenfield D, Craske MG. Moderators andnon-specific predictors of treatment outcome for anxiety disorders: acomparison of cognitive behavioral therapy to acceptance andcommitment therapy. Journal of Consulting and Clinical Psychology.2012;80(5):786.47. Fredriksen-Goldsen KI, Kim HJ, Shiu C, Goldsen J, Emlet CA. Successful agingamong LGBT older adults: physical and mental health-related quality of lifeby age group. Gerontologist. 2015;55(1):154–68.48. Davis JC, Bryan S, McLeod R, Rogers J, Khan K, Liu-Ambrose T.Exploration of the association between quality of life, assessed bythe EQ-5D and ICECAP-O, and falls risk, cognitive function and dailyfunction, in older adults with mobility impairments. BMC Geriatr.2012;12:65.49. Oh B, Cho B, Choi HC, Son KY, Park SM, Chun S, et al. The influence oflower-extremity function in elderly individuals’ quality of life (QOL): ananalysis of the correlation between SPPB and EQ-5D. Arch Gerontol Geriatr.2014;58:278–82.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitDavis et al. Health and Quality of Life Outcomes  (2015) 13:101 Page 10 of 10


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