Does Perturbation-Based BalanceTraining Prevent Falls? SystematicReview and Meta-Analysis ofPreliminary RandomizedControlled TrialsAvril Mansfield, Jennifer S. Wong, Jessica Bryce, Svetlana Knorr, Kara K. Patterson
Background. Older adults and individuals with neurological conditions are at anincreased risk for falls. Although physical exercise can prevent falls, certain types ofexercise may be more effective. Perturbation-based balance training is a novel inter-vention involving repeated postural perturbations aiming to improve control of rapidbalance reactions.
Purpose. The purpose of this study was to estimate the effect of perturbation-based balance training on falls in daily life.
Data Sources. MEDLINE (1946–July 2014), EMBASE (1974–July 2014), PEDro(all dates), CENTRAL (1991–July 2014), and Google Scholar (all dates) were the datasources used in this study.
Study Selection. Randomized controlled trials written in English were includedif they focused on perturbation-based balance training among older adults or indi-viduals with neurological conditions and collected falls data posttraining.
Data Extraction. Two investigators extracted data independently. Study authorswere contacted to obtain missing information. A PEDro score was obtained for eachstudy. Primary outcomes were proportion of participants who reported one or morefalls (ie, number of “fallers”) and the total number of falls. The risk ratio (proportionof fallers) and rate ratio (number of falls) were entered into the analysis.
Data Synthesis. Eight studies involving 404 participants were included. Partic-ipants who completed perturbation-based balance training were less likely to reporta fall (overall risk ratio�0.71; 95% confidence interval�0.52, 0.96; P�.02) andreported fewer falls than those in the control groups (overall rate ratio�0.54; 95%confidence interval�0.34, 0.85; P�.007).
Limitations. Study authors do not always identify that they have included per-turbation training in their intervention; therefore, it is possible that some appropriatestudies were not included. Study designs were heterogeneous, preventingsubanalyses.
Conclusions. Perturbation-based balance training appears to reduce fall riskamong older adults and individuals with Parkinson disease.
A. Mansfield, PhD, Toronto Reha-bilitation Institute, UniversityHealth Network, 550 UniversityAve, Room 11-117, Toronto,Ontario, Canada M5G 2A2; BrainSciences Program, SunnybrookResearch Institute, Toronto,Ontario, Canada; and Departmentof Physical Therapy and GraduateDepartment of Rehabilitation Sci-ences, University of Toronto,Toronto, Ontario, Canada.Address all correspondence to DrMansfield at: [email protected].
J.S. Wong, BSc, Toronto Rehabili-tation Institute, University HealthNetwork, and Graduate Depart-ment of Rehabilitation Sciences,University of Toronto.
J. Bryce, BSc, Toronto Rehabilita-tion Institute, University HealthNetwork.
S. Knorr, PhD, Toronto Rehabilita-tion Institute, University HealthNetwork.
K.K. Patterson, PhD, TorontoRehabilitation Institute, UniversityHealth Network; Department ofPhysical Therapy and GraduateDepartment of Rehabilitation Sci-ences, University of Toronto; andSchool of Physical Therapy, Fac-ulty of Health Sciences, WesternUniversity, London, Ontario,Canada.
[Mansfield A, Wong JS, Bryce J, et al.Does perturbation-based balancetraining prevent falls? Systematicreview and meta-analysis of prelim-inary randomized controlled trials.Phys Ther. 2015;95:700–709.]
Physiological impairments asso-ciated with neurological condi-tions (eg, stroke, Parkinson dis-
ease) and the “normal” aging processcan contribute to impaired balancecontrol and increased fall risk. Fallscan have direct negative conse-quences (ie, injuries) but also canresult in fear and anxiety, reducedindependent mobility, and increasedrisk of admission to long-term carefacilities. Previous work has gener-ally shown that physical exercisefocused on balance training can pre-vent falls1–5; however, there is lim-ited or no evidence for reduced fallsfollowing balance training in somecases (eg, for people with stroke6,7).It is possible that more specific exer-cise, focused on the mechanism ofoccurrence of falls, might be moreeffective for falls prevention.
Although many factors contribute toincreased risk for falls, a specific fallevent ultimately occurs when anindividual fails to recover from a lossof balance or postural perturbation.8
Postural perturbations can occur indaily life for a variety of reasons,including failure to control weightshifting during voluntary movementor experiencing a slip or trip whilewalking.9 Balance recovery reac-tions, such as swaying around theankles or hips, taking a step, orgrasping a handhold,10 are executedrapidly to prevent a fall following apostural perturbation. Individualswith impaired balance control andincreased fall risk often show diffi-culty controlling these balancerecovery reactions.11–13 Because allambulatory individuals are at risk forexperiencing a loss of balance duringdaily life, training to improve controlof balance recovery reactions may bean effective means of preventingfalls.14–16 Perturbation-based balancetraining (PBT) is a novel balancetraining intervention that incorpo-rates exposure to repeated posturalperturbations to evoke rapid balancereactions, enabling the individual to
improve control of these reactionswith practice. The alternative—train-ing voluntary movements (eg, voli-tionally executed stepping or reach-to-grasp movements)—will likely notlead to improved reactive balancecontrol due to the additional speedand stability requirements of balancereactions.15
Studies have shown that PBT canimprove speed and control of volun-tary movements17 and rapid balancereactions18–20 and can reduce occur-rence of “falls” (into a safety harness)following controlled postural pertur-bations in the laboratory.18,20,21 How-ever, the effect of PBT on risk of fallsduring daily life has not been conclu-sively demonstrated. Several studieswith small sample sizes have shownnonsignificant reductions in fall ratesin the trained group compared witha control group.22,23 Therefore, thepurpose of this study was to estimatethe effect of PBT on risk of falls indaily life among individuals atincreased risk for falls (ie, olderadults and individuals with neurolog-ical conditions) by conducting acomprehensive search for, and meta-analysis of, published and unpub-lished data from randomized con-trolled trials of PBT.
MethodDesign OverviewThis study involved systematicreview and meta-analysis, conductedaccording to Cochrane guidelines24
and reported according to thePRISMA statement and checklist.25
Data Sources and SearchesThe search strategy and inclusionand exclusion criteria for selectedstudies were determined a priori bythe investigators in collaborationwith an information specialist. Ter-minology around PBT is inconsistentin the literature. Some investigatorsrefer to “perturbation training,”whereas others use terms such as“agility training” or “dynamic bal-
ance training.” Additionally, thereare no relevant subject headingswithin literature databases to accu-rately encompass the content ofPBT. Thus, a multistage and iterativesearch strategy was used to identifystudies for inclusion in the review:
1. An initial search of MEDLINE(1946–current), EMBASE (1974–current), PEDro (all availabledates), and CENTRAL (1991–cur-rent) databases was conducted inJuly 2014. The database searcheswere conducted by the informa-tion specialist. An example of theMEDLINE search strategy is pro-vided in the Appendix; the searchterms were modified, as required,for other databases.
2. Relevant review articles and stud-ies that focused on PBT wereidentified based on either theabstract or full text of the article.
3. The reference lists of articlesidentified in step 2 were searchedto find additional studies thatmight have focused on PBT.
4. Steps 2 and 3 were repeated untilno new studies were identified.
5. Titles, abstracts, and full papers ofall identified studies were reviewedto determine if they met the inclu-sion criteria (see below). If a studymet all but the last criterion (ie, nofalls were reported in the paper),the authors were contacted todetermine if falls data existed. Ifsuch data were available, the studywas included in the analysis.
6. A cited reference search wascompleted using Google Scholar(all available dates) for all studiesselected in step 5 to identifyany studies that cited relevant arti-cles. This “snowball” samplingapproach has been used in othersystematic reviews (eg, Fegran
Does Perturbation-Based Balance Training Prevent Falls?
May 2015 Volume 95 Number 5 Physical Therapy f 701
et al,26 Groene et al,27 Sweeney etal28).
7. Steps 2 through 6 were repeateduntil no new studies wereidentified.
Bibliographic management was facil-itated using Endnote (version X5,Thomson Reuters, Philadelphia,Pennsylvania).
Study SelectionAll authors were involved in thesearch and review process. Eachabstract was reviewed by at least 2authors independently. Disagree-ments regarding study inclusionwere resolved by discussion. Theanalysis included studies publishedin English that met the following cri-teria: (1) focused on PBT (see defini-tion below), (2) included olderadults (�60 years old) or individualswith neurological conditions, (3)included a control group that did notdo any perturbation training over thecourse of the study (ie, crossoverdesigns were excluded), (4) ran-domly allocated participants togroups, and (5) collected data onfalls experienced in daily life. In thisstudy, perturbation-based balancetraining was defined as the inten-tional application of repeated pos-tural perturbations that cause a lossof balance over the course of a train-ing program with the goal of improv-ing whole-body reactive balancecontrol. Thus, we excluded studiesusing postural perturbations forrehabilitation of a single joint. Pertur-bations may be applied using equip-ment (eg, moving platform) or man-ually (eg, nudge from therapist).Although postural perturbations mayoccur during some challenging bal-ance activities (eg, standing on anunstable surface), such activities alsocan occur without postural perturba-tion with adequate feedforward bal-ance control; therefore, studies thatsolely included these activities wereexcluded. Additionally, to be consid-
ered “training,” there must be at least2 sessions; therefore, we excludedstudies examining within-sessionadaptations.
Data Extraction andQuality AssessmentThe following data were extractedfrom each selected paper: details ofthe population studied, details of thecontrol and PBT interventions (typeof training, frequency and durationof training sessions, duration of theentire training program), samplesize, falls monitoring duration, num-ber of participants in each groupwho reported one or more falls (ie,number of “fallers”), and total num-ber of falls reported by participantsin each group. Study authors werecontacted to obtain any missinginformation. Additionally, studieswere rated for methodological qual-ity according to the PEDro scale.29,30
The study received a score of 1 forPEDro items relating to key outcomemeasure if these items applied to col-lection and reporting of falls data.
Data extraction and PEDro scoringwere performed by 2 authors inde-pendently for each selected article.Disagreements between authorsregarding data extraction or PEDroscores were resolved by discussion.
Data Synthesis and AnalysisExtracted data were summarized in atable, and differences and similaritiesbetween studies were noted. Meta-analyses were conducted usingReview Manager (version 5.2, TheNordic Cochrane Centre, TheCochrane Collaboration). The 2 out-comes of interest were proportion offallers and fall rate. The genericinverse variance method was used,with the natural logarithm and stan-dard error of the risk ratio (propor-tion of fallers) and rate ratio (fallsrate) entered into the analysis. The I2
statistic was used to assess statisticalheterogeneity31; as there was signif-icant heterogeneity (I2�30%), ran-
Role of the Funding SourceDr Patterson was supported by aFocus on Stroke personnel awardfrom the Heart and Stroke Founda-tion and the Canadian StrokeNetwork.
ResultsWe identified 761 potential articlesusing the search strategy; Figure 1outlines the search process. Of thearticles identified, 7 met all inclusioncriteria.18,22,23,33–36 However, 2 ofthese studies involved several differ-ent types of exercise, and it was notclear that all participants assigned tothe PBT group actually received PBT.Thus, these 2 studies18,36 wereexcluded from the review. Threestudies met all but the last criterion(ie, reported falls experienced indaily life). However, falls data forthese studies were available else-where; for one of these studies, asubsequent article from the sameauthors reporting falls data was iden-tified,37 and for the other 2 studies,falls data were available from theauthors.19,38 Therefore, 8 studieswere included in the final analysis.Additional details for one study19
were obtained from a published pro-tocol.39 Table 1 summarizes the 8selected studies.
Description of StudiesThere were 404 participantsincluded in the 8 selected studies(202 completed PBT, and 202 werein the control groups). The selectedstudies included healthy19,34,37 andfrail22,38 older adults and individualswith Parkinson disease.23,33,35 Twostudies included frail older adults,which represented a heterogeneoussample of individuals with chronicconditions (eg, stroke, arthritis, cog-nitive impairment or dementia).Seven studies attempted to target
Does Perturbation-Based Balance Training Prevent Falls?
702 f Physical Therapy Volume 95 Number 5 May 2015
individuals at increased risk for falls;for example, by recruiting thosewith prior falls history, neurologicalconditions that increase fall risk, orfear of falling. One study includedlong-term care residents,22 whereasthe remaining studies includedcommunity-dwelling individuals.Falls monitoring periods ranged from2 weeks to 1 year. The experimentalintervention in 3 studies focusedsolely on PBT19,37,38; in all other stud-ies, PBT was included as part of abroader exercise program. An“active” control group was includedin 6 studies: for 3 studies, the controlintervention was specificallydesigned to have minimal effect onbalance control19,33,35; for 2 studies,the control intervention included
balance and mobility exercises34,38;and for 1 study, the control interven-tion included individualized exer-cises that may have benefited bal-ance control.22 Falls were typicallydefined as events where the personcame to rest unintentionally on theground, floor, or other lower level.40
All studies collected falls data via par-ticipant self-report; however, thestudy that involved individuals inlong-term care also included nursingreports of falls.22 Three studies useddiaries (or similar means) to collectfalls data prospectively19,33,38; 1study collected falls data retrospec-tively at the end of the monitoringperiod34; and 4 studies used a “con-tinuous retrospective” method, witha member of the research team con-
tacting participants daily,23 fort-nightly,37 or monthly22,35 to ask ifthey had fallen.
Intervention CharacteristicsIntervention characteristics aredescribed according to the FITT (ie,frequency, intensity, time, and typeof perturbation training) principle aswell as the overall intervention dura-tion. For the purpose of PBT, the“time” principle is equivalent to theamount of training per session (ie,number of perturbations). One studydid not report the prescribed inter-vention duration or the number orfrequency of sessions.34 Five studiesreported training session frequencyof 3 times per week,19,23,33,35,38 1study reported participants trained
94 duplicates removed
622 records excluded(not PBT)
131 full-text articles excluded74 not PBT10 healty young
participants31 not a randomized
controlled trial14 no falls data available 2 dose of perturbation
training unclear
761 records screened
139 full-text articlesassessed for eligibility
8 studies included inquantitative synthesis
(meta-analysis)
410 records identified throughdatabase searching
445 additional records identifiedthrough other sources
Iden
tifica
tion
Screen
ing
Eligibility
Includ
ed
Figure 1.Summary of search strategy. The initial database search yielded 410 abstracts. A further 445 studies were considered for inclusionthrough “snowball” sampling (as described in the text). The majority of studies were excluded because they did not focus onperturbation-based balance training (PBT). Eight studies were selected and included in the final analysis.
Does Perturbation-Based Balance Training Prevent Falls?
May 2015 Volume 95 Number 5 Physical Therapy f 703
Table 1.Characteristics of Studies Included in the Analysis
Moving platform perturbationsto stance to train steppingand grasping reactions, 48–64 perturbations per session(n�16)
1 y
30-min sessions, 3 times/week for 6 wk
Smania et al,33 2010 Individuals with idiopathicParkinson disease and posturalinstability (age range�50–79 y)
Joint mobilization, musclestretching, and motorcoordination exercisesnot designed to improvebalance control (n�27)
Balance training program thatincluded externalperturbations (push/pullfrom a physical therapist; 10min per session) (n�28)
1 mo
50-min sessions, 3 times/week for 7 wk
Lurie et al,34 2013 Older adults at risk for falling (agerange�65–96 y)
Individualized “standard”physical therapy,including strength,flexibility, balance, andmobility exercises, 3–17sessions of mean durationof 43 min (n�33)
Standard physical therapy plustreadmill-inducedperturbations, 1–19 sessionsof mean duration of 44 min(n�26)
3 mo
Rosenblatt et al,37
2013Healthy community-dwelling
older women (mean age�65 y)Nothing (n�80) Backward-directed treadmill
perturbations, four 1-hsessions over 2 wk, averageof 20 perturbations persessiona (n�82)
1 y
Shen and Mak,35
2014Individuals with idiopathic
Parkinson disease (meanage�64 y)
Lower-extremity strengthtraining (n�23)
Voluntary stepping and walkingwith external perturbationdelivered by a treadmill ormanually (n�22)
1 y
60-min sessions, 3 times/week for 8 wk (delivered in two 4-wkblocks with 4 wk of home exercise between blocks)
a Data provided by the study authors.
Does Perturbation-Based Balance Training Prevent Falls?
704 f Physical Therapy Volume 95 Number 5 May 2015
twice per week,37 and 1 studyreported that the frequency rangedfrom 1 to 3 sessions per week.22
When specified (6 studies), partici-pants were prescribed 4 to 24 train-ing sessions over 2 to 8weeks.19,23,33,35,37,38 Number of per-turbations prescribed per sessionranged from 20 to 64, for a total of 80to 1,536 perturbations over thecourse of the training pro-grams.19,23,37,38 Two studies speci-fied the amount of perturbationtraining in terms of the amount oftime (10 minutes per session33 or600 minutes total22) rather than thenumber of perturbations. Intensityof training, in terms of the magni-tude of the perturbation (eg, velocityor acceleration) was specified inonly 2 studies.22,38 Most studiesapplied perturbations either usingequipment readily available in phys-ical therapist practice (ie, tread-mills22,23,34,35,37) or manual externalperturbations (ie, a push or pull fromthe physical therapist33,35). Twostudies used a custom-designed mov-ing platform to deliver the posturalperturbations.19,38
Risk of Bias in Included StudiesThe PEDro scores ranged from 2 to 8out of a maximum of 10 (Tab. 2). Asis common in physical therapyresearch, blinding of participants tothe treatment was not possible, andno study met this criterion. Only onestudy33 blinded therapists to thetreatment; a different therapistadministered each intervention, andthey were not informed which wasthe control and which was theexperimental intervention. In addi-tion to these criteria, the least fre-quently met PEDro criteria were:falls data available for at least 85% ofparticipants initially allocated (5studies), allocation concealment (4studies), point measures and mea-sures of variability reported for fallsdata (4 studies), and blinding ofassessors who collected falls data (orfailure to report that assessors wereblinded; 4 studies).
Effects of InterventionData on proportion of fallers wereavailable for all studies (Tab. 3). Sixof the 8 studies reported that fewerparticipants in the PBT group expe-rienced falls following training com-
pared with the control group. Indi-viduals who completed PBT wereless likely to fall than those in thecontrol groups; the overall risk ratiofor all 8 studies combined was 0.71(95% CI�0.52, 0.96; P�.02; Fig. 2).Six of the 8 studies reported lowerfall rates in the PBT group comparedwith the control group. Overall, par-ticipants who completed PBTreported fewer falls in daily life thanthose in the control groups. The rateratio for all 8 studies combined was0.54 (95% CI�0.34, 0.85; P�.007;Fig. 3).
DiscussionThe results of this meta-analysis pro-vide evidence that PBT appears toboth reduce the likelihood of being afaller and reduce the frequency offalling among people at increasedrisk for falls (ie, older adults and indi-viduals with Parkinson disease).Importantly, the overall effect for fallrate (rate ratio�0.54) was lowerthan reported in previous meta-analyses of general balance trainingfor falls prevention in older adults(rate ratios�0.65–0.86).1–5 How-ever, the relative effectiveness of
a PEDro items: 1�eligibility criteria were specified; 2�participants were randomly allocated to groups; 3�allocation was concealed; 4�groups were similarat baseline; 5�participants were blinded to group allocation; 6�therapists were blinded to participant group allocation; 7�individuals who collected fallsdata were blinded to participant group allocation; 8�falls data were available for at least 85% of participants initially allocated to groups; 9�participantsreceived the treatment as allocated or intention-to-treat analysis was used; 10�statistical analysis of falls data were reported; 11�point estimates andvariability of falls data were reported. 0�no; 1�yes.b Not included in the final score.
Does Perturbation-Based Balance Training Prevent Falls?
May 2015 Volume 95 Number 5 Physical Therapy f 705
PBT compared with more “typical”balance training (ie, training thatrelies on maintaining stability ratherthan responding to instability, as isthe case for PBT) will need to beevaluated in future studies.
Previous work has demonstratedthat PBT improves control of reac-tions to postural perturbations in thelaboratory among healthy olderadults.19,20 Improved control of bal-
ance reactions may translate toimproved ability to respond to anunexpected “real world” loss of bal-ance, thereby preventing falls. Thereis evidence that the effects of PBTare specific to the nature of the per-turbations experienced in training.One study included in the currentmeta-analysis only applied “forwardfall” postural perturbations in train-ing (ie, simulated trips).37 Althoughthis study did not report a statisti-
cally significant reduction in totalfalls with training, there was a signif-icant decrease in the number of fallsresulting from trips. These resultsimply that training to respond to sim-ulated trips may specifically improvethe ability to respond to trips occur-ring in daily life but that these train-ing effects do not translate to othertypes of postural perturbations expe-rienced in daily life. Thus, we recom-mend that PBT programs include a
Table 3.Number of Participants Who Reported One or More Falls During the Follow-up Period (ie, Number of “Fallers”) and Number ofFalls Reported Are Presented for Each Studya
Study
Control Group Experimental Group
Number of Fallers Number of Falls Sample Size Number of Fallers Number of Falls Sample Size
Shimada et al,22 2004 6 11 11 5 8 15
Protas et al,23 2005 6 23b 9 5 10b 9
Maki et al,38 2008 3b 4b 4 2b 4b 4
Mansfield et al,19 2010 4b 4b 15 5b 6b 16
Smania et al,33 2010 19b 111 27 8b 36 28
Rosenblatt et al,37 2013 45b 91 80 46b 87 82
Lurie et al,34 2013 11 32b 33 5 10b 26
Shen and Mak,35 2014 13 18c 23 6 6c 22
Total 107 294 202 82 167 202
a The natural logarithm for the risk ratio (proportion of fallers) and rate ratio (number of falls) were entered into the meta-analysis (see Figs. 2 and 3).b Data provided by the study authors.c Fall rates reported from the start of the intervention period. Falls that occurred during the intervention period were not included in this count. Additionally,2 individuals from the control group and 1 participant from the experimental group were excluded from reporting of fall rates; thus, the sample size is 21per group for number of falls.
Figure 2.Results of meta-analysis for proportion of “fallers.” The risk ratio was calculated from the proportion of fallers (see Tab. 2). Studieswere heterogeneous (I2�31%). The natural logarithm of the risk ratio and standard error of the risk ratio was calculated based onthe proportion of fallers in each group for each study and included in the meta-analysis. The overall risk ratio was 0.71 (95%confidence interval�0.52, 0.96; P�.02). IV�inverse variance.
Does Perturbation-Based Balance Training Prevent Falls?
706 f Physical Therapy Volume 95 Number 5 May 2015
variety of perturbation modalities,directions, amplitudes, and concur-rent cognitive and motor tasks totrain balance reactions in varied sit-uations. Ultimately, this approachmay result in a greater reduction infall rates compared with PBT pro-grams that use a single perturbationmodality.
We included studies in the currentanalysis if they involved 2 or moresessions of perturbation training.However, it is possible that a singlesession of 24 perturbations is suffi-cient to bring about lasting improve-ments in reactive balance control20
and prevent falls in daily life.41 Threeof the studies in the current reviewaimed to achieve a dose of approxi-mately 700 to 1,150 perturbationsover several weeks of training.19,23,38
However, one study only includedapproximately 80 perturbations over4 sessions.37 Thus, the optimal doseof perturbation training for causinglasting changes in reactive balancecontrol must be determined. Addi-tionally, the potential benefit of“booster” training sessions after aninitial period of training should beexplored.41
Most of the included studies com-pared PBT with an “active” controlgroup. The characteristics of thecontrol activities varied widely fromlow-intensity stretching exerciseswith minimal physiological benefit19
to activities resembling current “bestpractice” for improving balance andmobility and for preventing falls (eg,strengthening exercises, activities tochallenge static and dynamic bal-ance, gait training).34 Existing evi-dence suggests that exercise, specif-ically exercise that challengesbalance control, can prevent falls.Evidence for novel interventions canonly be provided by comparing theeffect of the novel interventionagainst current best practice.42 Thus,future studies of novel interventionsfor falls prevention should include acontrol group that engages in activi-ties that have previously been shownto prevent falls.
LimitationsStudy quality was generally low tofair, with half of the studies scoring 5or less on the PEDro scale. Thus,despite the significant effect of per-turbation training on falls with thecurrent meta-analysis, the evidence
for perturbation training is currentlylimited by the quality of the studiesincluded. All studies included werepotentially biased by failure to blindresearch participants to the interven-tion. True blinding of research par-ticipants is not possible for exerciseinterventions of this nature43; how-ever, when study designs incorpo-rate an active control intervention,participants in the control groupmay believe that the intervention isbeneficial. One study attempted toblind the therapists administeringthe intervention by having differenttherapists administer the differentinterventions.33 However, thisapproach may have the effect of fur-ther biasing the results; as therewere 2 differences between thegroups (therapist and interventiontype), it is not clear if the differencesbetween the 2 groups were due todifferences in the intervention or todifferences in the way the interven-tions were administered by the ther-apists. Study authors did not alwaysreport the interventions withenough detail to be replicated. Inparticular, the intensity (magnitude)and number of perturbations werenot well described. Furthermore, it
Figure 3.Results of meta-analysis for rate of falls. The rate ratio was calculated from fall rates (ie, number of falls per person reported duringthe follow-up period; Tab. 2). Studies were heterogeneous (I2�73%). The natural logarithm of the rate ratio and standard error ofthe rate ratio were calculated based on the number of falls in each group for each study and included in the meta-analysis. The overallrate ratio was 0.54 (95% confidence interval�0.34, 0.85; P�.007). IV�inverse variance.
Does Perturbation-Based Balance Training Prevent Falls?
May 2015 Volume 95 Number 5 Physical Therapy f 707
was not always clear if the interven-tion was carried out as prescribed.Thus, there is a need for improvedreporting of balance training inter-vention characteristics.44
Terminology regarding PBT is notconsistent in current literature.Thus, a simple literature searchwould not have identified all appro-priate studies. However, we are rea-sonably confident that the multistagesearch strategy used, includingsnowball sampling, enabled us toidentify all relevant studies. Addition-ally, all meta-analyses are potentiallylimited by publication biases towardstudies reporting positive results.We attempted to address this limita-tion by seeking and including unpub-lished falls data from randomizedcontrolled trials of PBT. Further-more, only 2 studies included in thecurrent meta-analysis reported posi-tive results in terms of occurrence offalls33,35 (ie, a statistically significantreduction in falls with PBT comparedwith the control group). Finally, weidentified only 8 studies for inclusionin the meta-analysis. The study designsdiffered in several ways that couldhave contributed to differing resultsamong studies, including populationstudied, characteristics of training (eg,frequency, duration, method of pertur-bation, inclusion of nonperturbationbalance training), characteristics of thecontrol intervention, method of col-lection of falls data, and duration offalls follow-up period. Thus, we wereunable to conduct exploratory sub-analyses to determine if the effects ofPBT differ based on these differingstudy or population characteristics.
Perturbation-based balance trainingappears to reduce fall rates amongolder adults and individuals with Par-kinson disease. Additional work isneeded to determine if PBT is supe-rior to more traditional, well-established exercise programs forpreventing falls. Future work alsoshould consider the effect of PBT on
balance confidence, daily physicalactivity, and participation.
Dr Mansfield provided concept/idea/projectdesign. All authors provided writing anddata collection. Dr Mansfield provided dataanalysis and project management. Allauthors provided consultation (includingreview of manuscript before submission).The authors thank Jessica Babineau, BA,MLIS, for her assistance with the literaturesearch.
Dr Patterson was supported by a Focus onStroke personnel award from the Heart andStroke Foundation and the Canadian StrokeNetwork.
DOI: 10.2522/ptj.20140090
References1 Province MA, Hadley EC, Hornbrook MC,
et al. The effects of exercise on falls inelderly patients: a preplanned meta-analysis of the FICSIT trials. JAMA. 1995;273:1341–1347.
2 Robertson MC, Campbell AJ, Gardner MM,Devlin N. Preventing injuries in older peo-ple by preventing falls: a meta-analysis ofindividual level data. J Am Geriatr Soc.2002;50:905–911.
3 Chang JT, Morton SC, Rubenstein LZ, et al.Interventions for the prevention of falls inolder adults: systematic review and meta-analysis of randomised clinical trials. BMJ.2004;328:680–683.
4 Sherrington C, Whitney JC, Lord SR, et al.Effective exercise for the prevention offalls: a systematic review and meta-analy-sis. J Am Geriatr Soc. 2008;56:2234–2243.
5 Gillespie LD, Robertson MC, Gillespie WJ,et al. Interventions for preventing falls inolder people living in the community.Cochrane Database Syst Rev. 2012;9:CD007146.
6 Batchelor F, Hill K, Mackintosh S, Said C.What works in falls prevention afterstroke? A systematic review and meta-anal-ysis. Stroke. 2010;41:1715–1722.
7 Verheyden GS, Weerdesteyn V, PickeringRM, et al. Interventions for preventingfalls in people after stroke. CochraneDatabase Syst Rev. 2013;5:CD008728.
8 Maki BE, McIlroy WE. Postural control inthe older adult. Clin Geriatr Med. 1996;12:635–658.
9 Robinovitch SN, Feldman F, Yang Y, et al.Video capture of the circumstances of fallsin elderly people residing in long-termcare: an observational study. Lancet. 2013;381:47–54.
10 Shumway-Cook A, Woollacott MH. MotorControl: Translating Research Into Clini-cal Practice. 3rd ed. Philadelphia, PA: Lip-pincott Williams & Wilkins; 2007.
11 Schieppati M, Nardone A. Free and sup-ported stance in Parkinson’s disease: theeffect of posture and “postural set” on legmuscle responses to perturbation and itsrelation to the severity of the disease.Brain. 1991;114:1227–1244.
12 Maki BE, Edmondstone MA, McIlroy WE.Age-related differences in laterallydirected compensatory stepping behavior.J Gerontol A Biol Sci Med Sci. 2000;55:M270–M277.
13 Mansfield A, Inness EL, Wong JS, et al. Isimpaired control of reactive steppingrelated to falls during inpatient strokerehabilitation? Neurorehabil NeuralRepair. 2013;27:526–533.
14 Horak FB, Henry SM, Shumway-Cook A.Postural perturbations: new insights fortreatment of balance disorders. Phys Ther.1997;77:517–533.
15 Maki BE, McIlroy WE. Change-in-supportbalance reactions in older persons: anemerging research area of clinical impor-tance. Neurol Clin. 2005;23:751–783.
16 Pai Y-C, Bhatt TS. Repeated-slip training:an emerging paradigm for prevention ofslip-related falls among older adults. PhysTher. 2007;87:1478–1491.
17 Rogers MW, Johnson ME, Martinez KM,et al. Step training improves the speed ofvoluntary step initiation in aging. J Geron-tol A Biol Sci Med Sci. 2003;58:46–51.
18 Marigold DS, Eng JJ, Dawson AS, et al.Exercise leads to faster postural reflexes,improved balance and mobility, and fewerfalls in older persons with chronic stroke.J Am Geriatr Soc. 2005;53:416–423.
19 Mansfield A, Peters AL, Liu BA, Maki BE.Effect of a perturbation-based balance-training program on compensatory step-ping and grasping reactions in olderadults: a randomized controlled trial. PhysTher. 2010;90:476–491.
20 Bhatt T, Yang F, Pai Y-C. Learning to resistgait-slip falls: long-term retention incommunity-dwelling older adults. ArchPhys Med Rehabil. 2012;93:557–564.
21 Parijat P, Lockhart TE. Effects of moveableplatform training in preventing slip-induced falls in older adults. Ann BiomedEng. 2012;40:1111–1121.
22 Shimada H, Obuchi S, Furuna T, Suzuki T.New intervention program for preventingfalls among frail elderly people: the effectsof perturbed walking exercise using abilateral separated treadmill. Am J PhysMed Rehabil. 2004;83:493–499.
23 Protas EJ, Mitchell K, Williams A, et al. Gaitand step training to reduce falls in Parkin-son’s disease. Neurorehabilitation. 2005;20:183–190.
24 Higgins JPT, Green S. Cochrane Handbookfor Systematic Reviews of Interventions:Version 5.1.0. The Cochrane Collabora-tion. Updated March 2011. Available at:http://www.cochrane-handbook.org. Ac-cessed August 1, 2014.
25 Liberati A, Altman DG, Tetzlaff J, et al. ThePRISMA statement for reporting system-atic reviews and meta-analyses of studiesthat evaluate health care interventions:explanation and elaboration. PLoS Med.2009;6:e1000100.
Does Perturbation-Based Balance Training Prevent Falls?
708 f Physical Therapy Volume 95 Number 5 May 2015
26 Fegran L, Hall EO, Uhrenfeldt L, et al. Ado-lescents’ and young adults’ transitionexperiences when transferring from pae-diatric to adult care: a qualitative metasyn-thesis. Int J Nurs Stud. 2014;51:123–135.
27 Groene O, Botje D, Sunol R, et al. A sys-tematic review of instruments that assessthe implementation of hospital qualitymanagement systems. Int J Qual HealthCare. 2013;25:525–541.
28 Sweeney S, Obure CD, Maier CB, et al.Costs and efficiency of integrating HIV/AIDS services with other health services: asystematic review of evidence and experi-ence. Sex Transm Infect. 2012;88:85–99.
29 Maher CG, Sherrington C, Herbert RD,et al. Reliability of the PEDro scale for rat-ing quality of randomized controlled trials.Phys Ther. 2003;83:713–721.
30 de Morton NA. The PEDro scale is a validmeasure of the methodological quality ofclinical trials: a demographic study. Aust JPhysiother. 2009;55:129–133.
31 Higgins JPT, Thompson SG. Quantifyingheterogeneity in a meta-analysis. Stat Med.2002;21:1539–1558.
32 DerSimonian R, Laird N. Meta-analysis inclinical trials. Control Clin Trials. 1986;7:177–188.
33 Smania N, Corato E, Tinazzi M, et al. Effectof balance training on postural instabilityin patients with idiopathic Parkinson’s dis-ease. Neurorehabil Neural Repair. 2010;24:826–834.
34 Lurie JD, Zagaria AB, Pidgeon DM, et al.Pilot comparative effectiveness study ofsurface perturbation treadmill training toprevent falls in older adults. BMC Geriatr.2013;13:49.
35 Shen X, Mak MKY. Technology-assistedbalance and gait training reduces falls inpatients with Parkinson’s disease: a ran-domized controlled trial with 12-monthfollow-up. Neurorehabil Neural Repair.2015;29:103–111.
36 Halvarsson A, Franzen E, Faren E, et al.Long-term effects of new progressivegroup balance training for elderly peoplewith increased risk of falling: a random-ized controlled trial. Clin Rehabil. 2013;27:450–458.
37 Rosenblatt NJ, Marone J, Grabiner MD.Preventing trip-related falls by community-dwelling adults: a prospective study [Erra-tum in: J Am Geriatr Soc. 2014;62:1003].J Am Geriatr Soc. 2013;61:1629–1631.
38 Maki BE, Cheng KC-C, Mansfield A, et al.Preventing falls in older adults: new inter-ventions to promote more effectivechange-in-support balance reactions. JElectromyogr Kinesiol. 2008;18:243–254.
39 Mansfield A, Peters AL, Liu BA, Maki BE. Aperturbation-based balance training pro-gram for older adults: study protocol for arandomised controlled trial. BMC Geriatr.2007;7:12.
40 Ory MG, Schechtman KB, Miller P, et al.Frailty and injuries in later life, FICSIT tri-als. J Am Geriatr Soc. 1993;41:283–296.
41 Pai YC, Bhatt T, Yang F, Wang E. Pertur-bation training can reduce community-dwelling older adults’ annual fall risk: arandomized controlled trial. J Gerontol ABiol Sci Med Sci. 2014;69:1586–1594.
42 Dobkin BH. Confounders in rehabilitationtrials of task-oriented training: lessonsfrom the designs of the EXCITE and SCILTmulticenter trials. Neurorehabil NeuralRepair. 2007;21:3–13.
43 Andrews K. The limitations of randomizedcontrolled trials in rehabilitation research.Clin Rehabil. 1991;5:5–8.
44 Farlie MK, Robins L, Keating JL, et al.Intensity of challenge to the balance sys-tem is not reported in the prescription ofbalance exercises in randomised trials: asystematic review. J Physiother.2013;59:227–235.
Appendix.Sample Search Strategy (MEDLINE)
1. (perturbat* adj2 (train* or rehab* or exercis*)).tw.
2. ((platform* or “dynamic balanc*” or “slip” or “dynamic stabili*”) adj2 (train* or rehab* or exercis*)).tw.
3. 1 or 2
4. Limit 3 to humans
5. Remove duplicates from 4
Does Perturbation-Based Balance Training Prevent Falls?
May 2015 Volume 95 Number 5 Physical Therapy f 709
