REVIEW

http://dx.doi.org/10.17784/mtprehabjournal.2016.14.432

Do functional mobility tests predict the risk of falls in

community-dwelling elderly?
Ana Carolina Silva de Souza Moreira1, Giovana Zarpellon Mazo2, Eduardo Hauser3, Paula Bertolini de Paiva3,
Enaiane Menezes1, Fernando Luiz Cardoso4

ABSTRACT
Introduction: The falls are associated with morbidity and mortality in the elderly. Numerous of functional mobility clinical tests have
been created to identify older adults with potential for risk of falls. Objective: The purpose of this systematic review was to determine
the predictive validity of functional mobility tests to predict the risk of falls in community-dwelling elderly. Method: Articles in English
were searching in MEDLINE, SCOPUS and CINAHL. We found 18,520 documents and, after applying the inclusion and exclusion criteria,
11 articles were part of the final analysis. All articles analyzed included subjects over 60 years old. Results: The results showed that
the TUG Test has good discriminative validity for elderly non-institutionalized, but it does not provide an adequate predictive validity.
The TUG Test may not be enough as a unique basic screening tool to detect the risk of elderlies´ falling. Conclusion: It is suggested that
the TUG Test should be used in combination with other predictors of falling risk tools or should it be reconfigured for the different
levels of elderly physical active functionality.
Key Words: fall risk, older adults, validity tests, review, meta-analysis.

INTRODUCTION
One of the main causes of the loss of autonomy and of importance of the instrument used for decision-making in
independence of the elderly community residents are the relation to therapeutic and / or diagnostic / predictive conduct,
consequences of physical falls (1, 2), according to the database including the cost ratio of the instrument´s application, the
of the Unified Health System / Brazilian Ministry of Health(3), risks to which the evaluator and the patient will be submitted
in Brazil, between 1996 and 2005, about 24,645 elderly people and patient´s acceptability to such an instruments should
died due to falls, occupying the third place of mortality and also be considered when the indication and use of these
the first place among hospitalizations. In 2005, the prevalence instruments(6, 7).
of falls in the elderly was 34.8% in seven Brazilian states, Numerous instrument model methods (scales, tests,
and among those who fell, 55% reported a single fall in the questionnaires, among others) have been created among the
previous year. These data are in agreement with the data efforts to minimize and detect the risk of falls in the elderly
found in the United States, where about 30% to 40% of the population so that health professionals can identify individuals
elderly livings in a community suffer at least one fall in their with the potential for falls. The most commonly used tests
life, and this percentage increases to 60% when the elderly are the Balance Berg Scale (BBS), the Clinical Test of Sensory
have experienced a fall in previous year(4). Therefore, falls in Organization and Balance Test (CTSIB), the Functional Reach
the elderly become a public health problem, considering the Test (FRT), the Tinetti Balance Scale (TBS), One Leg Stand (OLS),
social and economic burden they generate(5). Tandem Stand Test (TS), Chair Stand Five times Test (CS-5), and
Given these considerations, there is a constant concern of The Timed Up and Go Test (TUG). To predict the risk of falls in
health professionals to prevent and mitigate the consequences the elderly, there is still no reference test considered “gold”,
of falls in the elderly´s health. In both clinical and community since the fall phenomenon is multifactorial, one of the factors
health, we must conform to available diagnostic and / or that explain the postural balance is the functional mobility of
predictive methods acceptable to the individual. The degree the individual.

Corresponding author: Ms. Ana Carolina Silva de Souza – Santa Catarina State University /UDESC. Centro da Ciências da Saúde e do Esporte/CEFID; Rua Pascoal Simone,
358; Coqueiros; Florianópolis/SC - Brazil; CEP: 88080-350. e-mail: [email protected]

Ph.D. Candidate in Graduate Program in Human Motion Sciences, Santa Catarina State University (UDESC), Santa Catarina (SC), Brazil.
1

Full list of author information is available at the end of the article.

Financial support: None.
Submission date 21 September 2016; Acceptance date 14 December 2016; Publication online date 27 December 2016

Manual Therapy, Posturology & Rehabilitation Journal. ISSN 2236-5435. Copyright © 2016. This is an Open Access article
distributed under the terms of the Creative Commons Attribution Non-Commercial License which permits unrestricted non-
commercial use, distribution, and reproduction in any medium provided article is properly cited.
Tests predict the risk of falls in community-dwelling elderly MTP&RehabJournal 2016, 14: 432

In the scientific environment, for clinical and epidemiological METHOD

research, content, criterion and construct validity are essential
requirements to obtain a good psychometric test, according to Inclusion Criterion
the sample profile. Since clinical trials are considered essential The following inclusion criteria were used: studies which
tools to diagnose and / or predict the possibility of risk of some enrolled elderly samples living in a community aged over
phenomenon. This systematic review aims to analyze the 60 years (1); Elderly sample without physical limitations (2);
tests in their validation criterion, essentially predictive one. Evaluation of physical balance by at least one clinical test (field
In the construct concept the test must measure exactly what work) (3); Presentation of at least one test to compare the
it intends, but also carry out this measurement with accuracy, results (4); Presentation of sensitivity and specificity values
without errors (validity of criterion). Predictive validity is part of the tests used (5); Publication of results in English, Spanish
of the criterion validity, which should be well understood for and or Portuguese (6).
the scientific community to which the sample profile of the Experimental or observational studies, studies with
samples exclusively of elderly people considered dependent,
instrument is appropriate to diagnose and / or predict the
fragile or with pre-established pathologies (neurological,
phenomenon. Thus, one can distinguish two types of validity
orthopedic, cardiovascular, vestibular system) were excluded
criterion: predictive and concurrent(8, 9).
in this study.
In this sense, there is a literature in the literature about
the most appropriate clinical tests to evaluate the risk of falls
in the elderly. Since current studies reporting that these tests Search Strategies
are having a “roof” or “floor” effect in detecting the risk of The research was conducted according to the Preferred
Reporting Items for Systematics Reviews and Meta-Analysis
falls in the elderly, such as: BBS (10, 11, 12, 13), TUG(14, 15, 16) and
(PRISMA recommendations) guidelines. The MEDLINE
other tests(17, 18). This entails a lack of evidence on which tests
(1966‑14/ 10/2014), SCOPUS (1960-16/ 10/2104) and CINAHL
would be best suited for use by health care professionals,
(1982-10/ 24/2014) databases were selected. The filters used
since they are poorly studied and analyzed in terms of their
as data search strategy were divided into three blocks: Elderly:
psychometric qualities.
Postural Balance and Study Type (APPENDIX I).
Insufficient clarity of instrument validity may mislead
researchers to make decisions in the clinical setting and in the
Article Selection
development of scientific research, and may underestimate
Two independent evaluators selected the studies after
or overestimate the results, resulting in potentially erroneous
reading the titles, excluding those that were not related to
actions in clinical decision-making or in data conclusions from
the purpose of the review. After the selection of the titles,
their empirical research. In addition, it is expected that in the
the evaluators analyzed the abstracts of the articles to identify
literature would be clear about the indications of each test
those that met the inclusion criteria and, subsequently, the
in relation to the different population profile, providing its
eligible studies were analyzed entirety for later inclusion in
construct, the dimension corresponding to the construct, test
the systematic review.
outcomes, collection time, test familiarization, training and
Evaluation of the methodological quality of the selected
manual detailed instruction to the researchers. Thus, when
studies
the aforementioned information is easily available to the For the methodological evaluation, the authors of the
researchers, it is believed that it would avoid the traditional present study developed a scale composed of 10 questions.
biases of publications generated by the lack of information This scale was developed from existing scales in the literature,
about the suitability and reliability of the instrument used in such as QUADAS and STARD(19) that did not meet the needs
different elderly populations. of the study objective. The questions elaborated refer to the
To obtain a more accurate indication of the discriminative presence of methodological aspects of the study evaluated,
ability and diagnostic accuracy of the functional balance tests containing the following questions:
as a clinical screening instrument for the identification of
elderly at risk of falling, a comprehensive systematic review 1. Does the article present a sample profile? (E.g., age,
and a meta-analysis were performed. In addition, an accurate sample size, faller and non-fallers, secondary diseases)
analysis of the quality of the studies was also performed. 2. Does the article present inclusion and exclusion criteria
Considering the lack of studies that demonstrate the predictive for the study?
capacity of functional mobility tests in the elderly, this study
3. Does the article provide a reference test?
aimed to analyze theoretically the predictive validity of these
available tests, as well as, to their methodological quality in 4. Does the article present the value of the reference test
terms of design and procedures. (sensitivity and specificity, cutoff point, etc)?

2
MTP&RehabJournal 2016, 14: 432 Moreira ACSS et al.

5. Does the article present cut-off points for analyzed Table 1. Shows 2 × 2 (two-by-two) table
tests? Test True positives(TP) False positives (FP) Total test positives:
positive a b a+b
6. Were the test results interpreted without influence
Test False negative (FN) True negatives (TN) Total test negatives:
among them? negative c d c+d

7. Has the application methodology of the tests been Total diseased: Total normal: Total population:
a+c b+d a+b+c+d
adequately described that could allow their later
replication? Meta-Analysis
8. Were the evaluators previously trained to apply the
We used the Q Chi-square statistic value to estimate the
heterogeneity of the individual studies that contribute to the
tests?
estimate pooled. The homogeneity was obtained by evaluating
9. Were the tests applied on all patients? (If not, on very the differences between the studies, that is, if they were larger
large sample, at random selection?) than expected just by chance. The ρ <0.05 in this analysis
indicates the presence of greater heterogeneity than would be
10. Have the tests been applied with rest interval so that expected by chance alone.
there wouldn´t be no influence on each other? Due to lack of standardization, different thresholds can
be used in included studies to define a positive test result.
All these questions was answered with the answers “yes”,
Differential threshold effect may be the reason for the
“partially” and “no”. With each result yes, it was assigned the detectable difference in sensitivities and specificities of the
value of 1 point whereas for the “partially” 0.5 and results precision test studies. The estimated accuracy of each study
“no” the zero value. in the curve (ROC) space (receiver operating characteristic)
The evaluation process of the selected articles was carried and the Spearman Correlation Coefficient between log (SEN)
out by four independent reviewers, whose analysis of each and log (1-SPE) were evaluated for threshold effect. A typical
question of the methodological evaluation was performed in “shoulder arm” plot in the ROC curve space and a strong positive
pairs at each step of the process. To resolve disagreements correlation suggested a threshold effect.
Statistical analysis is not always necessary in all systematic
among reviewers, a third reviewer evaluated all items involved.
reviews to check the test precision studies. The necessary
The systematic review was registered in PROSPERO under
condition for gathering the estimates is that the studies and
registration number CRD 42015026961. results should be reasonably homogeneous. Estimates can be
The values of true positive (TP), false positive (FP), true grouped by the fixed-effect model (FEM) or the random effects
false (TN), false negative (FN), sensitivity (SEN), specificity model (REM) to incorporate the variation between the studies,
(SPE), accuracy (ACC), positive predictive value (PPV), negative and the output can be represented graphically as funnel plots.
predictive value (VPN), positive likelihood ratio (PLR) and If heterogeneity due to the threshold effect is present, accuracy
negative likelihood ratio (NLR), in order to check the fallers. data can be gathered by fitting a summary ROC curve (SROC) and
SEN, SPE, ACC, VPP, VPN, PLR e NLR were calculated using the by area under the curve (AUC). The SROC curve summarizes and
following formulas: 1 to 7 and Table 1. gathers the true and false positive rates of different diagnostic
studies. The overall performance of diagnostic studies can be
visualized and reflected by an SROC curve without being affected
SEN=TP/(TP+FN) (1)
by a change in limit values. The best diagnostic modality would
produce a point in the upper left corner or coordinates (0.1)
SPE=TN/(FP+TN) (2) of the SROC space, representing 100% sensitivity (no false
negatives) and 100% specificity (false positives) at the individual
subject level. Likewise, AUC ranges from 1 to a perfect test that
ACC=(TP+TN)/(TP+FP+TN+FN) (3)
always diagnoses correctly, to 0 for a test that never does this
from individual studies or meta-analyzes. The homogeneity
PPV=TP/(TP+FP) (4) test, threshold effect analysis, pooled and specific weighted
sensitivity, SROC curve and sensitivity analysis were performed
NPV=TN/(FN+TN) (5) using Meta-Disc version 1.4.

RESULTS
PLR=SEN/(1−SPE) (6)
In the electronic databases used to search for articles,
2.607 were found in MEDLINE, 5.276 in SCOPUS and 11.464 in
NLR=(1−SEN)/SPE (7) CINAHL, totaling 18.520 documents. Of these, 1.890 articles

3
Tests predict the risk of falls in community-dwelling elderly MTP&RehabJournal 2016, 14: 432

were duplicated for a total of 16.630 articles. In the analysis evaluation of methodological quality using relevance criteria
of the persistent titles the research was selected 311 articles. to investigate the risk of falling (phenomenon) indicated that
After analysis of the abstracts, 97 articles were selected for only two studies had representative sample size. Overall, four
the entire reading. After analyzing the articles in full, 6 of them studies were classified as high quality studies, of which only
we could not get fully access, 80 were excluded because they one included representative sample.
did not meet the inclusion criteria. Thus, 11 were eligible for
the present study (Figure 1).
The methodological quality assessment (Table 2), from
the five studies (45.45%) were prospective and six (54.54%)
retrospective studies referring to the history of falling of the
elderly. It was verified that the articles that scored the most
were those performed by Dite et al.(20) (9.5 points), Rose et al.(21)
(9 points) and Wrisley and Kumar(22) (9 points). The lowest
scores were Truebloodet al.(23) (5 points); Greene et al.(24)
(5.5 points) and Tsutsuminoto et al.(25) (6 points). Among
the questions that were less pointed out and decreased
in the methodological quality of the articles scores are
the presentation of simple aspects such as “time to rest”
(Question 10) and “whether the evaluators were previously
trained” (Question 8). However, two questions were addressed
by all articles, whether “the tests were applied to all subjects
(or at random)” (question 9) and whether the tests were
analyzed without any influence on each other (question 6).
The sample profile and characteristics of the studies are
summarized in Table 3. It is noted that 81% of the studies
excluded elderly with cognitive deficit using the mini-mental
examination, 72.72% reported prevalent diseases, 45.45% the
type of medication administered, 63.63% reported the number
of elderly people who use gaiters and limitations of their daily
activities; 36.36% excluded elderly individuals who presented
balance deficiency; and 27.27% evaluated the state of the
visual, auditory and / or vestibular system. Complementary Figure 1. Flowchart of the steps of the systematic review studies.

Table 2. Methodological quality of systematic review studies.

Studies Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Total

Trueblood et al. (2001) (23)

± ✓ ✗ ✗ ✓ ✓ ± ✗ ✓ ✗ 5
Shumway-cook et al. ✓ ✓ ✓ ✗ ✓ ✓ ✓ ✗ ✓ ✗ 7
(2000) (28)
Rose et al. (2002)(21) ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✗ ✓ ✓ 9
Greene et al. (2010) (24)
± ± ✓ ✓ ✗ ✓ ± ✗ ✓ ✗ 5.5
Murphy et al. (2003) (29) ✓ ± ✗ ✗ ✓ ✓ ± ✓ ✓ ✓ 7
Wrisley; Kumar (2010) (22) ± ✓ ✓ ✓ ✓ ✓ ± ✓ ✓ ✓ 9
Dite et al. ✓ ± ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 9.5
(2002) (20)
Chiu et al. (2003) (16) ✓ ✓ ✗ ✗ ✓ ✓ ✓ ✗ ✓ ✓ 7
Stel et al. ✓ ± ✓ ✓ ✓ ✓ ✓ ✗ ✓ ✗ 7.5
(2003) (26)
Tsutsumimoto et al. ✓ ✓ ✗ ✗ ✓ ✓ ✓ ✗ ✓ ✗ 6
(2011)(25)
Greene et al. (2012) (27) ✓ ± ✓ ✓ ✓ ✓ ± ✗ ✓ ✗ 7
Caption: Q1: Question 1; Q2: Question 2; Q3: Question 3; Q4: Question 4; Q5: Question 5; Q6: Question 6; Q7: Question 7; Q8: Question 8; Q9: Question 9; Q10: Question 10;
✓: yes, ±: partial; ✗ not.

4
MTP&RehabJournal 2016, 14: 432 Moreira ACSS et al.

Table 3. Sample profile of the studies evaluated.

Trueblood et al. (2001)(23)

Wrisley; Kumar (2010)(22)

Murphy et al. (2003) (29)
Greene et al. (2010) (24)

Greene et al. (2012)(27)
Shumway-cook et al.

Dite et al. (2002) (20)

Rose et al.(2002) (21)

Chiu et al. (2003)(16)

Stel et al. (2003)(26)
(2000) (28)
Studies

Age group (years) > 60 > 65 > 60 > 60 > 60 60 -90 > 65 > 65 60 -85 > 60
Sample size 198 30 134 349 45 35 81 34 435 226
Secondary or specific
✗ ✓ ✓ ✗ ✓ ✓ ✓ ✗ ✓ ✓
diagnoses
Medications ✗ ✓ ✓ ✗ ✗ ✓ ✗ ✗ ✓ ✓
Sensorial deficits ✗ ✓ ✗ ✗ ✗ ✗ ✓ ✗ ✗ ✓
Balance deficits ✓ ✗ ✗ ✗ ✗ ✓ ✓ ✓ ✗ ✗
Cognitive state ✓ ✗ ✓ ✗ ✓ ✓ ✓ ✓ ✓ ✓
Limitation of Daily Life
✓ ✓ ✓ ✗ ✗ ✓ ✓ ✓ ✗ ✗
Activities
Gear device for
✗ ✓ ✓ ✗ ✗ ✗ ✓ ✓ ✓ ✓
Walking
Caption: ✓: yes, ±: partial; ✗ not.

Regarding the tests found in the literature, eight studies I² = 85.53%), respectively. The discriminatory TUG values ​​to
analyzed the Timed Up and Go test (TUG); Four Balance Berg identify the elderly fallers ranged from 10 to 20 seconds,
Scale (BBS); Three Functional Reach Test (FRT); Three Chair with a mean of 16.25 seconds. When analyzing the threshold
Stands Five times Test (CS-5); Two Tandem Stand Test (TS); effect (liminal), Spearman’s correlation coefficient was equal
Two POMA Balance (POMA-B); One POMA Gait (POMA-G); to -0.075, indicated a very weak association, not being
One POMA Balance and Gait (POMA GT); One Step Test (ST); statistically significant (p = 0.85). The post-test probability of
One 360º Turns (360 º T); One Semi-Tandem Stand Test (STST); an elderly person suffering a fall or not, considered a positive
One Side by Side Stand Test (SSST); One Clinical Test of Sensory and negative odds ratio was 2.68 (1.73-4.16) and X² of 34.4
Organization and Balance Test (CTSIB); One One Leg Stand (p <0.001), and 0.41 (95% CI: 0.27-0.64) and X² of 107.45
(OLS); One Four Square Step Test (FSST); One Up-to-go (UG); (p <0.001), respectively. The SROC curve synthesizes the
One 3m Walking Test (3WT) and; One Ordered Multi-Stepping sensitivity and false positive values (1- specificity), obtaining
Over Hoop (OMO) (Table 4). Of these, only the TUG and BERG a value of 0.85 of the area under the curve (AUC) and Cochran
Tests can be performed meta-analysis because they have Q of 0.78 as the highest common value of the sensitivity and
enough studies for the analytical tool. specificity. The overall Odds Ratio of these studies was 8.42
In the TUG test, the combined sample size was composed (95% CI: 3.4-20.8) and X² = 52.25 (p <0.001).
of 1113 elderly individuals over 60 years old who were able to On the four studies found on BBS, one did not analyze
complete the test. Of these 616 elderly people reported a history the predictive validity, the other three studies obtained the
of a retrospective or prospective falls. Follow-up periods varied combined sample size of 653 elderly individuals over 60 years
5 years retrospective and 2 years prospective. Sample sizes old who were able to complete the Scale. Of these 329 elderly
of the studies ranged from 30 to 226 participants. Six studies people reported a history of a retrospective or prospective
separated their sample between fallers and non‑fallers; And fall. Follow-up periods varied 5 years retrospective and
two studies divided the sample into multi‑fallers (with more 2 years prospective studies. Sample sizes of studies ranged
than two falls), not multi-fallers (one fall) and non-fallers from 34 to 329 subjects. Two studies separated their sample
(TABLE 4). Forest graphs of sensitivities and specificities of between fallers and non-fallers, and one study separate
9 subsets of data from all eight studies are shown in Figure 2. fallers, multi-fallers and non-fallers (TABLE 4). Sensitivity
Sensitivity values ​​ranged from 10% to 89% with an average values ​​ranged from 45% to 95.5% with an average of 58.7%
of 61.6% (95% CI: 57.2 -65.9%), and the specificity ranged (CI = 95%: 53.1-64%, X² = 29.95, p <0.0001, I² = 83%), and
from 62.9 to 100% with a mean of 76.5% (95% CI: 73-79.8%). the specificity ranged from 51% to 95.5% with a mean of
The heterogeneity test of sensitivity and specificity shows 71.3% (95% CI: 66-76.2%, X² = 30.73, p <0.0001, I² = 82%).
X² = 59.34 (P <0.0001, I² = 87.42%) and X² = 67.2 (P <0.0001, The discriminatory BBS scores to identify the elderly fallers

5
 Table 4. Characterization of the studies regarding the objectives, sample, characterization of falls, values of sensitivity and specificity of the tests and conclusion of the studies.

Sample History of fall

Author Objective Fall period Test SEN SPE CP Conclusion
Mean age /standard 2 or
N M H SQ 1Q
deviation +f

Determine the SEN and POMA-B 24% 91% 12/16

SPE of fall prediction
The tests were accurate to
tests and whether Total of sample: Six months POMA-G 21% 95% 8/12
Trueblood et al 2001(23) 189 143 37 99 73 26 identify individuals without
the combination of 77.9 (±7.26) (prospective)
risk of falling.
these tests is the best
TUG 10% 95% 20
predictor for falls
Verify the SEN and TUG 80% 100% 13.3
SPE of the TUG under TUG is a simple screening
conditions of one test. It is a sensitive
shumway-cook, Brauer e With falls: 78(±6). Six months TUGc 80% 93.3% 15
task and two tasks for 30 - - 15 _ 15 and specific measure of
Woolacot (2000)(28) No falls 86.2(±6) (retrospective)
identification of the probability for falls among
elderly who are prone TUGm 86.7% 93.3% 14.5 the elderly.
to falling

Determine if the
Tests predict the risk of falls in community-dwelling elderly

performance of the UG 78% 86% 8.5 TUG represents a valid

8 Up and Go Test can and effective method for
Rose, Jones e Luchese discriminate adults with With falls: 76.6(±6). One year discriminating possibility of
134 112 22 71 _ 63
(2002)(21) falls and without falls No falls: 78.4(± 6.5). (retrospective) TUG 71% 89% 10 falls occurrence in adults
and can be specific and with different ages, levels of
sensitive as a predictor activity and independence.

6
of falls as TUG BBS - - -

Show whether the use

of kinematic sensors TUG 58% 64.8% - The assessment of
in the body can be
Body Balance can be a
used to objectively to
With falls: 74 (±7,3); Five years fundamental strategy to
Greene et al 2010(24) quantify the TUG and 349 246 103 142 207 _
No falls: 70,1 (± 6,9) (retrospective) improve quality of life and
provide the amount of
can guide to fall prevention
analysis time, march
BBS 57.8% 64.2% strategies.
and stability time for
each TUG segment

Floor habil/
64% 100%
Transfer nãohabil
Several tests were able to
Identify the
CS-5 82% 82% 21s discriminate fallers and
performance measures
No falls: 71.2(±7.3) Fourtheen months not fallares, being the best
Murphy et al (2003) (29) that have the greatest 45 33 12 34 11
With falls: 79.6±6.5. (prospective) Habil/ predictors the test of “Time
potential to predict falls TST 55 94
não hábil of Rising of the Ground” and
in the elderly
“to Walk 15 meters”
POMA-B 55% 97% 12
FRT 73% 88% 20,3
Caption: TUG: Timed Up and Go test; TUGc: Timed Up and Go cognitive; TUGc: Timed Up and Go manual; BBS: Balance Berg Scale; UG: Get-Up and Go test; POMA-B: POMA Balance; POMA-G: POMA Gait; CS-5: Chair Stand Five Times
Test; ST: 30 s Sit to Stand Test; DGI: Dynamic Gait Index; FGA: Functional Gait Assessment; EMS: Elderly Mobility Scale; FSST: Four Step Square Test; OMO: Ordered Multi-Stepping Over Hoop; Functional Reach Test: FRT; SEN: Sensibility;
SPE: Specificity; CP: cutoff points.
MTP&RehabJournal 2016, 14: 432
 Table 4. Continued...
Sample History of fall
Author Objective Fall period Test SEN SPE CP Conclusion
Mean age /standard 2 or
N M H SQ 1Q
deviation +f

The aim of the study TUG 83% 96.6% 12.34

was to determine FGA seems to be able to
BBS - - -
Wrisley e Kumar (2010)) the concurrent, No falls: 71,5 (±7,2); six months predict falls and has good
(22) 35 18 17 29 _ 6
discriminatory and With falls: 77,7±7.8 (retrospective) DGI 100% 75.9% 19 concurrency with other
MTP&RehabJournal 2016, 14: 432

predictive validity of balance assessment tools.

FGA in the elderly FGA 100% 83% 20

To establish the
reliability and validity of FSST 89% 85% 15
a new Dynamic Clinical Multiples falls: The FSST was sensitive and
Balance Test (FSST). To 74±5,68; TUG 89% 63% 13 specific and had greater
six months
Dite et al. (2002)(20) assess the sensitivity, One fall: 81 _ _ 27 27 27 sensitivity and specificity in
(retrospective) ST 81% 67% 11S
specificity to identify 73,78±6.0No falls: relation to the 3 Tests with
subjects who fall and 74,14±6,07 which it was compared.
compare with 3 Tests of FRT
63% 59% 25cm
Balance and Mobility

Compare the Balance BBS 95,5% 76,5% 38

Scale of Berg and Two or more falls: POMA B-G 95,5% 95,5% 17 Although the fall is a
the Tinetti, Mobility (82,86±6.63) multifactorial event,
six months
Chiu, Au- Yeng, Lo. Scale in the elderly Only one fall: evaluations of this type
78 52 26 39 17 22 (retrospective)
(2003))(16) with the TUG in the (82,12±8,19) EMS 90,9% 95,5% 19 would require a lot of time

7
discrimination of No falls group: for application. Thus, BBS is a
elderly fallers and non- (81,86±7,82); strong predictor of falls.
fallers
TUG 77,3% 88,2% 24s.
Examine whether easily
measurable measures
of balance and muscle
strength can predict Measurements such as the
recurrent falls, as Tandem Stand and Manual
No falls or one fall:
well as, sophisticated One year Grip Strength can predict
Stel et al (2003)(26) (78,1±6,1)Two falls 435 238 197 333*; 102 TST 45% 74% 10s
measurements. (prospective) falls as well as sophisticated
or more (78,9±6,6)
Examine which of the measurements of body
modifiable risk factors balance.
were most strongly
associated with
recurrent falls.
Caption: TUG: Timed Up and Go test; TUGc: Timed Up and Go cognitive; TUGc: Timed Up and Go manual; BBS: Balance Berg Scale; UG: Get-Up and Go test; POMA-B: POMA Balance; POMA-G: POMA Gait; CS-5: Chair Stand Five Times
Test; ST: 30 s Sit to Stand Test; DGI: Dynamic Gait Index; FGA: Functional Gait Assessment; EMS: Elderly Mobility Scale; FSST: Four Step Square Test; OMO: Ordered Multi-Stepping Over Hoop; Functional Reach Test: FRT; SEN: Sensibility;
SPE: Specificity; CP: cutoff points.
Moreira ACSS et al.
 Table 4. Continued...

Sample History of fall

Author Objective Fall period Test SEN SPE CP Conclusion
Mean age /standard 2 or
N M H SQ 1Q
deviation +f

To investigate the
association of the
Ordered Multi Steeping
Over Hoop (OMO) No falls: The OMO is a good predictor
Tsutsumimoto et al. One year
with cognitive and (84,0±1,1) 59 49 10 45 14 OMO 86% 60% 21.9 for identifying fallers and not
(2013)(25) (prospective)
physical function and to With falls: (85,5±1,4) fallers.
investigate if the OMO
can predict incidences
of falls

To determine if a BBS 44,73% 73.62% 45

method based on
sensor data used in the
body can predict future
falls in the elderly The quantification of TUG
Tests predict the risk of falls in community-dwelling elderly

(27) in the community No falls 71,37 ±6,6 One year movements using body
Greene et al. (2012) 226 164 62 83 143
and compare the With falls: 71,75 ±6.9 (prospective) TUG 42,83% 62.62% 15.25s sensors can be an alternative
performance of the to evaluate future falls.
prediction of falls
in two standardized
methods of the same

8
dataset

Caption: TUG: Timed Up and Go test; TUGc: Timed Up and Go cognitive; TUGc: Timed Up and Go manual; BBS: Balance Berg Scale; UG: Get-Up and Go test; POMA-B: POMA Balance; POMA-G: POMA Gait; CS-5: Chair Stand Five Times
Test; ST: 30 s Sit to Stand Test; DGI: Dynamic Gait Index; FGA: Functional Gait Assessment; EMS: Elderly Mobility Scale; FSST: Four Step Square Test; OMO: Ordered Multi-Stepping Over Hoop; Functional Reach Test: FRT; SEN: Sensibility;
SPE: Specificity; CP: cutoff points.
MTP&RehabJournal 2016, 14: 432
MTP&RehabJournal 2016, 14: 432 Moreira ACSS et al.

Figure 2. Graph of the predictive validity analysis of the Timed Up and Go Test.

9
Tests predict the risk of falls in community-dwelling elderly MTP&RehabJournal 2016, 14: 432

ranged from 41.62 to 51.48; Multi-fallers 23,18 and; Non‑fallers DISCUSSION

from 48.59 to 54.00, with an average of 41.52 for fallers and This study aimed to analyze the predictive validity of
50.69 for non-ones. When analyzing the threshold effect the functional mobility tests to detect the risk of falls in
(liminal), Spearman’s correlation coefficient was equal to elderly residents in the community and independent., as
r = -0.800, indicating a strong association and not statistically well as, their methodological qualities in terms of design and
significant (p = 0.2). The post-test probability of an elderly procedures. The systematic review evidenced that the only
person suffering a fall or not, considered a positive and test with scientific discriminative validity for this population
negative odds ratio was 2.26 (X² = 10.52, p = 0.015) and 5.13 is the TUG, however it is not plausible to consider it as
(X² = 17.6; = 0.001), respectively. The SROC curve synthesizes “gold reference” test, since it does not present an adequate
the sensitivity and false positive values ​​( 1- specificity), predictive validity. The results of this study may have been
obtaining a value of 0.89 of the area under the curve (AUC) and influenced by the methodological quality criteria of the studies
Cochran Q of 0.82 as the highest common value of sensitivity that were not controlled by this study, although the majority
(72.7%) presented good methodological control according the
and specificity. The overall odds ratio of these studies had a
proposed criteria used here (See Table 3). It is noted, however,
value of 6.70 (95% CI: 1.75 - 23.88, X² = 18.72, p <0.001).
that there were major biases in the studies that might influence
Of the three studies that analyzed FRT, one did not find a
their results.
prospective difference between fallers and non-fallers, totaling
The review indicated that the mean TUG time of the fallers
two studies that reported sensitivity of the test. The combined
and non-fallers in all studies presented a significant difference,
sample size was composed of 126 elderly individuals aged suggesting that the test has the capacity to differentiate the
over 60 years were able to complete the scale. Of these fallers. In investigating the sensitivity and specificity of the test,
65 elderly people reported a history of a retrospective falls it was noted that the test has moderate diagnostic capacity.
with a time of six months and a prospective falls of 14 months. The diagnostic probability ratio (ORD) shows that the TUG
The sensitivity values were 64.6% (X² = 20.01, I² = 46%, has a good discriminative capacity to identify the fallers and
p <0.001) and specificity was 75.4% (X² = 1.88, I² = 14.9%, the non-fallers, discriminative capacity produced with the
p = 0.17). The post‑test probability of an elderly person falling studies included in this meta-analysis that proves that the
or non-falling, considered a positive and negative probability TUG is a good diagnostic test, with accuracy of 0.85 and the
coefficient ratio, was 2.89 and X² = 6.03 (p = 0.01) and 0.50 overall OR value of 8.4, besides of the Cronbach Q of 0.78.
and X² = 1.84 (p = 0.17), respectively. The overall odds ratio of However, the TUG cannot be considered with a predictive test
these studies had a value of 6.23 (X² = 4.49, p = 0.03). to predict falls in the elderly with precision, that is, cannot be
In the POMA Test, the combined sample size was composed considered a “gold standard test.” One of the reasons would
of 225 elderly individuals over 60 years old who were able be the heterogeneity found in the study, in which the test had
to complete the scale. Of these 41 elderly people reported a weak and non-significant correlation between sensitivity
a history of falls. Sensitivity values were 46.4% (X² = 6.20, and specificity, suggesting that there must be other factors
p = 0.01) and specificity was 85.8% (X² = 0.04, p = 0.83). resulting in variations in the estimated precision between the
The post-test probability of an elderly patient falling or not, individual studies. The variability presented in the test should
considered a positive and negative probability coefficient ratio, not be analyzed as an impediment to the analysis of the results,
was 3.76 and (X² = 3.34, p = 0.07), and 0.43 and X² = 3.94, but should be considered in careful evaluations when analyzing
the biases of publications.
p = 0.05), respectively. The overall odds ratio of these studies
The variability of the studies may be influencing the
had a value of 8.73 (X² = 4.13, p = 0.09).
results of the systematic review of the TUG Test, since of the
In the TS, the combined sample size was composed of
eight studies five are retrospective, impairing the criterion of
480 elderly subjects over the age of 60 years were able to
predictive validation of the test. In addition, studies conducted
complete the scale. Of these 113 elderly people reported
under different circumstances, differing in relation to the
a history of falls. Sensitivity values were 37% (X² = 5.12, criteria selection of the sample and the control of intrinsic
p = 0.02) and specificity was 82% (X² = 0.60, p = 0.44). factors that may influence the performance of the test,
The post-test probability of an elderly person suffering a fall indicating that even if we selected studies with the same type
or not, considered a ratio of positive and negative propensity of sample, the studies among themselves can present high
coefficient, was 2.96 (X² = 4.94, p <0.03) and 0.57 (X² = 2.96, homogeneity of subjects inside of samples. That is, not all the
p = 0.09), respectively. The overall odds ratio of these studies studies presented intrinsic and extrinsic control of the factors
was 5.47 (Q = 4.68, p = 0.03). that can have influencing the final result of the test. However,
Three studies using the CS-5 did not find a significant the problem with data homogeneity and inconsistent results,
difference between the fallers and non-fallers, as did the STST, when analyzed systematically, can provide important data on
SSST, CTSIB and OLS. test behavior in clinical practice where the population will be

10
MTP&RehabJournal 2016, 14: 432 Moreira ACSS et al.

varied. To test possible interactions of these variables and their balance and the predisposition to fall for elderly residents in
influence on accuracy it is necessary to carry out a study with the community.
this purpose, with an appropriate design. Three studies used the CS-5, none of which showed
At the BBS, the four studies found a significant difference discriminative ability to compare the mean between fallers
in mean score between the fallers and non-fallers, suggesting and non-fallers, so the studies did not assess the validity
that the test has the ability to differentiate the fallers. However, of the criterion. This test is widely used in the elderly as
of the four studies analyzed only three realized the predictive a simple test (40, 41, 42, 43, 44, 45) and in physical performance
validity of the instrument. The values ​​of sensitivity (58.7%) batteries(46), for indirect measurement of muscle resistance
and specificity (71.3%) calculated through the meeting of the of the knee joint - mainly quadriceps. The CS-5 has been
studies analyzed proved the high accuracy as a diagnostic test. demonstrated to be a simple and viable physical performance
The diagnostic odds ratio indicated had a good discriminative to request a decrease in daily living activity (ADL) and falls in
ability to identify the fallers and the non-fallers. However, the elderly(47, 48, 49). Nakazono et al.(50), concluded that CS-5
when analyzing the studies in detail, it can be observed and Chair Test (30 seconds) are simple and viable physical
that the results may have been biased in the meta-analysis. performance tests for the elderly and rehabilitation patients
Due to the fact that we found two studies (24,27) with the in the clinical context, to evaluate their muscular endurance.
same author in different period distances and design - one In addition to the other tests (360ºT, FSST, UG, Step Test Five
retrospective and the other prospective, both presented low Times, Step Test 15s, OMO Test), they presented a single study,
test sensitivity and moderate specificity. The other study is by indicated a coefficient of precision above of 0.80, suggesting
Chiu et al.(16) (2003), with retrospective case-control design, a probable excellent predictive validity, being considered
which analyzed between non-fallers and fallers that presented “gold standard” (51). Therefore, it is clinically important to
high sensitivity and specificity; And between non-fallers and develop studies that evaluate the predictive validity of these
tests to identify the risk of falls in this population.
multi-fallers that obtained excellent sensitivity and specificity
When analyzing the convergent validity of the tests in the
(95% in both). In this situation, perhaps the result of the
studies, the TUG presented greater frequency of convergent
study by Chiu et al.(16), may have influenced the results of the
validity with the other tests. The TUG Test showed weak
meta‑analysis performed here, thus not being able to conclude
correlation with CTSIB (on firm and unstable surface with
whether the instrument has good predictive validity.
closed eyes), moderate correlation with OLS, FRT and FGA;
Several studies have examined the validity of BBS in
and high correlation with the BBS, FSST and ST. In relation to
detecting the risk of falls in elderly residents in the community,
the convergent validity BBS presented moderate correlation
through factor analysis, ROC curve analysis, and sensitivity and
with FGA (r = 0.53) and high correlation with TUG (r = -0.76).
specificity in several languages(30, 31, 32, 33, 34, 35,). Studies point
It is understood that the Spearmam and Pearson correlation
to a Cronbach’s alpha ranging from 0.62 to 0.98; Sensitivity
tests will have a higher correlation index the closer to the
between 42% and 97%; Specificity between 26% and 92%; dimension of the construct between the tests. Thus, it is
Cut point between 45 and 50 points(10, 13, 30, 31, 32, 33, 34, 35, 30, 36, 37, 38, 39) . expected that the TUG presents a high correction between
Despite presenting a high alpha of Cronbach indicating a good the FSST, ST and BBS tests, since the TUG motor tasks are
internal consistency, as well as a good association between included or are similar to the motor tasks analyzed by the
item-factor and the other factors, this does not guarantee a FSST, ST and BBS.
good internal consistency in the measurement of the postural The systematic review has demonstrated the tendency of
control construct. In addition, several studies report that this the studies(22, 29, 22, 26) to publish only diagnostic tests with good
instrument has a “ceiling effect” (10, 11; 12, 13) and this reason is predictive validity, which leads to a scientific literature with
the justification used in the study by Wrisley and Kumar(22) did bias. Studies with negative results should be stimulated for
not analyze the predictive validity of this instrument, even their publications, with the intuition to offer to the scientific
finding that there was a significant difference between the environment the behavior of the tests in sample diversity,
fallers and non-fallers. In this sense, Souza et al.(39) reports protecting the researcher to use tests not advisable for its
that the BBS has no predictive validity in elderly patients population to be studied. Furthermore, in order to avoid
with knee osteoarthrosis, exposing some fragility in the test. this bias, studies should be designed with sufficient sample
However, the BBS was sensitive to individual differences in size of individuals for results to be credible and adequately
the sample, even in the absence of a good index of sensitivity disseminated. The accuracy of a diagnostic / predictive test
and specificity. is best determined by comparison with a standard reference
The POMA B and TS were selected only in two studies test, considered to be the gold standard. Currently, in the
each, which did not allow an in-depth analysis in the tests. literature, the standard reference test to predict falls in
The tests presented low sensitivity, but it does not clarify the the elderly has not yet been clarified. Even so, the review
aspect regarding the validity of the tests to detect altered proposed to select studies that presented a reference test

11
Tests predict the risk of falls in community-dwelling elderly MTP&RehabJournal 2016, 14: 432

(adjusted test), that is, that verified the predictive validity of is suggested that the TUG test can be used in association with
the target test and another test. The presence of a reference other predictors of fall risk to provide additional information
test reduces sampling rates in studies with non-representative on the identification of the elderly with the potential to suffer
sample size, especially in a control case study that the future falls.
number of individuals with the phenomenon to be studied
is equal to those that do not present what corresponds to a AUTHOR’S CONTRIBUTION
50% prevalence of phenomenon, so that the researcher can ACSSM: Systematic review and writing of the article; PBP: Systematic review
and writing of the article; EH: Systematic review; ECM: Systematic review; GZM:
evaluate the usefulness of the test according to its reality in Correction of the manuscript; FLC: Correction and translation of the manuscript.
clinical practice, that is, ascertain if the test underestimates or
overestimates the ability of the test to detect the phenomenon CONFLICTS OF INTEREST
in the population. The authors declare that they have no conflicts of interest in the research.
It is not possible with a single clinical trial to predict the
risk of falling in the elderly because the fall is a multifactorial AUTHOR DETAILS
phenomenon. However, it is essential to have the scientific 2. Professor in Graduate Program in Human Motion Sciences, Santa Catarina
State University (UDESC), Santa Catarina (SC), Brazil. 3. Master in Human
knowledge which discriminatory clinical tests are valid for
Motion Sciences, Santa Catarina State University (UDESC), Santa Catarina
the population profile to be studied. The systematic review (SC), Brazil. 4. Professor in Graduate Program in Human Motion Sciences and
pointed out that the only test with scientific discriminative Graduate Program on Education, Santa Catarina State University (UDESC),
validity for elderly residents in the community is the TUG, but Santa Catarina (SC), Brazil.
it is not possible to consider it as a gold reference test because
it does not present an adequate predictive validity. TUG may REFERENCES
not be sufficient as a simplified screening tool to detect the
1. Silvestre JA, Costa Neto MM. Abordagem do idoso em programas de
risk of falls, we suggest the use of TUG associated with other saúde da família. Cad Saúde Pública. 2003;19(3):839-47.
predictors of fall risk to provide additional information on the 2. Perracini MR, Ramos LR. Fatores associados a quedas em uma coorte de
identification of the elderly with the potential to suffer future idosos residentes na comunidade. Rev Saúde Pública. 2002; 36(6): 709-16.
falls. The meta‑analysis of fall risk factors points to fear of 3. Datasus. Banco de dados do Sistema Único de Saúde. 2010
falls, dementia, vertigo, gait deficiency, antiepileptic drugs, 4. Aschkenasy MT, Rothenhaus TC. Trauma and falls in the elderly. Emerg
muscle weakness, balance deficiency and environmental risk Med Clin North Am, 2006; 24: 413-32.
are predictors of falls in community-dwelling elderly people(52). 5. Englander F. Economic dimensions of slip and fall injuries. J Forensic Sci.
It is suggested that longitudinal studies be conducted in to 1996; 41(5): 733-46.

determine the causes and risk factors for falls in the elderly, 6. Zuckerman D. Hype in health reporting: “checkbook science” buys
distortion of medical news. Int. J. Health Serv. 2003; 33(2): 383-89.
verify the validity of the contents of functional mobility tests
7. Georg AE, Duncan BB, Toscano CM, Schmidt MI, Mengue S, Duarte C,
to detect the risk of falls. Besides that it is suggested studies to Polanczyk CA. Análise econômica de programa para rastreamento do
determine if the proposed tests and risk factors have predictive diabetes mellitus no Brasil. R. Saúde Públ. 2005; 39(3): 452-60.
validity over time to detect the risk of falls in the elderly and 8. Pasquali L. Instrumentação psicológica: Fundamentos e práticas. Porto
studies to monitor risk factors and functional mobility tests Alegre, Brasil: Artmed, 2010.
remain effective over time. 9. Pasquali L. Psicometria. Rev. Esc. Enferm. 2009; 43: 992-99.
10. Newton R. Balance screening of an inner city older adult population. Arch
Phys Med Rehabil. 1997; 78: 587-91.
CONCLUSION
11. Muir SW; Berg K; Chesworth B; SPEECHLEY M. Use of the Berg Balance
Although the systematic review evidenced that the studies
Scale for predicting multiple falls in community-dwelling elderly people:
present good methodological quality, they present great biases a prospective study. Phys Ther. 2008; 88(4): 449-59.
that compromised the result of the systematic review, such as: 12. Santos GM, Souza ACS, Virtuoso JF, Tavares GMS, Mazo GZ. Valores
Types of studies found - many studies presented a retrospective preditivos para o risco de quedas em idosos praticantes e não praticantes
design compromising the predictive validity of the test; Non- de atividade física por meio do uso da Escala de Equilíbrio de Berg. Braz
J Phys Ther. 2011, 15(2): 95-101.
representative sample size; Selection criteria; Control of intrinsic
13. Alzayer L, Beninato M, Portney LG. The accuracy of individual Berg Balance
and extrinsic factors; Concurrent validity of the tests to ascertain
Scale items compared with the total Berg score for classifying people with
the predictive validity of the target test and of another test; chronic stroke according to fall history. J Neurol Phys Ther. 2009, 33(3):
Besides other biases. 136-43.
Although the studies presented important biases, it can 14. Morris R, Harwood RH, Baker R, Sahota O. Armstrong S Masud T. A
be evidenced that the only test with discriminative scientific comparison of different balance tests in the prediction of falls in older
women with vertebral fractures: a cohort study. Age Ageing. 2007, 36(1):
validity for elderly residents in the community is the TUG test, 78-83.
but it cannot be considered as “gold reference test”, since its
15. Thrane G, Joakimsen RM, Thornquist E. The association between timed
predictive validity does not allow for this purpose. Considering up and go test and history of falls: the Tromsø study. BMC Geriatr 2007,
that the phenomenon of fall in the elderly is multifactorial, it 12(7): 1.

12
MTP&RehabJournal 2016, 14: 432 Moreira ACSS et al.

16. Chiu AY, Au-Yeung SS, Lo SK. A comparison of four functional tests in 36. Muir SW, Berg K, Chesworth B. Balance impairment as a risk factor
discriminating fallers from non-fallers in older people. Disabil Rehabil. forfalls in community-dwelling older adults Who are high functionin: a
2003, 25(1): 45-50. prospective study Phys Ther. 2010, 90(3): 338-47.
17. Pardasaney PK, Latham NK, Jette AM, Wagenaar RC, Ni P, Slavin MD, Bean
37. Holbein-Jenny MA, Billek-Sawhney B, Beckman E, Smith T. Balance in
JF. Sensitivity to change and responsiveness of four balance measures for
community-dwelling older adults. Phys Ther. 2012, 92(3): 388-97. personal care home residents: a comparison of the Berg Balance Scale,
18. Farrell MK, Rutt RA, Lusardi MM, Williams AK. Are scores on the physical the Multi-Directional Reach Test, and the Activities-Specific Balance
performance test useful in determination of risk of future falls in Confidence Scale J Geriatr Phys Ther. 2005, 28(2): 48-53.
individuals with dementia? J Geriatr Phys Ther. 2011, 34(2): p57-63. 38. Riddle DL, Stratford PW. Interpreting validity indexes for diagnostic tests:
19. Oliveira MRF, Gomes AC, Toscano C M. QUADAS e STARD: Avaliação an illustration using the Berg balance test. Phys Ther. 1999, 79(10); 939-48.
da qualidade de estudos de acurácia de testes diagnósticos Rev Saude
Publica, 2011, 45(2): 416-22. 39. Souza ACS, Santos GM. Sensibilidade da Escala de Equilíbrio de Berg em
indivíduos com osteoartrite Motriz: Rev Educ Fis. 2012, 18(2): 307-18.
20. Dite W, Temple VA. A clinical test of stepping and change of direction to
identify multiple falling older adults Arch Phys Med Rehabil. 2002, 83(11): 40. Reider NC, Naylor PJ, Gaul C Sensitivity and specificity of the minimal chair
1566-71. height standing ability test: a simple and affordable fall-risk screening
21. Rose DJ, Lucchese N, Wiersma LD. Development of a multidimensional instrument. J Geriatr Phys Ther. 2015, 38(2):90-5.
balance scale for use with functionally independent older adults Arch
41. Hirase T, Inokuchi S, Matsusaka N, Nakahara K, Okita M. A modified fall
Phys Med Rehabil. 2006, 87(11): 1478-85.
risk assessment tool that is specific to physical function predicts falls in
22. Wrisley DM, Kumar NA. Functional gait assessment: concurrent,
discriminative, and predictive validity in community-dwelling older adults. community-dwelling elderly people. J Geriatr Phys Ther. 2014, 37(4):
Phys Ther. 2010, 90(5): 761-73. 159-65.
23. Trueblood PR, Hodson-Chennault N, McCubbin A, Youngclarke D. 42. Buatois S, Perret GC, Gueguen R, Miget P, Vançon G, Perrin P, Benetos
Performance and impairment-based assessments among community A. A simple clinical scale to stratify risk of recurrent falls in community-
dwelling elderly: sensitivity and specificity. Aging (Albany NY) 2001, dwelling adults aged 65 years and older. Phys Ther. 2010, 90(4); 550-60.
24(1):2–6
43. Tiedemann A, Shimada H, Sherrington C, Murray S, Lord S. The
24. Greene BR, O’Donovan A, Romero-Ortuno R. Quantitative falls risk
assessment using the timed up and go test. Biomed Tech (Berl) 2010, comparative ability of eight functional mobility tests for predicting falls
57(12): 2918-26. in community -dwelling older people Age Ageing. 2008, 37(4): 430-35.
25. Tsutsumimoto K, Doi T, Misu S, Ono R, Hirata S. Can the Ordered Multi- 44. Buatois S, Miljkovic D, Manckoundia P, Gueguen R, Miget P, Vançon G,
Stepping Over Hoop test be useful for predicting fallers among older Perrin P, Benetos A. Five times sit to stand test is a predictor of recurrent
people? A preliminary 1 year cohort study Aging Clin Exp Res. 2013, 25(4):
falls in healthy community-living subjects aged 65 and older. J Am Geriatr
427-32.
Soc. 2008, 56(8):1575-77.
26. Stel VS, Smit JH, Plujim SMF. Balance and mobility performance as
treatable risk factors for recurrent falling in older persons. J Clin Epidemiol. 45. Schaubert KL, Bohannon RW. Reliability and validity of three strength
2003, 56(7): 659-68. measures obtained from community-dwelling elderly persons J Strength
27. Greene BR, Doheny EP, Walsh C. Evaluation of falls risk in community- Cond Res. 2005, 19(3): 717-20.
dwelling older adults using body-worn sensors. Gerontology 2012, 58(5): 46. Guralnik JM, Simonsick EM, Ferrucci L, Glynn RJ, Berkman LF, Blazer DG,
472-80.
Scherr PA, Wallace RB. A short physical performance battery assessing
28. Shumway-Cook A, Brauer S, et al. Predicting the probability for falls in
lower extremity function: association with self-reported disability and
community-dwelling older adults using the Timed Up & Go Test. Phys
Ther. 2000, 80(9): 896-903. prediction of mortality and nursing home admission. J Gerontol. 1994,
49(2): M85-94.
29. Murphy SL, Dubin JA, Gill TM. The development of fear of falling among
community-living older women: predisposing factors and subsequent fall 47. Ward RE, Leveille SG, Beauchamp MK, Travison T, Alexander N, Jette AM,
events. J Gerontol A Biol Sci Med Sci. 2003, 58(10): M943-M947. Bean JF. Functional performance as a predictor of injurious falls in older
30. Matsushima M, Yabe I, Uwatoko H, Shirai S, Hirotani M, Sasaki H. Reliability adults. J Am Geriatr Soc. 2015, 63(2): 315-20.
of the Japanese version of the Berg balance scale. Intern Med. 2014,
48. Cheng YY, Wei SH, Chen PY, Tsai MW, Cheng IC, Liu DH, Kao CL. Can sit-
53(15): 1621-4.
to-stand lower limb muscle power predict fall status? Gait Posture 2014,
31. Salavati M, Negahban H, Mazaheri M, Soleimanifar M, Hadadi M,
Sefiddashti L, Zahraee MH, Davatgaran K, Feizi A. The Persian version 40(3): 403-7.
of the Berg Balance Scale: inter and intra-rater reliability and construct 49. Zhang M, Kim JC, Li Y, Shapiro BB, Porszasz J, Bross R, Feroze U, Upreti
validity in elderly adults Disabil Rehabil. 2012, 34(20): 1695-98. R, Martin D, Kalantar-Zadeh K, Kopple JD. Relation between anxiety,
32. De Figueiredo KM, De Lima KC, Cavalcanti MAC, Guerra RO. Interobserver depression, and physical activity and performance in maintenance
reproducibility of the Berg Balance Scale by novice and experienced hemodialysis patients J Ren Nutr. 2014, 24(4): 252-60.
physiotherapists Physiother Theory Pract. 2009, 25(1):30-6
50. Nakazono T, Kamide N, Ando M. The Reference Values for the Chair Stand
33. Sahin F, Yilmaz F, Ozmaden A, Kotevolu N, Sahin T, Kuran B. Reliability and
validity of the Turkish version of the Berg Balance Scale J Geriatr Phys Test in Healthy Japanese Older People: Determination by Meta-analysis.
Ther. 2008, 31(1): 32-37. J Phys Ther Sci. 2014, 26(11):1729-31.
34. Wang CY, Hsieh CL, Olson SL, Wang CH, Sheu CF, Liang CC. Psychometric 51. Noronha APP, Alchieri JC. Reflexões sobre os instrumentos de avaliação
properties of the Berg Balance Scale in a community-dwelling elderly psicológica Em R Primi (Org) Temas em avaliação psicológica. Campinas:
resident population in Taiwan. J Formos Med Assoc. 2006, 105(12): 992-
Impressão Digital do Brasil 2002.
1000.
35. Miyamoto ST, Lombardi Junior I, Berg KO, Ramos LR, Natour J. Brazilian 52. Deandrea S, Lucenteforte E, Bravi F, Foschi R, La Vecchia C, Negri E. Risk
version of the Berg balance scale Braz J Med Biol Res. 2004, 37(9): 1411- factors for falls in community-dwelling older people: a systematic review
21. and meta-analysis. Epidemiology 2010, 21(5): 658-68.

13
Tests predict the risk of falls in community-dwelling elderly MTP&RehabJournal 2016, 14: 432

APPENDIX I. Search terms MEDLINE, SCOPUS and CINAHL.

MEDLINE Terms SCOPUS Terms CINAHL Terms

BLOCK ELDERLY: BLOCK ELDERLY BLOCK ELDERLY:
1 Aged/ Aged S1 (MH “Aged”)
2 agetw Age S2 TX Age
3 elderlytw Elderly S3 TX Elderly
4 oldertw Older S4 TX Older
5 (old* adj2 people)tw old* PRE/2people S5 TX (Old* n2 People)
6 (old* adj2 adult*)tw old* PRE/2 adult S6 TX (Old* n2 adult*)
7 aging/ aging S7 (MH “Aging”)
8 or/1-7 S8 S1 or S2 or S3 or S4 or S5 or S6 or S7
BLOCK STUDY TYPE BLOCK STUDY TYPE BLOCK STUDY TYPE
9 exp “Sensitivity and Specificity”/ “Sensitivity and Specificity” S9 (MH “Sensitivity and Specificity”)
10 sensitivitytw Sensitivity S10 TX Sensitivity
11 specificitytw specificity S11 TX Specificity
12 ((pre-test or pretest) adj probability)tw ((pre-test OR pretest) PRE/2 probability) S12 TX Post-test probability
13 post-test probabilitytw predictive value$ S13 (MH “Predictive Value of Tests”)
14 predictive value$tw likelihood ratio$ S14 TX (likelihood ratio*)
15 likelihood ratio$tw post-test probability S15 S9 or S10 or S11 or S12 or S13 or S14
16 or/9-15
BLOCK BALANCE BLOCK BALANCE BLOCK BALANCE
17 (control adj2 postur*)tw postural balance S16 TX (Control n2 postur*)
18 equilibriumtw postur* PRE/2 balance S17TX (Equilibrium)
19 (postur* adj2 equilibr*)tw balance S18 TX (Postur* n2 Equilibr*)
20 balancetw postur* PRE/2 equilibr* S19 TX Balance
21 (postur* adj2 balance)tw equilibrium S20 TX Balance, Postural
22 Postural Balance/ control PRE/2 postur* S21 S16 or S17 or S18 or S19 or S20
23 or/17-22

14