Content uploaded by Andrea Tricco
Author content
All content in this area was uploaded by Andrea Tricco on Nov 27, 2017
Content may be subject to copyright.
Comparisons of Interventions
for Preventing Falls in Older Adults
A Systematic Review and Meta-analysis
Andrea C. Tricco, PhD; Sonia M. Thomas,MSc; Are ti AngelikiVeroniki, PhD; Jemila S. Hamid, PhD; Elise Cogo, ND;
Lisa Strifler, MSc; Paul A. Khan, PhD; Reid Robson, MSc; Kathryn M. Sibley, PhD; Heather MacDonald, MSc;
John J. Riva, DC; Kednapa Thavorn, PhD; Charlotte Wilson, MSc; Jayna Holroyd-Leduc, MD; Gillian D. Kerr, MD;
Fabio Feldman, PhD; Sumit R. Majumdar, MD; Susan B. Jaglal, PhD; Wing Hui, MSc; Sharon E. Straus, MD, MSc
IMPORTANCE Falls result in substantial burden for patients and health care systems, and given
the aging of the population worldwide, the incidence of falls continues to rise.
OBJECTIVE To assess the potential effectiveness of interventions for preventing falls.
DATA SOURCES MEDLINE, Embase, Cochrane Central Register of Controlled Trials,
and Ageline databases from inception until April 2017. Reference lists of included studies
were scanned.
STUDY SELECTION Randomized clinical trials (RCTs) of fall-prevention interventions for
participants aged 65 years and older.
DATA EXTRACTION AND SYNTHESIS Pairs of reviewers independently screened the studies,
abstracted data, and appraised risk of bias. Pairwise meta-analysis and network meta-analysis
were conducted.
MAIN OUTCOMES AND MEASURES Injurious falls and fall-related hospitalizations.
RESULTS A total of 283 RCTs (159 910 participants; mean age, 78.1 years; 74% women) were
included after screening of 10 650 titles and abstracts and 1210 full-text articles. Network
meta-analysis (including 54 RCTs, 41596 participants, 39 interventions plus usual care)
suggested that the following interventions, when compared with usual care, were associated
with reductions in injurious falls: exercise (odds ratio [OR], 0.51 [95% CI, 0.33 to 0.79];
absolute risk difference [ARD], −0.67 [95% CI, −1.10 to −0.24]); combined exercise and vision
assessment and treatment (OR, 0.17 [95% CI, 0.07 to 0.38]; ARD, −1.79 [95% CI, −2.63 to
−0.96]); combined exercise, vision assessment and treatment, and environmental
assessment and modification (OR, 0.30 [95% CI, 0.13 to 0.70]; ARD, −1.19 [95% CI, −2.04 to
−0.35]); and combined clinic-level quality improvement strategies (eg, case management),
multifactorial assessment and treatment (eg, comprehensive geriatric assessment), calcium
supplementation, and vitamin D supplementation (OR, 0.12 [95% CI, 0.03 to 0.55]; ARD,
−2.08 [95% CI, −3.56 to −0.60]). Pairwise meta-analyses for fall-related hospitalizations
(2 RCTs; 516 participants) showed no significant association between combined clinic- and
patient-level quality improvement strategies and multifactorial assessment and treatment
relative to usual care (OR, 0.78 [95% CI, 0.33 to 1.81]).
CONCLUSIONS AND RELEVANCE Exercise alone and various combinations of interventions
were associated with lower risk of injurious falls compared with usual care. Choice of
fall-prevention intervention may depend on patient and caregiver values and preferences.
JAMA. 2017;318(17):1687-1699. doi:10.1001/jama.2017.15006
Editorial page 1659
Supplemental content
CME Quiz at
jamanetwork.com/learning
and CME Questions page 1706
Author Affiliations: Author
affiliations are listed at the end of this
article.
Corresponding Author: Sharon E.
Straus, MD, MSc, Director,
Knowledge TranslationProgram,
Li Ka Shing Knowledge Institute,
St Michael’s Hospital, 209 Victoria St,
East Bldg, Room 716, Toronto, ON,
M5B 1T8, Canada (sharon.straus
@utoronto.ca).
Research
JAMA | Original Investigation
(Reprinted) 1687
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
Data from the National Institute on Aging showed
that the 2-year prevalence of falls among individuals
aged 65 years or older was 36% in 2010.
1
Falls cause
a substantial burden to patients and health care systems.
2-4
In 2012, the cost of falls to Medicare was $31 billion.
5
Not
only can falls result in serious injury or death,
6,7
but older
adults who experience falls also report increased anxiety
and depression and reduced quality of life.
8,9
Given the
aging of the population worldwide, the incidence of falls is
expected to continue rising.
5
As such, preventing falls
among older persons is increasingly important for health
care systems.
Previous randomized clinical trials (RCTs) and system-
atic reviews have selectively examined fall-prevention
programs.
10-13
However, directly comparing more than 2
interventions using conventional meta-analysis has major
limitations. The key elements of an effective fall-prevention
program remain unclear, which has hampered implementa-
tion of effective interventions. Furthermore, a network
meta-analysis ranking all available fall-prevention interven-
tions and their combinations has not been conducted.
Therefore, a systematic review and network meta-analysis
on all available fall-prevention interventions for older
people were conducted.
Methods
Protocol
The systematic review protocol was developed using
guidance from the Preferred Reporting Items for System-
atic Review and Meta-analysis Protocols (PRISMA-P)
statement,
14
registered in PROSPERO (CRD42013004151),
and published.
15
Because the methods were reported
previously, they are described only briefly here. The
PRISMA-network meta-analysis extension
16
was used to
report results.
Eligibility Criteria
All types of RCTs (eg, cluster, crossover) examining
fall-prevention interventions (whether single or multifacto-
rial) for adults aged 65 years or older in all settings (eg, com-
munity, acute care) were included. Potential compar-
ators were usual care, other fall-prevention interventions,
and placebo.
Outcomes
The primary outcomes were the numbers of injurious falls
and fall-related hospitalizations. The secondary outcomes
were rate of falls, number of fallers, number of fall-related
emergency department visits, number of fall-related physi-
cian visits, number of fractures, costs (eg, to the health care
system), number of intervention-related harms (eg, muscle
soreness from exercise), and quality of life. Quality of
life was measured with the SF-12 or SF-36 physical and men-
tal summary component measures (range, 0-100) or the
EuroQol-5D VAS (range, 0-100), where 0 indicates maxi-
mum disability, and 100 indicates no disability.
17-19
Data Sources
MEDLINE, Embase, Cochrane Central Register of Controlled
Trials, and Ageline databases were searched from inception
until December 1, 2015 (see protocol for search strategy
15
).
Reference lists of included RCTs and relevant reviews were
scanned for additional RCTs. Study authors were contacted
for unpublished studies or additional data. An updated
search was conducted on April 19, 2017, which involved
screening, abstraction, and risk-of-bias appraisal by 2 review-
ers, working independently without additional reference
scanning or author contact regarding conference abstracts,
trial protocols, or non-English articles for any studies identi-
fied in the update.
Study Selection
After pilot-testing eligibility criteria for citations and full-text
articles, screening was conducted independently by pairs
of reviewers. Conflicts were resolved by a third reviewer.
Data Abstraction
Data abstraction was completed by independent pairs
of reviewers after pilot-testing of the data abstraction
form. Conflicts were resolved by a third reviewer. Interven-
tions were coded independently by a clinician (S.E.S.) and
a methodologist (A.C.T.) using a preestablished coding
guide (eTable 1 in the Supplement). Included interventions
were classified into the following broad categories: basic
falls risk assessment, calcium supplementation, cog-
nitive behavioral therapy, devices, diet modification,
electromagnetic field therapy and whole-body vibration,
environmental assessment and modification, exercise, floor
modifications, multifactorial assessment and treatment,
osteoporosis medications, podiatry assessment and treat-
ment, quality improvement strategies, social engagement,
surgery, vision assessment and treatment, and vitamin D
supplementation. Quality improvement strategies were
focused on increasing use of research in practice and were
classified at the health system, clinic and clinician, and
patient levels (Box 1 and Table 1).
20
Key Points
Question What types of fall-prevention programs may be
effective for reducing injurious falls in older people?
Findings In a network meta-analysis including 54 studies and
41 596 participants, exercise (odds ratio [OR], 0.51), combined
exercise, vision assessment and treatment, and environmental
assessment and modification (OR, 0.30), combined exercise,
and vision assessment and treatment (OR, 0.17), and combined
clinic-level quality-improvement strategies, multifactorial
assessment and treatment, calcium supplementation,
and vitamin D supplementation (OR, 0.12)were signif icantly
associated with reductions in injurious falls.
Meaning The analysis identified combinations of interventions
likely to be more effective than usual care for preventing
injurious falls.
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1688 JAMA November 7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
Risk-of-Bias Appraisal
The Cochrane Effective Practice and Organisation of Care
(EPOC) Group’s risk-of-bias tool was used to appraise in-
cluded studies.
21
This appraisal was conducted by pairs
of independent reviewers, with conflicts resolved by a third
reviewer. Small study effects were assessed for each out-
come (when >10 RCTs were available) using the comparison-
adjusted funnel plot and the netfunnel command (Stata 13.0)
22
with a fixed meta-analysis model and ordering treatmentsfrom
most to least effective according to clinical insight.
Data Synthesis and Analysis
Across all outcomes, pairwise random-effects meta-analysis
was conducted. Effect estimates are reported as odds ratios
(ORs) for dichotomous outcomes and mean differences for
continuous outcomes. Studies reporting dichotomous out-
comes with zero events across all groups were included in
the systematic review, but excluded from analysis. Studies
reporting continuous outcomes with the average effect esti-
mate but not reporting the associated measure of variance
(eg, standard deviation) were included in the analysis, with
standard errors imputed when feasible.
23,24
Random-effects network meta-analyses were con-
ducted for connected networks of included RCTs when
more than 10 RCTs were available and the number of RCTs
was greater than the number of intervention nodes. To sur-
mount small study effects,
25
smaller trials (n <100 partici-
pants) were excluded from analysis. Across all outcomes for
which network meta-analysis was possible, the transitivity
and consistency assumptions were assessed a priori.
26-28
In both network meta-analysis and inconsistency models,
common within-network, between-study variance across
intervention comparisons was assumed because the treat-
ments included in each network of trials were mostly
nonpharmacological. Subgroup and sensitivity network
meta-analyses were conducted for the primary outcome
with consideration of potential treatment effect modifiers.
Interventions were ranked using P scores
29
and presented in
a rank-heat plot.
30
For each network meta-analysis, the
overall risk for the control group (considering usual care
as the control) of the included studies was calculated via
the variance-stabilizing Freeman-Tukey double arcsine
approach.
31
A random-effects pairwise meta-analysis was
applied using inverse variance weights, and to facilitate
interpretation, summary group risks were back-transformed
to the initial scale. All network meta-analyses and consis-
tency assessments were conducted with R software (version
3.3.3)
32
using the netmeta package.
33
Results were summa-
rized using effect estimates (ORs or mean differences) and
their associated 95% CIs. Overall ORs, derived from each
network meta-analysis, were transformed to risk differ-
ences to allow judgment of the clinical importance of statis-
tically significant results.
34
Analysis details are provided in
eMethods in the Supplement.
Results
Study Selection
A total of 10 650 titles and abstracts and subsequently
1210 full-text articles were screened (Figure 1). Across
all outcomes, 283 RCTs and 20 companion reports were
Box 1. Components of Quality Improvement Strategies Classified
by Level of Implementation
Clinic Level
Quality improvement initiatives targeting the clinic or care team
include case management, team changes, electronic patient
registries, facilitated relay of information to clinicians, continuous
quality improvement, audit and feedback, staff education,
and clinician reminders
Patient Level
Quality improvement initiatives targeting the patient include
promotion of self-management, patient education, patient
reminders, and motivational interviewing
Health System Level
Quality improvement initiatives targeting the health system
include interventions with positive or negative financial incentives
directed at clinicians (eg, linked to adherence to some process of
care or achievement of some target outcome), positive or negative
financial incentives directed at patients, or system-wide changes
in reimbursement systems
Table 1. Components of Interventions to Prevent Falls
Intervention Component Abbreviation
Basic falls risk assessment bf
Calcium ca
Cognitive behavioral therapy cb
Clinic-level quality improvement cl-qi
Comprehensive podiatry assessment
and treatment
cp
Device—alarm de-al
Device—hip protector de-hp
Device—orthosis de-or
Dietary modifcations di
Environmental assessment
and modification
ea
Electromagnetic field therapy
and whole-body vibration
em+wb
Exercise ex
Flooring fl
Lavender la
Multifactorial assessment and treatment mf
Osteoporosis treatment op-tx
Patient-level quality improvement pa-qi
Social engagement so
Surgery—cataract su-ey
Surgery—hip su-hi
Surgery—pacemaker su-pm
Health system–level quality improvement sy-qi
Usual care uc
Vision assessment and treatment va
Vitamin D vi-d
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1689
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
included (citations in eReferences in the Supplement),
with a total of 159 910 participants (mean age 78.1 years;
74% women). Two included studies were available only as
conference abstracts,
35,36
and 14 other studies were identi-
fied by scanning reference lists of included studies and rel-
evant reviews.
37-50
Author Contact
Fifty-four authors were contacted, and responses were re-
ceived from 35 (response rate, 65%), which allowed inclusion
of 2 additional studies.
Study and Participant Characteristics
For 155 of the RCTs (54.8%), the mean age of participants
was between 74 and 84 years (Table 2; eTable 2 in the
Supplement). In 248 of the RCTs (87.6%), at least 50% of the
participants were women. One hundred sixty-nine RCTs
(59.7%) included a mixture of individuals with and without
a history of falls. The number of medications taken was not
reported in 172 RCTs (60.8%).
Two hundred fifty-five of the RCTs (90.1%) were pub-
lished in 2001 or later, and the studies were conducted in
Europe (121 [42.8%]), Australia or New Zealand (66 [23.3%]),
and other regions (Table 3; eTable 3 in the Supplement).
The setting for 142 studies (50.2%) was at home, 75 were in
a clinic ([26.5%]), and 72 were in the community (25.4%),
with some studies including multiple settings. One hundred
and fifty-one RCTs (53.4%) had an intervention duration
of 26 weeks or less, and 223 of the studies (78.8%) had
a duration (ie, from recruitment to the end of follow-up)
of 1 year or less.
Risk of Bias
Most RCTs had a low risk of bias for random sequence gen-
eration (187 [66.1%]), similar baseline outcome measures (212
[74.9%]), similar baseline characteristics (238 [84.1%]),
incomplete outcome data (221 (78.1%]), blinding (267
[94.3%]), and other components of bias (171 [60.4%]) (eTable
4 and eFigure 1 in the Supplement). However, a high propor-
tion had unclear risk of bias for allocation concealment
Table 2. Summary of Patient Characteristics
Characteristic
No. (%) of Randomized Clinical Trials
(N = 283)
Age, mean, y
64-73.9 56 (19.8)
74-83.9 155 (54.8)
≥84 36 (12.7)
Not reported 36 (12.7)
% Women
0-49.9 20 (7.1)
50-100 248 (87.6)
Not reported 15 (5.3)
History of falls
All 60 (21.2)
Mixed 169 (59.7)
None 0
Not reported 54 (19.1)
No. of medications taken
a
0-4 56 (19.8)
≥5 55 (19.4)
Not reported 172 (60.8)
a
Refers to the reported mean or median number of medications taken
by patients—not necessarily the number of medications taken daily.
Figure 1. Study Flow From LiteratureSearch
9454 Excluded
9191 Not a falls-prevention
intervention
3No relevant comparator
233 Not an RCT
27 Participants aged <65 y
907 Excluded
266 No relevant outcome
104 No abstractable data
99 Participants aged <65 y
35 Duplicate article
17 No relevant comparator
224 Not an RCT
162 Not a falls-prevention intervention
1210 Full-text articles assessed for eligibility
303 Studies included in meta-analysis
283 Primary publications
20 Companion reportsa
10
650 Records identified and screened
through database search
14 Articles identified from scanning
of reference lists
RCT indicates randomized
clinical trial.
a
If multiple publications reported
data from the same trial
(eg, reporting results at 1-year vs
2-year follow-up), the first was
identified as the primary publication
and any additional publications
were referred to as companion
reports to avoid double counting
data from the same trial.
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1690 JAMA November7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
(145 [51.2%]), contamination (190 [67.1%]), and selective out-
come reporting (183 [64.7%]). All comparison-adjusted fun-
nel plots suggested no evidence of publication bias (eFigure 2
in the Supplement).
Statistical Analysis
The design-by-treatment interaction model showed no evi-
dence of significant inconsistency across the network meta-
analysis (eTable 5 in the Supplement). Because of the large
number of results from the analysis, the overall results for
each outcome are presented and focus on statistically signifi-
cant intervention effects relative to usual care for network
meta-analyses (summarized in Box 2). The results from all
statistically significant treatment comparisons are available
in eTable 5 in the Supplement, and the results from all analy-
ses are posted on the Open Science Framework.
50
Pscores
were used to summarize results for the primary outcome,
and all results are presented in eTable 6, eTable 7, and eTable
8intheSupplement. The rank-heat plot, presented in eFig-
ure 3 in the Supplement, indicates that exercise is likely the
most effective intervention in terms of numbers of fallers,
injurious falls, fractures, and hip fractures.
Injurious Falls
Network meta-analysis for the primary outcome of injurious
falls included 54 RCTs (41 596 participants) with 39 interven-
tions plus usual care (Figure 2). The event rate for injurious
falls in the usual care group was 0.34 (95% CI, 0.24 to 0.44).
Across all 780 network meta-analysis comparisons, 101
(12.9%) were statistically significant (eTable 5 in the Supple-
ment). The following 4 interventions were associated with a
reduced risk of injurious falls relative to usual care:
•Exercise (OR, 0.51 [95% CI, 0.33 to 0.79]; absolute risk dif-
ference [ARD], −0.67 [95% CI, −1.10 to −0.24])
•Combined exercise and vision assessment and treatment
(OR, 0.17 [95% CI, 0.07 to 0.38]; ARD, −1.79 [95% CI, −2.63
to −0.96])
•Combined exercise, vision assessment and treatment, and en-
vironmental assessment and modification (OR, 0.30 [95% CI,
0.13 to 0.70]; ARD, −1.19 [95% CI, −2.04 to −0.35])
•Combined clinic-level quality improvement strategies,mul-
tifactorial assessment and treatment, calcium supplemen-
tation, and vitamin D supplementation (OR, 0.12 [95%CI, 0.03
to 0.55]; ARD, −2.08 [95% CI, −3.56 to −0.60])
The remaining 35 single or multifactorial interventions
were not significantly associated with the risk of injurious falls
relative to usual care. According to the P score results, com-
bined exercise and vision assessment and treatment was prob-
ably the most effective intervention (97% likelihood) to re-
duce injurious falls.
The results of subgroup analyses are summarized in
Table 4. For the 37 RCTs that had less than 75% women
(20 354 participants), which examined 27 treatments plus
usual care, the network meta-analysis results were consistent
with the main analysis. For the 44 RCTs with duration of 12
months or less (32 890 participants; examined 28 interven-
tions plus usual care), the network meta-analysis results
Table 3. Summary of Study Characteristics
Study Characteristic No. (%) of RCTs (N = 283)
Year of publication
1990-1995 10 (3.5)
1996-2000 18 (6.4)
2001-2005 66 (23.3)
2006-2010 77 (27.2)
2011-2015 87 (30.7)
2016-2017 25 (8.8)
Continent
a
Europe 121 (42.8)
Australia/New Zealand 66 (23.3)
North America 47 (16.6)
Asia 43 (15.2)
South America 4 (1.4)
Multicontinent 2 (0.7)
Study design
Parallel RCT 243 (85.9)
Cluster RCT 40 (14.1)
Site
Multicenter 155 (54.8)
Single center 127 (44.9)
Not reported 1 (0.4)
Settings
b
Home 142 (50.2)
Clinic 75 (26.5)
Community 72 (25.4)
Hospital 51 (18.0)
Long-term care facility 39 (13.8)
Retirement home 28 (9.9)
Not reported 9 (3.2)
Sample size (No. of patients)
Parallel RCTs
20-99 65 (23.0)
100-299 90 (31.8)
300-999 68 (24.0)
1000-9440 20 (7.1)
Cluster RCTs
68-500 21 (7.4)
501-1500 11 (3.9)
1501-5500 6 (2.1)
5501-10 558 2 (0.7)
No. of clusters, median (IQR) 17 (8.00-25.00)
Duration of Intervention, wk
<1-26 151 (53.4)
27-52 72 (25.4)
53-78 11 (3.9)
79-104 13 (4.6)
105-260 17 (6.0)
Not reported 19 (6.7)
Abbreviations: IQR, interquartile range; RCT,randomized clinical trial.
a
Continent refers to where the study was conducted; if not reported explicitly,
the location of the first author’s institution was used as a proxy.
b
The number of RCTs exceeds283 and percents total more than 100% because
some studies involved multiple settings.
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1691
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
were consistent with the main analysis for the interventions
of exercise vs usual care and combined clinic-level quality
improvement strategies, multifactorial assessment and treat-
ment, calcium supplementation, and vitamin D supplemen-
tation vs usual care; the remaining 2 significant comparisons
from the main analysis (combined exercise and vision assess-
ment and treatment vs usual care; combined exercise, vision
assessment and treatment, and environmental assessment
and modification vs usual care) were no longer included in
the network. Network meta-analysis was conducted for the
32 RCTs with participants who were younger than 80 years
old (24 869 participants), which examined 26 interventions
plus usual care. Compared with the main analysis, the same
interventions were associated with a reduced risk of injurious
falls except for the combination of clinic-level quality
improvement strategies, multifactorial assessment and treat-
ment, calcium supplementation, and vitamin D supplemen-
tation, which was no longer in the network. This finding was
consistent in another network meta-analysis involving the
40 RCTs (37 010 participants) for people with a mixed fall his-
tory (ie, some had fallen previously and some had not),
which examined 34 interventions plus usual care. A network
meta-analysis restricted to 11 RCTs involving 3830 patients
who had fallen previously and examining 9 interventions
plus usual care found that the combination of clinic-level
quality improvement strategies, multifactorial assessment
and treatment, calcium supplementation, and vitamin D
supplementation was associated with fewer injurious falls
than usual care (OR, 0.12 [95% CI, 0.04 to 0.44]; ARD, −2.08
[95% CI, −3.34 to −0.83]), while the remaining 3 comparisons
(exercise vs usual care; combined exercise and vision assess-
ment and treatment vs usual care; combined exercise, vision
assessment and treatment, and environmental assessment
and modification vs usual care) were no longer in the net-
work. Sensitivity analysis was conducted restricting the net-
work meta-analysis to 24 RCTs with a low risk of contamina-
tion bias (26 969 participants; 19 interventions plus usual
care); no intervention was associated with a lower risk of
injurious falls compared with usual care (eg, exercise vs usual
care, OR, 0.59 [95% CI, 0.29 to 1.18]; ARD, −0.53 [95% CI,
−1.23 to 0.17]), and 3 of the significant comparisons from
the main analysis were no longer in the network (combined
exercise and vision assessment and treatment vs usual care;
combined exercise, vision assessment and treatment, and
environmental assessment and modification vs usual care;
combined clinic-level quality improvement strategies, multi-
factorial assessment and treatment, calcium supplementa-
tion, and vitamin D supplementation vs usual care).
Overall, each of the 4 interventions that was associated
with better outcomes than usual care in the main analysis was
also associated with fewer injurious falls in 2 or more of the
additional analyses, although in some cases, an additional
analysis was not feasible for a particular subgroup.
Hospitalizations
For the primary outcome of fall-related hospitalizations, 20
RCTs (24 531 participants) with 25 interventions plus usual
care were included. Two pairwise meta-analyses were pos-
sible across all comparisons (eTable 9 in the Supplement).
There were no significant associations with hospitalizations
for combined clinic-level quality improvement strategies,
patient-level quality improvement strategies, and multifacto-
rial assessment and treatment relative to usual care (2 RCTs
[516 participants]; OR, 0.78 [95% CI, 0.33 to 1.81]; ARD, −0.03
[−0.10 to 0.093]) or for combined patient-level quality
improvement strategies and exercise relative to exercise
alone (2 RCTs [2126 participants]; OR, 1.12 [95% CI, 0.38 to
3.25]; ARD, 0.02 [−0.09 to 0.22]).
Emergency Department Visits
Eleven RCTs (2956 participants) with 12 interventions plus
usual care reported on emergency department visits. Only 1
pairwise meta-analysis (2 RCTs; 499 participants) was pos-
sible across all comparisons (eTable 9 in the Supplement). There
was no significant association with emergency department vis-
its for multifactorial assessment and treatment relativeto usual
care (OR, 1.24 [95% CI, 0.86 to 1.77]; ARD, 0.04[−0.03 t o 0.13]).
Outpatient Physician Visits
Twenty-one RCTs (17193 participants) with 32 interventions
plus usual care reported on physician visits. Only 1 pairwise
meta-analysis (2 RCTs; 681 participants) was possible
across all treatment comparisons (eTable 9 in the Supple-
ment). There was no significant association with outpatient
Box 2. Interventions Associated With Reduction of Outcome
Compared With Usual Care in Network Meta-analysis
Outcomes
Number of Injurious Falls
Exercise
Combined exercise and vision assessment and treatment
Combined exercise, vision assessment and treatment, and envi-
ronmental assessment and modification
Combined clinic-level quality improvement strategies, multifacto-
rial assessment and treatment, calcium supplementation, and vita-
min D supplementation
Number of Fallers
Exercise
Combined exercise, patient-level quality improvement strategies,
clinic-level quality improvement strategies, and multifactorial as-
sessment and treatment
Combined exercise, patient-level quality improvement strategies,
hip protectors, and environmental assessment and modification
Combined patient-level quality improvement strategies, clinic-
level quality improvement strategies, dietary modifications, cal-
cium supplementation, and vitamin D supplementation
Combined orthotics and exercise
Number of Fractures
Combined osteoporosis treatment, calcium supplementation, and
vitamin D supplementation
Number of Hip Fractures
Combined osteoporosis treatment, calcium supplementation, and
vitamin D supplementation
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1692 JAMA November 7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
physician visits for exercise relative to usual care (OR, 0.62
[95% CI, 0.32 to 1.18]; ARD, −0.10 [−0.19 to 0.04]).
Number of Fallers
Network meta-analysis for the outcome of number of fallers
included 158 RCTs, 107 300 participants, and 77 interven-
tions plus usual care (Figure 3). One RCT (0.6%) with 31 par-
ticipants (0.03%) was excluded from the network meta-
analysis because it had zero events across all groups. The event
rate for falls in the usual care group was 0.38 (95% CI, 0.33 to
0.43). Across all 3003 networkmeta-analysis comparisons, 200
(6.7%) were statistically significant (eTable 5 in the Supple-
ment). Of these, 5 interventions were associated with a lower
risk of patients experiencing a fall relative to usual care:
•Exercise (OR, 0.83 [95% CI, 0.70 to 0.99]; ARD, −0.19 [95%
CI, −0.36 to −0.01])
•Combined exercise, patient-level qualityimprovement strat-
egies, clinic-level quality improvement strategies, and mul-
tifactorial assessment and treatment (OR, 0.68[95% CI, 0.49
to 0.94]; ARD, −0.39 [95% CI, −0.72 to 0.06])
•Combined exercise, patient-level qualityimprovement strat-
egies, hip protectors, and environmental assessment and
modification (OR, 0.53 [95% CI, 0.29 to 0.97]; ARD, −0.63
[95% CI, −1.22 to −0.03])
•Combined patient-level quality improvement strategies,
clinic-level quality improvement strategies, dietary modifi-
cations, calcium supplementation, and vitamin D supple-
mentation (OR, 0.36 [95% CI, 0.14 to 0.93]; ARD, −1.03 [95%
CI, −1.99 to −0.08])
•Combined orthotics and exercise (OR, 0.22 [95% CI, 0.07 to
0.67]; ARD, −1.54 [95% CI, −2.67 to −0.40])
The remaining interventions were not significantly
associated with a lower risk of falls than usual care. One
intervention (combined exercise, patient-level quality
improvement strategies, and social engagement) was asso-
ciated with a higher risk that patients would experience falls
relative to usual care (OR, 5.13 [95% CI, 2.14 to 12.30]; ARD,
1.63 [95% CI, 0.76 to 2.51]).
Fractures
Network meta-analysis for the outcome of fractures included
68 RCTs, 86 491 participants, and 43 interventions plus usual
care (Figure 4). The event rate for fractures in the usual care
group was 0.07 (95% CI, 0.05 to 0.10). Across all 946 network
meta-analysis comparisons, 45 (4.8%) were statistically sig-
nificant (eTable 5 in the Supplement). Of these, 1 intervention
(combined osteoporosis treatment (eg, bisphosphonates),
calcium supplementation, and vitamin D supplementation)
was associated with a lower risk of fractures relative to usual
care (OR, 0.22 [95% CI, 0.09 to 0.54]; ARD, −1.51 [95% CI,
−2.41 to −0.62]). The remaining 42 interventions were not
significantly associated with a lower risk of fractures than
usual care.
Hip Fractures
Network meta-analysis for the outcome of hip fractures
included 39 RCTs, 52 281 participants, and 23 interventions
plus usual care (Figure 4). Four RCTs (9.1%) with 1877 par-
ticipants (3.5%) were excluded from the network meta-
analysis because they had zero events across all groups. The
event rate for hip fractures in the usual care group was 0.03
(95% CI, 0.02 to 0.04). Across all 276 network meta-analysis
Figure 2. Network Geometry for Injurious Falls
so
ex
ea
fl
va
pa–qi
cl–qi
mf
uc
de-al
ex+va
ex+vi–d
ex+pa–qi
ea+ex
ea+va
ea+pa–qi
pa–qi+va
cl–qi+mf
bf+pa–qi
bf+cl–qi
mf+pa–qi
ex+mf+pa–qi
ea+ex+va
ea+ex+vi–d
ea+ex+mf
cb+ex+pa–qi
cb+cl–qi+pa–qi
cl–qi+ex+pa–qi
cl–qi+ex+mf
cl–qi+mf+pa–qi
bf+cl–qi+pa–qi
bf+cl–qi+sy–qi
ea+ex+pa–qi+sy–qi
ca+ex+em+wb+vi–d
ca+cl–qi+mf+vi–d
cl–qi+ex+mf+pa–qi
cl–qi+mf+pa–qi+so
bf+cl–qi+ex+pa–qi
cl–qi+de–al+ex+mf+pa–qi
bf+ea+ex+pa–qi+va
Network geometry for 54
randomized clinical trials (41 596
patients). Each treatment node
indicates an intervention and is
weighted according to the number of
patients who received the particular
intervention. Each edge (line
connecting the nodes) is weighted
according to the number of studies
and directly compares the treatments
it connects. See Table 1 for
expansions of treatment
abbreviations. The coding guide,
which provides a description of each
intervention component, can be
found in eTable 1 of the Supplement.
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1693
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
Table 4.Subgroup Analyses of Ne tworkMeta-analysis for Injurious Falls Outcome
Comparison by Subgroup Studies, No. Patients, No.
Proportion With Event (95% CI)
Odds Ratio (95% CI)
Absolute Mean Difference
in Proportions (95% CI)
a
Intervention Control
Exercise vs Usual Care
Overall analysis 0.51 (0.33 to 0.79)
Participants <75% women 37 20 354 0.36 (0.16 to 0.59) 0.41 (0.29 to 0.53) 0.49 (0.31 to 0.78) −0.71 (−1.17 to −0.25)
Study duration ≤12 mo 44 32 890 0.30 (0.13 to 0.52) 0.33 (0.22 to 0.44) 0.48 (0.29 to 0.80) −0.72 (−1.23 to −0.22)
Age <80 y of age 32 24 869 0.25 (0.08 to 0.48) 0.35 (0.19 to 0.53) 0.44 (0.26 to 0.75) −0.81 (−1.35 to −0.28)
Mixed history of falling
b
40 37 010 0.36 (0.16 to 0.59) 0.37 (0.25 to 0.49) 0.49 (0.30 to 0.82) −0.71 (−1.22 to −0.20)
History of falling only
c
11 3830 0.16 (0.07 to 0.27) 0.24 (0.07 to 0.47) 0.90 (0.24 to 3.30) −0.11 (−1.41 to 1.19)
Low risk of contamination bias 24 26969 0.40 (0.00 to 0.96) 0.26 (0.15 to 0.37) 0.59 (0.29 to 1.18) −0.53 (−1.23 to 0.17)
Combined Exercise and Vision Assessment and Treatment vs Usual Care
Overall analysis 0.17 (0.07 to 0.38)
Participants <75% women 37 20 354 0.51 (0.42 to 0.59) 0.41 (0.29 to 0.53) 0.16 (0.07 to 0.39) −1.82 (−2.68 to −0.95)
Study duration ≤12 mo 44 32 890 NA 0.33 (0.22 to 0.44) NA
Age <80 y of age 32 24 869 0.51 (0.42 to 0.59) 0.35 (0.19 to 0.53) 0.17 (0.07 to 0.43) −1.76 (−2.66 to −0.85)
Mixed history of falling
b
40 37 010 0.51 (0.42 to 0.59) 0.37 (0.25 to 0.49) 0.16 (0.06 to 0.42) −1.82 (−2.77 to −0.86)
History of falling only
c
11 3830 NA 0.24 (0.07 to 0.47) NA
Low risk of contamination bias 24 26 969 NA 0.26 (0.15 to 0.37) NA
Combined Exercise, Vision Assessment and Treatment, and Environmental Assessment and Modification vs Usual Care
Overall analysis 0.30 (0.13 to 0.70)
Participants <75% women 37 20 354 0.65 (0.57 to 0.73) 0.41 (0.29 to 0.53) 0.30 (0.12 to 0.71) −0.22 (−2.09 to −0.35)
Study duration ≤12 mo 44 32 890 NA 0.33 (0.22 to 0.44) NA
Age <80 y of age 32 24 869 0.65 (0.57 to 0.73) 0.35 (0.19 to 0.53) 0.31 (0.13 to 0.78) −1.16 (−2.07 to −0.24)
Mixed history of falling
b
40 37 010 0.65 (0.57 to 0.73) 0.37 (0.25 to 0.49) 0.30 (0.11 to 0.78) −1.22 (−2.18 to −0.25)
History of falling only
c
11 3830 NA 0.24 (0.07 to 0.47) NA
Low risk of contamination bias 24 26 969 NA 0.26 (0.15 to 0.37) NA
Combined Clinic-Level Quality Improvement Strategies, Multifactorial Assessment and Treatment, Calcium Supplementation, and Vitamin D Supplementation vs Usual Care
Overall analysis 0.12 (0.03 to 0.55)
Participants <75% women 37 20 354 0.03 (0.00 to 0.07) 0.41 (0.29 to 0.53) 0.12 (0.03 to 0.56) −2.08 (−3.58 to −0.58)
Study duration ≤12 mo 44 32 890 0.03 (0.00 to 0.07) 0.33 (0.22 to 0.44) 0.12 (0.03 to 0.54) −2.08 (−3.56 to −0.61)
Age <80 y of age 32 24 869 NA 0.349 (0.191 to 0.527) NA
Mixed history of falling
b
40 37 010 NA 0.37 (0.25 to 0.49) NA
History of falling only
c
11 3830 0.03 (0.00 to 0.07) 0.24 (0.07 to 0.47) 0.12 (0.04 to 0.44) −2.08 (−3.34 to −0.83)
Low risk of contamination bias 24 26 969 NA 0.26 (0.15 to 0.37) NA
Abbreviations: NA, not applicable.
a
Odds ratios derived from each network meta-analysis were transformed to risk differences using
established methods.
34
b
Studies that included participants regardless of whether they had fallen in the past or not.
c
Studies that only included participants who had fallen in the past.
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1694 JAMA November 7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
comparisons, 9 (3.3%) were statistically significant (eTable 5
in the Supplement). Of these, 1 intervention (combined
osteoporosis treatment, calcium supplementation, and vita-
min D supplementation) was associated with a lower risk of
hip fracture relative to usual care (OR, 0.18 [95% CI, 0.05 to
0.62]; ARD, −1.70 [95% CI, −2.92 to −0.48]). An additional
22 interventions were not significantly associated with a
lower risk of hip fractures than usual care.
Quality of Life
Although 32 RCTs (18 521 participants) provided data on
quality-of-life measures (eTable 10 in the Supplement),
only 2 pairwise meta-analyses were possible (eTable 9 in
the Supplement). Exercise was not significantly associated
with improvement in quality of life as measured by the
SF-36 or SF-12 physical component summary score com-
pared with usual care (mean difference, −0.06 [95% CI,
−0.90 to 0.77]; 2 RCTs; 1206 participants). Similar results
were obtained for the SF-36 or SF-12 mental component
summary score (mean difference, 0.29 [95% CI, −1.00 to
1.58]; 2 RCTs; 1206 participants).
Harms
Fifty-seven of the RCTs (24558 participants) reported no
intervention-related harmful events in any study group, and
another 62 RCTs (39 596 participants) reported 1 or more
harms (eTable 11 in the Supplement). Only 2 pairwise meta-
analyses were possible (eTable 9 in the Supplement). Exer-
cise was not significantly associated with an increased risk of
muscle soreness compared with usual care (OR, 4.97 [95% CI,
0.35 to 70.38]; ARD, 0.13 [95% CI, −0.02 to 0.70]; 2 RCTs;
1021 participants). Supplementation with calcium and vita-
min D was not significantly associated with an increased risk
of gastrointestinal harm compared with usual care (OR, 1.05
[95% CI, 0.52 to 2.09]; ARD, 0.001 [95% CI, −0.01 to 0.03];
2 RCTs; 3853 participants).
Figure 3. Network Geometry for Fallers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18 19
20
21
22
23
24
25 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
Order
Code
Treatment
Abbreviation
1uc
2di
3so
4ex
5 em+wb
6ea
7fl
8 de-al
9 de-hp
10 va
11 su-ey
12 su-PM
13 ca
14 vi-d
15 op-tx
16 pa-qi
17 cb
18 cl-qi
19 bf
20 mf
21 di+ex
22 di+vi-d
23 ex+so
24 ex+va
25 ex+vi-d
26 ex+pa-qi
27 ex+sy-qi
28 ex+mf
29 ea+ex
30 ea+va
31 ea+pa-qi
32 de-hp+pa-qi
33 de-or+ex
34 ca+vi-d
35 op-tx+vi-d
36 pa-qi+va
37 pa-qi+vi-d
38 cb+ex
39 cb+pa-qi
Order
Code
Treatment
Abbreviation
40 cl-qi+de-hp
41 cl-qi+pa-qi
42 cl-qi+mf
43 bf+pa-qi
44 bf+cl-qi
45 mf+pa-qi
46 ex+pa-qi+so
47 ex+mf+pa-qi
48 ex+mf+sy-qi
49 ea+ex+va
50 ea+ex+vi-d
51 ea+ex+mf
52 ea+mf+sy-qi
53 cp+de-ft+pa-qi
54 ca+di+vi-d
55 ca+op-tx+vi-d
56 ca+pa-qi+vi-d
57 cb+ex+pa-qi
58 cb+cl-qi+pa-qi
59 cl-qi+ex+pa-qi
60 cl-qi+ex+mf
61 cl-qi+ea+pa-qi
62 cl-qi+de-hp+pa-qi
63 cl-qi+mf+pa-qi
64 bf+ex+pa-qi
65 bf+cl-qi+pa-qi
66 bf+cl-qi+sy-qi
67 ea+ex+pa-qi+sy-qi
68 de-hp+ea+ex+pa-qi
69 ca+cl-qi+mf+vi-d
70 ca+ea+pa-qi+vi-d
71 cl-qi+ex+mf+pa-qi
72 cl-qi+mf+pa-qi+so
73 bf+cp+ex+pa-qi
74 bf+cl-qi+ex+pa-qi
75 ca+cb+di+ex+vi-d
76 ca+cl-qi+di+pa-qi+vi-d
77 cl-qi+de-al+ex+mf+pa-qi
78 bf+ea+ex+pa-qi+va
Network geometry for 158 randomized clinical trials (107 300 patients). Each
treatment node indicates an intervention and is weighted according to the
number of patients who received the particular intervention. Each edge (line
connecting the nodes) is weighted according to the number of studies and
directly compares the treatments it connects. See Table 1 for expansions of
treatment abbreviations. The coding guide, which provides a description of
each intervention component, can be found in eTable 1 of the Supplement.
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1695
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
Discussion
Exercise alone and various combinations of interventions
were associated with lower risk of injurious falls compared
with usual care. Choice of intervention may depend on
patient and caregiver values and preferences. Combinations
of interventions, including exercise, vision assessment and
treatment, environmental assessment and modification,
multifactorial assessment and treatment, and vitamin D
supplementation were associated with preventing injurious
falls compared with usual care. The combination of exercise
Figure 4. Network Geometry for Fractures and Hip Fractures
uc
di
so
ex
fl
de–al
su–ey
ca
vi–d
op–tx
op–tx+vi–d
pa–qi
cl–qi
bf
mf
di+ex
ex+pa–qi
ex+mf
ea+va
ea+pa–qi
de–hp+pa–qi ca+vi–d
pa–qi+vi–d
cl–qi+de–hp
cl–qi+mf
bf+su–pm
bf+cl–qi
ea+ex+mf
cp+de–ft+pa–qi
ca+op–tx+vi–d
ca+pa–qi+vi-d
cb+cl–qi+pa–qi
cl–qi+ex+pa–qi
cl–qi+ex+mf
cl–qi+mf+pa–qi
bf+cl–qi+pa–qi
de–hp+ea+ex+pa–qi
ca+ea+pa–qi+vi–d
ca+cl–qi+mf+vi–d
cl–qi+ex+mf+pa–qi
cl–qi+mf+pa–qi+so
bf+cp+ex+pa–qi
ca+cl–qi+di+pa–qi+vi–d
Fractures
A
Hip fractures
B
uc
so
ex
fl
de–al
su–ey
ca
vi–d
pa–qi
bf
mf
ex+pa–qi
ea+va
de–hp+pa–qi
ca+vi–d
cl–qi+de–hp
cl–qi+pa–qi
cl–qi+mf
cp+de–ft+pa–qi
ca+op-tx+vi–d
cl–qi+de–hp+pa–qi
cl–qi+mf+pa–qi
de–hp+ea+ex+pa–qi
cl–qi+mf+pa–qi+so
cl–qi+de–al+ex+mf+pa–qi
A, Network geometry for 68
randomized clinical trials (86 491
patients). B, Network geometry for
39 randomized clinical trials (52 281
patients). Each treatment node
indicates an intervention and is
weighted according to the number of
patients who received the particular
intervention. Each edge (line
connecting the nodes) is weighted
according to the number of studies
and directly compares the treatments
it connects. See Table 1 for
expansions of treatment
abbreviations. The coding guide,
which provides a description of each
intervention component, can be
found in eTable 1 of the Supplement.
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1696 JAMA November 7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
and vision assessment and treatment was probably the
intervention most strongly associated with reduction in
injurious falls.
These results suggest that encouraging patients to exer-
cise, undergo a vision assessment, and consider osteoporo-
sis therapy (for those at risk), given the potential impact of
these interventions in preventing injurious falls. Other com-
binations of interventions to consider include exercise,
patient-level and clinic-level quality improvement strate-
gies, multifactorial assessment and treatment; exercise,
patient-level quality improvement strategies, hip protectors,
and environmental assessment and modification; and
orthotics and exercise in patients at risk for falls. The results
suggest focusing on implementing patient-level quality
improvement strategies (eg, patient education and patient
reminders) and clinic-level quality improvement interven-
tions (eg, audit and feedback) to increase uptake of this evi-
dence. The results also suggest that calcium and vitamin D
supplementation may decrease fractures, as may osteoporo-
sis therapy plus calcium and vitamin D supplementation.
However, the results indicate the need for a tailored ap-
proach; subgroup analyses showed that the combination of
exercise, environmental assessment and modification, and
multifactorial assessment and treatment was associated
with an increased risk of injurious falls among patients who
had fallen previously. In addition, the combination of exer-
cise, patient-level quality improvement strategies, and social
engagement was associated with increased fall risk. Exercise
may increase fall risk in some individuals because these
people become more mobile as their strength increases;
patients can be made aware of this situation, but any cau-
tionary advice must be balanced with the need to improve
mobility and avoid deconditioning. This finding also raises
the issue of the type of exercise to recommend, such as exer-
cise focused on enhancing balance.
52
Health care managers
might consider the use of clinic-level quality improvement
strategies, such as clinician reminders and audit and feed-
back, to increase uptake of multifactorial assessment and
treatment and of vitamin D supplementation to reduce inju-
rious falls. As well, for patients in long-term care settings,
hip protectors, environmental assessment and modification,
exercise, and patient-level quality improvement strategies
are potential options to reduce falls.
A 2015 systematic review and network meta-analysis
examining vision and risk factor interventions to prevent
falls included 7 RCTs with 2723 participants; interventions
combining vision assessment and treatment and multifacto-
rial assessment and treatment were found to be the most
effective.
53
The current systematic review was much larger
(with an additional 276 RCTs and 157187 participants; eRef-
erences in the Supplement) and includes information on the
effectiveness of quality improvement strategies and multi-
factorial assessment and treatment. Although the authors of
a systematic review on exercise
54
did not conduct network
meta-analysis, their results were similar to those reported
here, which suggests that exercise might be associated with
decreased falls in older adults, but the type of exercise
should be tailored to the individual.
The authors of RCTs included in this review could have
improved their studies by conducting adequate allocation
concealment (which is possible in any RCT), ensuring that
results were not influenced by contamination bias, and
reporting all outcomes. Most RCTs were completed within 6
months, but longer-term follow-up and confirmation of the
sustainability of these interventions are required. In addi-
tion, few studies were conducted in the acute care setting,
despite falls in hospitals often being considered during hos-
pital accreditation processes.
Strengths of the review process include reviewers work-
ing in pairs across all levels of screening, data abstraction, and
risk-of-bias appraisal; cleaning of the data by a third re-
viewer; and following the guidance of the International Soci-
ety for Pharmacoeconomics and Outcomes Research in con-
ducting statistical analyses.
55
This study had several limitations. The published
protocol included both the falls rate and costs as secondary
outcomes. However, because few studies reported these
outcomes consistently, these outcomes could not be ana-
lyzed directly; rather, data for falls rates were converted to
number of falls. As well, another outcome, quality of life, as
measured by SF-12 or SF-36 summary component measures
or EuroQol-5D, was added.
17-19
Some of the planned sub-
group analyses and sensitivity analyses were not conducted
because of insufficient data. Although the point esti-
mate was similar to the overall OR, the results were no lon-
ger statistically significant for the injurious falls network
meta-analysis when only studies with a low risk of contami-
nation bias were included. However, because most of the
studies (67%) were assessed as having an unclear risk of
contamination bias, the power of this sensitivity analysis
was limited by the lower number of studies that could be
included. This limitation suggests that improvements in
reporting are required. Most network meta-analyses
included numerous interventions, with sparse data for the
treatment comparisons; additional analyses, using the mod-
els suggested by Welton et al
56
and Caldwell and Welton
57
to account for sparseness, could be conducted in the fu-
ture. Scanning the reference lists of 32 additional studies
from the updated search and inclusion of an unpublished
conference abstract
58
and a non-English paper
59
were not
possible. Because of the large number of comparisons in the
network meta-analyses, multiplicity may have elevated
the rate of false positives in the statistically significant
results (type I error).
60,61
Although P scores are based on
the treatment effect estimates and their associated CIs,
it is recommended that the P score values be interpreted
along with the network meta-analysis point estimates and
their precision.
29
Conclusions
Exercise alone and various combined interventions were as-
sociated with lower risk of injurious falls compared with usual
care. Choice of intervention may depend on patient and care-
giver values and preferences.
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1697
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
ARTICLE INFORMATION
Accepted for Publication: September 28, 2017.
Author Affiliations: Knowledge Translation
Program, Li Ka Shing Knowledge Institute, St
Michael’s Hospital, Toronto, Ontario, Canada
(Tricco,Thomas, Veroniki, Hamid, Cogo, Strifler,
Khan, Robson, MacDonald, Thavorn, Wilson, Kerr,
Hui, Straus); Epidemiology Division, Dalla Lana
School of Public Health, University of Toronto,
Toronto,Ontario, Canada (Tricco); Institute of
Health Policy, Managementand Evaluation,
University of Toronto, Toronto, Ontario, Canada
(Strifler); Toronto Rehabilitation Institute,
University Health Network, Toronto, Ontario,
Canada (Sibley); Department of Community Health
Sciences, University of Manitoba, Winnipeg,
Manitoba, Canada (Sibley); Department of Family
Medicine, David Braley Health Sciences Centre,
McMaster University,Hamilton, Ontario, Canada
(Riva); Department of Health Research Methods,
Evidence, and Impact, McMaster University,
Hamilton, Ontario, Canada (Riva); Clinical
Epidemiology Program, The Ottawa Hospital
Research Institute, Ottawa, Ontario, Canada
(Thavorn); Departments of Medicine and
Community Health Sciences, University of Calgary,
Calgary, Alberta, Canada (Holroyd-Leduc);Older
Adult Program, Fraser Health, Surrey, British
Columbia, Canada (Feldman); Department of
Medicine, University of Alberta, Edmonton, Alberta,
Canada (Majumdar); Department of Physical
Therapy, University of Toronto, Toronto, Ontario,
Canada (Jaglal); Department of Medicine,
University of Toronto, Toronto, Ontario, Canada
(Straus).
Author Contributions: Drs Straus and Tricco had
full access to all of the data in the study and take
responsibility for the integrity of the data and the
accuracy of the data analysis.
Concept and design: Tricco,Sibley, Holroyd-Leduc,
Feldman, Majumdar,Jaglal, Straus.
Acquisition, analysis, or interpretation of data:
Tricco,Thomas, Veroniki, Hamid, Cogo, Strifler,
Khan, Robson, Sibley,MacDonald, Riva, Thavorn,
Wilson, Holroyd-Leduc, Kerr, Majumdar, Hui, Straus.
Drafting of the manuscript: Tricco,Thomas,
Veroniki, Cogo, Sibley, Straus.
Critical revision of the manuscript for important
intellectual content: All authors.
Statistical analysis: Veroniki, Hamid, Hui, Straus.
Obtained funding: Tricco,Sibley.
Administrative, technical, or material support:
Thomas, Cogo, Riva, Thavorn, Kerr, Straus.
Supervision: Tricco, Straus.
Other - Knowledge user input: Feldman.
Other - Editing & content analysis: Wilson.
Conflict of Interest Disclosures: All authors have
completed and submitted the ICMJE Form for
Disclosure of Potential Conflicts of Interest.
Dr Tricco reports receiving a Tier 2 Canada Research
Chair in Knowledge Synthesis grant. Dr Veroniki
reports receiving a Canadian Institutes of Health
Research (CIHR) Banting Postdoctoral Fellowship
Program grant. Dr Sibley reports receiving a Tier 2
Canada Research Chair in Integrated Knowledge
Translation in RehabilitationSciences grant. Dr Riva
reports board membership with the Ontario
Chiropractic Association. Dr Holyroyd-Leduc
reports working as an associate editor for the
Canadian Medical Association Journal. Dr Majumdar
reports support from the Faculty of Medicine and
Dentistry and the Faculty of Pharmacy and
Pharmaceutical Sciences, University of Alberta
(holds the Endowed Chair in Patient Health
Management). Dr Straus reports receiving a Tier 1
Canada Research Chair in Knowledge Translation
grant. No other disclosures were reported.
Funding/Support: This research was funded by a
CIHR Knowledge Synthesis Grant (KRS 289648).
Role of the Funders/Sponsors: CIHR had no role in
the design and conduct of the study; collection,
management, analysis, and interpretation of the
data; preparation, review,or approval of the
manuscript; and decision to submit the manuscript
for publication.
Additional Contributions: We thank Joseph
Beyene, PhD, McMaster University, for analyzing
preliminary data; Alexander Leung, MD, University
of Calgary, Sophia Tsouros, BHKin, University of
Ottawa Evidence-based Practice Center, Alana
Harrington, MSc, York University, Vera Nincic, PhD,
St Michael’s Hospital, and Geetha Sanmugalingham,
MSc, Tampa General Hospital, for screening and
abstracting data from some studies; Laure Perrier,
PhD, University of Toronto, for developing the
literature search strategies; Patricia Rios, MSc,
St Michael’s Hospital, for cleaning the data from the
updated search and helping update statistics files
and tables; Susan Le, HBSc, St Michael’s Hospital,
for formatting the manuscript and managing the
references; and Alissa Epworth, OCD, St Michael’s
Hospital, for updating our search and obtaining the
full-text articles. None of the aforementioned
individuals received compensation for their role in
the study beyond regular salary.Additionally, we
thank Becky Skidmore, MLS, Ottawa Health
Research Institute, who received compensation for
peer-reviewing the MEDLINE search strategy; and
Peggy Robinson, BSc, an independent editorial
consultant, who received compensation for
provision of copyediting services.
REFERENCES
1. Cigolle CT, Ha J, Min LC, et al. The epidemiologic
data on falls, 1998-2010: more older Americans
report falling. JAMA Intern Med. 2015;175(3):4 43-445.
2. Morrison A, Fan T, Sen SS, Weisenfluh L.
Epidemiology of falls and osteoporotic fractures:
a systematic review.Clinicoecon Outcomes Res.
2013;5:9-18.
3. Stel VS, Smit JH, Pluijm SM, Lips P.
Consequences of falling in older men and women
and risk factors for health service use and
functional decline. Age Ageing. 2004;33(1):58-65.
4. Alamgir H, Muazzam S, Nasrullah M.
Unintentional falls mortality among elderly in the
United States: time for action. Injury. 2012;43(12):
2065-2071.
5. Burns ER, Stevens JA, Lee R. The direct costs of
fatal and non-fatal falls among older adults—United
States. J Safety Res. 2016;58:99-103.
6. Gill TM, Murphy TE, Gahbauer EA, Allore HG.
Association of injurious falls with disability
outcomes and nursing home admissions in
community-living older persons. Am J Epidemiol.
2013;178(3):418-425.
7. Siracuse JJ, Odell DD, Gondek SP, et al. Health
care and socioeconomic impact of falls in the
elderly. Am J Surg.2012;203(3):335-338.
8. Kumar A, Carpenter H, Morris R, Iliffe S,
Kendrick D. Which factors are associated with fear
of falling in community-dwelling older people? Age
Ageing. 2014;43(1):76-84.
9. Parry SW, Deary V, Finch T, et al. The STRIDE
(Strategies to Increase confidence, InDependence
and Energy) study: cognitive behavioural
therapy-based intervention to reduce fear of falling
in older fallers living in the community—study
protocol for a randomised controlled trial. Trials.
2014;15:210.
10. Ray WA, TaylorJA, Brown AK, et al. Prevention
of fall-related injuries in long-term care:
a randomized controlled trial of staff education.
Arch Intern Med. 2005;165(19):2293-2298.
11. Dykes PC, Carroll DL, Hurley A, et al. Fall
prevention in acute care hospitals: a randomized
trial. JAMA. 2010;304(17):1912-1918.
12. Cumming RG, Sherrington C, Lord SR, et al;
Prevention of Older People’s Injury FallsPrevention
in Hospitals Research Group. Cluster randomised
trial of a targeted multifactorial intervention to
prevent falls among older people in hospital. BMJ.
2008;336(7647):758-760.
13. Hornbrook MC, Stevens VJ, Wingfield DJ, Hollis
JF, Greenlick MR, Ory MG. Preventing fallsamong
community-dwelling older persons: results from a
randomized trial. Gerontologist. 1994;34(1):16-23.
14. Moher D, Shamseer L, Clarke M, et al; PRISMA-P
Group. Preferred reporting items for systematic
review and meta-analysis protocols (PRISMA-P)
2015 statement. Syst Rev. 2015;4:1.
15. Tricco AC,Cogo E, Holroyd-Leduc J, et al.
Efficacy of falls prevention interventions: protocol
for a systematic review and network meta-analysis.
Syst Rev. 2013;2:38.
16. Hutton B, Salanti G, Caldwell DM, et al.
The PRISMA extension statement for reporting of
systematic reviews incorporating network
meta-analyses of health care interventions:
checklist and explanations. Ann Intern Med. 2015;
162(11):777-784.
17. Ware J Jr,Kosinski M, Keller SDA. A 12-item
Short-Form health survey: construction of scales
and preliminary tests of reliability and validity.Med
Care. 1996;34(3):220-233.
18. Ware J, KosinskiM, Dewey J. How to Score
Version Two of the SF-36 Health Survey. Lincoln, RI:
QualityMetric; 2000.
19. EuroQol Research Foundation. EQ-5Dhealth
questionnaire. http://www.euroqol.org/. Accessed
September 26, 2017.
20. Tricco AC,Ivers NM, Grimshaw JM, et al.
Effectiveness of quality improvement strategies on
the management of diabetes: a systematic review
and meta-analysis. Lancet. 2012;379(9833):2252-
2261.
21. Kim H, Yoshida H, Suzuki T. Falls and fractures in
participants and excluded non-participants of a fall
prevention exercise program for elderly women
with a history of falls: 1-year follow-up study. Geriatr
Gerontol Int. 2014;14(2):285-292.
Research Original Investigation Comparisons of Interventions for Preventing Falls in Older Adults
1698 JAMA November 7, 2017 Volume 318, Number 17 (Reprinted) jama.com
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
22. Chaimani A, Higgins JP,Mavridis D, Spyridonos
P, Salanti G. Graphical tools for network
meta-analysis in STATA. PLoS One. 2013;8(10):
e76654.
23. FurukawaTA, Barbui C, Cipriani A, Brambilla P,
Watanabe N. Imputing missing standard deviations
in meta-analyses can provide accurate results. J Clin
Epidemiol. 2006;59(1):7-10.
24. Littell J, Corcoran J,Pillai V. Systematic Reviews
and Meta-analysis. New York, NY: Oxford University
Press; 2008.
25. Nüesch E, Trelle S, ReichenbachS, et al.
Small study effects in meta-analyses of
osteoarthritis trials: meta-epidemiological study.
BMJ. 2010;341:c3515.
26. White IR, Barrett JK, Jackson D, Higgins JP.
Consistency and inconsistency in network
meta-analysis: model estimation using multivariate
meta-regression. Res Synth Methods. 2012;3(2):
111-125.
27. Jansen JP, Naci H. Is network meta-analysis as
valid as standard pairwise meta-analysis? it all
depends on the distribution of effect modifiers.
BMC Med. 2013;11:159.
28. Veroniki AA, Vasiliadis HS,Hig gins JP, Salanti G.
Evaluation of inconsistency in networks of
interventions. Int J Epidemiol. 2013;42(1):332-345.
29. Rücker G, Schwarzer G. Ranking treatments in
frequentist network meta-analysis works without
resampling methods. BMC Med Res Methodol. 2015;
15:58.
30. Veroniki AA, Straus SE, Fyraridis A, Tricco AC.
The rank-heat plot is a novel wayto present the
results from a network meta-analysis including
multiple outcomes. J Clin Epidemiol. 2016;76:193-
199.
31. Freeman MF, Tukey JW. Transformations related
to the angular and the square root. Ann Math Stat.
1950;21(4):607-611.
32. R: a language and environment for statistical
computing [computer program]. Vienna, Austria:
R Foundation for Statistical Computing; 2016.
33. Rucker G, Schwarzer G, KrahnU, Konig J.
Network meta-analysis using frequentist methods.
https://cran.r-project.org/web/packages/netmeta
/netmeta.pdf. Accessed September 26, 2017.
34. Murad MH, Montori VM, Walter SD, GuyattGH.
Estimating risk difference from relative association
measures in meta-analysis can infrequently pose
interpretational challenges. J Clin Epidemiol. 2009;
62(8):865-867.
35. Frohnhofen HH, Kandzia A, Willmann V, Joist A.
Hip fracture and hip protector in geriatric
in-hospital patients. Paper presented at: World
Congress on Osteoporosis (IOF WCO-ECCEO10);
May 5-8, 2010; Florence, Italy.
36. Patru S, Bighea AC, Marcu IR, Popescu RS.
Randomized controlled trial of exercise fall
prevention program in elderly women with
osteroporisis. Paper presented at: World Congress
on Osteoporosis (IOF WCO-ECCEO10); May 5-8,
2010; Florence, Italy.
37. Brown AI. Functional Adaptation to Exercisein
Elderly Subjects: Physiotherapy. Bently,Perth,
Western Australia: Curtin University of Technology;
2002.
38. Bunout D, Barrera G, AvendañoM, et al. Results
of a community-based weight-bearing resistance
training programme for healthy Chilean elderly
subjects. Age Ageing. 2005;34(1):80-83.
39. Cerny K, Blanks R, Mohamed O, et al.
The effect of a multidimensional exercise program
on strength, range of motion, balance and gait in
the well elderly.Gait Posture. 1998;7(2):185-186.
doi:10.1016/S0966-6362(98)90283-1
40. Chapuy MC, Arlot ME, Delmas PD, Meunier PJ.
Effect of calcium and cholecalciferol treatment for
three years on hip fractures in elderly women. BMJ.
1994;308(6936):1081-1082.
41. Chapuy MC, Pamphile R, Paris E, et al.
Combined calcium and vitamin D3 supplementation
in elderly women: confirmation of reversal of
secondary hyperparathyroidism and hip fracture
risk: the Decalyos II study.Osteoporos Int. 2002;13
(3):257-264.
42. Jäntti PO, Aho HJ, Mäki-Jokela PL,
Heikinheimo RJ. Hip protectors and hip fractures.
Age Ageing. 1998;27(6):758-759.
43. Jarvis N, Kerr K, Mockett S. A pilot study to
explore the feasibility of a randomised controlled
trial to determine the dose effect of physiotherapy
on patients admitted to hospital following a fall.
Practical Evidence. 2007;2:4-12.
44. Möller UO, Kristensson J, Midlöv P, Ekdahl C,
Jakobsson U. Effects of a one-yearhome-based
case management intervention on falls in older
people: a randomized controlled trial. J Aging Phys
Act. 2014;22(4):457-464.
45. Pfeifer M, Begerow B, Minne HW, AbramsC,
Nachtigall D, Hansen C. Effects of a short-term
vitamin D and calcium supplementation on body
sway and secondary hyperparathyroidism in elderly
women. J Bone Miner Res. 2000;15(6):1113-1118.
46. Reid IR, Mason B, Horne A, et al. Randomized
controlled trial of calcium in healthy older women.
Am J Med. 2006;119(9):777-785.
47. MacRae PG, Feltner ME, Reinsch S. 1-year
exercise program for older women: effects on falls,
injuries, and physical performance. J Aging Phys Act.
1994;2:127-142. doi:10.1123/japa.2.2.127
48. TrivediDP, Doll R, Khaw KT.Effect of four
monthly oral vitamin D3 (cholecalciferol)
supplementation on fractures and mortality
in men and women living in the community:
randomised double blind controlled trial. BMJ.
2003;326(7387):469.
49. Vetter NJ, LewisPA, Ford D. Can health visitors
prevent fractures in elderly people? BMJ.1992;
304(6831):888-890.
50. Tricco A, Thomas S. Open Science Framework.
Comparative efficacy of preventing falls in older
adults: a systematic review and network
meta-analysis. https://osf.io/ckuzj/. Accessed
September 26, 2017.
51. Zak M. Effectiveness of physiotherapeutic
regimens for enhancing the individual capability of
the elderly for unassisted coping after an incidental
fall. Fizjoterapia. 2005;13(2):12-19.
52. Lomas-Vega R, Obrero-Gaitán E, Molina-Ortega
FJ, Del-Pino-Casado R. Tai chi for risk of falls:
a meta-analysis. J Am Geriatr Soc. 2017;65(9):2037-
2043.
53. Zhang XY,Shuai J, Li LP. Vision and relevant risk
factor interventions for preventing falls among
older people: a network meta-analysis. Sci Rep.
2015;5:10559.
54. Sherrington C, Michaleff ZA, Fairhall N, et al.
Exercise to prevent falls in older adults: an updated
systematic review and meta-analysis. Br J Sports Med.
2016;bjsports-2016-096547.
55. Jansen JP, Trikalinos T,Cappelleri JC, et al.
Indirect treatment comparison/network
meta-analysis study questionnaire to assess
relevance and credibility to inform health care
decision making: an ISPOR-AMCP-NPC Good
Practice Task Forcereport . Value Health. 2014;17
(2):157-173.
56. Welton NJ, Caldwell DM, Adamopoulos E,
Vedhara K. Mixed treatment comparison
meta-analysis of complex interventions:
psychological interventions in coronary heart
disease. Am J Epidemiol. 2009;169(9):1158-1165.
57. Caldwell DM, Welton NJ. Approaches for
synthesising complex mental health interventions
in meta-analysis. Evid Based Ment Health. 2016;
19(1):16-21.
58. Hill AM, McPhail S, Waldron N, et al. Reducing
falls in hospital through education to change
patient and staff behaviour: a stepped wedge
cluster randomised controlled effectiveness trial.
Physiotherapy. 2015;101:e984.doi:10.1016/j.physio
.2015.03.1843
59. Kawabata T, Takemi Y,Murayama H, et al.
Effects of an intervention program for
community-dwelling elderly to improve frailty and
dietary habits [in Japanese]. Nihon Koshu Eisei Zasshi.
2015;62(4):169-181.
60. Higgins JP,Thompson SG. Controlling the risk
of spurious findings from meta-regression. Stat Med.
2004;23(11):1663-1682.
61. European Medicines Agency. Points to
Consider on Multiplicity Issues in Clinical Trials.
CPMP/EWP/908/99. www.ema.europa.eu. Accessed
September 26, 2017.
Comparisons of Interventions for Preventing Falls in Older Adults Original Investigation Research
jama.com (Reprinted) JAMA November 7, 2017 Volume 318, Number 17 1699
© 2017 American Medical Association. All rights reserved.
Downloaded From: by a St. Michael's Hospital User on 11/27/2017
