Solve 3 questions after reviewing attached articles

Use ‘questions’ word file to answer the questions

Solvethe questions below using the attached file. 12pt font size, Times new roman, 3 pages limit.

USE THE ATTACHTED ‘ARTICLE #1’ TO ANSWER THE QUESTIONS BELOW

Q1: Review the study titled “Consequences of bullying victimization in childhood and adolescence: A systematic review and meta-analysis” (Article#1 attachment), and answer the following:

A. Briefly describe of;

1. Search strategy

2. Data collection, quality assessment

3. Statistical analysis

B. Discuss the causality criteria as presented in the paper.

USE THE ATTACHTED ‘ARTICLE #2’ TO ANSWER THE QUESTIONS BELOW

Q2: Review the study titled “Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease”.

A. Describe the findings of this study as presented in figure 2.

B. Describe the publication bias as presented in figure 3.

C. Do you agree with the conclusion of the study? Justify your answer

USE THE ATTACHTED ‘ARTICLE #3’ TO ANSWER THE QUESTIONS BELOW

Q3: Review the study titled “Hydroxychloroquine in patients with COVID-19: A Systematic Review and metaanalysis”.

A. Briefly describe the search strategy and Inclusion criteria.

B. Describe the study results as presented in figures 2 and 3.

C. What are the limitations of the study?

D. Do you agree with the conclusion of the study? Justify your answer

See corresponding editorial on page 497.

Meta-analysis of prospective cohort studies evaluating the association
of saturated fat with cardiovascular disease1–

Patty W Siri-Tarino, Qi Sun, Frank B Hu, and Ronald M Kraus

ABSTRACT
Background: A reduction in dietary saturated fat has generally
been thought to improve cardiovascular health.
Objective: The objective of this meta-analysis was to summarize
the evidence related to the association of dietary saturated fat with
risk of coronary heart disease (CHD), stroke, and cardiovascular
disease (CVD; CHD inclusive of stroke) in prospective epidemio-
logic studies.
Design: Twenty-one studies identified by searching MEDLINE and
EMBASE databases and secondary referencing qualified for inclu-
sion in this study. A random-effects model was used to derive
composite relative risk estimates for CHD, stroke, and CVD.
Results: During 5–23 y of follow-up of 347,747 subjects, 11,006
developed CHD or stroke. Intake of saturated fat was not associated
with an increased risk of CHD, stroke, or CVD. The pooled relative
risk estimates that compared extreme quantiles of saturated fat in-
take were 1.07 (95% CI: 0.96, 1.19; P = 0.22) for CHD, 0.81 (95

CI: 0.62, 1.05; P = 0.11) for stroke, and 1.00 (95% CI: 0.89, 1.11;
P = 0.95) for CVD. Consideration of age, sex, and study quality did
not change the results.
Conclusions: A meta-analysis of prospective epidemiologic studies
showed that there is no significant evidence for concluding that
dietary saturated fat is associated with an increased risk of

CHD

or CVD. More data are needed to elucidate whether CVD risks are
likely to be influenced by the specific nutrients used to replace
saturated fat. Am J Clin Nutr 2010;91:535–46.

INTRODUCTION

Early animal studies showed that high dietary saturated fat and
cholesterol intakes led to increased plasma cholesterol concen-
trations as well as atherosclerotic lesions (1). These findings were
supported by associations in humans in which dietary saturated
fat correlated with coronary heart disease (CHD) risk (2, 3). More
recent epidemiologic studies have shown positive (4–10), inverse
(11, 12), or no (4, 13–18) associations of dietary saturated fat with
CVD morbidity and/or mortality.

A limited number of randomized clinical interventions have
been conducted that have evaluated the effects of saturated fat on
risk of CVD.Whereas some studies have shown beneficial effects
of reduced dietary saturated fat (19–21), others have shown no
effects of such diets on CVD risk (22, 23). The studies that
showed beneficial effects of diets reduced in saturated fat
replaced saturated fat with polyunsaturated fat, with the impli-
cation that the CVD benefit observed could have been due to an
increase in polyunsaturated fat or in the ratio of polyunsaturated

fat to saturated fat (P:S), a hypothesis supported by a recent
pooling analysis conducted by Jakobsen et al (24).

The goal of this study was to conduct a meta-analysis of well-
designed prospective epidemiologic studies to estimate the risk of
CHD and stroke and a composite risk score for both CHD and
stroke, or total cardiovascular disease (CVD), that was associated
with increased dietary intakes of saturated fat. Large prospective
cohort studies can provide statistical power to adjust for cova-
riates, thereby enabling the evaluation of the effects of a specif

nutrient on disease risk. However, such studies have caveat

including a reliance on nutritional assessment methods whose
validity and reliability may vary (25), the assumption that diets
remain similar over the long term (26) and variable adjustment
for covariates by different investigators. Nonetheless, a summary
evaluation of the epidemiologic evidence to date provides im-
portant information as to the basis for relating dietary saturated
fat to CVD risk.

SUBJECTS AND METHODS

Study selection

Two investigators (QS and PS-T) independently conducted
a systematic literature search of the MEDLINE (http://www.ncbi.
nlm.nih.gov/pubmed/) and EMBASE (http://www.embase.com

)

databases through 17 September 2009 by using the following
search terms: (“saturated fat” or “dietary fat”) and (“coronary” or
“cardiovascular” or “stroke”) and (“cohort” or “follow up”).

1 From the Children’s Hospital Oakland Research Institute, Oakland, CA

(PWS-T and RMK), and the Departments of Nutrition (QS and FBH) and

Epidemiology (FBH), Harvard School of Public Health, Boston, MA.
2 PWS-T and QS contributed equally to this work.
3 The contents of this article are solely the responsibility of the authors

and do not necessarily represent the official view of the National Center for

Research Resources (http://www.ncrr.nih.gov) or the National Institutes of

Health.
4 Supported by the National Dairy Council (PWS-T and RMK) and made

possible by grant UL1 RR024131-01 from the National Center for Research

Resources, a component of the National Institutes of Health (NIH), and NIH

Roadmap for Medical Research (PWS-T and RMK). QS was supported by

a Postdoctoral Fellowship from Unilever Corporate Research. FBH was

supported by NIH grant HL60712.
5 Address correspondence to RM Krauss, Children’s Hospital Oakland

Research Institute, 5700 Martin Luther King Junior Way, Oakland, CA

94609. E-mail: rkrauss@chori.org.

Received March 6, 2009. Accepted for publication November 25, 2009.

First published online January 13, 2010; doi: 10.3945/ajcn.2009.27725.

Am J Clin Nutr 2010;91:535–46. Printed in USA. � 2010 American Society for Nutrition 535

Studies were eligible if 1) data related to dietary consumption of
saturated fat were available; 2) the endpoints were nonfatal or
fatal CVD events, but not CVD risk factors; 3) the association of
saturated fat with CVD was specifically evaluated; 4) the study
design was a prospective cohort study; and 5) study participants
were generally healthy adults at study baseline. The initial
search yielded 661 unique citations, of which 19 studies met the
inclusion criteria and were selected as appropriate for inclusion
in this meta-analysis (Figure 1) (4–6, 8–11, 14–16, 18, 27–34).
All 4 investigators participated in the selection process (PS-T,
QS, FBH, and RMK). Additional reference searches were per-
formed by conducting a hand review of references from re-
trieved articles, and 3 more studies were identified (13, 17, 35).
Of the 22 identified studies, 10 studies provided data appropriate
to the constraints of this meta-analysis, specifically, relative risk
(RR) estimates for CVD as a function of saturated fat intake (10,
14–16, 28, 30–34). Where such data were not provided (n = 12),
data requests were made to investigators. Investigators from 6 of
the studies (4, 5, 8, 18, 29, 35) responded with the requested
data. A data set of the Honolulu Heart Study (9), obtained from
the National Heart, Lung, and Blood Institute (36), was used to
derive the RR estimates for this study. Investigators from 5 studies
either did not respond or could no longer access data sets (6, 11,
13, 17, 27). However, 4 of 5 of these studies provided data for
saturated fat intake as a continuous variable (6, 11, 13, 17). For
these studies, we derived RR estimates (see Statistical analysis)
for categorical saturated fat intake and CHD and/or stroke. The
study by Boden-Albala et al (27) was excluded because the RR
estimate published did not correspond to the values given for the
upper and lower bounds of the CI, ie, using these values to
derive the SE resulted in different estimates, and the authors did
not respond to our attempts to obtain the correct data.

Altogether, this meta-analysis included data from 21 unique
studies, with 16 studies providing risk estimates for CHD and
8 studies providing data for stroke as an endpoint. Data were de-
rived from347,747participants, ofwhom11,006 developedCVD.

Data extraction

Two authors (PS-T and QS) independently extracted and tab-
ulated data from each study using a standard extraction form.
Discrepancies were resolved via review of the original articles and
group discussion. From each study, we extracted information on
first author, publication year, disease outcome, country of origin,
method of outcome ascertainment, sample size, age, sex, average
study follow-up time, number of cases, dietary assessmentmethod
and the validity of the method, number of dietary assessments,
covariates adjusted, unit of measurement, RR of CVD comparing
extreme quantiles of saturated fat intake or per unit of saturated fat
intake, and corresponding 95% CIs, SEs, or exact P values.

Statistical analysis

RRs and 95% CIs were log transformed to derive corre-
sponding SEs for b-coefficients by using Greenland’s formula
(37). Otherwise, we used exact P values to derive SEs where
possible. To minimize the possibility that the association for
saturated fat intake may be influenced by extreme values, we
contacted the authors of studies in which RRs were presented as
per-unit increments of saturated fat intake and requested RRs
comparing extreme quantiles. For studies for which we did not
receive responses, we used the published trend RRs comparing
the 25th and 75th percentiles to estimate RRs comparing high
with low dichotomized saturated fat intakes (38).

Meta-analyses were performed by using STATA 10.0 (Stata-
Corp, College Station TX) and Review Manager 5.0 (The Nordic
Cochrane Centre, The Cochrane Collaboration, Copenhagen,
Denmark; http://www.cc-ims.net/RevMan). P , 0.05 was con-
sidered statistically significant. Random-effects models taking
into account both within-study and between-study variability
were used to estimate pooled RRs for associations of dietary
saturated fat with CHD risk. Relative to fixed-effects models,
random-effects models were more appropriate for the current
study because test statistics showed evidence of heterogeneity
among these studies. When several RRs were given for subgroup
analyses within a single study, random-effects models were used
to pool the RRs into one composite estimate.

We used the STATA METAINF module to examine the in-
fluence of an individual study on the pooled estimate of RR by
excluding each study in turn. We used the STATA METAREG
module to examine whether the effect size of these studies
depended on certain characteristics of each study, including age,
sex, sample size, duration of follow-up, whether disease out-
comes were confirmed by medical record review, and a score
evaluating overall study quality. This quality score was derived
from the following information: dietary assessment method
(where 5 points were given for diet records, 4 for validated FFQs,
3 for FFQs that were not formally validated, 2 for diet history, and
1 for 24-h recall), number of dietary assessments, and number of
adjusted established risk factors for CVD. Points were totaled to
construct a composite quality score for each study.

We further conducted secondary analyses to examine age- and
sex-specific effects (ie, age ,60 y compared with �60 y and

FIGURE 1. Study selection process. CHD, coronary heart disease; CVD,
cardiovascular disease; RR, relative risk. 1Three studies provided

outcome

data for both CHD and stroke.

536 SIRI-TARINO ET AL

T
A
B
L
E

S
tu
d
y

d
es
ig
n
ch
ar
ac
te
ri
st
ic
s

o
f

1
u
n
iq
u
e
p
ro
sp
ec
ti
ve

ep
id
em

io
lo
g
ic

st
ud
ie
s

o
f

sa
tu
ra
te
d
fa
t

ta
k
e

an
d
ri
sk

o
f
co
ro
n
ar
y
h
ea
rt
d
is
ea
se

(

C
H
D

)
o
r

st
ro
ke

1
S
tu
d
y

D
is
ea
se

o
u
tc
o
m
e

D
ie
ta
ry

as
se
ss
m

en
t

m
et
h
o
d

F
F
Q

it
em

V
al
id
at
io
n

o
f

n
u
tr
ie
n
ts

C
o
rr
el
at
io
n

w
it
h
d
ie
t

re
co
rd

2
V
al
id
at
io
n

in
st
ud
y

p
o
p
ul
at
io
n

M
u
lt
ip
le

as
se
ss
m
en
t

o
f
d
ie
t

A
ss
es
sm

en
t

in
te
rv
al

A
ve
ra
ge

d
u
ra
ti
o
n
o
f

fo
ll
ow

-u
p

y
y

C
H
D

S
h
ek
el
le

et
al
,
1
9
8
1
(1
7
)

F
at

C
H
D

In
te
rv
ie
w

(2
8

d
)

1
9
5

N
o
t
va
li
d
at
ed

—
—

B
as
el
in
e

—

2
0

M
cG

ee
et

al
,
1
9
8

(9
)

T
o
ta
l
C
H
D

2
4
-h

re
ca
ll
(1

d
)

—
N
A

—
—
B
as
el
in
e

—
1
0

K
u
sh
i
et

al
,
1
9
8
5
(1
3
)

F
at
al

C
H
D

D
ie
t-

h
is
to
ry

(t
im

e
fr
am

e
u
n
k
no
w

n
)

—
N
o
t
va
li
d
at
ed

—
—
B
as
el
in
e

—
2
0

P
o
sn
er

et
al
,
1
9
9
1
(1
6
)

T
o
ta
l
C
H
D
2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e

—
1
6

F
eh
il
y
et

al
,
1
9
9
3
(2
8
)

T
o
ta
l
C
H
D

7
-d

d
ie
t
re
co
rd

—
N
A
—
—
B
as
el
in
e

—
5

G
o
ld
b
o
ur
t
et

al
,
1
9
9
3
(3
5
)3

F
at
al
C
H
D
F
F
Q
(t
im
e
fr
am

e
u
n
k
n
ow

n
)

N
R

N
o
t
va
li
d
at
ed
—
—
B
as
el
in
e

—
2
3

A
sc
h
er
io

et
al
,
1
9
9
6
(4
)

T
o
ta
l
C
H
D
F
F
Q

(1
y
)

1
3
1

V
al
id
at
ed

0
.7
5

Y
es

T
w
ic
e

1
4

E
sr
ey

et
al
,
1
9
9
6
(6
)

F
at
al
C
H
D
2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e

—
1
2

M
an
n
et

al
,
1
9
9
7
(3
2
)

F
at
al
C
H
D
F
F
Q
(t
im

e
in
te
rv
al

u
n
k
no
w
n
)

N
R
N
o
t
va
li
d
at
ed
—
—
B
as
el
in
e

—
1
3

P
ie
ti
n
en

et
al
,
1
9
9
7
(1
5
)

T
o
ta
l
C
H
D
F
F
Q
(1
y
)

2
7
6

V
al
id
at
ed

0
.7
0

Y
es
B
as
el
in
e

—
6

B
on
if
ac
e
an
d
T
ef
ft
,
2
0
0
2
(5
)

F
at
al
C
H
D
F
F
Q
(t
im
e
fr
am
e
u
n
k
n
ow
n
)

3
0

V
al
id
at
ed

0
.3
4

Y
es
B
as
el
in
e
—
1
6

Ja
k
o
bs
en

al
,
2
0
0
4
(8
)

T
o
ta
l
C
H
D
7
-d
d
ie
t
re
co
rd
s
—
N
A
—
—
B
as
el
in
e
—
1
6

L
eo
sd
o
tt
ir
et

al
,
2
0
0
7
(1
4
)3

F
at
al
C
H
D
F
F
Q

(1
y
)
an
d
7
-d

m
en
u
-d
ia
ry

1
6
8

V
al
id
at
ed

0
.5
0

Y
es
B
as
el
in
e

—
8

O
h
et

al
,
2
0
0
5
(3
3
)

T
o
ta
l
C
H
D
F
F
Q
(1
y
)

1
1
6

V
al
id
at
ed

0
.6
8

Y
es

6
ti
m
es

2
–
4

2
0

T
u
ck
er

et
al
,
2
0
0
5
(

1
8

)

F
at
al
C
H
D
7
-d
d
ie
t
re
co
rd
s
—
N
A
—
—

4
ti
m
es

3
–
9

1
8

X
u
et

al
,
2
0
0
6
(1
0
)

T
o
ta
l
C
H
D
2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e

—
7

S
tr
o
k
e

M
cG
ee
et
al
,
1
9
8
4
(9
)3

T
o
ta
l
st
ro
ke

2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e
—
1
0
G
o
ld
b
o
ur
t
et
al
,
1
9
9
3
(3
5
)3
F
at
al

st
ro
k
e

F
F
Q
(t
im
e
fr
am
e
u
n
k
n
ow
n
)
N
R
N
o
t
va
li
d
at
ed
—
—
B
as
el
in
e
—
2
3

G
il
lm

an
et

al
,
1
9
9
7
(1
1
)

Is
ch
em

ic
st
ro
k
e

2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e
—
2
0

Is
o
et

al
,
2
0
0
1
(3
1
)

H
em

or
rh
ag
ic

st
ro
ke
F
F
Q
(1
y
)
1
1
6
V
al
id
at
ed
0
.6
8
Y
es
4
ti
m
es
2
–
4
1
4

H
e
et

al
,
2
0
0
3
(2
9
)

T
o
ta
l
st
ro
ke
F
F
Q
(1
y
)
1
3
1
V
al
id
at
ed
0
.7
5
Y
es

3
ti
m
es

4
1
4

Is
o
et

al
,
2
0
0
3
(3
0
)

H
em
or
rh
ag
ic
st
ro
ke
2
4
-h
re
ca
ll
(1
d
)
—
N
A
—
—
B
as
el
in
e

—
1
4

S
au
va
g
et

et
al
,
2
0
0
4
(3
4
)

F
at
al
st
ro
k
e

1
-d

d
ie
t
re
co
rd
—
N
A
—
—
B
as
el
in
e
—
1
4
L
eo
sd
o
tt
ir
et
al
,
2
0
0
7
(1
4
)3
Is
ch
em
ic
st
ro
k
e
F
F
Q
(1
y
)
an
d
7
-d
m
en
u
-d
ia
ry
1
6
8
V
al
id
at
ed
0
.5
0
Y
es
B
as
el
in
e
—
8

1
F
F
Q
,
fo
o
d
-f
re
q
u
en
cy

q
u
es
ti
o
n
n
ai
re
;
N
R
,
n
o
t
re
p
o
rt
ed
;
N
A
,
n
o
t
ap
p
li
ca
b
le
.

2
O
n
ly

p
er

ta
in
s
to

st
ud
ie
s
th
at

u
se
d
an

F
F
Q

to
m
ea
su
re

sa
tu
ra
te
d
fa
t
in
ta
k
e.

3
T
h
es
e
st
ud
ie
s
p
ro
v
id
ed

b
o
th

C
H
D

an
d
st
ro
k
e
o
u
tc
o
m
e
d
at
a.

META-ANALYSIS OF SATURATED FAT AND CVD 537

male compared with female) of saturated fat on CVD risk. These
secondary analyses were performed with only those studies that
provided stratified data according to these variables.

To examine the effects of replacing saturated fat with car-
bohydrate or polyunsaturated fat, we performed secondary meta-
analyses with studies that provided pertinent and extractable data.
Because energy from carbohydrate intake was excluded in fully
adjusted models (including adjustments for total energy and
energy from protein and fats other than saturated fat) in 6 of 21
studies, the regression coefficients of saturated fat could be
interpreted as the effects of isocalorically replacing carbohydrate
intakewith saturated fat (25). Similarly, there were 5 studies (4, 9,
29, 31, 33) in which the regression coefficients of saturated fat
could be interpreted as the effects of isocalorically replacing
polyunsaturated fat intake with saturated fat. Finally, because
consideration for total energy intake has been shown to be rel-
evant in the evaluation of nutrient-disease associations (39), we
also performed a subanalysis of studies (n = 15) that provided
total energy intake data (4, 6, 8–11, 15, 16, 18, 29–31, 33). Begg
funnel plots were used to assess potential publication bias (40).

RESULTS

The study design characteristics of the 21 studies identified by
database searches and secondary referencing that were included
in this meta-analysis (4–6, 8–11, 13–18, 28–35) are shown in

Table 1. Altogether, there were 16 studies that considered the
association of saturated fat with CHD and 8 studies that evalu-
ated the association of saturated fat with stroke. Dietary as-
sessments included 24-h recalls, food-frequency questionnaires
(FFQs), and multiple daily food records. The duration of follow-
up ranged from 6 to 23 y, with a mean and median follow-up of
14.3 and 14 y, respectively.

The baseline characteristics of the study participants are
provided in Table 2. The number of subjects in each study
ranged from 266 to 85,764. The age of participants ranged from
’30 to 89 y. There were 11 studies conducted exclusively in
men, 2 studies conducted exclusively in women, and 8 studies
that enrolled both men and women. There were 12 studies that
were conducted in North America, 6 in Europe, 2 in Japan, and
1 in Israel.

The level of adjustment for covariates varied according to
study (Table 3). Wherever possible, risk estimates from the
most fully adjusted models were used in the estimation of the
pooled RR. Studies that reported positive associations be-
tween saturated fat and CVD risk included the Lipid Research
Clinics Study (6), the Health Professionals Follow-Up

Study

(4), the Health and Lifestyle Survey (5), the Strong Heart Study
(10), and studies by Mann et al (32) and Jakobsen et al (8).
Notably, these positive associations were specific to subsets of
the study population, ie, younger versus older (6, 8, 10), women
versus men (5, 8), or for some, but not all, CHD endpoints (4).

TABLE 2

Baseline characteristics of participants of 21 unique prospective epidemiologic studies of saturated fat intake and risk of coronary heart disease (CHD) or

stroke1

Study

No. of

subjects Age Sex

Country of

residence Smoker

Disease

outcome

Method of

diagnosis

y %

CHD

Shekelle et al, 1981 (17) 1900 40–55 Male USA NR Fatal CHD DC

McGee et al, 1984 (9)2 8006 45–68 Male USA NR Total CHD MR and DC

Kushi et al,1985 (13) 1001 30–69 Male USA NR Fatal CHD DC

Posner et al, 1991 (16) 813 45–65 Male USA 41.6 Total CHD MR

Fehily et al, 1993 (28) 512 45–59 Male UK NR Total CHD DC

Goldbourt et al, 1993 (35)2 97673 �40 Male Israel NR Fatal CHD MR and DC
Ascherio et al, 1996 (4) 38,463 40–75 Male USA 9.5 Total CHD MR and DC

Esrey et al, 1996 (6) 4546 30–79 Both Canada 33.9 Fatal CHD MR

Mann et al, 1997 (32) 10,802 16–79 Both UK 19.5 Fatal CHD MR

Pietinen et al, 1997 (15) 21,930 50–69 Male Finland 100 Total CHD MR and DC

Boniface and Tefft, 2002 (5) 2676 40–75 Both UK NR Total CHD DC

Jakobsen et al, 2004 (8) 3686 30–71 Both Denmark NR Total CHD MR

Leosdottir et al, 2007 (14)2 28,098 45–73 Both Sweden 29.5 Total CHD DC

Oh et al, 2005 (33) 78,778 30–55 Female USA NR Total CHD MR and DC

Tucker et al, 2005 (18) 2663 34–80 Male USA 21.8 Fatal CHD MR and DC

Xu et al, 2006 (10) 2938 47–79 Both USA 29.7 Total CHD MR

Stroke

McGee et al, 1984 (9)2 8006 45–68 Male USA NR Total stroke MR and DC

Goldbourt et al, 1993 (35)2 97673 �40 Male Israel NR Fatal stroke MR and DC
Gillman et al, 1997 (11) 832 45–65 Male USA NR Ischemic stroke MR

Iso et al, 2001 (31) 85,764 34–59 Female USA NR Hemorrhagic stroke MR and DC

He et al, 2003 (29) 38,4633 40–75 Male USA 9.5 Total stroke MR and DC

Iso et al, 2003 (30) 4775 40–69 Both Japan 13.5 Hemorrhagic stroke MR and DC

Sauvaget et al, 2004 (34) 3731 35–89 Both Japan 26–35 Ischemic stroke DC

Leosdottir et al, 2007 (14)2 28,098 45–73 Both Sweden 29.5 Ischemic stroke DC

1 NR, not reported; MR, medical records; DC, death certificate.
2 These studies provided both CHD and stroke outcome data.
3 These numbers, as provided by the respective investigators, differ from those used in the original publications.

538 SIRI-TARINO ET AL

T
A
B
L
E
3

R
el
at
iv
e
ri
sk

(R
R
)
es
ti
m
at
es

fo
r
th
e
as
so
ci
at
io
n
o
f

sa
tu
ra
te
d
fa
t
in
ta
k
e

an
d
ri
sk

o
f
co
ro
n
ar
y
h
ea
rt
d
is
ea
se

(C
H
D
)
o
r
st
ro
ke

in
2
1
u
n
iq
u
e
p
ro
sp
ec
ti
ve

ep
id
em
io
lo
g
ic

st
ud
ie
s1

S
tu
d
y

S
ex

C
as
es

M
ed
ia
n
o
r
m
ea
n

sa
tu
ra
te
d
fa
t
in
ta
k
e

A
d
ju
st
ed

co
va
ri
at
es

M
u
lt
iv
ar
ia
te

ad
ju
st
ed

R
R
(9
5
%

C
I)

C
o
ro
n
ar
y
h
ea
rt
d
is
ea
se

st
ud
ie
s
S
h
ek
el
le
et
al
,
1
9
8
1
(1
7
)

(W
es
te
rn

E
le
ct
ri
c

S
tu
d
y)

al
e

F
at
al

C
H
D
:
2
1
5

1
6
.6
%

o
f

to
ta
l

en
er
g
y

A
g
e,

S
B
P,

ci
g
ar
et
te

s
p
er

d
ay
,

se
ru
m

ch
o
le
st
er
o
l,

al
co
ho
li
c
d
ri
n
k
s

p
er

m
o
n
th
,

B
M
I,

g
eo
g
ra
p
h
ic

o
ri
g
in

b
=
0
.0
31
,
P
=
0
.1
44

F
o
r
1
-u
n
it

in
cr
ea
se

in
sa
tu
ra
te
d
fa
t

M
cG
ee
et

al
,
1
9
8
4
(9
)

(H
o
n
o
lu
lu

H
ea
rt
S
tu
d
y
)2

M
al
e

T
o
ta
l
C
H
D
:
1
1
7
7

1
2
.7
%

o
f
to
ta
l

en
er
g
y

(a
g
e-
ad
ju
st
ed
)3

A
g
e,

to
ta
l
en
er
g
y

in
ta
k
e,

S
B
P,
B
M
I,

o
k
in
g
,

fa
m
il
y
h
is
to
ry

o
f
M
I,
p
h
y
si
ca
l

ac
ti
v
it
y,

in
ta
k
es

o
f

U
FA

,
al
co
ho
l,

p
ro
te
in
,

ca
rb
o
h
y
dr
at
e,

ve
g
et
ab
le

an
d
ch
o
le
st
er
o
l

R
R
m
e
n
,

6
0

y
=
0
.9
2
(0
.6
8
,
1
.2
3)

R
R
m
e
n
�

6
0
y
=
0
.7
0
(0
.4
1
,
1
.2
0)

P
o
ol
ed

R
R
=
0
.8
6
(0
.

6
7

,
1
.1
2)

P
fo
r
te
st

o
f
h
et
er
o
ge
n
ei
ty

=
0
.3
9

F
o
r
fi
ft
h

v
s
fi
rs
t
q
u
in
ti
le

K
u
sh
i
et
al
,
1
9
8
5
(1
3
)

(I
re
la
n
d
B
os
to
n
D
ie
t
H
ea
rt
S
tu
d
y)

M
al
e
F
at
al

C
H
D
:
1
1
0

1
6
.8
%

o
f
to
ta
l
en
er
g
y

A
g
e,
S
B
P,
se
ru
m

ch
o
le
st
er
o
l,
ci
g
ar
et
te

sm
o
k
in
g
,
al
co
ho
l
in
ta
k
e,

co
h
o
rt

b
=
0
.0
61
,
P

=
0
.0
5

F
o
r
1
-u
n
it
in
cr
ea
se
in
sa
tu
ra
te
d
fa
t
P
o
sn
er
et
al
,
1
9
9
1
(1
6
)

(F
ra
m
in
gh
am

S
tu
d
y
)

M
al
e

T
o
ta
l
C
H
D
:
2
1

4
5
–
55

y
o
ld
:

1
5
.2
%

o
f
to
ta
l
en
er
g
y
3

5
6
–
65

y
o
ld
:

1
4
.8
%

o
f
to
ta
l
en
er
g
y
3

V
ar
ia
b
le

o
f
in
te
re
st
,

en
er
g
y
in
ta
k
e,

p
h
y
si
ca
l
ac
ti
v
it
y,

se
ru
m
ch
o
le
st
er
o
l,
S
B
P,

le
ft

ve
n
tr
ic
u
la
r
h
y
p
er
tr
o
p
h
y,

ci
g
ar
et
te

sm
o
k
in
g
,

g
lu
co
se

in
to
le
ra
n
ce

M
et
ro
p
o
li
ta
n
re
la
ti
ve

w
ei
g
ht

R
R
4
5
–
5
5
y
=
0
.7
8
(0
.

6
1

,
1
.0
0
)

R
R
�
5
6
y
=
1
.0
6
(0
.8
6
,
1
.3
0)

P
o
ol
ed

R
R
=
0
.9
2
(0
.6
8
,
1
.2
4)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.0
6

F
o
r
re
co
m
m
en
d
ed

ve
rs
u
s
ac
tu
al

in
ta
k
e
(1
5
.2
%

v
s
1
0
%
)

F
eh
il
y
et
al
,
1
9
9
3
(2
8
)

(C
ae
rp
h
il
ly

S
tu
d
y)
M
al
e
T
o
ta
l
C
H
D
:
2
1

1
7
.3
%

fo
r
C
H
D

-f
re
e

su
b
je
ct
s
an
d
1
8
.1
%

fo
r
C
H
D

ca
se
s

o
n
e

R
R
=
1
.5
7
(0
.5
6
,
4
.4
2)

F
o
r
th
ir
d
v
s
fi
rs
t
te
rt
il
e

G
o
ld
b
o
ur
t
et

al
,
1
9
9
3
(3
5
)

(I
sr
ae
li
Is
ch
em

ic
H
ea
rt

D
is
ea
se
S
tu
d
y)
M
al
e
F
at
al

C
H
D
:
1
0
7
0

N
R
A
g
e,

b
lo
o
d
p
re
ss
u
re

se
ru
m
ch
o
le
st
er
o
l,

ev
er
-s
m

o
k
in
g
,
d
ia
b
et
es

p
re
va
le
n
ce

in
1
9
6
3

R
R
m
e
n
,
6
0
y
=
1
.0
5
(0
.8
7
,
1
.2
7)

R
R
m
e
n
�
6
0
y
=
0
.6
6
(0
.
4
4
,
1
.0
0)

4
P
o
ol
ed

R
R
=
0
.8
6
(0
.5
6
,
1
.3
5)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty
=
0
.0
5

F
o
r
fo
u
rt
h

v
s
fi
rs
t
q
u
ar
ti
le

A
sc
h
er
io
et
al
,
1
9
9
6
(4
)

(H
ea
lt
h
P
ro
fe
ss
io
n
al
s

F
o
ll
ow

-U
p
S
tu
d
y
)

M
al
e

T
o
ta
l
C
H
D
:
1
7
0
2

F
if
th

q
u
in
ti
le
:
1
4
.8
%

o
f
to
ta
l
en
er
g
y

F
ir
st
q
u
in
ti
le
:
7
.2
%

o
f
to
ta
l
en
er
g
y
A
g
e,

B
M
I,
sm

o
k
in
g
,
p
h
y
si
ca
l

ac
ti
v
it
y,
h
is
to
ry

o
f
h
y
p
er
te
n
si
o
n

o
r
h
ig
h
b
lo
o
d
ch
o
le
st
er
o
l,
h
is
to
ry

o
f
M
I
,

ag
e
6
0
y,

en
er
gy

in
ta
k
e,

fi
b
er

R
R
m
e
n
,
6
0
y
=
1
.2
4
(0
.8
7
,
1
.7
7)

R
R
m
e
n
�
6
0
y
=
1
.0
1
(0
.7
3
,
1
.4
1)

4
P
o
ol
ed

R
R
=
1
.1
1
(0
.8
7
,
1
.4
2)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.4
2

F
o
r
fi
ft
h
v
s
fi
rs
t
q
u
in
ti
le
E
sr
ey
et
al
,
1
9
9
6
(6
)

(L
ip
id

R
es
ea
rc
h
C
li
n
ic
s
S
tu
d
y)

B
o
th

F
at
al

C
H
D
:
9
2

3
0
–
59

y
o
ld
:
1
6
.8
%
fo
r
C
H
D

d
ea
th
s

an
d
1
5
.1
%

fo
r
n
o
n
-C
H
D

d
ea
th
s3

6
0
–
79

y
o
ld
:

1
3
.8
%

fo
r
C
H
D

d
ea
th
s
an
d
1
4
.3
%

fo
r
n
o
n
-C
H
D
d
ea
th
s3
A
g
e,

se
x
,
en
er
g
y
in
ta
k
e,

se
ru
m

li
p
id
s,

S
B
P,
ci
g
ar
et
te

sm
o
k
in
g

st
at
u
s,

B
M
I,

g
lu
co
se
in
to
le
ra
n
ce

R
R
,
6
0
y
=
1
.1
1
(1
.0
4
,
1
.1
8)

R
R
�
6
0
y
=
0
.9
6
(0
.8
8
,
1
.0
5)

P
o
ol
ed

R
R
=
0
.9
7
(0
.8
0
,
1
.1
8)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.4
0

F
o
r
1
-u
n
it
in
cr
ea
se
in
sa
tu
ra
te
d
fa
t
M
an
n
et
al
,
1
9
9
7
(3
2
)
B
o
th
F
at
al

C
H
D
:
4
5

M
en
:

T
h
ir
d
te
rt
il
e,

4
1
.0

g
/d

;

F
ir
st
te
rt
il
e,

1
4
.6

g
/d

W
o
m
en
:

T
h
ir
d
te
rt
il
e,

3
8
.1

g
/d
;

F
ir
st
te
rt
il
e,

1
3
.7

g
/d
A
g
e,

se
x
,
sm

o
k
in
g
,

so
ci
al

cl
as
s

R
R
=
2
.7
7
(1
.2
5
,
6
.1
3)

5
F
o
r
th
ir
d
v
s
fi
rs
t
te
rt
il
e

(C
o
nt
in
u
ed
)

META-ANALYSIS OF SATURATED FAT AND CVD 539

T
A
B
L
E
3
(C

on
ti
n
u
ed

)
S
tu
d
y
S
ex
C
as
es
M
ed
ia
n
o
r
m
ea
n
sa
tu
ra
te
d
fa
t
in
ta
k
e
A
d
ju
st
ed
co
va
ri
at
es
M
u
lt
iv
ar
ia
te
ad
ju
st
ed
R
R
(9
5
%
C
I)
P
ie
ti
n
en
et
al
,
1
9
9
7
(1
5
)

(A
lp
h
a-
T
o
co
p
he
ro
l,

B
et
a-
C
ar
o
te
n
e
S
tu
d
y
)

M
al
e

T
o
ta
l
C
H
D
:
6
3
5

F
if
th

q
u
in
ti
le
:
6
7
.5

g
/d

F
ir
st
q
u
in
ti
le
:
3
4
.7

g
/d
A
g
e,

tr
ea
tm

en
t
g
ro
u
p,

sm
o
k
in
g
,
B
M
I,

b
lo
o
d
p
re
ss
u
re
,
ed
u
ca
ti
o
n
,
in
ta
k
es

o
f

en
er
g
y,

al
co
ho
l,
fi
b
er
,
p
h
y
si
ca
l
ac
ti
v
it
y,

in
ta
k
es

o
f
li
n
o
le
ic

ac
id

an
d
tr
a
n
s
an
d

m
o
n
o
un
sa
tu
ra
te
d
fa
ts

R
R
=
0
.9
3
(0
.6
0
,
1
.4
4)

F
o
r
fi
ft
h
v
s
fi
rs
t
q
u
in
ti
le

B
o
n
if
ac
e
an
d
T
ef
ft
,
2
0
0
2
(5
)

(H
ea
lt
h
an
d
L
if
es
ty
le

S
u
rv
ey
)

B
o
th
F
at
al

C
H
D
:
1
5
5

M
en
:
4
7
.0

g
/d
3

W
o
m
en
:
3
4
.4

g
/d

3
A
g
e,

al
co
ho
l,
sm

o
k
in
g
,
ex
er
ci
se
,

so
ci
al
cl
as
s

R
R
m
e
n
,
6
0
y
=
1
.5
1
(0
.6
9
,
3
.3
1)

R
R
m
e
n
�
6
0
y
=
1
.0
1
(0
.5
7
,
1
.8
0)

R
R
w
o
m
en

,
6
0
y
=
1
.3
2
(0
.3
8
,
4
.5
7)

4
R
R
w
o
m
en

�
6
0
y
=
2
.3
4
(1
.0
2
,
5
.4
0
)4
,5

P
o
ol
ed

R
R
=
1
.3
7
(1
.1
7
,
1
.6
5)

5
P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.4
4

F
o
r
th
ir
d
te
rt
il
e
v
s
fi
rs
t
te
rt
il
e

Ja
k
o
b
se
n
et

al
,
2
0
0
4
(8
)
B
o
th

T
o
ta
l
C
H
D
:
3
2
6

M
en
:

1
9
.7
%

o
f
to
ta
l
en
er
g
y
W
o
m
en
:

1
9
.5
%

o
f
to
ta
l
en
er
g
y
F
at

in
ta
k
e
as

to
ta
l
en
er
g
y
in
ta
k
e,

co
h
o
rt
id
en
ti
fi
ca
ti
o
n
,
%

en
er
g
y
p
ro
te
in
,
%

en
er
g
y

o
th
er

fa
tt
y

ac
id
s,
fa
m
il
y
h
is
to
ry

o
f
M
I,
sm

o
k
in
g
,
p
h
y
si
ca
l
ac
ti
v
it
y,

ed
u
ca
ti
o
n
,
al
co
ho
l,

fi
b
er
,
ch
o
le
st
er
o
l,
S
B
P,

B
M
I

R
R
w
o
m
e
n
,
6
0
y
=
4
.7
8
(0
.9
5
,
2
4
.1
0
)4

R
R
w
o
m
en

�
6
0
y
=
1
.0
3
(0
.5
3
,
2
.0
0
)4

R
R
m
e
n
,
6
0
y
=
1
.0
1
(0
.4
8
,
2
.1
4)

R
R
m
e
n
�
6
0
y

=
0
.7
9

(0
.4
8
,
1
.2
9)

4
P
o
ol
ed

R
R
=
1
.0
3
(0
.6
6
,
1
.6
0)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.6
1

F
o
r
th
ir
d
te
rt
il
e
v
s
fi
rs
t
te
rt
il
e
L
eo
sd
o
tt
ir
et

al
,
2
0
0
7
(1
4
)

(M
al
m
o
D
ie
t
an
d
C
an
ce
r
S
tu
d
y
)

B
o
th

T
o
ta
l
C
H
D
:
9
0
8

M
en
:

F
o
u
rt
h
q
u
ar
ti
le
,
2
2
.3
%

o
f
to
ta
l
en
er
g
y
;

F
ir
st
q
u
ar
ti
le
,
1
2
.3
%

o
f
to
ta
l
en
er
g
y
W
o
m
en
:

F
o
u
rt
h
q
u
ar
ti
le
,
2
1
.8
%

o
f
to
ta
l
en
er
g
y
;

F
ir
st
q
u
ar
ti
le
,
1
2
.2
%

o
f
to
ta
l
en
er
g
y
A
g
e,

sm
o
k
in
g
h
ab
it
s,
al
co
ho
l

co
n
su
m
pt
io
n
,

so
ci
oe
co
n
o
m
ic

st
at
u
s,
m
ar
it
al

st
at
u
s,
p
h
y
si
ca
l
ac
ti
v
it
y,

B
M
I,
fi
b
er

in
ta
k
e,

an
d
b
lo
o
d
p
re
ss
u
re
.

R
R
w
o
m
e
n
=
0
.8
1
(0
.5
3
,
1
.2
4)

R
R
m
e
n
=
1
.0
2
(0
.7
6
,
1
.3
7
)

P
o
ol
ed

R
R
=
0
.9
5
(0
.

7
4

,
1
.2
1)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.3
8

F
o
r
fo
u
rt
h
v
s
fi
rs
t
q
u
ar
ti
le

O
h
et
al
,
2
0
0
5
(3
3
)

(N
u
rs
es
’
H
ea
lt
h
S
tu
d
y)

F
em

al
e

T
o
ta
l
C
H
D
:
1
7
6
6

F
if
th

q
u
in
ti
le
:
1
7
.6
%

o
f
to
ta
l
en
er
g
y
;

F
ir
st
q
u
in
ti
le
:
1
0
.1
%

o
f
to
ta
l
en
er
g
y
A
g
e,

B
M
I,
ci
g
ar
et
te

sm
o
k
in
g
,
al
co
h
o
l

in
ta
k
e,

p
ar
en
ta
l
h
is
to
ry

o
f
M
I,
h
is
to
ry

o
f
h
y
p
er
te
n
si
o
n,

m
en
o
p
au
sa
l

st
at
u
s,

h
o
rm

o
n
e
u
se
,

as
p
ir
in

u
se
,
m
u
lt
i

v
it
am

u
se
,
v
it
am

in
E
su
p
p
le
m
en
t
u
se
,

p
h
y
si
ca
l
ac
ti
v
it
y,
in
ta
k
es

o
f
en
er
g
y,

p
ro
te
in
,
ch
o
le
st
er
o
l,
M
U
FA

s,
P
U
FA

tr
a
n
s
fa
t;
a-
li
n
o
le
n
ic

ac
id
,
m
ar
in
e
n
2
3

fa
tt
y
ac
id
s,
ce
re
al

fi
b
er
,
an
d
fr
u
it

an
d
ve
ge
ta
b
le
s

R
R
=
0
.9
7
(0
.7
4
,
1
.2
7)

F
o
r
fi
ft
h
v
s
fi
rs
t
q
u
in
ti
le
T
u
ck
er

et
al
,
2
0
0
5
(1
8
)

(B
al
ti
m
o
re

L
o
n
gi
tu
d
in
al

S
tu
d
y
o
f
A
g
in
g
)

M
al
e
F
at
al

C
H
D
:
7
1

S
u
rv
iv
o
rs
:
1
2
.3
%

C
H
D

d
ea
th
s:

1
3
.8
%

O
th
er

d
ea
th
s:

1
4
.0

%
o
f
to
ta
l
en
er
g
y
3
A
g
e,
to
ta
l
en
er
g
y
in
ta
k
e,
B
M
I,
sm

o
k
in
g
,
al
co
h
o
l
u
se
,

p
h
y
si
ca
l
ac
ti
v
it
y
sc
o
re
,

su
p
pl
em

en
t
u
se
,
fr
u
it
an
d

ve
g
et
ab
le

in
ta
k
es
,
se
cu
la
r
tr
en
d

R
R
m
e
n
,
6
0
y
=
0
.5
7
(0
.1
4
,
2
.3
0)

R
R
m
e
n
�
6
0
y
=
2
.3
1
(0
.7
3
,
7
.2
7)

4
P
o
ol
ed

R
R
=
1
.2
2
(0
.3
1
,
4
.7
7)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.1
3

F
o
r
th
ir
d
te
rt
il
e
v
s
fi
rs
t
te
rt
il
e
(C
o
nt
in
u
ed
)

540 SIRI-TARINO ET AL

T
A
B
L
E
3
(C
on
ti
n
u
ed
)
S
tu
d
y
S
ex
C
as
es
M
ed
ia
n
o
r
m
ea
n
sa
tu
ra
te
d
fa
t
in
ta
k
e
A
d
ju
st
ed
co
va
ri
at
es
M
u
lt
iv
ar
ia
te
ad
ju
st
ed
R
R
(9
5
%
C
I)
X
u
et
al
,
2
0
0
6
(1
0
)

(S
tr
o
n
g
H
ea
rt
S
tu
d
y)

B
o
th

T
o
ta
l
C
H
D
:
1
3
8

F
o
u
rt
h
q
u
ar
ti
le
:

1
6
.5
%

o
f
to
ta
l
en
er
g
y

F
ir
st
q
u
ar
ti
le
:
7
.5
%

o
f
to
ta
l
en
er
g
y
V
ar
ia
b
le

o
f
in
te
re
st
as

%
o
f
en
er
g
y,

se
x
,
ag
e,

st
ud
y
ce
n
te
r,
d
ia
be
te
s

st
at
u
s,
B
M
I,
H
D
L
,
L
D
L
,
T
G
,

sm
o
k
in
g
,
al
co
ho
l
co
n
su
m
pt
io
n
,

h
y
p
er
te
n
si
o
n
,
en
er
gy

fr
om

p
ro
te
in
,

to
ta
l
en
er
g
y
in
ta
k
e

R
R
,
6
0
y
=
5
.1
7
(1
.6
0
,
1
6
.4
)5

R
R
�
6
0
y
=
0
.8
0
(0
.4
1
,
1
.5
4)

P
o
ol
ed

R
R
=
1
.9
1
(0
.3
1
,
1
1
.8
4
)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.0
0
6

F
o
r
fo
u
rt
h
q
u
ar
ti
le

v
s
fi
rs
t
q
u
ar
ti
le

S
tr
o
k
e
st
ud
ie
s

M
cG
ee
et
al
,
1
9
8
4
(9
)
(H
o
n
o
lu
lu
H
ea
rt
S
tu
d
y
)2
M
al
e

T
o
ta
l
st
ro
ke
:
4
9
2

1
2
.7
%
o
f
to
ta
l
en
er
g
y
(a
g
e-
ad
ju
st
ed
)3
A
g
e,
to
ta
l
en
er
g
y
in
ta
k
e,
S
B
P,
B
M
I,

sm
o
k
in
g
,
fa
m
il
y
h
is
to
ry

o
f
M
I,
p
h
y
si
ca
l
ac
ti
v
it
y,
in
ta
k
es
o
f
P
U
FA

s,
al
co
ho
l,

p
ro
te
in
,
ca
rb
o
h
y
dr
at
e,

ve
g
et
ab
le
s,

an
d
ch
o
le
st
er
o
l

R
R
m
e
n
,
6
0
y
=
0
.9
5
(0
.6
0
,
1
.5
0)

R
R
m
e
n
�
6
0
y
=
1
.2
3
(0
.6
6
,
2
.2
9)

2
P
o
ol
ed

R
R
=
1
.0
4
(0
.7
2
,
1
.5
0)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.5
2

F
o
r
fi
ft
h
v
s
fi
rs
t
q
u
in
ti
le
G
o
ld
b
o
ur
t
et
al
,
1
9
9
3
(3
5
)
(I
sr
ae
li
Is
ch
em
ic
H
ea
rt
D
is
ea
se
S
tu
d
y)
M
al
e
F
at
al

st
ro
ke
:
3
6
2

N
R
A
g
e,

b
o
d
y
h
ei
g
h
t,
b
lo
o
d

p
re
ss
u
re
,
sm

o
k
in
g
,
d
ia
b
et
es

R
R
m
e
n
,
6
0
y
=
0
.7
5
(0
.5
4
,
1
.0
5)

R
R
m
e
n
�
6
0
y
=
1
.2
6
(0
.7
0
,
2
.2
9)

4
P
o
ol
ed

R
R
=
0
.9
2
(0
.5
6
,
1
.5
1)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty
=
0
.1
3
F
o
r
fo
u
rt
h
q
u
ar
ti
le
v
s
fi
rs
t
q
u
ar
ti
le
G
il
lm
an
et
al
,
1
9
9
7
(1
1
)
(F
ra
m
in
gh
am
S
tu
d
y
)
M
al
e
Is
ch
em
ic

st
ro
k
e:

6
1

1
5
.0
%

o
f
to
ta
l
en
er
g
y
3
A
g
e,

to
ta
l
en
er
g
y,

S
B
P,
ci
g
ar
et
te

sm
o
k
in
g
,
g
lu
co
se

in
to
le
ra
n
ce
,

B
M
I,
p
h
y
si
ca
l
ac
ti
v
it
y,

le
ft

ve
nt
ri
cu
la
r
h
y
p
er
tr
o
p
h
y,

an
d
in
ta
k
es

o
f
al
co
ho
l
an
d

fr
u
it
an
d
ve
g
et
ab
le
s

R
R
is
c
h
e
m
ic

st
ro
k
e
=
0
.9
0
(0
.8
3
,
0
.9
6
)5

F
o
r
1
%

in
cr
ea
se
in
sa
tu
ra
te
d
fa
t
Is
o
et
al
,
2
0
0
1
(3
1
)
(N
u
rs
es
’
H
ea
lt
h
S
tu
d
y)
F
em
al
e
H
em

o
rr
ha
g
ic

st
ro
k
e:
7
4
F
if
th

q
u
in
ti
le
:
3
6
g
/d

F
ir
st
q
u
in
ti
le
:
2
0
g
/d

A
g
e,

sm
o
k
in
g
,
ti
m
e
in
te
rv
al
,

B
M
I,
al
co
ho
l
in
ta
k
e,

m
en
o
p
au
sa
l

st
at
u
s,
p
o
st
m
en
o
p
au
sa
l
h
o
rm

o
n
e

u
se
,
v
ig
o
ro
u
s
ex
er
ci
se
,
u
su
al

as
p
ir
in

u
se
,
m
u
lt
iv
it
am

in
s,

v
it
am

in
E
,
n
2
3
fa
tt
y
ac
id
s,

ca
lc
iu
m
,
to
ta
l
en
er
gy

in
ta
k
e,

q
u
in
ti
le
s
o
f
ch
o
le
st
er
o
l,
M
U
FA

s,
P
U
FA

s
(l
in
o
le
ic
),
ve
ge
ta
b
le

p
ro
te
in
,

tr
a
n
s
o
r
u
n
sa
tu
ra
te
d
fa
t,
an
im

al
p
ro
te
in
,

h
is
to
ry

o
f
h
y
p
er
te
n
si
o
n
,
d
ia
be
te
s,

an
d
h
ig
h
ch
o
le
st
er
o
l

R
R
=
1
.0
5
(0
.3
3
,
3
.3
9)

F
o
r
fi
ft
h
q
u
in
ti
le

v
s
fi
rs
t
q
u
in
ti
le
H
e
et
al
,
2
0
0
3
(2
9
)
(H
ea
lt
h
P
ro
fe
ss
io
n
al
s
F
o
ll
ow
-U
p
S
tu
d
y
)
M
al
e

T
o
ta
l
st
ro
ke
:
5
9
8

F
if
th

q
u
in
ti
le
:
3
1
g
/d

F
ir
st
q
u
in
ti
le
:
1
7
g
/d

B
M
I,
p
h
y
si
ca
l
ac
ti
v
it
y,
h
is
to
ry
o
f
h
y
p
er
te
n
si
o
n,

sm
o
k
in
g
st
at
u
s,

as
p
ir
in
u
se
,
m
u
lt
iv
it
am

in
u
se
,

al
co
ho
l
co
n
su
m
p
ti
on
,
p
o
ta
ss
iu
m
,

fi
b
er
,
v
it
am

in
E
,
fr
ui
t
an
d

ve
g
et
ab
le
s,
to
ta
l
en
er
g
y,

h
y
p
er
ch
ol
es
te
ro
le
m
ia
,

o
th
er

fa
ts
(M

U
FA
s,
P
U
FA
s,

an
d
tr
a
n
s
fa
ts
)

R
R
m
e
n
,
6
0
y
=
0
.7
2
(0
.3
5
,
1
.5
1)

R
R
m
e
n
�
6
0
y
=
0
.8
2
(0
.4
9
,
1
.3
6)

4
P
o
ol
ed

R
R
=
0
.7
9
(0
.5
2
,
1
.1
9)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty
=
0
.7
9
F
o
r
fi
ft
h
q
u
in
ti
le
v
s
fi
rs
t
q
u
in
ti
le
(C
o
nt
in
u
ed
)

META-ANALYSIS OF SATURATED FAT AND CVD 541

T
A
B
L
E
3
(C
on
ti
n
u
ed
)
S
tu
d
y
S
ex
C
as
es
M
ed
ia
n
o
r
m
ea
n
sa
tu
ra
te
d
fa
t
in
ta
k
e
A
d
ju
st
ed
co
va
ri
at
es
M
u
lt
iv
ar
ia
te
ad
ju
st
ed
R
R
(9
5
%
C
I)
Is
o
et
al
,
2
0
0
3
(3
0
)
B
o
th
H
em
o
rr
ha
g
ic
st
ro
k
e:
6
7

F
o
u
rt
h
q
u
ar
ti
le
:
1
7
.1

g
/d

F
ir
st
q
u
ar
ti
le
:
5
.2

g
/d
A
g
e,

se
x
,
to
ta
l
en
er
g
y
in
ta
k
e,

B
M
I,

h
y
p
er
te
n
si
o
n
,
d
ia
be
te
s,
to
ta
l
ch
o
le
st
er
o
l,

sm
o
k
in
g
st
at
u
s,
et
h
an
o
l
in
ta
k
e,

m
en
o
p
au
sa
l
st
at
u
s
(f
o
r
w
o
m
en
)

R
R
=
0
.3
0
(0
.1
3
,
0
.7
1)

5
F
o
r
fo
u
rt
h
q
u
ar
ti
le
v
s
fi
rs
t
q
u
ar
ti
le
S
au
va
g
et
et
al
,
2
0
0
4
(3
4
)

(A
d
u
lt
H
ea
lt
h
S
tu
d
y)

B
o
th
Is
ch
em
ic
st
ro
k
e:
6
0
N
R

A
g
e
an
d
se
x
st
ra
ti
fi
ed

an
d
ad
ju
st
ed

fo
r
ra
d
ia
ti
o
n
d
o
se
,
ci
ty
,
B
M
I,

sm
o
k
in
g
,
al
co
ho
l,
h
is
to
ry

o
f

h
y
p
er
te
n
si
o
n
an
d
d
ia
b
et
es

H
R
=
0
.5
8
(0
.2
8
,
1
.2
0)

F
o
r
th
ir
d
te
rt
il
e
v
s
fi
rs
t
te
rt
il
e
L
eo
sd
o
tt
ir
et
al
,
2
0
0
7
(1
4
)
(M
al
m
o
D
ie
t
an
d
C
an
ce
r
S
tu
d
y
)
B
o
th
Is
ch
em
ic

st
ro
k
e:
6
4
8

M
en
:
F
o
u
rt
h
q
u
ar
ti
le
,
2
2
.3
%
o
f
to
ta
l
en
er
g
y
;
F
ir
st
q
u
ar
ti
le
,
1
2
.3
%
o
f
to
ta
l
en
er
g
y
W
o
m
en
:
F
o
u
rt
h
q
u
ar
ti
le
,
2
1
.8
%
o
f
to
ta
l
en
er
g
y
;
F
ir
st
q
u
ar
ti
le
,
1
2
.2
%
o
f
to
ta
l
en
er
g
y
A
g
e,
sm
o
k
in
g
h
ab
it
s,
al
co
ho
l

co
n
su
m
p
ti
o
n
,
so
ci
oe
co
n
o
m
ic

st
at
u
s,
m
ar
it
al

st
at
u
s,
p
h
y
si
ca
l

ac
ti
v
it
y,
B
M
I,
fi
b
er
in
ta
k
e,
b
lo
o
d
p
re
ss
u
re

R
R
w
o
m
e
n
=
1
.2
6
(0
.8
1
,
1
.9
6)

R
R
m
e
n
=
1
.1
9
(0
.8
0
,
1
.7
7
)

P
o
ol
ed

R
R
=
1
.2
2
(0
.9
1
,
1
.6
4)

P
fo
r
te
st
o
f
h
et
er
o
ge
n
ei
ty

=
0
.8
5

F
o
r
fo
u
rt
h
v
s
fi
rs
t
q
u
ar
ti
le

1
M
I,
m
yo
ca
rd
ia
l
in
fa
rc
ti
o
n
;
M
U
FA

,
m
on
o
u
n
sa
tu
ra
te
d
fa
tt
y
ac
id
;
N
R
,
n
o
t
re
p
o
rt
ed
;
P
U
FA

,
p
o
ly
u
ns
at
u
ra
te
d
fa
tt
y
ac
id
;
S
B
P,

sy
st
o
li
c
b
lo
o
d
p
re
ss
u
re
;
H
R
,
h
az
ar
d
ra
ti
o;

T
G
,
tr
ig
ly
ce
ri
d
es
.

2
R
R
es
ti
m
at
es

w
er
e
d
er
iv
ed

fr
o
m

a
p
ro
v
id
ed

d
at
a
se
t
(3
6
).

3
V
al
u
es

re
p
re
se
n
t
th
e
m
ea
n
in
ta
k
e
o
f
sa
tu
ra
te
d
fa
tt
y
ac
id
.

4
D
at
a
w
er
e
p
ro
v
id
ed

b
y
st
ud
y
in
ve
st
ig
at
o
rs

o
n
re
qu
es
t.

5
S
ta
ti
st
ic
al
ly

si
gn
ifi
ca
nt

re
la
ti
o
n
.

542 SIRI-TARINO ET AL

Although the Strong Heart Study reported a positive association
between saturated fat and CVD in younger than in older in-
dividuals (ie, RR = 5.17; 95% CI: 1.6, 16.4), the fully adjusted
model that included adjustment for polyunsaturated fatty acids,
trans fats, and monounsaturated fatty acids was not statistically
significant (RR = 2.98; 95% CI: 0.66, 13.6).

In contrast, a number of studies did not show significant
associations of dietary saturated fat intake with CHD, including
the Western Electric Study (17), the Honolulu Heart Study (9),
the Ireland Boston Diet Heart Study (13), the Caerphilly Study
(28), the Framingham Heart Study (16), the Israeli Ischemic
Study (35), the Alpha-Tocopherol, Beta-Carotene Study (15), the
Nurses’ Health Study (33), the Malmo Diet and Cancer Study
(14), and the Baltimore Longitudinal Study of Aging (18).

With respect to stroke, although inverse associations of sat-
urated fat intake with hemorrhagic stroke were reported in 2
studies (11, 30), no association between saturated fat and stroke
was found in 6 other studies (9, 14, 29, 31, 34, 35). The relation
of saturated fat with ischemic versus hemorrhagic stroke may
differ given their different biological mechanisms, and consid-
eration of these 2 disease states as distinct endpoints may be
important.

Individual study estimates as well as the overall estimate for
CHD, stroke, and CVD are shown in Figure 2. Saturated fat
intake was not associated with an elevated risk of CHD, stroke,
or CVD as a composite outcome. The RRs (95% CIs) were 1.07

(0.96, 1.19) for risk of CHD, 0.81 (0.62, 1.05) for risk of stroke,
and 1.00 (0.89, 1.11) for overall CVD risk. Two of the 8 studies
included in the meta-analysis related to stroke examined hem-
orrhagic stroke exclusively (30, 31). When these 2 studies were
excluded from the meta-analysis, the pooled RR (95% CI) was
0.86 (0.67, 1.11).

We documented heterogeneity among studies that examined
saturated fat in relation to CHD (P = 0.04) or stroke (P = 0.01).
However, age (P = 0.16 for CHD, 0.40 for stroke), sex (P = 0.52
for CHD, 0.25 for stroke), sample size (P = 0.44 for CHD, 0.71
for stroke), duration of follow-up (P = 0.53 for CHD, 0.42 for
stroke), medical record review for CVD outcome confirmation
(P = 0.17 for CHD, 0.30 for stroke), and study quality as as-
sessed by a quality score (P = 0.62 for CHD, 0.70 for stroke)
could not explain this heterogeneity. Quality scores for each
study are provided in Supplementary Table 1 (see “Supple-
mental data” in the online issue).

No individual study had a particularly large influence on the
pooled estimate of RR for CVD, although Gillman et al’s (11) and
Boniface and Tefft’s (5) studies had relatively stronger effects on
the overall RR estimate than did other studies. The pooled RRs
for CVD were 1.03 (95% CI: 0.93, 1.14) after excluding Gillman
et al and 0.97 (95% CI: 0.88, 1.08) after excluding Boniface and
Tefft. When these 2 studies were excluded simultaneously, the
pooled RRs (95% CI) were 1.02 (0.94, 1.11) for CHD, 0.86 (0.65,
1.14) for stroke, and 1.00 (0.92, 1.10) for CVD, respectively.

FIGURE 2. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between saturated fat intake in relation to coronary
heart disease and stroke. 1Updated data were provided by respective investigators (4, 5, 8, 18, 29, 35) or derived from a provided data set (9, 36). SAT,
saturated fat intake; IV, inverse variance.

META-ANALYSIS OF SATURATED FAT AND CVD 543

Subgroup analyses evaluating the association of saturated fat
with CVD by sex or age (, or � 60 y) showed no significant
associations (see Supplementary Figures 1 and 2, respectively,
under “Supplemental data” in the online issue). In men, the
pooled RR (95% CI) of CVD in relation to saturated fat intake
was 0.97 (0.87, 1.08), whereas in women this figure was 1.06
(0.86, 1.32). The associations for saturated fat intake were
similar between participants who were younger than 60 y at
baseline and those who were older: the pooled RRs (95% CIs)
were 0.98 (0.84, 1.13) and 0.98 (0.86, 1.10), respectively. Fur-
ther stratification by both age and sex (ie, men or women
younger than 60 y and men or women older than 60 y) also
showed no significant associations between saturated fat and
CHD risk, although sample size may have been inadequate for
these analyses (see Supplementary Figures 3 and 4, respectively,
under “Supplemental data” in the online issue). The limited
number of studies excluded our ability to further stratify the
analysis by study outcome.

Of 21 studies, 15 studies adjusted for total energy intake in the
fully adjusted model. Secondary analyses conducted within these
studies showed results largely similar to the primary analysis (see
Supplementary Figure 5 under “Supplemental data” in the on-
line issue). Six studies further adjusted for energy from other
fats and protein, but left energy from carbohydrate out of the
fully adjusted model. In these studies, the pooled RR (95% CI)
was 0.98 (0.86, 1.13) for CHD, 0.93 (0.71, 1.21) for stroke, and
0.97 (0.86, 1.10) for overall CVD (see Supplementary Figure 6
under “Supplemental data” in the online issue). Similarly, 5
studies adjusted for energy from carbohydrate, protein, and fats
but not polyunsaturated fat. The pooled RR (95% CI) for these
studies was 1.07 (0.91, 1.25) for CHD, 0.95 (0.81, 1.13) for
stroke, and 1.02 (0.92, 1.14) for overall CVD (see Supplemen-
tary Figure 7 under “Supplemental data” in the online issue).

A funnel plot of the 21 studies that evaluated the association of
saturated fat with CVD is provided in Figure 3. The larger
studies at the top of the plot were somewhat more symmetrically
distributed than were the smaller studies at the bottom. This
suggests the heterogeneity of the study estimates as well
as possible publication bias favoring studies with significant
results.

DISCUSSION

This study sought to evaluate the effects of dietary saturated fat
on CVD risk by summarizing the data available from informative
epidemiologic studies and including, where possible and rele-
vant, supplementary information that had been provided on
request from investigators of the component studies. The con-
glomeration of data from 16 studies with CHD as an endpoint and
8 studies with stroke as the endpoint showed no association of
dietary saturated fat on disease prevalence after adjustment for
other nutrients wherever possible. Evaluation of the subset of
studies (n = 15) that adjusted for total energy, which has been
shown to be relevant in evaluating nutrient-disease relations
(39), yielded similar findings. This study had several strengths,
including the selection of prospective epidemiologic studies that
statistically adjusted for relevant covariates and the inclusion of
large studies with a significant number of incident cases. Fur-
thermore, the use of the random-effects model in our analyses
allowed for the heterogeneity of variance between studies.

A caveat of this study was its reliance on the accuracy of the
dietary assessments of the component studies, which may vary
depending on the method used (25). Underreporting of calories
has often contributed to the error associated with dietary
assessments, particularly in overweight individuals. Generally, 4-
to 7-d food records are considered to be the most accurate means
of dietary assessment, but suchmethods are generally not feasible
in large cohort studies. A single 24-h recall is relatively easy to
collect, but the information does not reflect long-term dietary
patterns. FFQs have become the method of choice in large ep-
idemiologic studies because they are inexpensive and can assess
long-term diets (25); however, this method is also subject to
random and systematic errors.

As part of a quality score, the method of nutrient assessment
was taken into account, and the risk estimates that each study
contributed were adjusted based on this quality score, which also
considered whether the dietary assessment method had been
validated or repeatedly performed as well as the number of
covariates included in the model. The latter criterion assumed
that investigators included all relevant covariates in their re-
gression models. Evaluation of the studies on the basis of this
quality score did not change the findings of this meta-analysis.

Only a limited number of studies provided data that enabled
the evaluation of the effects of isocalorically replacing saturated
fat with carbohydrate or polyunsaturated fat, and, as such, the
statistical power was diminished for the secondary analyses
restricted to these studies. Most recently, however, an analysis
conducted in a pooled cohort of studies showed a lower CHD risk
when saturated fat was replaced with polyunsaturated fat and
increased nonfatal myocardial infarction, but not fatal CHD, risk
when saturated fat was replaced with carbohydrate (24).

Inverse associations of polyunsaturated fat and CVD risk have
previously been reported (41, 42). Replacement of 5% of total
energy from saturated fat with polyunsaturated fat has been
estimated to reduce CHD risk by 42% (43). Notably, the amount
of dietary polyunsaturated fat in relation to saturated fat (ie, the P:
S ratio) has been reported to be more significantly associated with
CVD than saturated fat alone, with a reduced CHD risk found
with P:S ratios � 0.49 (44). Only 1 of the 21 studies that met
criteria for inclusion in this meta-analysis evaluated the relation
of the P:S ratio with CHD (14). No effect was seen in this study,

FIGURE 3. Funnel plot of studies of saturated fat intake in relation to
cardiovascular disease. Dotted lines are pseudo 95% CIs. The large studies at
the top of the plot were somewhat more symmetrically distributed than the
small studies at the bottom. This indicates publication bias favoring studies
with significant results. RR, risk ratio.

544 SIRI-TARINO ET AL

in which the average P:S ratio was ’0.4, nor was there an as-
sociation of P:S ratio with CVD in the Israeli Ischemic Heart
Study (U Goldbourt, personal communication, 2008). However,
these studies were relatively small.

Of note, in intervention trials that have shown protective
effects of reducing saturated fat, ie, the Veteran Affairs (19), Oslo
Diet Heart (20), and Finnish Mental Hospital (21) studies, the
calculated P:S ratios ranged from 1.4 to 2.4—values that are
much higher than the threshold of 0.49 abovewhich CHD risk has
been reported to be reduced (44). Relatively high P:S ratios
(1.25–1.5) were also observed in the Anti-Coronary Club Study,
an early trial that showed beneficial effects of a lower fat diet (30–
32% of total energy) (45). The presumed beneficial effects of
diets with reduced saturated fat on CVD risk may therefore be
dependent on a significant increase in polyunsaturated fat in the
diet. Existing epidemiologic studies and clinical trials support
that substituting polyunsaturated fat for saturated fat is more
beneficial for CHD risk than exchanging carbohydrates for
saturated fat in the diet, as described further elsewhere (46).

With respect to dietary carbohydrate, the type of carbohydrate
(ie, a high or low glycemic index) that replaces saturated fat is
likely important in influencing dietary effects on CVD risk
(47). However, there was insufficient information in the com-
ponent studies of this meta-analysis to permit examination of
this issue.

Our results suggested publication bias, such that studies with
significant associations tended to be received more favorably for
publication. If unpublished studies with null associations were
included in the current analysis, the pooled RR estimate for CVD
could be even closer to null. Furthermore, despite several indi-
cations in the published literature that sex and age may modify
the association of saturated fat with CHD (5, 6, 8, 10), we did not
observe effects of these variables on CHD risk. The lack of an
association may have been related to limited statistical power.

Although an inverse association of saturated fat with stroke
risk has been previously described (48), saturated fat intake was
not significantly associated with risk of stroke in the current meta-
analysis. The exclusion of 2 studies conducted in Japan (30, 34),
where saturated fat intake is known to be significantly lower than
in Western populations, did not substantially change the RR
estimate (RR = 0.90; 95% CI: 0.70, 1.15). The exclusion of 2
studies that evaluated hemorrhagic stroke specifically (30, 31)
also did not alter the RR estimate; however, these findings were
likely limited by statistical power.

In conclusion, our meta-analysis showed that there is in-
sufficient evidence from prospective epidemiologic studies to
conclude that dietary saturated fat is associated with an increased
risk of CHD, stroke, or CVD. However, the available data were
not adequate for determining whether there are CHD or stroke
associations with saturated fat in specific age and sex subgroups.
Furthermore, there was insufficient statistical power for this
meta-analysis to assess the effects on CVD risk of replacing
specific amounts of saturated fat with either polyunsaturated fat
or carbohydrate. Finally, nutritional epidemiologic studies pro-
vide only one category of evidence for evaluating the relation of
saturated fat intake to risk for CHD, stroke, and CVD. An overall
assessment requires consideration of results of clinical trials as
well as information regarding the effects of saturated fat on
underlying disease mechanisms, as discussed elsewhere in this
issue (46).

We thank David Boniface, Alberto Ascherio, Dariush Mozaffarian,

Katherine Tucker, Uri Goldbourt, and Marianne Jakobsen for providing data

requested for inclusion in this meta-analysis. The Honolulu Heart Program is

conducted and supported by the National Heart, Lung, and Blood Institute

(NHLBI) in collaboration with the Honolulu Heart Program Study Investiga-

tors. This manuscript was prepared by using a limited access data set obtained

from the NHLBI and does not necessarily reflect the opinions or views of the

Honolulu Heart Program or the NHLBI.

The authors’ responsibilities were as follows—PWS-T, QS, FBH, and

RMK: selected the studies for inclusion in the meta-analysis; PWS-T and

QS: extracted data from the studies and wrote the manuscript; QS: performed

the statistical analyses; and FBH and RMK: provided significant advice and

consultation. No conflicts of interest were reported.

REFERENCES
1. Anitschkow N. A history of experimentation on arterial atherosclerosis

in animals. Cowdry’s arteriosclerosis: a survey of the problem. Springfield,
IL: Charles C Thomas, 1967:21–44.

2. Kato H, Tillotson J, Nichaman MZ, Rhoads GG, Hamilton HB. Epi-
demiologic studies of coronary heart disease and stroke in Japanese
men living in Japan, Hawaii and California. Am J Epidemiol 1973;97:
372–85.

3. Keys A, Aravanis C, Blackburn HW, et al. Epidemiological studies re-
lated to coronary heart disease: characteristics of men aged 40-59 in seven
countries. Acta Med Scand Suppl 1966;460:1–392.

4. Ascherio A, Rimm EB, Giovannucci EL, Spiegelman D, Stampfer M,
Willett WC. Dietary fat and risk of coronary heart disease in men: cohort
follow up study in the United States. BMJ 1996;313:84–90.

5. Boniface DR, Tefft ME. Dietary fats and 16-year coronary heart disease
mortality in a cohort of men and women in Great Britain. Eur J Clin
Nutr 2002;56:786–92.

6. Esrey KL, Joseph L, Grover SA. Relationship between dietary intake
and coronary heart disease mortality: lipid research clinics prevalence
follow-up study. J Clin Epidemiol 1996;49:211–6.

7. Garcia-Palmieri MR, Sorlie P, Tillotson J, Costas R Jr, Cordero E,
Rodriguez M. Relationship of dietary intake to subsequent coronary
heart disease incidence: the Puerto Rico Heart Health Program. Am J
Clin Nutr 1980;33:1818–27.

8. Jakobsen MU, Overvad K, Dyerberg J, Schroll M, Heitmann BL. Di-
etary fat and risk of coronary heart disease: possible effect modification
by gender and age. Am J Epidemiol 2004;160:141–9.

9. McGee DL, Reed DM, Yano K, Kagan A, Tillotson J. Ten-year in-
cidence of coronary heart disease in the Honolulu Heart Program: re-
lationship to nutrient intake. Am J Epidemiol 1984;119:667–76.

10. Xu J, Eilat-Adar S, Loria C, et al. Dietary fat intake and risk of coronary
heart disease: the Strong Heart Study. Am J Clin Nutr 2006;84:894–902.

11. Gillman MW, Cupples LA, Millen BE, Ellison RC, Wolf PA. Inverse
association of dietary fat with development of ischemic stroke in men.
JAMA 1997;278:2145–50.

12. Mozaffarian D, Rimm EB, Herrington DM. Dietary fats, carbohydrate,
and progression of coronary atherosclerosis in postmenopausal women.
Am J Clin Nutr 2004;80:1175–84.

13. Kushi LH, Lew RA, Stare FJ, et al. Diet and 20-year mortality from
coronary heart disease. The Ireland-Boston Diet-Heart Study. N Engl J
Med 1985;312:811–8.

14. Leosdottir M, Nilsson PM, Nilsson JA, Berglund G. Cardiovascular
event risk in relation to dietary fat intake in middle-aged individuals:
data from The Malmo Diet and Cancer Study. Eur J Cardiovasc Prev
Rehabil 2007;14:701–6.

15. Pietinen P, Ascherio A, Korhonen P, et al. Intake of fatty acids and risk of
coronary heart disease in a cohort of Finnish men. The Alpha-Tocopherol,
Beta-Carotene Cancer Prevention Study. Am J Epidemiol 1997;145:
876–87.

16. Posner BM, Cobb JL, Belanger AJ, Cupples LA, D’Agostino RB, Stokes
J III. Dietary lipid predictors of coronary heart disease in men. The
Framingham Study. Arch Intern Med 1991;151:1181–7.

17. Shekelle RB, Shryock AM, Paul O, et al. Diet, serum cholesterol, and
death from coronary heart disease. The Western Electric study. N Engl J
Med 1981;304:65–70.

18. Tucker KL, Hallfrisch J, Qiao N, Muller D, Andres R, Fleg JL. The
combination of high fruit and vegetable and low saturated fat intakes is

META-ANALYSIS OF SATURATED FAT AND CVD 545

more protective against mortality in aging men than is either alone: the
Baltimore Longitudinal Study of Aging. J Nutr 2005;135:556–61.

19. Dayton S, Pearce M, Hashimoto S, et al. A controlled clinical trial of
a diet high in unsaturated fat in preventing complications of athero-
sclerosis. Circulation 1969;40(suppl 2):1–63.

20. Leren P. The Oslo Diet-Heart Study: eleven-year report. Circulation
1970;42:935–42.

21. Turpeinen O, Karvonen MJ, Pekkarinen M, Miettinen M, Elosuo R,
Paavilainen E. Dietary prevention of coronary heart disease: the Finnish
Mental Hospital Study. Int J Epidemiol 1979;8:99–118.

22. Frantz ID Jr, Dawson EA, Ashman PL, et al. Test of effect of lipid
lowering by diet on cardiovascular risk. The Minnesota Coronary Sur-
vey. Arteriosclerosis 1989;9:129–35.

23. Morris JN, Ball KP, Antonis A, et al Controlled trial of soya-bean oil in
myocardial infarction. Lancet 1968;XXX:693–9.

24. Jakobsen MU, O’Reilly EJ, Heitmann BL, et al. Major types of dietary
fat and risk of coronary heart disease: a pooled analysis of 11 cohort
studies. Am J Clin Nutr 2009;89:1425–32.

25. Willett W. Nutritional epidemiology. 2nd ed. New York, NY: Oxford
University Press, 1998.

26. Kromhout D, de Lezenne Coulander C, Obermann-de Boer GL, van
Kampen-Donker M, Goddijn E, Bloemberg BP. Changes in food and
nutrient intake in middle-aged men from 1960 to 1985 (the Zutphen
Study). Am J Clin Nutr 1990;51:123–9.

27. Boden-Albala B, Elkind MS, White H, Szumski A, Paik MC, Sacco RL.
Dietary total fat intake and ischemic stroke risk: the Northern Manhattan
Study. Neuroepidemiology 2009;32:296–301.

28. Fehily AM, Yarnell JW, Sweetnam PM, Elwood PC. Diet and incident
ischaemic heart disease: the Caerphilly Study. Br J Nutr 1993;69:303–14.

29. He K, Merchant A, Rimm EB, et al. Dietary fat intake and risk of stroke
in male US healthcare professionals: 14 year prospective cohort study.
BMJ 2003;327:777–82.

30. Iso H, Sato S, Kitamura A, Naito Y, Shimamoto T, Komachi Y. Fat
and protein intakes and risk of intraparenchymal hemorrhage among
middle-aged Japanese. Am J Epidemiol 2003;157:32–9.

31. Iso H, Stampfer MJ, Manson JE, et al. Prospective study of fat and
protein intake and risk of intraparenchymal hemorrhage in women.
Circulation 2001;103:856–63.

32. Mann JI, Appleby PN, Key TJ, Thorogood M. Dietary determinants of
ischaemic heart disease in health conscious individuals. Heart 1997;78:
450–5.

33. Oh K, Hu FB, Manson JE, Stampfer MJ, Willett WC. Dietary fat intake
and risk of coronary heart disease in women: 20 years of follow-up of
the nurses’ health study. Am J Epidemiol 2005;161:672–9.

34. Sauvaget C, Nagano J, Hayashi M, Yamada M. Animal protein, animal
fat, and cholesterol intakes and risk of cerebral infarction mortality in
the adult health study. Stroke 2004;35:1531–7.

35. Goldbourt U, Yaari S, Medalie JH. Factors predictive of long-term
coronary heart disease mortality among 10,059 male Israeli civil ser-
vants and municipal employees. A 23-year mortality follow-up in the
Israeli Ischemic Heart Disease Study. Cardiology 1993;82:100–21.

36. National Heart, Lung, and Blood Institute. Honolulu Heart Program
limited data set. Bethesda, MD: National Heart, Lung, and Blood In-
stitute, 2008.

37. Greenland S. Quantitative methods in the review of epidemiologic lit-
erature. Epidemiol Rev 1987;9:1–30.

38. Ding EL, Song Y, Malik VS, Liu S. Sex differences of endogenous
sex hormones and risk of type 2 diabetes: a systematic review and
meta-analysis. JAMA 2006;295:1288–99.

39. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake
in epidemiologic studies. Am J Clin Nutr 1997;65(suppl):1220S–8S;
discussion 1229S–31S.

40. Egger M, Altman DG, Smith GD, eds. Systematic reviews in health
care: meta-analysis in context. London, United Kingdom: BMJ Books,
2001.

41. Laaksonen DE, Nyyssonen K, Niskanen L, Rissanen TH, Salonen JT.
Prediction of cardiovascular mortality in middle-aged men by dietary
and serum linoleic and polyunsaturated fatty acids. Arch Intern Med
2005;165:193–9.

42. Soinio M, Laakso M, Lehto S, Hakala P, Ronnemaa T. Dietary fat
predicts coronary heart disease events in subjects with type 2 diabetes.
Diabetes Care 2003;26:619–24.

43. Hu FB, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of
coronary heart disease in women. N Engl J Med 1997;337:1491–9.

44. Hu FB, Stampfer MJ, Manson JE, et al. Dietary saturated fats and their
food sources in relation to the risk of coronary heart disease in women.
Am J Clin Nutr 1999;70:1001–8.

45. Christakis G, Rinzler SH,ArcherM,Maslansky E. Summary of the research
activities of the anti-coronary club. Public Health Rep 1966;81:64–70.

46. Siri-Tarino PW, Sun Q, Hu FB, Krauss RM. Saturated fat, carbohydrate,
and cardiovascular disease. Am J Clin Nutr 2010;91:502–9.

47. Barclay AW, Petocz P, McMillan-Price J, et al. Glycemic index, gly-
cemic load, and chronic disease risk–a meta-analysis of observational
studies. Am J Clin Nutr 2008;87:627–37.

48. Corvol JC, Bouzamondo A, Sirol M, Hulot JS, Sanchez P, Lechat P.
Differential effects of lipid-lowering therapies on stroke prevention:
a meta-analysis of randomized trials. Arch Intern Med 2003;163:
669–76.

546 SIRI-TARINO ET AL

Consequences of bullying victimization in childhood and
adolescence: A systematic review and meta-analysis

Sophie E Moore, Rosana E Norman, Shuichi Suetani, Hannah J Thomas, Peter D Sly, James G Scott

Sophie E Moore, Peter D Sly, Child Health Research Centre,
the University of Queensland, South Brisbane, QLD 4101,
Australia

Rosana E Norman, Institute of Health and Biomedical Innova
tion, Queensland University of Technology, Kelvin Grove, QLD
4059, Australia

Rosana E Norman, School of Public Health and Social Work,
Queensland University of Technology, Kelvin Grove, QLD 4059,
Australia

Shuichi Suetani, Queensland Centre for Mental Health
Research, the Park Centre for Mental Health, Wacol, QLD 4076,
Australia

Shuichi Suetani, Faculty of Medicine, the University of
Queensland, Herston, QLD 4029, Australia

Hannah J Thomas, James G Scott, the University of Queensland
Centre for Clinical Research, the University of Queensland,
Herston, QLD 4029, Australia

James G Scott, Metro North Mental Health, Royal Brisbane and
Women’s Hospital, Herston, QLD 4029, Australia

Author contributions: Norman RE and Scott JG designed
the study, supervised the systematic review and metaanalysis
and supervised the writing of the manuscript; Moore SE and
Suetani S conducted the systematic review; Moore SE conducted
the metaanalysis and wrote the first draft of the manuscript;
Thomas HJ drafted sections of the manuscript related to bullying
measurement and supervised the manuscript content; all authors
contributed to and approved the final manuscript.

Conflict-of-interest statement: The authors have no conflicts of
interest to declare.

Data sharing statement: No additional data is available.

Open-Access: This article is an openaccess article which was
selected by an inhouse editor and fully peerreviewed by external
reviewers. It is distributed in accordance with the Creative
Commons Attribution Non Commercial (CC BY-NC 4.0) license,
which permits others to distribute, remix, adapt, build upon this
work noncommercially, and license their derivative works on

different terms, provided the original work is properly cited and
the use is non-commercial. See: http://creativecommons.org/
licenses/by-nc/4.0/

Manuscript source: Invited manuscript

Correspondence to: James G Scott, Associate Professor,
the University of Queensland Centre for Clinical Research, the
University of Queensland, Bowen Bridge Rd, Herston, QLD
4029, Australia. james.scott@health.qld.gov.au
Telephone: +61736368111
Fax: +61736361166

Received: September 13, 2016
Peer-review started: September 14, 2016
First decision: October 21, 2016
Revised: December 4, 2016
Accepted: December 27, 2016
Article in press: December 28, 2016
Published online: March 22, 2017

Abstract
AIM
To identify health and psychosocial problems associated
with bullying victimization and conduct a meta-analysis
summarizing the causal

evidence.

METHODS
A systematic review was conducted using PubMed,
EMBASE, ERIC and PsycINFO electronic databases up
to 28 February 2015. The study included published
longitudinal and cross-sectional articles that examined
health and psychosocial consequences of bullying
victimization. All meta-analyses were based on quality-
effects models. Evidence for causality was assessed
using Bradford Hill criteria and the grading system
developed by the World Cancer Research Fund.

RESULTS
Out of 317 articles assessed for eligibility, 165 satisfied
the predetermined inclusion criteria for meta-analysis.

META-ANALYSIS

March 22, 2017|Volume 7|Issue 1|WJP|www.wjgnet.com

Submit a Manuscript:

http://www.wjgnet.com

/esps/

DOI: 10.5498/wjp.v7.i1.60

World J Psychiatr 2017 March 22; 7(1): 60-76

ISSN 2220-3206 (online)

World Journal of
PsychiatryW J P

Statistically significant associations were observed
between bullying victimization and a wide range
of adverse health and psychosocial problems. The
evidence was strongest for causal associations between
bullying victimization and mental health problems such
as depression, anxiety, poor general health and suicidal
ideation and behaviours. Probable causal associations
existed between bullying victimization and tobacco and
illicit drug use.

CONCLUSION
Strong evidence exists for a causal relationship between
bullying victimization, mental health problems and
substance use. Evidence also exists for associations
between bullying victimization and other adverse health
and psychosocial problems, however, there is insufficient
evidence to conclude causality. The strong evidence
that bullying victimization is causative of mental illness
highlights the need for schools to implement effective
interventions to address bullying behaviours.

Key words: Bullying; Victimization; Systematic review;
Meta-analysis; Child; Adolescent

Core tip: There is convincing evidence of a causal
association between exposure to bullying victimization in
children and adolescents and adverse health outcomes
including anxiety, depression, poor mental health,
poor general health, non-suicidal self-injury, suicidal
ideation and suicide attempts. It is probable that bullying
victimization also causes an increased risk of cigarette
smoking and illicit drug use. This review highlights that
bullying victimization is associated with a wide and
diverse range of problems and reinforces the need for
effective interventions to be implemented in schools to
address the high prevalence of children and adolescents
engaging in bullying behaviours.

Moore SE, Norman RE, Suetani S, Thomas HJ, Sly PD, Scott
JG. Consequences of bullying victimization in childhood and
adolescence: A systematic review and meta-analysis. World J
Psychiatr 2017; 7(1): 60-76 Available from: URL: http://www.
wjgnet.com/2220-3206/full/v7/i1/60.htm DOI: http://dx.doi.
org/10.5498/wjp.v7.i1.60

INTRODUCTION
Bullying victimization among children and adolescents
is a global public health issue, well-recognised as a
behaviour associated with poor adjustment in youth[1].
There is evidence suggesting bullying victimization in
children and adolescents has enduring effects which
may persist into adulthood[2-4]. Bullying victimization is
most commonly defined as exposure to negative actions

repeatedly and over time from one or more people, and
involves a power imbalance between the perpetrator(s)
and the victim[5]. Traditional bullying includes physical
contact (pushing, hitting) as well as verbal harassment
(name calling, verbal taunting), rumour spreading,
intentionally excluding a person from a group, and
obscene gestures. In recent years cyberbullying has
emerged as a significant public health problem[5-8].

The estimated prevalence of bullying victimization
is wide-ranging, with 10% and 35% of adolescents
experiencing recurrent bullying victimization[9-16].
While contextual and cultural differences influence
prevalence estimates[17], this variation is most frequently
explained by differences in measurement strategy[18-20].
As a result, researchers continue to call for greater
consensus in the definition and measurement of bullying
behaviours[17,21,22]. Cook et al[17] examined the variability
in prevalence of bullying victimization in a meta-analysis,
and more recently Modecki et al[20] synthesised studies
measuring both traditional bullying and cyberbullying.
Mean prevalence was 36% for traditional bullying
victimization and 15% for cyberbullying victimization[20].
There was significant overlap between bullying victimi
zation in traditional and online settings[20]. A meta-
analysis by Kowalski et al[23] showed that the strongest
predictor of cyber-victimization was traditional bullying
victimization.

Many studies have examined adverse health and
psychosocial problems associated with bullying victimi-
zation. Those most commonly reported are mental
health problems, specifically depression, anxiety, self
harm, and suicidal behaviour[2,14,24-28]. Over the past
two decades researchers have conducted a number of
systematic reviews to examine the relationship between
bullying victimization and ill mental health.

The first systematic investigation by Hawker and
Boulton[1] was a meta-analysis of cross-sectional
studies of peer victimization published between 1978
and 1997. The authors reported victimization was
significantly associated with depression, loneliness,
reduced selfesteem and selfconcept, as well as anxiety.
To understand the temporal sequence between peer
victimization and mental health problems, Reijntjes
and colleagues conducted a pair of meta-analyses of
longitudinal studies to examine internalizing (depres
sion, anxiety, withdrawal, loneliness, and somatic
complaints) and externalizing behaviours (aggression
and delinquency) and peer victimization[29,30]. They
examined two prospective paths: (1) peer victimization
at baseline and changes in internalizing and externalizing
problems at a second time point; and (2) internalizing
and externalizing problems at baseline and changes in
peer victimization at follow-up. The two meta-analyses
demonstrated internalizing and externalizing behaviours
are both antecedents and consequences of bullying
victimization[29,30].

Another meta-analytic review on bullying victimi-

Moore SE et al . Consequences of bullying victimization

61WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

zation and depression by Ttofi et al[31] found those
children who were bullied at school were twice as likely
to develop depression compared to those who had
not been bullied. In addition, another meta-analytic
review found those children involved in any bullying
behaviour were more likely to develop psychosomatic
problems[32]. Finally, three systematic reviews have
shown an association between bullying victimization
and increased risk of adolescent suicidal ideation and
behaviours[33-35].

In contrast to mental health, there is mixed evi
dence for the relationship between bullying victimi-
zation and substance use. Some studies report that
bullying victimization is associated with a reduced risk
of engaging in harmful alcohol use in later life[16,28],
whereas others suggest that being bullied may result in
an increased probability of later harmful alcohol use[36,37].
Similarly, some studies have shown an association
between being bullied and later illicit drug use and
smoking[36-39], whereas others have found no association
at all[2,14,24,40].

The association between bullying victimization and
psychosocial problems, such as academic achievement
and school functioning/connectedness and criminal
behaviour has also been examined. A metaanalysis by
Nakamoto and Schwartz[41] found a small but significant
negative association between bullying victimization
and academic achievement. However, Kowalski et al[23]
found no significant relationship between cyberbullying
victimization and academic achievement. Another
study found those exposed to bullying victimization in
adolescence were at increased risk of involvement in
criminal behaviour such as carrying a weapon[40].

There are now a large number of studies examining
associations between bullying victimization and a wide
range of adverse health and psychosocial problems.
However, many of these have not been systematically
examined and many existing systematic reviews did not
include cyberbullying. Furthermore, although associa-
tions exist, it is unclear if there is a causal relationship.
It is plausible that there are common factors that
predispose individuals to being bullied in childhood but
independently also increase the risk of adverse health
and other psychosocial problems. Rigorous appraisal
is required to consider both the possibility of a causal
association but also other plausible explanations for any
significant associations. Given the variation between
studies, this study aimed to investigate adverse
outcomes of both traditional and cyber bullying victimi-
zation and conduct a meta-analysis to summarize each
association. Furthermore, we critically evaluated whe-
ther sufficient evidence existed to establish a causal
relationship between bullying victimization and each of
the adverse health and psychosocial problems. This is
the first study to complete a summary of the evidence
for all adverse health and psychosocial problems that
are potentially a consequence of traditional and cyber
bullying victimization.

MATERIALS AND METHODS
This study followed the recommendations from the
PRISMA 2009 revision[42] and the guidelines outlined
by the Meta-analysis of Observational Studies in
Epidemiology[43] (Supplementary material S1). Methods
and inclusion and exclusion criteria were specified in
advance in the review protocol (Supplementary material
S2).

Inclusion and exclusion criteria
This systematic review and meta-analysis included
studies meeting the following inclusion criteria: (1)
reported original, empirical research published in a
peer reviewed journal; (2) examined the relationship
between exposure to bullying victimization as a child
or adolescent and one or more consequences of the
bullying exposure; and (3) is a population based
study. This study did not examine a particular type of
bullying victimization therefore all direct and indirect
forms of bullying including cyberbullying were included.
Included studies reported odds ratios (ORs) and
confidence intervals (CIs) comparing those exposed to
bullying victimization and those not exposed to bullying
victimization or, alternatively, provided information from
which effect sizes (ORs and CIs) could be calculated
between those exposed to bullying victimization and an
outcome.

Search strategy
Four electronic databases (PsycINFO, ERIC, EMBASE
and PubMed) were used to search for literature on the
adverse correlates of bullying victimization as either a
child or adolescent from inception up to 28 February
2015. The search was not restricted to the English
language nor by any other means. The searches of the
databases were conducted using the terms: “bullying”,
“bullied”, “harassment”, “intimidation”, “victimization”
along with “child” and “adolescent”. As this study
aimed to examine all correlates that were potentially a
consequence of bullying victimization, the search terms
used in conjunction with those above were broader
terms such as “outcome” “harm” “consequence” and
“risk”. In addition, reference lists of selected studies
were screened for any other relevant study and articles
in languages other than English were translated
(Supplementary material S2).

Data collection and quality assessment
The full text of articles that met all inclusion criteria
were retrieved and examined. Data extracted using a
data extraction template included publication details,
country where study was conducted, methodological
characteristics such as sample size and study design,
exposure and outcome measures, type of bullying and
frequency (Supplementary material S2). Each study
was then subjected to a quality assessment in order
for the reviewers to rate the quality of each study.

62WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Moore SE et al . Consequences of bullying victimization

classification criteria), severity of the bullying (frequent
– at least once a month, vs sometimes – less than once
a month), age of bullying victimization (before 13 years
of age vs after 13 years of age), and type of study
(prospective vs cross-sectional).

RESULTS
A total of 8231 articles were primarily identified by
the search, of which 3270 were duplicates. Titles
and abstracts for the 4961 remaining unduplicated
references were reviewed and 15 additional articles were
found from reference lists. From reviewing the title and
abstracts a further 4659 articles were excluded. This left
317 articles meeting the following criteria: (1) original
research extracted from a peer reviewed journal; and
(2) examined the bullying victimization as a child or
adolescent and one or more outcomes. Of the 317
articles reviewed, a further 152 articles were excluded
as they did not use a population based sample or did
not report enough information to calculate an effect
size. The remaining 165 articles provided evidence of
an effect size for bullying victimization and an outcome
(Figure 1) either odds ratio with confidence intervals
or provided data which enabled the calculation of effect
sizes. The majority (n = 142) were from high income
regions. There were far fewer studies (n = 22) from

Two reviewers independently reviewed the included
articles and completed the quality assessment and any
disagreements were resolved by a third reviewer. The
quality assessment tool was based on the Newcastle-
Ottawa Scale for assessing the quality of observational
studies[44] as used by Norman et al[45] (Supplementary
material S2).

Statistical analysis
Following the method used by Norman et al[45], MetaXL
version 2.1[46], an addin for Microsoft Excel was used
in this study to conduct the meta-analysis. ORs were
chosen as the summary measure. Heterogeneity was
assessed using the Cochran’s Q and I 2 statistics[47].
This meta-analysis used a quality effects model[48], a
modified version of the fixedeffects inverse variance
model that additionally allows greater weight to be
given to studies of higher quality vs studies of lesser
quality. The quality effects model avoids the limitation
in random-effects models of returning to equal weight-
ing irrespective of sample size if heterogeneity is
large[47,48]. Furthermore, in order to address the effects
of important study characteristics and explore hetero
geneity this study conducted subgroup analyses,
dependent on data availability, for sex of participants in
the sample, geographic location and income level (high
income vs low-to-middle income as per the World Bank

63WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Records identified through
database searching

(n = 8231)
Additional records

identified through other
sources (n = 15)

Records after duplicates removed
(n = 4961)

Records screened
(n = 4976)

Full-text articles
assessed for eligibility

(n = 317)

Id
en

ti
fic

at
io

n
S
cr

ee
ni

ng
E
lig

ib
ili

In
cl

ud
ed

Studies included in
qualitative synthesis

(n = 165)

Studies included in quantitative
synthesis (meta-analysis)

(n = 165)
Prospective cohort (n = 57)
Cross-sectional (n = 108)

Records excluded
(n = 4659)

Full-text articles excluded,
with reasons (n = 152)

n = 120 (not enough
information to calculate
effect size)
n = 32 (did not use
population based
samples)

PRISMA 2009 Flow Diagram

Figure 1 PRISMA flow diagram showing process of study selection for inclusion in systematic review and meta-analyses.

Moore SE et al . Consequences of bullying victimization

low- and middle-income countries, and only one study
utilized cross-national samples from different income-
level countries. Of the articles included, 57 had a
prospective cohort design and the remaining 108 were
cross-sectional. The majority of studies measured self-
reported bullying victimization. Some were from samples
collected from a state or regions where as others were
nationally representative (Supplementary material S3).

Bullying victimization in children and adolescents and
mental health
Bullying victimization in children and adolescents was
associated with a wide range of adverse mental health
outcomes (Table 1) including poor mental health (OR =
1.60; 95%CI: 1.421.81), syndromes such as depression
and anxiety, and symptoms and behaviours such as
psychotic symptoms, suicidal ideation and attempts.
Specifically, those exposed to bullying victimization had
an increased risk of depression (OR = 2.21; 95%CI:
1.343.65). This association remained significant for all
the sub-group analyses including prospective studies,
age bullying occurred, sex and severity of the bullying.
A dose response existed between being “sometimes
bullied” and “frequently bullied” and depression (OR
= 1.78; 95%CI: 1.392.28 and OR = 3.26; 95%CI:
2.45-4.34 respectively). In comparing high- and low-
to-middle income countries, there was no significant
difference in the odds of developing depression. Those
exposed to bullying victimization were significantly
more likely to experience anxiety (OR = 1.77; 95%CI:
1.342.33) and exposure to bullying victimization was
associated with a wide range of anxiety spectrum
disorders such as social phobia and post-traumatic stress
disorder. This association remained after conducting
subgroup analyses including study type, sex and severity
of the bullying; however, the association between
bullying victimization and anxiety was not significant in
children under 13 years (Table 1).

Bullying victimization was also associated with non-
suicidal selfinjury (OR = 1.75; 95%CI: 1.402.19)
and increased risk of suicidal ideation (OR = 1.77;
95%CI: 1.562.02) and the association remained
significant for all subgroup analyses (Table 1). A dose
response existed between being “sometimes bullied”
and “frequently bullied” and suicide ideation (OR = 1.53;
95%CI: 1.281.82 and OR = 2.59; 95%CI: 2.063.25
respectively). Bullying victimization was associated with
an increase in suicide attempts (OR = 2.13; 95%CI:
1.66-2.73). Subgroup analysis showed both males and
females were approximately three times more likely
to attempt suicide if they were bullied (OR = 2.93;
95%CI: 1.655.18 and OR = 2.89; 95%CI: 1.525.49
respectively). There was nearly a fourfold increase
in suicide attempts for individuals who experienced
frequent bullying victimization (OR = 3.77; 95%CI:
2.55-5.58) (Table 1). When comparing high income
countries to those with low and middle income, the
odds of bullying victims developing suicidal ideation or

attempting suicide were similar.
Although bullying victimization in children and

adolescents was associated with the pooling of all be-
havioural problems (OR = 1.37; 95%CI: 1.181.59),
this association was not significant in prospective cohort
studies and no doseresponse was observed. Diagnoses
of disruptive behavioural disorders were not associated
with bullying victimization in children and adolescents
(Table 1).

Bullying victimization in children and adolescents and
substance use
Table 2 presents the associations between bullying
victimization and substance use. When all studies were
pooled together there was a significant association
between bullying victimization and alcohol use (OR =
1.26; 95%CI: 1.001.58). A subgroup analysis showed
a significant association between bullying victimization
and the risk of tobacco use for prospective studies (OR =
1.62; 95%CI: 1.311.99). Furthermore, a dose response
was present with frequent bullying victimization being
associated with tobacco use (OR = 3.19; 95%CI:
1.19-8.58), whereas no significant association was
found with those who were “sometimes bullied” (Table 2).

Bullying victimization was associated with an
increased risk of illicit drug use (OR = 1.41; 95%CI:
1.10-1.81). Subgroup analysis revealed that the
association between bullying victimization and increased
risk of using illicit drugs was significant in both cross-
sectional (OR = 2.43; 95%CI: 1.424.15) and
prospective studies (OR = 1.27; 95%CI: 1.121.44).
A subgroup analysis also revealed bullying victims in
low-to-middle income countries are at an increased
risk of illicit drug use (OR = 4.05; 95%CI: 2.187.55)
compared with bullying victims in high income countries
(OR = 1.31; 95%CI: 1.151.48). No significant
association was found in a sub group analysis examining
cannabis and bullying victimization in children and
adolescence (Table 2).

Bullying victimization in children and adolescents and
other health outcomes
Table 3 presents the association between bullying
victimization and other health outcomes. Bullying
victimization was associated with increased risk of
somatic symptoms, the most common being stomach
ache (OR = 1.76; 95%CI: 1.532.03), sleeping
difficulties (OR = 1.73; 95%CI: 1.462.05), headaches
(OR = 1.64; 95%CI: 1.381.94), dizziness (OR = 1.64;
95%CI: 1.381.95), and back pain (OR = 1.67; 95%CI:
1.43-1.95). Bullying victimization was also associated
with an increased risk of being overweight and obese (OR
= 1.68; 95%CI: 1.212.33 and OR = 1.78; 95%CI:
1.42-2.21, respectively). These associations were
significant for crosssectional studies only and there was
no dose response.

When all studies were pooled, bullying victimization in
children and adolescents was associated with increased

64WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Moore SE et al . Consequences of bullying victimization

65WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Data points Pooled OR 95%CI
lower bound

95%CI
upper bound

Cochran’s
Q

I ²
(%)

Test for
heterogeneity

(P value)

Poor mental health
Pooling all 39 1.6 1.42 1.81 303.79 87.49 < 0.01 Study type Retrospective/cross-sectional 25 1.8 1.44 2.25 211.78 88.67 < 0.01 Prospective cohort 14 1.39 1.29 1.49 22.12 41.24 0.05 Sex Male 3 2.49 1.86 3.32 0.44 0 0.8 Female 3 2.38 1.41 4 6.95 71.22 0.03 Twins 3 1.41 1.27 1.56 2.5 20.09 0.29 Severity of bullying Sometimes 8 1.5 1.27 1.76 47.68 85.23 < 0.01 Frequent 8 1.52 1.18 1.95 51.4 86.38 < 0.01 Anxiety Pooling all 58 1.77 1.34 2.33 3816.23 98.51 < 0.01 Anxiety 32 1.56 1.39 1.75 434.61 92.87 < 0.01 Social phobia 8 2.48 1.59 3.86 11.01 36.41 0.14 Generalised anxiety disorder 2 2.83 1.38 5.84 0.11 0 0.74 PTSD 12 6.41 1.93 21.22 497.11 97.79 < 0.01 Specific phobia 1 2.4 1 5.6 - - - Separation anxiety disorder 1 4.6 2 10.6 - - - Panic disorder 1 3.1 1.5 6.5 - - - Agoraphobia 1 4.6 1.7 12.5 - - - Study type Retrospective/cross-sectional 39 2.02 1.21 3.38 3697.48 98.97 < 0.01 Prospective cohort 19 1.29 1.06 1.55 84.03 78.58 < 0.01 Age of bullying Less than 13 yr 13 1.4 0.58 3.41 123.12 90.25 < 0.01 Older than 13 yr 45 1.81 1.29 2.56 3688.82 98.81 < 0.01 Sex Male 16 1.84 1.3 2.59 112.23 86.63 < 0.01 Female 15 2.46 1.74 3.48 124.39 88.74 < 0.01 Severity of bullying Sometimes 7 1.46 1.06 2 43.41 86.18 < 0.01 Frequent 25 2.47 1.94 3.14 122.47 80.4 < 0.01 Geographic location and income level Low-to-middle income 22 2.41 1.75 3.32 175.97 88.07 < 0.01 High income 36 1.67 1.24 2.25 3441.17 98.98 < 0.01 Depression Pooling all 92 2.21 1.34 3.65 14525.32 99.37 < 0.01 Major depressive disorder 2 2.27 0.68 7.57 2.07 51.63 0.15 Study type Retrospective/cross-sectional 63 1.95 1.24 3.07 2594.97 97.61 < 0.01 Prospective cohort 29 3.03 1.31 6.98 4583.05 99.39 < 0.01 Age of bullying Less than 13 yr 36 2.11 1.63 2.72 544.9 93.58 < 0.01 Older than 13 yr 56 2.29 1.24 4.23 13806.72 99.6 < 0.01 Sex Male 27 2.07 1.48 2.89 443.84 94.14 < 0.01 Female 21 2.13 1.18 3.86 313.5 93.62 < 0.01 Severity of bullying Sometimes 15 1.78 1.39 2.28 78.61 82.19 < 0.01 Frequent 28 3.26 2.45 4.34 224.43 87.97 < 0.01 Geographic location and income level Low-to-middle income 13 2.53 1.75 3.68 143.98 91.67 < 0.01 High income 79 2.15 1.26 3.68 14351.72 99.46 < 0.01 Psychotic symptoms Specific psychiatric symptoms 6 2.07 1.49 2.87 20.57 75.69 < 0.01 Non-clinical psychotic experiences 9 2.68 2.03 3.54 15.1 47.03 0.06 Psychotic symptoms 5 2.73 1.97 3.77 10.86 63.16 0.03 Personality disorders Anti-social personality disorder 2 0.58 0.15 2.28 2.53 60.48 0.11 Borderline personality disorder 3 2.2 1.4 3.46 4.8 58.31 0.09 Eating disorders Bulimia nervosa 1 3 1.4 6.2 - - - Anorexia nervosa 1 0.004 0 251 - - -

Table 1 Associations between bullying victimization in children and adolescents and mental health outcomes

Moore SE et al . Consequences of bullying victimization

66WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Non-suicidal self injury
Pooling all 30 1.75 1.4 2.19 749.02 96.13 < 0.01 Study type Retrospective/cross-sectional 21 1.55 1.09 2.22 721.15 97.23 < 0.01 Prospective cohort 9 1.65 1.34 2.02 19.39 58.75 0.01 Sex Male 6 4.86 3.35 7.07 13.56 63.12 0.02 Female 4 2.7 2 3.65 8.92 66.37 0.03 Twins 2 2.57 1.79 3.7 0.12 0 0.73 Severity of bullying Sometimes 6 1.57 1.09 2.25 34.04 85.31 < 0.01 Frequent 7 2.52 1.6 3.97 54.49 88.99 < 0.01 Suicidal ideation Pooling all 105 1.77 1.56 2.02 2093.5 95.03 < 0.01 Study type Retrospective/cross-sectional 86 1.8 1.56 2.09 2037.46 95.83 < 0.01 Prospective cohort 19 1.68 1.38 2.05 38.98 53.82 < 0.01 Age of bullying Less than 13 yr 22 1.85 1.48 2.3 74.4 71.77 < 0.01 Older than 13 yr 83 1.75 1.51 2.03 1984.06 95.87 < 0.01 Sex Male 21 1.95 1.64 2.32 76.6 73.89 < 0.01 Female 18 2.15 1.84 2.52 33.15 48.72 0.01 Severity of bullying Sometimes 16 1.53 1.28 1.82 35.19 57.38 < 0.01 Frequent 21 2.59 2.06 3.25 49.83 59.87 < 0.01 Geographic location and income level Low-to-middle income 11 1.31 1.06 1.61 60.02 83.34 < 0.01 High income 94 1.8 1.43 2.26 1894.91 95.09 < 0.01 Suicide attempt Pooling all 48 2.13 1.66 2.73 1110.46 95.77 < 0.01 Suicidal attempt/non-suicidal self injury 3 2.97 1.68 5.23 6.33 68.42 0.04 Study type Retrospective/cross-sectional 40 2.03 1.46 2.84 1105.65 96.47 < 0.01 Prospective cohort 8 2.04 1.38 3.01 4.34 0 0.74 Age of bullying Less than 13 yr 11 2.11 1.65 2.69 11.49 12.98 0.32 Older than 13 yr 37 1.52 0.82 2.83 579.75 93.79 < 0.01 Sex Male 7 2.93 1.65 5.18 15.38 54.5 0.02 Female 7 2.89 1.52 5.49 24.23 71.11 < 0.01 Severity of bullying Sometimes 9 2.19 1.71 2.8 7.52 0 0.48 Frequent 12 3.77 2.55 5.58 28.31 61.14 < 0.01 Geographic location and income level Low-to-middle income 4 1.91 1.07 3.43 22.18 86.47 < 0.01 High income 44 2.17 1.69 2.8 1084.61 96.04 < 0.01 Behavioural problems Pooling all 54 1.37 1.18 1.59 862.4 93.85 < 0.01 Study type Retrospective/cross-sectional 29 1.18 0.99 1.41 311.21 91 < 0.01 Prospective cohort 25 1.56 0.94 2.58 413.53 94.2 < 0.01 Sex Male 9 1.35 0.68 2.67 450.06 98.22 < 0.01 Female 7 1.99 0.97 4.1 88.07 93.19 < 0.01 Twins 2 1.19 0.94 1.5 7.31 86.31 0.01 Severity of bullying Sometimes 8 1.95 0.92 4.1 243.71 97.13 < 0.01 Frequent 8 2.26 0.76 6.69 163.51 95.72 < 0.01 Externalising behaviours 20 1.29 1.12 1.5 157.09 87.9 < 0.01 Delinquent/deviant behaviour 24 1.99 1.24 3.2 423.78 94.57 < 0.01 Missed school 7 1.49 0.99 2.23 38.4 84.37 < 0.01 Disruptive behavioural disorders Attention deficit hyperactivity disorder 1 2.6 0.8 8.5 - - - Oppositional defiant disorder 1 0.8 0.3 2.5 - - - Conduct disorder 1 2.6 0.8 8.8 - - - Other mental health outcomes (not included above) Nervousness 9 1.82 1.51 2.2 135.59 94.1 < 0.01 Powerlessness 2 1.06 0.95 1.18 1.64 39.18 0.2 Feeling low 7 2.26 1.66 3.08 299.47 98 < 0.01 Irritability or bad temper 7 1.82 1.51 2.2 125.07 95.2 < 0.01 Feel helpless 5 3.2 2.01 5.09 273.51 98.54 < 0.01

Moore SE et al . Consequences of bullying victimization

risk of sexual behaviour problems (OR = 1.51; 95%CI:
1.01-2.25) which included teenage pregnancy, early
onset of sexual activities and risky sexual behaviour.

Subgroup analyses were not significant with the excep
tion of those frequently bullied. Although bullying
victimization was associated with increased likelihood of

67WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Feeling tense 3 3.07 2.06 4.56 3.02 33.75 0.22
Unhappy/sad 12 1.25 0.55 2.84 668.61 98.35 < 0.01 Worried 4 1.27 1.09 1.47 314.35 99.05 < 0.01 Afraid 3 2.68 1.18 6.09 9.49 78.92 0.01

Data points Pooled OR 95%CI
lower bound
95%CI
upper bound
Cochran’s
Q
I ²
(%)
Test for
heterogeneity
(P value)

Alcohol use
Pooling all 53 1.26 1.00 1.58 10328.18 99.5 < 0.01 Study type Retrospective/cross-sectional 38 1.28 0.88 1.84 10256.15 99.64 < 0.01 Prospective cohort 15 1.19 0.87 1.62 67.87 79.37 < 0.01 Sex Male 6 0.61 0.49 0.77 4.89 0.00 0.43 Female 4 0.88 0.50 1.57 6.36 52.86 0.1 Severity of bullying Sometimes 6 1.72 0.84 3.50 86.17 94.20 < 0.01 Frequent 13 1.53 0.78 3.03 332.27 96.39 < 0.01 Frequency of alcohol consumption Sometimes 24 1.52 1.08 2.13 8416.94 99.73 < 0.01 Frequent 29 0.99 0.86 1.14 251.81 88.88 < 0.01 Age of bullying Less than 13 yr 16 1.23 0.93 1.63 1618.14 99.07 < 0.01 Older than 13 yr 37 1.31 0.96 1.80 6896.71 99.48 < 0.01 Geographic location and income level Low-to-middle income 11 1.37 0.75 2.49 8328.21 99.88 < 0.01 High income 42 1.08 0.91 1.27 462.84 91.14 < 0.01 Tobacco use Pooling all 35 1.36 0.96 1.92 418.71 91.88 < 0.01 Study type Retrospective/cross-sectional 26 1.17 0.59 2.31 394.92 93.67 < 0.01 Prospective cohort 9 1.62 1.31 1.99 11.48 30.33 0.18 Sex Male 3 0.97 0.59 1.58 8.23 75.7 0.02 Female 3 0.51 0.37 0.68 1.78 0 0.41 Severity of bullying Sometimes 4 1.89 0.83 4.33 71.38 95.8 < 0.01 Frequent 4 3.19 1.19 8.58 39.85 92.47 < 0.01 Frequency of smoking Sometimes 28 1.36 0.89 2.06 400.72 93.26 < 0.01 Frequent 7 1.35 1.00 1.84 16.28 63.16 0.01 Illicit drug use Pooling all 34 1.41 1.10 1.81 677.62 95.13 < 0.01 Study type Retrospective/cross-sectional 11 2.43 1.42 4.15 297.3 96.64 < 0.01 Prospective cohort 23 1.27 1.12 1.44 67.23 67.28 < 0.01 Sex Male 7 1.04 0.81 1.33 32.84 81.73 < 0.01 Female 5 1.17 1.03 1.33 3.26 0.00 0.52 Severity of bullying Sometimes 7 1.22 0.78 1.90 160.47 96.26 < 0.01 Frequent 8 1.14 0.43 3.00 465.43 98.50 < 0.01 Geographic location and income level Low-to-middle income 5 4.05 2.18 7.55 98.61 95.94 < 0.01 High income 29 1.31 1.15 1.48 103.73 73.01 < 0.01 Cannabis only all 9 1.42 0.96 2.12 23.32 65.70 < 0.01 Study type Retrospective/cross-sectional 1 2.46 1.53 3.95 - - - Prospective cohort 8 1.36 0.90 2.05 18.14 61.41 0.01

Table 2 Associations between bullying victimization in children and adolescents and substance use

Moore SE et al . Consequences of bullying victimization

68WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Data points Pooled OR 95%CI
lower bound
95%CI
upper bound
Cochran’s
Q
I ²
(%)
Test for
heterogeneity
(P value)

Somatic symptoms
Unspecified psychosomatic symptoms 25 2.00 1.54 2.60 232.02 89.66 < 0.01 Stomach ache 25 1.76 1.53 2.03 138.73 82.7 < 0.01 Sleeping difficulties 24 1.73 1.46 2.05 574.91 96 < 0.01 Headache 26 1.64 1.38 1.94 169.16 85.22 < 0.01 Bedwetting 3 2.51 1.44 4.37 4.93 59.45 0.08 Feeling tired 2 2.68 1.39 5.19 1.22 17.87 0.27 Poor appetite 2 2.23 1.60 3.12 0 0 0.95 Back pain 8 1.67 1.43 1.95 73.53 90.48 < 0.01 Skin problems 1 1.82 1.33 251 - - - Dizziness 9 1.64 1.38 1.95 76.57 89.55 < 0.01 Eating and weight related problems Binge eating 2 2.66 1.68 4.22 0.57 0 0.45 Non-diet soft drink consumption 1 1.21 1.04 1.41 - - - Skips breakfast 6 1.41 1.20 1.65 11.89 57.94 0.04 Underweight

Pooling all 2 1.27 0.73 2.21 0 0 0.96
Sex
Male 1 1.28 0.69 2.37 – – –
Female 1 1.24 0.19 2.29 – – –
Overweight
Pooling all 14 1.68 1.21 2.33 82.69 84.28 < 0.01 Study type Retrospective/cross- sectional 12 1.99 1.39 2.85 65.45 83.19 < 0.01 Prospective cohort 2 0.98 0.64 1.49 1.92 47.97 0.17 Sex Male 7 1.22 0.99 1.49 8.04 25.4 0.23 Female 7 2.22 1.28 3.84 50.17 88.04 < 0.01 Severity of bullying Sometimes 2 1.32 1.00 1.74 0.09 0 0.77 Frequent 6 1.14 0.88 1.47 7.37 32.2 0.19 Obese Pooling all 13 1.78 1.42 2.21 14.68 18.28 0.26 Study type Retrospective/cross-sectional 10 1.97 1.53 2.53 7.22 0 0.61 Prospective cohort 3 1.57 0.89 2.77 6.89 70.97 0.03 Sex Male 6 1.94 1.45 2.60 4.88 0 0.43 Female 6 2.15 1.57 2.94 2.22 0 0.82 Severity of bullying Sometimes 2 1.63 1.11 2.38 0.52 0 0.47 Frequent 6 2.09 1.59 2.75 5.26 4.86 0.39 Sexual behaviour problems Teen parent 5 1.26 0.81 1.97 15.11 73.53 < 0.01 Risky sexual behaviour 4 2.28 0.95 5.48 23.43 87.2 < 0.01 Early onset of sexual activities 3 1.44 0.90 2.30 7.38 72.91 0.02 Pooling all 12 1.51 1.01 2.25 85.66 87.16 < 0.01 Study type Retrospective/cross-sectional 3 1.77 0.42 7.52 54.97 96.36 < 0.01 Prospective cohort 9 1.34 0.98 1.84 23.88 66.51 < 0.01 Severity of bullying Sometimes 2 0.81 0.51 1.28 2.46 59.32 0.12 Frequent 4 2.38 1.05 5.41 27.55 89.11 < 0.01 Health services utilised Pooling all 16 1.20 0.99 1.45 34.37 56.36 < 0.01 Study type Retrospective/cross-sectional 14 1.14 0.94 1.39 27.91 53.42 0.01 Prospective cohort 2 1.54 0.65 3.61 4.43 77.43 0.04 Sex Male 7 1.17 0.95 1.43 3.54 0 0.74 Female 7 1.41 1.12 1.77 11.16 46.25 0.08 General medication use Pooling all 12 1.16 0.80 1.70 117.98 90.68 < 0.01 Study type Retrospective/cross-sectional 11 0.99 0.56 1.75 112.26 91.09 < 0.01

Table 3 Associations between bullying victimization in children and adolescents and other health outcomes

Moore SE et al . Consequences of bullying victimization

poor general health (OR = 1.83; 95%CI: 1.452.31) and
this association persisted when restricted to prospective
cohort studies (OR = 1.56; 95%CI: 1.072.28). There
was no consistent increase in utilisation of health ser-
vices or medications in those exposed to bullying victimi
zation during childhood or adolescence (Table 3).

Bullying victimization in children and adolescents and
academic and social functioning
The association between bullying victimization and
functioning at school was inconsistent. There was a robust
association between bullying victimization in childhood
or adolescence and poor academic achievement. Those
who had exposure to bullying victimization were more
likely to have poor academic achievement (OR = 1.33;
95%CI: 1.061.66), whilst those with good academic
achievement were less likely to have been exposed to
bullying victimization (OR = 0.71; 95%CI: 0.600.85);
however, all studies except one were crosssectional.
Bullying victimization was not associated with later
financial or occupational functioning.

Similarly, there were inconsistent associations bet-
ween bullying victimization and social problems. Those

exposed were approximately twice as likely to report
loneliness (OR = 1.89; 95%CI: 1.392.57) and poor life
satisfaction (OR = 2.26; 95%CI: 1.413.60) and were
significantly less likely to have a good quality of life (OR
= 0.85; 95%CI: 0.780.93). Bullying victimization was
not consistently associated with low self-esteem, social
problems, or criminal behaviours (Table 4).

DISCUSSION
This paper provides the most comprehensive critical
analysis of the association between bullying victimization
and a wide range of health and psychosocial problems.
The primary and sub-group analyses allow for inter-
pretation of the evidence of causality within the Bradford-
Hill Framework, based on the following: Biological
plausibility, the temporal relationship of the association,
strength and consistency of the association, the presence
of a dose-response relationship, and whether an alter-
nate explanation for the associations is possible[49].
We used the grading system developed by the World
Cancer Research Fund[50] as used in the Global Burden of
Disease study as a guideline for evaluation of the level of

69WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Prospective cohort 1 1.67 1.09 2.58 – – –
Sex
Male
Medication for headache 2 1.43 1.06 1.93 2.34 57.21 0.13
Medication for stomach-ache 2 1.09 0.72 1.65 1.9 47.5 0.17
Female
Medication for headache 2 1.19 0.98 1.45 1.33 24.79 0.25
Medication for stomach-ache 2 1.23 1.01 1.5 0.27 0 0.61
Severity of bullying
Sometimes 5 1.26 0.99 1.59 15.46 74.13 < 0.01 Frequent 5 1.72 1.11 2.67 24.9 83.93 < 0.01 Over the counter drug misuse 3 0.95 0.19 4.66 76.34 97.38 < 0.01 Psychotropic medication use Pooling all 13 1.28 0.72 2.26 205.76 94.17 < 0.01 Study type Retrospective/cross-sectional 11 0.95 0.32 2.8 195.34 94.88 < 0.01 Prospective cohort 2 1.31 0.66 2.6 5.61 82.18 0.02 Sex Male Medication for nervousness 2 1.32 0.42 4.1 14.98 93.32 < 0.01 Medication for sleeping 2 1.89 1.33 2.67 1.59 37.28 0.21 Female Medication for nervousness 2 1.97 1.49 2.59 1.06 5.88 0.3 Medication for sleeping 2 1.83 1.42 2.36 0.27 0 0.6 Severity of bullying Sometimes 6 1.66 1.26 2.18 18.54 73.04 < 0.01 Frequent 6 1.88 1.17 3.03 31.93 84.34 < 0.01 Prescription drug misuse 3 0.92 0.17 5.07 88.16 97.73 < 0.01 Poor general health Pooling all 29 1.83 1.45 2.31 133.31 79 < 0.01 Study type Retrospective/cross-sectional 22 1.71 1.21 2.42 112.06 81.26 < 0.01 Prospective cohort 7 1.56 1.07 2.28 19.72 69.57 < 0.01 Sex Male 9 1.95 1.16 3.27 20.5 60.98 0.01 Female 9 2.36 1.11 5.04 52.78 84.84 < 0.01 Severity of bullying Sometimes 4 2.16 1.10 4.26 9.15 67.21 0.03 Frequent 4 6.96 2.17 22.35 16.72 82.05 < 0.01

Moore SE et al . Consequences of bullying victimization

70WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Data points Pooled OR 95%CI
lower bound
95%CI
upper bound
Cochran’s
Q
I ²
(%)
Test for
heterogeneity
(P value)

Poor school functioning
Pooling all 6 1.10 0.87 1.38 82 93.9 < 0.01 Study type Retrospective/cross-sectional 3 1.24 1.22 1.27 0.33 0 0.85 Prospective cohort 3 0.90 0.76 1.08 8.15 75.46 0.02 Severity of bullying Sometimes 1 0.96 0.88 1.04 - - - Frequent 1 0.98 0.76 1.19 - - - Academic achievement Poor academic achievement Pooling all 6 1.33 1.06 1.66 11.17 55.25 0.02 Study type Retrospective/cross-sectional 6 1.33 1.06 1.66 11.17 55.25 0.02 Good academic achievement Pooling all 4 0.71 0.60 0.85 8.81 65.97 0.07 Study type Retrospective/cross-sectional 3 0.86 0.8 0.92 2.89 30.69 0.58 Prospective cohort 1 0.46 0.28 0.76 - - - Sex Male 2 1.24 0.88 1.74 2.49 59.8 0.65 Female 2 1.32 0.99 1.75 1.4 28.7 0.84 Severity of bullying Sometimes 1 0.88 0.83 0.93 - - - Frequent 1 0.80 0.70 0.93 - - - Poor financial and occupational functioning Pooling all prospective cohort 16 1.14 0.87 1.50 92.97 83.86 < 0.01 Severity of bullying Sometimes 3 1.00 0.9 1.11 0.04 0 0.98 Frequent 3 0.81 0.61 1.07 2.68 25.32 0.26 Social isolation Loneliness Pooling all 13 1.89 1.39 2.57 3120.66 99.62 < 0.01 Study type Retrospective/cross-sectional 13 1.89 1.39 2.57 3120.66 99.62 < 0.01 Sex Male 4 2.58 1.62 4.10 222.21 98.65 < 0.01 Female 3 3.92 1.95 7.90 19.53 89.76 < 0.01 Severity of bullying Sometimes 2 2.09 1.98 2.20 0.39 0 0.53 Frequent 4 4.12 2.24 7.60 23.32 87.13 < 0.01 Self esteem Pooling all 14 0.99 0.92 1.07 93.73 86.13 < 0.01 Study type Retrospective/cross-sectional 4 1.13 0.83 1.54 76.58 96.08 < 0.01 Prospective cohort 10 0.97 0.93 1.01 12.32 26.93 0.2 Sex Male 5 0.96 0.88 1.06 20.65 80.63 < 0.01 Female 4 0.95 0.88 1.03 5.74 47.7 0.13 Severity of bullying Sometimes 5 0.99 0.95 1.04 1.61 0 0.81 Frequent 5 0.95 0.87 1.04 9.65 58.54 0.05 Social problems Pooling all 22 1.02 0.74 1.42 427.13 95.08 < 0.01 Study type Retrospective/cross-sectional 5 2.86 1.42 5.76 38.09 89.5 < 0.01 Prospective cohort 17 0.89 0.74 1.06 72.36 77.89 < 0.01 Sex Male 1 2.89 1.45 5.73 - - - Female 1 8.10 4.60 14.26 - - - Severity of bullying Sometimes 3 0.9 0.83 0.96 0.5 0 0.78 Frequent 3 0.81 0.72 0.92 3.67 45.49 0.16 Criminal behaviour Pooling all 33 1.04 0.78 1.39 133.36 76.01 < 0.01 Carrying a weapon 8 1.59 1.27 1.98 19.16 63.47 0.01

Table 4 Associations between bullying victimization and academic and social functioning

Moore SE et al . Consequences of bullying victimization

evidence.

Temporality
In this meta-analysis, both longitudinal (n = 57) and
cross-sectional (n = 108) studies showed associations
between bullying victimization and many adverse health
and psychosocial problems. Prospective studies provided
evidence of a temporal relationship showing bullying
victimization preceded the later adverse consequences.

A temporal relationship exists between bullying
victimization and outcomes such as anxiety, depression,
non-suicidal self-injury, suicide ideation and suicide
attempts. As poor mental health is also a known risk
factor for bullying victimization[51], it is with caution we
say that an independent temporal relationship exists
between bullying victimization and these adverse
mental health outcomes. Many studies did not control
for preexisting mental health and could be reporting
a continuation of preexisting psychopathology and
not a direct outcome of the bullying victimization.
Nonetheless, two recent studies have found that even
when controlling for preexisting mental health, bullying
victimization was strongly associated with later adverse
mental health consequences such as non-suicidal self-
injury and depression[27,28].

Strength of the association
Both prospective and population-based studies demon-
strated significant associations between bullying
victimization and adverse health and psychosocial
problems. After adjusting for confounding variables,
there was generally a reduction in the strength of
these associations. Furthermore, the magnitude of
the associations diverged depending on the sub-group
analysis performed. Despite some variability, bullying
victimization was found to significantly increase the
likelihood of mental ill health suggesting significant and
robust associations.

Consistency of the association
Consistency of the associations between bullying
victimization and mental ill health was demonstrated in
the estimated effect sizes across studies. It is possible
that publication bias affected the results for some of the

outcomes. Direction of the association (as estimated
through risk estimates) was consistent across different
geographic regions, samples, study designs, and income
levels investigated, particularly for anxiety, depression,
non-suicidal self-injury, suicide ideation and suicide
attempts (Supplementary material Figures S4, S5,
S7, S8, S9). Inconsistent associations were observed
for certain outcomes such as behavioural problems
(Supplementaty material Figure S6).

Dose-response relationship
Available evidence suggests that experiencing more
severe or frequent forms of adversity in childhood
increases the risk of adverse outcomes compared to
a lower exposure to adversity[45,52-56], particularly for
mental health problems. Similarly, this study demon-
strated a dose-response relationship between bullying
victimization and detrimental effects on health, in
particular for mental health problems. After summarizing
the evidence through a meta-analysis, dose-response
relationships were observed between bullying victimi-
zation and depression, suicide ideation, cigarette smok-
ing and loneliness[45,52-56]. An increase in the dose of
bullying victimization (frequent vs sometimes) resulted
in non-significantly greater point estimates for other
problems such as anxiety, medication use (general and
psychotropic), suicide attempts and non-suicidal self-
injury.

Plausibility
Due to a lack of animal models, the majority of infer
ences for biological plausibility arise from observational
rather than experimental data. However, one model
of social defeat in rats has been used to understand
bullying victimization[57,58]. Two male rats are placed into
a cage together, and after fighting, one rat becomes
dominant and the other subordinate. The subordinate rat
experiences social defeat and after a single experience
demonstrates signs of stress. One study found that the
subordinate rat demonstrated behaviours representative
of depression in humans when exposed to multiple social
defeats over several weeks[59].

Observational data has also been used to explain the
association between bullying victimization in childhood

71WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Violent offense/behaviour 6 1.25 1.01 1.56 2.46 0 0.78
Study type
Retrospective/cross-sectional 9 1.01 0.47 2.14 106.83 92.51 < 0.01 Prospective cohort 24 1.05 0.92 1.19 25.72 10.58 0.31 Sex Male 11 1.00 0.82 1.22 13.78 27.43 0.18 Female 4 0.70 0.46 1.04 0.38 0 0.94 Severity of bullying Sometimes 5 0.97 0.75 1.26 7.37 45.74 0.12 Frequent 8 1.22 0.86 1.74 16.33 57.13 0.02 Other outcomes reported Good quality of later life 6 0.85 0.78 0.93 17.42 71.29 < 0.01 Poor life satisfaction 6 2.26 1.41 3.60 3.79 0 0.58 Problematic internet usage 1 2.36 1.58 3.54 - - - Picked on by siblings 1 1.69 1.38 2.07 - - -

Moore SE et al . Consequences of bullying victimization

and adolescence and the later development of mental
health problems. First, early adverse experiences (i.e.,
bullying victimization) that occur during vulnerable
developmental periods can cause neurobiological[60,61] or
inflammatory[62] changes expressed as illnesses in later
life[61]. Moreover, those individuals exposed to frequent
bullying victimization who develop mental health
problems may self-medicate their distress and negative
emotions with alcohol, illicit drugs, medications, tobacco
or disengaging from school.

Taking into account both the limited animal stu-
dies[57-59] and observational studies[60-62], it can be under-
stood as to why bullying victimization can affect the
immediate and long-term health and non-health related
outcomes of the individual.

Consideration of alternate explanations
The relationship between bullying victimization and
adverse health and psychosocial problems are thought
to be complex and influenced by both genetics and
environmental factors; however, there are limited twin
studies available to inform these associations[27,63-66].
One study[64] found that being bullied in childhood is an
environmentally mediated contributing factor to poor
childhood mental health. Another found victimized twins
were more likely to self-harm than their non-victimized
twin sibling[27]. Exposure to bullying victimization has
also been found to be associated with socioeconomic
status[51,67] which is also known to play a role in the
development of mental health problems and other
health and non-health related outcomes[68].

It is further acknowledged that the association
between bullying victimization and adverse outcomes
is not necessarily an independent relationship. As early
emotional and behavioural problems are known risk
factors for bullying victimization, without adequate
statistical adjustment, some studies may risk reporting
preexisting psychopathology rather than a direct
outcome of bullying. The available evidence suggests a
complex relationship between genetics and environment
and neither can solely explain the relationship between
bullying victimization and adverse outcomes. Even
though some of the effects of bullying victimization
on adverse outcomes reported may be a result of
confounding factors, generally the association with
mental health problems was significant after controlling
for potential confounding factors.

Assessment of causality
Using the grading system developed by the World
Cancer Research Fund (WCRF)[50] as a guideline for
evaluation of the level of evidence, we concluded that
there was “convincing evidence” for a causal relationship
between bullying victimization and anxiety, depression,
poor general and mental health, non-suicidal self-injury,
suicide attempts, and suicide ideation. This evidence
was based on a substantial number of epidemiological
studies identified in this systematic review including
prospective observational studies of sufficient size,
duration, and quality showing consistent effects. In
addition, the association was considered biologically
plausible. We concluded that “probable evidence” of
a causal relationship existed between exposure to
bullying victimization and illicit drug and tobacco use
based on the epidemiological evidence. Possible causal
associations existed between bullying victimization and
lower academic achievement, alcohol use, loneliness,
obesity, overweight and psychosomatic symptoms.
This evidence was based mainly on findings from cross
sectional studies and a few prospective studies showing
inconsistent associations between exposure and disease.
More studies are needed to support these tentative
associations, which are also considered to be biologically
plausible.

All other significant associations reported in this study
were classified as having insufficient evidence of a causal
relationship (Table 5). This is not suggesting that there
is no causal relationship. Further research is needed to
better examine if any associations that exist are causal
or due to other confounding factors. Furthermore, the
use of WCRF grading system, although appropriate
for dietary risk factors, might not be adequate for
psychosocial factors particularly newly emerging risks.

Limitations
While we followed rigorous methodological steps, some
limitations are notable. As studies with non-significant
findings are less likely to be published, there may be a
publication bias within this meta-analysis resulting in
the association between bullying victimization and some
adverse outcomes being overstated[69,70]. Additionally,
inconsistencies would have occurred in the analysis
due to methodological differences in the way bullying
victimization is defined and measured throughout the
studies as there is no consensus on the best way to

72WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Strength of evidence Adverse health or psychosocial problem

Convincing Anxiety; depression; poor mental health; poor general health; non-suicidal self-injury; suicide attempts; suicide ideation
Probable Tobacco use; illicit drug use
Possible Alcohol use; psychotic symptoms; increased use of health services in females; lower academic achievement; social isolation;

loneliness; psychosomatic symptoms, overweight and obesity
Insufficient Binge eating; bulimia nervosa; borderline personality disorder; behavioural problems; carrying a weapon; general

medication use; health services sought; poor financial and occupational functioning; psychotropic medication use; poor
school functioning; sexual behavioural problems; poor life satisfaction

Table 5 Strength of evidence for a causal relationship between bullying victimization and adverse health or psychosocial problems

Moore SE et al . Consequences of bullying victimization

measure bullying victimization[18,19]. In order to address
this, a quality effects model was used giving higher
scores to those studies which provided respondents
with a definition and utilised a validated measure of
bullying. There are also methodological issues in regards
to the adverse outcomes reported, as some have been
self-reported, while others were reported by teachers,
parents, clinicians or through objective measures. This
issue was also addressed with the use of a quality
effects model in which higher quality scores were given
to those studies where standardised validated diagnostic
instruments were used to assess the outcome relative
to those where outcomes were self-reported on a non-
validated scale[44]. In spite of this methodology, the
assessment of exposure to bullying and the assessment
of a wide range of outcomes remains a challenge.
In particular, there will always be some uncertainty
pertaining to the measurement of bullying, especially
when retrospectively reported as a result of the
respondent’s subjective perception of the actions and
behaviours of others.

As a research question involving bullying victimization
can only be observational and not experimental, a further
limitation of this meta-analysis are those limitations that
come with observational studies[71]. First, we acknow-
ledge the issue of confounding. It is appropriate to adjust
for these confounders in the statistical analyses by either
stratification or multivariate analysis[71]. Although many
studies controlled for socio-demographic and other
variables[2,27], some reported unadjusted odds ratios
between bullying victimization and adverse outcomes,
or provided only basic adjustment for sex and age[72,73].
This was addressed in this meta-analysis through the
use of the quality score of studies where confounding
factors were not adequately adjusted and by conducting
further analyses where data were available[44]. Generally,
after controlling for the effects of confounding variables,
the associations between bullying victimization and
adverse outcomes were attenuated. The majority of
studies included in this meta-analysis did not identify
individuals who were both victims and perpetrators of
bullying. Previous research has suggested those who are
both perpetrators and victims are at even greater risk of
adverse mental health outcomes[28]; however, we were
unable to confirm this with the current study.

In the majority of primary analyses of the associa-
tion between bullying victimization and adverse out-
comes, significant heterogeneity was present. This
heterogeneity remained significant in most subgroup
analyses even after controlling for study quality in the
quality effects models[44].

In conclusion, evidence suggests a causal relation-
ship between bullying victimization and mental health
outcomes. There were also associations between
bullying victimization and other adverse health and
psychosocial problems which require further research
to accurately measure the negative impact of bullying
victimization and the broad health and economic

costs. Through the implementation of school wide
interventions that involve the entire school community
(i.e., staff, students, and parents) bullying behaviour
is considered a modifiable risk factor[25,74]. This review
highlights the increased likelihood of a wide and
diverse range of problems that are experienced by
those exposed to bullying victimization. These findings
reinforce the need for implementation of effective
interventions in schools to address the high prevalence
of children and adolescents engaging in bullying
behaviours.

COMMENTS
Background
Bullying victimization (including traditional and cyberbullying) among children
and adolescents is a global public health issue, well-recognised as a behaviour
associated with poor adjustment in youth. There is evidence suggesting bullying
victimization in children and adolescents has enduring effects which may persist
into adulthood.

Research frontiers
There have been many studies examining the association between bullying
victimization in children and adolescents and adverse health and social
problems. However, many of these have not been systematically examined and
existing systematic reviews did not include cyberbullying. Furthermore, although
associations exist, it is unclear if there is a causal relationship.

Innovations and breakthroughs
The authors found convincing evidence of a causal relationship between bullying
victimization in children and adolescents and adverse health outcomes including
anxiety, depression, poor mental health, poor general health, non-suicidal
self-injury, suicidal ideation and suicide attempts. It is probable that bullying
victimization also causes an increased risk of cigarette smoking and illicit drug
use.

Applications
Given the convincing evidence of a causal association, there is an urgent need
for effective interventions to be implemented in schools to address the high
prevalence of children and adolescents engaging in bullying behaviours.

Peer-review
This is an important topic on consequences of bullying victimization in childhood
and adolescence. This area for sure needs more attention. The authors have
done a great job presenting a large systematic review and meta-analysis of
studies correlating the history of bullying victimization with different mental health
problems in childhood and adolescence.

REFERENCES
1 Hawker DS, Boulton MJ. Twenty years’ research on peer victimi-

zation and psychosocial maladjustment: a meta-analytic review
of cross-sectional studies. J Child Psychol Psychiatry 2000; 41:
441-455 [PMID: 10836674 DOI: 10.1111/1469-7610.00629]

2 Copeland WE, Wolke D, Angold A, Costello EJ. Adult psychiatric
outcomes of bullying and being bullied by peers in childhood
and adolescence. JAMA Psychiatry 2013; 70: 419-426 [PMID:
23426798 DOI: 10.1001/jamapsychiatry.2013.504]

3 Takizawa R, Maughan B, Arseneault L. Adult health outcomes
of childhood bullying victimization: evidence from a five-decade
longitudinal British birth cohort. Am J Psychiatry 2014; 171:
777-784 [PMID: 24743774 DOI: 10.1176/appi.ajp.2014.13101401]

4 Sigurdson JF, Undheim AM, Wallander JL, Lydersen S, Sund AM.

73WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

COMMENTS

Moore SE et al . Consequences of bullying victimization

The longterm effects of being bullied or a bully in adolescence on
externalizing and internalizing mental health problems in adulthood.
Child Adolesc Psychiatry Ment Health 2015; 9: 42 [PMID:
26300969 DOI: 10.1186/s13034-015-0075-2]

5 Olweus D. Bullying at school: What we know and what we can do.
Oxford, UK: Blackwell, 1993

6 Olweus D. Bullying at school: basic facts and effects of a school
based intervention program. J Child Psychol Psychiatry 1994;
35: 1171-1190 [PMID: 7806605 DOI: 10.1111/j.1469-7610.1994.
tb01229.x]

7 Smith PK, Mahdavi J, Carvalho M, Fisher S, Russell S, Tippett N.
Cyberbullying: its nature and impact in secondary school pupils.
J Child Psychol Psychiatry 2008; 49: 376-385 [PMID: 18363945
DOI: 10.1111/j.1469-7610.2007.01846.x]

8 Sourander A, Brunstein Klomek A, Ikonen M, Lindroos J,
Luntamo T, Koskelainen M, Ristkari T, Helenius H. Psychosocial
risk factors associated with cyberbullying among adolescents: a
population-based study. Arch Gen Psychiatry 2010; 67: 720-728
[PMID: 20603453 DOI: 10.1001/archgenpsychiatry.2010.79]

9 Cross D, Shaw T, Hearn L, Epstein M, Monks H, Lester L, Thomas
L. Australian covert bullying prevalence study (ACBPS). Perth,
Australia Child Health Promotion Research Centre, Edith Cowan
University, 2009

10 Jansen PW, Verlinden M, Dommissevan Berkel A, Mieloo C,
van der Ende J, Veenstra R, Verhulst FC, Jansen W, Tiemeier
H. Prevalence of bullying and victimization among children in
early elementary school: do family and school neighbourhood
socioeconomic status matter? BMC Public Health 2012; 12: 494
[PMID: 22747880 DOI: 10.1186/1471-2458-12-494]

11 Rigby K, Slee PT. Bullying among Australian school children:
reported behavior and attitudes toward victims. J Soc Psychol 1991;
131: 615-627 [PMID: 1798296 DOI: 10.1080/00224545.1991.9924
646]

12 Solberg ME, Olweus D. Prevalence estimation of school bullying
with the Olweus Bully/Victim Questionnaire. Aggress Behav 2003;
29: 239-268 [DOI: 10.1002/ab.10047]

13 Due P, Holstein BE, Soc MS. Bullying victimization among 13 to
15-year-old school children: results from two comparative studies
in 66 countries and regions. Int J Adolesc Med Health 2008; 20:
209-221 [PMID: 18714557 DOI: 10.1515/IJAMH.2008.20.2.209]

14 Sourander A, Jensen P, Rönning JA, Niemelä S, Helenius H,
Sillanmäki L, Kumpulainen K, Piha J, Tamminen T, Moilanen I,
Almqvist F. What is the early adulthood outcome of boys who bully
or are bullied in childhood? The Finnish “From a Boy to a Man”
study. Pediatrics 2007; 120: 397-404 [PMID: 17671067 DOI:
10.1542/peds.2006-2704]

15 Hemphill SA, Kotevski A, Herrenkohl TI, Bond L, Kim MJ,
Toumbourou JW, Catalano RF. Longitudinal consequences of
adolescent bullying perpetration and victimisation: a study of
students in Victoria, Australia. Crim Behav Ment Health 2011; 21:
107-116 [PMID: 21370296 DOI: 10.1002/cbm.802]

16 Nansel TR, Overpeck M, Pilla RS, Ruan WJ, SimonsMorton
B, Scheidt P. Bullying behaviors among US youth: prevalence
and association with psychosocial adjustment. JAMA 2001; 285:
2094-2100 [PMID: 11311098 DOI: 10.1001/jama.285.16.2094]

17 Cook CR, Williams KR, Guerra NG, Kim TE. Variability in the
prevalence of bullying and victimization: A cross-national and
methodological analysis. In: Jimerson SR, Swearer S, Espelage
Dorothy L. Handbook of bullying in schools: An international
perspective. New York: Routledge, 2010: 347-362

18 Griffin RS, Gross AM. Childhood bullying: Current empirical
findings and future directions for research. Aggress Violent Beh
2004; 9: 379-400 [DOI: 10.1016/S1359-1789(03)00033-8]

19 Shaw T, Dooley JJ, Cross D, Zubrick SR, Waters S. The Forms
of Bullying Scale (FBS): validity and reliability estimates for a
measure of bullying victimization and perpetration in adolescence.
Psychol Assess 2013; 25: 1045-1057 [PMID: 23730831 DOI:
10.1037/a0032955]

20 Modecki KL, Minchin J, Harbaugh AG, Guerra NG, Runions KC.
Bullying prevalence across contexts: a meta-analysis measuring

cyber and traditional bullying. J Adolesc Health 2014; 55: 602-611
[PMID: 25168105 DOI: 10.1016/j.jadohealth.2014.06.007]

21 Thomas HJ, Connor JP, Scott JG. Integrating traditional bullying
and cyberbullying: Challenges of definition and measurement in
adolescents – a review. Educ Psychol Rev 2015; 27: 135-152 [DOI:
10.1007/s10648-014-9261-7]

22 Ybarra ML, Boyd D, Korchmaros JD, Oppenheim JK. Defining
and measuring cyberbullying within the larger context of bullying
victimization. J Adolesc Health 2012; 51: 53-58 [PMID: 22727077
DOI: 10.1016/j.jadohealth.2011.12.031]

23 Kowalski RM, Giumetti GW, Schroeder AN, Lattanner MR.
Bullying in the digital age: a critical review and meta-analysis
of cyberbullying research among youth. Psychol Bull 2014; 140:
1073-1137 [PMID: 24512111 DOI: 10.1037/a0035618]

24 Coomber K, Toumbourou JW, Miller P, Staiger PK, Hemphill SA,
Catalano RF. Rural adolescent alcohol, tobacco, and illicit drug use:
a comparison of students in Victoria, Australia, and Washington
State, United States. J Rural Health 2011; 27: 409-415 [PMID:
21967385 DOI: 10.1111/j.1748-0361.2010.00360.x]

25 Vreeman RC, Carroll AE. A systematic review of school-based
interventions to prevent bullying. Arch Pediatr Adolesc Med 2007;
161: 78-88 [PMID: 17199071 DOI: 10.1001/archpedi.161.1.78]

26 Klomek AB, Sourander A, Kumpulainen K, Piha J, Tamminen T,
Moilanen I, Almqvist F, Gould MS. Childhood bullying as a risk
for later depression and suicidal ideation among Finnish males. J
Affect Disord 2008; 109: 47-55 [PMID: 18221788 DOI: 10.1016/
j.jad.2007.12.226]

27 Fisher HL, Moffitt TE, Houts RM, Belsky DW, Arseneault L, Caspi
A. Bullying victimisation and risk of self harm in early adolescence:
longitudinal cohort study. BMJ 2012; 344: e2683 [PMID: 22539176
DOI: 10.1136/bmj.e2683]

28 Moore SE, Norman RE, Sly PD, Whitehouse AJ, Zubrick SR, Scott
J. Adolescent peer aggression and its association with mental health
and substance use in an Australian cohort. J Adolesc 2014; 37:
11-21 [PMID: 24331300 DOI: 10.1016/j.adolescence.2013.10.006]

29 Reijntjes A, Kamphuis JH, Prinzie P, Boelen PA, van der Schoot
M, Telch MJ. Prospective linkages between peer victimization and
externalizing problems in children: a meta-analysis. Aggress Behav
2011; 37: 215-222 [PMID: 21433031 DOI: 10.1002/ab.20374]

30 Reijntjes A, Kamphuis JH, Prinzie P, Telch MJ. Peer victimization
and internalizing problems in children: a meta-analysis of longitudinal
studies. Child Abuse Negl 2010; 34: 244-252 [PMID: 20304490 DOI:
10.1016/j.chiabu.2009.07.009]

31 Ttofi MM, Farrington DP, Lösel F, Loeber R. Do the victims of
school bullies tend to become depressed later in life? A systematic
review and meta‐analysis of longitudinal studies. J Aggress Confl
Peace Res 2011; 3: 63-73 [DOI: 10.1108/17596591111132873]

32 Gini G, Pozzoli T. Association between bullying and psychosomatic
problems: a meta-analysis. Pediatrics 2009; 123: 1059-1065 [PMID:
19255040 DOI: 10.1542/peds.2008-1215]

33 Holt MK, VivoloKantor AM, Polanin JR, Holland KM, DeGue S,
Matjasko JL, Wolfe M, Reid G. Bullying and suicidal ideation and
behaviors: a meta-analysis. Pediatrics 2015; 135: e496-e509 [PMID:
25560447 DOI: 10.1542/peds.2014-1864]

34 van Geel M, Vedder P, Tanilon J. Relationship between peer
victimization, cyberbullying, and suicide in children and adolescents:
a meta-analysis. JAMA Pediatr 2014; 168: 435-442 [PMID: 24615300
DOI: 10.1001/jamapediatrics.2013.4143]

35 Kim YS, Leventhal B. Bullying and suicide. A review. Int J Adolesc
Med Health 2008; 20: 133-154 [PMID: 18714552 DOI: 10.1515/
IJAMH.2008.20.2.133]

36 Tharp-Taylor S, Haviland A, D’Amico EJ. Victimization from
mental and physical bullying and substance use in early adolescence.
Addict Behav 2009; 34: 561-567 [PMID: 19398162 DOI: 10.1016/
j.addbeh.2009.03.012]

37 Goebert D, Else I, Matsu C, Chung-Do J, Chang JY. The impact of
cyberbullying on substance use and mental health in a multiethnic
sample. Matern Child Health J 2011; 15: 1282-1286 [PMID:
20824318 DOI: 10.1007/s10995-010-0672-x]

38 Niemelä S, BrunsteinKlomek A, Sillanmäki L, Helenius H, Piha J,

74WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Moore SE et al . Consequences of bullying victimization

Kumpulainen K, Moilanen I, Tamminen T, Almqvist F, Sourander
A. Childhood bullying behaviors at age eight and substance use
at age 18 among males. A nationwide prospective study. Addict
Behav 2011; 36: 256-260 [PMID: 21146319 DOI: 10.1016/
j.addbeh.2010.10.012]

39 Vieno A, Gini G, Santinello M. Different forms of bullying and
their association to smoking and drinking behavior in Italian
adolescents. J Sch Health 2011; 81: 393-399 [PMID: 21668879
DOI: 10.1111/j.1746-1561.2011.00607.x]

40 Nansel TR, Craig W, Overpeck MD, Saluja G, Ruan WJ. Cross-
national consistency in the relationship between bullying behaviors
and psychosocial adjustment. Arch Pediatr Adolesc Med 2004; 158:
730-736 [PMID: 15289243 DOI: 10.1001/archpedi.158.8.730]

41 Nakamoto J, Schwartz D. Is peer victimization associated with
academic achievement? A meta-analytic review. Soc Dev 2010; 19:
221-242 [DOI: 10.1111/j.1467-9507.2009.00539.x]

42 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting
items for systematic reviews and meta-analyses: the PRISMA
statement. Ann Intern Med 2009; 151: 264-269, W64 [PMID:
19622511 DOI: 10.1371/journal.pmed.1000097]

43 Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie
D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis of
observational studies in epidemiology: a proposal for reporting.
Meta-analysis Of Observational Studies in Epidemiology (MOOSE)
group. JAMA 2000; 283: 2008-2012 [PMID: 10789670 DOI:
10.1001/jama.283.15.2008]

44 Wells G, Shea B, O’connell D, Peterson J, Welch V, Losos M, Tugwell
P. The Newcastle-Ottawa Scale (NOS) for assessing the quality of
nonrandomised studies in meta-analyses. Ottawa: Ottawa Hospital
Research Institute, 2000. [Accessed 2016 Feb]. Available from: URL:
http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp

45 Norman RE, Byambaa M, De R, Butchart A, Scott J, Vos T. The
long-term health consequences of child physical abuse, emotional
abuse, and neglect: a systematic review and meta-analysis. PLoS
Med 2012; 9: e1001349 [PMID: 23209385 DOI: 10.1371/journal.
pmed.1001349]

46 Barendregt J, Doi SA. MetaXL version 2.1 [computer program].
Brisbane: EpiGear International, 2012

47 Doi SA, Barendregt JJ, Mozurkewich EL. Meta-analysis of
heterogeneous clinical trials: an empirical example. Contemp
Clin Trials 2011; 32: 288-298 [PMID: 21147265 DOI: 10.1016/
j.cct.2010.12.006]

48 Doi SA, Thalib L. A quality-effects model for meta-analysis.
Epidemiology 2008; 19: 94-100 [PMID: 18090860 DOI: 10.1097/
EDE.0b013e31815c24e7]

49 Hill AB. The environment and disease: association or causation?
Proc R Soc Med 1965; 58: 295-300 [PMID: 14283879]

50 World Cancer Research Fund. American Institute for Cancer
Research. Food, Nutrition, Physical Activity, and the Prevention of
Cancer: a Global Perspective. Washington, DC: AICR, 2007

51 Jansen DE, Veenstra R, Ormel J, Verhulst FC, Reijneveld SA.
Early risk factors for being a bully, victim, or bully/victim in
late elementary and early secondary education. The longitudinal
TRAILS study. BMC Public Health 2011; 11: 440 [PMID:
21645403 DOI: 10.1186/1471-2458-11-440]

52 Von Schoon I, Montgomery SM. The relationship between
early life experiences and adult depression. Z Psychosom Med
Psychoanal 1997; 43: 319-333

53 Brunstein Klomek A, Marrocco F, Kleinman M, Schonfeld IS,
Gould MS. Bullying, depression, and suicidality in adolescents.
J Am Acad Child Adolesc Psychiatry 2007; 46: 40-49 [PMID:
17195728 DOI: 10.1097/01.chi.0000242237.84925.18]

54 van der Wal MF, de Wit CA, Hirasing RA. Psychosocial health
among young victims and offenders of direct and indirect bullying.
Pediatrics 2003; 111: 1312-1317 [PMID: 12777546]

55 Due P, Holstein BE, Lynch J, Diderichsen F, Gabhain SN, Scheidt
P, Currie C. Bullying and symptoms among school-aged children:
international comparative cross sectional study in 28 countries. Eur J
Public Health 2005; 15: 128-132 [PMID: 15755782 DOI: 10.1093/
eurpub/cki105]

56 Due P, Hansen EH, Merlo J, Andersen A, Holstein BE. Is
victimization from bullying associated with medicine use among
adolescents? A nationally representative crosssectional survey in
Denmark. Pediatrics 2007; 120: 110-117 [PMID: 17606568 DOI:
10.1542/peds.2006-1481]

57 Björkqvist K. Social defeat as a stressor in humans. Physiol
Behav 2001; 73: 435-442 [PMID: 11438372 DOI: 10.1016/
s0031-9384(01)00490-5]

58 Watt MJ, Burke AR, Renner KJ, Forster GL. Adolescent male rats
exposed to social defeat exhibit altered anxiety behavior and limbic
monoamines as adults. Behav Neurosci 2009; 123: 564-576 [PMID:
19485563 DOI: 10.1037/a0015752]

59 Koolhaas J, Hermann P, Kemperman C, Bohu sB, Van Den
Hoofdakker R, Beersma D. Single social defeat in male rats
induces a gradual but long-lasting behavioral change: A model of
depression. Neurosci Res Commun 1990; 7: 35-41

60 Anda RF, Felitti VJ, Bremner JD, Walker JD, Whitfield C, Perry
BD, Dube SR, Giles WH. The enduring effects of abuse and related
adverse experiences in childhood. A convergence of evidence
from neurobiology and epidemiology. Eur Arch Psychiatry Clin
Neurosci 2006; 256: 174-186 [PMID: 16311898 DOI: 10.1007/
s00406-005-0624-4]

61 Shonkoff JP, Boyce WT, McEwen BS. Neuroscience, molecular
biology, and the childhood roots of health disparities: building
a new framework for health promotion and disease prevention.
JAMA 2009; 301: 2252-2259 [PMID: 19491187 DOI: 10.1001/
jama.2009.754]

62 Copeland WE, Wolke D, Lereya ST, Shanahan L, Worthman
C, Costello EJ. Childhood bullying involvement predicts low-
grade systemic inflammation into adulthood. Proc Natl Acad Sci
USA 2014; 111: 7570-7575 [PMID: 24821813 DOI: 10.1073/
pnas.1323641111]

63 Arseneault L, Cannon M, Fisher HL, Polanczyk G, Moffitt TE,
Caspi A. Childhood trauma and children’s emerging psychotic
symptoms: A genetically sensitive longitudinal cohort study. Am J
Psychiatry 2011; 168: 65-72 [PMID: 20952460 DOI: 10.1176/appi.
ajp.2010.10040567]

64 Arseneault L, Milne BJ, Taylor A, Adams F, Delgado K, Caspi
A, Moffitt TE. Being bullied as an environmentally mediated
contributing factor to children’s internalizing problems: a study
of twins discordant for victimization. Arch Pediatr Adolesc
Med 2008; 162: 145-150 [PMID: 18250239 DOI: 10.1001/
archpediatrics.2007.53]

65 Arseneault L, Walsh E, Trzesniewski K, Newcombe R, Caspi
A, Moffitt TE. Bullying victimization uniquely contributes to
adjustment problems in young children: a nationally representative
cohort study. Pediatrics 2006; 118: 130-138 [PMID: 16818558
DOI: 10.1542/peds.2005-2388]

66 Brent BK. Increased risk of psychotic symptoms at 12 years in
children who have been maltreated by adults or bullied by peers.
Evid Based Ment Health 2011; 14: 65 [PMID: 21764864 DOI:
10.1136/ebmh.14.3.65]

67 von Rueden U, Gosch A, Rajmil L, Bisegger C, RavensSieberer
U. Socioeconomic determinants of health related quality of life
in childhood and adolescence: results from a European study. J
Epidemiol Community Health 2006; 60: 130-135 [PMID: 16415261
DOI: 10.1136/jech.2005.039792]

68 Currie J. Healthy, wealthy, and wise: Socioeconomic status, poor
health in childhood, and human capital development. J Econ Lit
2009; 47: 87-122 [DOI: 10.2307/27647135]

69 Dubben HH, Beck-Bornholdt HP. Systematic review of publication
bias in studies on publication bias. BMJ 2005; 331: 433-434 [PMID:
15937056 DOI: 10.1136/bmj.38478.497164.F7]

70 Egger M, Smith GD. Misleading meta-analysis. BMJ 1995; 310:
752-754 [PMID: 7711568 DOI: 10.1136/bmj.310.6982.752]

71 Jepsen P, Johnsen SP, Gillman MW, Sørensen HT. Interpretation of
observational studies. Heart 2004; 90: 956-960 [PMID: 15253985
DOI: 10.1136/hrt.2003.017269]

72 Fekkes M, Pijpers FI, Verloove-Vanhorick SP. Bullying behavior
and associations with psychosomatic complaints and depression

75WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Moore SE et al . Consequences of bullying victimization

in victims. J Pediatr 2004; 144: 17-22 [PMID: 14722513 DOI:
10.1016/j.jpeds.2003.09.025]

73 McMahon EM, Reulbach U, Corcoran P, Keeley HS, Perry IJ,
Arensman E. Factors associated with deliberate self-harm among Irish
adolescents. Psychol Med 2010; 40: 1811-1819 [PMID: 20056025

DOI: 10.1017/s0033291709992145]
74 Scott JG, Moore SE, Sly PD, Norman RE. Bullying in children and

adolescents: a modifiable risk factor for mental illness. Aust N Z J
Psychiatry 2014; 48: 209-212 [PMID: 24317152 DOI: 10.1177/000
4867413508456]

P- Reviewer: Alavi N, Classen CF S- Editor: Ji FF L- Editor: A
E- Editor: Wu HL

76WJP|www.wjgnet.com March 22, 2017|Volume 7|Issue 1|

Moore SE et al . Consequences of bullying victimization

Published by Baishideng Publishing Group Inc
8226 Regency Drive, Pleasanton, CA 94588, USA

Telephone: +1-925-223-8242
Fax: +1-925-223-8243

E-mail: bpgoffice@wjgnet.com
Help Desk: http://www.wjgnet.com/esps/helpdesk.aspx

http://www.wjgnet.com

WJPv7i1-Back Cover

Since January 2020 Elsevier has created a COVID-19 resource centre with

free information in English and Mandarin on the novel coronavirus COVID-

19. The COVID-19 resource centre is hosted on Elsevier Connect, the

company’s public news and information website.

Elsevier hereby grants permission to make all its COVID-19-related

research that is available on the COVID-19 resource centre – including this

research content – immediately available in PubMed Central and other

publicly funded repositories, such as the WHO COVID database with rights

for unrestricted research re-use and analyses in any form or by any means

with acknowledgement of the original source. These permissions are

granted for free by Elsevier for as long as the COVID-19 resource centre

remains active.

Journal Pre-proof

“Hydroxychloroquine in patients with COVID-19: A Systematic Review and meta-
analysis.”

Awadhesh Kumar Singh, Akriti Singh, Ritu Singh, Anoop Misra

PII: S1871-4021(20)30136-3

DOI: https://doi.org/10.1016/j.dsx.2020.05.017

Reference: DSX 1672

To appear in: Diabetes & Metabolic Syndrome: Clinical Research & Reviews

Received Date: 6 May 2020

Revised Date: 7 May 2020

Accepted Date: 7 May 2020

Please cite this article as: Singh AK, Singh A, Singh R, Misra A, “Hydroxychloroquine in patients
with COVID-19: A Systematic Review and meta-analysis.”, Diabetes & Metabolic Syndrome: Clinical
Research & Reviews (2020), doi: https://doi.org/10.1016/j.dsx.2020.05.017.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

https://doi.org/10.1016/j.dsx.2020.05.017

“Hydroxychloroquine in patients with COVID-19: A Sy stematic Review and

Meta-analysis.”

Types of article: Research article

Authors name in order: Awadhesh Kumar Singh, Akriti Singh, Ritu Singh,

Anoop Misra

Authors affiliation:

Awadhesh Kumar Singh: M.D; D.M, Senior Consultant, Diabetes &

Endocrinology, G.D Hospital &

Diabetes Institute, Kolkata, West Bengal, India.

Akriti Singh: MBBS, Medical Resident, College of Medicine and JNM Hospital,

Kalyani, Nadia, West Bengal, India.

Ritu Singh: M.D, Senior Consultant, Gynaecology & Obstetrics, G.D Hospital &

Diabetes Institute, Kolkata, West Bengal, India.

Anoop Misra: M.D, Director, Fortis C-DOC Hospital for Diabetes and Allied

Sciences, New Delhi, India.

Highlights:

• The role of hydroxychloroquine in the treatment of COVID-19 is not fully

known.

• While initial studies with hydroxychloroquine showed some ray of hope,

recent studies that have emerged found either no benefit or a possible harm

in COVID-19.

• This meta-analysis showed no benefit on viral clearance, although a

significant increase in death was observed with hydroxychloroquine in

patients with COVID-19, compared to the control.

Abstract:

Backgrounds and Aims

The role of hydroxychloroquine (HCQ) in the treatment of COVID-19 is not fully

known. We studied the efficacy of HCQ compared to the control in COVID-19

subjects on a. viral clearance measured by reverse transcriptase polymerase chain

reaction (RT-PCR) and, b. death due to all cause.

Methods

PubMed, Scopus, Cochrane and MedRxiv database were searched using the

specific keywords up to April 30, 2020. Studies that met our objectives were

assessed for the risk of bias applying various tools as indicated. Three studies each

that reported the outcome of viral clearance by RT-PCR and death due to all cause,

were meta-analyzed by applying inverse variance-weighted averages of

logarithmic risk ratio (RR) using a random effects model. Heterogeneity and

publication bias were assessed using the I2 statistic and funnel plots, respectively.

Results

Meta-analysis of 3 studies (n=210) on viral clearance assessed by RT-PCR showed

no benefit (RR, 1.05; 95% CI, 0.79 to 1.38; p=0.74), although with a moderate

heterogeneity (I2=61.7%, p=0.07). While meta-analysis of 3 studies (n=474)

showed a significant increase in death with HCQ, compared to the control (RR,

2.17; 95% 1.32 to 3.57; p=0.002), without any heterogeneity (I2=0.0%, p=0.43).

Conclusions:

No benefit on viral clearance but a significant increase in mortality was observed

with HCQ compared to control in patients with COVID-19.

Keywords: Hydroxychloroquine, COVID-19, viral clearance, outcomes, death

1. Introduction:

Scientist and physicians are working at heightened pace to research the treatment

of coronavirus infection (COVID-19). Several potential candidate drugs have been

tried in COVID-19. From these list of candidate drugs, two anti-malarial drugs

came into limelight for following reasons. Initial studies found both chloroquine

(CQ) and its derivative hydroxychloroquine (HCQ) inhibits SARS-CoV-2

effectively in vitro [1-3]. This led clinicians to believe that both drugs may have

good potential in the treatment of COVID-19.

First report of human trial came from China. A commentary by Gao et al [4]

referring to 15 Chinese trials (whose complete results are still not available),

claimed benefit with CQ in inhibiting the exacerbation of pneumonia, improving

lung imaging findings, promoting a virus-negative conversion, and shortening the

disease in more than 100 patients. One study from these 15 Chinese trials,

conducted by Chen et al [5] later showed data of 62 patients and found that HCQ

significantly improved the clinical recovery (fever and cough) and pneumonia

assessed by chest CT scan, compared to the control. However, a close look into

this randomized control trial (RCT) found that the endpoints specified in the

published protocol differed from those reported. First, the trial was originally

supposed to report the results from two different dosage of HCQ on clinical and

radiological outcome, although only the report of higher dose HCQ was reported

finally. Second, the trial was stopped prematurely [6]. Another study from France,

a non-randomized trial of HCQ (n=36) by Gautret et al [7] also reported a

significant effect of HCQ and HCQ plus azithromycin (AZ) in lowering viral load

and viral clearance compared to control as measured by reverse-transcriptase

polymerase chain reaction (RT-PCR). However, this study was widely criticized

due to the poor trial design, unreliable conclusions, no clinical endpoints,

assessments made on day 6 despite a planned 10 days trial, different value of Cycle

threshold for RT-PCR, and derivation of results after excluding six patients from

the HCQ arm [8]. The publishing journal’s society also subsequently declared that

the trial by Gautret and Colleagues did “not meet the Society’s expected standard”

[9].

Nevertheless, based on these limited observational and anecdotal evidence, several

guidelines across the world allowed both these drugs in the treatment of COVID-

19 [10]. Interestingly, Indian Council of Medical research hurriedly issued a

guideline and additionally recommended the use of CQ and HCQ as a prophylactic

agent in the close contacts including the health care workers [11]. Surprisingly,

based on these emerging developments, US President while addressing the nation

on pandemic claimed CQ and HCQ as a “game changer” in the treatment of

COVID-19. The consequence of this announcement resulted in FDA issuing an

Emergency Use Authorization (EUA) to use both the drugs in the treatment of

COVID-19 on March 28, 2020. Historically, this new EUA represents the second

time when FDA has ever used any emergency authority to permit use of a

medication for an unapproved indication. Earlier, an investigational neuraminidase

inhibitor, peramivir was given similar EUA by FDA during the 2009-2010 for

severely ill patients with H1N1 influenza. Although later an RCT failed to show

any benefit of peramivir in severely ill hospitalized patients with influenza,

compared to the placebo. Nonetheless, peramivir is approved only for

uncomplicated influenza since 2014.

Since several newer studies of HCQ on COVID-19 have recently become

available, we aimed to study its effect on COVID-19 on two important objective

outcomes. These two important outcomes include – a. viral clearance by RT-PCR

negativity and, b. death due to all cause. In addition, we have also compiled the

results from all the studies that have studied the efficacy and safety of HCQ in

COVID-19, including non-controlled trials.

2. Methods:

This study was conducted in accordance with the Preferred Reporting Items for

Systematic Reviews and Meta-Analyses (PRISMA) [12]. However, this study has

not been registered in the International Prospective Register of Systematic Reviews

(PROSPERO).

2.1 Search strategy and Inclusion criteria:

Three authors (AKS, AS and RS) systematically searched the PubMed, Scopus,

Cochrane library and MedRxiv data base up to April 30, 2020. The key terms

searched were ‘‘Hydroxychloroquine’’ OR ‘‘HCQ’’ (All Fields) OR “viral

clearance” OR “death” OR “clinical recovery” AND COVID-19 OR SARS-CoV-

2. We retrieved all the studies conducted with hydroxychloroquine in patients with

COVID-19 that was compared to control and explicitly reported at least one

outcome of interest which include viral clearance by transcriptase polymerase

chain reaction (RT-PCR) and or death due to all cause.

We excluded case reports, preclinical studies, studies that did not report outcomes

with HCQ in COVID-19, and studies that did not compare the outcomes with HCQ

compare to placebo or control. The studies that met our predefined inclusion

criteria were screened by three authors (AKS, RS and AS), and the studies that

entirely fulfilled our inclusion criteria were retrieved with their supplementary

appendix for further review. Any ambiguity during study selection was resolved

by mutual discussion and consensus. One study whose full text was available in

Chinese (abstract in English) was translated to English by Google translator and

one study was retrieved through hand search. A detailed PRISMA flow-diagram

for the search strategy is included in figure 1.

2.2 Assessment of bias and Statistical Analysis:

Four reviewers (AKS, AS, RS and AM) independently assessed the studies for risk

of bias ascertained through Jadad checklist, ROBINS-I tool and Newcastle-Ottawa

scale for randomized, non-randomized and observational studies, wherever

appropriate [13-15] and any disagreements were resolved through mutual

discussion and consensus. Scoring of these studies on risk of bias tools have been

outlined in supplementary table 1. A detailed PRISMA checklist has been

appended in supplementary table 2.

Comprehensive meta-analysis (CMA) software Version 3, Biostat Inc. Englewood,

NJ, USA was used to calculate all the statistical analyses. Seven studies were

retrieved that reported any outcome with HCQ compared to the control in COVID-

19. Three studies each reported for viral clearance measure by RT-PCR and the

outcome of death due to any cause. We meta-analyzed the pooled data of primary

outcomes of 3 trials that reported the rate of PCR negativity, and 3 trials that

reported the difference in mortality between HCQ and control arm. Since one RCT

by Chen et al reported resorption of pneumonia on chest computed tomography

(CT) as a primary outcome but neither reported RT-PCR negativity, nor the

mortality outcome, thus we did not include this study in the meta-analysis,

however the outcome of this study shall be discussed.

Estimates from all the eligible studies have been combined by applying inverse

variance-weighted averages of logarithmic risk ratio (RR), using random-effects

analysis. Heterogeneity was measured using Higgins I² and Cochrane Q statistic

[16]. Heterogeneity was considered as low (I2 <25%) or moderate (25-50%) or high

(>50%). All the p reported here are two-sided and a p value of < 0.05 is considered

to be statistically significant. We also evaluated the potential publication bias by

applying funnel plots using the “trim and fill” adjustment, rank correlation test and

the Egger’s test.

3. Results:

The overview of results including the risk of bias from all the 7 studies that

compared HCQ to the control in COVID-19 have been summarized in table 1 [5, 7,

17-21]. The meta-data that was used in this metanalysis has been also represented

in table 2. Table 3 summarizes the safety and efficacy of all the 10 trials conducted

with HCQ in COVID-19, to date [5, 7, 17-24]. One RCT by Chen et al [5] that is

not included in this meta-analysis found “any improvement” in pneumonia were

significantly higher in HCQ arm, compared to the control (80.6 vs. 54.8%,

p=0.048). Moreover, significant improvement in chest CT (more than 50%

absorption of pneumonia) was increasingly observed in HCQ arm, compared to the

control (61.3 vs. 16.1%, p=not reported).

Nevertheless, the meta-analysis of 3 studies (n=210) that reported the rate of PCR

negativity (figure 2) found no benefit with HCQ, compared to the control (RR,

1.05; 95% CI, 0.79 to 1.38; p=0.74), although with a moderate heterogeneity

(I2=61.7%, p=0.07). After the adjustment of publication bias, the Trim and Fill

imputed the RR of 0.99 with 95% CI 0.69 to 1.42 (supplementary figure SF1).

However, the meta-analysis of 3 trials (n=474) that reported the mortality outcome,

showed a significant (2-fold) increase in death in HCQ arm (figure 3), compared to

the control (RR, 2.17; 95% 1.32 to 3.57; p=0.002), without any heterogeneity

(I2=0.0%, p=0.43) and publication bias (supplementary figure SF2).

4. Discussion:

To our knowledge, this would be the most updated meta-analysis to report the

effect of HCQ on viral clearance and mortality outcome, compared to the placebo

that included 6 studies. Additionally, we have also analyzed the results from all the

10 studies available that have studied the efficacy and safety of HCQ in patients

with COVID-19 (table 3).

A recent meta-analysis published by Sarma et al [25] have showed no difference in

viral clearance and composite of death or clinical worsening with HCQ, while a

significant improvement in radiological progression was observed, compared to the

control. However, the meta-analysis by Sarma et al seems to have overlooked the

raw data and mistakenly included the wrong denominators. For example – they

included number of patients for HCQ plus azithromycin (n=20) in their analysis,

rather than HCQ alone (n=14), for the denominator for viral clearance. Similarly,

the number of patients included for the composite of death or clinical worsening in

HCQ arm was also overlooked and mistakenly reported in denominator (n=20),

rather than the actual number (n=26). We believe that these differences could have

changed the outcomes.

We do acknowledge a number of limitations in our analysis that include lesser

number of patients overall, lack of individual patient data, combining the results of

RCT with other non-randomized studies and the inclusion of pre-print version of

some of the unpublished studies. Moreover, outcomes are not adjusted for multiple

confounding factors and no sensitivity analysis were made. Besides, this

metanalysis was not registered at PROSPERO.

While this meta-analysis found no benefit of HCQ in the treatment of COVID-19

on viral clearance and there was a 2-fold increase in death compared to the control

arm, this could have been skewed by the one larger study that have shown a

significant harm with HCQ, even when other smaller studies found no significant

difference. For example, the study by Magagnoli et al (n=368) [21] found that

there was no difference in the requirement of mechanical ventilator (MV) and

death in patients who were on MV. However, the risk of death from any cause was

higher in the HCQ group (adjusted hazard ratio 2.61, 1.10-6.17, p=0.03), compared

to the control. Since this study contributed more than 84% of weight in this pooled

meta-analysis of 3 studies, the signal of significant death appears to emerge.

Moreover, relatively elderly patients (mean age 68 year) and more sick (moderate

to severe COVID-19) patients were studied in Magagnoli et al study, compared to

all other studies. Therefore, the purported benefit of HCQ in early or mild

COVID-19 as observed in studies by Chen et al [5] and Gautret et al [7] cannot be

entirely ruled out, from the result of this meta-analysis. It is also possible that

HCQ may have some benefit in early and mild COVID-19 but possibly harmful in

moderate to severe COVID-19.

Nevertheless, none of these studies attributed the harm of HCQ directly linked to

the cardiac side effect. However, a recent double-blind RCT, Cloro-Covid-19

conducted by Borba et al [26] hinted of high lethality with the higher dosage of

chloroquine (CQ). Higher dose of CQ was associated with 39% death, compared to

15% death in lower dose arm. Fatality rate with high dose of CQ was as high as

60% in patients with underlying heart disease. QTc prolongation was significant in

19% of cases on high dose CQ compared to 11% in low dose CQ arm. Although no

signals of torsade de pointes were noted in this trial, it is believed that increase in

mortality in this trial could be attributed to the combination of CQ with

azithromycin (AZ) and oseltamivir or lopinavir/ritonavir, all of which can prolong

QTc interval [27]. Similarly, emerging studies from France and USA have

increasingly cautioned for QTc prolongation with both HCQ and HCQ plus AZ.

While Bessière et al [28] reported (n=40) a prolonged QTc in 93% of the patients

receiving either HCQ or HCQ plus AZ; Mercuro et al [29] reported QTc

prolongation (n=90) in 20% of patients treated with HCQ alone or HCQ plus AZ.

These findings underscores the safety of HCQ in the light of negligible benefit

observed in some of these studies.

Despite several limitations of this meta-analysis, we feel this finding would instill

some degree of skepticism and shall help in curbing the exuberant use of over

enthusiastically claimed “magical” drug. Hopefully, large randomized controlled

trial such as DISCOVERY (EudraCT 2020-000936-23) and RECOVERY (UK),

that is currently studying the effect of HCQ in COVID-19 and comparing it with

other anti-viral drugs will finally decide its fate. Meanwhile, we believe that any

prudent clinician would follow a pragmatic approach and shall apply these drugs

only after assessing the potential risk versus uncertain benefit.

5. Conclusions:

While no benefit on viral clearance demonstrated by HCQ compared to the control

in patients with COVID-19, a significant 2-fold increase in mortality with the HCQ

warrants its use if at all, with an extreme caution, until the results from larger

randomized controlled trials are available.

Funding: Not funded

Conflict of interest: Nothing to declare

Ethical permission: Not required as this analysis do not involve patients directly.

References:

1. Wang M, Cao R, Zhang L, Yang X, Liu J, Xu M, et al. Remdesivir and

chloroquine effectively inhibit the recently emerged novel coronavirus (2019-

nCoV) in vitro. Cell Res 2020;30:269-271.

2. Yao X, Ye F, Zhang M, Cui C, Huang B, Niu P, et al. In Vitro Antiviral Activity

and Projection of Optimized Dosing Design of Hydroxychloroquine for the

Treatment of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2).

Clin Infect Dis 2020;5801998:pii:ciaa237.

3. Liu J, Cao R, Xu M, Wang X, Zhang H, Hu H, et al. Hydroxychloroquine, a less

toxic derivative of chloroquine, is effective in inhibiting SARS-CoV-2 infection in

vitro. Cell Discov 2020;6:16.

4. Gao J, Tian Z, Yang X. Breakthrough: chloroquine phosphate has shown

apparent efficacy in treatment of COVID-19 associated pneumonia in clinical

studies. Biosci Trends 2020;14:72-3. 10.5582/bst.2020.01047 32074550

5. Chen Z, Hu J, Zhang Z, et al. Efficacy of hydroxychloroquine in patients with

COVID-19: results of a randomized clinical trial. Version 2. medRxiv

2020.03.22.20040758. [Preprint.] 10.1101/2020.03.22.20040758

6. Yan D, Zhang Z. Therapeutic effect of hydroxychloroquine on novel

coronavirus pneumonia (COVID-19). Chinese Clinical Trials Registry.

http://www.chictr.org.cn/showproj.aspx?proj=48880

7. Gautret P, Lagier JC, Parola P, etal . Hydroxychloroquine and azithromycin as a

treatment of COVID-19: results of an open-label non-randomized clinical trial. Int

J Antimicrob Agents 2020:105949. 10.1016/j.ijantimicag.2020.105949 32205204

8. Lenzer J. Covid-19: US gives emergency approval to hydroxychloroquine

despite lack of evidence. BMJ 2020;369:m1335. 10.1136/bmj.m1335 32238355

9. International Society of Antimicrobial Chemotherapy. Statement on IJAA paper.

Accessed April 23, 2020. https://www.isac.world/news-and-publications/official-

isac-statement

10. Singh AK, Singh A, Saikh A, Singh R, Misra A. Chloroquine and

hydroxychloroquine in the treatment of COVID-19 with or without diabetes: A

systematic search and a narrative review with a special reference to India and other

developing countries. Diabetes Metab Syndr. 2020 May-June; 14(3): 241–246.

11. Indian Council for Medical Research. Recommendation for empiric use of

hydroxychloroquine for prophylaxis of SARS-CoV-2 infection.

https://icmr.nic.in/sites/default/files/upload_documents/HCQ_Recommendation_2

2March_final_MM_V2 . Accessed 3 April

2020.

12. Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, et al.

Preferred reporting items for systematic review and meta-analysis protocols

(PRISMA-P) 2015: elaboration and explanation. BMJ 2015; 350: g7647

13. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of

randomized clinical trials: is blinding necessary? Control Clin Trials 1996;17:1–

12.

14. Sterne JAC, Hernan MA, Reeves BC, Savoic J, Berkman ND, Viswanathan M.

et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of

interventions. BMJ 2016;355:i4919

15. Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment

of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 2010;

25(9): 603-5.

16. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in

meta-analyses. BMJ 2003; 327: 557–60.

17. Jun C, Danping L, Li L, et al. A pilot study of hydroxychloroquine in treatment

of patients with common coronavirus disease-19 (COVID-19). J Zhejiang Univ

(Med Sci) 2020. doi: 10.3785/j.issn.1008-9292.2020.03.03

18. Tang W, Cao Z, Han M, Wang Z, Chen J, Sun W, et al. Hydroxychloroquine in

patients with COVID-19: an open-label, randomized, controlled trial. doi:

https://doi.org/10.1101/2020.04.10.20060558

19. Barbosa J, Kaitis D, Ryan F, Kim L, Xihui L. Clinical Outcomes of

Hydroxychloroquine in Hospitalized Patients with COVID-19: A Quasi-

Randomized Comparative Study. Bibliovid 2020. https://bibliovid.org/clinical-

outcomes-of-hydroxychloroquine-in-hospitalized-patients-with-covid-19-a-302

20. Mahevas M, Tran V-T, Roumier M, et al. No evidence of clinical efficacy of

hydroxychloroquine in patients hospitalized for COVID-19 infection with oxygen

requirement: results of a study using routinely collected data to emulate a target

trial. medRxiv 2020:2020.04.10.20060699.

21. Magagnoli J, Narendran S, Pereira F, Cummings T, Hardin HW, Sutton SS, et al.

Outcomes of hydroxychloroquine usage in United States veterans hospitalized with

Covid-19. doi: https://doi.org/10.1101/2020.04.16.20065920

22. Molina JM, Delaugeree C, Goff JL, et al. No evidence of rapid antiviral

clearance or clinical benefit with the combination of hydroxychloroquine and

azithromycin in patients with severe COVID-19 infection. Med Mal Infect (2020),

https://doi.org/10.1016/j.medmal.2020.03.006

23. Gautret P, Lagier JC, Parola P, et al. Clinical and microbiological effect of a

combination of hydroxychloroquine and azithromycin in 80 COVID-19 patients

with at least a six-day follow up: an observational study. IHU Méditerranée

Infection. 2020;27(1).

24. Million M, Lagier JC, Gautret O, et al. Early treatment of 1061 COVID-19

patients with hydroxychloroquine and azithromycin, Marseille, France. Bibliovid.

2020.

25. Sarma P, Kaur H, Kumar H, et al. Virological and clinical cure in COVID�19

patients treated with hydroxychloroquine: A systematic review and meta�analysis.

Journal of Medical Virology. Published on April 16, 2020.

https://doi.org/10.1002/jmv.25898

26. Borba MGS, Val FFA, Sampaio VS, et al; CloroCovid-19 Team. Effect of high

vs low doses of chloroquine diphosphate as adjunctive therapy for patients

hospitalized with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)

infection: a randomized clinical trial. JAMA Open. 2020;3(4):e208857.

doi:10.1001/jamanetworkopen.2020.8857

27. Fihn SD, Perencevich E, Bradley SM. Caution Needed on the Use of

Chloroquine and Hydroxychloroquine for Coronavirus Disease 2019. JAMA

Network Open. 2020;3(4.23):e209035. doi:10.1001/jamanetworkopen.2020.9035

28. Bessière F, Roccia H, Delinière A, et al. Assessment of QT Intervals in a case

series of patients with coronavirus disease 2019 (COVID-19) infection treated with

hydroxychloroquine alone or in combination with azithromycin in an intensive

care unit. JAMA Cardiol. Published online May 1, 2020.

doi:10.1001/jamacardio.2020.1787

29. Mercuro NJ, Yen CF, Shim DJ, et al. Risk of QT interval prolongation

associated with use of hydroxychloroquine with or without concomitant

azithromycin among hospitalized patients testing positive for coronavirus 2019

(COVID-19) infection. JAMA Cardiol. Published online May 1, 2020. doi:10.

1001/jamacardio.2020.1834

30.

Table 1: Studies of HCQ compared to placebo in patients with COVID-19

Study Types
of
studies

Country Age
(mean,
years)

N Case Control Severity of
COVID-19

HCQ
dose/day X
Days

Primary
outcome

Secondary
outcome

Improvement
in Primary
outcome

Improvement
in Secondary
outcome

Chen5
et al.*
(ChiCTR
2000029559)

RCT China 44.7 62 31 31 Mild/
moderate

400 mg/d X
5D

Time to
clinical
recovery and
improvement
of pneumonia
in chest CT

Yes NR

Jun17
et al.*
(NCT04261517)

RCT China NR 30 15 15 Mild/
moderate

400 mg/d X
5D

Viral load by
RT-PCR + vs.
– at day 7

NR No NR

Tang18
et al.*
(ChiCTR
2000029868)

RCT China 46 150 75 75 Mild/
Moderate
(84%)

1200 mg/d X
3D, followed
by 800 mg/d
X 2 wks.
(mild
/moderate
cases) or 3
wks. (severe
cases)

Viral load by
RT-PCR + vs.
– at day 28

Clinical
symptoms,
normalization
of laboratory
parameters and
chest radiology

No No. However,
reduction in
CRP and
symptoms
noted in HCQ
arm in post-
hoc analysis

Gautret7
et al.**

nRCT France 45.1 36 20# 16 Mild/
moderate

600 mg/d X
10D

Viral load by
RT-PCR + vs.
– at day 6

Improvement in
symptoms,
mortality

Yes NR

Barbosa19
et al.**

qRCT USA 62.7 63 32 31 Mild/
moderate

800 mg/d X
1-2D
followed by
200-400 mg
OD X 3-4D

Need to
escalate
respiratory
support and
rate of
intubation at
day 5

Change in
lymphocyte
count, NLR,
and mortality

No, rather
harm in
HCQ arm

No, direction
towards harm

Mahevas20 et
al.***

Retro France 60 181 84 97 Pneumonia
requiring
O2 Rx

600 mg/d X
7D

ICU transfer or
death from any
cause at day 7

All-cause
mortality at day
7,
Occurrence of
ARDS within 7
day

No No

Magagnoli21 Retro USA 68 368 210## 158 Mild/ NR Need for MV Death in No benefit. No

et al.*** moderate and death from
any cause

patients on MV Risk of death
due to any
cause was
higher in
HCQ arm

Molina22
et al

POS France 58.7 11 11 0 Fever and
O2 Rx
(severe)

600 mg/d X
10D + AZ
500mg on
day 1 and
250 mg 2-5
days

Viral load by
RT-PCR + vs.
– at day 5-6

NR No NR

Gautret23
et al

POS France 52.1 80 80 0 Mild
(92%)
/moderate

600 mg/d X
10D + AZ
500 mg on
day 1 and
250 mg/d X
4D

Need for O2
therapy or ICU
admission

Viral load,
length of
hospital stays

Yes Yes

Million 24
et al

POS France 43.6 1061 1061 0 Mild
(95%)
/moderate

600 mg/d X
10D + AZ
500 mg on
day 1 then
250 mg/d X
4D

death, negative
RT-PCR

NR Yes NR

* Quality assessed as 5/8 on Jadad checklist, ** Moderate quality on ROBINS I tool, *** Quality assessed as 7/8 on Newcastle-Ottawa Scale, #6 patients received HCQ plus AZ,
##113 received HCQ plus AZ, HCQ- hydroxychloroquine, AZ- azithromycin, RCT – randomized controlled trial, nRCT- Non-randomized controlled trial, qRCT- quasi-
randomized controlled trial, RT-PCR- reverse transcriptase polymerase chain reaction, ICU- intensive care unit, ARDS- acute respiratory distress syndrome, MV- mechanical
ventilators, NR- not reported, CT- computed tomography, D- days, d- daily, O2- oxygen, Rx- treatment, POS- prospective observational studies

Table 2: Meta-data for analysis and results

Study N Types of outcome
assessed

Outcome
assessed for, N

Events in HCQ
arm, n

Total case
on HCQ
arm, N

Events in
control arm,
n

Total control
arm, N

Relative risk, 95% CI, p
value

Chen et al 62 Absorption pf
pneumonia

62 25 31 17 31 1.47, 1.02-2.11, p=0.037

Jun et al 30 RT-PCR negativity 30 13 15 14 15 0.93, 0.73-1.18, p=0.55
Gautret et al 36 RT-PCR negativity 30# 8 14 2 16 4.57, 1.16-18.05, p=0.03
Tang et al 150 RT-PCR negativity 150 59 70 65 80 1.04, 0.90-1.20, p=0.622
Barbosa et al 63 Death 38* 2 17 1 21 2.47, 0.24-24.98, p=0.44
Mahevas et al 181 Death 181 3 84 4 97 0.87, 0.20-3.76, p=0.85
Magagnoli et al 368 Death 255## 27 97 18 158 2.44, 1.42-4.19, p=0.001
#6 patients on HCQ plus AZ not analyzed, ## 131 patients on HCQ plus AZ not analyzed, * 38 patients matched control analyzed, HCQ- hydroxychloroquine, AZ-
azithromycin, CI- confidence interval

Table 3: Descriptive results, adverse events and limitation of all the trials done with Hydroxychloroquine as on April 30, 2020.

Study Details of primary and secondary outcome Result of primary and secondary outcome Adverse events
noted

Limitations of
the study

Chen5 et al i. Clinical recovery is defined as the return of
body temperature (36.6 °C on the surface, ≤
37.2 °C under the armpit and mouth or ≤ 37.8
°C in the rectum and tympanic Membrane) and
cough relief, (slight or no cough) checked 3
times daily that maintained for more than 72 h.

ii. Pulmonary recovery is defined at three levels
as exacerbated, unchanged, and improved
(moderately improved when less than 50 % of
pneumonia were absorbed, and significantly
improved, when more than 50 % pneumonia
were absorbed in chest CT)

i. Recovery time from fever significantly
shortened in the HCQ arm compared to
control (2.2 vs. 3.2 days, p=0.0008). Cough
remission time was significantly reduced in
the HCQ arm compared to control (2.0 vs. 3.1
days, p=0.0016)

ii. Improvement in pneumonia were
significantly higher in HCQ arm compared to
control (80.6 vs 54.8%, p=0.048). Significant
improvement in chest CT were increasingly
observed in HCQ arm compared to control
(61.3 vs. 16.1%, p=nr).

i. Mild adverse
reactions noted in 2
patients from HCQ
arm. one developed a
rash, and one had
headache.

ii. Four of 62 patients
progressed to severe
COVID-19, all from
control arm and none
from HCQ arm.

Protocol violation
from original
plan. Not reported
the results from
lower dose HCQ
and premature
stoppage of the
trial. Detail use of
other antivirals in
control group is
not available.

Jun17 et al Primary endpoint was negative RT-PCR of
naso-pharyngeal for COVID-19 on days 7 after
randomization

i. RT-PCR negativity at day 7 in throat swabs
in HCQ arm versus control were similar (86.7
vs. 93.3% respectively, p>0.05).

ii. Median duration from hospitalization to
PCR negative were similar in HCQ arm and
placebo (4 vs. 2 days respectively, p>0.05)].
The median time for fever
normalization was similar (1 days) in both
arms.

iii. Radiological progression in CT chest was
noted less in HCQ group compared to control
(33.3 vs. 46.7% respectively, p=nr). .

Transient diarrhea
and abnormal liver
function were seen in
26.7% cases in HCQ
arm compared to
20% in controls
(p>0.05)

Manuscript
available in
Chinese language.

Tang18 et al i. The primary endpoint was PCR negativity for
COVID-19 at day 28.

ii. Secondary endpoints includes the
improvement of clinical symptoms such as

i. No difference in PCR negative conversion
rate between HCQ and control arm at day 28
(85.4 vs. 81.3%, p=0.341). The negative
conversion time in HCQ arm and control were
same (median 8 vs. 7 days; HR 0.846; 0.580-

Significantly higher
adverse events noted
in 30% of HCQ arm
compared to 8.8% of
control (p=0.001).

Selecting the virus
negative
conversion as the
primary
end-point might

fever (axillary temperature of ≤36.60C),
normalization of SpO2 (>94% on room air),
disappearance of respiratory symptoms (nasal
congestion, cough, sore throat, sputum
production and shortness of breath),
normalization of CRP, ESR, IL-6, TNF-α level
and lymphocyte count within 28-days. In
addition, PCR negativity at day 4, 7, 10, 14 or
21.

1.234; p=0.341).

ii. No difference in symptoms between two
arms within 28-days. No difference in PCR
negativity between two arms at
day 4, 7, 10, 14 or 21.

iii. A significantly greater reduction of CRP
observed in HCQ arm compared to control
(6.986 vs. 2.723 mg/l, p=0.045). A trend in
more rapid recovery of lymphopenia also
observed in HCQ arm compared to control.

iv. Post-hoc analysis (confounding effects of
anti-viral agents removed), found a significant
improvement in symptoms in HCQ arm
compared to control (HR 8.83, 1.09-71.3).

The most common
adverse event was
diarrhea in HCQ arm
compared to control
(10 vs. 0%, p=0.004).
Blurred vision seen in
1 patient on HCQ.

not be the most
appropriate
outcome. Issues to
ensure the fidelity
to the protocol by
investigators.

Gautret7 et
al

i. Primary endpoint was negative RT-PCR for
COVID-19 at day-6.

ii. Secondary outcomes include virological
clearance overtime, improvement in symptoms
(temperature, respiratory rate, length of stay at
hospital), mortality, and occurrence of side
effects.

i. Negative RT-PCR for COVID-19 was
significantly higher in HCQ arm (70 vs.
12.5% p= 0.001) compared to control at day 6.
Combination arm of HCQ plus AZ had
significantly higher PCR negativity compared
to HCQ alone and control (100 vs. 57.1 vs.
12.5%, p<0.001) at day 6. ii. No other details available for secondary outcome

One patient died in
HCQ arm on day 3
despite negative RT-
PCR. One patient
stopped HCQ due to
GI side effect

Poor trial design,
assessments made
on day 6 despite a
planned 10 days
trial, different
value of Cycle
threshold for RT-
PCR, and
derivation of
results after
excluding six
patients from the
HCQ arm

Barbosa19
et al

i. Primary outcome – mortality, effect on
escalation of respiratory support,

ii. Secondary outcome – hematology benefits
(absolute lymphocyte count and NLR)

i. Significantly higher respiratory
support needed at day 5 in HCQ arm
compared to control (p=0.013). HCQ
treatment were independent predictors of
escalation of respiratory support OR 7.18,
(1.50-34.51, p=0.014). In a matched subgroup
analysis (n=38) also shows escalated
respiratory support in HCQ arm compared to
control (p=0.041).

No torsade de pointes
noted

i. Baseline
requirement of O2
Rx or intubation
were significantly
higher in HCQ
arm compared to
control (p=0.012).

ii. Major errors in

ii. Increased trend towards worsening of NLR
in HCQ arm compared to control (p=0.051).

in Table 2. HCQ
arm showing 31
patients and
control arm 32
patients which is
just reverse to
table 1.

Mahevas20
et al

i. Primary outcome – composite of transfer to
the ICU and or death from any cause within 7
days.

ii. Secondary outcomes- all-cause mortality at
day 7 and the occurrence of ARDS within 7
days.

i. Transfer to the ICU or died within 7 days
were similar in HCQ arm compared to control
(20.2 vs 22.1%; RR 0.91, 0.47–1.80).
ii. Percentage of all-cause death at day 7 were
similar in HCQ arm compared to control (2.8
vs. 4.6%; RR, 0.61, 0.13-2.89).

iii. Percentage of patients who developed
ARDS within 7 days were similar in HCQ
arm and control (27.4 vs. 24.1%; RR 1.14,
0.65-2.00).

ECG changes were
noted in 9.5% of
cases in HCQ arm
that caused HCQ
discontinuation. ECG
changes includes
prolonged QTc, First-
degree AV block and
LBBB.

No random
assignment,
potential
unmeasured
confounders bias
and no propensity
match for some
important
prognostic
variables.

Magagnoli21
et al

i. Primary outcomes were death from any cause
and the need for mechanical ventilation

ii. Secondary outcome was death on those on
mechanical ventilator

i. Rates of death in the HCQ, HCQ+AZ, and
control arm were 27.8%, 22.1%, 11.4%,
respectively. Compared to control, the risk of
death from any cause was higher in the HCQ
group (adjusted HR 2.61, 1.10-6.17, p=0.03)
but not in the HCQ+AZ group (adjusted HR
1.14, 0.56-2.32, P=0.72).

ii. Rates of need of ventilation in HCQ,
HCQ+AZ, and control arm were 13.3%, 6.9%,
14.1%, respectively. The risk of ventilation
was similar in HCQ (adjusted HR 1.43, 0.53-
3.79, p=0.48), and HCQ+AZ arm (adjusted
HR 0.43, 0.16-1.12, p=0.09), compared to
control.

iii. Secondary outcome of death in patients
who required mechanical
ventilation was similar in HCQ (adjusted HR
4.08, 0.77-21.70, p=0.10), and HCQ+AZ arm
(adjusted HR 1.20, 0.25-5.77, p=0.82),
compared to the control.

Nothing reported Non-randomized,
retrospective,
selection bias,
residual
confounding, only
men, median age
>65 years and
majority were of
Black ethnicity.

Molina22

et al
Primary outcome was RT-PCR negativity at
day 5-6

RT-PCR was positive in 80% of cases (95%
CI 49–94) at days 5-6 after treatment.

One patient had
prolonged QTc on
HCQ+AZ and drug
was stopped

Significant
comorbidities
present and
majority of patient
had severe
COVID-19.

Gautret23

et al
i. Primary outcome was need for O2 therapy or
transfer to the ICU after at least three days of
treatment.

ii. Secondary outcome was PCR negativity and
length of stay in the ID ward

i. Majority of patients (81.3%) had favorable
outcome and were discharged. Only 15%
required
oxygen therapy.

ii. RT-PCR was negative in 83% of cases at
day 7, and 93% of cases at day 8.

iii. Mean time for discharge was 4.1 days with
a mean length of stay of 4.6 days.

Minor adverse events
reported with HCQ
including nausea,
vomiting and blurred
vision.

Results of six
patients from
previous trials by
Gautret et al were
included in this
study also.

Million 24
et al

Endpoints were death, negative RT-PCR i. Good clinical outcome and negative RT-
PCR were obtained in 91.7% within 10 days.
Prolonged viral carriage was observed in 4.4%
cases who had high viral load at diagnosis (p
< 0.01), however viral culture was negative at day 10. All except one had negative PCR at day 15. ii. Poor outcome was observed in 4.3% and more associated with older age (OR 1.11), severe cases (OR 10.05) and use of selective beta-blockers and ARBs (p<0.05).

No cardiac toxicity
was observed,
although no details of
assessment of cardiac
toxicity is available.

Biased associated
with all
observational
studies.
Moreover, same
groups of authors
may have biased
belief based on
positive results
from previous
trials.

RT-PCR – reverse-transcriptase-polymerase-chain-reaction, ARDS- acute respiratory syndrome, HCQ- hydroxychloroquine, AZ- azithromycin, CI-
confidence interval, ICU- intensive care unit, MV- mechanical ventilator, HR- hazard ratio, RR- relative risk, OR- odds ratio, nr- not reported, CT- computed
tomography, ESR- erythrocyte sedimentation rate, CRP- c-reactive protein, IL- interleukin, TNF- tumor necrosis factor, O2- oxygen therapy, ID- infectious
disease, ECG- electrocardiogram, AV- atrioventricular, LBBB- left bundle branch block

Conflict of Interest

We hereby declare that we have no conflict of interest related to this article

Awadhesh Kumar Singh

Akriti Singh

Ritu Singh

Anoop Misra

Turn in your highest-quality paper
Get a qualified writer to help you with

“ Solve 3 questions after reviewing attached articles ”

Get high-quality paper

NEW! AI matching with writer

Order an Essay Now & Get These Features For Free:

Turnitin Report

Formatting

Title Page

Citation

Outline

Place an Order