Anaemia and perinatal outcome in Port Moresby


Mola, G.; Permezel, M.; Amoa, A.B.; Klufio, C.A.

Australian and New Zealand Journal of Obstetrics and Gynaecology 39(1): 31-34

1999


In 1987, a computerized obstetric database was set up at the Port Moresby General Hospital. Between 1987 and 1992, 27,117 births took place. The mean haemoglobin value amongst the 83% of women in whom a haemoglobin value was tested was 10.0 +/- 1.7 g/dL. High stillbirth rates (94 per 1,000) were associated with a haemoglobin value < 6 g/dL. The stillbirth rate was slightly lower (14 per 1,000) in woman whose lowest haemoglobin value was in the range 10.0-10.9 g/dL than in those with a haemoglobin value > or = 11 g/dL (18 per 1,000). The stillbirth rate was increased in women with haemoglobin values > or = 14.0 g/dL. With respect to low birth-weight (< 2,500 g), the rates were also higher when the haemoglobin value was above 14.0 g/dL. The reason for these findings is not apparent and may be due to the impact of an uncharacterized confounding variable rather than the haemoglobin value.

Aust
NZ
J
Obstet
Gynnecol
1999;
39:
1:
31
Anaemia
and
Perinatal
Outcome
in
Port
Moresby
G.
Mo1a
1
DPH
(Syd.),
FRCOG,
FRACOG,
M.
Permezel
2
MD,
MRCP,
MRCOG,
FRACOG,
A.B.
Amoa
3
FWACS,
FRCOG
and
C.A.
Klufio
4
FWACS,
FRCOG,
FRCS
(Edin)
Department
of
Obstetrics
and
Gynaecology,
University
of
Papua
New
Guinea,
Port
Moresby,
New
Guinea
EDITORIAL
COMMENT:
We
accepted
this
paper
for
publication
because
it
provides
interesting
information
concerning
the
incidences
of
stillbirths
and
low
birth-weight
in
a
large
series
of
women
known
to
have
anaemia
at
some
time
during
pregnancy
but
not
necessarily
at
delivery.
Information
is
also
provided
concerning
fetal
distress
in
labour
(presence
of
a
'significant'
amount
of
meconium)
and
associations
with
maternal
age
over
35
years,
grand
multiparity
(not
defined),
multiple
pregnancy,
hypertension,
syphilis
and
highland
versus
coastal
region
of
origin.
The
authors
explain
that
malaria
is
common
in
this
population
and
may
have
affected
the
associations
noted.
Diagnosis
can
be
difficult
since
women
with
chronic
malarial
infection
may
not
reveal
parasitaemia
in
a
peripheral
blood
examination.
We
are
not
told
if
yaws
is
prevalent
in
this
community
and
whether
the
cases
of
syphilis
were
active
or
had
been
treated.
Discussion
with
the
authors
reveals
that
documentation
of
data
in
this
centre
did
not
allow
consideration
of
conditions
such
as
antepartum
haemorrhage,
premature
rupture
of
the
membranes,
prolapse
of
the
umbilical
cord,
diabetes
and
other
obstetric
complications
associated
with
increased
perinatal
mortality
rates.
The
authors
found
that
severe
degrees
of
maternal
anaemia
are
associated
with
the
highest
rates
of
stillbirth
but
this
finding
would
be
more
convincing
if
cases
of
antepartum
haemorrhage
had
been
excluded.
This
study
also
showed
a
higher
stillbirth
rate
in
pregnancies
where
the
haemoglobin
value
exceeded
14g/dL.
This
finding
warrants
further
examination
which
should
list
all
known
possible
causes
of
death
especially
placental
abruption
as
already
mentioned.
Study
of
the
placenta
is
also
important
in
women
with
anaemia
and
especially
those
who
have
pregnancy
complications
or
where
perinatal
death
occurs.
In
a
series
of
267
consecutive
women
with
singleton
pregnancies
and
severe
pregnancy
anaemia
(haemoglobin
value
<8g/dL)
previously
reported
from
Papua
New
Guinea,
the
placentas
were
relatively
small
in
all
11
perinatal
deaths
(A).
Cases
of
antepartum
haemorrhage
and
multiple
pregnancy
were
excluded
from
this
series
and
there
were
no
perinatal
deaths
when
the
mothers
had
severe
anaemia
and
placental
weight
exceeding
the
90th
percentile
(700
g)
(A).
Incerpi
and
colleagues
recently
reviewed
a
series
of
745
consecutive
stillbirths
and
concluded
that
in
evaluation
of
the
cause
of
stillbirths,
'primary
emphasis
should
be
placed
on
the
autopsy
and
placental
findings,
and
the
use
of
other
clinical
tests
may
be
individualized'
(B).
(A)
Beischer
NA,
Sivasamboo
R,
Vohra
S,
Silpisornkosal
S,
Reid
S.
Placental
hypertrophy
in
severe
pregnancy
anaemia.
J
Obstet
Gynaecol
Br
Commonw
1970;
77:
398-409.
(B)
Incerpi
MH,
Miller
DA,
Samadi
R,
Settlage
RH,
Goodwin
TM.
Stillbirth
evaluation:
What
tests
are
needed?
Am
J
Obstet
Gynaecol
1998;
178:
1121-1125.
1.
Associate
Professor,
Head
of
Department.
2.
Visiting
Professor
from
Mercy
Hospital
for
Women,
University
of
Melbourne.
3.
Consultant
Obstetrician.
4.
Professor,
former
Head
of
Department.
Address
for
correspondence:
Associate
Professor
G.
Mola,
Box
1421,
Boroko,
Papua
New
Guinea.
32
AUST
AND
NZ
JOURNAL
OF
OBSTETRICS
AND
GYNAECOLOGY
RESPONSE
FROM
AUTHORS:
The
definition
of
grand
multiparity
used
in
New
Guinea
was
parity
>4
which
was
the
definition
used
in
this
study.
Yaws
is
not
prevalent
in
our
population
nowadays
and
indeed
has
not
been
so
since
the
mass
eradication
programme
in
the
1950s
and
1960s.
Therefore
positive
treponema
serology
in
the
fertile
age
group
of
women
in
Papua
New
Guinea
today
is
most
likely
not
to
be
due
to
previous
yaws
infection.
All
women
with
positive
syphilis
serology
are
treated
as
soon
as
they
are
diagnosed;
however
because
of
late
booking,
many
women
are
not
treated
until
the
third
trimester,
and
the
occasional
woman
who
only
makes
1
or
2
visits
to
the
antenatal clinic
may
not
be
treated
until
after
delivery
or
until
an
intrauterine
fetal
death
has
occurred.
Glen
Mola
Head
of
Department
of
Obstetrics
and
Gynaecology
Summary:
In
1987,
a
computerized
obstetric
database
was
set
up
at
the
Port
Moresby
General
Hospital.
Between
1987
and
1992,
27,117
births
took
place.
The
mean
haemoglobin
value
amongst
the
83%
of
women
in
whom
a
haemoglobin
value
was
tested
was
10.0
±
1.7
g/dL.
High
stillbirth
rates
(94
per
1,000)
were
associated
with
a
haemoglobin
value
<6
g/dL.
The
stillbirth
rate
was
slightly
lower
(14
per
1,000)
in
woman
whose
lowest
haemoglobin
value
was
in
the
range
10.0-10.9
g/dL
than
in
those
with
a
haemoglobin
value
g/dL
(18
per
1,000).
The
stillbirth
rate
was
increased
in
women
with
haemoglobin
values
1.4.0
g/dL.
With
respect
to
low
birth-
weight
(<2,500
g),
the
rates
were
also
higher
when
the
haemoglobin
value
was
above
14.0
g/dL.
The
reason
for
these
findings
is
not
apparent
and
may
be
due
to
the
impact
of
an
uncharacterized
confounding
variable
rather
than
the
haemoglobin
value.
Relative
to
their
Australian
neighbours,
Papua
New
Guinea
women
endure
very
high
perinatal
and
maternal
mortality
rates.
Approximately
60%
of
women
deliver
at
home
without
supervision
with
stillbirth
rates
that
may
be
as
high
as
94
per
1,000
births
(1).
In
some
regions,
the
maternal
mortality
rate
may
approach
20
per
1,000
births,
although
poor
data
collection
makes
this
difficult
to
estimate
(2).
At
the
Port
Moresby
General
Hospital,
the
stillbirth
rate
over
the last
5
years
ranged
from
20-24
per
1,000
births
(3).
The
aim
of
this
study
was
to
determine
the
relationship
between
anaemia
and
adverse
pregnancy
outcome
amongst
27,177
pregnancies
at
the
Port
Moresby
General
Hospital
in
Papua
New
Guinea.
In
recent
years
considerable
controversy
has
arisen
as
to
whether
mild
or
moderate
anaemia
has
any
significant
adverse
effect
on
the
outcome
of
pregnancy.
Overviews
of
clinical
trials
do
not
reveal
any
substantial
benefit
from
routine
iron
and
folic
acid
supplementation
during
pregnancy
(4).
However,
these
reviews
are
almost
solely
concerned
with
studies
conducted
in
developed
countries.
Anaemia
in
pregnancy
remains
a
common
and
serious
problem
in
the
developing
world
and
is
commonly
multifactorial
in
origin
(5).
Sill
et
al
reported
that
44%
of
pregnant
women
attending
the
Port
Moresby
General
Hospital
had
a
haemoglobin
value
(Hb)
<10
g/dL
(6).
Antimalarial
antibodies
were
present
in
90%
of
these
women
but
only
7%
had
parasites
detected
on
blood
fflm
examination.
The
alpha-thalassaemia
gene
was
present
in
35%,
but
sickle
cell
disease
does
not
affect
Melanesians.
Ferritin
concentration
was
low
in
8%
and
folate
deficiency
was
recorded
in
3%.
Hookworm
infestation
is
common
in
Papua
New
Guinea
and
contributes
to
anaemia
in
pregnancy.
There
is
no
doubt
that
severe
anaemia
is
associated
with
serious
maternal
and
perinatal
morbidity
and
mortality
(7).
Where
previously,
blood
transfusion
has
been
employed
to
minimize
the
consequences
of
severe
anaemia,
the
increasing
prevalence
of
HIV
has
made
this
therapeutic
option
particularly
undesirable.
Even
with
testing
of
all
donated
blood,
the
'window
period'
between
infection
and
seroconversion
allows
transmission
of
the
virus.
HIV
is
of
course
a
more
potent
and
inevitable
cause
of death
than
anaemia
(8).
The
consequences
of
malarial
infection
in
pregnancy
include
intrauterine
growth
retardation,
premature
delivery
and
perinatal
mortality
(9).
It
is
a
particularly
dangerous
disease
in
primigravidas
and
women
without
immunity
from
previous
malarial
infection.
In
women
with
chronic
malarial
infection,
peripheral
blood
film
examination
does
not
always
reveal
parasitaemia
(10)
and
the
anaemia
itself
becomes
a
useful
surrogate
marker
of
malaria
(11,12).
In
this
situation,
treatment
and
prophylaxis
with
antimalarial
drugs
is
often
followed
by
a
rise
in
the
haemoglobin
value
even
where
it
has
not
been
possible
to
demonstrate
parasites.
G.
MOLA
ET
Al.
33
PATIENTS
AND
METHODS
Between
1987
and
1991,
data
were
collected
from
27,177
pregnant
women
who
were
booked
at
the
antenatal
clinics
in
or
around
Port
Moresby
and
delivered
in
Port
Moresby
General
Hospital.
At
delivery,
the
woman's
antenatal
and
confinement
records
were
recorded
on
a
computerized
database
in
the
labour
ward.
Every
woman
had
a
haemoglobin
estimation
at
booking.
(In
Port
Moresby,
the
average
gestational
age
at
booking
is
22
weeks).
If
the
first
haemoglobin
value
was
<10
g/dL,
the
haemoglobin
value
was
rechecked
monthly
until
the
anaemia
was
corrected.
Otherwise,
the
haemoglobin
value
was
only
rechecked
at
36
weeks'
gestation.
The
lowest
haemoglobin
value
recorded
during
the
pregnancy
was
the
value
recorded
in
the
database.
Analysis
of
the
data
was
made
using
the
Statistical
Package
for
Interactive
Data
Analysis
(SPIDA).
RESULTS
During
the
index
pregnancy
22,405
or
83%
of
women
in
the
database
had
a
haemoglobin
estimation.
The
haemoglobin
values
ranged
between
1.0
and
19.3
g/dL
with
a
mean
of
10.0
±
1.7
g/dL.
The
mean
haemoglobin
concentration
for
highland
women
was
10.3
±
1.6
g/dL
and
for
coastal
women
the
mean
was
10.0
±
1.8
g/dL.
Table
1.
Stillbirth
Rates
and
Anaemia
Haemoglobin
value
(g/dL)
Total
patients
Stillbirths
(%)
<6
53
5
(9.4%)
6-6.9
84
4
(4.8%)
7-7.9
285
8
(2.8%)
8-8.9
1,075
36
(3.3%)
9-9.9
2,379
53
(2.2%)
10-10.9
3,928
55
(1.4%)
11-11.9
5,507 85
(1.5%)
12-12.9
4,629
85
(1.8%)
13-13.9
2,716
50
(1.8%)
14-14.9
1,223
31
(2.5%)
>14.9
526
15
(2.9%)
Total
22,405
427
(1.9%)
Table
1
relates
the
stillbirth
rates
to
the
haemoglobin
levels.
It
can
be
seen
that
there
was
considerably
higher
fetal
wastage
(94
per
1,000)
in
women
with
a
haemoglobin
value
<6
g/dL;
stillbirth
rates
were
lowest
with
the
haemoglobin
value
in
the
range
10.0-10.9
g/dL
(14
per
1,000).
Table
2
shows
that
low
birth-weight
(LBW)
was
also
least
when
the
lowest
recorded
haemoglobin
value
in
pregnancy
was
in
the
range
10.0-
10.9
g/dL.
The
incidence
of
LBW
approached
40%
when
the
haemoglobin
value
was
less
than
6
g/dL.
Meconium-staining
of
the
amniotic
fluid
was
also
more
likely
in
the
presence
of
severe
anaemia
(table
3).
Unfortunately,
other
parameters
that
were
of
interest
were
either
not
recorded
on
the
database
(Apgar
scores)
or
impossible
to
calculate
(prematurity
and
small
for
dates
due
to
the
absence
of
any
reliable
assessment
of
gestational
age).
Table
2.
Low
Birth-weight
(<2,500
g)
and
Anaemia
Haemoglobin
value
(g/dL)
Total
patients
LBW
(%)
<6
53
21
(39.6%)
6-6.9
84
17
(20.2%)
7-7.9
285
54
(18.9%)
8-8.9
1,075
150
(14.0%)
9-9.9
2,379
274
(11.5%)
10-10.9
3,928
325
(8.3%)
11-11.9
5,507
499
(9.1%)
12-12.9
4,629
407
(8.8%)
13-13.9
2,716
241
(8.9%)
14-14.9
1,223 133
(10.9%)
>14.9
526
65
(12.4%)
Total
22,405
2,186
(9.8%)
LBW
=
low
birth-weight
Table
3.
Meconium
Contamination
of
the
Liquor*
and
Anaemia
Haemoglobin
value
(g/dL)
Total
patients
Meconium
(%)
<6
53
13
(24.5%)
6-6.9
84
13
(15.5%)
7-7.9
285
49
(17.2%)
8-8.9
1,075
170
(15.8%)
9-9.9
2,379
298
(12.5%)
10-10.9
3,928
480
(12.2%)
11-11.9
5,507
658
(11.9%)
12-12.9
4,629
596
(12.9%)
13-13.9
2,716
308
(11.3%)
14-14.9
1,223
151
(12.3%)
>14.9
526
62
(11.8%)
Total
22,405
2,798
(12.5%)
*
'Significant'
meconium
as
judged
as
++
or
more.
The
relationship
of
anaemia
to
stillbirth
was
further
analyzed
by
multiple
logistic
regression
to
exclude
obvious
confounding
variables
(table
4).
Severe
anaemia
(as
defined
by
a
haemoglobin
value
<8
g/dL)
remained
a
significant
independent
risk
factor
for
stillbirth
(odds
ratio
=
1.6).
As
expected,
syphilis
correlated
strongly
with
stillbirth
although
hypertension
did
not.
Women
from
Papua
New
Guinea
Highland
provinces
were
significantly
less
likely
to
have
a
stillbirth
and
older
women
more
likely.
Potential
correlates
of
anaemia
in
pregnancy
were
also
investigated
by
multivariate
analysis
(table
5).
Anaemia
was
more
common
in
women
with
a
multiple
pregnancy
(0.R.
=
1.8),
from
the
Coastal
provinces
of
the
country
(0.R.
=
1.6),
with
syphilis
(O.R.
=
1.5)
and
in
grandmultipara
(O.R.
=
1.2).
Increased
maternal
age
was
not
related
to
the
presence
of
anaemia.
34
AUST
AND
NZ
JOURNAL
OF
OBSTETRICS
AND
GYNAECOLOGY
Table
4.
Multivariate
Analysis
of
Factors
Related
to
Stillbirth
O.R.*
(95%
CI)
Syphilis
3.0
(2.0
-
4.5)
<0.001
Multiple
pregnancy
2.7
(1.8
-
4.0)
<0.001
Maternal
age
>35
years
2.0
(1.5
-
2.6)
<0.001
Haemoglobin
value
<8
g/dL
1.6
(1.3
-
2.0)
<0.001
Highland
region
of
origin
0.7
(0.6
-
0.9)
0.025
Hypertension
in
pregnancy
1.0
(0.8
-
1.3)
NS
*
Odds
ratio
with
95%
confidence
intervals;
NS
=
not
significant
at
1)=-0.05
Table
5.
Multivariate
Analysis
of
Factors
Related
to
Anaemia
O.R.*
(95%
CI)
Multiple
pregnancy
1.8
(1.4
-
2.3)
<0.001
Coastal
region
of
origin
1.6
(1.4
-
1.7)
<0.001
Syphilis
1.5
(1.2
-
1.9)
<0.05
Grand
multiparity
1.2
(1.1
-
1.4)
<0.01
Maternal
age
>35
years
1.0
(0.9
-
1.2)
NS
*
Odds
ratio
with
95%
confidence
intervals;
NS
=
not
significant
at
p=0.05
DISCUSSION
It
is
well
known
that
different
populations
have
different
rates
of
anaemia.
The
World
Health
Organization
has
collected
data
from
many
developing
countries
and
found
that
in
some
of
these
communities
more
than
two-thirds
of
pregnant
women
are
'anaemic'
when
assessed
by
the
WHO
definition
(haemoglobin
value
<11
g/dL)
(13).
The
situation
in
antenatal
women
from
Port
Moresby
is
similar
with
a
mean
haemoglobin
value
of
10.0
g/dL.
Malaria
is
endemic
on
most
of
the
coast
of
Papua
New
Guinea
but
is
not
present
in
most
of
the
Highland
regions.
The
strong
relationship
of
anaemia
with
`Coastal'
province
of
origin
is
therefore
expected.
Syphilis
remains
a
serious
problem,
particularly
in
the
Highland
provinces
of
Papua
New
Guinea.
It
correlates
independently
with
both
stillbirth
and
anaemia.
Pregnancy
is
clearly
one
of
the
main
factors
which
depletes
iron
stores
in
women
in
developing
countries
as
seen
by
the
higher
incidence
of
anaemia
in
grand
multipara.
Older
women
were
equally
likely
to
be
anaemic
as
younger
women.
Although
the
older
age
group
tend
to
be
of
higher
parity,
older
women
in
Papua
New
Guinea
have
more
social
standing
and
may
have
access
to
better
nutrition.
In
the
Port
Moresby
antenatal
population,
this
study
found
that
the
optimal
obstetric
performance
(as
evidenced
by
the
lowest
incidence
of
stillbirth
and
LBW)
was
seen
with
a
haemoglobin
value
in
the
range
10.0-10.9
g/dL.
Garn
analyzed
59,391
pregnancies
as
part
of
the
National
Collaborative
Perinatal
Program
(NCPP)
and
found
that
the
haemoglobin
values
associated
with
the
lowest
rates
of
intrauterine
fetal
death,
low
birth
weight
and
prematurity
were
10-12
g/dL
for
white
and
9-11
g/dL
for
black
women
(14).
It
may
be
that
the
optimal
haemoglobin
value
for
a
favourable
pregnancy
outcome
varies
between
different
populations.
A
large
trial
in
Finland
of
routine
versus
selective
administration
of
iron,
found
an
unexpected
increase
in
perinatal
mortality
in
the
`routinely'
supplemented
group
(15).
While
such
information
should
be
interpreted
cautiously,
further
exploration
is
indicated
(16).
Given
a
lack
of
apparent
adverse
sequelae
of
mild
degrees
of
anaemia
(as
defined
by
the
WHO),
what
are
the
therapeutic
implications?
Firstly,
the
pathological
causes
of
anaemia
(e.g.
malaria,
helminth
infestation,
iron
deficiency)
should
be
the
focus
of
the
therapeutic
thrust
rather
than
minor
haemoglobin
deficits.
Blood
transfusion
should
be
reserved
for
cases
of
severe
anaemia
not
amenable
to
other
measures,
particularly
in
view
of
significant
risks
of
viral
transmission.
In
this
series,
there
was
a
marked
increase
in
adverse
outcome
if
the
haemoglobin
value
was
less
6.0
g/dL
and
this
may
represent
a
suitable
threshold
for
blood
transfusion
in
the
Port
Moresby
pregnant
population.
References
1.
Papua
New
Guinea
National
Health
Plan
1993-1998.
Government
Printer,
PNG,
1993.
2.
Maternal
health
services
and
maternal
mortality
in
Papua
New
Guinea.
G
Mola.
Papua
New
Guinea
Med
J
1985;
28:
241-245.
3.
Annual
Reports
of
the
Division
of
Obstetrics
and
Gynaecology
at
Port
Moresby
General
Hospital
1992-1996.
4.
Mahommed
K.
Routine
iron
supplementation
in
pregnancy.
In
Enkin
M.W,
Kierse
MJNC,
Renfrew
MJ,
Nelson
JP,
eds.
Pregnancy
and
Childbirth
Module,
Cochrane
database
of
systemic
reviews,
Oxford
Cochrane
Updates
on
disk,
Update
software
1993.
5.
Fleming
AF.
The
aetiology
of
anaemia
in
pregnancy
in
Ndola,
Zambia,
Ann
Trop
Med
Parasitol
1989;
83:
37-49.
6.
Sill
RP,
Hill
AV,
Igo
JD.
Multifactorial
aetiology
of
anaemia
of
pregnancy
in
Port
Moresby,
Papua
New
Guinea,
PNG
Med
J
1987;
30:
193-198.
7.
Fleming
AF.
Tropical
Obstetrics
and
Gynaecology:
Anaemia
in
pregnancy
in
tropical
Africa,
Trans
R
Soc
Trop
Med
and
Hyg
1989;
83:
441-448.
8.
Zucker
JR,
Lackritx
EM,
Ruebush
TK
et
al.
Anaemia,
blood
transfusion
practices,
HIV
and
mortality
amongst
young
women
of
reproductive
age
in
western
Kenya.
Trans
R
Soc
Trop
Med
and
Hyg
1994;
88:
173-176.
9.
Brabin
BJ,
Ginny
M,
Sapau
J,
Galme
K,
Paino
J.
Consequences
of
maternal
anaemia
on
outcome
of
pregnancy
in
a
malaria
endemic
area
in
Papua
New
Guinea.
Ann
Trop
Med
Parasitol
1990;
84:
11-24.
10.
Felger
1,
Tavul
L,
Narara
A,
Genton
B,
Alpers
M,
Beck
HP.
The
use
of
polymerase
chain
reaction
for
more
sensitive
detection
of
Plasmodium
falciparum.
PNG
Med
J
1995;
38:
52-56.
11.
Warren
DA,
Molyneux
ME,
Beales
PE
Severe
and
complicated
malaria.
Trans
Roy
Soc
Trop
Med
and
Hyg
1990;
84:
Supp
2,
1-65.
12.Dolan
G,
ter-Kuille
FO,
Jacoutot
V
et
al.
Bed
nets
for
the
prevention
of
anaemia
in
pregnancy.
Trans
R
Soc
Trop
Med
Hyg,
1993;
87:
620-626.
13.World
Health
Organization.
The
prevalence
of
nutritional
anaemias
in
women
in
developing
countries.
WHO
Geneva
1979.
14.
Garn
SM,
Keating
MT,
Falkner
F.
Haematological
status
and
pregnancy
outcomes.
Am
J
Clin
Nutr
1981;
34:
115-117.
15.
Memminki
E,
Rimpela
U.
A
randomized
comparison
of
routine
versus
selective
iron
supplementation
during
pregnancy.
J
Am
Coll
Nutr
1991;
10:
3-10.
16.Mahomed
K.
Routine
iron
supplementation
in
pregnancy.
The
Cochrane
Library,
1998.