1. Conclusions
Public Health
Implications
Methods
Background Results
Evaluation Questions
Limitations
Recommendations
Evaluating existing systems for reducing child deaths: a state-level evaluation
of the Louisiana Child Death Review Program
Amanda Pereira, Lyn Kieltyka, Jane Herwehe, Cara Bergo
• Louisiana’s
child
mortality
rate
is
consistently
above
the
na5onal
average
• The
Louisiana
Child
Death
Review
(CDR)
Program
is
charged
with
reviewing
cases
of
unexpected
death
in
children
0
to
14
years
old
• Unexpected
child
deaths
are
those
resul5ng
from
“undiagnosed
disease,
or
trauma
in
which
the
surrounding
circumstances
of
which
are
suspicious,
obscure,
or
otherwise
unexplained”1
• A
25-‐member,
mul5-‐agency
state
panel
is
legisla5vely
tasked
with
comple5ng
an
annual
report
for
the
state
legislature
that
reviews
the
state
of
child
deaths
in
Louisiana
and
includes
recommenda5ons
for
preven5ng
future
deaths
• As
of
2015,
the
program
has
never
been
formally
evaluated
• As
part
of
the
2015
Graduate
Student
Epidemiology
Program,
the
Louisiana
CDR
program
was
evaluated
via
state-‐level
inputs,
outputs,
and
outcomes
Inputs
• Are
state
CDR
mee5ngs
effec5ve?
• Is
the
performance
of
the
CDR
database
adequate?
Outputs
• Is
CDR
producing
an
annual
legisla5ve
report?
Outcomes
• What
state-‐level
ac5ons
have
resulted
from
CDR?
Inputs
• Administer
an
engagement
and
sa5sfac5on
survey
to
state
CDR
panel
members
• Compare
roster
of
panel
members
to
aVendance
records
from
state
CDR
panel
mee5ngs
• Use
a
determinis5c
linkage
of
Vital
Records
death
cer5ficates
with
CDR
database
cases
to
determine
the
sensi5vity
and
posi5ve
predic5ve
value
(PPV)
of
the
overall
CDR
database,
manner
of
death,
autopsy
status,
race,
and
cause
of
death
• Use
logis5c
regression
to
assess
the
effect
of
turnover
of
regional
MCH
coordinators
on
the
propor5on
of
cases
reviewed
Outputs
• Review
Bureau
of
Family
Health
(BFH)
website
and
electronic
files
for
annual
CDR
reports
Outcomes
• Review
state
CDR
panel
mee5ng
minutes
for
prac5ce
changes,
programs,
system
improvements,
policies,
or
state
laws
aimed
at
preven5ng
future
deaths
Inputs
• State
mee5ngs
are
generally
effec5ve
in
terms
of
engagement
and
sa5sfac5on,
but
are
lacking
in
full
membership
and
aVendance
• The
CDR
database
is
not
performing
adequately
as
a
surveillance
system
as
it
is
not
capturing
all
true
cases
of
unexpected
child
death.
Abstracted
cases
are
also
not
always
true
cases
Outputs
• CDR
is
fulfilling
it’s
duty
of
producing
an
annual
report,
but
reports
are
not
based
on
current
data
Outcomes
• No
ac5ons
were
iden5fied
that
resulted
from
state
CDR
panel
ac5vi5es
Overall
• Areas
in
greatest
need
of
improvement
are
panel
membership
and
aVendance,
abstrac5on
of
all
eligible
cases,
classifica5on
of
SIDS
as
cause
of
death,
and
tracking
of
ac5ons
resul5ng
from
state
CDR
panel
ac5vi5es.
• Data
for
the
evalua5on
of
the
CDR
database
was
only
available
for
deaths
through
2013
• Staffing
informa5on
for
logis5c
regression
was
based
on
recall
of
regional
MCH
coordinators
• Review
of
CDR
panel
ac5vi5es
were
limited
to
available
mee5ng
minutes
(August
2011-‐March
2015)
• Surveillance
of
unexpected
child
deaths
in
Louisiana
can
be
improved
through
changes
in
mul5ple
stages
of
the
CDR
program
at
the
state-‐
level
• In
order
to
remain
effec5ve,
legisla5on
aimed
at
reducing
child
deaths
should
require
regular
evalua5ons
of
the
CDR
program
• Findings
from
this
evalua5on
can
be
shared
with
other
states
that
hope
to
evaluate
their
own
CDR
programs
Reference:
1.
Child
Death
Inves5ga5on,
Pub.
L.
No.
40,
§
2019.
Available
at:
hVps://legis.la.gov/Legis/Law.aspx?d=98002.
Updated
1999.
Accessed
6/8/2015.
This
project
was
supported
in
part
by
the
Health
Resources
and
Services
Administra5on
(HRSA)
of
the
U.S.
Department
of
Health
and
Human
Services
(HHS)
Title
V
MCH
Block
Grant
award.
This
informa5on
or
content
and
conclusions
are
those
of
the
author
and
should
not
be
construed
as
the
official
posi5on
or
policy
of,
nor
should
any
endorsements
be
inferred
by
HRSA,
HHS
or
the
U.S.
Government.
(or
any
other
disclaimer
needed)
Inputs
• Members
were
generally
engaged
and
sa5sfied
with
their
experience
on
the
state
CDR
panel
• 20
of
the
25
legisla5vely
mandated
posi5ons
were
filled
on
the
state
CDR
panel;
two
of
the
five
vacant
posi5ons
were
from
agencies
that
no
longer
exist
• An
average
of
39%
of
panel
members
aVended
each
mee5ng
• Sensi5vity=
67%;
All
true
cases
of
unexpected
child
deaths
are
not
being
captured
• Posi5ve
Predic5ve
Value
(PPV)=
84%;
some
abstracted
cases
of
unexpected
child
death
in
the
CDR
database
are
not
true
cases
Vital
Records
Eligible
Cases
Non-‐eligible
Cases
Total
CDR
Database
Abstracted
Cases
141
27
168
Non-‐abstracted
Cases
69
-‐-‐
69
Total
210
27
137
• Sensi5vity
and
PPV
of
manner
of
death,
autopsy
status,
race,
and
cause
of
death
was
over
90%
with
the
excep5on
of
SIDS
as
a
cause
of
death,
which
had
a
sensi5vity
of
29.4%
(sensi5vity
and
PPV
for
suicides
and
homicides;
other
and
unknown
race;
and
unknown
cause
of
death
was
low
due
to
small
cell
counts)
Outputs
• Reports
are
created
on
an
annual
basis
as
designated
by
legisla5on
• On
average,
there
is
a
3.5
year
gap
between
data
covered
in
CDR
reports
and
date
of
publica5on.
Outcomes
• No
state-‐level
prac5ce
changes,
programs,
system
improvements,
policies,
or
state
laws
resul5ng
from
CDR
panel
ac5vi5es
were
iden5fied
• Simple
logis5c
regression
indicated
there
was
a
3.58
(95%
confidence
interval:
2.05-‐6.25)
higher
odds
of
a
case
being
reviewed
in
regions
that
did
not
experience
turnover
in
the
posi5on
of
regional
MCH
coordinator
compared
to
those
that
did
Inputs
• Keep
an
updated
roster
of
state
CDR
panel
members
to
review
at
each
mee5ng
• Update
state
legisla5on
to
reflect
current
and
appropriate
membership
• Using
an
opera5onal
case
defini5on
with
ICD-‐10
codes,
establish
a
new
algorithm
to
beVer
iden5fy
cases
of
unexpected
child
death
Outputs
• Consider
using
provisional
vital
records
data
in
the
comple5on
of
CDR
reports
Outcomes
• Employ
a
tracking
system
for
ac5ons
discussed
at
state
CDR
mee5ngs
0%#
20%#
40%#
60%#
80%#
100%#
Natural#
Accidental#Suicide#
Hom
icide#
Undeterm
ined#
Autopsy#
W
hite##
Black#Other#
Unknow
n#
External#
SIDS#
Unknow
n#
SensiGvity#and#PosiGve#PredicGve#Value#of#Manner#of#Death,#
Autopsy#Status,#Race,#and#Cause#of#Death#
SensiGvity# PPV#
Manner
of
Death
Autopsy
Race
Cause
of
Death
Variable
1" 2" 3" 4" 5" 6" 7" 8" 9"
Reviewed" 27" 15" 16" 24" 11" 12" 23" 0" 17"
Not"Reviewed" 21" 13" 4" 9" 8" 3" 7" 19" 1"
0"
10"
20"
30"
40"
50"
60"
Number"of"Cases"
Cases"of"Unexpected"Child"Death"by"Review"Status"and"
Region""
!
Not"Reviewed" Reviewed"