REVIEW

p

aMo

m

llwn

rea

or

studies showed that either, free or total, triiodothyronine or thyroxine was lower in the group of

Introduction

Thyroid hormones padaptation of metabolillness (1). In hospitaalterations are very cincreased age or in thotriiodothyronine (T3) is

rather than T3, and increased catabolism of T3

European Journal of Endocrinology (2011) 164 147155 ISSN 0804-4643

DOI: 10.1530/EJE-10-0695q 2011 European Society of Endocrinologyto 3,3-diiodothyronine (T2) (46). With increasingseverity of illness, low total and free T4, and sometimeslow TSH, can be observed (6). Decrease in plasmaT4-binding globulin (TBG) or transthyretin as well asaccumulation of substances that lower the plasmathyroid hormone-binding capacity appear also to beimportant for the above-mentioned alterations inthyroid hormone levels during critical illness (4).

For the vast majority of patients, thyroid functionabnormalities observed during critical illness are

Methods

Data sources

To identify studies eligible for inclusion in this review,we searched PubMed, on February 1, 2009, applyingthe following combined search term: (thyroid diseaseOR thyroid OR hypothyroidism OR hyperthyroidism ORTSH OR T4 OR T3 OR thyroid hormones) AND (infectionOR sepsis OR bacteremia OR pneumonia OR nosocomialgroup of patients, which can be attributed to increaseddeiodination of thyroxine (T4) to reverse T3 (rT3),patients with sepsis or septic shock who had unfavorable outcome than in those who had favorableoutcome. Two other studies showed higher TSH values in the group of patients with unfavorableoutcome. No significant relevant findings were observed in the remaining study. Regarding thecorrelation of sepsis prognostic scoring systems with thyroid function tests, the three studies thatprovided specific relevant data showed variable findings.Discussion: Most of the relevant studies identified favor the concept that decreased thyroid function atbaseline might be associated with a worse outcome of patients with sepsis or septic shock. Althoughthese findings are not consistent, the role of thyroid function in affecting or merely predicting theoutcome of sepsis or septic shock merits further investigation.

European Journal of Endocrinology 164 147155

lay an important role in theic function to stress and criticallized patients, thyroid hormoneommon, particularly in those ofse with critical illness (2, 3). Lowcommonly observed in the latter

transient and do not represent an underlying thyroiddisease (7). Still, certain data suggest that themagnitude of thyroid hormone alterations in patientsadmitted to the intensive care unit (ICU) due to variouscauses is associated adversely with the patient outcome(811). In this review, we sought to evaluate system-atically the association between baseline thyroidfunction and the outcome of patients with, specifically,sepsis or septic shock.Association between thyroid funthe outcome of patients with sea systematic reviewAnna G Angelousi1, Drosos E Karageorgopoulos1, Anast1Alfa Institute of Biomedical Sciences (AIBS), 9 Neapoleos Street, 151 2311526 Athens, Greece and 3Department of Medicine, Tufts University Scho

(Correspondence should be addressed to M E Falagas at Alfa Institute of Bio

Abstract

Introduction: The severity of critical iabnormalities. We sought to evaluateshock is associated with the thyroid fucare unit (ICU).Methods: We performed a systematicResults: We included nine studies thatneonates, and two involved adults. Mlength of ICU stay was evaluated in theness is associated with various patterns of thyroid hormonehether the outcome of patients with, specifically, sepsis or septicction tests evaluated at diagnosis or admission in the intensive

view of relevant studies by searching PubMed.ll had a prospective cohort design. Seven involved children or

rtality was the outcome evaluated in eight studies, while theemaining study. In univariate analysis, six of the nine includedction tests at baseline andsis or septic shock:

sios M Kapaskelis1,2 and Matthew E Falagas1,2,3

arousi, Athens, Greece, 2Department of Medicine, Henry Dunant Hospital,l of Medicine, Boston 02153, Massachusetts, USA

edical Sciences (AIBS); Email: m.falagas@aibs.gr)Online version via www.eje-online.org

infection OR septic shock). The bibliographies of

cohort studies (either prospective or retrospective) or

Results

Characteristics of the included studies

exclusively neonates (12, 16), and the remaining two

In Table 1, we present data on the association between

Specifically, six of the nine studies included in thisreview showed, by univariate analysis, that amongpatients with sepsis or septic shock those with anunfavorable outcome had lower baseline serum levels of,either total or free, T3 or T4, than those with a favorableoutcome (1216, 19). In two other studies that involvedchildren and adults, who in both studies were admittedin the ICU due to septic shock, there was no difference inbaseline T3 or T4 between non-survivors and survivors

Articles excluded after detailedscreening according to specific

criteria(n=193)

Articles focusing on the association of thyroid function with other than sepsis/septic shock types of critical illness or with infections without sepsis/septic shock (n=42) Animal studies (n=51) Articles that did not give data for the outcome of patients with altered thyroid function (n=3) Review articles (n=64) Studies that provided data for thyroid function tests that did not refer to baseline (n=2) Case reports: (n=6) Articles studying other than thyroid hormones (e.g. adrenal hormones) (n=25)

9 individual articles qualifyingfor inclusion in our review

Figure 1 Flow diagram of the detailed process of the selection ofarticles for inclusion in our review.

148 A G Angelousi and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164serum thyroid hormone levels at baseline and theoutcome of patients with sepsis or septic shock.studies involved adults (18, 19). Mortality was theoutcome evaluated in eight of the studies (1, 1219),while in the remaining study, the outcome evaluated wasthe length of stay in the pediatric ICU (PICU) (13). In six ofthe included studies, it was specifically reported that theblood samples for thyroid hormone measurements weretaken before administration of dopamine (1, 1216, 19).

Association of thyroid hormones at baselinewith outcomeFrom the literature searches that we performed inPubMed, we retrieved a total of 3633 articles. After firstscreening, based on the title and the abstract, we selected202 articles for further detailed evaluation, after which,we identified nine studies as eligible for inclusion in thisreview (1219). The process of selecting articles forinclusion in this review is depicted graphically in Fig. 1.

All of the nine included studies had a prospec-tive cohort design. Five of the studies involved children(1, 1315, 17), two additional studies involvedclinical trials that provided data on the associationbetween thyroid hormones at baseline and the outcomeof patients of any age with sepsis or septic shock. Weconsidered baseline thyroid function tests as thosereferring to the time of diagnosis of sepsis or the time ofadmission in the ICU. We specifically evaluated English,French, German, Italian, and Spanish language articles.

Data extraction

Data extracted from each of the included studies werethose referring to the study design, the characteristicsof the study population and the subgroups of patientscompared, the baseline values of thyroid hormones,and their statistical association with the patientoutcome through univariate or multivariate analysis,where reported. We also extracted data on thecorrelation of the thyroid hormones at baseline withsepsis prognostic scores.relevant articles were also hand searched. We per-formed an updated PubMed search on October 1, 2010.

Study selection criteria

We selected for inclusion in our review, casecontrol orwww.eje-online.orgArticles selected for furtherevaluation after first screening of

title and abstract (n=202)

Potentially relevant articlesretrieved from PubMed

(n=3633)

Table1

Ass

ocia

tion

betw

een

baselin

eth

yro

idhorm

ones

and

outc

om

eof

patients

with

sepsis

or

septic

shock

indiff

ere

nt

stu

die

s.

Study

Study

design

Characteristics

ofwholestudy

population

Comparisonof

patientgroups

(unfa

vora

ble

vers

us

favora

ble

outc

om

e),n

Agein

years,

median

(range)or

meanG

S.D.

Sex

(male

s/

tota

l)

Valuesofthyroid

horm

onesin

thecomparedgroups

Comparisonin

univariate

analysis

Comparisonin

multivariate

analysis

Thyro

idhorm

one

levels

,units

Unfa

vora

ble

outc

om

e

Favora

ble

outc

om

eS

ignifi

cance

Sig

ni-

ficance

Variable

s

ente

red

in

the

model

Independent

pre

dic

tor

variable

s

(12)

Pro

spective

case

contr

ol

stu

dy

Cases:

143

new

born

s

adm

itte

din

PIC

U

with

sepsis

due

to

Staphylococcusaureus,

Streptococcuspneumoniae,

Escherichiacoli,Klebsiella

pneumoniae,Pseudomonas

aeruginosa,Klebsiella

oxytoca

Contr

ols

:149

healthy

new

born

s

Sepsis

non-

surv

ivors

vers

us

sepsis

surv

i-

vors

,20

vs

123

All

were

term

-or

pre

term

-

neonate

s

72/1

43

T3,

ng/d

l

(meanG

S.D

.)

T4,

ng/d

l

(meanG

S.D

.)

TS

H,

mIU

/l

(meanG

S.D

.)

112.8G

51.0

5.9G

1.5

2.2G

1.4

172.2G

61.5

7.1G

2.3

4.1G

2.1

P!

0.0

01

P!

0.0

1

P!

0.0

1

P!

0.0

01

aN

AE

uth

yroid

sick

syn-

dro

me

(1)

Pro

spective

cohort

stu

dy

Tw

enty

-four

child

ren

adm

itte

din

PIC

Uw

ith

septic

shock

due

to

pulm

onary

,C

NS

,or

GI

infe

c-

tions

due

tobacte

rem

iaby

Escherichiacoli

orStrepto-

coccuspneumoniae

or

Staphylococcusaureus

Non-s

urv

ivors

vers

us

surv

i-

vors

,12

vs

12

313/2

4T

3,

ng/d

l

(media

n(9

5%

CI)

)

T4,mg/d

l

(media

n(9

5%

CI)

)

fT3,

pg/m

l

(media

n(9

5%

CI)

)

fT4,

ng/d

l

(media

n(9

5%

CI)

)

TS

H,mIU

/ml

(media

n(9

5%

CI)

)

40

(40.0

040.2

3)b

4.4

5(1

.96.0

)b

1.8

5(0

.570.9

5)b

0.7

7(0

.570.9

5)b

1.2

1(0

.272.9

6)

0.2

6(0

.220.8

8)

NS

NS

NS

NS

PZ

0.0

4

NA

NA

NA

(13)

Pro

spective

cohort

stu

dy

Fort

y-f

our

child

ren

inP

ICU

who

surv

ived

from

menin

gococcal

septic

shock

(31

had

menin

-

gococcem

ia)

Patients

with

long

vers

us

short

PIC

Usta

yc,

11

vs

33

2.5

(0.1

15.7

)vs

5(0

.316.1

)

6/1

1vs

21/3

3

T3,

nm

ol/l

(geom

etr

ic

meanG

S.E

.M.)

T4,

nm

ol/l

(geom

etr

ic

meanG

S.E

.M.)

fT4,

pm

ol/l

(geom

etr

ic

meanG

S.E

.M.)

rT3,

nm

ol/l

(geom

etr

ic

meanG

S.E

.M.)

T3/r

T3

(geom

etr

ic

meanG

S.E

.M.)

TS

H,

mIU

/l

(geom

etr

ic

meanG

S.E

.M.)

0.2

4(0

.200.2

8)d

Low

in50%

,

Hig

hin

50%

,

Hig

h

Low

Norm

al

0.3

8(0

.360.4

0)d

norm

alin

50%

norm

alin

50%

PZ

0.0

22

(rZK

0.3

5)e

NS

(rZK

0.3

7)e

NS

NS

NS

NS

NS

P!

0.0

5

NS

P!

0.0

5

P!

0.0

5

NS

Dopam

ine

adm

inis

-

tration,

T4

on

adm

is-

sio

n,

IL6

on

adm

is-

sio

n,

age,

gender,

tim

eto

firs

t

pete

chia

IL6,

T4

Thyroid function and sepsis outcome 149EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164

www.eje-online.org

Table1Continued

Study

Study

design

Characteristics

ofwholestudy

population

Comparisonof

patientgroups

(unfa

vora

ble

vers

us

favora

ble

outc

om

e),n

Agein

years,

median

(range)or

meanG

S.D.

Sex

(male

s/

tota

l)

Valuesofthyroid

horm

onesin

thecomparedgroups

Comparisonin

univariate

analysis

Comparisonin

multivariate

analysis

Thyro

idhorm

one

levels

,units

Unfa

vora

ble

outc

om

e

Favora

ble

outc

om

eS

ignifi

cance

Sig

ni-

ficance

Variable

s

ente

red

in

the

model

Independent

pre

dic

tor

variable

s

(14)

Pro

spective

cohort

stu

dy

Fort

y-fi

ve

child

ren

adm

itte

din

PIC

Uw

ith

menin

gococcal

sepsis

or

septic

shock

Non-s

urv

ivors

vers

us

septic

shock

surv

ivors

and

sepsis

surv

ivors

,8

vs

30

and

7

1.1

(0.5

9.4

)vs

4.3

(0.1

16.1

)and

6.1

(2.3

12.2

)

(P!

0.0

5)

7/8

vs

19/3

0

vs

4/7

T3,

nm

ol/l

(media

n(I

QR

))

T4,

nm

ol/l

(media

n(I

QR

))

fT4,

pm

ol/l

(media

n(I

QR

))

rT3,

nm

ol/l

(media

n(I

QR

))

T3/r

T3

(media

n(I

QR

))

TS

H,

mIU

/l

(media

n(I

QR

))

TB

G,

mg/l

(media

n(I

QR

))

0.4

9(0

.340.6

1)

38

(2546)

15

(1321)

0.7

5(0

.551.2

1)

0.7

1(0

.330.8

0)

0.8

8(0

.521.1

4)

9(6

10)

0.4

0(0

.300.5

2)

and

0.4

5(0

.420.6

1)

56

(4973)

and

86

(6892)

18

(1520)

and

19

(1420)

1.4

0(1

.111.7

3)

and

1.1

7(1

.001.2

8)

0.3

0(0

.200.4

4)

and

0.4

1(0

.360.5

8)

0.8

4(0

.591.1

8)

and

0.8

0(0

.651.6

1)

12

(1015),

and

18

(1419)

NS

and

NS

P!

0.0

5and

P!

0.0

5

NS

and

NS

P!

0.0

5and

P!

0.0

5

P!

0.0

5and

P!

0.0

5

NS

P!

0.0

5and

P!

0.0

5

NS

NS

NS

NS

NS

NS

NS

T4,fT

4,T

3,rT

3,

T3/r

T3,T

S-

H,T

BG

,

age,gen-

der,

cort

i-

sol,

NE

FA

/alb

u-

min

ratio

,

time

tofir

st

pete

chia

,

IL6

IL6

(15)

Pro

spective

cohort

stu

dy

Seventy

-tw

ochild

ren

adm

itte

din

the

PIC

Uw

ith

sepsis

or

septic

shock

due

tobacte

rem

iaby

Pseudomonasaeruginosa,

Klebsiella

pneumoniae,Sta-

phylococcusaureus,

or

Escherichiacoli

Non-s

urv

ivors

vers

us

surv

ivors

,

26

vs

46

Sepsis

cases:

2.5G

2.3

,septic

shock

cases:

3.0G

2.3

35/7

2T

3,

nm

ol/l

(meanG

S.D

.)

T4,

nm

ol/l

(meanG

S.D

.)

fT3,

pm

ol/l

(meanG

S.D

.)

fT4,

pm

ol/l

(meanG

S.D

.)

TS

H,

mIU

/l

(meanG

S.D

.)

0.5

8G

0.1

6

64.5G

15.8

0.0

16G

0.0

04

12.7

7G

3.2

2

4.2G

2.2

1.0

0G

0.1

6

105.7

8G

19.3

5

0.0

30G

0.0

05

20.6

4G

3.4

8

4.9G

2.1

P!

0.0

01

P!

0.0

01

P!

0.0

01

P!

0.0

01

NS

NA

NA

NA

NA

NA

NA

NA

(16)

Pro

spective

cohort

stu

dy

Fort

y-n

ine

neonate

sw

ith

sepsis

(docum

ente

dbacte

rem

iain

11

and

menin

gitis

in8)

Non-s

urv

ivors

vers

us

surv

ivors

,

16

vs

33

0.7

3G

0.0

3N

DT

3,

ng/d

l

(meanG

S.D

.)

T4,mg/d

l

(meanG

S.D

.)

TS

H,mIU

/ml

(meanG

S.D

.)

57.7G

21.1

4.6G

3.1

9.0G

13.4

116.5G

50.6

7.3G

4.0

9.6G

10.5

P!

0.0

01

P!

0.0

1

NS

NA

NA

NA

NA

NA

(17)

Pro

spective

cohort

stu

dy

Tw

enty

-six

child

ren

adm

itte

din

PIC

Uw

ith

sepsis

due

tobac-

tere

mia

byNeisseriameningi-

tidis

Non-s

urv

ivors

vers

us

surv

ivors

,

8vs

18

0.8

3vs

2.4

(P!

0.0

5)

16/2

6T

3,

nm

ol/l

(media

n(I

QR

))

T4,

nm

ol/l

(media

n(I

QR

))

fT4,

pm

ol/l

(media

n(I

QR

))

rT3,

nm

ol/l

(media

n(I

QR

))

T3/r

T3,

(media

n(I

QR

))

TS

H,

mE

/l

(media

n(I

QR

))

0.5

3(0

.430.7

6)

38

(2546)

14

(1321)

0.7

5(0

.550.9

7)

0.7

6(0

.640.8

5)

0.9

7(0

.521.5

6)

0.3

8(0

.260.4

2)

44

(3956)

16

(1519)

1.4

4(0

.991.8

5)

0.4

3(0

.170.3

5)

0.2

9(0

.150.5

4)

P!

0.0

5

NS

NS

P!

0.0

1

P!

0.0

1

P!

0.0

1

NA

NA

NA

NA

NA

NA

NA

NA

(18)

Pro

spective

cohort

stu

dy

Tw

enty

-seven

adults

adm

itte

din

the

ICU

with

septic

shock

due

topneum

onia

,perito

nitis

,

urinary

tract

infe

ction,

or

bac-

tere

mia

Non-s

urv

ivors

vers

us

surv

ivors

,

12

vs

15

50G

19

18/2

7T

3,

ng/m

l

(meanG

S.D

.)

T4,

ng/m

l

(meanG

S.D

.)

TS

H,mIU

/ml

(meanG

S.D

.)

0.0

38G

0.1

9

46.1G

19.7

0.4

4G

0.6

1

0.0

38G

0.1

6

44.6G

9.7

0.7

9G

0.6

8

NS

NS

PZ

0.1

8(N

S)

NA

NA

NA

NA

NA

150 A G Angelousi and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164

www.eje-online.org

(1, 18). In the remaining study, which evaluatedTable1Continued

Study

Study

design

Characteristics

ofwholestudy

population

Comparisonof

patientgroups

(unfa

vora

ble

vers

us

favora

ble

outc

om

e),n

Agein

years,

median

(range)or

meanG

S.D.

Sex

(male

s/

tota

l)

Valuesofthyroid

horm

onesin

thecomparedgroups

Comparisonin

univariate

analysis

Comparisonin

multivariate

analysis

Thyro

idhorm

one

levels

,units

Unfa

vora

ble

outc

om

e

Favora

ble

outc

om

eS

ignifi

cance

Sig

ni-

ficance

Variable

s

ente

red

in

the

model

Independent

pre

dic

tor

variable

s

(19)

Pro

spective

cohort

stu

dy

Thirty

-seven

adults

with

sepsis

Non-s

urv

ivors

vers

us

surv

ivors

,

15

vs

22

57.6G

17.8

37/3

7T

3,

ng/d

l

(meanG

S.D

.)

T4,mg/d

l

(meanG

S.D

.)

fT4I,

(meanG

S.D

.)

rT3,

ng/d

l

(meanG

S.D

.)

TS

H,mIU

/ml

(meanG

S.D

.)

30.4

0G

13.4

0

5.5

0G

1.7

0

3.4

7G

0.8

9

41.2

0G

18.2

0

1.8

5G

0.8

1

52.5G

19.6

0

7.2

0G

2.8

0

4.5

3G

1.7

9

40.8

0G

12.9

0

1.8

6G

0.6

8

P!

0.0

01

P!

0.0

5

P!

0.0

5

NS

NS

NA

NA

NA

NA

NA

NA

NA

(P)I

CU

,(p

edia

tric

)in

tensiv

ecare

unit;

GI,

gastr

oin

testinal

tract;

(f/r

)T

3,

(fre

e/r

evers

e)

triio

doth

yro

nin

e;

fT4(I

),fr

ee

thyro

xin

e(index);

TB

G,

thyro

xin

e-b

indin

gglo

bulin

;IQ

R,

inte

rquart

ilera

nge;

CI,

confidence

inte

rval;

NS

,non-s

ignifi

cant;

NA

,non-a

vaila

ble

;N

EF

A,

non-e

ste

rified

fatt

yacid

levelin

the

alb

um

infr

action

of

blo

od

seru

m.

aB

ym

ultiv

ariate

analy

sis

euth

yro

idsic

ksyndro

me

was

rela

ted

by

augm

ente

dm

ort

alit

y(P!

0.0

01).

bV

alu

es

for

the

whole

stu

dy

popula

tion.

cS

hort

PIC

Usta

ycases

was

defined

as!

7days

and

long

sta

yas

more

than

7days.

dC

Iextr

acte

dfr

om

gra

phs.

er,

Spearm

ans

corr

ela

tion

coeffi

cie

nt.

Thyroid function and sepsis outcome 151EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164children with meningococcal sepsis admitted to thePICU, those who did not survive had higher baselineserum levels of T3 than those who survived, while nodifference was detected in the total and free T4 levels(17). In this study, non-survivors also had higherbaseline levels of TSH.

Higher baseline TSH values in non-survivors werealso observed in another study that evaluated childrenwith septic shock admitted to the PICU. This study didnot demonstrate differences in the levels of T3 or T4between non-survivors and survivors (1). On thecontrary, lower baseline TSH was associated withhigher mortality in a study that evaluated septicneonates admitted in the ICU (12); this was alsoaccompanied with lower T3 and T4. The baseline TSHlevels in the remaining six studies did not differbetween patients with an unfavorable outcomecompared with those with a favorable outcome. Inaddition, a lower rT3 value and a higher T3/rT3 ratiowere associated with an adverse outcome in twostudies (14, 17), whereas non-significant findings wereobserved in the other two studies that assessed thesame parameters (13, 19).

Three of the included studies further evaluated theabove-described associations in multivariate analysis.One of these studies showed that in pediatric patientswho survived meningococcal septic shock, lowerbaseline serum T4 was an independent predictor ofan unfavorable outcome, specifically prolonged ICUstay (13). The baseline interleukin 6 (IL6) level wasalso independently associated with this outcome. Thelength of ICU stay was estimated to increase by 15%for every doubling of the IL6 concentration and 16%for every 10 nmol/l decrease in the T4 level (13). Thesecond study that included pediatric ICU patients withmeningococcal sepsis or septic shock showed that theeffect of the thyroid function tests on mortality lostsignificance in the multivariate analysis, and that IL6was the only independent predictor of this outcome(14). In the remaining study, which was performed onnewborns admitted to the ICU with bacterial sepsis,the presence of euthyroid sick syndrome was indepen-dently associated with mortality (12).

In Table 2, we present data on the correlation ofthyroid function tests at baseline with various sepsisprognostic scores. Such associations were reported inthree of the nine included studies, which all evaluatedchildren with meningococcal sepsis or septic shockadmitted to the PICU (13, 14, 17). In one of thesestudies, both the pediatric risk of mortality (PRISM)score and the sequential organ failure assessment scorehad a moderate negative correlation with the baselineT4 value (14). In another study, the PRISM score had arather weak positive correlation with baseline TSH (17).In the remaining study, no significant relevant associ-ations were observed (13).www.eje-online.org

Table 2 Correlation between baseline thyroid hormones and sepsis prognostic scores in patients with sepsis or septic shock included in

r

linti

152 A G Angelousi and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164Discussion

The majority (six out of nine) of the evaluated relevantstudies showed that lower baseline thyroid hormonevalues, either T3 or T4, are associated with worseoutcome for patients with sepsis or septic shock. Theabove observations apply mainly to children rather thanto adults, as the latter group was evaluated in only twoof the included studies. Still, definite conclusions on theabove issue cannot be drawn on the basis of the dataavailable herein, as the associations observed were notalways consistent between the included studies, orwithin each study with regard to different thyroidfunction tests evaluated. Moreover, the association ofthyroid hormones at baseline and the outcome ofpatients with sepsis or septic shock was not commonlyevaluated in appropriate multivariate models to adjustfor the effect of potential confounders.

It is interesting that one of the included studiesshowed that, among the 26 children with meningo-coccal sepsis studied, the 18 who survived had lower T3levels at PICU admission than the 8 who did not survive(17). This finding stands in contrast to the rest of the

different studies.

Study

Baseline sepsis prognosticscore, median (range) ormeanGS.D. (patients withunfavorable versusfavorable outcome)

Cor

T3

(13) PRISM: 24 (1246) vs20 (935)

NS

(14) PRISM: 32 (2343) vs(21 (835) and 9 (513))

NS

SOFA: 15 (1319) vs(9 (317) and 2 (06))

(17) PRISM: 31 (2934) vs17 (1417)

rZ0.29(NS)

fT3, free triiodothyronine; fT4, free thyroxine; TBG, thyroxine-binding globuassessment score; APACHE, acute physiologic and chronic health evaluacorrelation-coefficient.included studies, which either showed the opposite orno relevant association. One potential explanation forthe discordant findings of the study in this regard is thatten of the survivors received treatment with dopamine,whereas the non-survivors received norepinephrine ordobutamine. Dopamine can suppress the pituitaryrelease of TSH and thus potentially the production ofT3 (20), while norepinephrine is believed to stimulatethe secretion of TSH (21). Notably, the non-survivors inthis study also had higher TSH levels (17). Yet, it wasnot specifically stated whether dopamine adminis-tration preceded the collection of blood samples for thethyroid hormone measurements. However, in the greatmajority of the remaining studies, thyroid hormonemeasurements were performed prior to dopamine use(1, 1215, 19).

www.eje-online.orgIn addition, in the study, in which higher T3 levelswere observed in non-survivors of meningococcal septicshock, the age of the non-survivors was significantlylower than that of the survivors (10 vs 29 monthsrespectively) (17). The changes that occur in thyroidhormone levels during the first month of life could, atleast in part, account for the differences in the T3 levelsobserved in this study. Specifically in neonates, theplasma thyroid hormone levels are typically elevatedcompared with those in older children (22). This isrelated to the TSH surge that occurs in the immediatepostnatal period and to the elevated TBG levelssecondary to maternal estrogen (22). In normal termneonates, the total and free T4 levels fall gradually overthe next 46 weeks but remain higher than in olderchildren and adults for a few months. The T3 levelsgradually reach those of infancy, between 2 and 12weeks of life (22, 23). In early neonatal life, there hasalso been observed an increased activity of type IIIdeiodinase (D3) enzyme with secondary increase in rT3levels and increase in the degradation of T3 to T2, whichreaches the adult range during the fourth postnatalmonth (22, 23).

elation of thyroid hormones with sepsis prognostic score

T4 fT3 fT4 TSH TBG

NS NS NS NS NS

rZK0.62(P!0.05)

NS NS NS rZK0.54(P!0.05)

rZK0.69(P!0.05)

rZK0.57(P!0.05)

rZK0.34(NS)

ND rZK0.41(NS)

rZ0.42(P!0.05)

ND

; PRISM, pediatric risk of mortality score; SOFA, sequential organ failureon score; NS, non-significant; ND, no data were reported; r, SpearmansIt is questionable whether the findings of our reviewcan be applicable to adult patients with sepsis or septicshock, who were evaluated in only two of the includedstudies. Relevant data suggest that children havedifferent hemodynamic responses to critical illness andseptic shock. Children more commonly suffer fromprogressive cardiac failure than adults and can respondto inotropic therapy (24). Mortality in pediatric septicshock is usually lower than in adults (24).

It is generally accepted that the alterations in thyroidhormones observed during critical illness constitute partof an adaptive metabolic response. This is based on thefinding that the great majority of patients recover normalthyroid function after the critical illness subsides (25). Ithas also been suggested that the decrease in metabolicfunction observed during the systemic inflammatoryresponse syndrome and the accompanying multiorgan

dysfunction may help protect cell survival (26). Still, antigen (38, 39). These actions can include the

of TSH (40). It has thus been hypothesized that the

Thyroid function and sepsis outcome 153EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164thyroid disorders are relatively common in the generalpopulation, with an estimated prevalence of 110% (27).The presence of even subclinical abnormalities might beimportant (28), as subclinical hypothyroidism has beenlinked to excess mortality in certain patient groups (28,29). Some studies have also found that a subset ofpatients with critical illness can have true hypothyroid-ism (11). High TSH or reduced rT3 can be suggestive ofsuch a diagnosis (11, 30). Thus, the findings of some ofthe studies included in our review that higher TSH, lowerrT3, or lower T3 and T4 are associated with anunfavorable outcome of patients with sepsis or septicshock could imply that, at least for a minority of thesepatients, thyroid hypofunction during sepsis or septicshock might influence per se the outcome of thiscondition. This hypothesis merits further evaluation inappropriately designed studies.

Some studies performed in animal models of sepsis orseptic shock support the hypothesis that relative thyroidinsufficiency is associated with a worse outcome. Forexample, lower baseline serum T4 and free T4 have beenassociated with decreased likelihood for survival inseverely ill dogs (puppies) with parvoviral diarrhea (31).In an experimental sepsis model in rats, those that werepreviously thyroidectomized showed abolishment of thehyperdynamic response that physiologically accom-panies early-stage sepsis. In the thyroidectomizedanimals (32), it was observed that the administrationof T4 reversed the decrease in survival.

Another experimental study in rats showed thatthyroid hormone supplementation during sepsis canincrease survival (33). A particular beneficial effect ofthyroid hormone administration in rats with experimen-tally induced sepsis has been documented on lungmechanics and histology, which was mainly attributedto enhanced synthesis of surfactant (34, 35). Still, otherstudies on experimentally induced sepsis have failed toshow a benefit of thyroid hormone replacement (7, 32,36). In humans, there is little evidence regarding thyroidhormone substitution during critical illness. Of note, somerelevant clinical studies have suggested that T4 supple-mentation can lead to reduction in the needs forvasoactive drug administration in circulatory shock (37).

Thyroid hormone supplementation during sepsis orseptic shock can lead to a reciprocal decrease in TSH.This issue needs particular attention, as there appears tobe a rather complex pathophysiological interplaybetween TSH and the immune system. Specifically,TSH has been shown to affect the function of varioustypes of hematopoietic and immune system cells thatexpress the TSH receptor (38, 39). Such cells mainlyinclude subsets of hematopoietic cells in the bonemarrow, as well as peripheral monocytes, dendriticcells, and T-lymphocytes. Furthermore, some of theabove cell types can produce biologically active TSH thatcan have autocrine and paracrine actions, which caninfluence the early stages of the immune response to animmune system cells can affect the systemic thyroidhormone activity.

It is debatable whether the data included in ourreview regarding the association between thyroidfunction and sepsis or septic shock can be extrapolatedto other types of critical illness. Several inflammatorycytokines, such as IL1b, IL6, and TNF-a, can suppress,via direct or indirect pathways, the thyroid function atdifferent levels (41, 42). In sepsis, the increase in theproduction of pro-inflammatory cytokines is morepronounced than that in other types of critical illness(43, 44). In this respect, baseline levels of thyroidhormones, including T4, T3, and TSH, can be substan-tially lower in septic patients than in non-septic patientswith critical illness of similar severity (45). The degreeof endothelial activation and dysfunction can also begreater in sepsis than in other types of critical illness,as reflected by higher levels of E-selectin, intercellularadhesion molecule-1, and von Willebrand factoractivity that have been observed in some studies(43, 44, 46).

The findings of our review regarding the associationbetween thyroid hormone abnormalities and the out-come of patients with sepsis or septic shock indicate thatthese abnormalities could be of prognostic value. In thestudies included in our review, the association betweenestablished sepsis prognostic systems and baselinethyroid hormones was at the most moderate. Thus,the prognostic value of thyroid hormones may beindependent of other prognostic markers. Likewise,other studies performed in critically ill patients haveshown that taking into consideration the baselinethyroid function tests can add to the predictive capacityof the APACHE score (11, 47).

In conclusion, the majority of the identified studiesthat evaluated the baseline thyroid function tests inpatients with sepsis or septic shock provide data thatfavor the existence of an association between lower T3 orT4 and worse outcome. Since thyroid hormone abnorm-alities are very common in septic patients, future studiesshould aim to more clearly establish the strength of theabove-mentioned association or even examine whethera causal relationship between thyroid hypofunction andadverse outcome exists. The role of the thyroid hormoneabnormalities as predictors of outcome of septic patientson top of the known risk prognostic scoring systemswarrants also further evaluation.

Declaration of interest

The authors declare that there is no conflict of interest that could beperceived as prejudicing the impartiality of the review reported.regulation of the synthesis and release of mediators ofinflammation, such as IL6 and tumor necrosis factor-a(TNF-a). Various immune system cells have also beenfound to express T3, a process that is under the influencewww.eje-online.org

Funding

syndrome in meningococcal sepsis: the impact of peripheralthyroid hormone metabolism and binding proteins. Journal ofClinical Endocrinology and Metabolism 2005 90 56135720.

relationship to survival in children with bacterial sepsis and septic

154 A G Angelousi and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164(doi:10.1210/jc.2005-0888)15 Yildizdas D, Onenli-Mungan N, Yapicioglu H, Topaloglu AK,

Sertdemir Y & Yuksel B. Thyroid hormone levels and theirThis review did not receive any specific grant from any funding agencyin the public, commercial or not-for-profit sector.

References

1 Lodha R, Vivekanandhan S, Sarthi M, Arun S & Kabra SK. Thyroidfunction in children with sepsis and septic shock. Acta Paediatrica2007 96 406409. (doi:10.1111/j.1651-2227.2007.00135.x)

2 Iglesias P, Munoz A, Prado F, Guerrero MT, Macas MC, Ridruejo E,Tajada P & Dez JJ. Alterations in thyroid function tests in agedhospitalized patients: prevalence, etiology and clinical outcome.Clinical Endocrinology 2009 70 961967. (doi:10.1111/j.1365-2265.2008.03421.x)

3 Simons RJ, Simon JM, Demers LM & Santen RJ. Thyroiddysfunction in elderly hospitalized patients. Effect of age andseverity of illness. Archives of Internal Medicine 1990 15012491253. (doi:10.1001/archinte.150.6.1249)

4 Warner MH & Beckett GJ. Mechanisms behind the non-thyroidalillness syndrome: an update. Journal of Endocrinology 2010 205113. (doi:10.1677/JOE-09-0412)

5 Sakharova OV & Inzucchi SE. Endocrine assessments duringcritical illness. Critical Care Clinics 2007 23 467490. (doi:10.1016/j.ccc.2007.05.007)

6 Peeters RP, Wouters PJ, Kaptein E, van Toor H, Visser TJ & Van denBerghe G. Reduced activation and increased inactivation of thyroidhormone in tissues of critically ill patients. Journal of ClinicalEndocrinology and Metabolism 2003 88 32023211. (doi:10.1210/jc.2002-022013)

7 Chopra IJ. Clinical review 86: euthyroid sick syndrome: is it amisnomer? Journal of Clinical Endocrinology and Metabolism 199782 329334. (doi:10.1210/jc.82.2.329)

8 Slag MF, Morley JE, Elson MK, Crowson TW, Nuttall FQ &Shafer RB. Hypothyroxinemia in critically ill patients as a predictorof high mortality. Journal of the American Medical Association 198124 543545.

9 Peeters RP, Wouters PJ, van Toor H, Kaptein E, Visser TJ & Van denBerghe G. Serum 3,3 0,5 0-triiodothyronine (rT3) and 3,5,30-triiodothyronine/rT3 are prognostic markers in critically illpatients and are associated with postmortem tissue deiodinaseactivities. Journal of Clinical Endocrinology and Metabolism 2005 9045594565. (doi:10.1210/jc.2005-0535)

10 Plikat K, Langgartner J, Buettner R, Bollheimer LC,Woenckhaus U, Scholmerich J & Wrede CE. Frequency andoutcome of patients with nonthyroidal illness syndrome in amedical intensive care unit. Metabolism 2007 56 239244.(doi:10.1016/j.metabol.2006.09.020)

11 Rothwell PM, Udwadia ZF & Lawler PG. Thyrotropin concen-tration predicts outcome in critical illness. Anaesthesia 1993 48373376. (doi:10.1111/j.1365-2044.1993.tb07006.x)

12 Kurt A, Aygun AD, Sengul I, Sen Y, Kurt AN & Ustundag B. Serumthyroid hormones levels are significantly decreased in septicneonates with poor outcome. Journal of EndocrinologicalInvestigation, 2010. (doi:10.3275/7261)

13 den Brinker M, Dumas B, Visser TJ, Hop WC, Hazelzet JA,Festen DA, Hokken-Koelega AC & Joosten KF. Thyroid function andoutcome in children who survived meningococcal septic shock.Intensive Care Medicine 2005 31 970976. (doi:10.1007/s00134-005-2671-8)

14 den Brinker M, Joosten KF, Visser TJ, Hop WC, de Rijke YB,Hazelzet JA, Boonstra VH & Hokken-Koelega AC. Euthyroid sickwww.eje-online.orgshock. Journal of Pediatric Endocrinology and Metabolism 2004 1714351442.

16 Das BK, Agarwal P, Agarwal JK & Mishra OP. Serum cortisol andthyroid hormone levels in neonates with sepsis. Indian Journal ofPediatrics 2002 69 663665. (doi:10.1007/BF02722699)

17 Joosten KF, de Kleijn ED, Westerterp M, de Hoog M, Eijck FC,Hop WCJ, Voort EV, Hazelzet JA & Hokken-Koelega AC. Endocrineand metabolic responses in children with meningoccocal sepsis:striking differences between survivors and nonsurvivors. Journal ofClinical Endocrinology and Metabolism 2000 85 37463753.(doi:10.1210/jc.85.10.3746)

18 Leon-Sanz M, Lorente JA, Larrodera L, Ros P, Alvarez J, Esteban AE& Landin L. Pituitarythyroid function in patients with septicshock and its relation with outcome. European Journal of MedicalResearch 1997 2 477482.

19 Mangas-Rojas A, Garca-Rojas JF, Barba Chacon A, Millan Nunez-Cortes J & Zamora-Madaria E. Changes in the hypophysealthyroid axis and their prognostic value in sepsis. Revista ClnicaEspanola 1990 187 395398.

20 Van den Berghe G, de Zegher F & Lauwers P. Dopamine suppressespituitary function in infants and children. Critical Care Medicine1994 22 17471753. (doi:10.1007/BF00866735)

21 Fukuda S. Correlations between function of pituitarythyroid axisand metabolism of catecholamines by the fetus at delivery. ClinicalEndocrinology 1987 27 331338. (doi:10.1111/j.1365-2265.1987.tb01159.x)

22 Cavallo L, Margiotta W, Kernkamp C & Pugliese G. Serum levels ofthyrotropin, thyroxine, 3,3 0,5-triiodothyronine and 3,30,5 0-triio-dothyronine (reverse T3) in the first six days of life. Acta PaediatricaScandinavica 1980 69 4347. (doi:10.1111/j.1651-2227.1980.tb07027.x)

23 Santini F, Chiovato L, Ghirri P, Lapi P, Mammoli C, Montanelli L,Scartabelli G, Ceccarini G, Coccoli L, Chopra IJ, Boldrini A &Pinchera A. Serum iodothyronines in the human fetus and thenewborn: evidence for an important role of placenta in fetalthyroid hormone homeostasis. Journal of Clinical Endocrinology andMetabolism 1999 84 493498. (doi:10.1210/jc.84.2.493)

24 Ceneviva G, Paschall JA, Maffei F & Carcillo JA. Hemodynamicsupport in fluid-refractory pediatric septic shock. Pediatrics 1998102 19. (doi:10.1542/peds.102.2.e19)

25 Attia J, Margetts P & Guyatt G. Diagnosis of thyroid disease inhospitalized patients: a systematic review. Archives of InternalMedicine 1999 159 658665. (doi:10.1001/archinte.159.7.658)

26 Singer M, De Santis V, Vitale D & Jeffcoate W. Multiorgan failure isan adaptive, endocrine-mediated, metabolic response to over-whelming systemic inflammation. Lancet 2004 364 545548.(doi:10.1016/S0140-6736(04)16815-3)

27 Stone MB & Wallace RB (eds). Committee on Medicare Coverage ofRoutine Thyroid Screening. Washington DC: National AcademiesPress, 2003.

28 Haentjens P, Van Meerhaeghe A, Poppe K & Velkeniers B.Subclinical thyroid dysfunction and mortality: an estimate ofrelative and absolute excess all-cause mortality based ontime-to-event data from cohort studies. European Journal ofEndocrinology 2008 159 329341. (doi:10.1530/EJE-08-0110)

29 Razvi S, Shakoor A, Vanderpump M, Weaver JU & Pearce SH. Theinfluence of age on the relationship between subclinicalhypothyroidism and ischemic heart disease: a meta-analysis.Journal of Clinical Endocrinology and Metabolism 2008 9329983007. (doi:10.1210/jc.2008-0167)

30 Burmeister LA. Reverse T3 does not reliably differentiatehypothyroid sick syndrome from euthyroid sick syndrome. Thyroid1995 5 435441. (doi:10.1089/thy.1995.5.435)

31 Schoeman JP & Herrtage ME. Serum thyrotropin, thyroxine andfree thyroxine concentrations as predictors of mortality incritically ill puppies with parvovirus infection: a model forhuman paediatric critical illness? Microbes and Infection 2008 10203207. (doi:10.1016/j.micinf.2007.11.002)

32 Moley JF, Ohkawa M, Chaudry IH, Clemens MG & Baue AE.Hypothyroidism abolishes the hyperdynamic phase and increasessusceptibility to sepsis. Journal of Surgical Research 1984 36265273. (doi:10.1016/0022-4804(84)90097-0)

33 Inan M, Koyuncu A, Aydin C, Turan M, Gokgoz S & Sen M.Thyroid hormone supplementation in sepsis: an experimentalstudy. Surgery Today 2003 33 2429. (doi:10.1007/s005950300004)

34 Dulchavsky SA & Bailey J. Triiodothyronine treatment maintainssurfactant synthesis during sepsis. Surgery 1992 112 475479.

35 Dulchavsky SA, Kennedy PR, Geller ER, Maitra SR, Foster WM &Langenbeck EG. T3 preserves respiratory function in sepsis.Journal of Trauma 1991 31 753758. (doi:10.1097/00005373-199106000-00004)

36 Little JS. Effect of thyroid hormone supplementation on survivalafter bacterial infection. Endocrinology 1985 117 14311435.(doi:10.1210/endo-117-4-1431)

37 Zuppa AF, Nadkarni V, Davis L, Adamson PC, Helfaer MA,Elliott MR, Abrams J & Durbin D. The effect of a thyroid hormoneinfusion on vasopressor support in critically ill children withcessation of neurologic function. Critical Care Medicine 2004 3223182322. (doi:10.1097/01.CCM.0000146133.52982.17)

38 Wang HC & Klein JR. Immune function of thyroid stimulatinghormone and receptor. Critical Reviews in Immunology 2001 21323337.

39 Klein JR. Physiological relevance of thyroid stimulating hormoneand thyroid stimulating hormone receptor in tissues other thanthe thyroid. Autoimmunity 2003 36 417421. (doi:10.1080/08916930310001603019)

40 Csaba G & Pallinger E. Thyrotropic hormone (TSH) regulationof triiodothyronine (T(3)) concentration in immune cells.Inflammation Research 2009 58 151154. (doi:10.1007/s00011-008-8076-8)

41 van der Poll T, Romijn JA, Wiersinga WM & Sauerwein HP. Tumornecrosis factor: a putative mediator of the sick euthyroid syndromein man. Journal of Clinical Endocrinology and Metabolism 1990 7115671572. (doi:10.1210/jcem-71-6-1567)

42 McIver B & Gorman CA. Euthyroid sick syndrome: an overview.Thyroid 1997 7 125132. (doi:10.1089/thy.1997.7.125)

43 Casey LC, Balk RA & Bone RC. Plasma cytokine and endotoxinlevels correlate with survival in patients with sepsis syndrome.Annals of Internal Medicine 1993 119 771777.

44 Cowley HC, Heney D, Gearing AJ, Hemingway I & Webster NR.Increased circulating adhesion molecule concentrations inpatients with the systemic inflammatory response syndrome: aprospective cohort study. Critical Care Medicine 1994 22 651657.(doi:10.1097/00003246-199404000-00022)

45 Monig H, Arendt T, Meyer M, Kloehn S & Bewig B. Activation ofthe hypothalamo-pituitaryadrenal axis in response to septic ornon-septic diseases implications for the euthyroid sick syndrome.Intensive Care Medicine 1999 25 14021406. (doi:10.1007/s001340051088)

46 Cummings CJ, Sessler CN, Beall LD, Fisher BJ, Best AM & Fowler AAIII. Soluble E-selectin levels in sepsis and critical illness. Correlationwith infection and hemodynamic dysfunction. American Journal ofRespiratory and Critical Care Medicine 1997 156 431437.

47 Chinga-Alayo E, Villena J, Evans AT & Zimic M. Thyroid hormonelevels improve the prediction of mortality among patientsadmitted to the intensive care unit. Intensive Care Medicine 200531 13561361. (doi:10.1007/s00134-005-2719-9)

Received 2 November 2010

Accepted 15 November 2010

Thyroid function and sepsis outcome 155EUROPEAN JOURNAL OF ENDOCRINOLOGY (2011) 164www.eje-online.org

Outline placeholderIntroductionMethodsData sourcesStudy selection criteriaData extraction

ResultsCharacteristics of the included studiesAssociation of thyroid hormones at baseline with outcome

DiscussionDeclaration of interestFundingReferences