Acta Pædiatrica ISSN 0803–5253

REGULAR ARTICLE

Ferritin levels in children with severe sepsis and septic shock

Pedro Celiny Ramos Garcia1,3 , Fernanda Longhi1,3 , Ricardo Garcia Branco (brancori@terra.com.br)2 , Jefferson Pedro Piva1,3 , Dani Lacks1,3 ,
Robert Charles Tasker2
1.Pediatric Intensive Care Unit, Department of Pediatrics, Hospital São Lucas da PUCRS, Porto Alegre, Brazil
2.Department of Paediatrics, School of Clinical Medicine, University of Cambridge, Cambridge, UK
3.Medical School, Pontificia Universidade Catolica do Rio Grande do Sul (PUCRS), Porto Alegre, Brazil

Keywords Abstract
Children, Ferritin, Intensive care, Iron metabolism,
Aim: To evaluate serum ferritin level in children with severe sepsis and septic shock and its
Prognosis, Sepsis
association with mortality.
Correspondence
Ricardo Garcia Branco, Department of Paediatrics, Method: A cohort study of 36 children aged 1 month–16 years with severe sepsis or septic shock
Box 116, University of Cambridge, School of Clinical requiring intensive care was conducted. Serum ferritin levels were measured at the time of diagnosis
Medicine, Addenbrookes Hospital, Hills Road, of sepsis and a ferritin index (FI = observed serum ferritin divided by the upper limit of normal ferritin
Cambridge, CB2 2QQ, United Kingdom.
Tel.: +44 1223 217715 | for age and gender) was calculated.
Fax: + 44 1223 586794 | Results: The median age (range) of the children was 6 (2–100) months. Ferritin was <200 ng/mL in
Email: brancori@terra.com.br 13 children, 200–500 ng/mL in 11 children and >500 ng/mL in 12 children. The mortality
Received associated with these groups was 23%, 9% and 58%, respectively. A ferritin >500 ng/mL was
31 May 2007; revised 10 September 2007;
accepted 17 September 2007. associated with a 3.2 (1.3–7.9) relative risk of death (p = 0.01). FI of 1.7 was the best cutoff value
for identifying those who died. In a logistic regression analysis, ferritin level and PRISM were
DOI:10.1111/j.1651-2227.2007.00564.x
independently associated with mortality.
Conclusions: Ferritin is raised in children with septic shock and high ferritin level is associated with poorer
outcome.

METHOD
INTRODUCTION
We undertook a prospective cohort study over 20 months
Ferritin is an iron storage protein highly conserved and
(January 2004–September 2005) in the Pediatric Intensive
widely distributed throughout nature (1). It has a major
Care Unit (PICU) at Hospital São Lucas da PUCRS, Porto
role in the complex process that regulates iron require-
Alegre, Brazil. The study was approved by our institutional
ment (2,3). Increased ferritin production is part of the acute
ethics review board, and children were enrolled in the study
phase reaction to infection and induces an iron-deficient
after parents gave informed consent.
milieu by reducing serum available iron (3–5). This pro-
All patients admitted to the PICU with severe sepsis
cess has been hypothesized to have evolved as a defence
or septic shock, aged 1 month–15 years, were consid-
mechanism to reduce iron available for invading organisms
ered potential subjects. The diagnosis of severe sepsis and
(4–6).
septic shock was based on the Society for Critical Care
In critical illness due to sepsis, a systemic inflammatory
Medicine and American College of Chest Physicians con-
response (SIR) is triggered and high levels of proinflamma-
sensus adapted for children (10). That is, proven or highly
tory cytokines are present in early phases of illness. Because
suspected infection, hypoperfusion, hypotension and two
proinflammatory cytokines such as interleukin 6, interleukin
or more of the following conditions: (a) body temperature
8, and tumor necrosis stimulate ferritin synthesis, ferritin
>38◦ C or <36◦ C; (b) heart rate >160/min in infants and
level in these patients should be raised. Indeed, in critically ill
>150/min in children, or higher than 2 standard deviations
adults, ferritin levels ranged from 340 to 830 ng/mL, which
(SD) above normal for age; (c) respiratory rate >60/min
is much higher than the level expected in a ‘normal’ inflam-
in infants and >50/min in children, or higher than 2 SD
matory response (6,7).
above normal for age; (d) white blood count (WBC) cell
The association between ferritin level and critical care
count >12000 cells/mm3 , or <4000 cells/mm3 , or band cells
outcome has also been described in adults. In postopera-
>10%. Septic shock was defined as persistent hypoperfu-
tive, critically ill adults’ ferritin level was associated with the
sion and hypotension after adequate fluid resuscitation, or
severity of illness (8) and, in adults with multiple trauma,
the need for inotropic or vasopressor therapy. Patients were
high ferritin was associated with an increased risk of devel-
excluded from the study if they had (1) a do not resusci-
oping multiple organ failure and acute respiratory distress
tate, or limitation of treatment order before PICU admis-
syndrome (9).
sion; (2) evidence of infection due to falciparum malaria;
In this study we have examined ferritin levels in critically
(3) family history or previous diagnosis of haemophagocytic
ill children with sepsis and assessed its association with out-
syndrome; (4) recipient of a blood transfusion in the last
come.


C 2007 The Author(s)/Journal Compilation 
C 2007 Foundation Acta Pædiatrica/Acta Pædiatrica 2007 96, pp. 1829–1831 1829
Ferritin in children with sepsis Garcia et al.

4 months; (5) proven or suspected hepatitis and (6) an Table 1 Characteristics of the population according to outcome
associated pathology known to affect ferritin levels. The vari- All children Survivors Nonsurvivors P∗
ables that were studied included demographic data, level (n = 36) (n = 25) (n = 11)
of nutrition (weight/height Z-score < 2 SD), need for me-
chanical ventilation, and days out of PICU within the first Boys 24 (67%) 16 (64%) 8 (73%) 0.7
28 days after admission (noted as 0 for children who died). Malnourished 13 (36%) 8 (32%) 5 (45%) 0.4
MV 31 (86%) 20 (80%) 11 (100%) 0.14
Pediatric Risk of Mortality (PRISM) score (11) was used to
Positive hemoculture 12 (33) 8 (32%) 4 (36%) 1.0
assess the risk of death on the day of admission. Pediatric
Age < 1 year 21 (58%) 15 (60%) 6 (55%) 0.7
Logistic Organ Dysfunction (PELOD) score (12) was used PRISM > 10% 15 (42%) 6 (24%) 9 (82%) <0.01
as an outcome measure of severity of illness. Blood culture Ferritin < 200 ng/mL 13 (36%) 10 (40%) 3 (27%) 0.3
findings, WBC count, alanine transaminase (ALT), aspartate Ferritin > 500 ng/mL 12 (33%) 5 (20%) 7 (64%) 0.02
transaminase (AST) and C-reactive protein (CRP) were also Ferritin index > 1.7 21 (58%) 10 (40%) 11 (100%) <0.01
recorded. Reference values for ALT, AST and CRP in our Log ferritin 2.49 ± 0.08 2.37 ± 0.09 2.76 ± 0.14 0.02
laboratory are 7–40 U/L, 13–40 U/L and <0.3 mg/dL, re-
MV = mechanically ventilated, PRISM = paediatric risk of mortality, CRP =
spectively. Patients were followed until PICU discharge or
C-reactive protein.
death. ∗
P-value for comparison between survivors and nonsurvivors.
Serum ferritin levels were measured at the time of di-
agnosis using a chemiluminescence technique (Immulite
chanical ventilation was required in 31 (86%) children and
2000, DPC/Médlab; Diagnostics Products Corp., Medlab
11 (30.6%) children died.
Produtos Medico-Hospitalares Ltda, Sao Paulo, Brazil).
A ferritin index (FI) was calculated using the ratio be-
tween observed ferritin and the upper limit of normal Serum ferritin
for age and sex. The upper limit of normal ferritin used The median serum ferritin was 303 (21–2210) ng/mL. Fer-
in boys aged 1–6 months, 7–12 months, 1–5 years, 6– ritin was <200 ng/mL in 13 children, 200–500 ng/mL in
9 years and 10–14 years was 410, 80, 24, 55 and 70 ng/mL, 11 children and >500 ng/mL in 12 children. The mortal-
respectively. The upper limit of normal ferritin used in girls ity in these groups was 23%, 9% and 58%, respectively.
aged 1–6 months, 7–12 months, 1–5 years, 6–9 years and Ferritin above 500 ng/mL had a 64% sensitivity and 80%
10–14 years was 340, 45, 24, 55 and 40 ng/mL, respectively specificity to predict death, and was associated with a 3.2
(13). (1.3–7.9) relative risk of death (p = 0.01). There was no asso-
In the analyses two cutoff values of serum ferritin ciation between level of ferritin and haemoglobin, CRP, AST,
were used, 200 ng/mL and 500 ng/mL. The first cutoff, ALP or age. Ferritin had a weak inverse correlation with
200 ng/mL, was used because it is the upper limit of fer- days out of PICU 28 days after admission (r2 = −0.21; p <
ritin reported in inflammatory conditions (6). The second 0.01) and a weak correlation with PELOD score (r2 = 0.23;
cutoff value, 500 ng/mL, was derived from levels reported p < 0.01). When considering only children with ferritin level
in critically ill adults. For example, ferritin level ranged from >200 ng/mL (n = 23), the association with PELOD was
340 to 830 ng/mL in critical illness and was higher in sur- moderately better (r2 = 0.61; p < 0.01).
gical patients and in those with multiple trauma (6–9). In Failure to increase ferritin above the upper limit usually
medical patients with ICU stay longer than 4 days ferritin observed in inflammation (200 ng/mL) (6) was not asso-
level was 471 ng/mL (14). ciated with increased mortality (relative risk 0.68 [0.23–
Categorical data were expressed as proportions, and 2], p = 0.4). However, among children with more severe
groups were compared using the chi-square test. Contin- presentation (PRISM > 10) 2 children had ferritin level
uous variables with normal distribution were expressed as <200 ng/mL and both died.
mean and SD, and comparisons between groups were per- The median FI was 2.4 (0.8–7.3). FI had an area under the
formed using the Pearson test. Continuous variables with ROC curve of 0.74 to predict outcome. The best cutoff value
nonparametric distribution were expressed as median with of FI to identify those who died was 1.7. The FI was >1.7 in
interquartile range. Serum ferritin level was normalized us- 58% of the children and there were no deaths in patients with
ing a log transformation. The area under the receiver opera- a value below this level (Table 1; sensitivity and specificity
tor characteristic (ROC) curve was used to assess the ability of 100% and 58%, respectively). The areas under the ROC
to predict adverse outcome. Regression analysis was per- curve using death as the outcome variable for ferritin and
formed using variables found to be significantly associated FI were 0.73 and 0.74, respectively. The relative risk of death
with outcome on univariate analysis. Statistical significance associated with an FI > 1.7 was 2.5 (1.6–4.1; p < 0.01).
was taken at p < 0.05.
Multiple regression
Univariate analysis identified PRISM, PELOD, PICU length
RESULTS of stay and ferritin (log transformed) as variables associated
Thirty-six children were enrolled in the study. There were 24 with mortality. Regression analysis showed ferritin level and
(67%) boys. In all 12 (33%) children with a positive blood PRISM to be independently associated with mortality (p =
culture. The median (range) age was 6 (2–100) months. Me- 0.04 and 0.04, respectively).

1830 
C 2007 The Author(s)/Journal Compilation 
C 2007 Foundation Acta Pædiatrica/Acta Pædiatrica 2007 96, pp. 1829–1831
Garcia et al.
DISCUSSION
In children with severe sepsis and septic shock, we found
high serum levels of ferritin similar to previous reports
in adults (4,6,14). In addition, we found that ferritin
>500 ng/mL is associated with worst outcome.
Our study, however, has some limitations that should be
addressed. First, we did not analyze the data according to
type of bacterial infection. Because the ability of different
bacteria to survive in low free-iron environments varies,
the host ferritin response and the associated outcome, may
be affected by differences in type of infecting agent (3,5).
Second, we have not analyzed the other components of
iron metabolism (e.g. free iron level, transferrin, ferritin
saturation, lactoferritin) that could also directly affect iron
availability.
Despite these limitations, the high levels of ferritin found
in our study were not unexpected. Ferritin synthesis is in-
creased by circulating cytokines known to be elevated in
sepsis (e.g. tumour necrosis factor alpha, interleukin-1). This
process is part of a complex mechanism that regulates iron
availability in the presence of an invading organism (2–5).
Because the rise in ferritin is thought to be a beneficial re-
sponse, we analyzed whether failure to increase ferritin to
a level above the level normally seen inflammation affected
outcome. Overall, failure to increase ferritin to 200 ng/mL
did not affect outcome.
The association between ferritin level and outcome in chil-
dren has not been reported before. We found that children
with ferritin >500 ng/mL had a 3-fold increase in risk of
death, and this association was independent of the PRISM
score. Our finding is consistent with previous reports in
adults showing an association between ferritin and sever-
ity of disease in sepsis (8). In our cohort, ferritin appears to
be an early prognostic indicator in children with sepsis. Fer-
ritin >500 ng/mL had a 64% sensitivity and 80% specificity
to predict mortality, and FI > 1.7 identified all children who
died, but with a lower specificity (58%). The areas under
the ROC curve for ferritin and FI were 0.73 and 0.74, re-
spectively. Because ferritin and PRISM were independently
associated with death, ferritin could be used in association
with current prognostic scores to improve predictive perfor-
mance in sepsis.
Finally, we have not evaluated the association between
ferritin and change in haemoglobin level. However, extrapo-
lation from adult data suggests that high levels of ferritin are
associated with an imbalance between the amount of iron
available for erythropoiesis and the amount of iron stored,
thereby increasing the susceptibility to anemia of chronic in-
fection and increasing the need for blood transfusion (15).
Whether this extrapolation is correct and whether it can be
reversed needs to be tested.

C 2007 The Author(s)/Journal Compilation 
C 2007 Foundation Acta Pædiatrica/Acta Pædiatrica 2007 96, pp. 1829–1831
Ferritin in children with sepsis
In summary, this study shows that ferritin is raised in chil-
dren with severe sepsis requiring intensive care and that it
may have potential as a prognostic indicator.
References
1. Theil EC. Ferritin: structure, gene regulation, and cellular
function in animals, plants, and microorganisms. Annu Rev
Biochem 1987; 56: 289–315.
2. Jurado RL. Iron, infections, and anemia of inflammation. Clin
Infect Dis 1997; 25: 888–95.
3. Bullen JJ, Rogers HJ, Spalding PB, Ward CG. Iron and
infection: the heart of the matter. FEMS Immunol Med
Microbiol 2005 43: 325–30.
4. Diaz J, Arribas JM, Vallina E, Maradona JA, Hevia C,
Blanco F. Acute-phase reactants in sepsi. Rev Clin Esp 1992;
191: 473–7.
5. Bullen J, Griffiths E, Rogers H, Ward G. Sepsis: the critical
role of iron. Microbes Infect 2000; 2: 409–15.
6. Darveau M, Denault AY, Blais N, Notebaert E. Bench-to-
bedside review: iron metabolism in critically ill patients. Crit
Care 2004; 8: 356–62.
7. Patteril MV, Davey-Quinn AP, Gedney JA, Murdoch SD,
Bellamy MC. Functional iron deficiency, infection and
systemic inflammatory response syndrome in critical illness.
Anaesth Intensive Care 2001; 29: 473–8.
8. Bobbio-Pallavicini F, Verde G, Spriano P, Losi R, Bosatra MG,
Braschi A, et al. Body iron status in critically ill patients:
significance of serum ferritin. Intensive Care Med 1989; 15:
171–8.
9. Sharkey RA, Donnelly SC, Connelly KG, Robertson CE,
Haslett C, Repine JE. Initial serum ferritin levels in patients
with multiple trauma and the subsequent development of
acute respiratory distress syndrome. Am J Respir Crit Care
Med 1999; 159: 1506–9.
10. Members of the American College of Chest Physicians/Society
of Critical Care Medicine Consensus Conference Committee.
American College of Chest Physicians/Society of Critical Care
Medicine Consensus Conference: definitions for sepsis and
organ failure and guidelines for the use of innovative therapies
in sepsis. Crit Care Med 1992; 20: 864–74.
11. Pollack M, Ruttiman U, Getson P. The Pediatric Risk of
Mortality (PRISM) score. Crit Care Med 1988; 16: 1110–6.
12. Leteurtre S, Martinot A, Duhamel A, Proulx F, Grandbastien
B, Cotting J, et al. Validation of the Paediatric Logistic Organ
Dysfunction (PELOD) score: prospective, observational,
multicentre study. Lancet 2003; 362: 192–7.
13. Brugnara C. Reference values in infancy and childhood. In:
Nathan DG, Orkin SH, editors. Nathan and Oski’s
hematology of infancy and childhood. 5th ed. Philadelphia,
PA: W.B. Saunders Company, 1998: xxii.
14. von Ahsen N, Muller C, Serke S, Frei U, Eckardt KU.
Important role of non diagnostic blood loos and blunted
erythropoietic response in anemia of intensive care patients.
Crit Care Med 1999; 27: 2630–9.
15. Weiss G. Modification of iron regulation by the inflammatory
response. Best Pract Res Clin Haematol 2005; 18 183–201.
1831