European Journal of Pediatrics

[Link]

ORIGINAL ARTICLE

Assessment of clinical outcome of children with sepsis

outside the intensive care unit
Federica Zallocco 1 & Patrizia Osimani 1 & Ines Carloni 1 & Vittorio Romagnoli 1 & Silvia Angeloni 1 & Salvatore Cazzato 1,2

Received: 12 January 2018 / Revised: 7 August 2018 / Accepted: 10 September 2018

# Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract
In 2016, in order to identify adult patients with sepsis who are likely to have poor outcomes, the Third International Consensus
Definitions Task Force introduced a new bedside index, called the quick Sepsis-related Organ Failure Assessment (qSOFA) score.
However, these new criteria have not been validated in the pediatric population. In this study, we sought to assess the qSOFA
score for children with sepsis, who are being treated outside the pediatric intensive care units. The qSOFA criteria were revised
and applied to a study population of 89 pediatric patients with sepsis, admitted in a pediatric tertiary referral center from 2006 to
2016. The analysis of prognostic performance of qSOFA score for the prediction of severe sepsis showed a sensitivity of 46%
(95% CI, 27–67%), a specificity of 74% (95% CI, 62–85%), a positive predictive value of 43% (95% CI, 34–52%), and a negative
predictive value of 77% (95% CI, 71–82%). The area under ROC curve for qSOFA score ≥ 2 was 0.602 (95% CI 0.492–0.705).
Conclusion: The qSOFA score showed a low accuracy to identify children in the pediatric ward at risk for severe sepsis.
Clinical tools are needed to facilitate the diagnosis of impending organ dysfunction in pediatric infection outside of the ICU.

What is Known:
• One of the major challenges for clinicians is to identify and recognize children with sepsis and impending organ dysfunction, in the emergency and in
the pediatric department.
• In 2016, members of the Sepsis-3 task force proposed qSOFA, an empirically derived score using simple clinical criteria, to assist clinicians in
identifying adult patients with sepsis at risk for poor outcome.
What is New:
• qSOFA demonstrated insufficient clinical value to be recommended as a screening tool for pediatric sepsis outside ICU.
• D-dimer level and blood glucose may be useful biomarkers to identify children at risk for severe sepsis.

Keywords Pediatric sepsis . qSOFA . Impending organ dysfunction . Pediatric infection . Severe sepsis

Abbreviations PICU Pediatric intensive care units

BP Blood pressure qSOFA Quick Sepsis-related Organ Failure Assessment
ICU Intensive care units PSCC Pediatric Sepsis Consensus Congress
IQR Interquartile range ROC Receiver operating characteristic
NPV Negative predictive values SD Standard deviation
PPV Positive predictive values SE Sensibility

Communicated by Nicole Ritz

* Salvatore Cazzato Vittorio Romagnoli

[Link]@[Link] [Link]@[Link]
Federica Zallocco Silvia Angeloni
[Link]@[Link] [Link]@[Link]
Patrizia Osimani 1
Pediatrics and Infectious Disease Unit, G. Salesi Children’s Hospital,
[Link]@[Link] Ancona, Italy
2
Ines Carloni Pediatrics and Infectious Disease Unit, Department of Mother and
[Link]@[Link] Child Health, G. Salesi Children’s Hospital, Via Corridoni 11,
60123 Ancona, Italy
 Eur J Pediatr

SIRS Systemic inflammatory response syndrome or younger. This finding was confirmed in a recent prospec-
SOFA Sequential [Sepsis-related] Organ Failure tive multicentric cohort study of children admitted in ICU for
Assessment infections [28]. The authors reported that the qSOFA score,
SP Specificity adapted according to age-specific cutoff, showed moderate
SS Severe sepsis accuracy to identify infected children at risk for worse
outcome.
However, Sepsis-3-based criteria have never been evaluat-
Introduction ed in pediatric wards, outside intensive care settings.
Moreover, to assist with early recognition of organ dysfunc-
Sepsis is a clinical syndrome resulting from a dysregulated tion, others pediatric sepsis screening scores have been devel-
systemic inflammatory response to infection, associated oped but they were validated only in the PICU setting and not
with a poor outcome, including high short-term mortality. recognized outside the PICU, where simpler tools, which in-
A recent systematic review and meta-analysis estimate an clude clinical variables that are easily obtained in the first
incidence of 3.0 million cases of sepsis in neonates and 1.2 hour, are needed [17, 24].
million cases in children younger than 20 years annually. In this study, we sought to assess the qSOFA score for
Mortality ranged from 1 to 5% for sepsis and 9 to 20% for children with sepsis, outside the PICU. In addition, we sought
severe sepsis (SS) [8]. to identify indicators that may be used for risk assessment in
It is quite complicated for clinicians to diagnose and children. Finally, we provided etiological factors and the clin-
assess pediatric sepsis due to the lack of specific signs ical characteristics of sepsis in children.
and symptoms and the rapid deterioration that occurs when
compensation fails. Children who develop signs of severe
sepsis with organ dysfunction, and especially those who Materials and methods
develop septic shock, are at the highest risk of life-
threatening and fatal complications. Study population
In 2005, the Pediatric Sepsis Consensus Congress (PSCC)
[10] agreed that the definition of pediatric sepsis requires the Medical records of children admitted to the Tertiary Center of
presence of systemic inflammatory response syndrome Pediatrics and Infectious Disease Unit, G. Salesi Children’s
(SIRS) in a child with suspected or proven infection. In the Hospital (Ancona, Italy) from January 1, 2006, to December
light of the high mortality associated with sepsis, the impor- 31, 2016, were retrospectively reviewed. The G. Salesi
tance of early detection and intervention was emphasized. Children’s Hospital is one of the largest pediatric healthcare
However, SIRS criteria have been criticized for their poor institutes in central and southern Italy and is a reference center
specificity because these are also met by children with non- for the care of children with infectious disease.
infectious diseases; furthermore, even infected children meet- The International Classification of Diseases, Ninth
ing these criteria often have no organ dysfunction [28, 30]. Revision, Clinical Modification (ICD-9-CM) discharge diag-
In 2016, a joint SCCM/European Society of Intensive Care nosis codes for sepsis (995.91); neonatal sepsis (771.81); se-
Medicine task force developed new definitions of adult sepsis vere sepsis (995.92), septic shock (785.52), and septicemia
and clinical criteria to identify patients likely to have this (038.9) were used to generate lists of potential cases. A target
deadly disorder (Sepsis-3 criteria) [32]. Sepsis, for adult pa- samples size of 129 records was selected.
tients, is now defined as a Blife-threatening organ dysfunction Data were collected on age, sex, history, clinical manifes-
caused by a dysregulated host response to infection^ and it is tations, laboratory investigations, and microbiological tests.
identified by an increase of, at least, two points in the Clinical and laboratory markers were taken in the first 24 h
Sequential [Sepsis-related] Organ Failure Assessment after sepsis onset, except for D-dimer that was collected at
(SOFA) score in patients with a suspicion of infection. The their worst value at any time during hospital stay. The onset
quick SOFA (qSOFA) score, a surrogate for SOFA in settings time for sepsis was defined as the earliest time during a hos-
in which all components of SOFA are not routinely measured, pital encounter when the patient meets criteria for sepsis.
was proposed to help identify patients with suspected infec- Children with sepsis were diagnosed according to the
tion who are being treated outside critical care units and likely criteria of SIRS, sepsis, SS, and septic shock by the PSCC
to develop complications of sepsis. in 2005 [10].
Recently, a pediatric version of the SOFA score (pSOFA) Neonates that were less than 72 h old were excluded from
was adapted and examined in a single-center retrospective the analysis.
study [21]. The authors found that the pSOFA score showed The criteria of qSOFA score [32] were revised and applied
an excellent discrimination for in-hospital mortality in a gen- for children, it included three parameters: respiratory rate >
eral pediatric intensive care units (PICU) population, 21 years 95th percentile for the age (1 point); Glasgow Coma Scale
Eur J Pediatr

score of less than 15 (1 point); decrease in systolic blood Results

pressure (BP) < 5th percentile for the age (1 point). The
qSOFA scores were defined by applying age-specific cutoffs Description of the study population
for respiratory rate and systolic blood pressure, as per the
PSCC definitions [10, 11]. Over the study period, 129 patient files were identified.
For each recruited patient, the three components of the Twenty-nine (23%) neonates were excluded because they
qSOFA were collected at their highest level in the first 24 h. were affected by early onset sepsis, 11 (9%) children were
The primary outcome was defined as development of se- excluded because they did not fulfill the criteria of sepsis,
vere sepsis (SS). In addition, secondary outcomes, defined as leaving 89 (69%) included for the final analysis (Fig. 1).
ICU admission, days of hospitalization, antibiotic treatment, Baseline characteristics of the study population are summa-
and duration of fever, were stratified for qSOFA score and SS. rized in Table 1.
At least one comorbidity was present in 5 (6%) patients:
three children were affected by hematology-oncology dis-
Statistical analysis
eases (Hodgkin disease in complete remission, acute lym-
phoblastic leukemia in maintenance therapy, and cyclic
Statistical comparisons were assessed for discrete and contin-
neutropenia in therapy with granulokine, respectively);
uous variables as appropriate. All non-normally distributed
one patient by congenital heart disease (septal atrial defect
variables were expressed as median and interquartile range
in hemodynamic compensation), one patient by HIV infec-
(IQR). Categorical variables were expressed as absolute num-
tion (CD4 count on admission: 509 cell/mmc). Only one
ber and percentage. Data non-normally distributed were ana-
patient died and he was the patient affected by cyclic neu-
lyzed using the Mann–Whitney U test, chi-square test (or
tropenia. Twenty-six patients (29%) fulfilled the criteria of
Fisher’s exact test) to compare continuous and dichotomous
severe sepsis.
variables respectively. To assess the performances of the
qSOFA in order to predict SS, we calculated diagnostic per-
formances: sensitivity (SE), specificity (SP), and negative and
Etiological factors
positive predictive values (NPV and PPV) for a qSOFA score
of 2 or higher. A receiver operating characteristic (ROC) curve
In all patients, a bacterial culture, at least, was performed.
was constructed and the corresponding area under the ROC
A pathogen was identified in blood culture in 23/70 (33%),
curve (AUROC) calculated. Furthermore, performance of
urine culture in 9/48 (19%), cerebrospinal fluid in 23/51
qSOFA to predict admission in ICU was compared with the
(45%), stool culture in 2/22 (9%), and swab culture in 7/12
definition of SS, established by the PSCC in 2005 [10]. The
(58%) of patients. The most commonly isolated bacteria
ROC curves were also obtained for the D-dimer and blood
are described in Table 2. A bacterial infection was docu-
glucose values as predictors of SS. To determine the most
mented from sterile sites in 42 (47%) patients. Viral diag-
appropriate cutoff value, which has the maximum sensitivity
nostics was performed in 37 (42%) patients. An exclusive
and specificity pair, Youden’s index was calculated.
viral infection was found in 4 (5%) patients. A combined
All statistical analyses were 2-tailed and p value < 0.05 was
infection (bacterial and viral) was identified in 2 (2%) pa-
considered significant.
tients. A patient was affected by fungal sepsis (Candida
For the lack of previous studies on qSOFA in children with
albicans). In 37 (42%) of 89 patients, sepsis was diagnosed
sepsis, the sample size was calculated based on the results of a
but no pathogen was isolated. The sites of infection are
previous study on adult data [9] looking for an in-hospital
described in Table 3.
mortality rate of at least 10%. The sample size was determined
to be 88 subjects assuming a power of 80% and confidence
interval of 95%. The study period was extended to achieve the
Clinical and laboratoristic characteristics stratified
stated patient number.
for severe sepsis
All analyses were performed using the statistical software
package SPSS software Version 20.0.
As shown in Table 4, children baseline characteristics with
non-SS were compared to those with SS, according to the
Ethical clearance criteria of PSCC [10]. Children with SS showed more frequent
tachypnea and abnormal heart rate as clinical sign of onset
The local ethics committee, CER (Regional Ethics Committee respect to patients with non-SS (p = 0.004 and p = 0.01, re-
of Marche), approved this single-center retrospective study spectively). The levels of D-dimer and blood glucose were
(reference number 2017-0064 OR). The research was carried significantly higher in patients with SS compared to non-SS
out in compliance with the Helsinki Declaration. (p = 0.001 and p = 0.003, respectively) (Table 4).
 Eur J Pediatr

Fig. 1 Flow diagram showing

study enrollment

Performance of qSOFA score for the prediction was found to be 0.729 (95% CI, 0.618–0.823), with a SE of
of severe sepsis 68% (95%CI, 47–85%), SP of 78% (95% CI, 64–88%), PPV
of 59% (95%CI, 39–77%), and NPV of 84% (95% CI, 71–
For patients with a qSOFA score < 2, the percentage of SS was 93%).
23% vs 43% with a qSOFA score of 2 or higher (p = 0.1). The The ROC analysis revealed that the serum glucose level of
AUROC curve for qSOFA was 0.602 (95% CI 0.492–0.705) 130 mg/dl was the best cutoff value for predicting SS. The
(Fig. 2). For the prediction of SS, qSOFA score had a SE of AUROC was found to be 0.720 (95% CI, 0.603–0.819), with
46% (95% CI, 27–67%), SP of 74% (95% CI, 62–85%), PPV a SE of 52% (95% CI, 30–74%), SP of 88% (95% CI, 77–
of 43% (95% CI, 34–52%), and NPV of 77% (95% CI, 71– 96%), PPVof 65% (95% CI, 38–86%), and NPVof 82% (95%
82%). CI, 70–91%).

Performance of D-dimer and blood glucose Secondary outcomes stratified for severe sepsis
for the prediction of severe sepsis and qSOFA

As a marker of SS, the most appropriate D-dimer cutoff value, Patients with SS, compared to non-SS, needed longer hospi-
calculated using a ROC curve, was 2568 μg/l. The AUROC talization, antibiotic treatment, and higher frequency of ICU

Table 1 Characteristic of study

population Characteristics Number (percentage) Missing data (%)

Males/females 58/31 (65%) 0 (0%)

Age, months 12 (1–75) 0 (0%)
Median (IQR)
Mean respiratory rate > 2 SD 48 (54%) 3 (3%)
Tachycardia/bradycardia 21 (24%) 4 (4%)
Temperature > 38.5 °C or < 36 °C 73 (82%) 2 (2%)
Frequency of organ dysfunction
Cardiovascular 7 (8%) N/A
Respiratory 15 (17%) N/A
Neurologic 53 (60%) N/A
Hematologic 13 (15%) N/A
Renal 2 (2%) N/A
Hepatic 12 (13%) N/A

N/A not applicable

Eur J Pediatr

Table 2 Bacteria isolated by cultures from 89 patients sepsis at risk for poor outcome [32]. Apparently, the use of
Type of pathogens isolated Number (percentage) qSOFA score in the emergency department setting resulted in
greater prognostic accuracy for in-hospital mortality than did
Streptococcus pneumoniae 7 (8%) either SIRS or SS [9]. Recently, a pediatric version of the
Streptococcus group B 4 (4%) SOFA score (pSOFA) was adapted and analyzed in critically
Streptococcus group A 1 (1%) ill children in PICU with confirmed or suspected infection.
Streptococcus group F 1 (1%) The study showed that pSOFA score had excellent discrimi-
Meningococcus spp 8 (9%) nation for in-hospital mortality, with an AUC of 0.94 (95% CI,
Meningococcus Y/W135 3 (3%) 0.92–0.95) [21]. However, the authors used a retrospective
Meningococcus B 2 (2%) design and data from a single PICU therefore limiting gener-
Escherichia coli 7 (8%) alizability of the findings.
Staphylococcus spp. 6 (7%) To our knowledge the current study is the only pediatric
Pseudomonas aeruginosa 2 (2%) study that assesses the validity of qSOFA score among pedi-
Enterococcus spp. 2 (2%) atric patients presenting outside the PICU. The results of our
Klebsiella pneumoniae 2 (2%) study show that qSOFA score has a low prognostic accuracy
Others 5 (6%) to predict SS. Compared to criteria of SS, defined by the
PSCC in 2005 [10], qSOFA has also a worse discriminative
value for predicting ICU admission. These findings are in
admission. No significant differences in secondary end points contrast with studies on adult population [6, 9]. Freund et al.
were found between patients with qSOFA score ≥ 2 compared [9] evaluated the predictive validity of qSOFA in a prospective
to qSOFA < 2 (Table 5). study of adult patients with suspected infection presenting to
the Emergency Department. The qSOFA performed better
Sensitivity analyses of qSOFA score and severe sepsis than SIRS criteria and SS to predict in-hospital mortality. In
for the prediction of ICU admission a recent study, Churpek et al. [6] found that qSOFA was more
accurate than SIRS for predicting in-hospital mortality and
The percentage of children admitted in ICU was 17% (5/29) for ICU transfer; however, the same study showed a lower prog-
patients with a qSOFA score ≥ 2 and 5% (3/60) for qSOFA score nostic performance among patients with suspected infection
< 2 (p = 0.07). The qSOFA score ≥ 2 had a SE of 63% (95% CI, outside the ICU. In another report, Raith et al. [26] evaluated
25–92%), SP of 71% (95% CI, 60–81%), PPVof 18% (95% CI, the predictive validity of qSOFA in a retrospective analysis on
6–37%), and NPV of 95.0% (95% CI, 86–99%). The AUROC adults and they found that the discriminatory performance of
for the SS was 0,750 (95% CI, 0.646–0.836), for qSOFA score it qSOFA in considering the outcome of hospital mortality or
was 0.669 (95% CI, 0.560–0.765) (Fig. 3). ICU length of stay of 3 or more days was not higher than SIRS
performance. We speculate that the low accuracy of qSOFA in
our pediatric population is due to the fact that SS can be
present without a qSOFA score ≥ 2. In fact, in our study, pa-
Discussion
tients with a qSOFA < 2 were affected by different forms of
organ dysfunction in addition to those defined by the qSOFA
In 2016, members of the Third International Consensus
score, such as hematologic and hepatic dysfunctions (Table 5).
Definitions for Sepsis and Septic Shock (Sepsis-3) task force
In addition, arterial hypotension, as confirmed by our study
proposed qSOFA, an empirically derived score using simple
(Table 4), remains a late sign of pediatric septic shock [4].
clinical criteria, to assist clinicians in identifying patients with
Therefore, qSOFA does not seem to be a useful and reliable
Table 3 Site of infection from 89 patients tool for the pediatrician outside the ICU. The greatest concern
is that this new sepsis score, in the pediatric population, un-
Site of infection Number (percentage) derestimates intervention at early stages of sepsis when the
syndrome may be most treatable.
Central nervous system infection 37 (42%)
Epidemiological studies concerning pediatric sepsis are few
Urinary tract infection 9 (10%)
and most of them involved children admitted in the pediatric
Respiratory system infection 8 (9%)
ICU. The Sepsis Prevalence, Outcomes, and Therapies
Gastrointestinal system infection 4 (5%)
(SPROUT) study collected data on a large number of children
Infection with unknown site 30 (34%)
with SS admitted to a PICU throughout 2013–2014 at 128
Miscellaneous infectionsa 10 (11%)
sites in 26 countries and it estimated a point prevalence of
a
Includes skin and soft tissue infection (1 patient) and throat or mouth 8.2% in PICU [36]. A recent multicenter prospective study
infection (seven patients) reported an incidence of neonatal and pediatric culture-
 Eur J Pediatr

Table 4 Clinical and laboratory characteristics, collected on admission*, stratified for non-severe sepsis and severe sepsis, according to the criteria of
International Pediatric Sepsis Consensus Conference 2005 [10, 11]

Characteristics Non-severe sepsis (63) Missing data (%) Severe sepsis (26) Missing data (%) p value

Sex N (%)
Males/females 39/24 (62%) 0 (0%) 19/7 (73%) 0 (0%) 0.3
Age, months [median (IQR)] 8 (1–71) 0 (0%) 17 (2–111) 0 (0%) 0.4
Respiratory rate > 2SD N (%) 27 (43%) 2 (3%) 21 (81%) 0 (0%) 0.004
Abnormal heart ratea N (%) 8 (13%) 3 (5%) 13 (50%) 0 (0%) 0.01
T > 38.5 °C or < 36 °C N (%) 51 (81%) 1 (2%) 22 (85%) 0 (0%) 0.8
Hypotensionb N (%) 1 (2%) 6 (10%) 2 (8%) 1 (4%) 0.5
GCS < 15 N (%) 45 (71%) 0 (0%) 20 (77%) 0 (0%) 0.8
WBC(/mmc) median (IQR) 19,610 (7540–22,700) 0 (0%) 15,295 (6325–23,555) 0 (0%) 0.7
Leukocyte count abnormalitiesc 51 (81%) 0 (0%) 19 (73%) 0 (0%) 0.6
Serum sodium (mEq/l) median (IQR) 136 (134–138) 3 (5%) 137 (133–140) 1 (4%) 0.6
Serum potassium (mEq/l) median (IQR) 4.4 (4.1-5.0) 3 (5%) 3.7 (3.4-4.4) 1 (4%) 0.001
Serum calcium (mg/dl) median (IQR) 9.2 (8.6-9.7) 5 (8%) 8.5 (8.0-9.2) 2 (8%) 0.007
Blood glucose (mg/dl) median (IQR) 104 (83–120) 11 (17%) 133 (104–180) 5 (19%) 0.003
D-dimer (μg/l) median (IQR) 1032 (471–2506) 9 (14%) 3444 (1463–6070) 1 (4%) 0.001

RR respiratory rate, T temperature, GCS Glasgow Coma Scale

*D-dimer and blood pressure collected to their worst level.
a
Tachycardia, defined as a mean heart rate > 95th percentile for age in the absence of external stimulus, chronic drugs, or painful stimuli; or otherwise
unexplained persistent elevation over a 0.5- to 4-h time period or for children < 1 yr. old bradycardia, defined as a mean heart rate < 10th percentile for
age in the absence of external vagal stimulus, blocker drugs, or congenital heart disease; or otherwise unexplained persistent depression over a 0.5-h time
period.
b
Systolic BP < 5th percentile for age.
c
Leukocyte count elevated or depressed for age (not secondary to chemotherapy-induced leukopenia) or 10% immature neutrophils.

proven bacterial sepsis of 25.1 per 100,000 children per year, rather than SIRS was used as the criterion and mortality in-
with an average 30-day in-hospital mortality of 7%. Organ creased to 17% when organ dysfunction was present [1].
dysfunction was present in 39% of episodes. Sepsis incidence In our study, SS accounted in about one-third (29%) of
decreased by two-thirds (8.3 per 100,000) if organ dysfunction cases of sepsis admitted in 10 years in a pediatric ward, and
this data confirms that sepsis remains a critical problem in
children even outside the ICU. The wideness gap in hospital
mortality (1–26%) reported in literature [1, 8, 25] may be
associated to the methods used to identify sepsis (i.e., different
administrative codes and different definitions). The low hos-
pital mortality in our study (1%) may be due to the character-
istics of the studied population selected in pediatric ward,
outside the ICU and hemato-oncology unit. The only child
who died was affected by cyclic neutropenia and it is known
that mortality in children with comorbidity is higher [36].
In our study, children with SS showed more frequent
tachypnea and abnormal heart rate as clinical sign of onset
than patients without SS, which could be attributed to the
pathophysiological mechanisms of organ system dysfunc-
tions. Respiratory rate is increased in response to tissue hyp-
oxia and metabolic acidosis. It is also a sign of primary focus
of infection in lungs, or early acute respiratory distress syn-
drome [20]. Tachycardia occurs early in response to falling
cardiac output and is the most significant physical findings
Fig. 2 ROC curves of qSOFA score for prediction of severe sepsis in septic shock [14].
Eur J Pediatr

Table 5 Secondary outcomes stratified for severe sepsis and qSOFA and frequency of organ dysfunction

qSOFA score (n) Severe sepsis (n)

Outcome < 2 (60) ≥ 2 (29) p value No (63) Yes (26) p value

ICU admission N (%) 3 (5%) 5 (17%) 0.071 2 (3%) 6 (23%) 0.007

Days of hospitalization [median (IQR)] 12 (10–21) 14 (11–46) 0.805 12 (10–20) 15 (12–39) 0.022
Duration of antibiotic treatment (days) [median (IQR)] 11 (8–15) 12 (4–16) 0.262 10 (7–14) 14 (11–23) 0.002
Duration of fever (hours) [median (IQR)] 48 (24–93) 48 (24–16) 0.254 48 (24–96) 60 (9–174) 0.6
Organ dysfunction
Respiratory N (%) 7 (12%) 8 (28%) 0.143 0 (0%) 15 (58%) < 0.001
Cardiovascular N (%) 2 (3%) 5 (17%) 0.075 0 (0%) 7 (27%) < 0.001
Neurologic N (%) 33 (55%) 20 (69%) 0.455 32 (51%) 21 (81%) 0.010
Hematologic N (%) 6 (10%) 7 (24%) 0.179 1 (2%) 12 (46%) < 0.001
Hepatic N (%) 7 (12%) 5 (17%) 0.765 2 (3%) 10 (38%) < 0.001
Renal N (%) 1 (2%) 1 (3%) 1.000 0 (0%) 2 (8%) 0.1

According to previous studies [7, 19, 29] levels of D-dimer in children after cardiac surgery, necrotizing enterocolitis
were higher in patients with SS than in those with non-SS and (NEC), brain injury, and in mechanically ventilated children
a significant correlation was found between D-dimer levels with bronchiolitis [5, 13, 31, 35, 38].
and days of hospitalization. Consumption and depletion of Our findings showed significant differences in calcium and
endogenous coagulation proteins occur frequently in patients potassium levels between SS and non-SS. Septic patients are
with SS, and the depletion of anticoagulant and fibrinolytic particularly at risk of hypocalcemia and hypokalemia. In par-
factors contributes to the microvascular fibrin deposition as- ticular, increased mortality and longer stay in ICU were re-
sociated with organ damage [2, 12, 15, 18]. ported in patients with low ionized calcium [3, 23, 33, 34].
Children with SS had higher blood glucose levels com- This study has some limitations which have to be pointed
pared to children with non-SS (Table 4). Hyperglycemia was out.
found to be associated with increased morbidity and mortality First, the impact of our findings must be tempered by the
rates for both adults and children admitted to ICU [16, 22, 37]. relatively small sample size and the low number of children
Hyperglycemia was described as a marker of poor prognosis with severe sepsis that required admission to ICU. However, a
recent study of Schlapbach et al. [28], involving children ad-
mitted to ICU due to infection, has obtained similar perfor-
mance of the qSOFA.
Second, as suggest by Rhee et al. [27], the assignment of
sepsis diagnoses is extremely variable among clinicians be-
cause there is no gold standard diagnostic test for sepsis.
Therefore, the use of the ICD9-CM codes to select the popu-
lation in the current retrospective study may have led to an
overestimation of CNS infections and to an underestimation
of respiratory infections and of the mortality for sepsis, limit-
ing the generalizability of our findings.
Finally, data collected retrospectively means that analysis
relied on medical records for all information and data could be
incomplete. Indeed, many patients did not have blood lactate
measurement, thus the latter was excluded from analysis,
resulting in a possible misclassification in the septic shock
category. The highest value of D-dimer level was recorded
on different days among patients and the trend was not ana-
lyzed: this could have biased the results to higher D-dimer
levels in SS group.
Fig. 3 ROC curves of qSOFA score and severe sepsis for prediction of In conclusion, large-scale study has not been performed to
admission in ICU refine pediatric sepsis screening tools. One of the major
 Eur J Pediatr

challenges for clinicians is to identify and recognize children Carneiro B (2017) Prognostic accuracy of Sepsis-3 criteria for in-
hospital mortality among patients with suspected infection present-
with sepsis and impending organ dysfunction, in the pre-
ing to the emergency department. JAMA 317:301–308
hospital setting, in the emergency and in the pediatric depart- 10. Goldstein B, Giroir B, Randolph A (2005) International pediatric
ment. Contrary to adult patients, our data did not confirm the sepsis consensus conference: definitions for sepsis and organ dys-
accuracy of the qSOFA score to stratify sepsis severity among function in pediatrics. Pediatr Crit Care Med 6:2–8
11. Goldstein B, Giroir B, Randolph A (2005) Letters to the editor.
pediatric patients. However, our study suggests, that D-dimer
Pediatr Crit Care Med 6:500–501
level and blood glucose may be useful biomarkers to identify 12. Green J, Doughty L, Kaplan SS, Sasser H, Carcillo JA (2002) The
children at risk for severe sepsis. tissue factor and plasminogen activator inhibitor type-1 response in
pediatric sepsis-induced multiple organ failure. Thromb Haemost
Authors’ contributions All authors had full access to all of the data for 87:218–223
this study and take responsibility for the integrity of the data and the 13. Hall NJ, Peters M, Eaton S, Pierro A (2004) Hyperglycemia is
accuracy of the data analysis. associated with increased morbidity and mortality rates in neonates
with necrotizing enterocolitis. J Pediatr Surg 9:898–901
14. Khilnani P, Singhi S, Lodha R, Santhanam I, Sachdev A, Chugh K,
Compliance with ethical standards Jaishree M, Ranjit S, Ramachandran B, Ali U, Udani S (2010)
Pediatric sepsis guidelines: summary for resource-limited countries.
Conflict of interest The authors declare that they have no conflict of Indian J Crit Care Med 14:41–52
interest. 15. Kidokoro A, Iba T, Fukunaga M, Yagi Y (1996) Alterations in
coagulation and fibrinolysis during sepsis. Shock 5:223–228
Ethical approval This retrospective study was approved by the Regional 16. Klein GW, Hojsak JM, Schmeidler J, Rapaport R (2008)
Ethics Committee of Marche (reference number 2017-0064 OR). Hyperglycemia and outcome in the pediatric intensive care unit. J
Pediatr 153:379–384
17. Leteurtre S, Duhamel A, Salleron J, Grandbastien B, Lacroix J,
Informed consent For this type of study, formal consent is not required.
Leclerc F, Groupe Francophone de Réanimation et d’Urgences
Pédiatriques (2013) PELOD-2: an update of the pediatric logistic
organ dysfunction score. Crit Care Med 41:1761–1773
References 18. Levi M, Van der Poll T, Ten Cate H, Van Deventer SJ (1997) The
cytokine-mediated imbalance between coagulant and anticoagulant
mechanisms in sepsis and endotoxaemia. Eur J Clin Investig 27:3–9
1. Agyeman PK, Schlapbach LJ, Giannoni E, Stocker M, Posfay-
19. Hu L, Zhu Y, Chen M, Li X, Lu X, Liang Y, Tan H (2016)
Barbe KM, Heininger U, Schindler M, Korten I, Konetzny G,
Development and validation of a disease severity scoring model
Niederer-Loher A, Kahlert CR (2017) Epidemiology of blood
for pediatric sepsis. Iran J Public Health 45:875
culture-proven bacterial sepsis in children in Switzerland: a
20. Mace SE (2001) The pediatric patient with acute respiratory failure:
population-based cohort study. Lancet Child Adolesc Health 1:
clinical diagnosis and pathophysiology. Pediatr Emerg Med Rep 6:
124–133
21–32
2. Asakura H, Ontachi Y, Mizutani T, Kato M, Ito T, Saito M,
21. Matics TJ, Sanchez-Pinto LN (2017) Adaptation and validation of a
Morishita E, Yamazaki M, Suga Y, Miyamoto KI, Nakao S
pediatric sequential organ failure assessment score and evaluation
(2001) Depressed plasma activity of plasminogen or alpha2 plas-
of the Sepsis-3 definitions in critically ill children. JAMA Pediatr
min inhibitor is not due to consumption coagulopathy in septic
171:e172352
patients with disseminated intravascular coagulation. Blood
22. McCowen K, Malhotra A, Bistrian B (2001) Stress-induced hyper-
Coagul Fibrinolysis 12:275–281
glycemia. Crit Care Clin 17:107–124
3. Barron R, Freebairn R (2010) Electrolyte disorders in the critically 23. Müller B, Becker KL, Kränzlin M, Schächinger H, Huber PR,
ill. Anaesth Intensive Care Med 11:523–528 Nylen ES, Snider RH, White JC, Schmidt-Gayk H, Zimmerli W,
4. Biban P, Gaffuri M, Spaggiari S, Zaglia F, Serra A, Santuz P (2012) Ritz R (2000) Disordered calcium homeostasis of sepsis: associa-
Early recognition and management of septic shock in children. tion with calcitonin precursors. Eur J Clin Investig 30:823–831
Pediatr Rep 4:e13 24. Patton L, Young V (2017) Effectiveness of provider strategies for
5. Branco RG, Tasker RC (2007) Glycemic level in mechanically the early recognition of clinical deterioration due to sepsis in pedi-
ventilated children with bronchiolitis. Pediatr Crit Care Med 8: atric patients: a systematic review protocol. JBI Database Syst Rev
546–550 Implement Rep 15:76–85
6. Churpek MM, Snyder A, Han X, Sokol S, Pettit N, Howell MD, 25. Pawar A, Raut A, Kalrao V, Jacob J, Godha I, Thomas R (2016)
Edelson DP (2017) Quick sepsis-related organ failure assessment, Etiology and clinical outcomes of neonatal and pediatric sepsis.
systemic inflammatory response syndrome, and early warning Arch Pediatr Infect Dis 4:e33602
scores for detecting clinical deterioration in infected patients outside 26. Raith EP, Udy AA, Bailey M, McGloughlin S, MacIsaac C,
the intensive care unit. Am J Respir Crit Care Med 195:906–911 Bellomo R, Pilcher DV (2017) Prognostic accuracy of the SOFA
7. Dhainaut JF, Shorr AF, Macias WL, Kollef MJ, Levi M, Reinhart score, SIRS criteria, and qSOFA score for in-hospital mortality
K, Nelson DR (2005) Dynamic evolution of coagulopathy in the among adults with suspected infection admitted to the intensive
first day of severe sepsis: relationship with mortality and organ care unit. JAMA 317:290–300
failure. Crit Care Med 33:341–348 27. Rhee C, Gohil S, Klompas M (2014) Regulatory mandates for
8. Fleischmann-Struzek C, Goldfarb DM, Schlattmann P, sepsis care reasons for caution. N Engl J Med 370:1673–1676
Schlapbach LJ, Reinhart K, Kissoon N (2018) The global bur- 28. Schlapbach LJ, Straney L, Bellomo R, MacLaren G, Pilcher D
den of paediatric and neonatal sepsis: a systematic review. (2018) Prognostic accuracy of age-adapted SOFA, SIRS,
Lancet Respir Med 6:223–230 PELOD-2, and qSOFA for in-hospital mortality among children
9. Freund Y, Lemachatti N, Krastinova E, Van Laer M, Claessens YE, with suspected infection admitted to the intensive care unit.
Avondo A, Occelli C, Feral-Pierssens AL, Truchot J, Ortega M, Intensive Care Med 44:179–188
Eur J Pediatr

29. Schwameis M, Steiner MM, Schoergenhofer C, Lagler H, Buchtele 34. Steele T, Kolamunnage-Dona R, Downey C, Toh CH, Welters I
N, Jilma-Stohlawetz P, Boehm T, Jilma B (2015) D-dimer and (2013) Assessment and clinical course of hypocalcemia in critical
histamine in early stage bacteremia: a prospective controlled cohort illness. Crit Care 17:R106
study. Eur J Inter Med 26:782–786 35. Verhoeven JJ, Hokken-Koelega AC, den Brinker M, Hop WC, van
30. Scott HF, Deakyne SJ, Woods JM, Bajaj L (2015) The prevalence Thiel RJ, Bogers AJ, Helbing WA, Joosten KF (2011) Disturbance
and diagnostic utility of systemic inflammatory response syndrome of glucose homeostasis after pediatric cardiac surgery. Pediatr
vital signs in a pediatric emergency department. Acad Emerg Med Cardiol 32:131–138
22:381–389 36. Weiss SL, Fitzgerald JC, Pappachan J, Wheeler D, Jaramillo-
31. Sharma D, Jelacic J, Chennuri R, Chaiwat O, Chandler W, Bustamante JC, Salloo A, Singhi SC, Erickson S, Roy JA, Bush
Vavilala MS (2009) Incidence and risk factors for perioperative JL, Nadkarni VM (2015) Global epidemiology of pediatric severe
hyperglycemia in children with traumatic brain injury. Anesth sepsis: the sepsis prevalence, outcomes, and therapies study. Am J
Analg 108:81–89 Respir Crit Care Med 191:1147–1157
32. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, 37. Wintergerst KA, Buckingham B, Gandrud L, Wong BJ, Kache S,
Annane D, Bauer M, Bellomo R, Bernard GR, Chiche JD, Wilson DM (2006) Association of hypoglycemia, hyperglycemia,
Coopersmith CM, Hotchkiss RS (2016) The third international and glucose variability with morbidity and death in the pediatric
consensus definitions for sepsis and septic shock (Sepsis-3). intensive care unit. Pediatrics 118:173–179
JAMA 315:801–810 38. Yates AR, Dyke PC, Taeed R, Hoffman TM, Hayes J, Feltes TF,
33. Song SK, Karl IE, Ackerman JL, Hotchkiss RS (1993) Increased Cua CL (2006) Hyperglycemia is a marker for poor outcome in the
intracellular Ca2+: a critical link in the pathophysiology of sepsis? postoperative pediatric cardiac patient. Pediatr Crit Care Med 7:
Proc Natl Acad Sci 90:3933–3937 351–355