doi:10.1111/jpc.

13849

ORIGINAL ARTICLE

Trend and outcome of sepsis in children: A nationwide cohort

study
Chia-Hung Yo,1 Tzu-Chun Hsu,2 Meng-Tse Gabriel Lee,2 Lorenzo Porta,3 Po-Yang Tsou,4 Yu-Hsun Wang,4
Wan-Chien Lee,2 Szu-Ta Chen5,6,7,8 and Chien-Chang Lee;2,9 on behalf of National Taiwan University Hospital
Health Economics and Outcome Research Group
1
Department of Pediatric Emergency Medicine, Far Eastern Memorial Hospital, New Taipei City, 2Department of Emergency Medicine, National Taiwan
University Hospital, 6Department of Pediatrics, National Taiwan University Children’s Hospital, 7Graduate Institute of Toxicology, College of Medicine,
National Taiwan University, Taipei, Departments of 5Pediatrics, 9Emergency Medicine, and General Medicine, National Taiwan University Hospital Yun-Lin
Branch, Yunlin County, Taiwan, 3Department of Biomedical and Clinical Sciences, Luigi Sacco Hospital, University of Milan, Milan, Italy, 4Department of
Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland and 8Department of Epidemiology, Harvard T.H. Chan School of
Public Health, Boston, Massachusetts, United States

Aim: The aim of this study was to investigate the trend of incidence and outcome of paediatric sepsis in a population-based database.
Methods: Children with sepsis were identified from the 23 million nationwide health insurance claims database of Taiwan. Sepsis was defined by
the presence of single ICD-9 code for severe sepsis or septic shock or a combination of ICD-9 codes for infection and organ dysfunction. We ana-
lysed the trend of incidence, mortality and source of infection in three age groups: infant (28 days to 1 year), child (1–9 years) and adolescent
(10–18 years).
Results: From 2002 to 2012, we identified 38 582 paediatric patients with sepsis, of which 21.3% were infants, 52.8% were children and 25.8%
were adolescents. The incidence of sepsis was 336.4 cases per 100 000 population in infants, 3.3 times higher than in children (101.5/100 000
cases) and 7.3 times higher than in adolescents (46.2/100 000 cases). While sepsis incidence decreased from 598.0 to 336.4 cases per 100 000
people in the infant population, it remained relatively unchanged in children and adolescents. For 90-day mortality, there were significant
decreases in all three age groups (absolute decrease of 5.0% for infants, 3.7% for children and 14.4% for the adolescents). In the infant population,
we observed a decrease in the incidence of lower respiratory tract infections, while the incidence of urinary tract infections remained
unchanged.
Conclusions: The incidence and mortality of sepsis among paediatric patients have decreased substantially between 2002 and 2012, especially
among infants. The widespread use of Haemophilus influenzae and pneumococcal vaccines in infants could be a possible explanation.

Key words: children; incidence; mortality; sepsis.

What is already known on this topic What this paper adds

1 Sepsis remains among the top 10 leading causes of death in chil-
dren and adolescents.
2 Although sepsis constitutes an important burden for children,
published population-based epidemiologic studies on paediatric
sepsis are scarce.
1 Using a population-based database with linkage to a national
death certificate, we found a significant decrease in infantile sep-
sis incidence from 2002 to 2012.
2 The decreasing infantile sepsis trend coincides with decreasing
incidence in lower respiratory tract infection.
3 A possible explanation could be the widespread use of Haemo-
philus influenzae and pneumococcal vaccines in infants.

Sepsis remains among the top 10 leading causes of death in

Correspondence: Dr Chien-Chang Lee, Health Economic Outcomes
Research Group and Department of Emergency Medicine, National Taiwan
University Hospital No. 7, Chung Shan S. Road, Zhongzheng District, Taipei
100, Taiwan. Fax: +886 2 2356 5926; email: cclee100@gmail.com;
hit3transparency@gmail.com
Conflict of interest: None declared.
Accepted for publication 4 December 2017.
children and adolescents. 1,2 In fact, the 2013 Global Burden of
Disease Study estimated that 5 million children <5 years of
age die of sepsis each year, almost 20% of all deaths in chil-
dren younger than 5 years world-wide.3 In middle- to low-
income countries, sepsis even accounts for 60–80% of lost
lives in childhood. Even in the United States, the annual inci-
dence of severe sepsis is relatively high, approximately 3.0
cases per 1000 children, claiming almost 4500 young lives
each year.4–6

Journal of Paediatrics and Child Health (2018) 1

© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Trend and outcome of sepsis in children
Over the past 10 years, several studies using administrative
data have reported a rising trend in the incidence of adult sepsis
and a declining trend in the sepsis mortality.7–10 However, there
have only been a few nationally representative epidemiological
studies on sepsis in children, mainly based on survey database
or registries.4,11 Whether these survey databases accurately cap-
tured sepsis cases remain controversial. In fact, the population
estimates were possibly subjected to inaccurate estimations of
sampling probability and geographical representativeness. Thus,
the true incidence and outcome of paediatric sepsis may have
been poorly grasped by public health workers, politicians and
many health-care professionals. A population-based nationwide
estimate of the incidence and outcome of sepsis in children is
urgently needed to make a decision about the distribution of
health-care resources, assignment of research priorities and
benchmarking across geographical regions and hospital systems.
As the published epidemiological studies on paediatric sepsis
mainly focus on selected populations of North America and
Europe, an epidemiological study investigating the situation of
Asian countries is needed to fully understand the world
situation.
Taiwan is a developed country with a high gross domestic
product per capita and low infant mortality rate. On a purchasing
power parity basis, Taiwan’s gross domestic product per capita in
2016 has a world ranking of 27th, which is slightly behind Swe-
den and United States, according to the Central Intelligence
Agency Fact Sheet. In 2004, Taiwan was ranked 11th among
high-income countries for crude infant mortality, and only six
countries (Sweden, Finland, Czech Republic, Belgium, Austria
and Germany) had infant mortality rates statistically significantly
lower than that of Taiwan. The current study aimed to perform
an epidemiology study on paediatric sepsis using a true
population-based nationwide administrative database. Further-
more, this database allows linkage to the national death certifi-
cate database, providing a robust estimate of sepsis mortality.
Using this database, we sought to determine the change in inci-
dence and mortality of sepsis in children over time.
Methods
Data source
This study is a cohort review of a prospectively collected national
health insurance (NHI) database. Taiwan’s NHI is a single-payer
health insurance programme, enrolling 99.8% of the entire
23 million Taiwanese population. This study used the entire data-
base, maintained by a government-run data co-ordination centre,
to enrol all children from 28 days of age to 18 years admitted for
sepsis from January 2002 to December 2012. The NHI claims
database contains administrative data for all health insurance
enrolees, whether on an outpatient or an inpatient basis. Data
available for analysis include demographics (age, gender, resi-
dence area, insurance premium); ICD-9-CM (International Clas-
sification of Diseases, 9th revision, Clinical Modification) codes
for diagnoses and procedures; admission source (emergency
department, nursing home or transfer from outside hospital);
admission and discharge dates; discharge disposition (home,
nursing home, transfer to another hospital and discharge against
medical advice, dead); and billing data for medications,
2 Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
C-H Yo et al.
radiological imaging studies, laboratory tests and supplies charged
to each patient. Death data were obtained from the linked
national death certificate registry database. The institutional
review board of National Taiwan University Hospital has
approved this study. As this is an electronic database study using
anonymised patients, requirement for informed consent was
waived.
Definitions
We identified all patients aged 18 years or younger from the dis-
charge records of emergency department or hospitalisation. Sepsis
was defined as the presence of a diagnosis of infection and at least
one acute organ dysfunction in any discharge diagnosis. For patients
with multiple hospitalisation records for sepsis in a given year, we
used the first eligible emergency department or hospitalisation
record as the index admission date. Operationally, children were
defined as having sepsis if any of the following three criteria was
met: (i) an ICD-9 code for severe sepsis (995.92); (ii) an ICD-9 code
for septic shock (785.52); or (iii) an ICD-9 code of infection plus at
least one ICD-9 code of organ dysfunction (modified Angus cri-
teria).12 The ICD-9 CM codes used for identification of the source of
infection (e.g. pneumonia, urinary tract infection, meningitis and
bacteraemia) are listed in Appendix S1 (Supporting Information).
Organ dysfunction was defined by a combination of ICD-9-CM and
procedure codes (Appendix S2, Supporting Information). We col-
lected information on chronic comorbid illnesses during the index
admission for infant group and in the 1-year period before the
index admission for young children and adolescents. Chronic illness
was categorised into 12 common diseases in this population, includ-
ing diabetes, chronic pulmonary disease, chronic heart diseases,
asthma, cerebral palsy, human immunodeficiency virus (HIV) infec-
tion/acquired immune deficiency syndrome (AIDS), chronic renal
disease, chronic liver disease, congenital gastrointestinal diseases,
solid tumour, haematological malignancy and autoimmune disor-
ders. We ascertained the post-sepsis survival status by discharge
records as well as linked death certificates.
Statistical analysis
We divided the patient cohort into three age subgroups to reflect
definitions for infants (28 days to 1 year), pre-school and school-
age children (1–9 years) and pre-adolescence children and
adolescents (10–18 years). We compared the demographic char-
acteristics and prevalence of key comorbid conditions, number
and type of organ dysfunctions, utilisation of invasive procedures
and outcomes across the study period. We presented continuous
variables by mean  standard deviation and categorical variables
by frequency and percentage. We defined the incidence of sepsis
as the number of each event divided by the number of enrolees
during each year, and we defined all-cause mortality as the num-
ber of deaths within 30 days divided by the total number of cases.
Then, we compared the longitudinal changes of incidence and
mortality of sepsis and septic shock using graphical methods. To
clarify whether the change of sepsis incidence was due to the
change in the population age and gender structure over time, we
calculated the age–gender standardised incidence and mortality
rate from 2002 to 2012 using the age and gender distribution in
2001 as the standard. We assumed that all cases of severe sepsis
C-H Yo et al.
in Taiwan were collected in this complete nationwide database,
and there was no possibility for chance variation; therefore, we
did not perform statistical tests to verify our results. All analyses
were performed using SAS version 9.4 (Cary, NC, USA).
Compliance with ethical standards
The Ethics Committee of National Taiwan University approved
the study. Exemption of informed consent was conceded due to
the retrospective study of an electronic database. Patients were
codified, and no personal information was recorded.
Results
We identified 38 582 patients <18 years of age who developed
sepsis between 2002 and 2012, of which 21.3% were infants
(aged 28 days to 1 year), 52.8% were young children (aged
1–9 years) and 25.8% were pre-adolescents or adolescents (aged
10–18 years) (Fig. 1). The overall incidence of paediatric sepsis
was determined to be 161.3 per 100 000 population, with a 30-
day mortality of 4.4%.
Baseline characteristics
Table 1 summarises the characteristics of sepsis patients in the
three age groups. In all three age groups, male was the predomi-
nant gender. The proportion of male patients decreased with age,
from 60.06% in the infant group to 54.7% in young children
Fig. 1 Assembling of study cohort.
Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Trend and outcome of sepsis in children
group and 55.56% in the adolescent group. The infant group had
the highest burden of chronic pulmonary disease, chronic heart
disease and congenital gastrointestinal disease. The children
group had a higher prevalence of asthma, cerebral palsy and
chronic liver disease. The adolescent group had the highest prev-
alence of diabetes, HIV infection, chronic renal disease, solid
tumour, haematological malignancies and autoimmune diseases.
The respiratory tract was the predominant source of infection in
all three age groups, followed by genitourinary tract, unclassified
bacterial, intra-abdomen and skin or skin structure infections.
Urinary tract infections had a higher prevalence in the infant and
adolescent groups, while intra-abdominal and skin or skin struc-
ture infections increased with age. For sepsis-related organ/sys-
tem dysfunction, acute respiratory failure was the most common
organ dysfunction observed in three age groups. Hepatic dysfunc-
tion was the second most common organ dysfunction in infants,
while haematological system dysfunction was the second most
common cause of organ dysfunction in young children and ado-
lescents. The prevalence of renal and metabolic system dysfunc-
tions increased with age. The adolescent patients had the highest
mortality, followed by infants and young children. The infants,
however, stayed longer in the intensive care unit and wards than
young children or adolescents.
Incidence of sepsis and septic shock
From 2002 to 2012, the incidence of sepsis in infants showed an
absolute decrease of 261.6 cases per 100 000 population, from
3
Trend and outcome of sepsis in children C-H Yo et al.

Table 1 Characteristics of sepsis hospitalisation in children aged <18 years, national health insurance claims database of Taiwan, 2002–2012

Infants (28 days to 1 year), Children (1–9 years), Adolescents (10–17 years),
Characteristic n = 8, 215 n = 20 369 n = 9998

Male gender, n (%) 4934 (60.06) 11 136 (54.67) 5455 (54.56)

Age, mean (SD) — 3.79  2.51 13.83  2.30
Underlying comorbidity, n (%)
Diabetes 1 (0.01) 19 (0.09) 105 (1.05)
Chronic pulmonary disease 251 (3.06) 149 (0.73) 25 (0.25)
Chronic heart diseases 1540 (18.75) 1383 (6.79) 241 (2.41)
Asthma 307 (3.74) 2129 (10.45) 416 (4.16)
Cerebral palsy 289 (3.52) 1759 (8.64) 764 (7.64)
HIV/AIDS 0 (0.00) 3 (0.01) 3 (0.03)
Chronic renal disease 86 (1.05) 159 (0.78) 185 (1.85)
Chronic liver disease 122 (1.49) 487 (2.39) 96 (0.96)
Congenital gastrointestinal diseases 267 (3.25) 233 (1.14) 27 (0.27)
Any tumour 89 (1.08) 1081 (5.31) 974 (9.74)
Haematological malignancy 39 (0.47) 620 (3.04) 569 (5.69)
Autoimmune disorder 263 (3.20) 878 (4.31) 1004 (10.04)
Source of infection, n (%)
Lower respiratory tract infection 4023 (48.97) 10 216 (50.15) 3735 (37.36)
Upper respiratory tract infection 2217 (26.99) 8770 (43.06) 3069 (30.7)
Genitourinary tract infection 1364 (16.6) 1665 (8.17) 1489 (14.89)
Intra-abdominal infection 175 (2.13) 589 (2.89) 462 (4.62)
Skin and skin structure infection 189 (2.3) 588 (2.89) 451 (4.51)
Biliary tract infection 4 (0.05) 54 (0.27) 35 (0.35)
Systematic fungal infection 284 (3.46) 496 (2.44) 383 (3.83)
Other bacterial disease 624 (7.6) 1062 (5.21) 601 (6.01)
System dysfunction, n (%)
Acute respiratory failure 5382 (65.51) 8051 (39.53) 4111 (41.12)
Central nervous system dysfunction 505 (6.15) 1779 (8.73) 665 (6.65)
Haematologic system dysfunction 1364 (16.6) 5514 (27.07) 1855 (18.55)
Hepatic system dysfunction 1509 (18.37) 4836 (23.74) 1628 (16.28)
Acute kidney injury 186 (2.26) 634 (3.11) 769 (7.69)
Outcome
30-day all-cause mortality, n (%) 504 (6.14) 913 (4.48) 643 (6.43)
Length of hospital stay, median (25–75 14 (6,33) 8 (4,20) 11 (5,31)
percentile)
Length of ICU stay, median (25–75 7 (3,15) 5 (2,11) 5 (2,11)
percentile)

AIDS, acquired immune deficiency syndrome; HIV, human immunodeficiency virus; ICU, intensive care unit; SD, standard deviation.

Fig. 2 Annual trend of sepsis incidence,

stratified by three different age groups.
( ), 28 days to 1 year; ( ),
1–9 years; ( ), 10–17 years.

4 Journal of Paediatrics and Child Health (2018)

© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
C-H Yo et al.
598.0 to 336.4 cases per 100 000 population, with an annual
decrease of 23.8 cases per 100 000 population (Fig. 2). Sepsis
incidence, however, remained relatively unchanged for children
and adolescents over the study period. Sepsis is exceedingly more
common in infants (336.4/100 000), with rates being 3.3-fold
higher during infancy compared with childhood (46.2/100 000)
and 7.3-fold higher compared with adolescence (101.5/100 000).
The sepsis incidence among infants did not decrease uniformly
every year. In fact, infantile sepsis incidence declined substan-
tially from 2002 to 2006 and then decreased slowly from 2006
to 2012.
Mortality of sepsis
There were substantial decreases in 30-day mortality (Fig. 3a)
and 90-day mortality (Fig. 3b) of sepsis in all three age groups of
children during the study period. For 30-day mortality, sepsis
mortality decreased similarly across all age groups. The absolute
decrease was 3.0% for the infant group (from 8.5 to 5.5%),
2.7% for children group (from 6.1 to 3.3%) and 4.6% for the
adolescent group (from 9.1 to 6.5%). For 90-day mortality, the
decrease in sepsis mortality for adolescents outpaced that of
Fig. 3 (a) Annual trend of 30-day mortal-
ity of sepsis, stratified by three different
age groups. (b) Annual trend of 90-day
mortality of sepsis, stratified by three dif-
ferent age groups. ( ), 28 days to
1 year; ( ), 1–9 years; ( ),
10–17 years.
Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Trend and outcome of sepsis in children
infants and children. The absolute decrease of 90-day mortality
was 5.0% for infants (from 12.8 to 7.9%), 3.7% for children
(from 6.5 to 2.8%) and 14.4% for the adolescents (from 22.4 to
8.0%). The total number of paediatrics with sepsis that died
within 90 days also decreased (418 deaths in 2002 to 213 deaths
in 2012).
Sources of infection
Trends of the two major sources of infection in children, lower
respiratory tract infection and urinary tract infection, are
shown in Figure 4. Compared with children and adolescent
groups, the infant group had a higher incidence of both lower
respiratory tract infections and urinary tract infections. For
the infant group, the incidence of lower respiratory tract infec-
tions decreased substantially, while the incidence of urinary
tract infections remained relatively unchanged during the
study period. For the children and adolescent groups, the inci-
dence of lower respiratory tract infections and urinary tract
infections remained relatively unchanged during the study
period.
5
Trend and outcome of sepsis in children
Discussion
In this nationwide, population-based, epidemiological study, we
found that the incidence and mortality of sepsis among paediatric
patients have decreased substantially during the study period.
The incidence of sepsis was the highest in infants, 7.3 times
higher than in adolescents and 3.3 times higher than in children.
The major sources of infection were respiratory tract and genito-
urinary tract infections. The incidence and mortality of lower
respiratory tract infections experienced a substantial decrease,
while those of the urinary tract infections remained relatively
unchanged.
In general, our study conforms to the previously published epi-
demiological studies on paediatric sepsis. In fact, we found that
the incidence of sepsis was disproportionately high in infants,
corroborating the findings in previous studies using different
databases and different ICD-9CM coding schemes for sepsis
patient identification. Furthermore, it has been reported that
infants have 3.2–9.2-fold higher incidence than older children.4,5
These variations in the incidence of sepsis in infants and older
children have been attributed to different underlying chronic dis-
eases; age-specific differences in the immune, cardiovascular and
metabolic responses to sepsis; infecting organisms, sites of infec-
tion; organ dysfunction and perinatal events.13,14 We observed
6 Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
C-H Yo et al.
Fig. 4 (a) Annual trend of incidence of
lower respiratory tract infection, strati-
fied by three different age groups.
(b) Annual trend of incidence of genito-
urinary tract infection, stratified by three
different age groups. ( ), 28 days to
1 year; ( ), 1–9 years; ( ),
10–17 years.
that after infancy, epidemiological borders between children and
adolescents are less distinct; however, age-related differences
continue to occur, especially regarding underlying chronic dis-
eases. The children group has a higher prevalence of asthma,
cerebral palsy and chronic liver disease, but the adolescent group
has the higher prevalence of diabetes, HIV infection, chronic
renal disease, tumour, haematological malignancies and autoim-
mune diseases. Furthermore, the mortality and trend we
observed were similar to those of North America.4,11,15 In fact, in
North America, the in-hospital mortality rate for paediatric sepsis
decreased from 10.6% in 2004 to 6.8% in 2012.15 In the same
period, in Taiwan, we observed a similar decrease (from 11.4 to
6.2%) in the 90-day mortality rate for paediatric sepsis. Improved
supportive care, early evidence-based bundle care and introduc-
tion of new vaccines are believed to have improved the outcome
of paediatric sepsis.16–20
The roll out of Haemophilus influenzae type B (HiB) vaccine and
the pneumococcal vaccine during the study period might have
had a substantial effect on the incidence and mortality trend of
paediatric sepsis. The HiB vaccine was available in Taiwan since
1996, but it was not covered by the government health insur-
ance. According to Taiwan’s Centres for Disease Control, the vac-
cination rate of HiB was less than 10% in 2002 but has gradually
increased to 50% in 2005.20 The universal Hib vaccination has
C-H Yo et al.
been implemented since 2010, and now, the vaccination rate has
reached 97%. The universal Hib vaccination has been associated
with a more than fivefold decrease in the incidence of invasive
HiB from 2000 to 2012.21 Likewise, the pneumococcal vaccine
for children, which was made commercially available in 2000 in
the United States, has been found to decrease the pneumococcal
disease in infants under 2 years of age by 58–66%.16–18 In Tai-
wan, the pneumococcal vaccine was commercially available in
2005, and healthy infants could be vaccinated with co-payment.
However, the vaccine was provided for free for infants with
immunocompromised conditions, those in vulnerable socio-
economic status and those with inadequate health-care
resources.22 From 2008 to 2012, the incidence of invasive pneu-
mococcal diseases decreased by 36.4% for infants under 2 years
of age, and this was attributed to the increase in pneumococcal
vaccination.23 Thus, the aforementioned significant decrease in
infantile sepsis incidence and mortality over time, especially the
discrepancy between respiratory tract infections and urinary tract
infections, might be related to childhood vaccination. Additional
large population-based epidemiological studies comparing infants
who were vaccinated and not might be warranted to evaluate the
widespread effects of vaccines.
Strength and limitation
The population-based nature of the study is a major strength and
allowed us to report a true sense of sepsis prevalence and out-
comes in ways that are lacking from other administrative data-
base reports. It is very difficult to obtain high-quality data for an
entire country, especially in Asia, and this is only possible
because of the existence of a single-payer insurance system and
good record keeping. Another strength is the robustness in the
death outcome ascertainment for each included participant. In
fact, studies based on hospital records may suffer from underesti-
mation of mortality due to early discharge or transfer of severe
patients prior to death. In addition, the ICD-9 coding practices
were not deeply affected by policy changes in Taiwan during the
study periods. Recent studies showed that the published guidance
for sepsis codes by the Centres of Medicare and Medicaid Services
in 2003 and the introduction of medical severity diagnosis-related
group (MS-DRG) systems for sepsis in 2007 led to the upcoding
practice in the US hospitals and thus the spurious increase of sep-
sis incidence and pseudo-improvement in the sepsis outcome.24,25
Knowledge on the Surviving Sepsis Campaign guideline is widely
disseminated in Taiwan, but there was no additional incentive
from the government-run NHI during the study period. Thus, the
systematic upcoding practice of sepsis in children was not likely
in Taiwan. In addition, the dramatic difference in the trend of
incidence and mortality between the three age groups provides a
strong piece of evidence against the pseudo-improvement of pae-
diatric sepsis outcome due to upcoding.
There are also several limitations for this study. First, results of
our study are derived from administrative datasets, which rely on
correct data coding, and lack any microbiology information.
Therefore, misclassification of sepsis could be possible in this and
other similar studies. The nature of misclassification is believed to
be random, and previous studies find the administrative dataset
to yield similar paediatric sepsis incidence as clinical dataset.26
Second, the Angus sepsis identification strategy that we followed
Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Trend and outcome of sepsis in children
has been reported to be more inclusive than other identification
strategy and may overestimate the incidence of sepsis.15 Instead
of just including patients with the diagnosis of severe sepsis and
septic shock, we included all possible sources of infections that
can lead to organ dysfunction. The Angus strategy was originally
designed for adult patients. There are important developmental
differences in the host response to infection at the organ system
level between children and adults. For example, children with
sepsis are less amenable to fluid resuscitation and require vaso-
pressor at an earlier stage. The prevalent use of non-invasive
nasal continuous positive airway pressure ventilation in infants
may avoid the use of mechanical ventilation in some children.
Therefore, use of the vasopressor treatment code to define shock
and the mechanical ventilation procedure code to define respira-
tory failure may thus capture a different sepsis population in chil-
dren. Whether the use of Angus coding strategy in children
would lead to under- or over-estimation of sepsis incidence in
children requires further clarification. A validation study that
compares the ICD-9-defined sepsis with clinical diagnosis in a
large electronic medical record database is urgently required.
Third, the nationwide database management team does not allow
finer age categorisation as the small number of patients in each
year’s subcategories may expose some participants’ identities and
violate the privacy protection principles. We were also unable to
study the sepsis epidemiology in the neonate group because the
neonates have another unique insurance registration system,
which was not available. Fourth, Taiwan is an Asian country that
is economically well developed, and results from Taiwan cannot
easily be compared to many less-developed Asian countries that
have no national insurance scheme. Lastly, although there was
no obvious coding policy or reimbursement rate changes for sep-
sis during the study period, the growing acceptance of a clinical
‘sepsis syndrome’ during this time period may also have inflated
the incidence and decreased the mortality of paediatric sepsis.
Conclusions
Using a large nationwide database and combination coding strategy
for sepsis identification, we observed a different trend of sepsis inci-
dence in the Taiwanese paediatric population. In fact, unlike what
is observed in most western countries, the overall sepsis incidence
has declined significantly over time. The infant group had the high-
est burden of sepsis but also experienced the greatest decrease in
the sepsis burden. We confirmed with previous findings that the
mortality of sepsis in infant, children and adolescent groups has
decreased over time. This study also documented that the trend of
lower respiratory tract infections decreased substantially in infants,
while the trend of urinary tract infections remained unchanged.
The challenges ahead are to understand the key determinants of
these results, especially the widespread effects of vaccines. As more
and more children survive diseases that were previously fatal, such
as leukaemia, these comorbid conditions will have an even greater
impact on both the incidence and outcome of sepsis in paediatric
patients in the future.
Acknowledgements
We thank the staff of the Core Labs at the Department of Medical
Research in National Taiwan University Hospital for technical
7
Trend and outcome of sepsis in children
support, Medical Wisdom Consulting Group for technical assis-
tance in statistical analysis and National Taiwan University Hospi-
tal Health Economics and Outcome Research Group for advice on
study design. This study is supported by the Taiwan National Sci-
ence Foundation Grant NSC 102-2314-B-002-131-MY3; Taiwan
National Ministry of Science and Technology Grants MOST 104-
2314-B-002-039-MY3 and MOST 106-2811-B-002-048; and Far
Eastern Memorial Hospital Grant FEMH-2016-C-028. No funding
bodies had any role in study design, data collection and analysis,
decision to publish or preparation of the manuscript.
References
1 The Global Burden of Disease Child and Adolescent Health Collabora-
tion, Kassebaum N, Kyu HH et al. Child and adolescent health from
1990 to 2015: Findings from the global burden of diseases, injuries,
and risk factors 2015 study. JAMA Pediatr. 2017; 171: 573–92.
2 Global Burden of Disease Pediatrics Collaboration, Kyu HH, Pinho C
et al. Global and national burden of diseases and injuries among chil-
dren and adolescents between 1990 and 2013: Findings from the
global burden of disease 2013 study. JAMA Pediatr. 2016; 170:
267–87.
3 Abubakar II, Tillmann T, Banerjee A. Global, regional, and national
age-sex specific all-cause and cause-specific mortality for 240 causes
of death, 1990-2013: A systematic analysis for the Global Burden of
Disease Study 2013. Lancet 2015; 385: 117–71.
4 Hartman ME, Linde-Zwirble WT, Angus DC, Watson RS. Trends in the
epidemiology of pediatric severe sepsis. Pediatr. Crit. Care Med.
2013; 14: 686–93.
5 Watson RS, Carcillo JA, Linde-Zwirble WT, Clermont G, Lidicker J,
Angus DC. The epidemiology of severe sepsis in children in the
United States. Am. J. Respir. Crit. Care Med. 2003; 167: 695–701.
6 Watson RS, Carcillo JA. Scope and epidemiology of pediatric sepsis.
Pediatr. Crit. Care Med. 2005; 6 (3 Suppl.): S3–5.
7 Fleischmann C, Scherag A, Adhikari NK et al. Assessment of global
incidence and mortality of hospital-treated sepsis. Current estimates
and limitations. Am. J. Respir. Crit. Care Med. 2016; 193: 259–72.
8 Bouza C, Lopez-Cuadrado T, Saz-Parkinson Z, Amate-Blanco JM. Epi-
demiology and recent trends of severe sepsis in Spain: A nationwide
population-based analysis (2006–2011). BMC Infect. Dis. 2014; 14:
717–30.
9 Stevenson EK, Rubenstein AR, Radin GT, Wiener RS, Walkey AJ. Two
decades of mortality trends among patients with severe sepsis: A
comparative meta-analysis. Crit. Care Med. 2014; 42: 625–31.
10 Gaieski DF, Edwards JM, Kallan MJ, Carr BG. Benchmarking the inci-
dence and mortality of severe sepsis in the United States. Crit. Care
Med. 2013; 41: 1167–74.
11 Ruth A, McCracken CE, Fortenberry JD, Hall M, Simon HK, Hebbar KB.
Pediatric severe sepsis: Current trends and outcomes from the Pedi-
atric Health Information Systems database. Pediatr. Crit. Care Med.
2014; 15: 828–38.
12 Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J,
Pinsky MR. Epidemiology of severe sepsis in the United States: Analy-
sis of incidence, outcome, and associated costs of care. Crit. Care
Med. 2001; 29: 1303–10.
8 Journal of Paediatrics and Child Health (2018)
© 2018 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
C-H Yo et al.
13 Piekkala P, Kero P, Erkkola R, Sillanpaa M. Perinatal events and neona-
tal morbidity: An analysis of 5380 cases. Early Hum. Dev. 1986; 13:
249–68.
14 Wheeler DS, Wong HR, Zingarelli B. Pediatric sepsis – Part I: “Children
are not small adults!” Open Inflamm. J. 2011; 4: 4–15.
15 Balamuth F, Weiss SL, Neuman MI et al. Pediatric severe sepsis in US
children’s hospitals. Pediatr. Crit. Care Med. 2014; 15: 798–805.
16 Centers for Disease Control and Prevention. Direct and indirect
effects of routine vaccination of children with 7-valent pneumococcal
conjugate vaccine on incidence of invasive pneumococcal disease –
United States, 1998–2003. MMWR Morb. Mortal. Wkly. Rep. 2005;
54: 893–7.
17 Herz AM, Greenhow TL, Alcantara J et al. Changing epidemiology of
outpatient bacteremia in 3- to 36-month-old children after the intro-
duction of the heptavalent-conjugated pneumococcal vaccine.
Pediatr. Infect. Dis. J. 2006; 25: 293–300.
18 Centers for Disease Control and Prevention. Invasive pneumococcal dis-
ease in children 5 years after conjugate vaccine introduction – Eight states,
1998–2005. MMWR Morb. Mortal. Wkly. Rep. 2008; 57: 144–8.
19 Dellinger RP, Levy MM, Rhodes A et al. Surviving sepsis campaign:
International guidelines for management of severe sepsis and septic
shock, 2012. Intensive Care Med. 2013; 39: 165–228.
20 Bréchot N, Luyt C-E, Schmidt M et al. Venoarterial extracorporeal
membrane oxygenation support for refractory cardiovascular dys-
function during severe bacterial septic shock. Crit. Care Med. 2013;
41: 1616–26.
21 Centers for Disease Control. Invasive Haemophilus influenzae Type b
(Hib). Taipei: Centers for Disease Control; 2017. Available from: http://
www.cdc.gov.tw/english/info.aspx?treeid=e79c7a9e1e9b1cdf&nowtreeid=
e02c24f0dacdd729&tid=C9B5C59280B1867E [accessed 16 June 2017].
22 Wei S-H, Chiang C-S, Chen C-L, Chiu C-H. Pneumococcal disease and
use of pneumococcal vaccines in Taiwan. Clin. Exp. Vaccine Res.
2015; 4: 121–9.
23 Chiang CS, Chen YY, Jiang SF et al. National surveillance of invasive
pneumococcal diseases in Taiwan, 2008–2012: Differential temporal
emergence of serotype 19A. Vaccine 2014; 32: 3345–9.
24 Gohil SK, Cao C, Phelan M et al. Impact of policies on the rise in sep-
sis incidence, 2000–2010. Clin. Infect. Dis. 2016; 62: 695–703.
25 Walkey AJ, Lagu T, Lindenauer PK. Trends in sepsis and infection
sources in the United States. A population-based study. Ann.
Am. Thorac. Soc. 2015; 12: 216–20.
26 Weiss SL, Parker B, Bullock ME et al. Defining pediatric sepsis by dif-
ferent criteria: Discrepancies in populations and implications for clini-
cal practice. Pediatr. Crit. Care Med. 2012; 13: e219–26.
Supporting Information
Additional Supporting Information may be found in the online
version of this article at the publisher’s web-site:
Appendix S1. Codes associated with infection and organ
dysfunction.
Appendix S2. Codes associated with organ dysfunction.