Current Medical Research & Opinion
6, 2014, 1065–1069
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 07/31/15
For personal use only.
Vol. 30, No.
0300-7995
doi:10.1185/03007995.2014.887003
Brief report
Gastrointestinal bleeding in hospitalized children
in the United States
Chaitanya Pant
Department of Medicine, Kansas University Medical
Center, Kansas City, KS, USA
Senthilkumar Sankararaman
Department of Pediatrics, Louisiana State University
Health Sciences Center, Shreveport, LA, USA
Abhishek Deshpande
Department of Medicine, Medicine Institute, Cleveland
Clinic, Cleveland, OH, USA
Mojtaba Olyaee
Department of Medicine, Kansas University Medical
Center, Kansas City, KS, USA
Michael P. Anderson
Department of Biostatistics and Epidemiology,
University of Oklahoma Health Sciences Center,
Oklahoma City, OK, USA
Thomas J. Sferra
Department of Pediatrics, Case Western Reserve
University School of Medicine, UH Rainbow Babies &
Children’s Hospital, Cleveland, OH, USA
Address for correspondence:
Thomas J. Sferra MD, Martin and Betty Rosskamm
Chair in Pediatric Gastroenterology, Chief, Pediatric
Gastroenterology & Nutrition, Rainbow Babies and
Children’s Hospital, 11100 Euclid Avenue, Suite 737,
Cleveland, OH 44106, USA.
Tel: +1 216 844 1765;
thomas.sferra@uhhospitals.org
Keywords:
Adolescent – Child – Epidemiology – Gastrointestinal
bleeding – Incidence – Outcomes – Pediatric
Accepted: 14 January 2014; published online: 4 February 2014
Citation: Curr Med Res Opin 2014; 30:1065–9
! 2014 Informa UK Ltd www.cmrojournal.com
Article ST-0409.R1/887003
All rights reserved: reproduction in whole or part not permitted
Abstract
Objective:
To investigate the epidemiology of GI bleeding in hospitalized children in the United States.
Methods:
Data were obtained from the Healthcare Cost and Utilization Project Kids’ Inpatient Database, Agency for
Healthcare Research and Quality for the year 2009. The data were weighted to generate national-level
estimates.
Results:
There were 23,383 pediatric discharges with a diagnosis of GI bleeding accounting for 0.5% of all
discharges. Children with a GI bleed compared to those without were more likely to be male (54.5% vs.
45.8%; P50.001), older (children 11 years; 50.8% vs. 38.7%; P50.001), and admitted to a teaching
hospital (70.5% vs. 56.4%; P50.001). Children 11–15 years of age had the highest incidence of GI
bleeding (84.2 per 10,000 discharges) and children less than 1 year of age the lowest (24.4 per 10,000
discharges). The highest incidence of GI bleeding was attributable to cases coded as blood in stool (17.6 per
10,000 discharges) followed by hematemesis (11.2 per 10,000 discharges). Those with a GI bleed had a
higher co-morbid burden (12.3% vs. 2.3%; P50.001) and severity of illness (40.1% vs. 14.5%;
P50.001). The highest mortality rates associated with GI bleeding were observed in cases with
intestinal perforation (8.7%) and esophageal perforation (8.4%). GI bleeding was independently
associated with a higher risk of mortality (aOR 1.68, CI 1.53–1.84).
Conclusions:
Our results describe the epidemiology of GI bleeding in hospitalized children within the United States. We
found a substantial risk of mortality attributable to GI bleeding in this patient population. Our study is limited
by the exclusion of non-hospitalized children, the reliance on ICD-9-CM codes and the absence of
longitudinal follow up of patients.
Background
Gastrointestinal (GI) bleeding in a child is a common reason for referral to a
pediatric gastroenterologist1. Despite this, data regarding the epidemiology of GI
bleeding in the pediatric population within North America are limited, consist-
ing of a few single-center studies many of which were published over 20 years
ago. Moreover, it is difficult to draw generalizable conclusions from these studies
as they differ in patient care setting, type of GI bleeding investigated, and overall
study objective. In a retrospective cohort study published in 1979 from a pedi-
atric tertiary referral center, hematemesis accounted for 0.2% of 316,020 emer-
gency department visits2. In 1994, Teach and Fleisher reported that of more than
40,000 visits to the Boston Children’s Emergency Department, 0.3% of children
had a presenting complaint of rectal bleeding3. In a 1992 prospective study,
GI bleeding in children Pant et al. 1065
 Current Medical Research & Opinion Volume 30, Number 6 June 2014
63 (6.4%) upper GI bleeds were detected among 984 chil-
dren admitted to a pediatric ICU over a 55 week period4.
Higher rates (up to 25%) of upper GI bleeding in children
also from the ICU setting have been reported in two smal-
ler prospective studies5,6.
Currently, there are no large, multi-center studies that
have investigated the epidemiology of GI bleeding in chil-
dren. Our objective was to define the recent incidence and
associated mortality of GI bleeding in hospitalized chil-
dren using a United States nationwide administrative
database.
Materials and methods
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 07/31/15
Data source
We used the 2009 Kids’ Inpatient Database (KID) for the
year 2009 for this study. The KID is part of the Healthcare
Cost and Utilization Project (HCUP) sponsored by the
United States Agency for Healthcare Research and
Quality. Individual hospitalizations in which the patient
age is 20 or less at the time of admission are maintained in
the KID as unique entries (the unit of observation is the
inpatient stay record; individual patients cannot be
For personal use only.
tracked across multiple hospitalizations). Discharge
weights are provided to calculate national level estimates
for the extracted data. For 2009, the KID contains
7,370,203 weighted pediatric discharges from 4121 com-
munity hospitals (as defined by the American Hospital
Association; nonfederal, short-term general and specialty
hospitals accessible by the general public) from 44 states.
Diagnoses and ICD-9-CM
We used the International Classification of Diseases,
Ninth Revision, Clinical Modification (ICD-9-CM) diag-
nostic codes to identify GI bleeding in the KID. Our use of
ICD-9-CM codes were identical to, and the classification
of GI bleeding is based on, the methods of Zhao and
Encinosa7 who reported hospitalizations for GI bleeding
in adults from the HCUP Nationwide Inpatient Sample
(NIS); the KID and NIS are from the same family of data-
bases. We categorized GI bleeding into three major classes:
upper GI bleeding, lower GI bleeding, and unspecified GI
bleeding. Unspecified GI bleeding is comprised of two dis-
tinct ICD-9-CM codes: 578.9 (hemorrhage of the GI tract;
unspecified) and 578.1 (blood in stool). ICD-9-CM 578.1
can refer to hematochezia (generally lower GI bleeding) or
melena (generally upper GI bleeding).
Variables
Patient characteristics included age, gender, income,
health insurance coverage, and comorbid conditions.
Race was not included since it is not reported by all
1066 GI bleeding in children Pant et al.
states in the KID. We categorized patients into the follow-
ing age groups: 51, 1–5, 6–10, 11–15, and 16–20 years of
age. Normal, uncomplicated, in-hospital births were
excluded from our study. Median household income was
classified according to income quartiles (1st–4th) per
HCUP documentation based on patient residence.
Insurance status was grouped into private, public
(Medicaid/Medicare), and none (no pay, self-pay, or
other). Hospital characteristics included teaching status
(teaching versus non-teaching), location (urban versus
rural) and region (Northeast, Midwest, South, or West).
Severity of illness (minor, moderate, major, or extreme)
was categorized per the hospitalization’s All Patient
Refined Diagnosis-Related Group (APR-DRG) code8,9.
The examined comorbid conditions consisted of the 29
disease states contained within the Elixhauser comorbidity
index10. This is a widely used index in which higher scores
indicate a greater comorbid disease burden11,12. We also
extracted patient information pertaining to in-hospital
mortality, hospital charges and hospital length of stay
(each of which is coded within the database). Data were
analyzed and are presented using weights provided in the
KID to derive national level estimates.
Statistical methods
Statistical analyses were performed using SAS version 9.2
(SAS institute, Cary, NC, USA). Categorical data are
reported as frequencies and percentages and the chi-
square test was used for comparing differences between
these variables. Rates of GI bleeding are reported as per
10,000 hospital discharges. The threshold for significance
for these analyses was P50.05. To assess the independent
effect of GI bleeding on mortality a multiple variable logis-
tic regression was performed with the presence or absence
of death as the dichotomous outcome variable. The mul-
tiple variable logistic regression model was built by select-
ing variables that had a logical association with mortality
(GI bleeding, demographics [age, sex, hospital region,
location and teaching status, median household income,
and payer type], severity of illness categorized as per the
APR-DRG code and the comorbid burden determined by
the Elixhauser scale). We felt that given the very large
number of patients to number of variables under study,
overfitting the model was not a problem. We tested all
between-variable estimated correlation coefficients and
determined that multicollinearity was not a problem.
Adjusted odds ratio (aOR) and 95% confidence intervals
(CIs) are reported.
Results
In 2009, there were 23,383 pediatric hospitalizations with
a diagnosis of GI bleeding. Table 1 details the character-
istics of these hospitalizations in comparison to those
www.cmrojournal.com ! 2014 Informa UK Ltd
 Current Medical Research & Opinion Volume 30, Number 6 June 2014

Table 1. Characteristics of cases with and without a discharge diagnosis of Table 2. Comorbidity burden, severity of disease, and outcome measures
gastrointestinal bleedinga. of cases with and without a discharge diagnosis of gastrointestinal
bleedinga.
Variableb GIB Without GIB
(n ¼ 23,383) (n ¼ 4,533,603) Variableb GIB Without GIB
(n ¼ 23,383) (n ¼ 4,533,603)
Age (median, IQR, years) 11 (15) 3 (17)
Age group (%) Elixhauser Comorbidity Score 1.1 (1.2) 0.4 (0.8)
51 year 18.9 39.8 (mean  SD)
1–5 years 18.6 14.0 Elixhauser Comorbidity Score
6–10 years 11.7 7.5 (% by score)
11–15 years 15.4 9.4 0 38.6 74.3
16–20 years 35.4 29.3 1 32.7 17.6
Sex (%) 2 16.4 5.9
Female 45.5 54.2 3 12.3 2.3
Male 54.5 45.8 Severity of illness (%)
Hospital setting (%) Minor 26.0 49.1
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 07/31/15

Rural 7.1 11.2 Moderate 34.0 36.4

Urban 92.9 88.8 Major 23.5 11.8
Teaching status (%) Extreme 16.6 2.7
Teaching 70.5 56.4 LOS (median, IQR, days) 3 (5) 2 (2)
Non-teaching 29.5 43.6 Charges (median, IQR, $) 18,651 (39,654) 9461 (15,558)
Geographic region (%) Mortality (%) 2.8 0.6
Northeast 15.7 17.1
a
Midwest 23.2 21.5 All data are weighted to produce national estimates.
b
South 37.2 29.3 For each variable there are significant differences (P50.001) between
West 23.9 22.1 patients with and without GIB.
Household income (%) GIB, gastrointestinal bleed; IQR, interquartile range; LOS, length of stay;
1st quartile 29.0 32.7 SD, standard deviation.
2nd 25.6 26.1
3rd 23.9 22.7
For personal use only.

4th 21.4 18.5

Insurance status (%)
Private 44.9
Public 40.7
None 14.2
a
All data are weighted to produce national estimates.
b
For each variable there are significant differences (P50.001) between
patients with and without GIB.
GIB, gastrointestinal bleed; IQR, interquartile range.
without GI bleeding. Overall, children with GI bleeding
differed significantly from those without. Those with a GI
bleed were older (median age 11 versus 3 years; P50.001),
more likely to be male (54.5% vs. 45.8%, P50.001), and
cared for in a teaching hospital (70.5% vs. 56.4%,
P50.001). The incidence of GI bleeding in hospitalized
children was 51.3 per 10,000 hospitalizations (0.5% of all
hospitalizations). Upper GI bleeding was over three times
more frequent than lower GI bleeding (incidence, 22.2 vs.
6.8 per 10,000 hospitalizations). However, in the majority
of cases there was not a differentiation between upper
and lower (designated as unspecified, 24.0 per 100,000
hospitalizations). Although children 16–20 constituted
the greatest overall percentage of GI bleeding (35.4%),
children 11–15 years of age had the highest overall inci-
dence of GI bleeding (84.2 per 10,000 discharges)
while children age less than 1 year of age had the lowest
overall incidence (24.4 per 10,000 discharges). Upper GI
bleeding was more frequent than lower GI bleeding in all
age groups. This location of bleeding was identified in
31.7% of infants (51 year of age) and over 40% at all
other ages.
Overall, children with a GI bleed during a hospitaliza-
43.2 tion were sicker than the general hospitalized pediatric
46.6
10.0 population (Table 2). Patients with GI bleeding had a
greater co-morbid burden (comorbidity score 3; 12.3%
vs. 2.3%; P50.001) and greater severity of illness as
assessed by APR-DRG codes (major or extreme illness;
40.1% vs. 14.5%; P50.001). Children with GI bleeding
also had longer median hospital stays (3 days vs. 2 days;
P50.001), greater median hospital charges ($18,651 vs.
$9461; P50.001), and higher overall mortality (2.8% vs.
0.6%; P50.001) compared to children without GI bleed-
ing. Using a multiple variable logistic regression model
that adjusted for demographic and comorbid conditions,
we determined that GI bleeding was independently
associated with a higher risk of mortality (aOR 1.68, CI
1.53–1.84).
Table 3 describes the incidence of GI bleeding in chil-
dren by anatomical site and the corresponding mortality.
The highest incidence of GI bleeding in the HCUP-KID
was attributable to cases coded as blood in stool (17.6 per
10,000 discharges) followed by hematemesis (11.2 per
10,000 discharges). The highest mortality associated
with GI bleeding was observed in cases with intestinal
(8.7%) and esophageal perforation (8.4%). Children
with unspecified GI bleeding accounted for the highest
overall incidence (24.0 per 10,000 discharges) followed
closely by cases of upper GI bleeding (22.2 per 10,000
discharges) while the incidence of lower GI bleeding
in children was substantially lower (6.8 per 10,000 dis-
charges). The mortality, however, was significantly

! 2014 Informa UK Ltd www.cmrojournal.com GI bleeding in children Pant et al. 1067

 Current Medical Research & Opinion Volume 30, Number 6 June 2014

higher in children with lower GI bleeding compared to bleeding (84.2 per 10,000 discharges) while children age
bleeding from other sites (P50.001). less than 1 year had the lowest overall incidence (24.4 per
Table 4 describes the incidence of GI bleeding (upper, 10,000 discharges).
lower, unspecified, and overall) by age. Children aged
11–15 years had the highest overall incidence of GI
Discussion
Table 3. Incidence and associated mortality by type of GI bleeda.
In this study, we evaluated the epidemiology of GI bleed-
Type Total Incidence Mortality ing in hospitalized children in the United States. Our data
(per 10,000 (%) represents national-level estimates and is therefore free
discharges)
from single-center or regional-level biases. Our results
Location indicate that hospitalized children with GI bleeding were
All 23,383 51.3 2.8 more likely to be male, older and admitted to teaching
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 07/31/15

Upper 10,107 22.2 2.0 hospitals and have more severe illness and a greater comor-
Lower 3106 6.8 4.0
Unspecified 10,927 24.0 3.4 bid burden compared to children without GI bleeding.
Diagnosis Previous studies have noted the association of upper GI
Blood in stool 8021 17.6 1.7 bleeding in children with male gender and the presence of
Hematemesis 5092 11.2 1.5
Unspecified 2978 6.5 7.8 concurrent of medical disorders2,13. The median length of
Rectal bleeding 1904 4.2 1.3 stay, the hospital costs and the mortality were significantly
Gastritis/duodenitis 1806 4.0 2.4 higher in the GI bleeding group. After performing multiple
Intestinal perforation 1155 2.5 8.7
Mallory–Weiss 1063 2.3 0.1 variable regression analysis, GI bleeding in hospitalized
Gastric ulcer 828 1.8 2.3 children remained an independent risk factor for mortal-
Duodenal ulcer 649 1.4 1.8 ity. The incidence of lower GI bleeding in hospitalized
Esophageal variceal 450 0.99 3.8
children was lower compared to the incidence of upper
For personal use only.

Esophageal perforation 275 0.60 8.4

Esophageal, other 153 0.34 4.6 and unspecified GI bleeding; however, the mortality was
Peptic ulcer, unspecified 97 0.21 3.1 higher in this group. National estimates suggest that in
Esophageal ulcer 86 0.19 0.0
Gastrojejunal ulcer 65 0.14 1.5 hospitalized adult patients, the incidence of lower GI
Angiodysplasia, intestine 30 0.07 3.3 bleeding is also lower compared to upper and unspecified
Angiodysplasia, stom/duod 25 0.05 0.0 GI bleeding; although the overall incidence of lower GI
Diverticular disease 20 0.04 0.0
bleeding is higher in adults, presumably due to a much
a
All data are weighted to produce national estimates. greater incidence of malignancy and diverticulosis7.

Table 4. Age-associated incidencea, percentage, and mortality by type of GI bleedb.

Age group (years)

All ages 51 1–5 6–10 11–15 16–20

All
Incidence 51.3 24.4 68.1 79.6 84.2 61.8
% across age groups 100 18.6 18.6 11.7 15.4 35.4
% within age group – 100.0 100.0 100.0 100.0 100.0
Mortality (%) 4.2 3.2 2.1 2.7 2.0
Upper
Incidence 22.1 7.7 29.6 34.5 36.4 30.2
% across age groups 100 13.9 18.7 11.8 15.5 40.2
% within age group – 31.7 43.4 43.4 43.2 49.0
Mortality (%) 4.2 2.3 0.9 2.5 1.1
Lower 6.8
Incidence 3.0 5.9 8.8 12.4 10.0
% across age groups 100 17.8 12.0 9.8 17.1 43.3
% within age group – 12.5 8.6 11.1 14.8 16.3
Mortality (%) 10.0 3.8 2.0 3.6 2.4
Unspecified
Incidence 24.0 14.3 35.0 38.7 38.5 23.5
% across age groups 100 23.6 20.4 12.2 15.1 28.8
% within age group – 58.5 51.3 48.6 45.7 38.1
Mortality (%) 3.2 4.0 3.2 2.7 3.4
a
Incidence is calculated as the number of cases per 10,000 discharges for each specified age group. All data are weighted to produce
national estimates.
b
All data are weighted to produce national estimates.

1068 GI bleeding in children Pant et al. www.cmrojournal.com ! 2014 Informa UK Ltd


When cases of GI bleeding were categorized by anatomical
site, children with blood in stool represented the highest
incidence followed by hematemesis. The highest mortality
was observed in cases with gut perforation (intestinal or
esophageal), which may be secondary to the subsequent
development of peritonitis or sepsis. Age stratification of
cases revealed that the incidence of GI bleeding was lowest
in infants and highest in the 11–15 year age group,
although children in the 16–20 year age group accounted
for more than one-third of total GI bleeding cases.
A possible explanation for the age-related increase in GI
bleeding is a co-incident increase in comorbid conditions
(e.g. portal hypertension) that directly lead to hemorrhage
Curr Med Res Opin Downloaded from informahealthcare.com by Nyu Medical Center on 07/31/15
from the GI tract.
Our study has several limitations. Our results pertain
to hospitalized children without accounting for those
who were assessed and discharged from the emergency
department. The use of large data sets might show differ-
ences without clinical significance (e.g. the differences we
observed for geographic region, household income, and
insurance status). Due to our reliance on ICD-9-CM
codes for this study, the number of upper and lower GI
bleeds have been likely underestimated because of the
unspecified GI bleed category (including blood in stool
For personal use only.
(hematochezia/melena – 578.1; hemorrhage of gastrointes-
tinal tract unspecified – 578.9). Also, since the database
does not allow for the longitudinal follow up of patients, an
undetermined number of individuals may be have been
captured more than once in our analyses. We did not
include race in our analysis since it is not reported by all
states in the KID. Finally, in our analyses, we considered
the ICD-9 code for hematemesis as a distinct entity.
We recognize that children may have been diagnosed
with hematemesis at their initial assessment and later
coded more specifically for upper GI bleeding on the
basis of subsequent investigative tests such as endoscopy.
The same criticism is applicable to our use of the ICD-9
code for ‘‘blood in stool’’ (melena).
Conclusion
In conclusion, GI bleeding is associated with a consider-
able risk for mortality in hospitalized children. These
results add to our knowledge regarding the epidemiology
of GI bleeding in children and set the foundation for
studies examining specific risk factors for this serious
health problem. The results of these types of studies can
contribute to the development of methods to prevent GI
bleeding in children.
! 2014 Informa UK Ltd www.cmrojournal.com
Current Medical Research & Opinion Volume 30, Number 6 June 2014
Transparency
Declaration of funding
This study was not funded.
Declaration of financial/other relationships
C.P., S.S., A.D., M.O., M.P.A., and T.J.S. have disclosed
that they have no significant relationships with or financial
interests in any commercial companies related to this study or
article.
CMRO peer reviewers may have received honoraria for
their review work. The peer reviewers on this manuscript
have disclosed that they have no relevant financial relationships.
Acknowledgments
The authors acknowledge all of the HCUP Data Partners
that contribute to HCUP (http://www.hcup-us.ahrq.gov/db/
datapartners.jsp).
References
1. Wolfram WAL, Arensman RM, Bensard D, et al. Pediatric gastrointestinal
bleeding. Available at: http://emedicine.medscape.com/article/1955984-
overview [Last accessed 8 July 2013]
2. Cox K, Ament ME. Upper gastrointestinal bleeding in children and adoles-
cents. Pediatrics 1979;63:408-13
3. Teach SJ, Fleisher GR. Rectal bleeding in the pediatric emergency depart-
ment. Ann Emerg Med 1994;23:1252-8
4. Lacroix J, Nadeau D, Laberge S, et al. Frequency of upper gastrointestinal
bleeding in a pediatric intensive care unit. Crit Care Med 1992;20:35-42
5. Cochran EB, Phelps SJ, Tolley EA, et al. Prevalence of, and risk factors for,
upper gastrointestinal tract bleeding in critically ill pediatric patients. Crit Care
Med 1992;20:1519-23
6. Lopez-Herce J, Dorao P, Elola P, et al. Frequency and prophylaxis of upper
gastrointestinal hemorrhage in critically ill children: a prospective study com-
paring the efficacy of almagate, ranitidine, and sucralfate. The gastrointestinal
hemorrhage study group. Crit Care Med 1992;20:1082-9
7. Zhao Y, Encinosa, W. Hospitalizations for gastrointestinal bleeding in 1998
and 2006. HCUP Statistical Brief #65. Rockville, MD: Agency for Healthcare
Research and Quality, 2008. Available at: http://www.hcup-us.ahrq.gov/
reports/statbriefs/sb65.pdf [Last accessed 8 July 2013]
8. Heaton PC, Tundia NL, Schmidt N, et al. National burden of pediatric
hospitalizations for inflammatory bowel disease: results from the 2006
Kids’ Inpatient Database. J Pediatr Gastroenterol Nutr 2012;54:477-85
9. Ma MH, Bai HX, Park AJ, et al. Risk factors associated with biliary pancreatitis
in children. J Pediatr Gastroenterol Nutr 2012;54:651-6
10. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use
with administrative data. Med Care 1998;36:8-27
11. Werner RM, Asch DA, Polsky D. Racial profiling: the unintended conse-
quences of coronary artery bypass graft report cards. Circulation
2005;111:1257-63
12. Zhu H, Hill MD. Stroke: the Elixhauser index for comorbidity adjustment of
in-hospital case fatality. Neurology 2008;71:283-7
13. Freedman SB, Stewart C, Rumantir M, et al. Predictors of clinically significant
upper gastrointestinal hemorrhage among children with hematemesis.
J Pediatr Gastroenterol Nutr 2012;54:737-43
GI bleeding in children Pant et al. 1069